Table Of ContentDraftversion February5,2008
PreprinttypesetusingLATEXstyleemulateapjv.6/22/04
SPITZER OBSERVATIONS OF THE DUSTY WARPED DISK OF CENTAURUSA
Alice C. Quillen
DepartmentofPhysicsandAstronomy,UniversityofRochester,Rochester,NY14627
Mairi H. Brookes, Jocelyn Keene, Daniel Stern, Charles R. Lawrence, Michael W. Werner
JetPropulsionLaboratory,4800OakGroveDrive,Pasadena,CA91109
Draft versionFebruary 5, 2008
ABSTRACT
Spitzer mid-infraredimages ofthe dusty warpeddisk inthe galaxyCentaurusAshow a parallelogram-shapedstruc-
6
ture. We successfully model the observed mid-infrared morphology by integrating the light from an emitting, thin,
0
and warped disk, similar to that inferred from previous kinematic studies. The models with the best match to the
0
morphology lack dust emission within the inner 0.1 to 0.8kpc, suggesting that energetic processes near the nucleus
2
have disturbed the inner molecular disk, creating a gap in the molecular gas distribution.
n
Subject headings: galaxies: structure – galaxies: ISM – galaxies: individual (NGC 5128) – galaxies:
a
peculiar
J
6
1. INTRODUCTION metry axis, becomes increasingly corrugated as a func-
1 tion of time. The short timescale estimated since the
CentaurusA(NGC5128)isthe nearestofallthe giant
v merger in NGC5128 is approximately consistent with
radiogalaxies. Because of the disk of gas and dust in its
5
the timescale suggested by the presence of tidal debris
3 centralregions,CentaurusAissuspectedtobetheprod-
andbytheshell-likefeaturescontainingatomichydrogen
1 uct of a merger of a small gas rich spiral galaxy with
(Schiminovich et al.1994;Peng et al.2002)andmolecu-
1 a larger elliptical galaxy (Baade & Minkowski 1954).
largas(Charmandaris et al.2000). Analternativemodel
0 Numerical simulations of such mergers predict large
accounting for the warped disk is the polar ring model
6 shell-like features (Hernquist & Quinn 1988) that have
0 been observed in CentaurusA over a large range of by Sparke (1996), consistent with the polar orbit of the
/ radii (Malin et al. 1983; Peng et al. 2002). Some con- disk implied by the galaxy isophotes in the outer galaxy
h
(Malin et al. 1983; Haynes et al. 1983; Peng et al. 2002)
p tain atomic and molecular gas, implying that they
but requiring a longer timescale to account for the twist
- were formed from material stripped from a galaxy rel-
o atively recently, fewer than a few galactic rotation pe- of the warp.
r riodsago(Schiminovich et al.1994;Charmandaris et al. More recent observations of the central region include
st 2000). Tidal features associated with this debris have submillimeter imaging with the Submillimeter Com-
a mon User Bolometer Array (SCUBA) and mid-IR imag-
also been identified, supporting the relatively short (few
v: times 108 years) estimated timescale since the merger ing with the Infrared Camera (ISOCAM) on the In-
frared Space Observatory (ISO) satellite (Mirabel 1999;
i took place (Peng et al. 2002).
X Leeuw et al.2002). Atthesewavelengths,the dustydisk
In its central regions, NGC5128 exhibits a well rec-
r ognized, optically-dark band of absorption across its isseeninemissionratherthanabsorption. At100–200pc
a from the nucleus the galaxy contains a molecular cir-
nucleus. This dusty disk was first modeled as a
cumnuclear disk that has been studied in molecular line
transient warped structure by Tubbs (1980). Bland
emission (Israel et al. 1990, 1991) and resolved in Paα
(1986), Bland et al. (1987), Quillen et al. (1992), and
emission (Schreier et al. 1996; Marconi et al. 2001). For
Nicholson et al. (1992) found that the kinematics of
a recent summary of the wealth of observational stud-
the ionized and molecular gas are well modeled by
iescarriedoutonthis peculiarandactivenearbygalaxy,
a warped disk composed of a series of inclined con-
seethe comprehensivereviewby Israel(1998). Basedon
nected rings undergoing circular motion (as also ex-
the discussion by Israel (1998), we adopt a distance to
plored for other galaxies with peculiar morphology by
CentaurusA of 3.4Mpc. At this distance 1′ on the sky
Steiman-Cameron et al. 1992). The model explored by
corresponds to ∼1kpc.
Quillen, Graham & Frogel (1993) modified the kine-
In this manuscript we focus on the geometry of the
matic model by Quillen et al. (1992) to fit the morphol-
dusty disk in CentaurusA as seen from Spitzer Infrared
ogy of the absorptive, dusty disk seen in near-infrared
ArrayCamera(IRAC) images. Theseimagesresolvethe
images and proposed a timescale of about 200 million
structure of the disk more clearly than previous obser-
years since the core of an infalling spiral galaxy reached
vations and also show the disk out to larger radii. They
and merged with the elliptical galaxy nucleus. An ini-
provide us with an opportunity to study the geometry
tiallyflatdisk,misalignedwiththegalaxyprincipalsym-
ofthewarpeddiskinmuchhigherdetailthanpreviously
Electronicaddress: [email protected] possible. Observationsaredescribedin§2. Ourgeomet-
Electronicaddress: [email protected] ric model is described in §3. A discussion and summary
Electronicaddress: [email protected] follow in §4.
Electronicaddress: [email protected]
EElleeccttrroonniiccaaddddrreessss:: [email protected][email protected] 2. OBSERVATIONS
2
Figures 1 and 2 present images of NGC5128 taken on starlight in the longer wavelength 5.8 and 8µm bands.
2004 February 10 in the 3.6, 4.5, 5.8 and 8.0µm broad- In visible or near-infraredbands, folds can correspond
band filters (channels 1-4) of the Spitzer Infrared Array toregionswheretheabsorptionfromdustobscuresback-
Camera (IRAC; Fazio et al. 2004). In each filter, fixed ground starlight. In this case folds that are closer to
cluster observing mode was used for these observations the observer absorb more backgroundstarlight from the
to produce a map of 5×6 points, at which five dithered galaxy. Thus some of the features seen in emission in
exposures were taken. The exposure time per frame was the IRAC images resemble and coincide with absorption
12s. Additionalshorterexposure,0.4sframes,werealso bands previously seen in near-infrared images. In Fig-
taken to correct for possible saturation in the longer ex- ure 3 we show a color map made from 2Micron All Sky
posureframes. Thecoverageofthemapateachposition Survey (2MASS) images from the 2MASS large galaxy
on the sky varies between three and ten frames, with atlas(Jarrett et al.2003). Thesouth-easternedgeofthe
an average coverage of six frames, corresponding to an parallelogramin the IRAC images lies in the same loca-
exposure time of 72s. tion as the absorption band ∼ 10′′ to the south east of
Before mosaicing, the basic calibrated data (BCD) the nucleus prominent in the near-infraredimages. This
frames were corrected for artifacts using the IRAC arti- is consistent with the study by Leeuw et al. (2002), who
factmitigationcode (excluding the pulldowncorrection) compared submillimeter images to near-infraredimages.
available from the Spitzer Science Center contributed The edge of emission ∼ 1′ to the north of the nucleus
software pages1. The final mosaiced images were pro- (atasurfacebrightnessbelow thatofthe parallelogram)
ducedfromtheditheredframesbyapplyingtheMOPEX in the 8.0µm IRAC image corresponds to the top of the
software2 to the corrected BCD frames. dustlaneprominentinopticalimagesofthegalaxy. This
Here we focus on the central 5′×5′ region in the final correspondstotheextinctionfeatureinthenear-infrared
mosaics. Themosaicedimagesprovideaplatescalewith color map to the north of the nucleus. From comparing
1′.′2 pixels; FWHMs of the point spread functions are theIRACimagestothe2MASSimagesweinferthatthe
1.7, 1.7, 1.9, 2′.′0 in channels 1–4, respectively. The rms southern side of the parallelogram is closer to the ob-
noise levels in these images are 0.025, 0.024, 0.070, and server than the northern side. Because the dust on this
0.060MJysr−1 in channels 1–4, respectively. The above side lies in front of the plane perpendicular to the line
sensitivitiesagreewiththepredictionsoftheSENS-PET of sight containing the galaxy nucleus, it absorbs more
sensitivityestimatorinlongerwavelengthschannels. The backgroundstarlight and so causes a deeper band of ex-
3.6µm channel contains emission from the stellar com- tinction in the near-infrared images. The northern side
ponent of the galaxy that extends over a large portion oftheparallelogramliesontheoppositesideandsoisnot
of the field of view. This makes the 3.6µm image about seen in the near-infrared images. Likewise the oval edge
halfassensitiveastheSENS-PET predictionforthedust of emission ∼ 1′ to the north of the galaxy nucleus (see
emission. Fig. 2) is nearer the observer than that on the opposite
The IRAC images presentedhere are deeper than pre- edge∼1′tothesouthofthenucleus,andonlythenorth-
viousISOCAMimages(Mirabel(1999)),andhavehigher ernsidecausesanabsorptionfeatureinthenear-infrared
angular resolution than previous submillimeter images images and color maps (see Fig. 3).
((Mirabel 1999; Leeuw et al. 2002)). In the inner few
arcminutes, the IRAC images exhibit a parallelogram 3. MODELINGOFTHEWARPEDDISK
shape in emission (see Fig. 2). This shape, previously At many observed wavelengths, the morphology of
seen at lower angular resolution, was interpreted as an CentaurusA is well reproduced by geometric mod-
S-shape, possibly associated with shocks (Mirabel 1999; els of a warped disk (Bland 1986; Bland et al. 1987;
Leeuw et al. 2002). A parallelogram morphology has Quillen et al. 1992; Nicholson et al. 1992; Quillen et al.
beenseenpreviouslyinothergalaxies. Forexample,dust 1993; Sparke 1996). We describe and extend such mod-
absorption features in the SO galaxy NGC4753 exhibit eling here.
a parallelogram shape, and have been modeled with a A warped disk can be described as a series of tilted
warped twisted disk by Steiman-Cameron et al. (1992). rings, each with a different radius r. We assume that
When an optically thin warped disk is seen in emission, the gasand dustare evenlydistributed oneachring and
the edges of folds in the disk correspond to regions of each ring is smoothly connected to those at larger and
higher surface brightness. Multiple folds are seen along smallerradii. Wefollowthenotationandframeworkused
the line of sight (e.g., Bland 1986; Bland et al. 1987; by Quillen et al. (1992, 1993) to describe the geometry
Quillen et al. 1993; Nicholson et al. 1992) implying that of a warped disk. Each ring is described by two angles,
somepartsofthediskareseenthroughotherpartsofthe a precession angle α(r) and an inclination angle ω(r).
disk. In this case the morphology would not be nearly These angles are given with respect to an assumed prin-
symmetric across the origin (r → −r), as observed, un- cipal axis of the underlying elliptical galaxy. This axis
less the outer disk was nearly optically thin in the mid- requires two angles to describe, χ, corresponding to the
infrared. We infer that the disk probably has a low nor- position angle counter clockwise from north of the axis
mal optical depth in the mid-infrared (if observed face- on the sky, and an inclination angle, ϑ, that describes
on), though the bright edges of the folds and individual the tilt of this galactic axis with respect to the line of
clumps in the disk may not be optically thin. The par- sight(seeFig.6byQuillen et al.1993). Weassumethat
allelogram is present in all four IRAC bands, though it the galaxy is axisymmetric and not triaxial so a third
ismostprominentcomparedtothediffuseemissionfrom angle is not requiredto describe its orientation. We also
assumethatthegalaxyshapeisfixed,andnottumbling.
1http://ssc.spitzer.caltech.edu/archanaly/contributed/browse.html Our description for the ring projection angles can be
2 http://ssc.spitzer.caltech.edu/postbcd/ related to those of previous works. While Quillen et al.
3
(1992,1993)describedtheanglesofthewarpeddiskwith precessionangle used previously by Quillen et al. (1993)
respect to the principal axis of the underlying elliptical from matching the near-infrared morphology is shown
galaxy, Bland et al. (1987) and Nicholson et al. (1992) as a dotted line in this figure and is similar to our best
matched the Hα velocity field with a tilted ring model matching model. The Spitzer data allow us to better
describing the orientation of the rings with respect to study the outer parts of the disk than possible with the
the line of sight. This model fit the velocity field by old 12CO(2-1) spectra and near-infrared images. Thus
adjustingtheringinclinationasafunctionofradius,i(r), there are some differences in the precession angle α(r)
and position angle counter clockwise from north, p(r). between our old model and our newer one, as shown in
To produce a model image of the mid-infrared emis- Figure 6. To better match the morphology in the outer
sion,wemustconsiderallemittingandabsorbingregions parts of the disk we allowedthe disk to twist further (to
along the line of sight at each position on the sky. This higher a higher value of α) before decreasing at larger
is much simpler when the system is optically thin. In radius.
this case, we sum all emission along the line of sight at Our model predicts that the disk alternates between
eachpositiononthe sky. The nearsymmetryofthe disk havingthe southernsideandnorthernsidenearest,with
suggests that we can use an optically thin approxima- folds both above the nucleus. The near-infrared images
tion to model this disk. While individual clouds could showastrongdustabsorptionfeaturetothesouthwestof
contain optically thick regions, the bulk of the emitted thenucleus correspondingto aninnerfold,andaweaker
mid-infrared light is likely to reach the observer. When feature north of the nucleus corresponding to an outer
the disk is optically thin, brighter areas correspond to fold (Quillen et al. 1993). In Figure 5 we show the rings
regions that appear folded from the perspective of the comprising our warp model, projected on the sky. The
viewer. Here we neglect emission from a spherical com- nearer semi-circle of each ring is shown in red, whereas
ponentassociatedwiththe starsandonlyconsideremis- the more distant semi-circle is shown in blue. The disk
sion from a thin but warped disk. atr ∼60′′ is nearestthe observeronthe south-eastside.
Ournumericalprocedurebeginsbyrandomlysampling This regioncorrespondsto a fold inthe disk that is seen
x,y positions inthe plane perpendicular to the principal in the near-infrared color map, 10′′ to the south-east of
axisoftheellipticalgalaxy. Ateachpositionwecompute thenucleus(seeFig.3). Atr &100′′thenorthernsideof
a z-coordinate based on a smooth (spline) function for the disk is again nearest the observer. This corresponds
thediskprecessionandinclinationangles,α(r)andω(r). to the northern dust lane seen in the the optical images
Thepointsthatspecifythesesplinefunctionsarelistedin and the near-infrared color map about 45′′ north of the
Table1. The coordinatesforeachpointarethenrotated nucleus.
into the viewer’s frame. Points along each line of sight
3.1. Comparison to previous studies on the geometry of
aresummedtoproduceamodelimage. Forthefunctions
the warped disk
α(r) and ω(r), we began with those from Quillen et al.
(1993) and slightly varied the angles at different radii In Figure 7a,b, we show the inclination and position
to achieve a better match to the observed morphology. anglesasa functionofradiuswith respectto the viewer;
Matching was done by visual comparison to the IRAC functions that can be directly compared to those found
images. No minimizing fit to the image data values was byNicholson et al.(1992)fromtheirkinematicfitstothe
done. Our procedure is adequate to understandthe pro- Hα velocity field (see their Fig. 9a,b). Nicholson et al.
jection affects associated with an emitting warped disk. (1992)adoptedadistanceof3Mpctothegalaxy,sotheir
Future and more intensive modeling would be required linearscaleis15%differentthanours;wehavecorrected
to do a multi-dimensional and multi-wavelength fit. for this difference by rescaling the given distances. We
To construct a model, we must consider the thickness see from Figure 7 that the position angles of their tilted
of the disk and its brightness distribution. Because the ringfitarequitesimilartothosepredictedbyourmodel.
disk is not infinitely thin we randomly chose slight off- The warp shape originally designed to match the ve-
setsfromthez-coordinatecomputedfromtheprecession locity field in molecular gas (Quillen et al. 1992) re-
and inclination angles. The size of the offset depends sembled that which matched the Hα velocity field by
on a disk aspect ratio k(r)=h/r, where h(r) is the disk Nicholson et al. (1992). Both models included an in-
thicknessasafunctionofradius. Weassumedforthedisk clined disk that tilted so that gas rings at different radii
aspectratioapowerlawform,k(r)=k50(r/50′′)βz. The alternated between being retrograde and prograde with
intensity of the emission contributed at each point was respect to the observer. This was also a characteris-
then computed assuming that the disk volume emissiv- tic of the high inclination warp model for NGC4753
ity integrated through the disk vertically is a power law by Steiman-Cameron et al. (1992). The tilted ring fit
function of radius, ǫ(r) ∝ r−βe. We also allowed an in- by Nicholson et al. (1992) to the velocity field did not
nergapintheradialsurfacebrightnessprofile,rgap. The specify which side of the each ring was closest the
contributionfromtheactivenucleusandthe∼100pccir- viewer. However Bland (1986), Bland et al. (1987),
cumnucleardisk arenottakenintoaccountinourmodel Nicholson et al. (1992), and Quillen et al. (1993) used
sobyagapwemeanadeficitwithinrgap andthatofthe the visible andnear-infraredimagestobreakthis degen-
circumnuclear disk at ∼6′′. eracy. Thediskistiltedsothatitisnearerthevieweron
A modelintensity image computedas described above the southern side for intermediate radii; accounting for
is shown in Figure 4 along with the 8.0µm IRAC image thesouthernedgeofthedustlaneseeninopticalimages,
of the galaxy. The numerical parameters of this model andnearerthevieweronthenorthernsideatlargestradii
are summarized in Table 1. The angle precession and to account for the northern dust lane seen in the optical
inclination angular functions, α(r) and i(r), are shown and near-infrared images. These flips in orientation can
inFigure6asasolidanddashedlines,respectively. The be seen from Figure 7b. They occur where the inclina-
4
tionwithrespecttotheviewercrosses90◦atwhichpoint radii, a nested set of parallelogram shapes can be seen
the disk is edge-on. (e.g., as in M84; Quillen & Bower 1999). However in
The model by Quillen et al. (1993) had a somewhat CentaurusA, an outer oval is seen in the outer disk (see
lower inclination for the galaxy principal axis, ϑ = 65◦, Fig. 8 and 5) that is slanted in the opposite direction as
instead of 75◦, used here. Their model also hada higher the inner parallelogram. This difference in the direction
disk inclination with respectto the galaxyprincipalaxis ofthe slantis a featureofthe signchangeinthe slope of
at smaller radii. There is redundancy in the model be- the precession angle.
tween the inclination of the disk with respect to the The precessionrate of a gas ring inclined with respect
galaxy axis, ω(r), and the tilt of the galaxy principal to the underlying galaxy undergoing circular motion is
axis, ϑ. The maximum and minimum ring inclinations proportionaltotheellipticityofthegalacticgravitational
correspond to values set by ϑ ± ω (see discussion by potential and the angular rotation rate of the ring (e.g.,
Quillen et al.1992). Toexhibittheparallelogramshape, Tubbs 1980; Sparke 1984). If the self gravity of the ring
the disk inclination with respect to the viewer must go is important then it too can affect the precession rate
above and below 90◦ causing folds in the disk. This re- (Sparke 1986; Arnaboldi & Sparke 1994; Sparke 1996).
stricts the models to a range of values ϑ+ω ∼100◦. The direction of precession depends on the orientation
Here, we find a somewhatbetter match (see Fig. 8) to of the ring and galaxy(whether a polar ring or not) and
themid-infraredimageswithamodelthathasdecreasing whether the galaxy is prolate or oblate. Most models
inclination (with respect to our estimated galaxy princi- assumethatfollowinga mergerthe gasanddustaredis-
pal axis) at larger radii. However this decrease may be tributed in a plane that is misaligned with the principal
spurious. To compute the model at sufficient resolution axesofthegalaxy. Tobeconsistentwiththedirectionof
to compare to the inner region, we cannot well sample the twist in the outer parts of CentaurusA (decreasing
points over a large region. By accounting for every ob- α(r) at r & 100′′), the galaxy can either be prolate and
servedfeaturewemayhaveachievedabettermorphology the ring locatednear the equatorialplane (Quillen et al.
match with a model that has exaggerated variations in 1993) or the galaxy can be oblate and the ring would be
the ring angles. In other words, the angular variations nearly polar (Sparke 1996).
mayhavebeencompressedintoasmallerradiithanthey Thereversalinthesignoftheslope(dα/dr)isunlikely
should be. to be caused solely by the shape of the rotation curve.
To be more certain of the orientation of the outer ThepredictedrotationcurveofCentaurusAincreasesat
disk, better kinematic constraints on this outer disk smallerradius;allthe waytoaradiusof10′′ (seeFig.13
are needed. By specifying a choice for the principal by Marconi et al.2005). Consequently the angularrota-
axis of the galaxy, we have chosen to describe the ge- tionrateshouldincreaseastheradiusdecreasesnearlyall
ometry of the warped disk with respect to a particu- the way to the galaxy center. The rotation curves used
lar axis. This choice helped Quillen et al. (1993) com- by Quillen et al. (1993) and Sparke (1996) were nearly
pare the shape of the disk to the predictions of sim- solid-body at small radii, and so underestimated the an-
ple merger and precessing ring models. However, it gular precession rate at small radii (within ∼1.5′ of the
is not necessarily significant that the disk inclination nucleus).
as measured with respect to our assumed galaxy axis The reversal in the sign of the slope (dα/dr) could
varies slowly with radius. An additional complexity be due to a drop in the galaxy eccentricity (exploited
not considered with our choice of projection angles is by both Quillen et al. 1993; Sparke 1996) that would re-
that the galaxy may be triaxial, or significantly vary duce the precession rate at small radii. To account for
in shape and orientation with radius (e.g., as consid- thechangeinslopeofαatr ∼100′′,Quillen et al.(1993)
ered by Arnaboldi & Sparke 1994). The galaxy could assumed a sharp cutoff in the ellipticity of the gravita-
be in the process of dynamic relaxation following the tional potential of the galaxy as a function of radius.
merger. Analytical models or one-dimensional integra- For r < 80′′, the potential ellipticity was much reduced
tions fail to capture the complexity of more detailed nu- in the model by Quillen et al. (1993) compared to that
mericalsimulations,particularfornearequalmassmerg- outside it. The galaxy isophotes are best viewed in the
ers (e.g., Mihos 1999; Mihos & Hernquist 1996). Imag- short wavelength 3.6µm IRAC image where the stellar
ing and kinematic studies of the outer galaxy suggest lightmostcontributesto the flux andthe extinctionand
that these additional degrees of freedom are important emission from dust is minimized compared to that at
(Malin et al. 1983; Schiminovich et al. 1994; Hui et al. shorterwavelengthsinthenear-infrared. Galaxyisopho-
1995; Peng et al. 2002). tal contours for this image are shown in Figure 9.
We first examine the isophotes at 3.6µm to search
3.2. In context with the dynamical warp models
for evidence of the large scale stellar bar proposed by
The model precession angle α(r) reaches a maximum Mirabel (1999). If such a bar were responsible for the
at r ∼ 100′′, decreasing at both larger and smaller radii parallelogramshape in the mid-infrared images, the bar
(see Fig. 6). This implies that the corrugated disk is would be viewed at intermediate inclinations (not edge-
twisted in one direction for r . 100′′ and in the op- on), and so should be evident in the isophote shapes.
posite direction for r & 100′′. Since the near-infrared However, the isophotes do not exhibit a region of flat
morphology depends on which side of each ring is closer surface brightness or a change in ellipticity or position
to the observer, and our model is similar to that used angle overa shortrange in radius. These features would
to match the near-infrared images (also see discussion be expected at the end of a stellar bar. We conclude
by Bland et al. 1987; Nicholson et al. 1992), this pecu- that there is no evidence for a large scale,few kpc sized,
liar change in the handedness of the twist is likely to be stellar bar at the heart of CentaurusA.
real. Whenthehandednessofthetwististhesameatall The galaxy isophotal ellipticity does decrease in the
5
inner regions. By fitting contours to the 3.6µm image The model that is most similar to the IRAC images
outside the emitting disk, we measured a galaxy isopho- (shownin Fig. 4) contains a gapin the dust distribution
tal ellipticity for the stellar distribution (ǫ = 1− b/a, withouterradiusr ∼50′′. Wecomparethemodelshown
wherebandaarethesemi-minorandsemi-majoraxes)of in Figure 4 to a similar one that lacks the deficit in the
ǫ∼0.1 at r ∼4′, ǫ∼0.05 at r ∼2′, and ǫ∼0 at r<1′. dustdistribution. ThismodelisshowninFigure10a,and
Because the gravitational potential is a convolution of does not match the observedmorphologyas well as that
the density profile with a 1/r function, the gravitational containing a gap in the dust distribution. The smooth
potentialcontoursaresmootherthantheassociatedden- continuationoftheprecessionangleintotheinnerregion
sity distribution. However,the isophotes are likely to be results in an edge-on disk with respect to the viewer at
rounder and smoother than the actual density distribu- some point within r =60′′. This causes the sharpbright
tion. While the drop in background galaxy ellipticity linear feature at r < 60′′ seen in Figure 10a that is not
exploitedbyQuillen et al.(1993);Sparke(1996)intheir exhibited by the IRAC images.
dynamical models is real, it may not be sharp enough If the disk has a lower inclination with respect to the
to account for the abrupt drop in the precessionrate in- viewer, the observed surface brightness is reduced. We
ferredinthecentralregionofthegalaxyfromthechange considerthepossibilitythatthediskprecessionangledif-
in the slope of the precession angle α at r ∼100′′. fersintheinnerregionfromthatexpectedfromasmooth
We now consider the role of the mass in the disk. continuation of our model. To test this possibility, we
Sparke (1996) showed that the self gravity of the disk computedamodelthathasaflattened(non-twisted)disk
couldincreasetheprecessionrateandvarytheprecession at r . 60′′. This model is shown in Figure 10b. The
axisinthe centralregion. Sparke(1996)onlyconsidered sharp,bright,inner feature alongthe east-westdirection
the mass in the molecular and atomic gas components. seeninFigure10aisnotevidentinFigure10b. However,
Quillen et al. (1993) noted that there were extensions in the morphology of this model also does not display the
the K-band isophotes that were not reproduced by the characteristics of our preferred model shown in Figure 4
purelyabsorptivedisk model. The K-bandisophotesare that better resembles the IRAC images. In particular,
extended ata radius of about60′′ in the 15 mag/arcsec2 the triangular features to the south-east and north-west
contour (also see the dereddened contour map shown in of the nucleus that are seen in the near-infrared color
Figure 9 by Marconi et al. 2005). We can compare the mapsonthe south-eastside (Quillen et al.1993)arenot
massinapossiblediskgasandstellarcomponenttothat as good a match to those of the IRAC images. To re-
in the underlying galaxy. With a circular rotational ve- move the bright inner feature of Figure 10a, the preces-
locity of 250kms−1, a total mass of ∼ 1010M⊙ is en- sion angle must remain above α > 270◦. However, the
closed within a radius of 1kpc. The mass in molecular curved triangle edge of emission in the parallelogram is
gas in the same region is a few times 108M⊙ (Phillips not present if the precessionangle does not steeply drop
et al. 1987, corrected for the difference in the assumed between a radius of 100′′ and 50′′. We have tried mod-
distance). The level of 15magarcsec−2 in K-band cor- elswithbothincreasinganddecreasingprecessionangles
responds to a surface density of 4500M⊙pc−2 assuming near the nucleus, finding no improved match to the ob-
a mass to light ratio at K-band of 0.5 (e.g., Silge et al. servedmorphology. Ourbestmatchisthemodelwithan
2005). The actual surface density would be ∼ 10 times inner gap in the dust distribution. We explored models
lower than this, taking into account the mean disk axis with a lower but constant surface density within the es-
ratio on the sky. This surface density can be compared timatedgapradiusof∼50′′, andcanexcludethosewith
to the estimated surface density of molecular gas, or a a surface density in the gap that is above 1/5 of that at
few hundred M⊙pc−2. This suggests that at least an the outer gap edge.
equal mass exists in stars in the disk as in gas. Previous studies have discussed the possibility of a
Comparing the total mass in gas and stars in the disk deficit in the gas distribution in the same region as we
to that in the underlying spherical component, we esti- findadeficitinthedustdistribution. At100–200pcfrom
mate that a few percent of the total mass within 1kpc the nucleus there is a circumnuclear molecular disk that
liesinthedisk. Themassinthediskislikelytobeafew has been studied in molecular line emission (Israel et al.
times larger than that used by Sparke (1996) to account 1990,1991). Adeficitintheionizedgasdistributionout-
forthe diskgeometry. We supportthe finding bySparke sidethecircumnucleardiskwasseeninthePaαkinemat-
(1996) that the self-gravityin the disk is important,and ics by Marconi et al. (2001). In Paα, the circumnuclear
soshouldsignificantlyaffectthediskprecessionrate. Fu- disk at a radius of r ∼ 6′′ is seen, and so is emission at
ture modifications of the prolate model should take this significantly lower velocities (hence inferred larger radii)
into account, as the self-gravityin the disk couldchange and significantly higher velocities (within the sphere of
the direction of precession in this region. If the outer influenceofthemassiveblackhole). HoweverPaαemis-
disk is prolate, then a reversalin the twist could be due sion is lacking at intermediate radii and velocities. This
to the self-gravity of the disk in the inner region. If the lackofemissionwouldbe expectediftherewereadeficit
outer galaxy is oblate, then the ring is polar and the of gaseous material between the radii of ∼ 6′′ (set by
model by Sparke (1996) would account for the reversal the estimated outer radius of the circumnuclear disk,
in the twist of the disk. Both dynamical models could Schreier et al. 1996; Marconi et al. 2000, 2001) and 50′′
be updated to include better estimates of the mass in (estimated from our model).
the disk, the galaxy isophotes at 2.2 and 3.6µm, and an The rotation curve previously fit to the CO and Hα
improved rotation curve based on the light distribution. kinematics rose linearly (solid body) within a radius of
1′ of the nucleus (Quillen et al. 1992; Nicholson et al.
3.3. A gap in the dusty disk between a radius of 6′′ and
1992). Because an edge-on gas ring appears linear on a
50′′
position-velocity diagram, a gas disk with an inner hole
6
can mimic or be confused with a gas disk extending all the dusty disk at 6′′ .r .50′′. A gap exists in same re-
the way to the nucleus in a galaxy with a linearly rising gionin the gas distribution. Marconi et al. (2001) saw a
rotationcurve. Nicholson et al.(1992)showedthatthere deficit of Paα emission near the nucleus at intermediate
wasadiscrepancybetweentheirmeasuredrotationcurve velocities. Nicholson et al.(1992)suggestedthatthedis-
andthatexpectedfromar1/4ordeVaucoulerslaw. They crepancy between the measured linearly rising rotation
listed a possible hole in the HII region distribution as a curvewithinaradiusof60′′mightbeexplainedbyahole
possiblecauseforthisdiscrepancy. Alinearlyrisingrota- in the ionized gas distribution. It is not easy to deter-
tioncurvewithinaradiusof1′ ofthe nucleusis notcon- mine if there is a gap in the dust distribution since the
sistentwiththeK-bandsurfacebrightnessprofile(seethe infraredsurface brightness depends on the inclination of
rotation curve predicted in Figure 13 by Marconi et al. the disk and the disk is highly corrugated. If the disk
2005). The apparent region of solid body rotation need twists to lower inclinations (closer to face-on) at small
not be accountedfor with a galacticbar, as proposedby radii, it would have a lower surface brightness near the
Mirabel (1999), and can be better explained by a gap in nucleus. We have explored models with smoothly vary-
the gas and dust distribution. ingradialsurfacebrightnessdistributions,andsmoothly
varying precession and inclination angles. We find that
4. SUMMARYANDDISCUSSION only models with a deficit of dust interior to ∼ 50′′ re-
By integrating the light through an emitting, opti- semblethemid-infraredimages. Weexcludemodelsthat
cally thin, dusty, and warped disk, we have successfully havemorethan∼1/5thesurfacedensitywithinr=50′′
matched the morphology of CentaurusA seen in mid- asatthat radius. We conclude that there is a gapin the
infrared IRAC images. We confirm previous proposals gas and dust distribution between 0.1 and 0.8kpc from
that the disk morphology is well explained by a warped the nucleus. The inner radius of the gap we have taken
disk (Bland 1986; Bland et al. 1987; Quillen et al. 1993; from studies of the circumnuclear disk (Marconi et al.
Nicholson et al. 1992; Leeuw et al. 2002) rather than a 2001;Israel et al.1991),andtheouterradiusisestimated
barred one (Mirabel 1999). The disk is nearly edge-on from our model.
with respect to the viewer, but tilts so that folds ap- It is interesting that only the region between the cir-
pear above and below the galaxy equator. There is a cumnuclear disk (100–200pc) and ∼ 0.8kpc has been
fold south of the nucleus at a radius r ∼60′′ responsible depleted of gas and dust. While we find no evidence
for high extinction seen in near-infrared images, and a for a large scale stellar bar in CentaurusA, a dense gas
fold north of the nucleus at larger radii responsible for diskcouldhaveexhibiteddynamicalinstabilitiesdeposit-
the northern edge of the dust lane seen in optical and inggasintothecircumnucleardisk(e.g.,Shlosman et al.
images. Extinctionfeatures seenin the near-infraredex- 1989). Energetic star formation or activity associ-
tinction map correspond to folds in the disk that are ated with the black hole can deplete and evacuate the
located nearer the observer and so absorb more back- central region of a galaxy (e.g., Springel et al. 2005;
groundstellarlightfromthegalaxy. Inthemidinfrared, Veilleux et al. 2005), though it is not clear how a cir-
however,folds on both the near and opposite side of the cumnuclear disk would be protected from or reformed
galaxy correspond to bright emission features. The in- following this activity. Nearby galaxies exhibit evacu-
ner folds account for the parallelogram shape, while the ated central regions or circumnuclear rings. For exam-
outerfoldsaccountforthenorthernedgeofthedustlane ple, the Circinus galaxy has a ∼ 500pc radius molecu-
seeninopticalimagesandafainterovalofemissionseen lar ring (Curran et al. 1998) and contains a Seyfert nu-
outside the parallelogramin the IRAC images. cleus. M82containsa1kpcradiuscircumnuclearmolecu-
The disk geometry we use to match the mid-infrared larring;however,apreviousepochofstarformationhas
morphology is similar to that found previously by occurred within this ring (e.g., Forster Schreiber et al.
Quillen et al. (1993), and is also similar to that required 2003). Futureobservationalstudiesmaydifferentiatebe-
to fit the CO and Hα velocity field (Bland et al. 1987; tween the role of star formation, dynamical instabilities
Quillen et al. 1992; Nicholson et al. 1992). The geomet- andnuclearactivityindisruptingthe gasanddustydisk
ricwarpmodelsbyQuillen et al.(1993);Nicholson et al. in CentaurusA.
(1992) have been predictive; they provide good matches The best matching geometric warp model requires a
to the mid-infrared morphology. Some differences in the change in the slope of the precession angle at a radius
precession angle exist in the model at r & 100′′ com- of about r ∼ 100′′. Two previous models account for
pared to previous work. Previous CO, Hα spectra and this twist. Quillen et al. (1993) used a model in which
near-infrared imaging were not sensitive enough to pro- the galaxywasprolateinits outerregionandthe galaxy
vide tight kinematic constraints on the outer disk. The ellipticityabruptlydroppedtozerowithinr ∼80′′. How-
Spitzer data have allowed us to extend the model past ever,thisabruptdropisnotconsistentwiththeisophote
r ∼ 100′′ compared to previous models and see closer shapesat3.6µm. Aprolatemodel(withtheadvantageof
in to the nucleus where the extinction is high at shorter a relatively short timescale) might account for the disk
wavelengths. Better constraints on the disk geometry geometry if modified to include the self-gravity of the
couldbeachievedinthefuturebyfitting observationsat disk. The polar ring model by Sparke (1996) naturally
more than one wavelength. For example, modern high accountsfor the reversal,but requiresa longertimescale
resolutionkinematic observations that are fit along with to operate. These dynamical models could be updated
theSpitzer datausingthesamemodelwouldallowmuch to use a more accurate mass distribution and rotation
stronger constraints on the disk geometry and gas and curve. Improvementstothesemodelsmayleadto better
dust distribution than the slight modification to previ- understandingofthegalaxymergerthatcreatedCentau-
ous geometric models that we have used here. rusA’s peculiar morphology,as well as the merger’s role
The warp disk model suggests that there is a gap in in feeding the active galactic nucleus.
7
WethankJossBland-Hawthorn,VaroujanGorjianand work was in part provided by National Science Founda-
Vassilis Charmandaris for helpful comments and sug- tion grant AST-0406823, and the National Aeronautics
gestions. We thank the Research School of Astronomy and Space Administration under Grant NNG04GM12G
and Astrophysics of the Australian National University issued through the Origins of Solar Systems Program.
andMountStromloObservatoryforhospitalityandsup- We acknowledgesupportby awardHST-GO-10173-09.A
port for ACQ during Spring 2005. Support for this through the Space Telescope Science Institute.
REFERENCES
Arnaboldi,M.&Sparke,L.S.1994, AJ,107,958 Mihos,C.1999, Ap&SS,266,195
Baade,W.,&Minkowski,R.1954,ApJ,119,215 Mihos,J.C.,&Hernquist,L.1996, ApJ,464,641
Bland,J.1986,Ph.D.thesis,Univ.Sussex Nicholson, R. A., Bland-Hawthorn, J., & Taylor, K. 1992, ApJ,
Bland,J.,Taylor,K.,&Atherton, P.D.1987,MNRAS,228,595 387,503
Charmandaris,V.,Combes,F.,&vanderHulst,J.M.2000,A&A, Peng, E. W., Ford, H. C., Freeman, K. C., & White, R. L. 2002,
356,L1 AJ,124,3144
Curran, S. J., Johansson, L. E. B., Rydbeck, G., & Booth, R. S. Phillips, T. G., Ellison, B. N., Keene, J. B., Leighton, R. B.,
1998, A&A,338,863 Howard, R. J., Masson, C. R., Sanders, D. B., Veidt, B., &
Fazio,G.G.etal.2004,ApJS,154,10 Young,K.1987,ApJ,322,L73
Forster Schreiber, N. M., Genzel, R., Lutz, D., & Sternberg, A. Quillen,A.C,,Graham,J.R.,&Frogel,J.A.1993,ApJ,412,550
2003, ApJ,599,193 Quillen,A. C.,de Zeeuw, P. T.,Phinney, E. S.,& Phillips,T. G.
Haynes, R.F.,Cannon, R. D.,& Ekers R.D.1983, ProcASA,5, 1992,ApJ,391,121
241 Quillen,A.C.,&Bower,GaryA.1999, ApJ,522,718
Hernquist,L.,&Quinn,P.J.1988, ApJ,331,682 (CentaurusA)
Hui,X.,Ford,H.C.,Freeman,K.C.,Dopita,M.A.1995,ApJ,449, Schiminovich, D., van Gorkom, J. H., van der Hulst, J. M., &
592 Kasow,S.1994,ApJ,423,L101
Israel,F.P.1998, A&ARv,8,237 Schreier,E.J.,Capetti,A.,Macchetto,F.,Sparks,W.B.,&Ford,
moleculardisk H.J.1996, ApJ,459,535
Israel, F. P., van Dishoeck, E. F., Baas, F., Koornneef, J., Black, Shlosman,I.,Frank,J.,&Begelman,M.C.1989,Nature,338, 45
J.H.,&deGraauw,T.1990,A&A,227,342 Silge, J.D., Gebhardt, K., Bergmann, M., & Richstone, D. 2005,
CentaurusA-Probingthecircumnucleardisk AJ,130,406
Israel, F. P., van Dishoeck, E. F., Baas, F., de Graauw, T., & Sparke,L.S.1984,MNRAS,211,911
Phillips,T.G.1991,A&A,245,L13 Sparke,L.S.1986,MNRAS,219,657
Jarrett, T. H.,Chester, T., Cutri, R.,Schneider, S. E. & Huchra, Sparke,L.S.1996,ApJ,473,810
J.P.2003,AJ,125,525 Springel, V., Di Matteo, T., & Hernquist, L. 2005, MNRAS, 361,
Leeuw,L.L.,Hawarden,T.G.,Matthews,H.E.,Robson,E.I.,& 776
Eckart,A.2002, ApJ,565,L131 Steiman-Cameron,T.Y.,Kormendy,J.,Durisen,R.H.1992, AJ,
Mirabel,I.F.,etal.1999,A&A,341,667 104,1339
Malin,D.F.,Quinn,P.J.,&Graham,J.A.1983,ApJ,272,5 Tubbs,A.D.1980,ApJ,241,969
Marconi, A., Schreier, E. J., Koekemoer, A., Capetti, A., Axon, Veilleux, S., Cecil, G., & Bland-Hawthorn, J. 2005, ARA&A, 43,
D.,Macchetto, D.,&Caon,N.2000, ApJ,528,276 769
Marconi,A.,Capetti,A.,Axon,D.J.,Koekemoer,A.,Macchetto,
D.,&Schreier,E.J.2001,ApJ,549,915
Marconi, A., Pastorini, G., Pacini, F., Axon, D.J., Capetti, A.,
Macchetto,D.,Koekemoer,A.M.,&Schreier,E.J.2006,ApJ,in
press,astroph-0507435
8
Fig. 1.— a)Thecentralregionoftheimageat3.6µmtakenbytheIRACcameraonboardtheSpitzerSpaceTelescope. Theemission
is shown on a log scale. At 3.6 and 4.5µm starlight peaking near the galaxy center is seen in addition to emission from dust. At longer
wavelengths emissionfromdustisprimarilyseen.
TABLE 1
Modelparameters
Description Parameter Value
Galaxyprincipalaxis
PAonsky χ 20◦
Tilt ϑ 75◦
Dustintensity
Dustemissivityindex βe 4.0
Aspectratio k50 0.1
Aspectratioindex βk 0.9
Radiusofinnerhole rgap 50′′
Note. — The precession angle for this model, α(r), is a spline function that interpolates between 10 points in the format (radius in
arcseconds, α in degrees): (0.2, 180), (36.0, 210 ), (57, 270), (80, 345), (89, 375), (100, 400), (135, 380), (155,345), (240,280), (400,240).
Theinclinationangleω(r)isasplinefunctionthatinterpolates betweenthefollowingpoints(0.2,25),(100,20),(260, 8),(400,8).
9
Fig. 2.— The same region as Fig. 1 but a color composite of the IRAC images at 3.6, 5.8 and 8.0µm. This figure is shown on a log
scale. Theblackandwhiteversionofthisfigureonlyshowstheemissionatat8.0µm.
10
Fig. 3.— Color map showing the log of the H-band divided by the J-band 2MASS images by Jarrettetal. (2003). Lighter shadding
corresponds to regions of heavier extinction. Folds in the disk both nearer and more distant the observer are seen in the mid-infrared
images. Howeverfoldsnearertheobserverabsorbmorebackgroundstarlightandsoaremoreprominentinthenear-infraredimages.
Fig. 4.— OntheleftistheIRAC8.0µmimage. OntherightisthemodelforthewarpeddiskwithparametersgiveninTable1.