Disentangling the circumnuclear environs of Centaurus A: II. On the nature of the broad absorption line
D. Espada, A. B. Peck, S. Matsushita, K. Sakamoto, C. Henkel, D. Iono, F. P. Israel, S. Muller, G. Petitpas, Y. Pihlstroem, G. B. Taylor, D. V. Trung
aa r X i v : . [ a s t r o - ph . C O ] J u l Draft June 24 2010
Preprint typeset using L A TEX style emulateapj v. 03/07/07
DISENTANGLING THE CIRCUMNUCLEAR ENVIRONS OF CENTAURUS A:II. ON THE NATURE OF THE BROAD ABSORPTION LINE
D. Espada , A. B. Peck , S. Matsushita , K. Sakamoto , C. Henkel , D. Iono , F. P. Israel , S. Muller , G. Petitpas , Y. Pihlstr¨om , G. B. Taylor , and D. V. Trung Draft June 24 2010
ABSTRACTWe report on atomic gas (H I ) and molecular gas (as traced by CO(2–1)) redshifted absorptionfeatures toward the nuclear regions of the closest powerful radio galaxy, Centaurus A (NGC 5128).Our H I observations using the Very Long Baseline Array allow us to discern with unprecedentedsub-parsec resolution H I absorption profiles toward different positions along the 21 cm continuum jetemission in the inner 0 . ′′ ∼
55 km s − ). First, the broad H I line is more prominent towardthe central and brightest 21 cm continuum component than toward a region along the jet at a distance ∼
20 mas (or 0.4 pc) further from it. This suggests that the broad absorption line arises from gaslocated close to the nucleus, rather than from diffuse and more distant gas. Second, the differentvelocity components detected in the CO(2–1) absorption spectrum match well other molecular lines,such as those of HCO + (1–0), except the broad absorption line that is detected in HCO + (1–0) (andmost likely related to that of the H I ). Dissociation of molecular hydrogen due to the AGN seems to beefficient at distances r .
10 pc, which might contribute to the depth of the broad H I and molecularlines. Subject headings: galaxies: elliptical and lenticulars — galaxies: individual (NGC 5128) — galaxies:structure — galaxies: ISM INTRODUCTION
The analysis of absorption lines arising from atomicand molecular gas in galaxies is a powerful technique tostudy the physical and chemical properties of the inter-stellar medium (ISM). Only a handful of nearby galaxiesare known to show atomic (H I ) and molecular absorp-tion features. H I and molecular absorption line studiesare not only relevant to deduce the properties of the ISMalong specific lines of sight in nearby galaxies (e.g. Guptaet al. 2007), but are also essential for an interpretation ofsimilar data toward higher redshift galaxies (e.g. Mulleret al. 2006; Combes 2008; Henkel et al. 2009). Studying Academia Sinica, Institute of Astronomy and Astrophysics,P.O. Box 23-141, Taipei 10617, Taiwan. Harvard-Smithsonian Center for Astrophysics, 60 Garden St.,Cambridge, MA 02138, USA; [email protected]. Instituto de Astrof´ısica de Andaluc´ıa - CSIC, Apdo. 3004,18080 Granada, Spain. Joint ALMA Office, Av. El Golf 40, Piso 18, Las Condes, San-tiago, Chile. Harvard-Smithsonian Center for Astrophysics, SubmillimeterArray, 645 North A‘ohoku Place, Hilo, 96720, USA. National Radio Astronomy Observatory, P.O. Box O, Socorro,NM 87801, USA. Max-Planck-Institut f¨ur Radioastronomie, Auf dem H¨ugel 69,53121 Bonn, Germany. Nobeyama Radio Observatory, NAOJ, Minamimaki, Mi-namisaku, Nagano, 384-1305, Japan. Sterrewacht Leiden, Leiden University, Postbus 9513, 2300 RALeiden, the Netherlands. Onsala Space Observatory, SE-43992 Onsala, Sweden. Department of Physics and Astronomy, MSC07 4220, Univer-sity of New Mexico, Albuquerque, NM 87131, USA. Center for Quantum Electronics, Institute of Physics, Viet-nam Academy of Science and Technology, 10 DaoTan, ThuLe,BaDinh, Hanoi, Vietnam. both H I and molecular gas absorption lines at the sametime can reveal the interface between both componentsof the ISM at the circumnuclear regions where dissocia-tion of molecular gas is expected as a result of energeticprocesses triggered by Active Galactic Nuclei (AGNs).At a distance of D ≃ (Harris et al. 2009),Centaurus A (Cen A, NGC 5128) is the nearest giantelliptical exhibiting nuclear activity, with powerful radiojets extending almost 10 ◦ on the sky and with prominentatomic and molecular emission as well as absorption fea-tures detected toward its nuclear regions (Israel 1998). Itis thus a unique source to study the molecular to atomicinterface. At this distance, high spatial resolution can beachieved: 1 ′′ corresponds to 18 pc projected linear size.For a comprehensive general review of Cen A, see Israel(1998).One of the most prominent features in Cen A is thedust lane along its minor axis, which is composed of alarge amount of gas and dust within a warped disk-likestructure seen nearly edge-on. An H I mass of (10 ± × M ⊙ extends out to a radius of at least r =7 ′ (or about 7 kpc), while it is observed to be mostlyabsent within the inner arcmin (or 1 kpc projected linearsize) (van Gorkom et al. 1990; Schiminovich et al. 1994;Israel 1998). A comparable amount of molecular gas iscontained within the inner r ≃ ′ (2 kpc), M H ≃ × M ⊙ (Eckart et al. 1990b). About 30% – 40% of thetotal molecular gas content is located within the inner 1 ′ (1 kpc), M H = (8.1 ± × M ⊙ (Espada et al. 2009,hereafter Paper I). A circumnuclear disk of molecular gas This distance is used throughout this paper.
D. Espada et al.is resolved in the inner kiloparsec, with a size of 24 ′′ × ′′ (430 ×
215 pc) in diameter, and with different kinematicswith respect to the molecular gas at larger radii (PaperI).Along the NE – SW direction (perpendicular to thecircumnuclear molecular gas disk), there exists a spectac-ular radio-continuum jet extending from the nucleus (pcscale) to the outer radio lobes (hundreds of kpc scale).Subparsec-scale VLBI (Very Large Baseline Interferom-etry, including the VSOP - VLBI space observatory pro-gramme) high resolution radio-continuum observations(e.g. at 2.3, 4.8, 8.4, 22.2 and 43 GHz) show that thestructure of Cen A is rather complex, consisting of abright nuclear jet and a fainter counterjet, as well as abright central component (e.g. Tingay et al. 1998; Tin-gay & Murphy 2001; Horiuchi et al. 2006). Note that thiscomponent is not the nucleus itself since the nucleus isnot bright at frequencies below 10 – 20 GHz due to syn-chrotron self-absorption and free-free absorption from adisk or torus of ionized gas (Jones et al. 1996; Tingay& Murphy 2001). The compact core becomes visible atmm/submm wavelengths (Kellermann et al. 1997) whereits spectral index is quite flat (see Fig. 6 in Meisenheimeret al. 2007).H I absorption lines in this source were found for thefirst time four decades ago (e.g. Roberts 1970). Sincethen, VLA aperture synthesis observations with resolu-tions of the order of ∼ ′′ (van der Hulst et al. 1983;Sarma et al. 2002) have shown H I absorption lines indifferent regions along the jet. They are distributed to-ward both the core and the inner jet (extended along 30 ′′ )and are composed of a set of lines with widths of typi-cally 10 km s − . The most prominent H I absorptioncomponent is located close to the systemic velocity ofthe galaxy ( V LSR = 541.6 ± − , Israel 1998) . To-ward the compact bright continuum component, two red-shifted lines are seen at V = 578 km s − and 598 km s − ,as well as a broad absorption component (extending upto V = 620 km s − with a width of ∆ V ≃
55 km s − ).Sarma et al. (2002) showed that the latter is only presenttoward the central component, but not along the otherregions of the jet. This sets an upper limit to the size ofthe absorbing material of about 20 ′′ , or 360 pc as seen inprojection.Molecular absorption lines from cm to submm wave-lengths have also been detected for several species. Atcm wavelengths narrow molecular lines (a few km s − width) close to the systemic velocity have been observedin several molecules such as OH, H CO, C H and NH (Gardner & Whiteoak 1976a,b; Seaquist & Bell 1990; vanLangevelde et al. 1995, 2005). Single-dish observationshave been performed at mm/submm wavelengths for thelowest transitions of CO, HCN, HNC, HCO + , H CO + ,N H + , C H, CN and CS (Eckart et al. 1990a; Israel et al.1990; Wiklind & Combes 1997; Eckart et al. 1999). Inaddition to narrow lines close to the systemic velocity,both narrow and broad absorption features are seen atredshifted velocities in HCO + (1–0), OH 18cm, and toa minor extent in HCN(1–0), HNC(1–0) and CS(2–1),covering a velocity range from ∼
570 to 620 km s − (e.g.Seaquist & Bell 1990; Eckart et al. 1990b; Israel et al. All velocities in this paper are radio LSR. Conversion fromLSR to heliocentric velocity is V Hel = V LSR + 2.4 km s − . I observations using the VeryLarge Baseline Array (VLBA) and the phased Very LargeArray (VLA) which allow us to derive the properties ofthe main H I absorption features seen toward the con-tinuum source at 21 cm with unprecedented sub-parsecresolution (versus the ∼
180 pc resolution in past VLAexperiments). This is complemented by CO(2–1) obser-vations (and isotopologues) at the Submillimeter Array(SMA , Ho et al. 2004) (Paper I) from which we canextract a CO(2–1) absorption profile with minimal con-tamination by emission. We introduce our VLBA H I and SMA CO(2–1) observations and the data reductionin §
2, where we present the continuum maps and ab-sorption spectra. In § I absorption components and the physi-cal properties of their corresponding regions. We presentour results on the CO(2–1) and CO(2–1) absorptionlines in § I absorption lines withthose of molecular lines in §
5. We discuss the results in § § OBSERVATIONS AND DATA REDUCTION
The array configuration, phase center, primary andsynthesized beam, achieved noise and spectral resolutionof our VLBA H I and SMA CO(2–1) observations areshown in Table 1. In the following we explain in moredetail our observational strategies and data reduction. VLBA H I Observations
The VLBA H I observations were carried out on 26November 1994. Due to its low declination (Dec. ≃− ◦ ), Cen A was observable for 2 hours with 8 of the 10VLBA antennas and the phased VLA. The observationswere made using 2 IFs of 2 MHz each. The IFs werecentered on the redshifted H I line ( ν rest = 1420.405752MHz, at a velocity V LSR = 541.6 km s − ). Global fringe-fitting was performed using AIPS task FRING with asolution interval of 6 minutes. Amplitude calibration wasderived using the antenna gain and system temperaturerecorded at each antenna. Bandpass calibration was done The VLBA and VLA are operated by the National Radio As-tronomy Observatory, a facility of the National Science Foundationoperated under cooperative agreement by Associated Universities,Inc. The Submillimeter Array is a joint project between the Smith-sonian Astrophysical Observatory and the Academia Sinica Insti-tute of Astronomy and Astrophysics, and is funded by the Smith-sonian Institution and the Academia Sinica. isentangling the circumnuclear environs of Centaurus A. II 3using the calibrator 3C273 (J1229+0203). The contin-uum was obtained from line-free channels. Continuumsubtraction was done in Difmap (Shepherd et al. 1995)using a model of
CLEAN components made from severalline-free channels. Subsequent editing and imaging of alldata was also done using
Difmap , using natural weight-ing. The resulting Half Power Beam Width (HPBW) is36.3 × × − , and the total velocitycoverage of a single IF is ∼
430 km s − . SMA CO(2–1) Observations
Details of the CO(2–1) observations using the SMA inits compact configuration are reported in Paper I. CO(2–1), CO(2–1) and C O(2–1) lines were simultaneouslyobserved. We used two independent sources, 3C273 andCallisto, to check the consistency of our bandpass cali-bration. Both solutions were nearly identical in most ofthe spectral windows, and in particular where the CO(2–1) lines were located. The continuum toward Cen A at1.3 mm was found to consist of an unresolved source witha flux of S . = 5.9 ± − at the AGN po-sition which was used to calibrate the data of Cen A it-self (Paper I). No continuum subtraction was performed.The velocity resolution is 1 km s − .In order to minimize undesired emission, we obtain theCO(2–1) absorption spectrum by directly fitting the am-plitude of the interferometric visibilities using a pointsource model fixed at the position of the unresolved con-tinuum source. This was done with the task UV_FIT inthe
GILDAS reduction package. The CO(2–1) spectrumextracted using visibilities from all baselines still showssome emission at velocities V = 420 – 510 km s − and V = 620 – 680 km s − . This corresponds to extended emis-sion from the circumnuclear regions (Paper I). A new fitto the visibility data with projected baseline lengths ≥
50 m yielded a flatter spectral baseline. This is due toa combination of the spatial filtering capabilities of theinterferometer, beam smearing and probably a lack ofCO(2–1) emission toward the central ∼ ′′ ( ∼
90 pc).The typical extent that we take into account by remov-ing the shorter baselines can be roughly estimated as ≤ λ/D ≃ ′′ ( ∼
72 pc), where λ is the wavelength of theobservations and D the average baseline length.We show in Figure 1 the resulting CO(2–1) spectrawith and without visibilities with baseline lengths largerthan 50 m. For the CO(2–1) and C O(2–1) transi-tions, however, we retained all visibilities since there isno significant emission down to our noise level. Theirresulting noise per channel, σ = 0 . σ =0.3 Jy). H I ABSORPTION FEATURES
21 cm Continuum: Nuclear Jet
The 21 cm continuum map in the inner 0 . ′′ ∼ ′′ , or 180 pc) found by van der Hulst et al. (1983)and Sarma et al. (2002) (labeled as “Nucleus”). We findthat this central component further comprises a more compact bright component and a nuclear jet extendingup to 0 . ′′
23 to the NE, with a similar P.A. ≃ ◦ as theinner 21 cm jet extending up to 55 ′′ (van der Hulst et al.1983; Sarma et al. 2002). Note that van Langevelde et al.(2005) present a VLBI 18 cm continuum map with a sub-arcsecond resolution where the central bright componentis seen. However, the rest of the nuclear jet is not asclearly detected as in our map.While the peak flux density is 603 mJy beam − andthe noise level of our maps is 2 mJy beam − , note thatthe uncertainty of the absolute flux density measurementis larger. Uncertainties in flux calibration are estimatedto be about 5% ( S = 600 ±
30 mJy beam − ). Inaddition, on-source errors due to deconvolution are ex-pected. For instance, Tingay & Murphy (2001) estimatedthat the corresponding flux density error in their 2.2 GHzVLBA data of Cen A due to this effect is about 30% forthe brightest components. This source of error is ex-pected to be larger for our lower observing frequency,since the uv coverage is less uniform.As mentioned in §
1, the core of the radio source hasa very strongly inverted spectrum since the continuumemission is seen through a circumnuclear ionized gas diskwhich is opaque at wavelengths longer than 13 cm (or fre-quencies lower than 2.3 GHz) due to free-free absorptionand to self-absorbed synchrotron emission (Jones et al.1996; Tingay & Murphy 2001). The compact bright com-ponent in Figure 2 is thus part of the approaching jetand is offset from the actual AGN. Tingay & Murphy(2001) showed that there is an absolute offset of 12.5 masto the E and 10.5 mas to the N between the 8.4 GHzand 2.2 GHz central components. This corresponds toa separation of 0.3 pc along the jet axis. We show inFig. 2 the estimated location of the AGN (8.4 GHz cen-tral component), assuming that the bright component ofour 1.4 GHz map coincides with that of the 2.2 GHz map,and also that position variability between the differentobserving dates is not significant. We do not detect anycontinuum emission from the counter-jet toward the SWto the 2 mJy beam − noise level of our VLBA maps. Identification of H I Absorption Features
Figure 3 shows the H I spectrum integrated over theentire continuum source. We find 5 main H I absorp-tion lines: four relatively narrow lines and an underlyingbroad line. These absorption lines, all red-shifted withrespect to the systemic velocity, were detected using theVLA by van der Hulst et al. (1983) and Sarma et al.(2002), although toward an unresolved component of size ∼ ′′ – 10 ′′ . We adopt the nomenclature used in Sarmaet al. (2002) for the most prominent lines (in roman nu-merals, IV, SV, I, II and III).We derive from Gaussian fits the mean velocities ( V ),peak flux densities ( S ) and Full Widths at Half Maxi-mum (FWHM) of the different lines in our H I profile,as shown in Table 2. We first fit the broad line usingthe redshifted wing at V >
600 km s − , and then fit theremaining lines individually. An exception is line I/II,which is a blend of lines I and II in the H I spectra ofSarma et al. (2002). Being almost unresolved with our3.3 km s − resolution, we decided to compute them to-gether. The H I absorption features in our spectra arelocated at V = 541, 552, 575 and 595 ( ±
1) km s − (IV,SV, I/II, III, as outlined in Figure 3). The most promi- D. Espada et al.nent lines show FHWM values of about 5 – 10 km s − .A fit to the broad absorption line yields a FWHM = 53 ±
10 km s − centered at 578 ± − . The final fitand the resulting residual from the observed H I profileare presented in Figure 3.The great advantage of our VLBA data is that wecan obtain the H I absorption spectrum toward differ-ent positions over the continuum emission. Figure 2shows the four most prominent continuum sources wherewe obtained H I absorption profiles. Their relative off-sets with respect to the phase center are: (∆ α ,∆ δ ) =(+73,+69) mas, (+63,+58) mas and (+20,+10) mas(namely, positions a , b and c ), corresponding to sep-arations of 1.8 pc, 1.5 pc and 0.4 pc respectively; aswell as toward the brightest component at (+1, −
1) mas(namely, position d ). The continuum flux density peaksof the different components are S = 11, 23, 145 and603 mJy beam − , respectively.While we detect H I absorption features only close tothe systemic velocity (components IV and SV) in posi-tions a and b , in position c and d the profiles show twoadditional redshifted H I absorption components I/II andIII at V = 574 – 576 km s − and 595 – 596 km s − (Fig-ure 4). An additional shallow redshifted wing in position d is clearly seen at velocities V >
600 km s − , which cor-responds to the broad absorption line. The fitted Gaus-sian parameters are presented in Table 3, and the fitsare shown in Figure 4. To fit these lines, we set as initialvalues the velocities and FWHM found for the H I ab-sorption lines integrated over the entire continuum. Sincethe flux densities refer to their corresponding continuumemission, we calculated next the optical depths and H I column densities in order to compare between differentpositions. H I Optical Depths and Column Densities
We calculated the H I optical depths, τ HI , as I obs [ Jy ] = I cont e − τ HI , where I obs is the observed flux in absorptionand I cont is the measured continuum flux. Absolute fluxuncertainties do not have any effect on the optical depthsince it is derived from the I obs / I cont ratio. In columns 5and 6 of Table 2 and 3 we present the H I optical depthpeak (peak τ HI ) and integrated optical depths ( R τ HI dV )integrated over the entire continuum source and for eachindependent position, respectively. The upper limits cor-responding to non-detected lines are calculated as 3 σ (∆ V δv ) / , where ∆ V is the FWHM in Table 2 and δv is the channel spacing of 3.3 km s − .In Figure 5 we plot τ HI toward the four different posi-tions a , b , c and d along the 21 cm nuclear jet. The mainabsorption line is component SV, with peak τ HI ≥ d ).Component I/II is characterized by peak τ HI = 0.29 ± c , while peak τ HI = 0.41 ± d . The 3 σ upper limit obtained in position b is peak τ HI < b if ofcomparable depth as in position c or d .It is surprising that we detect the broad line compo-nent against the position d but it is weaker toward theadjacent continuum region, position c . Figure 6 shows a close-up of the τ HI in both locations. A fit to thebroad component (assuming that a Gaussian functionapplies) gives integrated optical depths of R τ HI d V = 3.6 ± − for position d while it is R τ HI d V = 0.5 ± − in position c (Table 3). That the broad ab-sorption line in position c is weaker is reinforced by thesmaller optical depths in the spectral regions between thelines SV, I/II and III at V = 560 km s − and 590 km s − with respect to position d (Figure 6). This implies thatthe H I opacity of the broad line component varies withina projected linear scale of less than 0.4 pc.Finally, we calculate the H I column densities (Column7 in Tables 2 and 3), using the following relation: N (HI) / T s [cm − K − ] = 1 . × Z τ HI dV [km s − ]where T s is the spin temperature. The H I column densi-ties in the different lines cover the range N (H I ) /T s ∼ (1– 12) × cm − K − . The broad line toward position d has a corresponding N (H I )/ T s that is comparable tothe deepest narrow lines. CO(2–1) ABSORPTION FEATURES
In our 1.3 mm observations, with a restoring beam of6 . ′′ × . ′′ ×
40 pc), we found an unresolved con-tinuum source at the galactic center (Paper I). The innerjet observed at cm wavelengths has a steep radio spec-trum so the contribution at mm wavelengths is negligi-ble (Israel et al. 2008). Therefore our 1.3 mm continuumemission is dominated by the nuclear continuum emis-sion. The angular size of the central continuum sourceis 0.5 ± ± CO(2–1) and C O(2–1) (the latter two taking into ac-count all visibilities) are shown in Figure 7 b ), c ) and d ),respectively. Absorption features have been detected inCO(2–1) and CO(2–1), but not in C O(2–1) at a 3 σ level (= 0.3 Jy). CO(2–1) Line Characterization
The CO(2–1) absorption spectrum spans a range of540 < V <
620 km s − and is composed of 6 narrowlines. The narrow absorption features are labelled asoutlined in Figure 7 a ): lines , and correspond tothe absorption lines close to the systemic velocity, at V = 539, 544 and 550 km s − , and , and to absorptionlines at higher velocities, located at V = 575, 606 and613 km s − (Table 4). Component is clearly detectedin the profile with all visibilities, but it turns out to beweak with only the longest baselines due to the increaseof noise level. CO(2–1) absorption line is located in themost likely systemic velocity of the galaxy. Absorptionlines , and are much deeper than the higher velocitycomponents, with component the most prominent.Three CO(2–1) absorption lines are also detected,corresponding to features , , in the CO(2–1) profile.An additional line corresponding to CO(2–1) line wastentatively detected. Features and are not detectedat a 3 σ level.The linewidths are comparable in most cases to ourspectral resolution. Because of the limited velocity reso-lution of our SMA observations (1 km s − ) and the nar-rowness of the absorption lines, we do not attempt to fitisentangling the circumnuclear environs of Centaurus A. II 5Gaussian functions to the several CO(2–1) line compo-nents. Line widths in the CO(2–1) profile of components and are about ∆ V = 3 – 5 km s − , while the re-maining lines have even smaller line widths ∆ V ≃ − . Lines with widths less than 1 – 2 km s − arenot (or insufficiently) resolved. Note that in those casesthe optical depths are underestimated. CO(2–1) Optical Depths and Column Densities
Since the radio core is small ( ∼ , and ) have depths > f ≥ depthof absorption for a normalized spectrum (continuum fluxis unity), the covering factor is f > f of unity for all the components.We have calculated the CO(2–1) optical depths τ CO(2 − , in the same manner, I obs [ Jy ] = I cont e − τ CO(2 − (Table 4), as those of H I . The optical depths for anyundetected CO(2–1) and C O(2–1) lines have beencalculated using a 3 σ level upper limit. CO(2–1) com-ponent is saturated since the depth of the absorptionline nearly equals the continuum level, and τ CO(2 − ishighly uncertain due to non-linearity of the logarithmicfunction. High values of peak τ CO(2 − show that com-ponents , and are optically thick. CO(2–1) compo-nents , and , as well as all of the CO(2–1) lines, arecharacterized by peak τ CO(2 − . τ CO(2 − for components , and are τ CO(2 − = 2.1 ± ± ≥ ± τ CO(2 − = 1.8 +0 . − . , 1.9 +0 . − . and ≥
3. Israel et al.(1991) also gives an optical depth for component of τ CO(2 − = 0.4 ± ± CO(2–1) lines, the peak τ CO(2 − valuesare: 0.11 ± ± ± ± , , and . Eckart et al. (1990a)gives a peak optical depth of 0.4 for the line close to V =550 km s − and non-detections for the rest of the lines.This value is well in agreement with our component .We derive column densities of each CO(2–1), CO(2–1) and C O(2–1) lines using the following equation: N (CO) = 8 πν c A ul g u Q ( kT ex ) exp ( E J /kT ex )(1 − exp ( − hν/kT ex )) Z τ CO(2 − dV, where we assume that LTE conditions apply. E J de-notes the energy of the lower level of the transition,T ex the excitation temperature, Q ( T ex ) = P J (2 J +1) exp ( − E J /kT ex ) the partition function, and A ul , theEinstein coefficient of each isotopologue. We use T ex =10 K, although it could vary depending on the propertiesand location of the molecular clouds. Indeed, a gradientin excitation temperature has been inferred using a nom-inal temperature ratio of CO(2–1) and CO(1–0) emissionline ratios by Israel et al. (1990, 1991): T ex ≃
10 K at r > ′ ( > r < ′′ ), T ex ≃
25 K. The different detected lines show CO column densi-ties N (CO) spanning two orders of magnitude (Table 4),from (23 ± × cm − (component ) to ≥ × cm − (component ). N (CO) increases approx-imately an order of magnitude if a T ex = 100 K is used.The CO column density is in the range < × cm − (component ) and (145 ± × cm − (com-ponent ).We list in Table 4 the H column densities ( N (H ))assuming abundance ratios [CO]/[H ] = 8.5 × − (Fr-erking et al. 1982), [ CO]/[H ] = 1 × − (Solomonet al. 1979) and [C O]/[H ] = 1.7 × − (Frerking et al.1982). The H column densities as obtained from CO(2–1) should be similar to those obtained with CO(2–1).However, N (H ) obtained with CO(2–1) are systemati-cally smaller than those obtained using CO(2–1). The[CO]/[ CO] ratios for the six lines in order of increas-ing velocity are: 17 ±
2, 5 ± ≥ >
3, 17 ± >
5. These abundance ratios are considerably smallerthan the assumed [CO]/[ CO] = 85, similar to the lo-cal value. This deviation in the abundance ratios mightbe due to a combination of the following reasons: i ) dif-ferent excitation temperatures for CO and CO, ii )optical depth effects, iii ) different filling factors and iv )chemistry, such as selective photodissociation (of CO).However, low abundance ratios of ∼
25 have been ob-served towards the Galactic center and PKS 1830-211(see Table 7, Muller et al. 2006) [ C]/[ C] and [ O]/[ O] Isotopic Ratios
We can roughly estimate the [ C]/[ C] and[ O]/[ O] isotopic ratios from the [CO]/[ CO] and[ CO]/[C O] ratios, assuming that the molecular abun-dance ratios can be decomposed into isotopic abundanceratios of the atoms. If we assume that [ CO]/[ CO]= [ C]/[ C], then we obtain that [ C]/[ C] ≃ C]/[ C] ratio is a measure of the primaryto secondary processing: while C mostly arises frommassive stars, C is predominantly produced from re-processing of C from earlier stellar generations (e.g.Wilson & Rood 1994).From our observations we derive lower limits fromthe abundance ratio [ CO]/[C O]. It ranges from > >
12. This is somewhat less restrictive than the[ CO]/[C O] >
30 obtained by Wild et al. (1997) us-ing the J=1–0 absorption line, although their spectralresolution is a factor 2 lower ( ∼ − ). Higherspectral resolution observations, with a channel width of0.23 km s − , had been previously carried out by Israelet al. (1991). Israel et al. report a 2 σ detection in theC O line for component , which gives a value for theabundance ratio [ CO]/[C O] ≃ ±
3. This wouldbe in agreement with our lower limit, but not with thatin Wild et al. (1997).We can then obtain [ O]/[ O] from our previous es-timate of [ C]/[ C]. O and O form in massive starsand are predicted to behave similarly, except at earlystages of massive star formation since O is a primary el-ement and O is a secondary one (Prantzos et al. 1996).The ratio is therefore expected to decrease with time andstellar processing (e.g. Muller et al. 2006).By adopting [ C]/[ C] = 5 – 17 and [ CO]/[C O] ≃ ± O]/[ O] ∼ −
220 . This ratio would be in agree- D. Espada et al.ment with values ∼ −
200 in nearby starburst galax-ies such as NGC 253 and NGC 4945 (Wang et al. 2004;Harrison et al. 1999; Henkel et al. 1993).Overall, the [ C]/[ C] and [ O]/[ O] ratios in Cen Aare within those found toward the Galactic center andstarburst galaxies (Henkel et al. 1993). Thus, ourcalculated [ C]/[ C] and [ O]/[ O] ratios using COlines suggest within the uncertainties that the molecu-lar clouds producing the absorption lines close to thesystemic velocity might have been affected by relativelylong-term starburst conditions.On the other hand, note that Wiklind & Combes(1997) derived an isotopic ratio [ C]/[ C] >
70 for com-ponent from its [HCO + ]/[H CO + ] ratio, which yields[ O]/[ O] >
200 – 840. This seems to contradict our re-sults and shows the need of a systematic chemical studyusing different isotopic molecular lines. COMPARISON OF ATOMIC AND MOLECULARABSORPTION LINES
In this section we compare the variations of τ HCO + (1 − / τ CO(2 − and τ CO(2 − / τ HI in order to inferthe variation of molecular to atomic abundances withvelocity. As a zeroth-order approximation we considerthree assumptions: i ) absorption lines of different species(both molecular and atomic) that have similar veloci-ties are physically related; ii ) temperature (and spin)differences vary similarly for different species so thatthe ratios are reasonably constant; and iii ) molecularabundance ratios are reasonably constant ([CO]/[H ] and[HCO + ]/[H ]). Molecular gas as traced by CO(2–1) andHCO + (1–0) The different velocity components detected in theCO(2–1) absorption spectrum correspond well to othermolecular lines, such as those of HCO + (1–0). We choseHCO + (1–0) for comparison because it traces moleculargas with different physical conditions (its critical den-sity is two orders of magnitude larger than that tracedby CO(2–1)). It is also the best S/N and the high-est spectral resolution ( ≃ − ) absorption profilethat can be found in the literature (Wiklind & Combes1997). The CO(2–1) absorption profile can be comparedto that of HCO + (1–0) (Figure 7 e ), which results fromthe sum of 17 Gaussian components used by Wiklind &Combes (1997) to parametrize the observed HCO + (1–0)absorption spectrum. The more prominent lines of theHCO + (1–0) spectrum are centered at the same veloci-ties and have similar widths to their CO(2–1) counter-parts. Note that no variability was found in differentepochs for the HCO + (1–0) absorption spectrum (Wik-lind & Combes 1997), so we do not expect to find vari-ability in our CO(2–1) profiles either.On the other hand, the high velocity component ofthe HCO + (1–0) absorption profile (Wiklind & Combes1997; Eckart et al. 1999) (Figure 7 e ) is more prominentthan in our CO(2–1) spectrum (Figure 7 b ). The broadcomponent of HCO + (1–0) absorbs roughly 15% of thecontinuum emission (Wiklind & Combes 1997). If theCO(2–1) broad absorption feature were as deep as theHCO + (1–0) broad line then we would have detected it.We derive the abundance ratio assuming T ex = 10 Kfor CO(2–1) and T ex = 5 K for HCO + (1–0) (Wiklind & Combes 1997). For lines within 540 km s − V <
570 km s − , the total abundances of HCO + and COare N (HCO + ) ≃ × cm − (Table 8, Wiklind& Combes 1997) and N (CO) ≃ × cm − , whichprovides the ratio [HCO + ]/[CO] ≃ × − . In thehigh velocity component, V >
570 km s − , the ratio isfive times larger than in the low velocity components,[HCO + ]/[CO] ≃ × − .The [HCO + ]/[CO] ratio is even larger within570 km s − < V <
600 km s − . While [HCO + ]/[CO] ≃ × − for components at V >
600 km s − , in thevelocity range between 570 km s − < V <
600 km s − it is a factor 10 larger, [HCO + ]/[CO] ≃ × − . Thisdifference arises in the velocity range corresponding tothe broad absorption component. Atomic versus molecular gas content
In Figure 8 we present a comparison between the H I and CO(2–1) absorption features. It is important to notethat we are tracing both the H I component d and molec-ular gas in the line of sight within a region located closeby( . I absorption featuresare located between V = 544 and 600 km s − , in COand other molecules the main absorption features occurclose to the systemic velocity, and then at larger veloc-ities, V >
605 km s − . The H I absorption features areremarkably more prominent at 560 < V <
610 km s − than their molecular counterparts. The two components and of the CO(2–1) and HCO + (1–0) spectra at V>
605 km s − are located in the same velocity range asthe H I broad absorption feature which is seen up to V = 640 km s − .Next we compare the molecular to atomic abundanceratio assuming T ex = 10 K for CO(2–1) and T s = 100 Kfor H I . First, in the low velocity component the gas ismostly in its molecular phase. The CO(2–1) line seemsto be related to the H I component IV, and it yields aratio N (H )/ N (H I ) = (7.25 × cm − ) / (0.9 × cm − ) ≃
8. The N (H )/ N (H I ) ratio corresponding toCO(2–1) line is similar to that of the CO(2–1) line . The H I absorption line SV and CO(2–1) line , theratio of H to H I column densities is N (H )/ N (H I ) ≥ (1800 × cm − ) / (12 × cm − ) = 1.5.In the high velocity component located between570 km s − < V <
600 km s − , the gas is predominantlyeither atomic or molecular gas traced by HCO + (1–0),but not by CO(2–1). H I lines I/II and III containa total of N (H I ) ≃ (13 . ± × cm − . If spa-tially linked to molecular clouds with the same kine-matics, they would only correspond to CO(2–1) line ,which contains a smaller amount of molecular gas, N (H ) ≃ (26 ± × cm − . The ratio is N (H )/ N (H I ) ≃ + (1–0) in this ve-locity range. Wiklind & Combes (1997) estimated it tobe N (HCO + )/ N (H I ) ≃ × − , or N (H )/ N (H I ) ≃ + ]/[H ] = 10 − –10 − . Even with the contribution of HCO + in the ve-locity range between 570 km s − < V <
600 km s − , weobtain a slightly smaller [H ]/[H I ] ratio than in the lowisentangling the circumnuclear environs of Centaurus A. II 7velocity component.Finally, from V >
600 km s − , CO(2–1) components and are within the H I broad line. There N (H I ) =7 × cm − (broad line), to be compared to the N (H ) ≃ . × cm − , which yields a ratio N (H )/ N (H I ) ≃ DISCUSSION
Warped Disk or Non-Circular / Infalling Motions?
Although there is a consensus that narrow absorptionlines with velocities close to the systemic velocity (LowVelocity Component -LVC-, in the velocity range 540km s − < V <
560 km s − , following the nomenclatureby Wiklind & Combes 1997) represent gas far from thenucleus ( ∼ − < V <
640 km s − ).A schematic view of proposed models to describe thealmost edge-on disk of CenA is presented in Fig. 9. Thisfigure includes edge-on and face-on views of the proposedmodels within the inner 2 kpc, as well as a qualitativeinterpretation of the kinematic location of the absorp-tion lines. A warped and thin disk model (e.g. Quillenet al. 2006; Neumayer et al. 2007; Quillen et al. 2009)reproduces well the observed emission lines in terms ofdistribution and kinematics. However, this model is notable to reproduce by itself the existence of the redshiftedcomponents (HVC) because it assumes pure circular or-bits (Eckart et al. 1999) (see Figure 9 a ). It is necessaryto augment this model with either diffuse gas (Eckartet al. 1999) (see Figure 9 b ) or non-circular/infalling mo-tions (e.g., as suggested in van der Hulst et al. 1983 andPaper I; see Figure 9 c ). Since the implications of thelatter two models diverge, we discuss next what is themost plausible one.It is remarkable that we detect the broad line of theHVC against the brightest continuum source (position d )but it is found to be much weaker toward position c just ∼ c ) can provide the observed differences betweenpositions c and d , since one would expect higher veloc-ities closer to the nucleus. On the other hand, this isnot expected in the warped disk with diffuse gas model(Figure 9 b ) since this gas should be distributed at highlatitudes up to about a few hundreds of parsecs.In addition, the physical locations of the different ab-sorbing regions in the line of sight derived from thewarped disk with diffuse gas model (Figure 9 b ) are com-plex. The model predicts that LVC lines and (Fig-ure 7) arise at r = 1.7 – 1.9 kpc and are separated by upto ∆ z ≃
160 pc from the disk, line should be located r = 200 pc and ∆ z ≃
0, and the HVC lines , and at r = 200 – 600 pc and ∆ z ≃
300 pc (see Figure 9 b ).In this model, within r <
600 pc, the nearer the corre-sponding absorbing region is to the nucleus the closer thecorresponding line velocity is to the systemic velocity.Finally, no component within the inner r = 200 pc istaken into account in Eckart et al.’s model, although alarge concentration of molecular gas is found there (Pa-per I). A contribution to the molecular gas profile of thiscomponent should be seen as an extension of the narrow line (arising from r = 200 pc at ∆ z = 0 pc in Eckartet al.’s model). The limit imposed by Eckart et al. (1999)was set at r = 200 pc so that the line reproduces theobserved one. However, without that constraint, line is difficult to be reproduced in that model.These arguments suggest that the nature of the HVCsis not just diffuse gas within a warped disk far from thenucleus, and that non-circular motions and/or infallinggas must be important mechanisms to explain the natureof this component. Within this scenario, molecular lines4, 5 and 6 and atomic lines I/II and III would arise atradii r . r & c ). An X-ray Dominated Region (XDR)?
Since a powerful AGN resides in this galaxy, and asignificant amount of gas is located within the inner r = 220 pc (Paper I), it is reasonable to think that X-rayradiation plays a major role modifying the chemical com-plexity (i.e., dissociating molecular hydrogen into atomicgas, ionizing and heating, etc.).In fact, the AGN of Centaurus A is characterized bya X-ray luminosity of L X ∼ × erg s − . Theincident flux at a distance of 1 pc is thus quite large, F X ∼
300 erg s − cm − . On the other hand, it is not likelythat it has such a strong effect at distances ∼
100 pc,with fluxes of the order of F X ∼ − cm − . Thegas column density attenuating these X-rays is estimatedto be ∼ cm − , probably by a cloud that entirelysurrounds the nucleus or by a torus-like structure (Evanset al. 2004, and references therein).The LVC is likely arising from cold molecular gas, withkinetic temperatures T ∼
10 K and average hydrogendensities of the order of n ∼ cm − , which seemsto be appropriate for the disk as seen in emission (e.g.Eckart et al. 1990b) as well as in absorption (e.g. Wik-lind & Combes 1997; Muller & Dinh-V-Trung 2009). Lessinformation is present in the literature for the physicalconditions of the HVC, since this weak component is dif-ficult to detect.Observationally, Wiklind & Combes (1997) show thatthe average of [HCO + ]/[HCN] ∼ + ]/[HCN] ∼ ∼ F X =160 erg s − cm − and n H = 10 . cm − (or both F X and n H two ordersof magnitude lower). This numerical ratio agrees wellwith the observations. However, under certain circum-stances, a PDR could also reproduce the given moleculargas ratios.X-rays do not lead to strong dissociation of CO andit can be present in an XDR at elevated temperatures(Spaans 2008). Warm CO gas produces emission origi-nating from high rotational transitions. Bright CO emis-sion lines have been seen for the first three J levels, andlarger T ex (20 – 30 K) have been inferred for molecularclouds close to the nucleus (e.g. Israel et al. 1991). Onthe other hand, absorption lines of the lower J transitionswill preferentially sample the (excitationally) coldest gas.In the nuclear regions, where higher densities and kinetictemperatures are expected, only high-J transitions of COwould have such broad lines as seen in HCO + (1–0), asit has been suggested with previous CO(3–2) single-dish D. Espada et al.data (Wiklind & Combes 1997). This is likely the reasonwhy we did not detect the broad line in CO(2–1), butmay well be detected in higher transitions of CO. Thiscould be a good test to distinguish the physical conditionof the molecular gas closer to the nucleus.The molecular lines at V >
605 km s − , including ab-sorption features and , are probably molecular clumpsclose to or within the H I region from which the H I broadabsorption component arises. We consider the possibil-ity that these features are located close to the nucleus(as suggested in Figure 9 c ). For an X-ray illuminatedgas like these clouds, the physical conditions of the gasthat surrounds the AGN can be studied using the effec-tive ionization parameter, ξ eff , which is proportional tothe ratio of incident (and attenuated) X-ray photon fluxto gas density of the cloud (Maloney et al. 1996). Theeffective ionization parameter of a molecular cloud canbe calculated as: ξ eff ∼ . × − L / ( N . n r )where the L = L X /10 is the X-ray luminosity of hardphotons (energies > − , n the gas den-sity of the cloud in 10 cm − , r the distance to the X-rayemitting source in pc, and N the attenuating gas col-umn density in 10 cm − (Maloney et al. 1994, 1996).The molecular cloud regions are exposed to an X-ray lu-minosity from Cen A’s AGN of L X = 4.8 × ergs s − in the hard energy range 2 –10 keV (Evans et al. 2004).If we assume for the molecular gas arising in the twohigh velocity absorption features and a T ex = 10 K,a size of the 1.3 mm continuum of 0.01 pc (Kellermannet al. 1997), and a size of the cloud of R ≃
10 pc thenthe molecular clouds would be characterized by densitiesof about n ∼ cm − .The gas column densities that are attenuating the X-rays are estimated to be ∼ cm − (Evans et al. 2004).By assuming a distance to the nucleus of r = 10 pc, thenit yields a parameter ξ eff = 7 × − . For r = 1 pc then ξ eff = 0.1. A ξ eff > − means that the gas will be pre-dominantly in its atomic phase and it indicates that themolecular clouds must be at r >
10 pc in order to sur-vive under the above assumptions. This is in agreementwith the smaller [H ]/[H I ] ratio found in the HVC withrespect to the LVC ( § Physical properties of the circumnuclear H I As mentioned in § I features (i.e. I, I/II and III) arebetween 5 – 10 km s − . If one neglects line broadeningdue to rotation and turbulence, and attributes the ob-served FWHM only to thermal motions in the gas, theline widths shown in Table 2 would indicate tempera-tures between 800 and 5,700 K. Turbulence can accountfor a linewidth of about ∼ − , at least within ourown Galaxy (Burton 1988). Taking this into account,the actual gas temperature is probably between 20 and2500 K.The nature of the gas from which the broad absorp-tion lines arise are much less understood. The underlyingbroader H I absorption line, with a width of ∼
55 km s − ,cannot be attributed solely to thermal motions since itwould give too high temperatures ( ∼ I component is probably a blend arising from several com-plexes rather than a single one. Since it only showsredshifted velocities, it is likely that this component islargely affected by systematic kinematic contributions,such as non-circular and/or infalling motions, and closeto the nucleus, mostly within a distance . c to d , and 0.3 pc from position d to theactual nucleus; as mentioned in § i = 70 ◦ ), thenthe disk-like feature associated with the broad H I linewith column densities of about N (H I ) ∼ × cm − would have a (non-projected) radius r . I and other molecular lines such as HCO + (1–0)may suggest that the corresponding absorbing molecularregions are physically connected.The atomic material arising from the broad compo-nent is likely contributing to fuel the powerful AGN ofCentaurus A. Assuming that this H I region is infallingwith a velocity of ∼
37 km s − (mean velocity minus thesystemic velocity), a column density of N (H I ) ∼ × cm − , and a size of 2 pc, we estimate an accretionrate of 2 × − M ⊙ yr − , enough to power the requiredaccretion rate of ∼ × − M ⊙ yr − obtained from itsisotropic radio luminosity (e.g. van Gorkom et al. 1989). SUMMARY AND CONCLUSIONS
We present H I and CO(2–1) absorption features to-ward the nuclear regions of Centaurus A (NGC 5128),using the VLBA and the SMA respectively.Our H I observations allow us to discern with sub-pcresolution the absorption profiles toward the nuclear jetin the inner 5.4 pc. The H I absorption lines are com-posed of a system of 4 main narrow lines as well as anunderlying broad line (FWHM ∼
55 km s − ), in agree-ment with previous works using lower spatial resolutionobservations. Interestingly, the underlying broad absorp-tion line in our data is seen to be more prominent towardthe central 21 cm continuum component (base of the nu-clear jet, although not the nucleus itself) than towardthe continuum emission at ≥ CO(2–1) profile were identified with previously observed molec-ular lines such as HCO + (1–0) (e.g. Wiklind & Combes1997; van Langevelde et al. 1995). Overall, our calculatedisentangling the circumnuclear environs of Centaurus A. II 9isotopic ratios [ C]/[ C] ∼ O]/[ O] ∼
70 – 220 seem to indicate that the molecular cloudsproducing the observed narrow absorption lines close tothe systemic velocity have been in a relatively long-termstarburst environment as in nearby starburst galaxies.We did not detect any counterpart of the broad absorp-tion line in the CO(2–1) spectrum as that found in H I or other molecular lines, although the noise level in ourCO(2–1) profile would have been sufficient to detect thisfeature if it were as strong as that of HCO + (1–0).Since both our H I (position d ) and molecular absorp-tion profiles arise from molecular clouds in front of a sim-ilar continuum emission region within ∼ I to H are no-ticeably different at different velocities. While the lowvelocity components, V <
560 km s − are dominated bymolecular gas, the atomic phase gas seems to dominateat 560 < V <
600 km s − when the H I line is comparedwith the molecular gas content as traced by the CO(2–1) line. Although the molecular to atomic gas ratio isnot drastically different when we consider the molecu-lar gas traced by HCO + (1–0), still there is a trend fora larger atomic content for components at higher veloci-ties. These distinct signatures suggest that the physicalproperties, location and chemistry of the gas producingthe broad line is different to those of the narrow lines.Since the gas corresponding to the broad line is likelynot far from the AGN, we argue that its properties mightbe modified by X-rays, which penetrate much fartherthan UV. The observed abundances of species such asHCO + and HCN were estimated to be ∼ − , andabundance ratios [HCO + ]/[HCN] ≥ ∼ + (1–0). Since X-rays do not strongly dissociateCO molecules and they can coexist at high temperatures,higher transitions of CO other than CO(2–1) could tracethe broad line component, as has been suggested withprevious CO(3–2) single-dish data.Because of its kinematics, the two narrow red-shiftedlines at V >
600 km s − found in our CO(2–1) spectrum,as well as in previous HCO + (1–0) spectrum might beclose to the AGN and falling toward it. These molecularclouds are likely at a distance &
10 pc from the AGN sothat they can survive the X-ray emission. On the otherhand, at distances .
10 pc most of the gas will be inits atomic phase, material that is likely contributing topower the AGN of Centaurus A.We thank the SMA and NRAO staff members whomade the observations reported here possible. We alsothank A. Sarma for providing the 21 cm continuum VLAdata, and S. Martin and I. Jimenez-Serra for interest-ing discussions. This research has made use of NASA’sAstrophysics Data System Bibliographic Services, andhas also made use of the NASA/IPAC ExtragalacticDatabase (NED) which is operated by the Jet Propul-sion Laboratory, California Institute of Technology, un-der contract with the National Aeronautics and SpaceAdministration. DE was supported by a Marie CurieInternational Fellowship within the 6 th European Com-munity Framework Programme (MOIF-CT-2006-40298).
Facilities:
VLBA,SMA
REFERENCESBurton, W. B. 1988, The structure of our Galaxy derived fromobservations of neutral hydrogen (Galactic and ExtragalacticRadio Astronomy), 295–358Combes, F. 2008, Ap&SS, 313, 321Eckart, A. et al. 1990a, ApJ, 365, 522—. 1990b, ApJ, 363, 451Eckart, A., Wild, W., & Ageorges, N. 1999, ApJ, 516, 769Espada, D. et al. 2009, ApJ, 695, 116Evans, D. A. et al. 2004, ApJ, 612, 786Frerking, M. A., Langer, W. D., & Wilson, R. W. 1982, ApJ, 262,590Gardner, F. F., & Whiteoak, J. B. 1976a, Proceedings of theAstronomical Society of Australia, 3, 63—. 1976b, MNRAS, 175, 9PGupta, N. et al. 2007, ApJ, 654, L111Harris, G. L. H., Rejkuba, M., & Harris, W. E. 2009, ArXiv e-prints0911.3180Harrison, A., Henkel, C., & Russell, A. 1999, MNRAS, 303, 157Henkel, C. et al. 1993, A&A, 268, L17—. 2009, A&A, 500, 725Ho, P. T. P., Moran, J. M., & Lo, K. Y. 2004, ApJ, 616, L1Horiuchi, S., Meier, D. L., Preston, R. A., & Tingay, S. J. 2006,PASJ, 58, 211Israel, F. P. 1998, A&A Rev., 8, 237Israel, F. P., Raban, D., Booth, R. S., & Rantakyr¨o, F. T. 2008,A&A, 483, 741Israel, F. P., van Dishoeck, E. F., Baas, F., de Graauw, T., &Phillips, T. G. 1991, A&A, 245, L13Israel, F. P. et al. 1990, A&A, 227, 342Jones, D. L. et al. 1996, ApJ, 466, L63Kellermann, K. I., Zensus, J. A., & Cohen, M. H. 1997, ApJ, 475,L93Maloney, P. R., Begelman, M. C., & Rees, M. J. 1994, ApJ, 432,606 Maloney, P. R., Hollenbach, D. J., & Tielens, A. G. G. M. 1996,ApJ, 466, 561Meijerink, R., & Spaans, M. 2005, A&A, 436, 397Meisenheimer, K. et al. 2007, A&A, 471, 453Muller, S., & Dinh-V-Trung. 2009, ApJ, 696, 176Muller, S., Gu´elin, M., Dumke, M., Lucas, R., & Combes, F. 2006,A&A, 458, 417Neumayer, N. et al. 2007, ApJ, 671, 1329Prantzos, N., Aubert, O., & Audouze, J. 1996, A&A, 309, 760Quillen, A. C. et al. 2008, MNRAS, 384, 1469—. 2006, ApJ, 645, 1092Quillen, A. C., Neumayer, N., Oosterloo, T., & Espada, D. 2009,ArXiv e-prints 0912.0632Roberts, M. S. 1970, ApJ, 161, L9Sarma, A. P., Troland, T. H., & Rupen, M. P. 2002, ApJ, 564, 696Schiminovich, D., van Gorkom, J. H., van der Hulst, J. M., &Kasow, S. 1994, ApJ, 423, L101Seaquist, E. R., & Bell, M. B. 1990, ApJ, 364, 94Shepherd, M. C., Pearson, T. J., & Taylor, G. B. 1995, in Bull.Amer. Astron. Soc., Vol. 27, , 903Solomon, P. M., Sanders, D. B., & Scoville, N. Z. 1979, ApJ, 232,L89Spaans, M. 2008, EAS Publications Series, Vol. 31, ed. C. Kramer,S. Aalto, & R. Simon, 47–520Tingay, S. J. et al. 1998, AJ, 115, 960Tingay, S. J., & Murphy, D. W. 2001, ApJ, 546, 210van der Hulst, J. M., Golisch, W. F., & Haschick, A. D. 1983, ApJ,264, L37van Gorkom, J. H. et al. 1989, AJ, 97, 708van Gorkom, J. H., van der Hulst, J. M., Haschick, A. D., & Tubbs,A. D. 1990, AJ, 99, 1781van Langevelde, H. J., Pihlstr¨om, Y., & Beasley, A. 2005, Ap&SS,295, 249 a)b)
Fig. 1.— a ) CO(2–1) normalized intensity spectra obtained from all the visibilities using a fit in the UV plane assuming a point-likefeature for the continuum emission. It clearly fails to eliminate most of the emission arising from the circumnuclear molecular gas (Espadaet al. 2009). The arrow indicates the systemic velocity of the galaxy ( V = 541.6 km s − ). b ) Same as a ) but only using visibilities withprojected baselines larger than 50 m. The contamination by emission is efficiently diminished. Fig. 2.—
Nuclear jet of the 21 cm continuum emission of Cen A in the inner 0 . ′′ S = 600 ±
30 mJy beam − . Note that this is not associated with the nucleus itself but withthe base of the nuclear jet ( § × σ , where σ = 2 mJy beam − . The grey scale ranges from −
10 to 603 mJy beam − . The restoring beam ischaracterized by a half power beam width HPBW = 36.3 × . ◦
8, as is shown by the filled ellipse inthe lower right of the plot. Note that 1 mas corresponds to 0.018 pc.van Langevelde, H. J., van Dishoeck, E. F., Sevenster, M. N., &Israel, F. P. 1995, ApJ, 448, L123Wang, M. et al. 2004, A&A, 422, 883Wiklind, T., & Combes, F. 1997, A&A, 324, 51 Wild, W., Eckart, A., & Wiklind, T. 1997, A&A, 322, 419Wilson, T. L., & Rood, R. 1994, ARA&A, 32, 191 isentangling the circumnuclear environs of Centaurus A. II 11
Fig. 3.— H I absorption profile of Cen A, integrated over the entire 21 cm continuum region probed by our VLBA experiment (Fig. 2).The profile is mainly composed of five lines: IV, SV, I/II and III, plus an underlying broader line with a width of 53 ±
10 km s − , whosewing is clearly seen at V >
600 km s − . The velocity resolution is 3.3 km s − and the rms noise of 3 mJy beam − per channel. The solidline is the sum of the Gaussian fits to the individual lines (dashed lines, as specified in Table 2). The plot at the bottom shows the residualfrom this fit, with a rms σ = 0.01 Jy. Fig. 4.— H I absorption lines toward single pixels at positions a , b , c and d as outlined in Figure 2 (see bottom right of each panel), aswell as the individual Gaussian fits and underlying final fitted curve. The residuals from the fit are characterized by a σ as indicated onthe lower left of each panel. Fig. 5.— H I optical depths for the four different positions along the 21 cm jet ( a , b , c ) and brightest component ( d ), as outlined inFigure 2. Positions are indicated at the upper right of each panel. Fig. 6.—
Fits to the H I optical depth ( τ ) profiles at positions c (left) and d (right). A zoom in optical depth from -0.05 to 0.15 isincluded to emphasize the weak lines. isentangling the circumnuclear environs of Centaurus A. II 13 c)a)b) d)e)1 2 3 4 5 6 Fig. 7.— a ) CO(2–1) normalized intensity spectrum using all visibilities. Numbered labels indicate the different molecular gas absorptioncomponents. b ) CO(2–1) spectrum only with projected baselines longer than 50 m. c ) CO(2–1) spectrum. The three main features closeto the systemic velocity are detected. d ) C O(2–1) spectrum. No absorption lines have been detected with our sensitivity. No emissionis seen either in the CO(2–1) or the C O(2–1) spectra, therefore we include all the visibilities. e ) Gaussian fits to the single-dishHCO + (1–0) spectrum (Wiklind & Combes 1997). The arrow indicates the systemic velocity ( V = 541.6 km s − ). Fig. 8.—
Comparison of the CO(2–1) (dashed line) and the H I absorption profiles (solid line). The velocity resolution is 1 km s − and 3.3 km s − for CO(2–1) and H I , respectively. !" >?@ DDB
F G DE H
DDB
DDB
F G DE H I,%3!(
DFD
DFD DFD
Fig. 9.—
Scheme showing the edge-on and face-on views of the proposed models within the inner 2 kpc, as well as the correspondingabsorption lines: a ) warped thin disk (as in Quillen et al. 2005), b ) warped disk with diffuse gas (Eckart et al. 1999), and c ) warped diskwith a contribution of non-circular motions (weak bar) model (Espada et al. 2009). isentangling the circumnuclear environs of Centaurus A. II 15 TABLE 1Main parameters of the VLBA H I and SMA CO(2–1) observations . H I VLBA observations
Date 1994 November 26R.A. of phase center (J2000) 13 h m s ◦ ′ . ′′ × × . ◦ − ) 541.6Total bandwidth 2 MHz ( ∼
430 km s − )Spectral resolution (km s − ) 3.3rms noise (3.3 km s − ) 3 mJy beam − rms (continuum map) 2 mJy beam − CO(2–1) SMA observations
Date 2006 April 5R.A. of phase center (J2000) 13 h m s ◦ ′ ′′ On source time (hr) 3FWHM of synthesized beam 2 . ′′ × . ′′ ×
108 pc), PA = 0 . ◦ − ) 550Total bandwidth 2 GHz ( ∼ − ) in each sideband(separated by 10 GHz)Spectral resolution (km s − ) 1rms noise (1 km s − ) 0.1 Jy beam − (0.3 Jy beam − for baselines >
50 m)
TABLE 2Parameters of H I absorption lines integrated over the entire 21 cm continuum Absorption V Peak S (a) FWHM (a)
Peak τ R τ d V N (HI)/ T s feature (km s − ) (mJy beam − ) (km s − ) (km s − ) (cm − K − )IV 541 ± − ±
19 5 ± ± ± . ± . × SV 552 ± − ±
20 8.5 ± ± ± ± × I, II 575 ± − ±
17 12.6 ± ± ± ± × III 595 ± − ±
20 5.2 ± ± ± . ± . × Broad 578 ± − ±
21 53 ±
10 0.07 ± ± ± × Flux density peak and FWHM are calculated from a Gaussian fit.
TABLE 3Parameters of H I absorption lines toward individual positions (a) Absorption V Peak S FWHM Peak τ R τ d V N (H I )/ T s feature (km s − ) (mJy beam − ) (km s − ) (km s − ) (cm − K − )a)IV (541) < < < < × SV c ± − ± < < < × I/II (575) < < < < × III (595) < < < < × Broad (578) < < < < × b)IV (541) <
17 (5) < < < × SV 552 ± ± ± ± ± ± × I/II (575) <
17 (14) < < < × III (595) <
17 (6) < < < × Broad (578) <
17 (50) < < < × c)IV 541 ± ± ± ± ± ± × SV 551 ± ± ± ± ± ± × I/II 576 ± ± ± ± ± ± × III 596 ± ± ± ± ± ± × Broad 578 ± ± ±
10 0.01 ± ± ± × d)IV 540 ± ±
10 3 ± ± ± ± × SV 551 ± ±
10 7.7 ± ± ± ± × I/II 574 ± ±
10 12.8 ± ± ± ± × III 595 ± ±
10 5.6 ± ± ± ± × Broad 577 ± ±
10 55 ±
10 0.06 ± ± ± × Velocity ( V ), flux density peak (Peak S ), FWHM (full width half maximum), peak optical depth (Peak τ ) and integrated optical depth( R τ d V ) were derived from Gaussian fits. Velocities in parenthesis are assumed values in the fitting. Uncertainties in the flux density peakand the peak optical depth correspond to the residuals after Gaussian fitting (see Fig. 4). Uncertainties in the integrated optical depth arecalculated as the uncertainty of the area of the Gaussian profile. Upper limits in the integrated optical depth are 3 σ , and are calculated as3 × σ × (∆ V δv ) / , where ∆ V is the velocity of the integrated profile and δv the channel spacing. Line SV in position a is detected, butthe optical depth cannot be calculated since the absorption depth is larger than the continuum level due to the noise. Thus, we considerit as an upper limit. isentangling the circumnuclear environs of Centaurus A. II 17 TABLE 4Parameters of the CO(2–1), CO(2-1) and C O(2-1) absorption lines
Isotope Line
V S
Peak τ R τ dV N (CO) ( a ) N (H ) ( b ) (km s − ) (Jy beam − ) (km s − ) (10 cm − ) (10 cm − ) CO(2-1)
539 0.72 ± ± ± ±
10 725 ±
544 1.20 ± ± ± ± ±
550 0.16 ± ≥ ≥ ≥ ≥
575 5.33 ± ± ± ± ±
606 2.66 ± ± ± ± ±
613 4.61 ± ± ± ± ± CO(2-1)
541 5.30 ± ± ± ± ±
545 4.90 ± ± ± ± ±
552 3.33 ± ± ± ± ± – – – < . < <
607 5.80 ± ± ± ± ± – – – < . < < O(2-1) – – – – < . < < a Isotopic CO column densities.We assume a filling factor of unity, local thermodynamic equilibrium (LTE) and an excitation temperature T ex = 10 K for the absorbing molecular gas ( § is likely saturated, so the optical depth and the columndensities are lower limits. CO lines and and C O lines are not detected, thus we use upper limits taken at a 3 σ level. b H column density ( N (H )) derived from the CO(2–1) absorption lines (and isotopes). To derive the N (H ) from the CO(2–1) lines, weuse an abundance ratio of [CO]/[H ] = 8.5 × − (Frerking et al. 1982), [ CO]/[H ] = 1 × − (Solomon et al. 1979) and [C O]/[H ]= 1.7 × − (Frerking et al. 1982) ([CO]/[ CO] = 85 and [CO]/[C18