Movatterモバイル変換

• Top
• Abstract
• 1. Introduction
• 2. Observations and data handling
• 3. Results
• 4. Analysis and discussion
• 5. Conclusions
• Acknowledgments
• References
• List of tables
• List of figures

© ESO, 2014

1. Introduction

Disks of dense dust and gas deeply embedded in the stellar body of a giant elliptical galaxy are the more easily identified remnants of smaller gas-rich galaxies that have fallen in. They are a transient phenomenon: on varying time-scales, the gas will be consumed by accretion onto a central black hole, expulsion by jets emanating from the nucleus, and by the in-situ formation of new stars rejuvenating the host galaxy stellar population. All three processes are currently active in NGC 5128, the host of the Fanaroff-Riley class I (FR I) radio source Centaurus A (Cen A).

The properties of these embedded disks are probably different from those of the interstellar medium (ISM) in spiral or star-burst galaxies (see e.g. the review by Henkel & Wiklind1997). Kiloparsec-sized embedded thin disks are directly exposed to the intense but UV-poor combined radiation from all stars in the host elliptical. There is little differential rotation, and the lack of shear may locally favour the formation of massive groups of luminous stars in the disk affecting their surrounding ISM. On much smaller sub-kiloparsec scales, the dynamics of a denser circumnuclear disk may be only loosely related to those of the larger kiloparsec-sized extended disk. Instead, in such a circumnuclear disk both dynamics and excitation expected to be tied more closely to the properties of the super-massive black hole in the nucleus.

The emission lines from carbon and carbon monoxide are key to understanding the properties of the ISM in galaxies, as they provide almost all the cooling of the dense neutral gas. Warm and tenuous gas is traced by ionised carbon ([CII] ), warm and dense gas by neutral carbon ([CI] ) and carbon monoxide (CO), and cold and dense gas by (CO) and its isotopologues. Although the two neutral carbon [CI] lines at 492 and 809 GHz and CO lines with rest frequencies up to about 1000 GHz can be measured from the ground (albeit with increasing difficulty), the ionised carbon [CII] fine-structure line at 1.9 THz requires a platform above most or all of the atmosphere.

NGC 5128 is the nearest (D = 3.84 Mpc) giant elliptical (review by Israel1998). Its central black hole is still accreting in the aftermath of a merger with a medium-sized late-type galaxy a few hundred million years ago (Graham1979; Struve et al.2010). The remnant ISM of the merged galaxy has redistributed itself into a warped, thin disk (“the extended thin disk” – hereafter called the ETD – which shows in projection as a dark band crossing the optical image of the galaxy (Dufour et al.1979; Nicholson et al.1992). It has comparable amounts of atomic (HI) and molecular H₂ gas, and its total mass is uncertainly estimated at1.5 × 10⁹  M_⊙, about two per cent of the enclosed dynamical mass. On significantly smaller scales, the nuclear black hole of5 × 10⁷  M_⊙ (Cappellari et al.2009) is obscured by a compact circumnuclear disk (CND) that feeds the black hole (Israel et al.1990).

Unlike the ETD, the compact CND is oriented at right angles to the luminous radio and X-ray jets emanating from the Centaurus A nucleus. This suggests that the CND and the jets are somehow connected, but it is not yet clear how that may be. The CND is a well-defined entity with a diameter of20′′ (400 pc). It has been seen in FIR continuum (Hawarden et al.1993) and¹²CO line emission (Espada et al.2009), but was not yet studied in detail.

In this paper, we combine single-dish CO and [CI] observations of the Cen A nucleus obtained from the ground over many years, and from space. Although observations of the Cen A nucleus distinguishing CND and ETD in the lower (J_upper ≤ 3) transitions of CO are found in a few earlier papers (Israel et al.1990; Israel1992; Rydbeck et al.1993), all data in this paper have much higher signal-to-noise ratios and are presented here for the first time.

2. Observations and data handling

2.1.Herschel-SPIRE

The nucleus of Centaurus A was observed with the Spectral and Photometric Imaging Receiver and Fourier-Transform Spectrometer (SPIRE-FTS – Griffin et al.2010) onboard theHerschelSpace Observatory¹ (Pilbratt et al.2010) in the single-pointing mode with sparse image sampling. The FTS has two detector arrays (the SLW: wavelength range 303–671μm corresponding to a frequency range 447–989 GHz, and the SWW: wavelength range 194–313μm corresponding to a frequency range 959–1544 GHz). The observations are summarised in Table1. The data were processed and calibrated using HIPE version 9.0.0. The spectral range covered the  ¹²CO lines in theJ = 4–3 toJ = 13–12 transitions as well as the two submillimeter [CI] lines, which were all detected; however, no  ¹³CO lines were detected. The spectral resolution of 1.21 GHz was insufficient to resolve any of the lines in the SLW spectrum, but may start to resolve some of the lines in the SSW spectrum, particularly the [NII] line. Line fluxes were extracted by fitting a sinc-Gaussian function to the line profile. The beam FWHM values are given in the on-lineHerschel-SPIRE manual; they range from 29′′ to 42′′ for the SLW, and from 16.8′′ to 21.1′′ for the SSW. At the overlap between the two arrays, at about 310μm or 964 GHz, the beam size jumps from 21.1′′ to 37.3′′ going from SSW to SLW. The results are shown in Fig.3, and summarised in Tables2 and4.

Fig. 1 Baseline-subtracted line profiles towards the center of NGC 5128 (Centaurus A) obtained with the ground-based telescopes SEST, JCMT, and APEX (top two rows) and with the HIFI instrument on-board of theHerschelSpace Observatory (bottom two rows). Vertical scale is${\mathit{T}}_{\mathrm{A}}^{\mathrm{\ast }}$ in Kelvin, horizontal scale isVLSR in  km   s-1. Species, transition and telescope used are identified at the top left corner of each panel. The HIFI  12CO results for theJ = 8–7,J = 10–9, andJ = 13–12 transitions are the averages of the profile towards the nucleus and the adjacent positions separated from that by10′′. Note that the deep absorption apparent in the13CO (1–0) profile is real; its depth to negative values reflects the weakness of the line emission relative to the (subtracted) continuum strength at the same frequency. For more details, see Sect. 2 and Tables2 and3.

Fig. 2 Baseline-subtracted line profiles towards the center of NGC 5128 (Centaurus A) obtained with the APEX telescope and the HIFI instrument on-board theHerschelSpace Observatory. The HIFI [CI] and [CII] profiles for the nuclear position, and the positions offset by± 10′′ from the nucleus, are shown separately. The NW offset profile is represented by a dashed line, the SE offset profile by a solid line. Species, transition and telescope used are identified at the top left corner of each panel. Vertical scale is${\mathit{T}}_{\mathrm{A}}^{\mathrm{\ast }}$ in Kelvin, horizontal scale isVLSR in  km   s-1. For more details, see Table4.

2.2.Herschel-HIFI

The center of Centaurus A was also observed with the Heterodyne Instrument for the Far Infrared (HIFI – de Graauw et al.2010) on-boardHerschelas part of the guaranteed time key programme HEXGAL (PI: R. Güsten). Observations were carried out in fast-chopping dual-beam switch mode using a wobbler throw of 3′ for all observations, and are summarised in Table1. Calibration was achieved through hot-cold absorber measurements. The data were recorded using the wide-band acousto-optical spectrometer, consisting of four units with a bandwidth of 1 GHz each, covering the 4 GHz IF for each polarisation with spectral resolutions of 1 MHz. Data were reduced using the HIPE and CLASS software packages. For each scan, we combined the four sub-bands in each polarisation to create a 4 GHz spectrum. We subtracted first-order baselines. For each line, we inspected the result in each polarisation (H and V) separately. The continuum and line amplitudes agreed within 15%. We have used the calibration in Table 5.5 of the on-line HIFI Observer’s Manual to convert antenna temperatures to main-beam temperatures and flux densities. Between 480 and 960 GHz, the main-beam efficiency is almost constant, dropping fromη_mb = 0.76 toη_mb = 0.74, and the antenna temperature to flux conversion factor likewise changes only little from 464 to 472 Jy/K.

With the exception of the C¹⁸O(5–4) and [NII] lines (neither of which was detected), we observed all lines at three positions: the center, and two positions offset by 10′′ to either side in a position angle of 45° in order to separately sample emission from the CND. The central position was observed with reasonably good signal-to-noise ratios in theJ = 5–4,J = 7–6,J = 9–8, andJ = 13–12 transitions of  ¹²CO, as were theJ = 13–12  ¹²CO NW and SE offset positions. The other CO measurements had much shorter durations resulting in poorly defined spectral profiles, but the [CI] and [CII] lines were again observed with (very) good signal-to-noise ratios at all positions. As shown in Tables2–4, the HIFI beam sizes ranged from38″ at 576 GHz to11″ at 1900 GHz. The profiles observed with HIFI are shown in Figs.1 and2.

2.3. APEX 12m

Between 2007 and 2011 we have used the Vertex Antennentechnik APEX² 12-m telescope (Güsten et al.2006) to observe the nucleus of NGC 5128 in theJ = 3–2  ¹³CO transition at 330 GHz, theJ = 4–3 andJ = 6–5  ¹²CO transitions at 461 and 691 GHz respectively, and the two submillimeter [CI] transitions at 492 and 809 GHz. The location of APEX at the high elevation of 5105 m renders it very suitable to high-frequency observations from the ground. The observations were made with the First Light APEX Submillimeter Heterodyne (FLASH) dual-frequency receiver (Heyminck et al.2006) and the Carbon Heterodyne Array (CHAMP+) receiver (Güsten et al.2008; Kasemann et al.2006), both developed by the Max Planck Institut für Radioastronomie in Bonn (Germany). Main-beam efficiencies were 0.73, 0.60, 0.56, and 0.43 at operating frequencies of 352, 464, 650, and 812 GHz. At the same frequencies, the antenna temperature to flux density conversion factors were 41, 48, 53, and 70 Jy/K, respectively, with beam sizes ranging from18″ to7.7″.

All observations were done under excellent weather conditions with typical overall system temperatures of 2100 K for CHAMP+-I (SSB, 690 GHz), 7500 K for CHAMP+-II (SSB, 800 GHz), and 1100 K for FLASH-I (DSB, 460 GHz), 290 K for APEX-1 and 230 K for APEX-2a (both SSB). Calibration errors are estimated at 15 to 20%. Observations were taken using fast Fourier transform spectrometer (FFTS; Klein et al.2006) back-ends for all instruments, except CHAMP+, for which only the two central pixels were attached to the FFTS back-ends. Other CHAMP+ pixels were attached to the MPI Array Correlator System (MACS) back-ends. FFTS back-ends are able to reach resolutions of 0.12 MHz (0.045  km   s^-1 at 800 GHz), while the MACS units were used at a resolution of 1 MHz (0.36  km   s^-1 at 800 GHz). APEX absolute pointing accuracy is~2′′ (rms), but its pointing on track is accurate to0.6′′ (rms). All observations were taken using position switching with reference positions in azimuth ranging from 600′′ to 3600′′. The results are shown in Figs.1 and2, and summarised in Tables2–4.

2.4. JCMT 15m

The 15mJames Clerk MaxwellTelescope (JCMT)³ on top of Mauna Kea (Hawaii) was used between 2003 and 2005 to measure theJ = 3–2 transitions of  ¹²CO and  ¹³CO at 345 and 330 GHz respectively towards the nucleus of Centaurus A. At the observing frequencies, the beam size was 14′′ and the main-beam efficiency was 0.62. The antenna temperature to flux density conversion was 29.4 Jy/K. We used the dual-polarisation receiver RxB and the Digital Autocorrelating Spectrometer (DAS) back-end in wide-band mode, with total band-widths of 920 MHz (800 km s^-1) and 500 MHz (435  km   s^-1), providing velocity resolutions of 0.65 and 0.325  km   s^-1 respectively. As seen from Hawaii, Centaurus A culminates at an elevation of only 23° above the horizon. We observed the object only when above 20°, i.e. about 2 h per day. All observations were taken in beam-switching mode with a throw of 3′ in azimuth. The results are shown in Fig.1, and summarised in Tables2 and3.

2.5. SEST 15m

We have used the 15m Swedish ESO Submillimetre Telescope⁴ (SEST; Booth et al.1989) on top of Cerro La Silla (Chile) to observe theJ = 1–0 HCN,J = 1–0, andJ = 2–1  ¹²CO and  ¹³CO transitions with angular resolutions of 57′′, 45′′, and 23′′ respectively. In order to convert observed antenna temperatures to main-beam temperatures, we use efficienciesη_mb = 0.75, 0.70, and 0.50 respectively. Similarly, we derive flux densities from the observed antenna temperatures by applying the respective conversion factors 25, 27, and 41 Jy/K. All observations were made in the double-beam switching mode, with a throw of 12′ and a frequency of 6 Hz, producing excellent baseline stability. In the early period 1989–1993, we used a relatively noisy Schottky barrier diode receiver. Between 1996 and 2003 we obtained spectra with the more sensitive SIS receiver, using high- and low-resolution AOS back-ends in parallel. The data presented in this paper were taken with the latter, which had a total bandwidth of 500 MHz (later 1 GHz), and a resolution of 1 MHz (later 1.4 MHz). The results are shown in Fig.1, and summarised in Tables2 and3.

Fig. 3 Left: full submillimeter spectrum of the center of NGC 5128 (Centaurus A) obtained with the SPIRE instrument on board theHerschelSpace Observatory (See Sect. 2). Species and transitions are identified throughout. The jump in the continuum at 944 GHz is caused by the different angular resolutions of the SLW and the SSW (see Sect. 2). The SPIRE continuum contains a contribution from the point source nucleus of about 8.2 Jy at 460 GHz, slowly decreasing with frequency. The remaining continuum is due to extended thermal emission from dust, increasing with frequency. Vertical scale is flux in Jansky, horizontal scale is frequency in GHz. For more details, see Tables2 and3.Right: comparison of SPIRE CO line fluxes of the NGC 5128 center with those of the starburst galaxy M 82, the AGN+starburst NGC 1068, and the (U)LIRGs Arp 193, Arp 220, and NGC 6240. Mrk 231 has not been marked separately as its spectral ladder is identical to that of NGC 6240. The ratio of the galaxy CO line flux to the Cen A CO line flux is shown for eachJ transition, as observed by SPIRE without correction for finite beam-size, source extent, and beam efficiency. The vertical line separates measurements obtained with the SSW from those obtained with the SLW in a beam roughly twice as wide.

3. Results

3.1. Complex nature of the observed nuclear line profiles

The complexity of the individual line profiles is particularly obvious in the lowerJ transitions of  ¹²CO (Fig.1). These illustrate the presence of the various contributions that are due to physically distinct components in the line of sight to the nucleus (see Israel1992,1998). These are: (a) very broad line emission at300      km   s^-1 < V_LSR < 800      km   s^-1 sampling the rapidly rotating compact CND, fully covered by beams larger than20′′, but progressively more resolved in smaller beams; (b) narrower line emission roughly at450      km   s^-1 < V_LSR < 650      km   s^-1, most prominent in the lower  ¹²CO transitions, that samples the much larger and fully resolved ETD, and (c) a pattern of absorption against the nuclear continuum point source (but not the extended dust continuum) in the range500      km   s^-1 < V_LSR < 625      km   s^-1. Because the continuum emission from the nucleus is completely unresolved even in the smallest observing beams, the absorption features sample the molecular ISM in both the ETD and CND along the line of sight to the nucleus in a very narrow pencil-beam, less than0.1′′ across (i.e.<2 pc).

In the remainder of this paper, we will first briefly discuss the continuum emission, and then concentrate on theline emission, i.e. contributions (a) and (b). Analysis and discussion of theline absorption(contribution (c)), based on profiles with higher spectral resolution than shown here, will be presented in a subsequent paper.

3.2. Nuclear continuum emission

In the beams used to obtain the velocity-resolved profiles, the (variable) flux of the nuclear point source is dominant, and dust continuum emission is negligible (cf. Israel et al.2008). However, at the higher frequencies shown in Fig.3, there is a steadily increasing contribution by thermal emission from dust extended over the beam area. The unresolved nuclear continuum spectrum can be approximated by a power-lawF_ν    ∝    ν^-0.36 (Meisenheimer et al.2007). From APEX-1 and APEX-2a observations of the continuum underlying the HCO⁺ line emission at 267 GHz and 355 GHz made in June, 2010 (i.e. close to the SPIRE observing date of August, 2010) we extrapolate a flux densityF_461  GHz    =    8.2 ± 0.3 Jy. As this is indistinguishable from the value implied by Fig.3, we conclude that at the lowest frequencies observed with SPIRE the continuum flux is still wholly due to the nucleus. At the highest observed SPIRE frequencies around 1500 GHz, the nuclear flux has decreased toF_1500  GHz    ≈    5 Jy, and contributes no more than15% to the continuum flux measured in the aperture.

3.3. Spectral line flux distributions

3.3.1. Integral CND/ETD line profiles

First, we will analyse the observed (CND/ETD) spectral line emission without attempting to separate it into the individual ETD and CND contributions. This allows us to use well-defined integrated line intensities (such as those from SPIRE) up to theJ = 12–11  ¹²CO transition, not hampered by the uncertainties inherent in component decomposition, and suitable for comparison with measurements of other galaxy centers. This is particularly important for the highest observed line transitions, which suffer from increasingly poor baseline-definition and signal-to-noise ratios in the velocity-resolved HIFI measurements. The measured emission line intensities, and the different beam sizes used, are given in Tables2–4.

In constructing the NGC 5128 CO spectral line ladder we need to take into account the variation in beam sizes. The CND and the ETD have finite but different extent, and the observed emission line fluxes must be normalised to the same beam-size. To accomplish this, we have taken the ALMA Band 6 high-resolution mosaic of theJ = 2–1  ¹²CO emission from Centaurus A obtained in the public ALMA Calibration and Science Verification program. From these data, we have constructed a series of maps at spatial resolutions identical to those of our single-dish line observations. From each of these maps, we have extracted the central (nuclear) line profile and determined its integral value. Assuming that the distribution on the sky of CO emitting clouds is identical for all transitions, we have determined the beam normalisation factors (BNFs) required to relate the emission line fluxes to one another. These factors are included in Table5.

When more than one independent measurement was available, we took the error-weighted mean to derive the flux. In all these cases, the derived normalised fluxes agree reasonably well with one another, and also with values obtained from linear interpolation between measurements with beams larger respectively smaller than22′′. For instance, our normalised flux for the [CII] line is only10% above the value interpolated from the HIFI (11′′ beam) and ISO-LWS (70′′ beam) measurements published by Unger et al. (2000). This is quite gratifying, as the large beam normalisation factors appropriate to such small beams are at the limit of reliability.

3.3.2. Absorption correction and normalisation

Since the absorption occurs against the continuum emission from a nucleus very much smaller than any of the beams used, the absorption line intensity, unlike the emission line intensity, is independent of beam size. We have determined the magnitude of the absorption by subtracting the integrated line fluxes from the Gaussian-fitted line fluxes. This amount is once again added to the normalised emission line flux to finally yield the normalised and corrected lineluminositieslisted in Table5. We have adopted an uncertainty in the absorption line flux of 50%.

Because of the poor definition of the line profiles beyond theJ = 7–6 transition, we have not attempted to correct for absorption losses at the higher  ¹²CO transitions. We believe that the effect of this on the analysis is limited, as Table5 shows that the magnitude of the flux absorbed against the nuclear continuum decreases steadily fromJ = 2–1  ¹²CO upwards. This is probably due to decreasing absorption optical depths at increasingJ levels, and to the slow flux decrease of the nuclear continuum itself.

For the  ¹³CO transitions we only give the normalised and corrected luminosities. Because of the relatively low signal-to-noise ratios of these weaker lines, we have not attempted to process the individual profiles. Instead, we have divided the  ¹²CO luminosities by the  ¹²CO/   ¹³CO ratios presented in Table3. This is more accurate because these were obtained by fitting, in each transition, the  ¹³CO profiles to the much more accurate  ¹²CO profiles.

4. Analysis and discussion

4.1. Comparison of the central CO ladders in different galaxies

CO line fluxes extracted from SPIRE spectra have been published for a limited number of active (mostly Seyfert) galaxies (Pereira-Santaella2013) and also for the relatively nearby NGC 1068 (Spinoglio et al.2012). SPIRE data are also available for star-burst galaxies including the nearby M 82 (Panuzzo et al.2010) as well as the more distant and much more energetic Markarian 231 (Van der Werf et al.2010), Arp 220 (Rangwala et al.2011), NGC 6240 (Meijerink et al.2013), Arp 193 (Papadopoulos et al.2013). The latter are all (ultra)luminous infrared galaxies, or (U)LIRGs. Due to their much greater distances (D = 42–107 Mpc) they were covered in their entirety, whereas the SPIRE aperture sampled only the central region in the nearby galaxies. Because M 82 is at the same distance as Centaurus A, theHerschelmeasurements with SPIRE (Panuzzo et al.2010) and HIFI (Loenen et al. 2010) are directly comparable to those presented here. Unlike NGC 5128, M 82 has a star-burst center lacking an easily identifiable nucleus. In NGC 1068, the SPIRE-SLW aperture covers both the Seyfert AGN and the star-burst, but the SSW aperture covers only the AGN and its surroundings (cf. Fig. 2 in Spinoglio et al.2012). It is of interest to compare the SPIRE line spectra of NGC 5128 and these galaxies and identify any differences.

The right-hand diagram in Fig.3 shows, for each CO transition, the observed line flux ratio of the luminous star-burst galaxies and the NGC 5128 center. No errors are indicated, as the formal extraction errors are very small so that the uncertainties are dominated by systematic effects. All the low-frequency (SLW) flux ratios are seen to increase with transition. However, at the highest (SSW) frequencies, the flux ratios with Arp 220, Arp 193 and the M 82 center are constant, whereas ratios for NGC 6240 (and the ULIRG Markarian 231 which has an identical CO ladder – see Van der Werf et al.2010; Meijerink et al. 2012) keep increasing. This pattern suggests that the most-highly excited gas fraction in NGC 5128, emitting in the highest (SSW) transitions, is similar in nature to the corresponding gas phase in the star-burst galaxies, whereas the bulk of the molecular gas in NGC 5128 (emitting in the lower transitions of the SLW region) is much less highly excited than most of the gas in the star-burst galaxies.

The NGC 5128 CO ladder peaks in theJ = 5–4  ¹²CO transition (corresponding to an upper energy level temperatureT = E_U/k = 83 K), and then decreases to low levels barely reaching a quarter of the peak luminosity in theJ = 10–9 transition (see, for instance, Fig.5, and also Table6). The CO emission from all bright-star-burst galaxy peaks in theJ = 7–6 (155 K) orJ = 8–7 (200 K) transitions. These galaxies have CO ladders that are relatively flat between theJ = 5–4 and theJ = 13–12 (500 K) transitions. Their SPIRE FTS spectra show strong CO lines up to the highest observed frequencies near 1550 GHz (J = 13–12), whereas the spectrum of NGC 5128 exhibits CO lines drowning in the noise beyondJ = 9–8 near 1000 GHz. In M 82, resembling the (U)LIRGs but more modestly excited, the CO ladder peaks in theJ = 7–6 transition, and CO line luminosities decline more rapidly than in the (U)LIRGs but not nearly as fast as in NGC 5128. For instance, theJ = 10–9 transition in the M 82 center still has60% of the luminosity in the peak transition, versus26% for the NGC 5128 center. The comparison with the resolved AGN-galaxies is more difficult to make, but except for NGC 7582 their CO ladders peak in theJ = 4–3/J = 5–4 transitions, only slightly below the Cen A peak. All AGN ladders drop significantly with increasing transition. This behaviour is illustrated by the CO(10–9) to CO(5–4) luminosity ratios summarized in Table6. The lowest ratio (corresponding to the steepest CO ladder high frequencydrop) is presented by the NGC 5128 center, followed by the average AGN; the average star-burst galaxy shows ariseinstead. Thus, the central NGC 5128 CO ladder is the “coolest” of all galaxies considered.

4.2. Remarkably strong [CI] emission lines

NGC 5128 has aJ = 2–1/J = 1–0 [CI] luminosity ratio of 2.6 (corresponding to a line brightness temperature ratio of 0.6), indistinguishable from both M 82 and NGC 1068, and more generally in-between themeanvalues found for the AGN centers and the star-burst disks (cf. Table6).

However,both[CI] lines in the center of NGC 5128 are rather bright with respect to the CO lines, unlike the situation in the other galaxies (Table6). In NGC 5128, the 809 GHz [CI](2–1) line is almost six times stronger than its 807 GHz CO(7–6) neighbour. This very high NGC 5128 [CI]/CO(7–6) ratio is unparalleled. The star-burst galaxies show the weakest [CI] lines; the AGNs have double their relative [CI] line intensity but this still falls far short of the NGC 5128 value. The 492 GHz [CI] line is also more intense than the nearby 461 GHz CO(4–3) line. Corresponding [CI]/CO(4–3) line ratios are much lower in the luminous star-burst galaxies. They are also lower (but not as much) in the AGNs, the Milky Way center, and in the 15 nearby galaxies of various type observed from the ground by Israel & Baas (2002), and Hitschfeld et al. (2008). Among these, high ratios as in NGC 5128 are found only in NGC 3079, NGC 4826, NGC 4945, M 51, and Circinus. At least three of the latter galaxies have a nuclear outflow. Such high ratios provide a strong hint that an excitation mechanism other than PDR is required (Israel2005), as does the very high [CI]/  ¹³CO(2–1) ratio in NGC 5128. Indeed, the observed [CI] and CO line intensities are inconsistent with e.g. the PDR models by Kaufman et al. (1999). In these models, the observed [CI]492 GHz lineintensityrequires a densityn_o ≈ 3000      cm^-3, and a very high radiation fieldG ≥ 5 × 10⁶  G_o, whereas the intersection of the [CI]809 GHz/[CI]492 GHz and [CI]492 GHz/CO(1–0) lineintensity ratiosdefine a densityn_o ≈ 1000      cm^-3 and a very low radiation fieldG ≈ 10  G_o.

Finally, by comparing the intensities of a particular line obtained with different apertures, we may obtain an estimate for the size of the area emitting in that line. A very compact source will produce the same brightness in all apertures independent of size, whereas a very extended source will rapidly become brighter with increasing aperture size. We conclude from the COJ = 4–3, COJ = 7–6, and [CI]J = 1–0 data in Tables4 and2 that the [CI] emission arises from a much more compact region than theJ = 4–3  ¹²CO emission but that the sizes of the regions emitting in theJ = 7–6  ¹²CO and [CI] transitions are not very different (implying that theJ = 7–6  ¹²CO is also much more compact than theJ = 4–3  ¹²CO distribution).

Thus, the ladder of CO intensities from the (AGN-dominated) central region of NGC 5128 is quite distinct from the CO ladders representing extragalactic star-burst environments. In NGC 5128, the CO ladder intensities decrease more rapidly and most of the molecular gas is distinctly less excited than in the other galaxies. Most remarkably, NGC 5128 has [CI] line emission (much) stronger than that of the nearest CO lines. This is quite unlike other galaxies, whether normal or in possession of an AGN or a star-burst, where the [CI] lines are less prominent.

Fig. 4 Position-velocity maps of molecular line emission from the central region of Centaurus A, in position angle 125° counter-clockwise from north (data summarised in Tables2 and3). Horizontal scales are velocityV(LSR) in  km   s-1, vertical scales are offsets from the nucleus in arcsec. Thepanelson theleftshow emission from  12CO; theJ = 1–0,J = 2–1, andJ = 3–2 transitions are shown fromtop to bottom, respectively. Thepanelson therightshow emission from  13CO (J = 1–0 atthe top,J = 2–1 inthe middle). TheJ = 1–0 HCN transition is atbottom right. TheJ = 1–0 maps have resolutions of45′′−55′′, all other maps have an effective resolution of23′′. In all three  12CO maps, the contours are at multiples of 50 mK in main-beam brightness temperature. In the  13CO maps contours are at multiples of 5 mK (J = 1–0) and 10 mK (J = 2–1). The HCN map contours are at multiples of 5 mK. Strong absorption is clear in all panels nearVLSR = 550     km   s-1. Emission from the rapidly rotating compact nuclear disk becomes progressively more clear with increasingJ-transition in the  12COpanels, and in the HCN map.

4.3. Determination of individual CND and ETD CO ladders

Notwithstanding the advantages of analysing the combined ETD and CND line profiles as observed, the fact remains that the ETD and CND are different features, with potentially different physical properties. Unfortunately, a unique decomposition of the observed profiles is not straightforward, especially in case of the highJ transitions observed withHerschel-HIFI which are relatively noisy, and lack well-defined baselines. Yet, such a decomposition is required if we are to analyse the properties of the CND separately from its surroundings.

To gain a perspective on the CND, we have collected in Fig.4 position-velocity (pV) maps of the inner100′′ of NGC 5128 in the lower  ¹²CO and  ¹³CO transitions and in the HCNJ = 1–0 transition, in position angle PA = 125° anti-clockwise from north. This is along the heart line of the ETD; the more compact CND has PA≈ 145°. Besides the data shown above, we have usedJ = 1–0  ¹³CO data from Wild et al. (1997) andJ = 3–2  ¹²CO data from Liszt (2001). The ETD is the diagonal feature dominating the  ¹²CO maps, and the CND is the almost horizontal feature in the map center that is most obvious in the HCN map. The absorption lines close to the systemic velocity are (almost) saturated, and most prominent in the lower-frequencyJ = 1–0 HCN,  ¹²CO, and  ¹³CO maps. Map line intensities are in temperature units, in which the (subtracted) continuum emission and the associated absorption drops with increasing frequency.

The difference between the HCN map on the one hand, and the  ¹²CO and  ¹³CO maps on the other hand is striking. The ETD signature (diagonal feature) is very clear in  ¹²CO and  ¹³CO, but absent in HCN where only the CND signature (horizontal feature) is very clear. Towards the ETD. HCN intensities are at least six times weaker than towards the CND. This factor is a lower limit because of the strong absorption towards the nucleus affecting the measured line intensity. Because the critical density for excitation of the HCNJ = 1–0 line is of the order of10⁶      cm^-3, the HCN map implies that essentially all dense gas in the line of sight towards the Cen A nucleus is actually concentrated in the CND.

The high CO transitions, fromJ = 6–5 onwards, predominantly trace this same dense molecular gas. The COJ = 7–6 transition, for instance, has a critical densityn_crit    =    4 × 10⁵      cm^-3. This is comparable to that of the HCNJ = 1–0 emitting gas that we have shown to be limited to the CND. Indeed, the line profiles of the higherJ CO transitions in Fig.1 clearly show the signature of the rapidly rotating CND, and provide little or no evidence for a (narrow-line) contribution from the ETD. We have also noted in the preceding section that theJ = 7–6 CO emission is much more compact than theJ = 4–3 CO emission. We are therefore quite confident that from theJ = 7–6 transition onwards, the observed CO line emission is predominantly due to the CND.

The same conclusion cannot be drawn for the lower CO transitions fromJ = 1–0 toJ = 5–4, where the complex nature of the velocity-resolved profiles in Fig.1 clearly implies significant contributions from both the CND (broad plateau) and the ETD (narrower peak). We have estimated the relative contributions as follows.

Analysis of the maps in Fig.4 shows that the profile widths from the CO in the ETD result from a contribution caused by a change of rotation velocity across the beamΔ   V_rot = 2.9 ± 0.3      km   s^-1/arcsec (156 ± 16      km   s^-1/kpc) and an intrinsic contribution with a velocity FWHM of92 ± 3      km   s^-1 which is independent of the observing beam. This beam-independent contribution implies the existence of a significant line-of-sight velocity dispersion⟨ v_r ⟩  = 39 ± 1      km   s^-1 in the ETD.

This result allowed us to decompose the line profiles shown in Fig1 into Gaussian components corresponding to the maximum and the minimum CND contribution respectively. In both procedures the ETD contribution was represented by a Gaussian with a fixed velocity half-width corresponding to the relevant beam-size, a central velocity allowed to deviate from the systemic velocity (V_LSR = 540      km   s^-1) by at most 10  km   s^-1 to take into account small pointing errors, and leaving only the amplitude as a completely free parameter.

TheminimumCNDemission(CND_min) was found by fitting the observed profiles withtwoadditional Gaussians with central velocities fixed atV_LSR = 440 ± 10      km   s^-1 andV_LSR = 635±15      km   s^-1 respectively, and leaving both amplitude and velocity width as free parameters. The sum of the two CND Gaussians has a minimum around the systemic velocity, thus maximising the ETD contribution. In the CND_min decomposition, we find roughly equal amounts of flux for the CND and for the normalized ETD in all transitions (see Table7).

ThemaximumCNDemission(CND_max) was found by fitting the observed profiles with onlyoneadditional Gaussian with free parameters. This second CND Gaussian peaks at roughly the same (systemic) velocity as the first Gaussian representing the ETD contribution, which thus is minimised. Because the CND is not a completely filled disk (cf. Israel et al.1990), a single Gaussian component fitted to the wings of the observed profile will effectively overestimate the CND contribution. In this decomposition, the ETD contribution in the normalized beam is only a fifth to a fourth of the CND luminosity.

Table7 summarises the results for both decompositions (CND_min and CND_max) in the CO transitions up toJ = 6–5. The last two columns contain the mean of these two decompositions which we believe that best represents the actual situation. The CND:ETD flux/luminosity ratio in a22′′ (410 pc) beam is typically 2:1 in the lower transitions. Taking into account the beam filling factor of the CND, its mean line surface brightness exceeds that of the ETD by about a factor of four in these transitions.

4.4. Radiative transfer modelling

4.4.1. LVG modelling

Fig. 5 Results of the PDR/XDR model fitting to the NGC 5128 CO ladders.Left: fit to the CND CO ladder with three PDR models;center: fit to the CND CO ladder with two PDR models and one XDR model;right: fit to the ETD CO ladder with two PDR modelss.

We have modelled the observed  ¹²CO and  ¹³CO line intensities and ratios with the RADEX large velocity gradient (LVG) radiative transfer models (Jansen1995; Jansen et al.1994; Hogerheijde & van der Tak2000;http://www.strw.leidenuniv.nl/~michiel/ratran/). These codes provide model line intensities as a function of three input parameters per molecular gas phase: molecular gas kinetic temperatureT_k, densityn(H₂), and the CO velocity gradientN(CO)/dV. By comparing model to observed lineratios, we may identify the physical parameters best describing the actual conditions. We present both the result of modelling the observed integral CO ladder and those of the derived individual CND and ETD CO ladders.

In the modelling, we assume a constant CO isotopical abundance [  ¹²CO]/[  ¹³CO] = 40 throughout. This value is close to values found in various galaxy centers (Mauersberger & Henkel1993; Henkel et al.1993,1994,1998; Bayet et al.2004). We identify acceptable fits by searching a grid of model parameter combinations (T_k = 10–150 K,n(H₂) =10²–10⁵      cm^-3, andN(CO) /dV =6 × 10¹⁵–3 × 10¹⁸      cm^-2) for (combined) line ratios matching those observed. In the case of two-phase models, the relative contribution of the two components is treated as a free parameter. The physical gas properties undoubtedly encompass a wider range of temperatures and densities than suggested by even a two-phase model. However, lacking a physical model for the distribution of clouds and their sources of excitation, two phases is the maximum that can be considered fruitfully, especially as only the well-defined lowerJ CO transitions have  ¹³CO observations that allow us to break theT_k − n(H₂) degeneracy inherent to the  ¹²CO ladder.

We find that no combination of parameters from a single-phase gas provides a satisfactory fit to the observed line intensities. This is not unexpected in view of the profile complexity involving the two distinct contributions from ETD and CND. Consequently, we have also modelled the  ¹²CO and  ¹³CO simultaneously with two molecular gas phases. Although the available  ¹³CO measurements break much of the strong temperature-density degeneracy inherent to  ¹²CO measurements, we still find three different possible solutions.

The first set of solutions puts all gas at an elevated temperature ofT_kin = 150 K. In this set, two thirds of the emission comes from a low-density first phase (n(H₂) = 100      cm^-3), and one third from a moderate-density second phase (n(H₂) = 1000      cm^-3).

In the second set of possible solutions, the two phases have similar modest densities and elevated temperatures. They differ mainly in their CO velocity gradient (3 × 10¹⁶      cm^-2/   km   s^-1 versus4.5 × 10¹⁷      cm^-2/   km   s^-1, for the second phase). About80−90% of the emission arises from the first phase with kinetic temperatures of 100–150 K, and densities that could be as low as 100  cm^-3 or as high as 1000  cm^-3, but are most likely around 500  cm^-3. The remaining20−10% arises a slightly warmer, and slightly more tenuous second phase (T_kin ≈ 150 K,n(H₂) = 100−500   cm^-3).

The final set of solutions combines an again very similar first phase (T_kin = 100 K,n(H₂) = 500  cm^-3), with a much denser and colder second phase (T_kin = 20–30 K,n(H₂) = 10⁵   cm^-3).

We emphasize that the different physical environments represented by the solutions are observationally indistinguishable. Moreover, they are not mutually exclusive but may apply simultaneously and sample a multi-phase structure of the ISM: very dense, cold clumps embedded in a more tenuous warm molecular gas.

Thus, the LVG analysis of the integral line profile requires the presence of significant amounts of warm gas at temperatures of 100–150 K at moderate densities of typically500      cm^-3, but does not distinguish between the situation where this is all, where two thirds of the emission is from amore tenuouswarm gas (100      cm^-3), or where10−15% of the emission comes from amuch more dense, coldgas. As it turns out, only the latter case fulfils the additional constraints posed by considering the CND and the ETD separately.

When we fit LVG models to the individual CND and ETD components, once again no single phase model provides a reasonable fit to the observed CO line ladders. However, the CND CO ladder is well-fitted by two gas phase components in a narrow range, involving a dense (n_H2 = 10⁴      cm^-3) and a much more tenuous (n_H2 = 300 ± 200      cm^-3) gas. The properties of the dense gas are well-defined, with a low temperatureT_kin = 25 ± 5. The temperature of the tenuous gas temperature is not so tightly constrained. It may be atT_kin = 60−100 K, in which case it will be responsible for three quarters of the CO emission in theJ = 2–1 transition, but it may also be at a much lowerT_kin = 20−30 K in which case itsJ = 2–1 emission share will be about50−60%.

The physical parameters of the ETD gas are better determined and differ markedly from those of the CND gas. Most (65 ± 15%) of the emission in theJ = 2–1 transition is from a gas with a well-established high density (n_H2 = 10⁴      cm^-3) and low temperatureT_kin = 30 K. The remainder of the gas (responsible for35 ± 15% of the emission) has a very low density (n_H2 = 100      cm^-3) at an elevated temperatureT_kin = 100 ± 50 K.

4.4.2. PDR/XDR modelling

To further investigate the physical conditions and excitation mechanisms, we have also applied the PDR/XDR models by Meijerink & Spaans (2005) and Meijerink et al. (2007). In XDRs, the excitation is dominated by X-ray photo-ionization heating, i.e., the Coulomb interaction of keV electrons with thermal electrons. The heating efficiency of X-rays is of the order of 10 to 40 percent, much more efficient than the photo-electric heating (~0.3 to1.0 percent) in PDRs. The ionisation is driven by primary and secondary X-ray ionisations, and the ionisation fraction can be larger thanx_e ~ 0.1, three orders of magnitude higher than in PDRs. The high degree of ionisation is able to drive an active ion-molecule chemistry and makes it possible to maintain high abundance levels of molecules at high temperatures,T > 300 K. This expresses itself in much larger column densities of warm gas in XDRs, which betray themselves by much more intense higherJ CO transitions.

The PDR/XDR code requires three input parameters: a densityn, a surface area covering factor (in this case relative to that of component PDR2), and an incident UV flux in units of the one-dimensional Habing (1968) fieldG₀ (=1.6 × 10^-3 erg  cm^-2 s¹), or an X-ray energy fluxF_X (in units of  cm^-2 s¹), respectively. We constrained the PDR parameters to be fitted by using the relatively well-established gas volume densities supplied by the LVG modelling for the first two gas components. The resulting fits are shown in Fig.5. The boxes on the left and in the center show fits to the CO ladder of the CND using three PDRs, and two PDRs and one XDR, respectively. The box on the right e shows the fit to the CO ladder of the ETD in the same line of sight, which requires only two PDRs. Throughout, the major contributor to the low-J emission of both  ¹²CO and  ¹³CO (PDR1) is well-represented by the selected densityn = 300   cm^-3 and a moderate incident radiation fieldG = 300   G₀. It has a large surface filling factor compared to PDR2, the component dominating the mid-J transitions (3 ≤ J < 8). PDR2 requires a more intense incident radiation fieldG = 10^3.75−10^4.75   G₀ with higher densitiesn = 3 × 10⁴   cc for the CND andn = 10⁴   cc for the ETD. Its relatively small surface filling factor implies that the dense, strongly irradiated component exists mostly or entirely in clumped form. We note that the fits are not unique, and the ones presented here are merely those with the lowest chi-square values. Especially the radiation field is not well-constrained. The CO emission emitted by PDRs is not very sensitive to the radiation field, as it originates mostly from the UV-shielded part of the clouds.

The third component is a significant contributor only in the highJ transitions (J ≥ 8). Its existence is required by the CND CO ladder which reliably extends to the higherJ transitions (Sect. 4.3). This high-energy component represents either extreme PDR conditions (PDR3), or X-ray irradiation (XDR). The extreme PDR case must have very high densities of the order ofn = 10⁶      cm^-3 (consistent with the detection of HCN emission from the CND) and radiation fieldsG = 55   000   G₀. It has a slightly smaller filling factor than that of PDR2, and also has a three times lower total column density,N_H ~ 7.3 × 10^-21 cm^-2. On the other hand, if it represents an XDR, the density does not need to exceed that of PDR2 (n = 10^4.5      cm^-3), with an X-ray radiation field of${\mathit{F}}_{}\mathrm{=}\mathrm{9.0}$ erg cm² s¹ and a column density ofN_H = 3 × 10²² cm^-2. In that case, its surface filling factor is a few times than that of PDR2 (but still much less than that of PDR1). Equally good fits are, however, obtained for XDR phases with column densities ranging fromN_H = 10²² to10²³ cm^-2, and concomitant X-ray fluxes ofF_X of 5 and 16 erg cm^-2 s^-1. respectively. As Fig.5 shows, the observational data do not significantly constrain the CND physics because both PDR3 and (a variety of) XDR models provide comparably good fits to the CO ladder. Observationally, the appropriate models can only be identified by CO line intensities at e.g. theJ = 15–14 transition and beyond which are presently lacking.

4.5. Beam-averaged molecular gas properties

In order to derive overall molecular gas column densities from the LVG data, we assume that only about a quarter of all carbon is in the gas-phase (δ_C = 0.27), the remainder being tied down in dust grains. In a forthcoming paper, we find that the ISM of both the CND and the ETD is characterized by a metalicity of about 0.7–0.8 times that of the Solar Neighbourhood. We assume a [C]/[H] elemental abundancex_c = 1.6 × 10^-4 which leads us to expect expect a neutral gas-phaseN_C/N_H = 4.5 × 10^-5. The chemical models by van Dishoeck & Black (1988), updated by Visser et al. (2009) show a strong dependence of the ratio of atomic carbon to molecular carbon monoxide column densities on the total carbonN_C = N(C) + N(CO) and molecular hydrogenN(H₂) column densities. Thus, each value ofN(H₂) is associated with a unique value ofN(C)/N(CO)and a unique value ofN_C = N(C) + N(CO) which can be derived from the models.

From the LVG analysis, we find very similar beam-averaged column densities for the CND (N(CO) = 0.34 × 10¹⁸) and for the ETD (N(CO) = 0.29 × 10¹⁸  cm^-2). We also find that in both the CND and the ETD about two to three times more carbon is in atomic than in molecular form, and we obtain a beam-averaged total hydrogen columnN_H = (3.4 ± 0.1) × 10²²      cm^-2 for the CND, andN_H = (1.0 ± 0.1) × 10²²      cm^-2 for the ETD. Taking into account that the ETD fills all of the normalized beam, and the CND only half of it, we derive a four times higher molecular gas surface-filling factor (0.10 ± 0.01) for the CND.

The molecular gas column density is found by subtracting the neutral hydrogen HI from the total hydrogen column density. Measurements of the HIemissionsuggest its contribution is quite small in the central few hundred parsecs (see Struve et al.2010). On the other hand, HIabsorptionline measurements imply very high HI column densities (typically10²²−10²³      cm^-2), but these apply to a pencil-beam area of order10⁵ smaller and are related to material much less extended than the scale of the CND. Neglecting a possible (but probably small) contribution by atomic hydrogen, we find for the CND a beam-averaged molecular hydrogen column densityN(H₂) = (1.7 ± 0.1) × 10²²      cm^-2. The corresponding mass isM_gas = 4.7 ± 0.5   10⁷   M_⊙, including a35% helium contribution. In the ETD the HI column may not be negligible. ForN(HI) ≤ 0.15 × 10²²      cm^-2 (cf. Struve et al.2010) we obtain for the ETDN(H₂) = 0.9 × 10²²      cm^-2.

The PDR model analysis produces higher gas masses. The pure PDR-model yields a total gas massM_gas = 1.5 × 10⁸   M_⊙ which includes a35% contribution of helium. A fraction of 0.95 of this mass resides in the lowest-excitation phase (PDR1), whereas the densest, highest-excitation gas (PDR3) mass is of the order of a per cent of the total. If we assume that the highest observed COJ transitions come from gas excited by X-rays rather than by UV-photons, the mass of this phase can be substantially higher. For instance, if we assume an XDR density identical the PDR2 gas, we obtain a (helium-corrected) gas massM_gas = 1.7 × 10⁸   M_⊙, of which15% is in the X-ray irradiated phase. Thistotalgas mass is only slightly higher than that of the pure PDR case. In the fraction of the ETD measured by our beam, the mass density is no more than a quarter of that of the CND, and only15% is in a denser, more excited phase (PDR2). As shown in Fig.5, there is no third phase of very dense gas highly excited by X-rays or their equivalent. In the pure PDR case, the mass of dense gas in the CND is not much higher than that in the ETD-beam. However, in the XDR case the mass of dense gas is ten times higher in the CND than in the ETD-beam. This is in better agreement with the observation that the HCN emission (Table3 and Fig.4) traces dense gas almost exclusively in the CND.

In the PDR/XDR models the atomic gas mass fraction ranges between one-third to half of the total gas mass. This is because quite large radiation fields are needed to obtain PDR excitation conditions that are able to reproduce the observed line fluxes. This can, however, be mitigated by including a small amount of mechanical heating (cf. Kazandjian et al.2012,2013). When sloshing motions of the gas are responsible for part of the total heating, especially in the regions where the gas is shielded from UV radiation, the incident UV fluxes can be substantially lowered (yielding a smaller atomic hydrogen fraction), while obtaining the same CO model fluxes.

The physical properties of the NGC 5128/Centaurus A circumnuclear disk are summarised in Table9. For the best estimate of the CND mass we use the geometric mean of the LVG and PDR results. This brings the mass toM_gas = 8.4 × 10⁷   M_⊙ with a correspondingN(H₂)-to-${\mathit{I}}_{\hspace{0.17em}{\mathrm{12}}^{}\mathrm{CO}}$ ratioX =4 × 10²⁰  cm^-2/   K   km   s^-1, about twice the “standard” Milky Way ratio. We cannot determine a useful mass value for the small fraction of the ETD included in our beams, but from our analysis we estimate anX-value not very different, and possibly a bit closer to that of the Milky Way. This is the first time the CND gas mass has been determined rather than guessed. For instance, a much lower CND mass was given by Espada et al. (2009), but this was based entirely on an assumed very lowX-factor of 0.4X_GAL. This appears to be appropriate for “normal” AGNs, but does not adequately describe the Cen A situation.

In Table9 we also compare the integrated CO luminosity (using the models to extrapolate the observed CO line luminosities up to theJ = 15–14 transition) with the observed neutral and ionized carbon luminosities. In case of significant X-ray excitation, the total CO luminosity may be somewhat higher than given here, because of the contribution of the very high-J transitions. Nevertheless, from the comparison in Table9 we may conclude that neutral carbon cooling of the CND is at most half of that provided by carbon monoxide. In contrast, the cooling by ionized carbon is about an order of magnitude higher than the CO cooling.

5. Conclusions

• 1.

  We have measured  ¹²CO fluxes from the central region of NGC 5128 (Centaurus A) in transitions up toJ = 12–11, as well as the fluxes from the submillimeter [CI] and [CII] lines. In addition, we have presented high-S/N velocity-resolved profiles of  ¹²CO emission in theJ = 1–0 throughJ = 5–4 and of  ¹³CO emission in theJ = 1–0 throughJ = 3–2 transitions.

• 2.

    ¹²CO line luminosities, normalized to a22′′ beam (410 pc) increase up toJ = 5–4, and then steadily drop to low values. This drop is more pronounced than that seen in Seyfert AGNs, and much more pronounced than seen in star-burst galaxies.

• 3.

  Both [CI] lines are more luminous than the adjacent (J = 4–3, andJ = 7–6)  ¹²CO lines. This behaviour is not seen in other (star-burst or AGN) galaxies, and is thus unique to the Cen A center. The [CI] line intensities and their ratio to CO line intensities are inconsistent with standard PDR model predictions.

• 4.

  Detailed analysis shows that up toJ = 6–5, about a third of the  ¹²CO flux observed in the normalised22′′ beam is contributed by the ETD embedded in NGC 5128, and unrelated to the compact CND just contained within the normalised beam. The CND contains a high proportion of very dense gas, in contrast to the ETD. At transitions aboveJ = 6–5, the  ¹²CO emission is completely dominated by the CND contribution.

• 5.

  We have decomposed the observed CO spectral line ladder into individual (total) CND and (representative) ETD CO ladders. The ETD ladder peaks in theJ = 4–3/J = 5–4 transition and then drops rapidly. The CND ladder peaks in theJ = 6–5 andJ = 7–6 transitions before dropping more slowly.

• 6.

  LVG and PDR/XDR model analysis of the CND  ¹²CO fluxes shows that most of the molecular gas mass resides in a relatively cool (T_kin = 25–80 K), not very dense (n_H2 ≈ 300      cm^-3) gas phase if the CND is heated exclusively by UV photons (PDR).

• 7.

  A small fraction of the gas in the CND is more highly excited and has much higher densities (typically3 × 10⁴      cm^-3).

• 8.

  In the CND but not in the ETD a third, more highly excited, high-density phase must also be present, either in the form of an extreme PDR or in the form of an XDR.

• 9.

  The CND has a total gas massM_CND = 8.4 × 10⁷   M_⊙ (uncertain by a factor of two) which is about10% of the mass of the much larger ETD.

• 10.

  The CO-H₂ conversion factorX_CND is4 × 10²⁰      K   km   s^-1 (uncertain by a factor of two), about twice the local Milky Way factorX_SN.

Acknowledgments

We thank Markus Schmalzl for help with the ALMA SV data. We also thank all facility observers, engineers, and support scientists who by their often anonymous efforts have made possible the collection of data presented in this paper.

References

All Tables

All Figures

Fig. 1 Baseline-subtracted line profiles towards the center of NGC 5128 (Centaurus A) obtained with the ground-based telescopes SEST, JCMT, and APEX (top two rows) and with the HIFI instrument on-board of theHerschelSpace Observatory (bottom two rows). Vertical scale is${\mathit{T}}_{\mathrm{A}}^{\mathrm{\ast }}$ in Kelvin, horizontal scale isVLSR in  km   s-1. Species, transition and telescope used are identified at the top left corner of each panel. The HIFI  12CO results for theJ = 8–7,J = 10–9, andJ = 13–12 transitions are the averages of the profile towards the nucleus and the adjacent positions separated from that by10′′. Note that the deep absorption apparent in the13CO (1–0) profile is real; its depth to negative values reflects the weakness of the line emission relative to the (subtracted) continuum strength at the same frequency. For more details, see Sect. 2 and Tables2 and3.

In the text

Fig. 2 Baseline-subtracted line profiles towards the center of NGC 5128 (Centaurus A) obtained with the APEX telescope and the HIFI instrument on-board theHerschelSpace Observatory. The HIFI [CI] and [CII] profiles for the nuclear position, and the positions offset by± 10′′ from the nucleus, are shown separately. The NW offset profile is represented by a dashed line, the SE offset profile by a solid line. Species, transition and telescope used are identified at the top left corner of each panel. Vertical scale is${\mathit{T}}_{\mathrm{A}}^{\mathrm{\ast }}$ in Kelvin, horizontal scale isVLSR in  km   s-1. For more details, see Table4.

In the text

Fig. 3 Left: full submillimeter spectrum of the center of NGC 5128 (Centaurus A) obtained with the SPIRE instrument on board theHerschelSpace Observatory (See Sect. 2). Species and transitions are identified throughout. The jump in the continuum at 944 GHz is caused by the different angular resolutions of the SLW and the SSW (see Sect. 2). The SPIRE continuum contains a contribution from the point source nucleus of about 8.2 Jy at 460 GHz, slowly decreasing with frequency. The remaining continuum is due to extended thermal emission from dust, increasing with frequency. Vertical scale is flux in Jansky, horizontal scale is frequency in GHz. For more details, see Tables2 and3.Right: comparison of SPIRE CO line fluxes of the NGC 5128 center with those of the starburst galaxy M 82, the AGN+starburst NGC 1068, and the (U)LIRGs Arp 193, Arp 220, and NGC 6240. Mrk 231 has not been marked separately as its spectral ladder is identical to that of NGC 6240. The ratio of the galaxy CO line flux to the Cen A CO line flux is shown for eachJ transition, as observed by SPIRE without correction for finite beam-size, source extent, and beam efficiency. The vertical line separates measurements obtained with the SSW from those obtained with the SLW in a beam roughly twice as wide.

In the text

Fig. 4 Position-velocity maps of molecular line emission from the central region of Centaurus A, in position angle 125° counter-clockwise from north (data summarised in Tables2 and3). Horizontal scales are velocityV(LSR) in  km   s-1, vertical scales are offsets from the nucleus in arcsec. Thepanelson theleftshow emission from  12CO; theJ = 1–0,J = 2–1, andJ = 3–2 transitions are shown fromtop to bottom, respectively. Thepanelson therightshow emission from  13CO (J = 1–0 atthe top,J = 2–1 inthe middle). TheJ = 1–0 HCN transition is atbottom right. TheJ = 1–0 maps have resolutions of45′′−55′′, all other maps have an effective resolution of23′′. In all three  12CO maps, the contours are at multiples of 50 mK in main-beam brightness temperature. In the  13CO maps contours are at multiples of 5 mK (J = 1–0) and 10 mK (J = 2–1). The HCN map contours are at multiples of 5 mK. Strong absorption is clear in all panels nearVLSR = 550     km   s-1. Emission from the rapidly rotating compact nuclear disk becomes progressively more clear with increasingJ-transition in the  12COpanels, and in the HCN map.

In the text

	Fig. 5 Results of the PDR/XDR model fitting to the NGC 5128 CO ladders.Left: fit to the CND CO ladder with three PDR models;center: fit to the CND CO ladder with two PDR models and one XDR model;right: fit to the ETD CO ladder with two PDR modelss.
In the text

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.