Mid-Infrared Interferometry of Active Galactic Nuclei: an Outstanding Scientiﬁc Success of the VLTI

Astronomical Science

Mid-infrared Interferometry of Active Galactic Nuclei: an Outstanding Scientiﬁc Success of the VLTI

Klaus Meisenheimer1 and below the torus are visible. The ob- time (Leinert et al., 2003). The sensitivity David Raban2 ject then appears as a Seyfert 2 galaxy. required for most AGN observations 1, 3 Konrad Tristram Spectropolarimetric observations of (correlated flux Fcorr ≤ 1 Jy in the N-band) Marc Schartmann1,4,5 Seyfert 2 galaxies, showing broad lines in can only be reached by the combination Walter Jaffe 2 scattered light, support this idea (see of two Unit Telescopes (UTs) of the VLTI. Huub Röttgering 2 review by Antonucci, 1993). The UV-opti- The highest sensitivity for detecting Leonard Burtscher1 cal light which is trapped by dust in the and tracking the interferometric fringes is torus should heat the dust to a few hun- obtained by inserting a prism into the dred Kelvin, and the dust should re-radi- interferometric beams, that spectrally dis- 1 Max-Planck-Institut für Astronomie, ate in the mid-infrared. Indeed, the Spec- perses the N-band light with a spectral Heidelberg, Germany tral Energy Distributions (SEDs) of both resolution ,25. For brighter objects a 2 Sterrewacht Leiden, the Netherlands Seyfert 1 and Seyfert 2 galaxies display grism with a higher resolution of 250 can 3 Max-Planck-Institut für Radioastro- signatures of AGN heated dust between be used. In both cases MIDI delivers nomie, Bonn, Germany ,3L and 30 µm. It is an open issue two spectra (phase-shifted by 180 de- 4 Max-Planck-Institut für Extraterres- whether dust obscuration plays a similar grees) onto its detector, containing spec- trische Physik, Garching, Germany role in radio galaxies. tral and interferometric information at 5 Universitäts-Sternwarte, München, the same time. A special analysis pipeline Germany Before the advent of the VLT Interferom- is needed to extract this information. We eter (VLTI), the size, shape and internal use the Expert Work Station (EWS) pipe- structure of the torus remained unknown, line developed in Leiden by Walter Jaffe. Active Galactic Nuclei (AGN) are pow- although the mid-infrared spectra located ered by accretion onto a supermassive the dust within a few parsec of the core. Observations of the scientific target have black hole. The unified scheme for Single 8-m-class telescopes cannot re- to be complemented by standard star strongly accreting AGN postulates that solve mid-infrared structures of this size. observations obtained with an identical the central engine is enshrouded by a Even in the L-band (3.6 µm) a diffrac tion- instrumental set-up. The essential result of doughnut-shaped structure of gas and limited 8-m telescope is limited to 93 mas the pipeline analysis is a spectrum of the dust – the so-called torus. We report resolution (Full Width at Half Maximum, (calibrated) correlated flux Fcorr(L ) in the observations with the MID-Infrared FWHM). At the distance of nearby range 7.8 to 13.2 µm (see Figure 1b, c, d).

Interferometric Instrument (MIDI) at the Seyfert galaxies, such as NGC 1068 and Fcorr(L ) corresponds to the Fourier Trans- VLT Interferometer, which resolve the NGC 4151 (14 Mpc), this corresponds to form of the source emission evaluated tori in the nearest Seyfert 2 galaxies, 6.5 parsec. at a coordinate (called ‘uv-point’ or ‘base- and suggest a complex structure, con- line’) given by the projected separation sisting of a compact inner disc embed- The situation changed dramatically between the telescopes as viewed from ded in a patchy or filamentary outer in December 2002, when MIDI, the MID- the source. Spatial information about the torus. The prominent nearby radio gal- Infrared Interferometric Instrument, source structure can be obtained com- axy Centaurus A, however, shows little became operational at the VLTI. MIDI paring Fcorr(L ) at different uv-points. To sign of a torus. Instead, its mid-infrared observes in the N-band (wavelengths 8 to the actually measured uv-points can be emission is dominated by non-thermal 13 µm). Using the widest telescope sepa- added the total flux Ftot(L ) registered by radiation from the base of the radio jet. ration (UT1–UT4) of 125 m, the width of a single telescope, essentially equivalent Thus, not all classes of AGN contain a the point-spread function at 8 µm is only to an observation with zero baseline (see thick torus. 7 mas, or 0.5 parsec at the distance of Figure 1a). Different baselines can either NGC 1068. But at the start of MIDI’s op- be realised by using different telescope erations two major questions remained: combinations or by observing the target The unified scheme for Active Galactic first, would MIDI be sensitive enough during its movement across the sky with Nuclei (AGN) explains various types of to reach extragalactic targets? Second, a fixed telescope combination. AGN by a line-of-sight effect: it postulates would observations with a handful of that the central engine – an accreting baselines allow us to reconstruct the dust As evident from Figure 1, the AGN spec- supermassive black hole – is embedded distribution in the torus and thus provide tra between 8.5 and 12.5 µm are often in a doughnut-shaped torus of gas and scientifically meaningful insights? This dominated by a broad absorption trough, dust. Thus, the hot accretion disc and the article demonstrates that today both caused by silicate dust grains. The exact surrounding Broad Line Region (BLR) is questions can be answered unequivo- profile of this ‘silicate feature’ depends only visible when looking along the torus cally: yes! on the chemical composition, size and axis. This is the case in Seyfert 1 galaxies, crystalline structure of the grains. Thus the optical spectra of which are charac- the N-band interferometry of an AGN not terised by a blue continuum and broad Mid-infrared interferometry with MIDI only resolves the spatial structure of the emission lines. In an edge-on case, how- nuclear dust but also can give insight ever, the direct view onto the core is MIDI operates as classical stellar interfer- into the dust properties within the inner blocked by the dusty torus and only nar- ometer of the Michelson type. It com- few parsecs. row emission lines from regions above bines the beams of two telescopes at a

36 The Messenger 133 – September 2008 25 of 14.4 Mpc, one parsec corresponds to is about 3.5 pc, but its exact shape re- (a) an angular scale of 14 mas, i.e. parsec- mains to be determined by shorter base- 1 20 scale structures can just be resolved with lines along the East-West direction. MIDI at the VLTI.

) y

J The major axis of the hot component is ( 15

u The earliest MIDI observations of perfectly aligned with a spur of water l F

l NGC 1068 were obtained half a year after masers extending about 20 mas towards a

t 10 o

T MIDI became operational, during VLTI NW from the (radio-)core, although the Science Demonstration Time (SDT). Jaffe relative astrometric position cannot be 5 et al. (2004) demonstrated for the first determined. Surprisingly, the orientation time that a compact, geometrically thick of its minor axis (P. A. = 48˚), which might dust structure – as expected for the dust mark the symmetry axis of an inclined 8 9 10 11 12 13 Wavelength (µm) torus – indeed exists in Seyfert 2 galax- disc, does not fit well to the source axis ies. Essentially only two visibility points as determined from outflow phenome- 6 were observed at that time. The corre- na. The inner radio jet points almost ex- 5 Ozone lated fluxes were best modelled by two actly North (P. A. = 2˚), while the ionisa- components, a small, relatively hot one tion cone opens between P. A. , –5 – 4 (T > 800 K, diameter about 1 pc), embed- –30˚. For the standard torus scenario this 3 ded in a larger component of 320 K and is a puzzle: the open funnel which allows 2 about 3.5 pc diameter. the ionising UV-photons to escape should be caused by the angular momentum 1 (b) New observations with MIDI (Raban et al., barrier and thus be aligned with the rota-

) 2008b) cover the uv-plane much better: tion axis of the gas distribution. How y J ( 2.5 15 visibility points were obtained with the could a tilted disc form out of this gas?

u tele scope combinations UT1–UT3, Perhaps the hot inner component is l 2 F UT1–UT4, and UT2–UT3. An additional not a rotationally supported structure d 1.5 e

t measurement with the orthogonal base- (disc) but rather a filament or hot channel. a l 1 e line UT3–UT4 proved essential for the r r o 0.5 (c) following results. To study the details of Further insight into the dust properties C the silicate absorption profile, the higher can be inferred from the depth of the sili- (R ,230) resolution grism was used. cate feature. In the total flux, which is 2.5 2 Even with this more complete uv-cover- 1.5 age, direct image reconstruction is not possible because MIDI observes only two 1 telescopes at a time and rapid atmos- 0.5 ((dd) pheric phase shifts cannot be recovered 0 by phase closure techniques. The meas- 8 10 12 ured Fcorr(L ) spectra for different baselines Wavelength (µm) still have to be interpreted by simple models. Remarkably, a model of two compo- 40 –40 Figure 1. Results of MIDI observations of NGC 1068. nents with Gaussian brightness distribu- (mas) (a) Total flux F ( ): the contribution of the hot com- tot L tion and black-body spectrum describes ponent is shown in red, that of the extended compo- the correlated flux data reasonably well. nent in blue. (b) Correlated flux Fcorr(L ) obtained with a 40-m baseline orientated along position angle With the inclusion of the longest VLTI P. A. = 36˚. The red dotted line gives the model fit baselines UT1–UT3 and UT1–UT4, the and the blue shaded area shows the contribution of measurements perfectly constrain the 1 pc the extended component. (c) F (L ) for 52 m base- –40 corr size, shape and orientation of the hot, line along P. A. = 112˚. (d) Fcorr(L ) for 97 m baseline along P. A. = 36˚. The comparison between (b) and inner component of the dust torus: major (c) shows that the hot component is more extended axis 20 mas (1.4 pc FWHM), oriented Figure 2. Observational model of the dust torus in (better resolved) in SE–NW direction. along P. A. = 138˚. It is rather elongated, NGC 1068. A hot component (yellow) is embedded in an extended cooler component (brown). The with an axis ratio of 0.25, indicating a orientation of the radio axis is indicated by a purple geometrically thin (disc-like?) structure. dotted line and the blue wedge gives the open- The dust torus in NGC 1068 Only a lower limit, T > 800 K, can be set ing angle of the ionisation cone, observed on 100-pc to its temperature. The lack of short scales. The first AGN observed with MIDI was the baselines, < 50 m in the East-West direc- 1 prototypical Seyfert 2 galaxy NGC 1068. tion makes the determination of the over- Such baselines are provided by the Auxiliary Tele- scopes (ATs). A MIDI observation programme It is the brightest extragalactic N-band all size and shape of the more extended with the ATs is under way and has already detected source in the southern sky. At its distance ‘torus component’ uncertain. Its diameter fringes from NGC 1068.

The Messenger 133 – September 2008 37 Astronomical Science Meisenheimer K. et al., Mid-infrared Interferometry of Active Galactic Nuclei

dominated by the outer component (Fig- 200 ure 1a), the absorption optical depth at 10 µm is moderate, ,0.4, whereas the depth towards the inner component is al- 100 most ﬁve times larger at ,1.9 (Figure 1b, ) s

c, d). Obviously, most of the dust column a m is located in the outer component. ( 0 C E D

The dust torus in the Circinus galaxy –100

The Circinus galaxy at a distance of 4 Mpc is the closest Seyfert galaxy. It –200 shows all signs of a classical Seyfert 2: 200100 0–100 200 1000–100 –200 narrow allowed and forbidden emis- RA (mas) RA (mas) sion lines, strong silicate absorption and a heavily absorbed X-ray spectrum. An Figure 3. The dust torus in the Circinus galaxy. The component would also have to be inter- extended cone of emission line gas and left panel shows the smooth model (composed preted as a rotationally supported disc. of two Gaussian brightness distributions, the right the presence of broad lines in the po- panel visualises the best-fitting patchy model. larised optical flux (caused by scattering) Dashed lines indicate the opening angle of the ioni- It is worth noticing that the depth of the provide direct evidence that the cen- sation cone. silicate absorption towards the dust com- tral engine is hidden from our direct view ponents in Circinus shows a different be- behind a substantial amount of dust. haviour than that observed in NGC 1068. Circinus is a spiral galaxy seen almost ness with that of a black body leads to a In Circinus the depth of the silicate fea- edge on. Thus, several magnitudes covering factor of only 20 %. Moreover, ture towards the inner component is shal- of visual extinction might be caused by the observed correlated flux values are lower than towards the outer component. the dust lanes in the spiral disc, behind poorly reproduced by the smooth Gaus- Obviously the dust column through the which the nuclear region is located. sian model, but rather seem to ‘wiggle’ outer dust component is not very high around it when plotting them as function and the absorption trough is partly filled The high southern declination of Circinus of baseline orientation. In order to test by silicate emission from the dust disc. (D = –65˚) makes it an almost ideal target whether a patchy brightness distribution for the VLTI: it can be observed for up to could improve the fit, we modified the 12 hours during long winter nights, thus smooth Gaussian distribution by a fore- The radio galaxy Centaurus A allowing the projected baseline orienta- ground screen of randomly distributed tion between each of two UTs to swing variations in transmission. A thousand The radio galaxy Centaurus A (= NGC by up to 180˚ due to the earth’s rotation. different screens were realised, the im- 5128) plays a key role in extra galactic In five observing runs during the MIDI ages were Fourier-transformed, and com- astronomy: at a distance of only 3.8 Mpc guaranteed time observation programme, pared with the observed correlated it is the closest large elliptical galaxy, the we have been able to collect 21 visibility fluxes. Indeed, we found several patchy closest galaxy merger and the closest points, most of them with the shortest screens which reproduce the observa- VLTI baselines UT2–UT3 and UT3–UT4 tions much better than the smooth mod- 60 (Tristram et al., 2007). They provide the el. The best-fitting model is displayed most complete uv-coverage obtained for in the right panel of Figure 3. Interestingly )

s 40 any extragalactic target so far. enough, it shows a bright patch on the a m (

axis of the ionisation cone. We regard this t e s

As in NGC 1068, at least two compo- as evidence that our interferometric data f f 20 O nents with Gaussian brightness distribu- contain hints for the existence of hotter h t tion are required to model the correlated dust close to an open funnel which con- u o S

ﬂuxes: a compact component (major ﬁnes the ionising radiation. -

h 0 t axis 0.4 pc and axis ratio 0.2); and an r o almost round extended component The size and orientation of the inner, N (FWHM 1.9 pc, see Figure 3). Contrary to disc-like component again ﬁt very well to –20 the case of NGC 1068, the colour tem- the known disc of water masers which 200 300 400 500600 700 peratures of the inner and outer compo- show a Keplerian rotation pattern (Fig- 60 40 20 0 –20 nents both lie around 300 K, differing by ure 4). Although the location of the dust East-West Offset (mas) less than 50 K. However, the outer com- emission with respect to the maser disc ponent does not seem to be smoothly cannot be determined by our MIDI ob - Figure 4. Overlay of the compact dust component ﬁlled with dust at a constant temperature. servations, it is very likely that both discs in Circinus over the location of the (warped) disc of Comparing its average surface bright- are co-spatial: in this case the inner dust water masers (from Gallimore et al., 2004).

38 The Messenger 133 – September 2008 violent AGN. At its distance, 1 pc corre- N the torus include the widely spread and Baseline sponds to 53 mas. The radio source can orientation continuous distribution in X-ray absorbing be traced over seven orders of magni- hydrogen column densities between 50 tude in angular scales, from the VLBI jets ˚ Seyfert 1 and Seyfert 2 galaxies and sev- J

e 4 t T

(a few mas) to the outer lobes (several a eral cases in which an AGN changed its

x U

is –

degrees). Extinction in the dust lane 3 broad line spectrum, indicating a change T

of the merging spiral galaxy severely ob- U in central obscuration. Radiative trans- scures our view towards the nucleus E fer calculations of ‘clumpy’ torus models of Centaurus A. Thus observations at showed that another problem of the infrared wavelengths are mandatory (see continuous torus models – namely their

Meisen heimer et al., 2007, and references ˚ prediction of a strong silicate emission 9 therein for more details). ˚ ± in Seyfert 1 galaxies, which is rarely ob- 7 3 12 T –U served – can be solved by shadowing T2 Centaurus A was observed in 2005 with 10 mas U effects in a clumpy structure (Nenkova et MIDI using two telescope combinations: 0.19 pc al., 2002). In a recent study we demon- UT3–UT4 and UT2–UT3. With both strate by fully 3D radiative transfer calcu- combinations two visibility points were Figure 5. Sketch of our model for the N-band emis- lations (Schartmann et al., 2008) that a obtained, separated by about two hours. sion from the central parsec of Centaurus A. An wide variety of cloud distributions is able unresolved point source is surrounded by a faint The projected baseline with UT3–UT4 dust disc. to reproduce the observed mid-infrared was orientated roughly perpendicular to spectra. Moreover, when simulating inter- the parsec scale radio jet, while UT2–UT3 ferometric observations of such a clumpy was aligned with it (Figure 5). We found this ‘synchrotron core’ as the base of the torus, we find similar ‘wiggles’ in the that the mid-infrared emission is margin- radio jet (for details see Meisenheimer et correlated fluxes to those observed in ally resolved perpendicular to the jet axis al., 2007). Our interferometric results on Circinus. with a 60-m projected baseline, whereas Centaurus A demonstrate that mid-infra- it remains unresolved along the jet axis. red radiation processes are not restricted Despite the success of radiative transfer Accordingly, we conclude that the 8 to to thermal dust emission. models in explaining the infrared SEDs of 13 µm emission from the core of Centau- AGN, they cannot solve the stability prob- rus A is dominated by an unresolved The thermal dust emission from the core lem pointed out by Krolik & Begelman: point source (FWHM < 6 mas), which of Centaurus A is very feeble, more than how could the geometrically thick distri- contributes between 50 % and 80 % of 20 times weaker than that of the Circinus bution of clouds be maintained? To the total flux at 13 µm and 8 µm, respec- galaxy at the same distance. We think address this question a hydrodynamical tively. The extended component is tiny that both a lack of dust in the inner par- model is required that simulates a real- (FWHM , 30 mas), and seems elongated sec and the absence of a sufficiently istic mass injection into the torus and fol- perpendicular to the radio axis (see strong heating source are responsible for lows the evolution of the gas clouds. We sketch in Figure 5). However, a better uv- this. Certainly, Centaurus A neither con- are currently developing a torus model coverage (including longer baselines) tains a torus which severely blocks our for Seyfert galaxies that starts from a will be required to constrain the size, line of sight nor a UV-optically bright number of assumptions. The centre of the shape and orientation of this extended central accretion disc. Most likely, the galaxy harbours a massive young stellar component more accurately. We interpret accretion onto its black hole happens via cluster (age between 40 and 100 million the extended component as dust emis- an advection dominated accretion flow years). Stellar mass loss via planetary sion from a small, inclined disc (diameter (ADAF), which is very inefficient in con- nebulae and stellar winds injects gas and about 0.6 pc). The unresolved component verting accretion power m˙ c 2 i n t o radiation. dust into the system, while frequent su- is identified with the non-thermal ‘syn- pernova explosions stir up the gas. Lo- chrotron core’ of Centaurus A, since we cally the gas gets compressed and the find that – after correcting for the fore- Models of the torus subsequent cooling instability leads to ground extinction of AV = 14 mag (derived the formation of dense and cool filaments from the depth of the silicate absorp- The concept of a doughnut-shaped (see Figure 6). In between the filaments tion) – its flux level and spectrum lies per- ‘torus’, continuously filled with gas and cavities of very hot plasma form over- fectly on the extrapolation of the power- dust, is an oversimplified geometrical pressured regions, which expand radially law spectrum observed at millimetre picture. Already 20 years ago, Krolik & along the density gradient. Thus the cool wavelengths. Together with photometry at Begelman (1988) pointed out that it must filaments also become radially stretched. shorter wavelengths (from HST and the consist of a large number of individual The cool gas and dust streams inwards AO camera NACO at the VLTI) the flux of clouds orbiting around the AGN core. along the filaments and accumulates in a the unresolved point source fits perfectly However, frequent cloud–cloud collisions very dense turbulent disc with a few par- to a canonical synchrotron spectrum: it would make such a system very unstable: sec radius. is characterised by a rather flat power-law within a few orbital timescales it should –0.36 Fv t v , cutting off exponentially at a settle into a geometrically thin disc. Other In a second step, the radiative transfer 13 frequency vc = 8 × 10 Hz. We interpret arguments for a clumpy sub-structure of through the simulated density distribution

The Messenger 133 – September 2008 39 Astronomical Science Meisenheimer K. et al., Mid-infrared Interferometry of Active Galactic Nuclei

–24.0 –22.0 –20.0 2.04.0 6.08.0 –40 –200 20 40

LOG (g/cm3) LOG T (K) 40 40 40 40

30 30

20 20 20 20

10 10 c) p

( 0 0 0 0

z –10 –10

–20 –20 –20 –20

–30 –30

–40 –40 –40 –40

0 10 20 30 40 01020 30 40 –40–20 02040 r (pc) r (pc)

Figure 6. Hydrodynamical torus model. The left and N-band to be observed with MIDI. The But seen from the VLT, the uv-coverage middle panels show the gas density and tempera- preliminary target list was selected from of this Seyfert 1 galaxy will always re- ture in a meridional slice. The right panel displays the image at 12 µm which would be observed from an AGN with known N-band flux > 1 Jy. main very limited. The closest southern edge-on view onto this torus. The simulations refer to Since most of the available N-band pho- Seyfert 1 galaxy which is bright enough an AGN that is about five times more luminous than tometry was obtained in large aper- for MIDI observations, NGC 3783, is three NGC 1068. tures, it was necessary to observe all tar- times more distant than NGC 4151. In gets with TIMMI2 at the 3.6-m telescope order to obtain a direct comparison, more (beam size 0.7) to get the core flux at distant (and luminous) Seyfert 2 galaxies is calculated (assuming a standard gas- L = 12 µm. The final target list (Table 1) have to be studied as well. to-dust ratio in all cells below sublimation contains all southern AGN with SN (core) temperature). The emerging mid-infrared > 300 mJy (Raban et al., 2008). 13 of images (right panel in Figure 6) reproduce the targets have been observed during Synthesis the filamentary density structure. They the snapshot survey, two more were can explain the ‘patchy’ outer torus ob- tried by other observers. From 11 of these At the first sight, our results for the dust served in Circinus rather nicely. It should 15 targets, MIDI could detect interfero- structures in the Seyfert 2 galaxies be noted, however, that the central tur- metric fringes (column 6 in Table 1). Three NGC 1068 and Circinus look quite similar: bulent dust disc appears dark in our sim- of the sources, for which MIDI observa- they both contain an elongated inner ulations. A set of torus models is gener- tions were attempted, could not be ob- component which is embedded in a ated by varying the mass injection and served since their nuclei were too faint for larger dust distribution, heated to about supernova rates. Observing those under the adaptive optics system MACAO. Only 300 K. The observed difference in torus various aspect angles can well account one source, the star burst nucleus in size is expected from the fact that for the wide spread in hydrogen column NGC 253, seems too extended to show NGC 1068 is about 10 times more lumi- in Seyfert galaxies (over three orders of an interferometric signal. nous than Circinus. In both sources magnitude) while the change in silicate the inner component is aligned with the depth (from absorption to moderate emis- Most targets have been observed only location of water masers. sion) remains limited. with the shortest baseline UT2–UT3, and remain unresolved within the errors (which On the other hand, one might argue that are dominated by the measurement of the differences between both objects are

MIDI observations of distant AGN the total flux Ftot). Additional observations even more significant: only in NGC 1068 with longer baselines will be required to do we find dust heated to almost the sub- In addition to the detailed studies de- determine the size and flux of their dust limation temperature, while in Circinus scribed above, we carried out an AGN tori (if present). Despite its northern de- any strong temperature gradient between snapshot survey during the guaranteed clination (+40˚), we recently managed to the innermost dust and outer parts of time observations of the MIDI consor- observe the nearest Seyfert 1 galaxy, the torus is absent. Moreover, the elonga- tium. The survey tried to identify all those NGC 4151 with the VLTI. It is clearly re- tion of the hot dust component in AGN, which are bright enough in the solved at 10 µm with a 60-m baseline. NGC 1068 appears significantly tilted with

40 The Messenger 133 – September 2008 Name Type z Scale S N (core) MIDI Remarks Table 1. Target list and results of the [mas/pc] [mJy] AGN snapshot survey carried out dur- *NGC 1068 S2 0.00379 14.0 15 000 X well observed (16 visibility points), see text ing MIDI guaranteed time observations (GTO). Targets marked by * were NGC 1365 S1.8 0.00546 11.0 610 X marginally resolved in snapshot survey released from the GTO list early. A IRAS 05189-2524 S2 0.0425 1.0 550 AO correction with MACAO failed cross in the column ‘MIDI’ indicates MCG-5-23-16 S1.9 0.00827 5.7 650 X done in snapshot survey successful MIDI observations (X: com- Mrk 123 S1 0.0199 2.5 640 X done in snapshot survey plete interferometric measurement, x: fringes detected, but unstable weather NGC 3281 S2 0.01067 4.4 620 AO failed on nucleus, nearby star not used conditions prohibited complete obser- *NGC 3783 S1 0.00973 5.0 590 X observed by Beckert et al. (in prep.) vation). NGC 4151 S1 0.00182 14.0 1400 X resolved in snapshot survey Centaurus A RG 0.00332 53.0 1220 X first results with short baselines, see text IC 4329A S1 0.01605 3.1 420 x fringes detected *Mrk 463 S1 0.0504 1.0 340 not yet tried Circinus S2 0.00145 50.0 9 700 X well observed (21 visibility points), see text NGC 5506 S2 0.00618 8.0 910 AO correction with MACAO failed NGC 7469 S1 0.01631 3.1 410 x fringes detected NGC 7582 S2 0.00539 9.0 320 not yet tried 3C 273 QSR 0.1583 0.3 350v X two interferometric measurements NGC 253 core LE 0.00080 57.3 110 0 – no fringes detected (Hönig, priv. comm.) respect to the source axis as defined by NGC 1068. Thus the question arises: is flow onto the black hole might be equally the radio jet and the ionisation cone, there such a thing as the standard torus important. whereas the dust disc in Circinus seems in Seyfert galaxies? In any case, the to fit perfectly into an axisymmetric torus ‘torus’ possesses a complex structure, model. The outer torus in Circinus ap- which not only appears different (due References pears patchy or filamentary as predicted to line-of-sight effects) but may differ in- Antonucci, R. 1993, ARA&A, 31, 473 by hydrodynamical models. The low ab- trinsically between individual AGN. This is Gallimore, J. F., Baum, S. A. & O’Dea, C. P. 2004, sorption depth in the silicate feature not necessarily in conflict with the es- ApJ, 613, 794 towards the inner component indicates sential assumption of the unified scheme: Jaffe, W., Meisenheimer, K., Röttgering, H., et al. that our line of sight onto the dust disc it is still possible that Seyfert 1s and 2004, Nature, 429, 47 Krolik, J. H. & Begelman, M. C. 1988, ApJ, 329, 702 is not severely blocked by the outer Seyfert 2s are intrinsically the same class Leinert, C., Graser, U., Richichi, A., et al. 2003, structure and most of the large hydrogen of objects. In order to verify this generic The Messenger, 112, 13 column towards the X-ray core must assumption, one has to prove that sim- Meisenheimer, K., Tristram, K. R. W., Jaffe, W., et al. be located within a radius , 0.2 pc. In ilar tori, as in NGC 1068 and Circinus, 2007, A&A, 471, 453 Nenkowa, M., Ivezi, Z. & Eliitzur, M. 2002, ApJL, 570, contrast, NGC 1068 exhibits a huge dust also exist in Seyfert 1 galaxies. The de- L9 column towards the hot component. tection of an extended component in Raban, D., Heijligers, B., Röttgering, H., et al. 2008a, Here most of the absorbing gas and dust NGC 4151 with MIDI marks a promising A&A, in press (arXiv:0804.2395) is located outside a radius of ~ 1 pc. first step in this direction. Finally, our re- Raban, D., Jaffe, W., Röttgering, H., et al. 2008b, MNRAS, in press sults on Centaurus A demonstrate that Schartmann, M., Meisenheimer, K., Camenzind, M., From these differences it seems evident the absence of broad emission lines can- et al. 2008, A&A, in press (arXiv:0802.2604) that the torus in the Circinus galaxy is not always be explained by an obscuring Tristram, K. R. W., Meisenheimer, K., Jaffe, W., et al. not just a scaled-down version of that in torus. Intrinsic properties of the accretion 2007, A&A, 474, 837

Colour images of the brightest galaxies in four gal- increasing stellar mass, i.e. a rough time sequence. axy groups at redshift ~ 0.36, formed by combining The brightest galaxies in the left two images have VIMOS B, V and R band images (20 × 2 0 s e c t i o n s gravitationally bound bright companions. See ESO shown). The galaxies are ordered from left to right in Science Release 24/08 for more details.

The Messenger 133 – September 2008 41