cygnus a: stars, dust and cones

Cygnus A: stars, dust and cones

Neal Jackson,1 Clive Tadhunter2 and William B. Sparks3

1 NRAL Jodrell Bank, University of Manchester, Macclesfield SK11 9DL2 Department of Physics, University of Sheffield, Hounsfield Road, Sheffield S3 7RH3 Space Telescope Science Institute, 3700 San Martin Drive, Baltimore, MD 21218, USA

Accepted 1998 July 20. Received 1998 July 20; in original form 1998 March 25

A B S T R A C TWe present new Hubble Space Telescope (HST ) continuum and spectral line images of theradio galaxy Cygnus A. The images show much complex structure in the central kpc2.Continuum images show the central dust lane in detail, allowing detailed maps of E(B¹V) tobe constructed; the dust appears to follow a roughly Galactic extinction law. The emission-linecomponents are resolved in the line images and investigated in detail. A clear ‘opening cone’morphology is found, especially in the lines of Ha and [O i]. Blue condensations are seen in thesouth-eastern emission component and surrounding the central region. These are almostcertainly due to star formation, which began <1 Gyr ago as deduced from the colour of theregions. More extended blue continuum is also seen and corresponds to the blue polarizedcomponent detected by other recent spectropolarimetric observations.

Key words: galaxies: active – galaxies: individual: Cygnus A – galaxies: jets – galaxies:nuclei.

1 I N T RO D U C T I O N

Cygnus A has been the prototypical radio galaxy for over 40 yearsand only now is its basic structure becoming clear. A review ofdevelopments before 1996 is given by Carilli & Barthel (1996).

The object is a strong radio source, consisting of two bright radiolobes, separated by 160 kpc1 and arranged symmetrically about aradio core which provides the energizing jets. In the central few kpcis found a region of irregular optical line and continuum emissionsuperimposed on the red galaxy light. This emission has a double(NW–SE) structure; the north-western component emits proportio-nately more line emission than the south-eastern component. Aweaker central condensation is seen between the two major com-ponents on HST and high-resolution ground-based images. Accu-rate positioning of the radio core with respect to the astrometricframe of the HST image is not easy, but the radio jet appearscoincident with a gap in the eastern component.

Many authors (e.g. Barthel 1989) have suggested that everypowerful radio galaxy in fact contains a quasar nucleus, hiddenfrom direct view by obscuring material, arranged in a flatteneddistribution, along the line of sight. Oblique views of the hiddenquasar may be obtained by scattering of light from the centralregion into our line of sight by dust or electrons outside the plane ofthe obscuring material. This model predicts that the light soscattered should be polarized and have the spectrum of the hiddenquasar, including broad lines. Cygnus A has long been classified as

a radio galaxy, due to its predominantly narrow-line opticalspectrum, but there has long been speculation that at least someof the off-nuclear continuum is scattered nuclear light (Pierce &Stockton 1986; Tadhunter, Scarrott & Rolph 1990). There is nowconsiderable evidence for a hidden nucleus and that this nucleus isthat of a quasar. First, broad lines have been seen in the ultravioletregion of the spectrum (Antonucci, Hurt & Kinney 1994). Second,after a long and fruitless search for polarized broad lines (Goodrich& Miller 1989; Jackson & Tadhunter 1993), a high-sensitivityinvestigation with the Keck II telescope (Ogle et al. 1997) hassucceeded in detecting them. Both these results suggest that lightfrom the central regions is scattered into our line of sight by dust orelectrons outside the plane of the obscuring material. Ogle et al.estimate that this scattered light contributes about 15 per cent to thetotal light from the central regions at 5500 A, although this fractionincreases at shorter wavelength. The luminosity of the quasar can beestimated and appears to be at the low end of the quasar luminosityrange, in agreement with earlier estimates (Tadhunter et al. 1990).Further evidence for a hidden quasar comes from the inferredpresence of an absorbed power-law component in X-rays (Uenoet al. 1994).

However, there is some evidence that other factors contribute tothe narrow-line spectrum. Stockton, Ridgway & Lilly (1994) andTadhunter, Metz & Robinson (1994) have studied the extended ,5-kpc-wide region which emits the narrow lines, and find a number ofline ratios which cannot be simply explained by assuming that theyare photoionized by a hidden nucleus. The most puzzling problem isthe unusual strength and morphology of [N ii], which is difficult toreproduce by any simple model which assumes solar abundance.Tadhunter et al. (1994) suggest that there may be some contribution

Mon. Not. R. Astron. Soc. 301, 131–141 (1998)

q 1998 RAS

1 We use H0 ¼ 65 km s¹1 Mpc¹1 (e.g. Kundic et al. 1997), q0 ¼ 0:5throughout, which corresponds to 1:15 arcsec ¼ 1 kpc at the redshift ofCygnus A.

to the gas excitation from a distributed ionization source, and thatautoionizing shocks such as those investigated by Sutherland,Bicknell & Dopita (1993) may be present. Such shocks may bedriven by the radio jet’s passage through the line-emitting regions.Other quantities, such as the [O iii]/[O i] ratio, are easier toreconcile with the hidden-quasar model. This ratio increasestowards the centre of the cone, consistent with anisotropic escapeof ionizing photons bounded by the putative torus.

The properties of the nuclear regions of Cygnus A may also beinfluenced by the environment of the cluster in which it lies, whichmay also account for the unusual strength of the radio emission(Barthel & Arnaud 1996). In this cluster, a cooling flow is inferredto exist from X-ray observations (e.g. Arnaud et al. 1984; Reynolds& Fabian 1996), with mass deposition rates of up to 250 M( yr¹1.An obvious sink for this mass deposition is star formation, evidencefor which may be sought in the optical continuum emission.

In this paper we investigate several areas. First, we use the high-resolution images to constrain the geometry of the system. Inparticular, we argue that the structures we see in the lower-ioniza-tion-line images allow quite detailed statements to be made aboutthe geometry of the anisotropic escape of photons from the centre.We present new astrometry and attempt to place the radio coreaccurately with respect to the optical image. Further constraints arepresented on blue, compact star-forming regions within the centralregion and on the mechanisms which produce the optical continuumemission. Finally, we discuss the emission-line ratios seen in ourimages in an attempt to disentangle the effects of reddening andphotoionization.

2 H S T O B S E RVAT I O N S

Table 1 summarizes the HST observations. All observations weretaken using the Wide Field and Planetary Camera II, which was

installed in 1993 December during the first HST servicing missionand which corrects for the spherical aberration of the opticalsystem. The continuum images were taken using the standardbroad-band filters. In this programme the filters F336W andF814W were used, and images were extracted from the archivefor the filters F450W and F622W (in each case the number refers tothe approximate central wavelength of the filter in nm). Discussionof the F450Wand F622W images is presented briefly by Lynds et al.(1994). All continuum images were recorded by the PlanetaryCamera (PC) chip, which has a pixel scale of 0.0455 arcsec, andhave been background-subtracted using average values of back-ground emission away from the galaxy.

Line images were obtained using the Linear Ramp Filters(LRFs). These filters produce an image whose central wavelengthdepends on position on the chip, and the field of view at any givenwavelength is relatively small, though enough to encompass thecentral regions of Cygnus A. Images of Ha were obtained on the PCchip. These images are likely to contain strong contamination from[N ii]. All other line images were obtained on one of the three WideField (WF) chips, which have a pixel scale of ,0:996 arcsec.

Flat-fielding was performed for the LRF images using flat-fieldssupplied by STScI.2 Cosmic rays were rejected by combining theimages using the task crrej in the stsdas package, distributed bySTScI within the noao iraf software. All images were rotated usingthe orientat keyword in the FITS header of the data file. They werethen registered to each other using the nrao aips package by fitting tothe maximum of the light distribution (using the maxfit task) of a star00: 8 E, 40: 5 S of the central condensation of Cygnus A (except for the336-nm image in which the star is not visible at sufficient signal-to-noise ratio: in this case the image was aligned by eye to put the

132 N. Jackson, C. Tadhunter and W. B. Sparks

q 1998 RAS, MNRAS 301, 131–141

2 The observation numbers of the flat-fields used were, e391433ru for [O i]and Ha, e380934eu for [O ii], and e3809354u for [O iii].

Table 1. HST/WFPC2 observations of Cygnus A. In addition to these observations,further images in the F450W and F622W filters were extracted from the archive.

Date Observation number Filter Chip Exp. time Continuum/line(sec)

960408 u30x0101t FR418 WF2 1400 [O ii]960408 u30x0102t FR418 WF2 900 [O ii]960408 u30x0103t FR418 WF2 500 [O ii]960408 u30x0104t FR418 WF2 500 [O ii]960408 u30x0105t FR418 WF2 700 [O ii]960408 u30x0106t FR418 WF2 500 [O ii]960408 u30x0107t FR418 WF2 500 [O ii]960408 u30x0108t FR418 WF2 500 [O ii]960408 u30x0109t F814W PC1 500 I-band continuum960408 u30x010at F814W PC1 500 I-band continuum960408 u30x010bt FR680 WF2 1400 [O i]960408 u30x010ct FR680 WF2 1100 [O i]960408 u30x010dt FR680 WF2 500 [O i]960408 u30x010et FR680 WF2 500 [O i]960408 u30x010ft F336W PC1 600 U-band continuum960408 u30x010gt F336W PC1 400 U-band continuum960409 u30x010ht F336W PC1 500 U-band continuum960409 u30x010it F336W PC1 500 U-band continuum960409 u30x010jt FR680 PC1 700 Ha

960409 u30x010kt FR680 PC1 400 Ha

951122 u30x0201t FR533 WF3 400 [O iii]951122 u30x0202t FR533 WF3 400 [O iii]951122 u30x0203t FR533 WF3 500 [O iii]951122 u30x0204t FR680 PC1 400 Ha

features of Cygnus A in the right places). Detailed description of theadjustment of the HST frame to the radio frame is given in subsequentsections. Flux calibration of each frame was performed using thephotflam keyword in the header of each file.

The process of continuum subtraction of the line images thenfollowed. This is complicated by the fact that the continuum imagesthemselves are significantly contaminated by emission lines. Table2 gives a summary of the continuum filters used and the extent towhich they are contaminated by each line. The F555W image(Jackson et al. 1996) was not used, as it is contaminated by thevery strong [O iii] line at essentially 100 per cent transmission.

Pure continuum images were extracted from the broad-bandimages using the relationship

IC ¼

ICO ¹ SBLO

BCOfLILO

� �1 ¹ S

BLO

BCOfL

� � ;

where IC represents the corrected flux, and ICO the flux in thecontinuum image. For each line, fL represents the fractionalsensitivity of the filter to the line, tabulated above; BLO and BCO

represent the bandwidths of the narrow-band and broad-band filters,which are tabulated by STScI (Burrows et al. 1997); and ILO

represents the flux measured in the narrow-band image. Not alllines have available narrow-band images. In these cases we extra-polated from lines for which we did have images, using flux ratiosgiven by Osterbrock & Miller (1975) and Osterbrock (1983) torelate high-ionization lines to [O iii] and low-ionization lines to Ha.Continuum subtraction was finally performed on the line imagesusing the corrected continuum image IC. The only narrow-bandimage for which this made a significant difference to the structuresobserved was the [O i] image in the FR680 filter. All other lines hada sufficient equivalent width so that the continuum subtraction didnot change the counts appreciably in the regions of interest.

The final continuum and line images are presented in the panelsof Figs 1 and 2. Note that the pixel scale is different in differentimages. The pixels in the [O i], [O ii] and [O iii] images areapproximately twice as big as those in the Ha and continuumimages. The signal-to-noise ratio is variable. In particular, becauseof the relatively low transmission in the blue and the low galacticlatitude of Cygnus A (68), the 336-nm image is of very low signal-to-noise ratio. The observation u30x0204t was not used in the

composition of the Ha image due to the presence of a large,grazing-incidence cosmic ray in much of the nuclear region.

3 D I S C U S S I O N

3.1 Nature and centre of the cone

The position of the radio nucleus has in the past not been obvious.Vestergaard & Barthel (1993) obtained ground-based astrometryfor their study of Cygnus A with the Nordic Optical Telescope. Ithas not hitherto been possible, however, to achieve an accuratealignment of the HST and radio frames to better than the 0.5 arcsecachieved by Jackson et al. (1996). More accurate alignment requireseither an unlikely coincidence such as a nearby Hipparcos star, orfurther observations.

Accordingly, astrometry has been obtained of nearby stars withthe Carlsberg Automatic Meridian Circle on La Palma by theCAMC team. The positions are given in Table 3.

Astrometry has been attempted based on star B from Table 3,which normally lies on the same chip as the Cygnus A image. AGaussian has been fitted to each HST image wherever the image ofthe star is unsaturated and where the star falls well within the chip.This was possible for four images; those of [O i], [O iii], Ha and450-nm continuum. An offset has then been performed to the radioposition given by Carilli & Perley (private communication toVestergaard & Barthel 1993) of 19h59m28s

:348, 408440 020: 17(J2000). In Fig. 3 we show the resulting astrometry with thenominal position of the radio core with respect to the opticalstructure. The crosses represent the nominal 70-mas error of theCAMC astrometry. However, it is obvious from the figure that thereis an additional contribution of approximately the same order fromthe scaling and offsetting on the HST chips. (It should be pointedout, however, that in no case have we extrapolated from the star onone chip to the object on another.)

At the very least, these results make it clear that the radionucleus lies within the central condensation in the opticalstructure. There is a hint that it may lie just to the south-west ofthe peak of the optical emission, although the peak lies within theformal error box.

The results of Ogle et al. (1997) have made it clear that there is asubstantial contribution to the continuum light from photonsescaping anisotropically from the central quasar nucleus.

Cygnus A: stars, dust and cones 133

q 1998 RAS, MNRAS 301, 131–141

Table 2. Contaminating lines present in the four continuum images. Each line is given with its restwavelength and the nominal fractional sensitivity of the filter at the wavelength to which it is carried byCygnus A’s redshift. The sensitivities were estimated from the filter curves published by STScI (Burrows etal. 1997).

Filter Line Wavelength Sensitivity Filter Line Wavelength Sensitivity

F814W [S ii] 6731 0.75 F450W Hb 4861 0.49[S ii] 6717 0.64 [Heii] 4686 0.83[N ii] 6584 0.013 Hg 4340 0.87Ha 6563 0.006 [Ne iii] 3869 0.79

F622W [N ii] 6584 0.06 [Oii] 3727 0.64Ha 6563 0.11 F336W [Ne v] 3426 0.22

[N ii] 6548 0.19 Mgii 2879 0.008[O i] 6360 0.8[O i] 6300 0.85[N ii] 5754 0.75[N i] 5169 0.39

[O iii] 5007 0.0003

Accordingly, it comes as little surprise that our Ha image (Fig. 2)shows line emission concentrated in the region of solid angle closeto the radio jet. As well as the north-western component, there isconsiderable filamentary emission. The whole is bounded by arelatively bright outline of approximately paraboloidal shape.There are also areas where little emission is seen, but it has alreadybeen argued that there is patchy dust obscuration scattered through-out the central region, based on ground-based imaging (Shaw &Tadhunter 1994; Stockton et al. 1994; Jackson et al. 1996).

In the simplest versions of the ‘unified schemes’ of radio galaxiesand quasars, which the HSTand Keck spectroscopy and polarimetryresults suggest apply to Cygnus A, radio galaxies contain a hiddenquasar nucleus and the photons escape in regions not blocked by thetorus. This results in a cone-like distribution of line emission, whichis in fact seen in a small number of objects. The most convincingexamples of this are to be found in Seyfert galaxies such as NGC5252 (Tsvetanov & Tadhunter 1989) and NGC 5728 (Wilson et al.

1993). However, the presence of a cone-like distribution in CygnusA is controversial. Stockton et al. (1994) regard the morphology asbeing in essence a ‘mini-spiral’ based on their ground-basedimages, the structure being due mostly to star formation. Weargue later that star formation is indeed present. However, thereare two reasons that lead us to believe that we are seeing an‘ionization cone’. The first is the polarization work of Ogle et al.(1997). The second is that the Ha image shows both edges of thealleged ‘cone’ quite clearly (Fig. 2) despite some dust in thevicinity. However, the shape is not quite conical; the boundariesappear instead to be roughly paraboloidal. Fig. 4 shows the [O i]image, with the boundaries of the paraboloidal cone-edge drawn in.The position angle of the radio jet is also shown.

Why, in Cygnus A, do we appear to be seeing a cone openingangle that decreases with radius? One possible explanation is that,rather than resembling a brick wall, the edge of the torus is not sharpbut has a column density that increases relatively gradually with


q 1998 RAS, MNRAS 301, 131–141

DE

CL

INA

TIO

N (

J200

0)

RIGHT ASCENSION (J2000)19 59 28.55 28.50 28.45 28.40 28.35 28.30 28.25 28.20 28.15

40 44 04

03

02

01

00

Figure 1. Line-subtracted continuum images: 336 nm (top left), 450 nm (top right), 622 nm (bottom left), 814 nm (bottom right).

DE

CL

INA

TIO

N (

J200

0)


40 44 04

03

02

01

00

DE

CL

INA

TIO

N (

J200

0)


40 44 04

03

02

01

00

DE

CL

INA

TIO

N (

J200

0)


40 44 04

03

02

01

00

angle away from the radio jet. This, combined with r¹2 fall-off,means that the density of ionizing photons will fall below anyparticular critical level along a locus that is not a straight line. Theinside of the cone would be expected to be relatively brighter than

the edges, and this is not the case. However, the continuum imagesreveal the presence of considerable extinction in front of the cone,which blocks out much of the region between the nucleus and thenorth-western component.


q 1998 RAS, MNRAS 301, 131–141

DE

CL

INA

TIO

N (

J200

0)


40 44 04

03

02

01

00

Figure 2. Continuum-subtracted line images (see text). [O i](top left), [Oii](bottom left), Ha (top right), [Oiii] (bottom right).

Table 3. Positions of stars near to Cygnus A, measured with the CarlsbergAutomatic Meridian Circle. Positions are given in J2000 coordinates with anassociated error. The errors are likely to be larger than uncertainties in alignmentof the radio and optical reference frames.

Star RA (J2000) Dec (J2000) Error Observing epoch V magnitude

A 19 59 29.730 +40 45 15.17 00: 06 1996.45 12.42B 19 59 30.079 +40 44 04.66 00: 07 1996.50 14.30C 19 59 31.519 +40 43 52.39 00: 13 1996.59 15.56

DE

CL

INA

TIO

N (

J200

0)


40 44 04

03

02

01

00

DE

CL

INA

TIO

N (

J200

0)


40 44 04

03

02

01

00

DE

CL

INA

TIO

N (

J200

0)


40 44 04

03

02

01

00

3.2 The continuum images

The three continuum images reveal a large amount of detailedstructure (Fig. 1). There appear to be three major components: theold galaxy component, blue compact condensations and moreextended blue regions. These will be discussed in turn.

3.2.1 The old galaxy component

The host galaxy is clearly visible, outside the small region coveredby Fig. 1, as a smooth, elliptical brightness distribution with a

semimajor axis at PA, ¹ 208. In order to fit the galaxy, we havetaken the 450-, 622- and 814-nm images as a starting point. Thesehave been cleaned of stars and low-level cosmic rays by hand usingthe CLEAN program in FIGARO. The galaxy has then been subtractedusing the IMSURFIT package in IRAF, taking in each case the fit regionas the region outside a circle of radius 1.5 arcsec from the centre ofthe galaxy. This has the effect of interpolating into the centralregions without the fit being affected by the components associatedwith the active nucleus and line emission at the centre. The order ofthe spline has been adjusted until visual inspection reveals asatisfactory fit without the introduction of obvious artefacts orresiduals. A ninth-order spline was generally required to fit allareas of the image satisfactorily. A cut along the major axis of theresulting galaxy fit shows a light distribution that is flat with radiusuntil about 1.1 arcsec from the nucleus, and then drops off at a rateof about 3.5 mag arcsec¹2 arcsec¹1=4, similar to the profile found byStockton et al. (1994, their fig. 4). The major axis of the fit withinthe central regions is along PA,1408, although visual inspection ofthe more distant isophotes (> 2 arcsec) shows a position angle closeto the Stockton et al. value of 1668. The reason is likely to be thatsome residual components beyond 1.5 arcsec and associated withthe active galaxy may affect the fit slightly. After subtraction of thegalaxy fit, the flux in the central region (about 2 arcsec) of eachimage has then been examined.

The raw flux densities in the central regions in these models of thegalaxy are about 2.2, 3.2 and 3:4 × 10¹17erg cm¹2 s¹1A¹1 persquare arcsecond in the 450-, 622- and 814-nm bands respectively.Correcting for the reddening of E(B¹V).0.4 in our own Galaxy(e.g. Spinrad & Stauffer 1982; Shaw & Tadhunter 1994) gives aratio of 1:1.3:1.1 in Fl between these three wavelengths. Bycomparison with the spectral synthesis models of Bruzual &Charlot (1993), this can be seen to agree well with the expectedcolours of a 5–10 Gyr-old stellar population. This provides addi-tional confirmation of the photometry together with existing red-dening determinations.

3.2.2 Blue compact condensations: star formation?

It is clear from comparisons of the continuum images in Fig. 1 that


q 1998 RAS, MNRAS 301, 131–141

DE

CL

INA

TIO

N (

J200

0)


40 44 04

03

02

01

00

Figure 3. The 814-nm image, with the derived position of the radio corefrom astrometry performed on four frames (Ha, [O i], [O iii] and 450-nmcontinuum). Errors on the astrometry are about 70 mas due to the standardstar position.

DE

CL

INA

TIO

N (

J200

0)


40 44 04

03

02

01

00

Figure 4. The [O i] image, with the edges of the paraboloidal boundary ofthe emitting region drawn. The arrowed line represents the position angle ofthe radio jet, and starts from our best estimate of the radio core position fromthe CAMC astrometry.

DE

CL

INA

TIO

N (

J200

0)


40 44 04

03

02

01

00

1

2

34

56

7 8

Figure 5. 622-nm image, with the major blue condensation regions marked(see text).

the bluest regions are concentrated in three places: in part of thenorth-western component, very close to where the radio jet goesthrough it (cf. Jackson et al. 1996), within the south-easterncomponent, and to the north of the main structure. Lynds et al.(1994), Stockton et al. (1994) and Jackson et al. (1996) have allsuggested that we are here seeing regions of star formation. Thecondensations are reminiscent of the young star clusters seen in thenearby active galaxy NGC 1275 (Holtzman et al. 1992).

In Fig. 5 some of the main compact blue regions are identified.Regions 2–5 are completely within the south-eastern componentand have sizes of ,0:1 arcsec. Three other condensations, 6, 7 and8, lie just to the north of the main structure. The emission in thenorth-western component is more homogeneous and covers a largerarea. In this section we consider three alternative mechanisms forproducing these compact blue regions: nebular continuum (e.g.Dickson et al. 1995), scattered light from the hidden nucleus (Ogleet al. 1997) and, finally, star formation.

It is likely that some nebular continuum is in fact present(Dickson 1997; Ogle et al. 1997). However, for typical physicalconditions, the equivalent width of Hb over nebular continuum is,2000 A (see the case of PKS 2250¹41 discussed by Dickson et al.1995). Assuming the observed Ha/Hb to be 4, this suggests that ablue condensation of continuum flux density ,3 × 10¹19

erg cm¹2 s¹1A¹1 should produce about 2× 10¹15 erg cm¹2 s¹1 inline flux. The Ha image is slightly difficult to measure, as emissionfrom the blue condensations sits on a plateau of more generalemission within the south-eastern component (and because it iscontaminated by [N ii] emission), but is probably about a factor of 3lower than this. In condensation 8, which lies away from the mainemitting regions and is therefore easy to measure, the deficit in Ha

photons is a factor of 10 and may be due to residual continuumcontamination of the emission-line image. We therefore concludethat, while nebular continuum may be contributing, the totalcontribution to the blue condensations is likely to be about 10 percent. This is also the conclusion reached independently by Ogle etal. (1997) from detailed fitting of their spectropolarimetric data.

Scattered light is also a possibility for the blue condensations.One may speculate that they represent clumps of scattering materialin the beam of the hidden quasar. Again, however, one would expectgas to be associated with the dust, and therefore one would expect

emission lines. Since the blue condensations form a substantial partof the south-eastern component, it would then be very difficult toexplain the significantly lower emission line equivalent width in thesouth-eastern component than in the north-western component (e.g.van den Bergh 1976).

We now consider the possibility of star formation. It has beenproposed for some time that stars could be formed within galaxieswhich, like Cygnus A, lie at the centre of a cooling flow cluster(Fabian, Nulsen & Canizares 1982; Reynolds & Fabian 1996).Imaging and spectroscopy of a number of central cluster galaxieshas recently supported the interpretation that stars are formeddirectly as mass drops out of the cooling flow (Cardiel, Gorgas &Aragon-Salamanca 1995; Melnick, Gopal-Krishna & Terlevich1997; Smith et al. 1997). In the nearby cD galaxy Hydra A,Balmer absorption lines corotating with the gas in a circumnucleardisc have been found (Melnick et al. 1997), suggesting formation ofO–B stars near the centre. Melnick et al. suggest a cooling flow asthe origin of the star formation, arguing that the colours of the starssuggest that any merger must have occurred less than 10 Myr ago.

It is also possible, however, that the star formation is induced bymerger (e.g. Fritze-von Alvensleben & Gerhard 1994), and this maybe particularly true in the case of relatively isolated radio galaxies(Tadhunter, Dickson & Shaw 1996). Holtzman et al. (1992) haveargued that NGC 1275, which has similar compact regions of bluelight, contains stars which have formed as a result of a merger, andthat the merger origin is supported by the lack of dispersion in thecolours of the blue condensations – implying that they formed at thesame time.

The continuum energy distribution of each of the compact blueregions in Cygnus A is shown in Fig. 6, together with predictions ofcolours for stellar populations of various ages from an initialstarburst (Bruzual & Charlot 1993). The colours of the regionshave been corrected for reddening of E(B¹V) ¼ 0:4. The 100-Myrline drawn in the figure assumes that the star formation has a longcharacteristic time-scale (>3 Gyr); instantaneous burst models givemuch flatter predicted spectra unless t # 10 Myr. From the figure itcan be seen that, if the compact blue regions are the result ofstarburst activity, it must have happened much less than 1 Gyr agoand possibly as little as a few Myr ago. If the reddening in theseregions is greater than E(B¹V) ¼ 0:4, the limit becomes even morestringent. The sum of the flux densities in regions 1–6 is about3×10¹18 erg cm¹2 s¹1A¹1, corresponding to about 3×107L(. Thisprobably does not represent all the luminosity associated with eachepisode of star formation, which is expected to last ,107yr (Heck-man, Armus & Miley 1990) at a star formation rate of,100 M( yr¹1 (e.g. Colina & Perez-Olea 1992).

The compact blue regions are not clearly seen on the emission-line images. This is unlikely to be a sampling, signal-to-noise ratioor resolution effect, as they are barely visible on the Ha image,which is very similar to the 622-nm image in terms of resolution.

How can this be related to the models of star formation? The lackof detailed correspondence with the radio jet, seen also in somehigh-redshift radio galaxies (Best, Longair & Rottgering 1996),suggests that this star formation is not induced by the radio jetdirectly. The direct cause is likely to be shocks; these may propagateoutward after the passage of the radio jet or be associated with anongoing merger event. Stockton et al. (1994) argue for the latter. Inthe case of Cygnus A, however, there is also evidence for a coolingflow (Reynolds & Fabian 1996) as well as evidence for denseenvironments such as that provided by a Faraday rotation measureof $1000 rad m¹2, which is similar to that observed in Hydra A.This may provide the gas for the star formation. There is, however, a


q 1998 RAS, MNRAS 301, 131–141

Figure 6. Plot of log (flux) against log (wavelength) for the regionsidentified in Fig. 5. Regions are: 1 — cross; 2 — asterisk; 6 — triangle;4 — x; 7 — cross in circle; 5 — square; 8 — circle with dot; 3 — circle.Lines corresponding to tSFR ¼ 3 Gyr starbursts from Bruzual & Charlot(1993) are also plotted.

relatively low dispersion in age between the blue condensations(Fig. 6), which could suggest a merger origin.

3.2.3 Extended blue regions: scattered light

In addition to the compact blue regions seen in and around thesouth-eastern component, more diffuse blue regions are seen in bothmajor optical components in the centre of Cygnus A. Themore extended regions dominate the light from the north-westerncomponent and form a low-level background in the south-easterncomponent to the blue condensations discussed in the lastsection.

The results of Ogle et al. (1997) leave little doubt that thisextended continuum is due to scattered quasar light. The modelsresulting from spectropolarimetry suggest that ,60, ,40 and ,20

per cent of the light at 450, 622 and 814 nm, respectively, is due tothis component (which agrees well with the value of ,60 per centnear Hb deduced by Osterbrock 1983). We have used the AIPS

routine TVSTAT to integrate the light in each of the continuum images(see the last section for details) along the slit used by Ogle et al.(1997, their fig. 2). An estimate of the background old-galaxycomponent has then been subtracted, taken using an average in theregions just outside the central condensations. Of the background-subtracted light, the region in the north-western component con-tributes about 30, 25 and 20 per cent in the 450-, 622-nm and 814-nm images, respectively. Although much of the polarized, scatteredcontinuum arises in the north-western component, a significantamount must also lie in the south-eastern component. This is inaccordance with the finding of Ogle et al. that the south-easterncomponent contains significant polarized continuum emission.


q 1998 RAS, MNRAS 301, 131–141

Figure 7. The inferred extinction in the 450-, 622- and 814-nm frames, in magnitudes, defined as 2.5 log (observed flux/flux in fitted galaxy component). For the622- and 814-nm frames, dotted contours are given at ¹1, ¹0.9...¹0.1 mag, and solid contours at 0.1, 0.2...1.0 mag. For the 450-nm frame, dotted contours aregiven at ¹3, ¹2.7...¹0.3 mag, and solid contours at 0.3, 0.6...3.0 mag.

3.3 Ionization and reddening

3.3.1 Amount of extinction

Many studies have previously attempted to quantify the reddeningwhich occurs between us and the optically emitting components inCygnus A. There are two parts to the reddening: a considerableforeground Galactic contribution (Cygnus A lies at b ¼ 68),together with extinction within the Cygnus A host galaxy. Thesubject is reviewed by Shaw & Tadhunter (1994) and Tadhunter etal. (1994), based on earlier work and their own spectroscopy andthat by Stockton et al. (1994). The likely situation is that theforeground reddening is approximately equal to E(B¹V) ¼ 0:36(e.g. Spinrad & Stauffer 1982), and that additional reddening withinCygnus A occurs. The maximum total reddening inferred from Hb/Hg ratios (Tadhunter et al. 1994) is E(B¹V) ¼ 1, although theaverage total reddening over the central regions is E(B¹V),0.7.The continuum images (Fig. 2), however, make it plain that thereddening is extremely patchy and is probably considerably greaterthan E(B¹V) ¼ 1 in the centre of the obscured area just to the north-west of the nucleus.

We have used the galaxy-subtraction procedure outlined earlierto estimate the quantity of dust present. The ‘extinction’, inmagnitudes, has been calculated from the ratio of the flux observedin the raw frames to the interpolated flux in the galaxy. The resultsare shown in Fig. 7. Note that the ‘extinction’ is negative wherecontinuum emission from regions associated with the nuclearactivity is present. Pixel-by-pixel plots of inferred extinction atdifferent wavelengths show a good correlation, with systematicallyhigher extinction at shorter wavelengths. For extinction in our ownGalaxy (e.g. Howarth 1983), A814 , 1:79E(B¹V), A622 , 2:65E(B¹V) and A450 , 3:9E(B¹V). The observed reddening ratiosseen in Cygnus A are consistent with the Galactic ratio.

The implied E(B¹V) varies with position, but reaches a max-imum of about 0.5. This is a lower limit on the obscuration, as itassumes that the only emission taking place along this line of sight

comes from the smooth galaxy component. It represents only theobscuration intrinsic to Cygnus A, as any obscuration in our ownGalaxy would affect also the smooth component of Cygnus A. Thelimit, however, agrees reasonably well with the estimate for the‘dust lane’ of E(B¹V) ¼ ,0.4 derived by Shaw & Tadhunter (1994)from emission-line ratios.

Assuming a Galactic dust-to-gas ratio (Bohlin, Savage & Drake1978), this implies a gas column number density ofNðHÞ ¼ 3 × 1021 atom cm¹2, corresponding to a gas mass of2:2 × 107M(.

3.3.2 Distribution of extinction: dust lane?

Many authors have suggested that the distribution of extinction inCygnus A is the result of a dust lane, similar to those seen inCentaurus A (Baade & Minkowski 1954) and in other more power-ful radio galaxies. Recent examples include 3C 270 (Mahabal et al.1996), and in a large HST survey of 0.1< z < 0:5 radio galaxies, deKoff et al. (1996) find many dust features including dust lanes in thecentral ,1 kpc. Frequent associations with disturbed morphologiesand tidal tails lead de Koff et al. to suggest a merger origin for theoptical structure.

The three-colour continuum images allow investigation of thereddening distribution across the central region. Unfortunately, it isdifficult to disentangle the reddening from the unknown emissiondistribution, and this problem has already allowed only lower limitson E(B¹V) to be derived. The major reddening occurs along a NE–SW line running between the nuclear component and the north-western component. Less intense reddening occurs along a parallelline, south of the nucleus. These two regions correspond exactly tothe dust lanes identified by Stockton et al. (1994, their fig. 3). TheHST pictures reinforce the impression that the reddening is patchy.

Stockton et al. (1994) suggested that the structure of the centralregions could be interpreted as a ‘mini-spiral’ system, associatedwith a merger. They proposed that a secondary infrared componentseen 1.3 arcsec north of the nucleus is, in fact, the nucleus of amerging companion, and that the dust lanes lie along the edges ofmini-‘spiral arms’. The new data provide some support to the


q 1998 RAS, MNRAS 301, 131–141

DE

CL

INA

TIO

N (

J200

0)


40 44 04

03

02

01

00

Figure 8. Image of the ratio of Ha/[O iii], blanked in areas of low signal-to-noise ratio. The grey-scale runs from 0 (white) to 2.5 (black). Note theconcentration of [Oiii] in a narrower ‘cone’ than the Ha distribution.

DE

CL

INA

TIO

N (

J200

0)


40 44 04

03

02

01

00

Figure 9. Ratio of [O iii]/[O i] in the central region of Cygnus A. The grey-scale runs from 0 (white) to 30 (black).

hypothesis, as most of the blue condensations lie in the placesproposed by Stockton et al. as the sites of star formation, andcondensations 6–8, though some distance away from the majoremission components, lie around the edge of the northern dust lane.

3.3.3 Line ratios

The line ratios provide evidence in support of the ionization-conemodel discussed earlier. Fig. 8 shows an image of the ratio of Ha/[O iii]. It is evident that the [O iii] emission occupies a considerablysmaller region of solid angle than Ha. It is, of course, possible toaccount for this difference of a factor of 3 in the ratio by postulatingdifferential reddening. However, to achieve such a factor between[O iii] and Ha one needs an extinction at V of about 4 mag, andthere is no evidence for such high values within the central region.Nor is it possible to appeal to different pixel scales and effectiveresolutions: the [O i] image, which has the same pixel scale as the[O iii] image, shows the boundaries of the conical region quiteclearly and the [O iii] image does not.

Fig. 9 shows the [O iii]/[O i] ratio across the source, blanked inregions of low signal-to-noise ratio in the [O i] image. Here the mostobvious feature is the relative brightness of [O iii] towards the outeredges of the north-west component. These regions are also thebluest within the component, and again one may wonder ifdifferential reddening is intervening. We do not believe so, fortwo reasons. First, in Fig. 10 we plot the [O iii]/[O i] ratio pixel bypixel. A clear correlation is evident, and the straight-line nature ofthis plot suggests that reddening is not responsible. Second,Tadhunter et al. (1994), in their spectroscopic study of thisregion, found a similar variation in [O iii]/[O i] across the north-western component which did not vanish when corrected for red-dening by use of the Ha/Hb ratio.

A similar attempt to derive ionization parameter variations hasbeen made using the [O iii]/[O ii] ratio (Fig. 11), but this is difficultdue to the low signal-to-noise ratio in the [O ii] image. Theionization state appears highest in the central and north-westerncomponents.

4 C O N C L U S I O N S

We have presented new HST images of the prototypical powerfulradio galaxy Cygnus A. We amplify recent spectropolarimetryresults by localizing the blue continuum emission associated withthe polarized optical component seen by Ogle et al. (1997). We alsolocalize blue, condensed clumps of forming young stars, which area considerable distance from the radio jet and therefore not due todirect jet-induced star formation. The gas which forms the stars islikely to have come either from the cooling flow in which Cygnus Ais embedded or from a merger. Formation activity must have beenoccurring considerably less than 109 yr ago. New astrometry hasenabled us to pin down the position of the optical nucleus withrespect to the radio core and line images to confirm earlier resultsabout the ionization gradients across the source. The structure of thecontinuum and line emission provides the clearest example yet ofcone-like structure in a powerful radio galaxy, which is particularlynoticeable in the emission-line images.

AC K N OW L E D G M E N T S

This research was based on observations with the Hubble SpaceTelescope, obtained at the Space Telescope Science Institute, whichis operated by Associated Universities for Research in AstronomyInc. under NASA contract NAS5-26555. The positions of thenearby stars to Cygnus A were obtained using the CarlsbergAutomatic Meridian Circle (CAMC) on La Palma and commu-nicated to us by Drs Leslie Morrison and Bob Argyle.

REFERENCES

Antonucci R., Hurt T., Kinney A., 1994, Nat, 371, 313Arnaud K. A., Fabian A. C., Eales S. A., Jones C., Forman W., 1984,

MNRAS, 211, 981Barthel P. D., 1989, ApJ, 336, 606Barthel P. D., Arnaud K. A., 1996, MNRAS, 283, L45Best P. N., Longair M. S., Rottgering H. J. A., 1996, MNRAS, 280, L9Bohlin R. C., Savage B. D., Drake J. F., 1978, ApJ, 224, 132


q 1998 RAS, MNRAS 301, 131–141

Figure 10. Pixel-by-pixel plot of [O iii]/[O i] in the central region of CygnusA.

DE

CL

INA

TIO

N (

J200

0)


40 44 04

03

02

01

00

Figure 11. Ratio of [O iii]/ [O ii] in the central region of Cygnus A, blankedin regions where the [Oii] image is of low signal-to-noise ratio. The grey-scale runs from¹0.17 (white) to 8.12 (black).

Bruzual G., Charlot C., 1993, ApJ, 405, 538Burrows C. J. et al., 1997, in Burrows C. J., ed., Wide Field and Planetary

Camera 2 Instrument Handbook. STScI PublicationCardiel N., Gorgas J., Aragon-Salamanca A., 1995, MNRAS, 277, 502Carilli C., Barthel P. D., 1996, A&AR, 7, 1Colina L., Perez-Olea D., 1992, MNRAS, 259, 709de Koff S., Baum S. A., Sparks W. B., Biretta J., Golombek D., Macchetto F.,

McCarthy P., Miley G. K., 1996, ApJS, 107, 621Dickson R., 1997, PhD thesis, Univ. SheffieldDickson R., Tadhunter C., Clark N., Morganti R., 1995, MNRAS, 273, L29Fabian A. C., Nulsen P. E. J., Canizares C. R., 1982, MNRAS, 201, 933Fritze-von Alvensleben U., Gerhard O. E., 1994, A&A, 285, 751Goodrich R. W., Miller J. S., 1989, ApJ 346, L21Heckman T. M., Armus L., Miley G. K., 1990, ApJS, 74, 833Holtzman J.A. et al., 1992, AJ, 103, 691Howarth I., 1983, MNRAS, 203, 301Jackson N., Tadhunter C. N., 1993, A&A, 272, 105Jackson N., Tadhunter C. N., Sparks W. B., Miley G. K., Macchetto F., 1996,

A&A, 307, L29Kundic, T. et al., 1997, ApJ, 482, 75Lynds R. et al., 1994, BAAS, 26, 941Mahabal A., Kembhavi A., Singh K. P., Bhat P. N., Prabhu T. P., 1996, ApJ,

457, 598Melnick J., Gopal-Krishna, Terlevich R., 1997, A&A, 318, 337

Ogle P. M., Cohen M. H., Miller J. S., Tran H. D., Fosbury R. A. E., GoodrichR. W., 1997, ApJ, 482, L37

Osterbrock D. E., 1983, PASP, 95, 563Osterbrock D. E., Miller J. S., 1975, ApJ, 197, 535Pierce M. J., Stockton A., 1986, ApJ, 305, 204Reynolds C. S., Fabian A. C., 1996, MNRAS, 278, 479Shaw M., Tadhunter C., 1994, MNRAS, 267, 589Smith E. P., Bohlin R. C., Bothun G. D., O’Connell R. W., Roberts M. S.,

Neff S. G., Smith A. M., Stecher T. P., 1997, ApJ, 478, 516Spinrad H., Stauffer J. R., 1982, MNRAS, 200, 153Stockton A., Ridgway S. E., Lilly S., 1994, AJ, 108, 414Sutherland R. S., Bicknell G. V., Dopita M. A., 1993, ApJ, 414, 510Tadhunter C. N., Scarrott S. M., Rolph C. D., 1990, MNRAS, 246, 163Tadhunter C., Metz S., Robinson A., 1994, MNRAS, 268, 989Tadhunter C. N., Dickson S., Shaw M. A., 1996, MNRAS, 281, 591Tsvetanov Z., Tadhunter C. N., 1989, Nat, 341, 422Ueno S., Koyama K., Nishida M., Yamauchi S., Ward M. J., 1994, ApJ, 431,

L1van den Bergh S., 1976, ApJ, 210, L63Vestergaard M., Barthel P. D., 1993, AJ, 105, 456Wilson A. S., Braatz J. A., Heckman T. M., Krolik J. H., Miley G. K., 1993,

ApJ, 419, L61

This paper has been typeset from a TEX=LATEX file prepared by the author.


q 1998 RAS, MNRAS 301, 131–141

cygnus a: stars, dust and cones

Documents