The evidence for jet-cloud interactions in a sample of high/intermediate-redshift radio galaxies

33 Pages·0.63 MB·English

by  C. Solorzano-Inarrea

Mon.Not.R.Astron.Soc.000,000–000 (0000) Printed1February2008 (MNLATEXstylefilev1.4)
The evidence for jet-cloud interactions in a sample of
high/intermediate-redshift radio galaxies
C. Sol´orzano-In˜arrea1,2, C. N. Tadhunter1, D. J. Axon3
1Department of Physics and Astronomy, Universityof Sheffield, Sheffield, S3 7RH
2Department of Physics and Astronomy, Universityof Leeds, Leeds, LS29JT
3Department of Physical Sciences, Universityof Hertfordshire, Hatfield, AL10 9AB
1
0
0 1February2008
2

n
a ABSTRACT
J We present detailed long-slit spectroscopic observations of a sample of four powerful
5 radio galaxies, with intermediate/high redshifts (0.47 < z < 0.81). The observations
2 cover the rest-wavelength range 2600 ˚A- 6600 ˚A, chosen to include important diag-
nostic emission lines ([NeV]λ3426, [OII]λ3727, [NeIII]λ3869, Hβ, [OIII]λ5007), which

1 are also measured in optical observations of low-redshift radio galaxies. In two of
v
the galaxies (3C 352 and 3C 435A) the radio sources are of the same scale as the
7
emission-line regions, whereas in the other two (3C 34 and 3C 330) the radio sources
5
are extended on a larger scale than the emission-line structures. We find that the
4
extended regionsof all the galaxiespresenthighly disturbed kinematics,consisting of
1
0 line-splitting (∆v ∼ 1000 km s−1) and/or underlying broad components (FWHM =
−1
1 1000— 1500km s ). These features aredifficult to explainin terms of gravitational
0 motioninthehaloesofthehostgalaxies.Rather,itislikelythattheyaretheresultof
/ strong shocks driven through the ISM/IGM by the radio sources. These observations
h
suggest that jet-induced shocks have an important effect on the emission-line proper-
p
ties even in sources in which the radio structures are on a much larger scale than the
-
o emission-line structures.
r While the emission-line kinematics provide strong evidence for shock accelera-
t
s tion, the dominant ionization mechanism for the emission-line gas remains uncertain.
a
We have compared the optical diagnostic line ratios of the galaxies in our sample
:
v with various ionization models, including: pure shock ionization, shocks including
i a photoionized precursor, power-law photoionization and photoionization including
X
matter-bounded clouds. We find that both pure-shock ionization and power-law pho-
r toionization model predictions fail to provide good fits to the data. On the other
a
hand,onindividual diagnosticdiagrams,models forshockswhichinclude a photoion-
ized precursor are consistent with the results for the majority of the EELR of the
galaxies in our sample, and photoionization including matter-bounded clouds models
also give reasonable fits to some of the EELR. However, in terms of the positions of
the points relative to the model sequences on the diagnostic diagrams,there is a lack
of consistency from diagram to diagram. The diagnostic diagram involving the line
ratios [OIII](4959+5007)/4363and HeII(4686)/Hβ is particularly problematic in this
regard. Overall, our results suggest that, if the EELR are shock-ionized, one or more
of the assumptions implicit in the shock models may need to be reconsidered.
In addition, we have investigated the nebular continuum contribution to the UV
excess in the galaxiesin our sample.We find a substantialnebular emissioncontribu-
tionto the UV continuuminallthe cases,inthe range∼ 10—40 per cent.However,
after the subtraction of the nebular component, a significant UV excess remains in
the extended nebulae of most of the objects.
Key words: galaxies: active – galaxies:jets – galaxies:kinematics and dynamics.
c 0000RAS
(cid:13)
2 C. Sol´orzano-In˜arrea et al.
1 INTRODUCTION low-surface-brightness emission-line structures at large dis-
tancesfromtheradiojetaxis.Thesestructuresareinconsis-
Most powerful radio galaxies have extended emission-line
tentwiththeilluminationmodel,sincepartsofthemlieout-
regions (EELR) extendingto large distances from their nu-
side the ionization cones predicted by the unified schemes,
clei(5—100kpc)(Tadhunter1986).TheseEELRofferthe
andrathersuggestthattheionizingeffectsoftheradiocom-
possibility ofinvestigatingtheorigin oftheactivityandthe
ponentscan extendfar from the radio jet axes.
origin of the gas itself. However, if we are going to use the
At higher redshifts (z > 0.5) there appears to be more
EELRinthisway,weneedtoknowtheextenttowhichthe
evidence for shocks than in the population of low-redshift
emission-linemorphologiesandkinematicsreflecttheintrin-
radiogalaxies.Atsuchredshiftspowerfulradiogalaxiesfre-
sicpropertiesofthegaseoushaloes,andtheextendtowhich
quentlyshowhighlycollimatedUVcontinuumandemission-
they reflect the effects of the activity. Thus, it is crucial to
linestructureswhicharecloselyaligned withtheradioaxes
understand theionization mechanisms in these sources.
(McCarthy et al. 1987; Chambers, Miley & van Breugel
The nature of the mechanisms involved in producing
1987; Best, Longair & R¨ottgering 1996). Such collimated
theobservedkinematicsandionizationoftheEELRremains
structures are not what would be expected on the basis of
highly uncertain. Currently, the two most accepted models
illumination by the broad radiation cones predicted by the
are photoionization by a central active nucleus, and shock-
unifiedschemes.Inaddition,manyhigh-redshiftradiogalax-
ionization by the interaction between the radio-emitting
ies present highly disturbed emission-line kinematics along
components and the ambient gas. In the latter model, the
theradioaxis,whicharedifficulttoexplainintermsofpure
so-called jet-cloud interaction model, the advancing radio
gravitational motions (McCarthy, Baum & Spinrad 1996).
jet drives strong shocks into the warm clouds of gas in the
This suggests that processes other than AGN illumination
haloes of the host galaxies. The regions directly behind the
maybeatwork.Indeed,manyofthesefeaturesobservedin
shocksarecompressedandheatedtoveryhightemperatures
powerfulhigh-redshiftradiogalaxiesareconsistentwiththe
and line emission is produced by two distinct mechanisms:
jet-cloud interaction model described above and with the
(a)asthepost-shockcloudscoolbyemittinglineradiation,
results of detailed studies of jet-cloud interactions in low-
and (b) photoionization of the cool, dense, post-shock gas
redshift radio galaxies.
andtheprecursorgasaheadoftheshockbythehotshocked
Inaddition,recentworkbyBest,R¨ottgering&Longair
gas. It is likely that no single mechanism, but a combina-
(2000a),atredshiftsz 1,hasdemonstratedthatthereisa
tion ofboth AGNphotoionization andjet-cloud interaction ≃
strong evolution of the emission-line properties of powerful
is required to explain the kinematics and ionization of the
radio galaxies with radio size. They find that radio sources
EELR overthewhole range of redshifts.
with small linear sizes have emission-line properties which
Atlowredshifts(z<0.3),thepropertiesoftheEELRin
are in agreement with the theoretical shock ionization pre-
mostradiogalaxies appeartobeconsistent withtheillumi-
dictions,andshowhighlydistortedvelocityprofiles.Incon-
nationoftheambientgasbytheradiation fieldofanactive
trast, the larger radio sources, where the shock fronts have
nucleus(e.g. Fosbury1989), and theemission-line kinemat-
passed beyond the emission-line regions, present emission-
ics can be explained in terms of pure gravitational motion
lineratiosconsistentwithAGNphotoionization,withhigher
in the haloes of the galaxies (Tadhunter, Fosbury & Quinn
ionization statesandnarrowerlinewidthsthanthoseofthe
1989b; Baum, Heckman & van Breugel 1990). This illumi-
smaller radio sources.
nation modelis inagreement with theunifiedschemes(e.g.
However,inapparentcontradictionwiththeresultsob-
Barthel 1989), in which it is proposed that all active galax-
tained at z 1, Villar-Mart´ın, Tadhunter & Clark (1997),
ies are intrinsically identical but, dueto orientation effects, ≃
basingtheirstudyontheUVlineratiosofasampleofradio
their appearance varies.
galaxies at z>2, found that theEELR appear to be mainly
However, some nearby radio galaxies show clear ev-
photoionized by the central AGN in the very high redshift
idence for strong interactions between the radio-emitting
regime.
componentsandtheambientgas.Featuresobservedinthese
One problem with comparing the results from samples
galaxies include:
atdifferentredshiftsisthatopticalobservationssampledif-
close morphological association between radio and op- ferentrestwavelengthsastheredshiftvaries.Consequently,
tic•al emission; the diagnostic diagrams used to determine the ionization
complexkinematics(underlyingbroadcomponentsand mechanism involve lines of different ionization and excita-
spl•it narrow components); tion as the redshift changes, and there is a danger that our
large linewidths in the extendedgas; sensitivity to particular ionization mechanisms varies with
• wavelength.Becauseofthisproblem,ithasbeendifficultto
anticorrelationsbetweenlinewidthandionizationstate.
• determinethebalancebetweenshocksandphotoioinization,
Detailed studiesbyClark etal.(1997, 1998) andVillar- and how this balance varies with redshift.
Mart´ın et al. (1999b) present direct evidence for the dra- Inthispaperwepresentdetailedlong-slitspectroscopic
matic effect that shocks can have in the properties of the observationsoffourpowerfulhigh-redshift(0.47<z<0.81)
extendedemission-line gas ina sampleof low-redshift radio radio galaxies, aimed at investigating the balance between
galaxies. The emission lines of these galaxies are resolved shocks and AGN-photoionization. With these observations
in two main kinematic components: a low-ionization broad we extend the approach of combining kinematical and ion-
component, spatially associated with the radio structures, ization information — previously used in studies of low-
andahigh-ionization narrowcomponent,extendingbeyond redshiftradiogalaxies(Clarketal.1997,1998;Villar-Mart´ın
the radio structures. Furthermore, a recent study of two et al. 1999b) — to the higher redshift regimes. Crucially,
nearby radio galaxies (Tadhunter et al. 2000) has revealed these new data cover the same rest wavelength range as
c 0000RAS,MNRAS000,000–000
(cid:13)
Jet-cloud interactions in high-z radio galaxies 3
previous studies of low-redshift radio galaxies, thereby al- using the Starlink DIPSO spectral analysis package. Gaus-
lowingdirectcomparisonstobemadebetweenlowandhigh- sians were fitted to the profiles of the emission lines. The
redshift objects. measuredlinewidthswereallcorrectedforthespectralres-
olutionoftheinstrumentviatheGaussianquadraturetech-
nique(w=[w2 w2 ]12). Theinstrumental widths(cor-
obs− inst
responding to the instrumental resolution), derived using
2 SAMPLE SELECTION
the night-sky lines and verified with arc lines, are listed in
Our sample consists of four 3C radio galaxies imaged by Table 1.
McCarthy, Spinrad & van Breugel (1995), with intermedi- The data have not been corrected for Galactic redden-
ate redshifts in the range 0.47 < z < 0.81. The upper red- ing.Noneoftheobjectsinoursamplehavelowlatitudeand,
shift limit was chosen to ensure that the rest-wavelength therefore, the local extinction is not large for any of them
range3400 –5200 ˚Awas accessible with optical instrumen- (0.022 E(B-V) 0.059).
tation. The galaxies are thefollowing: 3C 34 (z=0.690), 3C A≤Hubble con≤stant of H0 = 50 km s−1 Mpc−1 and a
330 (z=0.550), 3C 352 (z=0.8067) and 3C 435A (z=0.471). deceleration parameterofq0 =0.5areassumed throughout
All four objects were previously known as having extended this paper. Details of the resulting angular scale for each
emission-lineregionsalignedalongtheirradioaxes.Wealso galaxy are given in Table 2.
selected the sources so as to have different radio sizes rela-
tive to the corresponding host galaxies. 3C 34 and 3C 330
are large doubleradio sources, in which theradio hot spots
have passed far beyond the EELR. 3C 352 and 3C 435A 4 RESULTS
aremedium-sizedoubleradiosourceswithcloseassociations
4.1 3C 34
betweentheradioandopticalstructures,andwithemission
lineand radioemission on asimilar scale. Although 3C352 4.1.1 Previous observations
isthehighestredshiftedgalaxyofoursample,itsEELRare
3C 34 is a large 46 arcsec (365 kpc) classical FRII double
large enough to allow a spatially resolved analysis.
radio source, at a redshift z = 0.69. It was first optically
identified by Riley, Longair & Gunn (1980). Imaging by
McCarthy, Spinrad & van Breugel (1995) reveals that the
3 OBSERVATIONS, DATA REDUCTION AND galaxy lies in a rich compact cluster environment, and an
ANALYSIS extended[OII]λ3727 emission region of high-surface bright-
ness aligned with the radio source. A recent analysis of the
Long-slitspectroscopicobservationswerecarriedoutonthe
environment around 3C 34, based on K-band images, con-
night15/8/93for3C34andonthenights20-21-22/7/96for
firms that the galaxy resides in a rich cluster environment
therestofthesample,usingtheISISdouble-armedspectro-
(Best2000b).Despitethefactthattheradiosourceislarger
graphonthe4.2mWilliam HerschelTelescopeonLaPalma
than the emission-line nebula, Best, Longair & R¨ottgering
(Spain). Details of these observations are presented in Ta-
(1997)reportpossibleevidenceforastrongjet-inducedstar
ble1.BlueandredspectraweretakenusingtheTek1(0.357
formation. Their Hubble Space Telescope images of 3C 34
arcsec/pix) and EEV3 (0.335 arcsec/pix) CCD detectors,
showalongnarrowregionofblueemission orientatedalong
respectively.Theslitwasorientedapproximatelyparallelto
theradioaxisanddirectedtowardstheradiohotspot.They
theradio axis for each galaxy (see Table 1 for details).
propose that this aligned emission is associated with a re-
The data-reduction was carried out using the IRAF
gionofmassivestarformation,inducedbythepassageofthe
software packageand wasperformed inseveral stages: bias-
radiojetthroughagalaxywithintheclustersurrounding3C
subtraction (using the unilluminated region of the chip);
34. Based on Fabry-P´erot images, Neeser, Meisenheimer &
cosmic-rayremoval(manuallyusingtheIMEDITtask);flat-
Hippelein(1997)proposethattheextendedemission-linere-
fielding (using the normalized flat-field frame); wavelength
gion in 3C 34 is photoionized by anisotropic UV radiation
calibration (using copper-argon comparison frames for the
emitted by the central AGN. They find that this interpre-
blue spectra and copper-neon for the red spectra); atmo-
tation is energetically viable, but insufficient to explain the
spheric extinction correction, flux calibration and sky sub-
line-emission kinematicsobservedin3C34. Therefore, they
traction.Frommeasurementsofnight-skyemissionlines,the
suggest that the kinematics in the extended regions of 3C
absolute uncertainties in the wavelength calibration are es-
34 are best explained as gas swept aside by the lateral ex-
timatedtobe 0.2˚Aforthe300Bgrating,and 0.4˚Afor
∼ ∼ pansion of theradio source lobes.
the158B and 158R gratings. Comparison between different
spectrophotometric standard star observations gives a flux
calibrationerrorof 5percentacrossthewavelengthrange
∼ 4.1.2 The emission-line structure
withineachdataset.Followingthebasicreduction,thetwo-
dimensional spectra were corrected for any distortion along Since the airmass was low (1.073) for the observations of
thespectraldirection,usingtheStarlinkFIGAROpackage. 3C 34 and the slit wide, the effects of differential atmo-
Then,theredspectraofallthegalaxies,except3C34,were spheric refraction can be neglected. The spatial profiles of
resampled in the spatial direction to have the same pixel [OII]λ3727(solidline)and[OIII]λ5007(dashedline)arepre-
scale as theblue spectra (0.357 arcsec/pix). sentedinFigure1(top).Theywerederivedinthefollowing
Thetwo-dimensionalspectrawereallshiftedtothecor- way:a40˚Awideslice,perpendiculartothedispersiondirec-
responding rest frame of each galaxy before the analysis. tion and centred on the appropriate emission line, was ex-
One-dimensionalspectrawerethenextracted,andanalysed tracted; then, to correct for the continuum contamination,
c 0000RAS,MNRAS000,000–000
(cid:13)
4 C. Sol´orzano-In˜arrea et al.
a slice of the same width, taken from an adjacent line-free fluxesarenormalisedtotheobserved-frameHβfluxforeach
region of thespectrum, was subtracted from thefirst one. aperture. The fluxes have not been corrected for intrinsic
Intheseprofiles,lineemissionisseenextendingapprox- reddening, given that the Hγ/Hβ ratio is consistent with
imately 16 arcsec (127 kpc) along the radio axis. The posi- thetheoreticalvaluecorrespondingtoCaseBrecombination
tions of the radio hot spots lie far outside the EELR, and (Osterbrock 1989).
thereforetheyarenotindicatedontheplots.Inthenucleus, Figure 4 shows the variation in the [OII](λ3727) /
the [OII] emission peak is coincident with the continuum [OIII](λ5007) line ratio along the radio axis. Apart from
centroid,whilethe[OIII]emissionpeaksslightlytotheeast. ahigh-ionizationregion totheeastofthenucleus,thespec-
In addition to the central nuclear peak, the spatial profiles trashowconsistentlyalowionizationstateovermostofthe
show a pronounced peak 4.5 arcsec west of the nucleus. nebula([OII]/[OIII] > 0.7).
∼
It is noticeable that the [OII] emission is stronger than the
[OIII] emission in both the nucleus and the western EELR.
4.1.5 The physical conditions
This indicates a low-ionization state.
Unfortunately, it was not possible to measure the electron
temperaturefor3C34fromthe[OIII](4959+5007)/4363be-
4.1.3 The emission-line kinematics cause the[OIII]λ4363 emission line was not detected.
Kinematic information was extracted from the one-
dimensional spectra by fitting single Gaussians to the pro-
4.1.6 The continuum emission
filesofthestrongemissionlines[OII]λ3727and[OIII]λ5007.
Figure 2 (top) shows the variation in the velocity centroids Ithasbeendemonstrated(Dicksonetal.1995)thattheneb-
of these emission lines along the radio axis. Both the [OII] ular continuum emission (a combination of free-free, free-
and [OIII]velocities varyin asimilar manner,although the boundrecombination and two-photoncontinua,plushigher
[OIII]lineemissiondoesnotextendtotheeastasfarasthe order Balmer lines) makes a major contribution to the UV
[OII] line emission. A strong splitting ( 1000 km s−1) in continua of powerful radio galaxies wherever the emission
both [OII] and [OIII] lines is observed a∼t a radial distance lines have large equivalent widths. However, there is still
of 2 to 4.5 arcsec (16 to 36 kpc) on thewest side of the nu- some controversy about the relative importance of theneb-
cleusalongtheradioaxis.Itcanbeseenthatthesplittingis ular component in the UV continuum emission. The theo-
almost symmetric with respect tothenuclearvelocity,hav- reticalCaseBnebularcontinuumcanbegeneratedbyusing
ing the two components relative shifts of + 500 and – the Hβ flux (e.g. Osterbrock 1989). This calculation is not
500kms−1.Theextractedspectrum(1.7x∼1.5arcsec2 ap∼er- particularlysensitivetodensityandtemperaturevariations,
ture centred 3.7 arcsec west of the continuum centroid) onlytheintrinsicreddeningcouldgivesomeuncertainty;but
showing the ∼high-velocity components to the [OII] line is thisisnotaproblemsincenosignificantreddeninghasbeen
presented in Figure 3. measured in 3C 34.
Thevariationsinthelinewidthsalongtheradioaxisfor The spatial profile of the nebular continuum was cal-
the [OII] and [OIII] emission lines are shown in Figure 2 culated in the following way: a spatial slice, perpendicular
(bottom). The linewidths are found to be moderate, about to thedispersion direction, containing theHβ emission line
400kms−1onaverage,forboth[OII]and[OIII]alongthedi- was extracted from the rest-frame 2-D spectrum. Then, an
rectionoftheslit.Acrossmuchofthegalaxy,thelinewidths adjacent continuum slice of the same width was subtracted
of [OII] and [OIII] are consistent, except for the small re- from the first one to obtain the continuum-free Hβ profile,
gionsat 1arcsecdistancefrom thenucleusatbothsides, which was used to produce the nebular continuum profile
where th∼ere are small but significant radial velocity differ- (byusing theNEBCONT routinein DIPSO).
encesbetween[OII]and[OIII].Also,onthewestsideofthe Figure 1 (bottom) presents the spatial profile of the
nucleus, at the location of the splitting, it can be seen that continuumemissionalongtheradioaxisof3C34.Thesolid
the [OII] line tends to be slightly broader than the [OIII] line shows the UV continuum emission from 3400 ˚A – 3700
line; however, this could be due to the fact that the [OII] ˚A,thedashed lineshowstheblue-continuumemission after
line is a doublet ([OII]λ3726.0 and [OII]λ3728.8). thesubtraction of thenebularcontinuumcontribution, and
These results agree with those obtained by Neeser, the dotted line shows the continuum emission from 4500
Meisenheimer & Hippelein (1997), based on Fabry-P´erot ˚A – 4800 ˚A, scaled to the peak of the nebular-subtracted
imaging. Althoughtheydonotdetectthelinesplitting,the blueprofile.Itisfoundthatthecontributionofthenebular
large linewidths ( 1200 km s−1) they find in the western emission is 12 percent on the nucleusand 11 per cent
EELRareconsiste∼ntwithunresolvedsplitting.Theyexplain on the east∼ern EELR, for the wavelength ran∼ge 3400 ˚A –
thisasbeingtheoverlapoftwodistinctsources(thecentral 3700 ˚A. From Figure 1 (bottom), it can be seen that the
one and thewestern extended region). nebular continuum cannot account for the UV continuum
excess observed on both sides of the nucleus, especially on
thewest side.
4.1.4 The emission-line spectra
Integrated blue and red spectra of the eastern EELR, the 4.2 3C 330
centralnuclearregion andthewestern EELR of3C 34, and
4.2.1 Previous observations
a 2-D spectrum showing the [OII]λ3727 emission line are
presentedinFigure20.Theintegratedfluxesoftheobserved 3C 330, with a redshift z = 0.550, was first optically iden-
emissionlinesinthesethreeregionsaregiveninTable3.The tified by Spinrad et al. (1976). It is a large 62 arcsec (458
c 0000RAS,MNRAS000,000–000
(cid:13)
Jet-cloud interactions in high-z radio galaxies 5
kpc)doubleradiosource.Ground-based[OIII]λ5007andra- doubletcomponentsseparately;therefore,thiscouldslightly
dio continuum images (McCarthy, van Breugel & Kapahi broadenthemeasured[OII]linewidths(FWHM).Inthenu-
1991)revealthat3C330hasasmalllobedistanceasymme- clear and northern extended region, both [OII] and [OIII]
try, yet shows a large asymmetry in its extended emission linewidths follow smoooth curves, but they vary in a more
lines,which arestrongeron thesideoftheradiolobecloser chaotic way along the southern EELR. The Hβ linewidth
tothenucleus(North-East).MorerecentHSTnarrow-band couldonlybemeasuredinthecentralregionsofthegalaxy.
images show a narrow cone-like structureon the north-east Inthenucleusandnorthernsideofthenucleus,thelinewidth
side of the galaxy, which has been interpreted in terms of of Hβ varies in a similar manner as the [OIII] linewidth.
illumination by a quasar hidden in the core of the galaxy However, on the southern side of the nucleus, it appears to
(McCarthy 1997). However, the putative “ionization cone” beintermediate between thoseof the[OII]and [OIII]emis-
is narrower than expected on the basis of the broad cones sion lines.
predictedbytheunifiedschemesforpowerfulradiogalaxies High-resolution long-slit spectra, taken at the same
(Barthel1989).ContinuumimagesbyMcCarthy,Spinrad& position angle (PA 230o) by McCarthy, Baum & Spinrad
vanBreugel(1995) indicatethatthisobject liesinacluster (1996), reveal that the [OIII] line emission extends 17
∼
andhasaclosecompaniontothenorth.Incontrast,ROSAT arcsec along the slit. The results they obtain in the kine-
observations of 3C 330 reveal a relatively low X-ray lumi- matics are similar to ours, although they find a smooth ve-
nosity (Crawford & Fabian 1996), which is not compatible locity curve for the [OIII] emission line. The amplitude of
with the presence of a rich cluster in the environment of both the velocity gradient and the linewidths for [OIII] are
3C 330. Based on long-slit spectroscopy, McCarthy, Baum in agreement with what we obtain.
& Spinrad (1996) carried out kinematic analysis of 3C 330, Figure 7 (top) shows the Hβ, [OIII]λ4959 and
finding a large, smooth velocity gradient along PA 230o in [OIII]λ5007 emission-line profiles from the red spectrum of
[OIII]λλ4959,5007. 3C330,correspondingtoa2.14x1.35arcsec2 aperturecen-
tred at 3.39 arcsec south-west of the nucleus. Initially, a
single Gaussian fit (dashed line) was considered, but it is
4.2.2 The emission-line structure clear that the [OIII]λ5007 profile cannot be reproduced by
this simple fit. The broad wing seen in the [OIII]λ5007 line
Figure 5 (top) shows the variation of the [OII]λ3727 (solid
suggests theexistenceof abroad component,inaddition to
line) and [OIII]λ5007 (dashed line) fluxes along the radio
thenarrowcomponent,inthekinematicstructureof3C330.
axis of 3C 330. Since the radio hot spots of 3C 330 are
The two-component Gaussian fits (dot-dash-dot lines) and
locatedoutsidetheEELR,theyarenotshowninthefigures.
the total fit (dashed line) are presented in Figure 7 (bot-
Lineemissionisseenextendingapproximately14arcsec
tom).ItcanbenoticedthattheHβ profileisdominatedby
(103 kpc) along the radio axis, with a central nuclear peak
theemission of thebroad component.
and aseparated peak at 2arcsec to thenorth-east of the
∼ The variation in the velocity centroids along the ra-
continuum centroid. It can be noticed that the off-nucleus
dio axis of 3C 330 for the narrow and broad components
peak issignificantly larger for [OII]thanfor [OIII],indicat-
seen in [OII]λ3727 and [OIII]λ5007 and the correspond-
ing that the ionization state is much lower in the northern
ing linewidths are shown in Figure 8. Because of the high-
extendedregion than in thenucleus.
resolution of the blue spectra, it was possible to fit each
componentofthe[OII]doubletseparately,ineachkinematic
component, assuming the low-density limit for the doublet
4.2.3 The emission-line kinematics
ratio. A splitting in the [OII] emission line is detected at a
In contrast to the other ojects in our sample, the radialdistanceof2.5to7arcsec(18to51kpc)south-westof
intermediate-resolutionR300Bgratingwasusedfortheblue the nucleus. Note that the velocity centroids of the narrow
observations of3C 330, and thelow-resolution R158R grat- componentsvaryinasimilarway,followingasmoothcurve,
ingfortheredobservations,givinginstrumentalresolutions for [OII] and [OIII] in the southern region; in contrast, the
of 4.3 ˚A and 7.4 ˚A, respectively. Figure 6 (top) shows the broad components have a more distorted behaviour, giving
variation in the velocity centroids of [OII]λ3727, Hβ and theshift to thetotal velocities shown in Figure 6 (top).
[OIII]λ5007 along the radio axis of 3C 330. The [OII], Hβ
and [OIII] velocities vary in a similar manner along the
4.2.4 The emission-line spectra
northernextendedregion, from 1to 5arcsec. However,a
∼
clear contrast in thevelocity centroidsof [OII]with respect Two-dimensionalspectrumshowingthe[OII]λ3727emission
to [OIII] starts to appear in the nuclear region and reaches line, and integrated blue and red spectra of four spatial re-
a maximum of 200 km s−1 of difference between both gions(southernEELR,nucleus,n1-andn2-northernEELR)
∼
velocities at the extreme of thesouthern EELR. Along this along the radio axis of 3C 330 are presented in Figure 21.
extendedregion,the[OII]velocitycentroidsfollowasmooth Theintegratedfluxesoftheobservedemissionlinesinthese
curve,butthe[OIII]velocitycentroidsvaryinachaoticway. regionsaregiveninTable4.Thefluxesarenormalisedtothe
Figure 6 (bottom) shows how the linewidths of observed-frame Hβ flux for each region. Since the Balmer-
[OII]λ3727, Hβ and [OIII]λ5007 vary along the radio axis line intensities (Hδ and Hγ) relative to Hβ are not signifi-
of 3C 330. The low-ionization [OII] line is slightly broader cantlylowerthanexpectedinCaseBrecombination(Oster-
thanthehigh-ionization[OIII]linealmosteverywherealong brock 1989) in the extended emission-line regions, the off-
theradioaxis,exceptatbothextremesoftheslitwherethe nuclearlineratioswerenotcorrectedforintrinsicreddening.
errors are bigger and it is not so clear. However, the [OII] However, there is evidence for significant reddening in the
line was fitted with a single Gaussian, instead of fitting the nuclearaperturewithbothHδandHγsignificantlylessthan
c 0000RAS,MNRAS000,000–000
(cid:13)
6 C. Sol´orzano-In˜arrea et al.
expected for Case B recombination (EB−V = 0.59 0.13). interferometryof3C352,andconcludethatthemorphology
±
The extinction implied by these results (A =2.4 0.5) is and velocity structure of the extended emission-line region
B
±
consistent with that inferred for the nuclear region of low- around3C352canbeexplainedintermsofcoolinggasion-
redshift radio galaxies (Tadhunter, Metz & Robinson 1994; ized and accelerated by the bowshock associated with the
Robinson et al. 2000). radio jet, while some ionization close to the centre could
Figure 9 shows the variations in the also be due to radiative ionization. Based on long-slit spec-
[OII]λ3727/[OIII]λ5007 and [OIII]λ5007/Hβ line ratios troscopy observations, Best, R¨ottgering & Longair (2000a)
along the radio axis of 3C 330. It can be seen that the nu- study the UV emission-line ratios [CIII]λ2326/[CII]λ1909
clear region is of high ionization with 10 < [OIII]/Hβ < and[NeIII]λ3869/[NeV]λ3426of3C352,andcomparethem
13. The ionization state decreases to the northern side of with theoretical predictions of shock and photoionization
the nucleus up to 1 arcsec, then rises up to 2 arcsec and models, finding that the line ratios are in agreement with
∼
decreasesagaintowardstheextremeofthenorthernEELR. shock models including a precursor region.
OnthesouthernEELRthe[OII]/[OIII]lineratioshowslittle
significant variation, but has a higher ionization state than
thenorthernEELR.Unfortunately,onlyonepointcouldbe
4.3.2 The emission-line structure
obtained along the southern region for the [OIII]/Hβ line
ratio. Figure 10 (top) presents the [OII]λ3727 (solid line) and
[OIII]λ5007(dashedline)spatialprofilesalongtheradioaxis
of 3C 352. They were derived by extracting spatial slices,
4.2.5 The physical conditions following thesameprocedureoftheothergalaxies (seeSec-
tion4.1.2).Itcanbeseenthatthe[OII]lineemissionextends
The electron temperature T was measured for the
e
10 arcsec (82 kpc) along the radio axis, while the [OIII]
central region and the EELR of 3C 330, by using ∼
line emission only extends 6 – 7 arcsec (53 kpc) along
the [OIII](4959+5007)/4363 line ratio. The [OIII]λ4363 ∼
thesamedirection.Theradiohotspotsinthisradiogalaxy
linewidth was constrained to be the same as that of
lie just outside the EELR, and they are not shown in the
[OIII]λ5007. The values obtained from the red spectra are
figures.
[OIII](4959+5007)/4363 = 33 7 for the southern EELR,
±
51 4 for thenucleus (after reddeningcorrection), 66 8 for
± ±
the n1-northern EELR and 44 7 for n2-northern EELR,
which implies electron temperat±ures ofT = 23000+5000 K, 4.3.3 The emission-line kinematics
e −3000
17600+500 K,15300+900 Kand19000+2000 K,respectively.
A den−si1t0y00of n = 10−08c00m−3 was assum−e1d50.0 Figure 11 (top) shows how the velocity centroids of
e [OII]λ3727 and [OIII]λ5007 vary along the radio axis of 3C
352. They follow a smooth rotation curve and both vary in
4.2.6 The continuum emission a similar manner.
The variation in the linewidths of [OII]λ3727 and
Figure 5 (bottom) shows spatial profile of the continuum [OIII]λ5007 is presented in Figure 11 (bottom). Both [OII]
emission along the radio axis of 3C 330. The solid line rep- and[OIII]linewidthsvaryinasimilar way.However,in the
resentstheprofileof theUVcontinuumemission (3200 ˚A– centralregionitappearsthatthelow-ionization [OII]lineis
3700˚A),thedashedlinerepresentstheUVcontinuumafter broader than the high-ionization [OIII] line. From this fig-
the subtraction of the nebular continuum (calculated from ure,itcanalsobenoticedthatthe[OIII]lineemission does
the rest-frame Hβ flux along the slit, in the same way as not extend as far as the [OII] line emission along the radio
that of 3C 34, by using the NEBCONT routine in DIPSO axis.
(seeSection4.1.6)),andthedottedlinerepresentsthegreen Figure 12 shows the Hβ, [OIII]λ4959 and [OIII]λ5007
continuum emission (5200 ˚A – 5700 ˚A), scaled to the peak emission-line profiles from thered spectrum of 3C352, cor-
of the nebular-subtracted UV profile. A UV excess can be respondingtoa1.07x1.56arcsec2 aperturecentredat0.71
observed approximately between 2 and 4 arcsec north-east arcsec south-east of thenucleus. A single Gaussian fit,con-
ofthenucleus.Itisfoundthatthecontributionofthenebu- sideringlineemissionfromonlyonekinematicalcomponent,
laremission is 50percentoftheUVcontinuumemission isshownatthetop.However,thebroadwingsin[OIII]sug-
∼
inthenucleusof3C330, and 35percentinthenorthern gest the existence of a broad component in the kinematic
∼
EELR,forthewavelengthrange3200˚A–3700˚A.However, structure of 3C 352. The two-component fits (dot-dash-dot
thenebularcontinuumcannotaccountfortheobservedUV lines)andtheresultingfit(dashedline)arepresentedatthe
excessatthenorth-eastsideofthenucleus,sincethisexcess bottom of Figure 12. It can be seen that the broad compo-
still remains after thenebularsubtraction. nent dominates theemission of Hβ.
Figure13(top)showsthevariation ofthevelocity cen-
troidsalongtheradioaxisof3C352forthebroad andnar-
4.3 3C 352
row components seen in [OII]λ3727 and [OIII]λ5007. The
4.3.1 Previous observations velocity centroids of Hβ are also plotted. The variation in
the linewidth along the radio axis for the two kinematical
3C352isacompact10.2arcsec(84kpc)doubleradiosource
componentsseenin [OII]λ3727 and[OIII]λ5007 isshown in
at redshift z = 0.8067 (McCarthy, Spinrad & van Breugel
Figure 13 (bottom) together with thelinewidth of Hβ.
1995). It was first studied optically by Smith et al. (1979)
who found a low-ionization emission-line spectrum. Hip-
pelein & Meisenheimer (1992) presentFabry-P´erot imaging
c 0000RAS,MNRAS000,000–000
(cid:13)
Jet-cloud interactions in high-z radio galaxies 7
4.3.4 The emission-line spectra 4.4.1 Previous observations
Figure22 showsintegrated blueandred spectraof thecen- The radio source 3C 435A, at a redshift z = 0.471 and
tral nuclear region, and the southern and northern EELR with an angular size of 14 arcsec (97 kpc), was discov-
of 3C 352, a 2-D spectrum showing the [OII]λ3727 emis- ered at a small projected distance from the powerful radio
sion line is also plotted. Fluxes of the emission lines in the source 3C 435B (McCarthy, van Breugel & Spinrad 1989),
threeregionsarelistedinTable5,andarenormalisedtothe but the two objects are at different redshifts. Narrow-band
observed-frame Hβ flux in each region. Given that the Hδ [OII] emission-line imaging shows presence of an extended
andHγ fluxesrelativetoHβ areconsistent with Case Bre- emission-line nebula that extends well beyond the compact
combination (Osterbrock 1989), no corrections for intrinsic radio source on both sides of the nucleus (van Breugel &
reddeninghavebeen made tothe line fluxes. McCarthy 1989). Bidimensional spectroscopy hasbeen per-
Thevariationinthe[OII]λ3727/[OIII]λ5007alongthe formedwiththeintegralfieldspectrographTIGER(Rocca-
radio axis is shown in Figure 14 (top). It can be observed Volmerange et al. 1994), showing at least five components
thattheionizationstatepeakswherethecontinuumcentroid around the central one. A scenario of gas being overpres-
issituated,andthenfallsoneithersideofthenucleus,more suredbyradioplasmaexpansion,orbytheinteractionofrel-
steeply towards thenorthern side, as theradius increases. ativisticelectronswiththeambientgas,wasfavoured.Kine-
Figure 14 (bottom) shows the [OIII]λ5007 / Hβ line maticanalysisbyMcCarthy,Baum&Spinrad(1996),based
ratio along theradio axis of 3C 352. Unfortunately,the Hβ on long-slit spectroscopy along PA 229o, shows a complex
linecouldonlybemeasuredinthenuclearregion.However, velocity profile, with an overall amplitude of afew hundred
fromthisfigureitcanbenoticedthatthenucleushasahigh km s−1, and apparent rotation in both the central galaxy
ionization state ([OIII]/Hβ 10). and in thecomponent lying 4 arcsec north-east.
∼ ∼
4.4.2 The emission-line structure
4.3.5 The physical conditions
The spatial profiles of the [OII]λ3727 (solid line) and
The electron temperature T was measured for the
e [OIII]λ5007(dashedline)emissionlinesarepresentedinFig-
central nuclear region of 3C 352 by using the
ure15(top).Theseprofileswereobtainedinthesamewayas
[OIII](4959+5007)/4363 line ratio; but this measurement
those of the other four radio galaxies, by extracting spatial
was not possible for the two EELR. The value obtained for
slices through thelong-slit spectra (see Section 4.1.2).
thenucleusfrom thered datais [OIII](4959+5007)/4363 =
46 7, which implies a temperature T = 18000+2000 K for The projected positions of the radio hot spots are in-
the±nuclearregion.Adensityofn =10e0cm−3wa−s10a0s0sumed. dicated on the plots by dashed vertical lines. Line emission
e
extends approximately 24 arcsec (166 kpc) along the radio
axis. In addition to the central galaxy, a companion can
beseen located at 4–5arcsec (30kpc)north-east of the
∼
4.3.6 The continuum emission continuumcentroidalongtheradioaxis(seealsoMcCarthy,
vanBreugel&Spinrad1989). Thecompanion issituatedin
The spatial profile of the continuum emission along the ra-
the interaction zone of the radio jet with the intergalactic
dioaxisof3C352isshowninFigure10(bottom).Thesolid
medium (Rocca-Volmerange et al. 1994). The fact that the
line shows theUV (2600 ˚A- 3700 ˚A) continuumprofile, the
[OII] line is relatively stronger in this component indicates
dashedlineshowstheUVcontinuumafterthesubtractionof
that it has lower ionization state than the central compo-
the nebular continuum contribution (derived from the rest-
nent.
frameHβfluxalongtheslit,byusingtheNEBCONTroutine
inDIPSO(seeSection4.1.6)),andthedottedlineshowsthe
4400 ˚A – 4800 ˚A continuum profile, scaled to the peak of
4.4.3 The emission-line kinematics
the nebular-subtracted UV profile. The emission from the
starsituatedjustabout6.5arcsecnorth-westof3C352was Since the slit positions and the grating angles were coinci-
subtracted before extracting the continuum profile of the dent for the red spectra taken on two different nights, the
radio galaxy. It is found that the nebular emission contri- individualframeswerecoaddedtoproduceasinglelong-slit
bution with respect to the continuum emission is 37 per spectra, and therefore increasing theS/N ratio.
∼
cent in the nucleus and in the southern EELR, and 28 Figure 16 (top) shows how the velocity centroids of
∼
per cent in the northern EELR, for the wavelength range [OII]λ3727 and [OIII]λ5007 vary along the radio axis of 3C
2600 ˚A – 3700 ˚A. There is no evidence for excess UV con- 435A. Both [OII] and [OIII] velocities are seen to vary in a
tinuumfollowing thesubtraction ofthenebularcontinuum: similar manner, with rotation curve-like velocity variations
the nebular-subtracted UV and longer wavelength contin- in both central and northern components.
uum spatial profile are not significantly different (although The variation in the linewidths of [OII]λ3727 and
thepresenceofthenearbybrightstarmakesthecomparison [OIII]λ5007 along the radio axis of 3C 435A is shown in
difficult). This behaviour is similar to that observed in the Figure 16 (bottom). The low-ionization [OII] line tends to
low-redshift jet-cloud interaction 3C 171 (Clark 1996). bebroaderthanthehigh-ionization[OIII]line,especiallyin
the region between the nucleus and the northern radio hot
spot. Also, in the northern EELR the [OIII] emission line
seemstobemuchnarrowerthanthe[OII]emission line, be-
4.4 3C 435A
yond the radio hot spot. In this case, the [OIII] linewidth
c 0000RAS,MNRAS000,000–000
(cid:13)
8 C. Sol´orzano-In˜arrea et al.
might have been affected by the subtraction of the night- 4.4.6 The continuum emission
sky line that falls on it, since the S/N ratio in theEELR is
The spatial profile of the continuum emission along the ra-
muchlower than in thecentralregions. However, the[OIII]
dio axis of 3C 435A is shown in Figure 15 (bottom). The
linewidth results agree with those of McCarthy, Baum &
solid line represents the UV (2600 ˚A – 3700 ˚A) continuum
Spinrad (1996).
emission,thedashedlinerepresentstheUVcontinuumafter
Figure 17 shows the [OII]λ3727 emission-line profile of
3C 435A, corresponding to a 2.50 x 1.56 arcsec2 aperture thesubtractionofthenebularcontinuumcontribution(cal-
culated from therest-frame Hβ fluxalong theslit, byusing
centredat1.43arcsecnorth-eastofthecontinuumcentroid.
the NEBCONT routine in DIPSO (see Section 4.1.6)), and
Initially,aone-componentGaussianfitwasconsidered(top),
thedottedlinerepresentsthegreen(5200 ˚A–6000 ˚A)con-
butthissimplefitcannotreproducethebroadwingof[OII],
tinuumprofile,scaledtothepeakofthenebular-subtracted
whichcanbeseenasemission fromabroadkinematiccom-
UV profile. It is found that the nebular emission with re-
ponent. The plot at the bottom of Figure 17 presents the
spect to the continuum emission is 10 per cent in the
two-component (narrowand broad) Gaussian fit(dot-dash- ∼
nucleus, and 21 per cent at the location of the northern
dot lines) and the total fit (dashed line), which reproduces ∼
component, for the wavelength range 2600 ˚A – 3700 ˚A. On
the[OII] profile muchbetter than theone-component fit.
thebasisofthecomparisonbetweenthenebular-subtracted
Figure 18 (top) shows how the velocity centroids vary
UVandtheredcontinuumslices,thereisevidenceforaUV
along the radio axis of 3C 435A for the broad and narrow
excessonbothsidesofthenucleus,especiallyatthelocation
components seen in [OII]λ3727. The [OIII]λ5007 velocity
ofthenorthernhotspot.Note that the extended UVcontin-
centroidsarealsoplotted.Nobroadcomponentcouldbefit-
uum emission (nebular-subtracted) extends over the entire
tedtothe[OIII]λ5007 emission lineat anyspatial location.
emission line nebula and well beyond the radio hot spot on
Itcanbeseenthatthenarrowcomponentof[OII]variesina
the north-east side of the galaxy.
similar way as the [OIII]velocity centroids, consistent with
rotation, while the broad component has a more distorted
andchaoticbehaviour.Thevariationinthelinewidthsalong
the radio axis for the two kinematic components seen in
5 DISCUSSION
[OII] is presented in Figure 18 (bottom). The linewidth of
the broad component varies between 800 – 1500 km s−1 5.1 Summary of results
∼
(FWHM), reaching the maximum values in the region be-
We summarise below some results derived in the previous
tween the continuum centroid and the northern radio hot
section:
spot.
Extremekinematics: Allthesourcesshowdisturbedoff-
•
nuclear emission line kinematics, including: line splitting in
4.4.4 The emission-line spectra 3C34and3C330;andthedetectionofbothbroad(FWHM
= 1000 – 1500 km s−1) and narrow (FWHM < 600 km
Integrated blue and red spectra of the southern EELR,
s−1) components in the extended emission line∼regions of
thecentralnuclearregion,thenortherncomponent andthe
3C 330, 3C 352 and 3C 435A. We emphasise the disturbed
northern EELR of 3C 435A, and a 2-D spectrum showing
kinematicsaredetectedinallthesourcesinoursample,not
the[OII]λ3727emissionlinearepresentedinFigure23.The
just those in which the radio and emission line structures
fluxes of the observed emission lines in the four regions are
havea similar scale.
listed in Table 6, and are normalised to theobserved-frame
Relationshipbetweenlinewidthandionizationstate: Al-
Hβ flux for each region. The line fluxes have not been cor- •
though we do not see evidence for an anticorrelation be-
rected for intrinsic reddening, given that the Hα/Hβ ratio
tween ionization state and linewidth as strong as seen in
is not significantly larger than the theoretical value corre-
some low-redshift jet-cloud interaction sources (Clark et al.
sponding to Case B recombination (Osterbrock 1989), sug-
1997, 1998; Villar-Mart´ın et al. 1999b), the low-ionization
gesting that theintrinsic reddening is not large.
[OII]λ3727 and Hβ emission lines tend to be broader than
Figure 19 shows the [OII]λ3727/[OIII]λ5007 and the
the high-ionization [OIII]λ5007 emission line in the regions
[OIII]λ5007/Hβ lineratiosalongtheradioaxisof3C435A.
wherethecomplexkinematicstructuresdescribedaboveare
From these plots it can be noticed that the region between
observed,especially in thecentral region of 3C 352.
the nucleus and the northern radio hot spot has the lowest
Overall ionization state: The ionization state peaks in
ionizationstate,1<[OIII]/Hβ <2,although,theionization •
the near-nucleus regions of the galaxies and decreases to-
state is seen to be low everywhere. It can also be seen that
wards the EELR. In particular, in 3C 435A a minimum in
the nucleus has higher ionization state than the northern
the ionization state is observed at 2 arcsec north-east of
companion located 4 – 5 arcsec to thenorth. ∼
∼ thenucleus, just behindthe radio hot spot.
High electron temperatures: The high electron temper-
•
atures measured in the extended regions of 3C 330 and in
4.4.5 The physical conditions
thecentralregion of3C352areinconsistentwithphotoion-
For 3C 435A, the [OIII]λ4363 was not measured because izationbyacentralAGN(Tadhunter,Robinson&Morganti
it was outside the wavelength range covered. Therefore, 1989a),buttheycanbeeasilyexplainedintermsofjet-cloud
the temperature diagnostic could not be obtained for this interactions (Villar-Mart´ın et al. 1999b).
galaxy. Extended UV continuum: The contribution of the neb-
•
ularcontinuumemission totheUVcontinuumissignificant
inallcases,withthenebularcontributionrangingfrom 10
∼
c 0000RAS,MNRAS000,000–000
(cid:13)
Jet-cloud interactions in high-z radio galaxies 9
percentinthenucleiof3C34and3C435A,to 40percent 1994; Villar-Mart´ın et al. 1999b) and/or in the turbulent
∼
in the extended regions of 3C 330. However, following sub- boundary layers of the radio jets is required. Furthermore,
traction of the nebular component, a significant UV excess fast winds from the quasars and starbursts associated with
remains in theextended nebulaein most of theobjects. thehostgalaxiescouldalsocontributetothelargevelocities
observedinthesesources(Heckman,Armus&Miley1990).
Further evidence for jet-cloud interactions is provided
5.2 Kinematics: The evidence for shocks
bytheexistenceofbroadkinematiccomponentsintheemis-
Inprevioussectionswehavepresentedevidenceforextreme sion lines of some of the radio galaxies. We have detected
emission-linekinematicsinalltheobjects.Belowwediscuss underlyingbroadcomponentswithlargelinewidths(FWHM
theimplications of these results. 1000 – 1500 km s−1) in the line profiles of three galaxies
∼
Line splitting is observed in the western EELR of 3C inoursample(3C330,3C352and3C435A),inadditionto
34 and 3C 330 (∆v 1000 km s−1 and 500 km s−1, re- narrow components (FWHM < 600 km s−1). The large 62
spectively). These v∼elocity shifts are lar∼ger than what we arcsec (458 kpc) radio source∼3C 330 shows an underlying
would expect from gravity of an isolated elliptical galaxy broad component in both southern and northern EELR, at
(∆v < 400 km s−1) (e.g. Tadhunter, Fosbury & Quinn radial distances of 1.5 – 4 arcsec (11 – 30 kpc) from the
∼
1989b, Baum, Heckman & van Breugel 1990). Previously, nucleus. Again, evidence for interaction between the radio
high-velocitycomponentshavealsobeenobservedintheex- and optical structures can be detected in the EELR, even
tendedregionsofpowerfulradiogalaxies (Tadhunter1991), when the major radio components have passed beyond the
bothatlowredshifts(inCygnusA)andathighredshifts(in EELR. In the case of 3C 352, in which the peaks of the ra-
3C265).Suchsplittings,whicharecharacteristicofexpand- dio lobes are located just outside the EELR, an underlying
ing shells of material, are most likely the result of interac- broadcomponenthasalsobeendetectedatbothsidesofthe
tionsbetweentheradioplasmaandtheinterstellarmedium. nuclear region. 3C 435A is theonly object of our sample in
3C 34 is a large double radio source, which extends 46 arc- which the radio hot spots have not passed yet beyond the
sec (365 kpc); however, the line splitting is detected at a EELR.Inthisobject, abroad kinematiccomponent canbe
radial distance of 2 to 4.5 arcsec (16 to 36 kpc) from detected in [OII]λ3727 at both sides of the nucleus. In all
∼ ∼
thenucleus.Similarly,3C330 isalarge 62arcsec (458kpc) of these cases, the linewidths of the broad components are
double radio source, but we detect a line splitting in [OII] largerthanwouldbeexpectedforgravitationalacceleration
at a radial distance of 2.5 to 7 arcsec (18 to 51 kpc) in the potential of a giant elliptical galaxy.
∼ ∼
fromthenucleus.Thisindicatesthattheeffectsoftheinter- Similar kinematically disturbed components have also
actionsbetweentheradiocomponentsandthecloudsofgas been observed at low redshifts in other powerful radio
are also important in galaxies with extended radio sources, galaxies with jet-cloud interactions. High-dispersion long-
in which the main radio emitting components have passed slit spectra of thelow-redshift radio galaxies 3C 171 (Clark
beyond theEELR. et al. 1998) and PKS 2250-41 (Villar-Mart´ın et al. 1999b)
As mentioned above, these splittings are too large to have revealed multi-component emission-line kinematics:
beduetogravitationalmotionsofthehostgalaxies.Onthe narrow components, with high-ionization state, and broad
other hand,it is not yet clear whether direct shock acceler- components, with low-ionization state. In the case of PKS
ation associated with the radio components is sufficient to 2250-41, the narrow component extends beyond the radio
accelerate the clouds to such high velocities (Villar-Mart´ın structure, while the broad component is found to be spa-
et al. 1999b). Considering the warm clouds (T 104 K) to tially associated with the radio emitting structures and is
beinequilibriumwithahotambientphase(T ∼107K),the believed to represent gas cooling behind the shock fronts.
∼
densitycontrast between thetwo phasesis 1000, and the As argued by Villar-Mart´ın et al. (1999b), it is likely that
∼
velocity of the clouds due to the advance of the radio jet is the gaseous component associated with the broad lines is
vc ∼vspρρhc (vs:shockvelocity;ρh:densityofthehotphase; accelerated by entrainment in the faster moving hot ISM
ρ :densityofthewarmclouds).Assumingthatthehotspot behind theshock front.
c
advancespeedis 0.01c—0.1c (Scheuer1995), theclouds However, we have to be cautious in interpreting the
will beaccelerated∼tovelocities of 100 —1000 km s−1 in broadcomponentsolelyintermsofjet-inducedshocks,since
∼
the direction of the jet, but the velocities perpendicular to abroadcomponentisdetectedwellbeyondthenorthernra-
theradioaxiswill bemuchsmaller. Bymeasuringtheratio diohotspotin3C435A.Thisissimilartothephenomenon
betweenthelengthoftheradioarmsandtheextensionper- observed in the high-redshift radio galaxy MRC1558-003,
pendiculartotheradioaxes,thelateralvelocitiescanbees- which presents high velocities and large linewidths in the
timated.Weobtainlateralextensionsrelativetothelengths EELR beyond the radio structures (Villar-Mart´ın, Binette
oftheradioarmsof 0.4inthecaseof3C34,and 0.15for & Fosbury 1999a). This suggests that another accelerating
∼ ∼
3C330 (seeradiomapsinBest, Longair &R¨ottgering1997 mechanism, apart from jet-cloud interactions, may be at
and McCarthy, van Breugel & Kapahi 1991, respectively), work.Nonetheless,eveninthecaseof3C435A,thebroadest
implyinglateralvelocitiesof 15—400kms−1.Giventhat emissionlinesaremeasuredinaregionbehindtheradiohot
∼
the radio sources 3C 34 and 3C 330 are likely to be close spot. Another common characteristic in the kinematics of
to edge-on and on much larger scale than the emission-line the galaxies in our sample is that the [OII]λ3727 emission
structures,weexpectthehigh-velocitycomponentsobserved line tends to be broader than the [OIII]λ5007 emission line
in these galaxies to be due to the lateral expansion of the intheregionswherethedisturbedkinematicsareobserved.
cocoons. Therefore, the bowshock acceleration may not be This could bebecause the[OII]line is adoublet and it was
sufficienttoexplainthelargevelocities.Someentrainmentof fitted with a single Gaussian. To quantify this broadening,
the clouds in thepost-shock wind (Klein, McKee & Colella wesimulated[OII]emissionlinesofdifferentwidthsandfit-
c 0000RAS,MNRAS000,000–000
(cid:13)
10 C. Sol´orzano-In˜arrea et al.
tedthemwithbothsingleandtwoGaussians.Wefoundthat tios with the predictions of both AGN photoionization and
fittingthetwocomponentsofthedoubletwithasingleGaus- shock-ionization models.
sianmadethelineonly 50kms−1 broaderthanifthetwo Figure 24 shows the resulting diagnostic diagrams,
∼
componentswerefittedseparately.However,ithappensthat which involve the line pairs: [NeV]λ3426 & [NeIII]λ3869,
[OII] is broader by approximately 300 km s−1 in 3C 34 at [OII]λ3727 & [OIII]λ5007, Hβ & [OIII]λ5007, HeII(4686)
thelocation ofthesplitting,andin3C330wherethebroad & Hβ and the electron temperature diagnostic line ratio
componentisdetected,correspondingalsotolow-ionization [OIII](4959+5007)/4363. Emission line pairs involving the
regions (see Figures 4 and 9 (bottom), respectively). In the same element are particularly useful for the diagnostic di-
case of 3C 352, the [OII] line is observed to be 200 km agrams because the dependence of the line ratios on the
s−1 broader than [OIII] in the central region, w∼ithin 1 elementalabundancesisrelatively minor(Baldwin, Phillips
∼
arcsec distance on either side of the nucleus, corresponding &Terlevich1981).Also,ifthetwoemissionlinesinthepair
with the region where the underlying broad component is arecloseinwavelength,thepossibleerrorsarisingfrom cal-
detected.In 3C 435A, the[OII]emission lineis observed to ibration and reddening corrections are minimized.
bemuchbroaderthan[OIII]intheregionbehindthenorth- To generate the AGN model predictions we used the
ernhotspot,coincidentwithalow-ionizationstate(seeFig- multipurposephotoionizationcodeMAPPINGS.Theresults
ure 19). In addition, at a radial distance of 6 – 12 arcsec areplottedassolidlines,whichrepresentthelineratiospro-
∼
northofthenucleusof3C435A,beyondthenorthernradio ducedbyoptically-thick,single-slab,power-lawphotoioniza-
hot spot,and coincident with ahigh-ionization-state region tionmodels(F να)withspectralindicesofα=-1.0,-1.5
ν
∝
(see Figure 19 (top)), the [OII] line appears to be 500 and-2.0(fromtoptobottom),andasequenceintheioniza-
kms−1 broaderthan[OIII].However,sincetheS/Nr∼atiois tion parameter covering the range 2.5 10−4 < U < 10−1.
low in thisextended region, the linewidth of [OIII]λ5007 in Adensityof100cm−3 andsolarabund×anceswereassumed.
this object is likely to be affected by the subtraction of the The shocks models are taken from Dopita & Suther-
night-skyline that falls on the[OIII] emission line. land(1995,1996).Weincludegridsofsimpleshocks(dashed
An anticorrelation between linewidth and ionization lines) and shocks with the photoionized precursor included
state is a common feature of radio galaxies in which jet- (50percentshocks+ 50percentprecursor) (dottedlines).
cloud interactions are taking place: low-ionization lines are The two main parameters for controlling the post-shock
broader than high-ionization lines. This phenomenon has emission line spectrum are the velocity of the shock (150
beenobservedinthepowerfullow-redshiftradiogalaxies3C v 500kms−1)andthemagneticparameter(0 B/√n
s
171(Clark etal.1998) andPKS2250-41 (Clark etal.1997, ≤ 4 µ≤Gcm−3/2). Each sequence in the Figures corr≤esponds
≤
Villar-Mart´ın et al. 1999b). This anticorrelation is difficult to a fixedB/√n valueand a changing v .
s
toexplainintermsoftheAGNilluminatingtheundisturbed In addition to the photoionization and shock models,
ambientgasinthehostgalaxy,butitcanbeeasilyexplained we have also plotted mixed-medium photoionization mod-
intermsofshocks.Thelineemission containscontributions els, including both matter-bounded and radiation-bounded
from both high-ionization-state gas, with narrow velocity clouds (dot-dash-dot line) from Binette, Wilson & Storchi-
dispersion, emitted in the undisturbed precursor zone, and Bergmann (1996). These combine an optically-thick zone
low-ionization-state gas, with a much broader velocity dis- (ionization-bounded[IB]component),illuminatedbyapho-
tribution, emitted by the disturbed cooling gas behind the ton spectrum that is initially filtered through an optically-
shock. Given the similarity between the kinematical prop- thin zone (matter-bounded [MB] component). The se-
erties of the objects in our sample and the low-redshift jet- quences of line ratios were derived by allowing the ratio
cloud interaction objects, it is plausible that the broad and (A ) of the solid-angle subtended from the nuclear pho-
M/I
narrow components in our sample arise in a similar way at toionizing source by the MB clouds relative to that sub-
all redshifts. tendedbytheIBcloudstovaryintherange10−4 A
≤ M/I ≤
Overall, the complex kinematics observed in the ex- 10. Since by definition the MB clouds lie somewhere in be-
tended regions of the galaxies in our sample indicate that tween the central source and the IB clouds, strictly speak-
the material has been highly perturbed and accelerated by ing, A cannot be lower than unity. However, based on
M/I
theinteraction between theradio emitting componentsand theunifiedschemesandduetoorientationeffects,somefrac-
theambientgas.Weseeevidenceforthisshockacceleration tion of theMB clouds located in theinner regions could be
eveninthosegalaxieswiththemoreextendedradiosources obscuredfromtheobserver(e.g.byadustytorus),therefore
(3C34and3C330),inwhichtheradiohotspotshavepassed giving a physical meaning to an apparent A < 1.
M/I
well beyondthe EELR. Forcomparison,wehavealsoincludeddatacorrespond-
ing to the EELR of low-redshift radio galaxies. The crosses
represent objects in which strong jet-cloud interactions are
5.3 Ionization mechanism: AGN photoionization
taking place: PKS 2250-41, 3C 171, 4C 29.30, Coma A
or jet-induced shocks ?
(Clark1996),andPKS1932-464(Villar-Mart´ınetal.1998).
Although it is likely that acomponent of theISMhas been The open triangles represent the photoionized radio galaxy
accelerated by jet-induced shocks in all the sources in our 3C321whichisconsistentwithmixed-mediumphotoioniza-
sample, this does not necessary imply that the clouds are tion models (Robinson et al. 2000).
shock-ionized.Forexample,cloudscouldbeacceleratedand Figure24(a)showsthe[OIII](5007)/Hβ lineratioplot-
compressed by jet-induced shocks, but then photoionized ted against [OII](3727)/[OIII](5007). This combination of
by an illuminating AGN in the core of the host galaxy af- emission line pairs was first used by Baldwin, Phillips &
ter they have cooled. To address the issue of the ionization Terlevich (1981) to distinguish between the four predomi-
mechanism,inthissectionwecomparethemeasuredlinera- nantionizationmechanisms(HIIregions,planetarynebulae,
c 0000RAS,MNRAS000,000–000
(cid:13)