The Link Between Extended Line Emission and … Observations and data reduction 85 and line...
Transcript of The Link Between Extended Line Emission and … Observations and data reduction 85 and line...
Chapter 4.
Evidence for shock excitation in
line-luminous brightest cluster galaxies
In this chapter, we investigate the emission-line properties of a sample of brightest cluster
galaxies that occupy the high end of the BCG emission-line luminosity distribution,
having L(Hα) � 1042 erg s−1. We have selected six galaxies that are known to produce
strong low-ionisation emission lines and, in some cases, to host extended optical emission-
line regions. These include three well-known and well-studied cool-core systems: Abell
2597, PKS 0745-19, and ZwCl 3146. Table 4.1 outlines the properties of the BCGs,
including integrated X-ray, radio, and Hα line luminosities. Throughout the rest of this
chapter, we refer to the BCGs by the abbreviated names given in brackets in the first
column of this table.
We describe WiFeS observations of each galaxy that map the extent of the optical
emission-line regions and provide large-scale kinematic information. The clusters are lo-
cated at higher redshifts than the low-luminosity systems discussed in previous chapters,
so the emission is less extended on the sky and is less well resolved spatially. We also
describe results from Gemini Multi-Object Spectrograph (GMOS) observations that re-
solve more detailed spatial and kinematic structure in the central regions of the nebulae
of the most distant systems. The discussion in this chapter is focused on the excita-
tion properties of the spectra. The detailed morphologies of the nebulae and kinematic
results are presented in Appendix A.
Table 4.1 Properties of the BCG sample
Cluster (abbrev.) BCG z DL LHαa L1.4GHz
b LXc
(Mpc)
Abell 2597 (A 2597) PKS 2322-12 0.0822 372 1.8 31 4.7RXC J1524.2-3154 (R 1524) 2MASX J15241295-3154224 0.1022 469 1.2 1.3 3.0PKS 0745-19 (PKS 0745) PKS 0745-19, PGC 021813 0.1028 472 3.3 63 23RXC J1504.1-0248 (R 1504) SDSS J150407.51-024816.5 0.2169 1070 13 8.5 30ZwCl 348 (Z 348) RX J0106.8+0103 0.2535 1270 6.3 7.1 7.4ZwCl 3146 (Z 3146) SDSS J102339.64+041110.7 0.2900 1490 17 2.0 30
a Total Hα luminosities in 1042 erg s−1, from this work.b 1.4GHz luminosities in 1024 WHz−1, from the NRAO VLA Sky Survey (NVSS; Condon et al.,1998).c X-ray luminosities in 1044 erg s−1; values for A 2597, PKS 0745, & Z 3146 from Rafferty et al. (2006),Z 348 from Ebeling et al. (2000), R 1504 from Bohringer et al. (2005), and R1524 from Bohringeret al. (2004).
81
82 4. Shock excitation in line-luminous BCGs
The results from this sample demonstrate some of the rich variety that is observed
in the emission properties of BCGs, but are consistent with an increasing contribution
by star formation in the galaxies with the most luminous emission lines. The results
are also compatible with the model of episodic AGN feedback in the cluster cores, but
higher spatial resolution would be needed to map the detailed behaviour of the gas in
the line-emitting filaments in these more distant clusters.
We measure significant variation in the optical emission-line ratios among the sample
and discuss excitation mechanisms that might contribute to this variation. We find that
shocks are likely to play an important role in exciting the emission-line spectrum observed
in the extended filaments of BCGs, in addition to stellar photoionisation.
4.1 Observations and data reduction
4.1.1 WiFeS observations
The galaxies were observed with WiFeS between May and November 2009. Table 4.2
summarises these observations. As described in previous chapters, the science obser-
vations were obtained in a pattern of two pairs of 30 minute exposures on a galaxy
separated by a 30 minute exposure of the sky. The scaled sky frame was subtracted
from each of the four galaxy frames in this sequence. A dither pattern of 2�� offsets was
applied to the position of each galaxy frame and a nearby reference star was used to
autoguide the galaxy exposures. Table 4.2 shows the total exposure times obtained for
each galaxy.
Observations of a spectrophotometric standard star were obtained during the night
with each set of galaxy observations and these were used to calibrate the absolute flux
scale of the observations. Dwarf stars with spectra that resemble a featureless blackbody
were also measured and used to remove atmospheric absorption features from the flux
standard star and galaxy spectra. The wavelength calibration was derived from CuAr
or NeAr arc lamp spectra obtained with the science observations. Quartz-iodine lamp
flat-field frames were observed at the beginning of each night. Twilight sky flat-field
frames were obtained on most nights, so that those used in processing an observation
were always taken within two nights of that observation.
Table 4.2 Summary of WiFeS observations
Cluster Scale Dates observed (2009) Exp. time PA(kpc arcsec−1) (hr) (deg)
Abell 2597 1.53 Oct 17-20 11.0 166RXC J1524.2-3154 1.88 May 1, 3, 25; Jul 20 4.5 180PKS 0745-19 1.90 Oct 16, 17-20, Nov 14-16 6.0 140RXC J1504.1-0248 3.52 May 25-26; Jul 24 4.0 255ZwCl 348 3.94 Oct 21, Nov 13-14 4.0 145ZwCl 3146 4.37 May 25-26 2.0 180
4.1 Observations and data reduction 83
These data were reduced using the WiFeS IRAF data reduction scripts, following the
process described in Section 3.2. The spectral sampling is 0.81 A pixel−1 in the reduced
blue cubes and 1.3 A pixel−1 in the red data cubes. The typical velocity resolution
achieved is ∼ 100 km s−1 FWHM. The pixels were binned by two in the spatial direction
along the slices to form 1��×1�� spatial pixels. As for the observations discussed in
previous chapters, each science frame was reduced individually. The frames were then
spatially aligned using the peak of the continuum emission and median-combined.
4.1.2 GMOS long-slit spectroscopy
Long-slit spectra of Z 3146 and Z 348 were also obtained with the Gemini Multi-Object
Spectrograph (GMOS; Allington-Smith et al., 2002) on the Gemini North telescope in
Hawaii. Four exposures of 20minutes were recorded for each galaxy between August and
November 2008 in the queue observing mode. Observations of the spectrophotometric
and smooth-spectrum standard star Feige 110, CuAr arc calibration lamp, and quartz-
iodine flat-field lamp were also obtained with each of the galaxy observations.
The observations were made with a 1��× 330�� slit, oriented at a position angle (PA)
of 145° for Z 348 and 337° for Z 3146. The pixels were binned by two in the spatial
direction, so the pixel scale along the slit is 0.15�� per pixel. The B600 grating was
used with the GG455 blocking filter. The detector pixels were binned by four in the
spectral direction to provide a spectral sampling of ∼ 1.86 A pixel−1. The typical velocity
resolution of the spectra is ∼ 160 km s−1. The wavelength coverage is 5960− 8840 A for
Z 3146 and 5760− 8640 A for Z 348.
The data were reduced using the standard Gemini IRAF package GMOS tasks. Twi-
light sky flat-field frames were used to correct the variation in the illumination along
the slit. The sky background was averaged over a length of ∼ 20�� along the slit close
to the position of the galaxy, to provide a good correction over the small fraction of the
slit (< 15��) that contains emission from the galaxy. The observations of standard star
Feige 110 were used to calibrate the absolute flux scale and to correct telluric absorption
features in the galaxy spectra. The galaxy frames were processed individually, and then
median-combined to remove cosmic rays and bad pixels.
4.1.3 Archival GMOS integral-field spectroscopy
Integral-field GMOS observations of R 1504 and Z 3146 are also publicly available in the
Gemini Science Archive1. The observations were part of observing program GN-2005A-
Q-67, by A. Edge and R. Wilman, and were obtained on 18 May and 28 June 2005.
Three consecutive 20minute science exposures were observed for each galaxy.
The observations employed the one-slit integral-field mode of GMOS, using the right,
or red, slit mask. This provides a 3.5��× 5.0�� field of view sampled by 500 hexagonal
lenslets that have a projected diameter of 0.2��. In this mode, a 1.75��× 5.0�� sky field,
1 http://www2.cadc-ccda.hia-iha.nrc-cnrc.gc.ca/gsa/
84 4. Shock excitation in line-luminous BCGs
imaged by 250 lenslets, is simultaneously observed at a position offset by one arcminute
from the object field. This blank sky frame was used to subtract the sky background
emission from the science data.
The B600 grating was used for these observations and the grating angle was chosen
to provide wavelength coverage that extends from the [O ii] line at 3727 A to the [N i]
line at 5200 A in each galaxy. These settings provide a spectral pixel sampling of 0.46 A,
and a typical spectral resolution of ∼ 100 km s−1.
Quartz-iodine lamp flat-field and arc lamp images were obtained during the night
with the science observations. Twilight flat-field observations from the observing nights
are also available in the archive. The bias frame and standard star (EG 131) observations
used in the data processing are baseline calibrations provided as part of the Gemini queue
observing program during the semester. The standard star observation was separated
by several weeks from the galaxy observations.
We reduced and calibrated the raw data using the GMOS data reduction scripts
provided in the Gemini IRAF package. The three frames obtained for each object were
median-combined before processing. The routine gmos.gfreduce was used to subtract
the detector bias level, flat-field the images, calibrate the wavelength scale, subtract the
sky background and calibrate the flux scale. The reduced spectra are resampled into
three-dimensional data cubes with 0.1��× 0.1�� spatial pixels.
4.1.4 Continuum and emission-line fitting
To extract the line emission from the data cubes and long-slit spectra, the continuum and
emission-line features of the galaxy spectra were fit simultaneously using the gandalf
(Gas AND Absorption Line Fitting) IDL software package. gandalf is an algorithm
that is designed to separate the relative contributions of the stellar continuum and
nebular emission in galactic spectra and to measure the gas emission and kinematics.
The software was developed by the SAURON team and is available from the SAURON
website2; see also Sarzi et al. (2006) for details. The fitting procedure also makes use of
the pPXF code of Cappellari & Emsellem (2004) to derive the stellar kinematics. The
gandalf algorithm employs the MPFIT package (Markwardt, 2009) to implement the
Levenberg-Marquardt least-squares minimisation method and determine the optimum
linear combination of continuum templates and Gaussian emission-line models to match
a spectrum.
We used single stellar population (SSP) synthesis models from Gonzalez-Delgado
et al. (2005)3 to form the template library for the continuum spectrum matching. These
have a spectral sampling of 0.3 A. The synthetic spectra cover a wavelength range of
3000− 7000 A, so that the region of spectrum from approximately the wavelength of the
[O ii] λλ3726, 3729 doublet to that of [S ii] λ6731 could be included in the continuum
2 http://www.strw.leidenuniv.nl/sauron3 https://w3.iaa.es/∼rosa/research/synthesis/HRES/ESPS-HRES.html
4.1 Observations and data reduction 85
and line fitting. This includes all of the strong optical emission lines in the WiFeS data
for each of the galaxies. We used the SSP library spectra formed from the Geneva
stellar evolutionary tracks, which cover a wide range of metallicities: twice solar, solar,
half-solar, and 1/10 solar, and ages from 1Myr to 10Gyr.
The model spectra were convolved with a Gaussian function to degrade the resolution
to that of the observed spectra. In the gandalf fitting procedure, the spectral regions
around the emission lines, residual sky features, and bad pixels are first masked while the
optimum combination of continuum models is found. The line profile models are then
added and the fit is repeated, masking only the regions of bad data, such as anomalous
sky line residuals. In our analysis, a multiplicative polynomial correction of up to eighth
order was also included in the model and no reddening corrections were performed during
the spectral fitting.
The emission lines were each modelled with a single Gaussian function. The strongest
hydrogen recombination line in each data cube (Hα, Hβ, or Hγ) was fit freely and
the velocity centroids and line widths of the other emission lines were constrained to
match those of the hydrogen line. Two exceptions were the [O ii] λλ3726, 3729 and
[N i] λλ5198, 5200 doublets, which are unresolved in the spectra. A single Gaussian
was fit to the blended lines. The velocity centroid of the [N i] profile was tied to the
relevant hydrogen line, while the [O ii] profile was fit independently. The emission-line
dispersions have been corrected for the instrumental resolutions. All line widths are
given as dispersions, rather than FWHM, unless otherwise stated. Section A.1 in the
appendix shows examples of the continuum and line fitting results in integrated central
spectra for each of the galaxies.
The gandalf routine returns a measurement of the ratio of the fitted line amplitude
to the residual noise in the spectrum A/N for each emission-line fit. This was used
to estimate the limiting sensitivity and statistical errors in the emission-line flux. The
parameter measures how much the fitted emission lines protrude above the noise in
the spectrum, and quantifies the relative uncertainty in the line fluxes by providing a
measurement of the statistical noise in the continuum-subtracted spectra (Sarzi et al.,
2006). We find that the A/N values returned are very similar to the errors measured for
the line fluxes using the method described in Chapters 2 and 3. We adopt a sensitivity
limit of A/N = 4 for all the line flux measurements. The corresponding line flux varies
slightly among the data cubes, but results in typical limiting line surface brightnesses of
∼ 5×10−17 erg s−1 cm−2 arcsec−2 in the WiFeS spectra, ∼ 2×10−15 erg s−1 cm−2 arcsec−2
in the GMOS IFS data, and ∼ 3 × 10−17 erg s−1 cm−2 arcsec−2 in the GMOS slit data.
Where we present emission-line properties derived from the line fits, only the results for
pixels in which A/N > 4 are included. In the emission-line maps in Appendix A, the
pixels with signal-to-noise below this threshold value are left blank.
As described in Chapter 3, we adopt the reddening law of Osterbrock (1989) and a
ratio of total to selective absorption R = AV /E(B−V ) = 3.1. We correct the measured
86 4. Shock excitation in line-luminous BCGs
emission-line fluxes for extinction using the measured Hα/Hβ line flux ratio, assuming
an intrinsic flux ratio of F (Hα)/F (Hβ) = 3.1. This value is suitable for systems in which
a component of collisional excitation is expected to contribute to the Hα flux (Veilleux &
Osterbrock, 1987). For each pixel in which both the Hα and Hβ lines are detected above
the sensitivity limit, we use the line flux ratio and reddening law to estimate the level
of extinction in the spectrum. We calculate a mean extinction value from all of these
pixels, and this is applied in pixels for which a local value can not be determined from
the Balmer decrement. These average extinction values are given in the third column
of Table 4.3 for each galaxy. The Galactic contribution to the extinction, obtained from
the dust measurements of Schlegel et al. (1998), is indicated in Table 4.3 for each of the
galaxies. The maps of emission-line fluxes that are presented have not been corrected
for extinction.
4.2 Results
In the following sections, we present the integrated spectral properties of the BCG
sample and discuss the excitation properties, based on the diagnostic optical line flux
ratios measured across the extended nebulae. Detailed descriptions of the morphological
and kinematic results for the individual galaxies are deferred to Appendix A. There, we
provide a short summary of background information from the literature for each galaxy,
and present and discuss the emission-line flux and velocity maps that are derived from
the data.
4.2.1 Integrated spectra
Figure 4.1 shows spectra from our WiFeS observations of the six brightest cluster galax-
ies, integrated over a 5��-diameter circular aperture centred at the peak of the galaxy
continuum emission. These spectra have been corrected for the mean level of extinction
measured in the filaments from the Hα/Hβ line flux ratio, as described in Section 4.1.4.
The total luminosities of the Hα and [N ii] emission from the extended emission-line
regions of the galaxies are shown in Table 4.3.
The emission-line spectra exhibit the strong forbidden and hydrogen recombination
lines that were observed in the lower-luminosity BCG nebulae described in Chapters 2
and 3: [O ii] λλ3726, 3729 (unresolved), [O iii] λλ4959, 5007, [N i] λ5198, 5200, [O i]
λλ6300, 6363, [N ii] λλ6548, 6583, [S ii] λλ6716, 6731, Hα λ6563, Hβ λ4861, and
Hγ λ4340. In addition, fainter lines that are also detected in the integrated spectra of
some or all of the high-luminosity systems presented here include: [Ne iii] λ3869, [Ne iii]
λ3967, [S ii] λλ4069, 4076 (unresolved), He ii λ4686, He i λ5876, [Ca ii] λλ7291, 7324,
[O ii] λ7330 (blended with [Ca ii]), Hδ λ4105, H� λ3970, and Hζ λ3889.
The integrated [N ii]/Hα line ratios are plotted against the total Hα luminosities
from the WiFeS data in Figure 4.2. The values for all of the high- and low-luminosity
4.2 Results 87
Figure 4.1 Spectra of each galaxy integrated over 5��-diameter aperture centred at the con-tinuum peak, ordered by increasing redshift. The dashed vertical lines indicate the wavelengthsof residual [N ii] λ5577 and [O i] λλ6300, 6363 emission lines from the sky (other OH sky lineresiduals are also evident in the spectra). These spectra have been corrected for the mean levelof extinction measured in the data, as indicated in each panel.
88 4. Shock excitation in line-luminous BCGs
Figure 4.2 Extinction-corrected,integrated Hα luminosities plottedagainst the integrated [N ii]/Hαline flux ratios for the galaxies inour sample. In addition, the openmarkers show data taken fromCrawford et al. (1999), from slitspectra of an X-ray-selected sampleof BCGs. These luminosities havebeen corrected for the value of H0
used in this work.
nebulae studied in this work are included, together with the sample from Crawford et al.
(1999). Our results show good agreement with the trend of decreasing line flux ratio as
the Hα luminosity increases, with a similar scatter among the values as was observed by
Crawford et al. (1999).
4.2.2 Optical diagnostic diagrams
Figure 4.3 shows flux ratios measured for each galaxy from the WiFeS data plotted on
the three commonly-used optical line flux ratio diagnostic diagrams (also known as BPT
diagrams; Baldwin et al., 1981; Veilleux & Osterbrock, 1987). We note that these plots
include only the values from pixels where the components of both line ratios are detected
above the limiting signal-to-noise ratio (A/N > 4). This restricts the data presented
here to the pixels in which the fainter [O iii] lines can be accurately measured, so that
the regions represented are the high-surface-brightness, and therefore inner, parts of the
nebulae. The line flux ratios measured in the nebulae of the low-line-luminosity galaxies
Table 4.3 Integrated emission-line fluxes and luminosities
BCG AV,gala AV,avg
b F (Hα) L(Hα) F ([N ii]) L([N ii])(mag) (mag)
A 2597 0.09 0.49 7.1± 0.5 1.8± 0.1 5.4± 0.4 1.4± 0.1R1524 0.60 0.96 2.0± 0.2 1.2± 0.1 2.6± 0.3 1.5± 0.2PKS 0745 1.62 1.65 3.6± 0.4 3.3± 0.3 3.7± 0.4 3.3± 0.3R1504 0.33 0.67 5.4± 0.8 13.0± 1.9 3.0± 0.5 7.2± 1.1Z 348 0.08 0.61 2.1± 0.4 6.3± 0.9 0.9± 0.1 2.6± 0.4Z 3146 0.09 0.80 3.4± 0.4 17.4± 2.1 2.7± 0.3 13.6± 1.6
Line fluxes are in units of 10−14 erg s−1 cm−2, and luminosities in 1042 erg s−1.The luminosities have been corrected for extinction.a Foreground Galactic extinction (in magnitudes) from Schlegel et al. (1998).b Mean extinction (in magnitudes) assumed in pixels where it can not be esti-mated from the Hα and Hβ fluxes and the adopted extinction law.
4.2 Results 89
NGC 4696, A 3581, 2A 0335+096, and Sersic 159-03, as described in Chapters 2 and 3,
are also shown on these diagnostic plots. For A 3581, 2A 0335+096, and Sersic 159-03,
the values measured both in the nucleus and in the inner filaments, as presented in
Table 3.4, are shown as open and filled markers, respectively.
The flux-ratio diagrams provide a useful way to classify the dominant excitation
mechanism in emission-line galaxies. The diagnostic diagrams presented here show the
classification schemes described by Kewley et al. (2006) and Schawinski et al. (2007),
which are based on theoretical and empirically-derived divisions between the classes of
AGN/Seyfert, H ii/star-forming galaxies, and LINER-type emission. The solid curved
line in the [N ii]/Hα – [O iii]/Hβ diagram is the ‘maximum starburst line’, an upper limit
of starburst activity determined by Kewley et al. (2001) from theoretical stellar photoion-
isation models. The dashed line is a semi-empirical boundary proposed by Kauffmann
et al. (2003) to modify the Kewley et al. (2001) scheme and separate purely star-forming
systems from those with composite spectra, to which AGN or LINER emission may
contribute in addition to stellar photoionisation. The divisions between the LINER and
AGN regions in the [S ii]/Hα – [O iii]/Hβ and [O i]/Hα – [O iii]/Hβ diagrams are the
empirical boundaries determined by Kewley et al. (2006) from an analysis of the spectra
of 85,000 galaxies in the Sloan Digital Sky Survey (SDSS). The corresponding dividing
line in the [N ii]/Hα – [O iii]/Hβ diagram was defined by Schawinski et al. (2007) using
the spectra of early-type emission-line galaxies from the SDSS.
The line ratios measured in our sample of BCGs are largely consistent with pure
LINER-type systems or composite LINER/H ii spectra. In the [N ii]/Hα – [O iii]/Hβ
diagram, the flux ratios for different galaxies form a sequence of decreasing [N ii]/Hα
flux ratio, moving from LINER-like values into the composite/starburst region of the
diagram. The [N ii]/Hα flux ratios decrease approximately as the total emission-line lu-
minosity of the nebulae increases. This reflects the trend seen in the integrated [N ii]/Hα
flux ratios of BCGs, shown in Figure 4.2. However, there is not a direct correspondence
between the [N ii]/Hα ratio and L(Hα). This variation will be discussed further in Sec-
tion 4.3.3 in the context of a mixing between the excitation mechanisms that contribute
in different galaxies.
The range of [N ii]/Hα flux ratios measured within individual nebulae is small rel-
ative to the variation between the values for different galaxies. The WiFeS data show
a larger range of [O iii]/Hβ flux ratio values than [N ii]/Hα flux ratios within each
of the galaxies. However, there are significantly larger uncertainties associated with the
[O iii]/Hβ flux ratios because both Hβ and [O iii] are fainter features than [N ii] and Hα.
The variation in the flux ratio measured from the WiFeS data is largely attributable to
this uncertainty. A contribution from AGN photoionisation in the nuclear regions may
also contribute to the spread of [O iii]/Hβ flux ratios, as discussed further below.
Though the BCGs are well-differentiated by the [N ii]/Hα line flux ratios, the
[S ii]/Hα and [O i]/Hα ratios behave differently. The [S ii]/Hα – [O iii]/Hβ plot shows
90 4. Shock excitation in line-luminous BCGs
Figure 4.3 The flux ratios from the WiFeS data are plotted here on the standard opticaldiagnostic diagrams. The black lines and labels show the divisions of the classification schemedescribed by Kewley et al. (2006) and Schawinski et al. (2007). The left column shows thefull parameter space occupied by emission-line galaxies as, for example, presented by Kewleyet al. (2006), and the right column shows a closer view of the regions surrounding our data.The values for the line-luminous nebulae where the line components have A/N > 4 are shown,together with flux ratios measured in the emission nebulae of NGC 4696, from Chapter 2, andA3581, 2A 0335+096, and Sersic 159-03, from Chapter 3. For the latter, the open points indicatenuclear fluxes and the solid points those from the extended filaments (as in Table 3.4).
4.2 Results 91
that systems with distinct [N ii]/Hα values have similar [S ii]/Hα flux ratios, though
the measurements are still distributed between the LINER and composite regions of the
diagnostic plot. In particular, the measurements for most luminous emission-line nebu-
lae overlap in this diagram. In addition, the measurements for all the galaxies are closely
clustered in the [O i]/Hα–[O iii]/Hβ plot. In both diagrams, the measured flux ratios are
located further toward the LINER classification region than in the [N ii]/Hα –[O iii]/Hβ
scheme. The flux ratio values fall entirely within the most extreme LINER region of
the [O i]/Hα – [O iii]/Hβ diagram. This diagram is the most sensitive to LINER ex-
citation (Kewley et al., 2001). The [O i]/Hα flux ratio is sensitive to the hardness of
the ionising field, and is enhanced by shocks. The high [O i]/Hα ratios may indicate a
strong contribution to the excitation from LINER mechanisms, even in the systems that
are likely to be dominated by star formation (e.g., Z 3146, Z 348, and R1504). As will
be discussed in Section 4.3.3, the enhanced [S ii]/Hα and [O i]/Hα ratios indicate the
presence of a source of supplementary heating that affects the partially-ionised regions
of the nebula where these lines originate. The emission is not dominated by a single
excitation mechanism.
The GMOS long-slit spectra measured for R 1504 and Z 3146 also include the diag-
nostic emission lines used to form the BPT diagrams. For clarity, we plot the results of
these observations separately in Figure 4.4. The left column shows flux-ratio measure-
ments for Z 348, where the two velocity components that are measured in the emission-
line profiles (see Section A.2.5.1) are shown in different colours. The right column shows
the results for Z 3146. The marker sizes represent the distance of the pixel along the
slit from the position of the continuum peak. These observations have slightly higher
sensitivity and better spatial resolution than the WiFeS data, though they sample a
limited region of the nebulae.
The flux ratios vary smoothly across the nebula. In both galaxies, there is a similar
radial trend in the values on each side of the nucleus, and this is is likely to be in
part caused by the influence of the central AGN. In the inner part of the nebulae, the
[O iii]/Hβ ratios decrease and the ratios of the flux in the [N ii], [S ii], and [O i] lines to
that in Hα increase. This trend is observed over the inner six to nine pixels, or 0.9−1.3��,
in all ratios for both galaxies. The observations of standard stars that were obtained
with the galaxy spectra indicate that the point-spread function (PSF) for an unresolved
source in the observations has a FWHM of 0.7− 0.8��. This implies that the influence of
an unresolved central source of emission may be spread over an angular distance of up to
∼ 1.4− 1.6��, the approximate full-width at zero-intensity of the PSF. This could cause
the smooth transition toward more AGN-like values of the emission-line ratios over the
central ∼ 1�� in both of these nebulae.
Beyond the central region, the data show the same systematic trend seen in other
data of decreasing [N ii]/Hα, [S ii]/Hα, and [O i]/Hα flux ratios as the [O iii]/Hβ flux
ratio decreases. This resembles a tendency toward more star-formation-dominated flux
92 4. Shock excitation in line-luminous BCGs
ZwCl 348 ZwCl 3146
Figure 4.4 Flux ratios from the GMOS slit spectra plotted on the common BPT opticaldiagnostic diagrams. The left column contains plots of the results for the two velocity componentsthat were fit to the emission from Z348, and right column shows the results for Z 3146. Therelative size of each marker represents the distance of the pixel from the continuum peak in theslit.
4.3 Discussion 93
ratios with distance from the nucleus, though the emission has low surface brightness
in these regions. The variation in line flux ratios is interpreted as changing mix of
excitation mechanisms in the emission regions. These results are consistent with a
significant contribution from stellar photoionisation in these line-luminous nebulae, and
are discussed further in Section 4.3.3.
4.3 Discussion
4.3.1 The role of shock excitation
As also seen in other studies (e.g., Edwards et al., 2009; Hatch et al., 2007; Ogrean
et al., 2010; Wilman et al., 2009), we observe significant variation of the emission-line
flux ratios within and among these line-luminous nebulae. This is generally explained by
spatially-varying excitation mechanisms. We discuss here evidence that shock excitation
is operating in conjunction with stellar photoionisation to ionise the gas in the emission
regions.
We suggested in Chapter 2 that shock excitation by intermediate-velocity shocks
could operate as the underlying mechanism that excites the low-ionisation atomic emis-
sion lines and strong molecular hydrogen emission in the extended emission-line regions
of BCGs. NGC 4696 hosts a relatively low-luminosity nebula and shows evidence for
only low levels of star formation. As a result, the emission regions have a LINER-like
spectrum without significant contribution from stellar photoionisation. We found that
this optical emission spectrum is well-matched by the intermediate-velocity shock mod-
els that were investigated. We consider now whether a similar mode of excitation can
explain the properties of the emission in the more luminous systems.
The velocity dispersion is another useful diagnostic of shock excitation. The ionising
flux produced by shocks scales with shock velocity, and so the relative emission-line
fluxes observed in the shocked gas also vary with velocity. If there is a significant
contribution to the emission from shocks, we expect to see a correlation between the line
flux ratios and the measured velocity dispersions in the emission regions (see e.g., Dopita
& Sutherland, 1995). Low velocity dispersions, on the order of several tens of kilometres
per second (e.g., Epinat et al., 2010), are expected in gas excited by photoionisation
in H ii regions. Therefore, emission-line widths significantly in excess of this value can
be diagnostic of shock excitation. Generally, the velocity dispersions in the extended
filaments of BCGs are high, as expected from shocks. Other effects may broaden the
measured dispersions: for example, the presence of multiple emission structures with
different velocities along the line of sight, or beam-smearing in the pixels where there
are steep gradients in the velocity distribution. We observe complex morphological and
kinematic structures in the BCG filaments and such effects likely increase the scatter in
the measured line dispersions. These factors may obscure the diagnostic in observations,
but we expect that some systems will be affected to a lesser degree and provide a clear
94 4. Shock excitation in line-luminous BCGs
Abell 2597
RXC J1524.2-3154
PKS 0745-19
Figure 4.5 Plots of velocity dispersion versus line flux ratio for, from left to right: [N ii]/Hα,[S ii]/Hα, [O i]/Hα, and [O iii]/Hβ, from the WiFeS observations of A 2597 (top), R 1524(centre), and PKS0745 (bottom). The colour of each marker indicates the distance of the pixelfrom the galaxy nucleus, according to the colour scale to the right of the plots for each galaxy.
signal of this correlation.
In Figures 4.5 and 4.6, we plot the measured velocity dispersions of the line emission
from our WiFeS observations of each galaxy as a function of the flux ratios of [N ii]/Hα,
[S ii]/Hα, [O i]/Hα, and [O iii]/Hβ. The plots contain values from all pixels where both
emission lines are detected above the signal-to-noise ratio criterion of A/N = 4. The
[O iii]/Hβ figure for R 1524 is excluded because the [O iii] line is only detected above
this limit in two pixels. In these diagrams, the marker colours represent the distance
from the nucleus of the galaxy, according to the colour scale shown.
Each of the diagrams for A 2597, R 1524, and PKS0745 in Figure 4.5 shows a cor-
relation between the velocity dispersions and line flux ratios. The results for Z 348, in
Figure 4.6, are also consistent with these trends. This characteristic signature of shock-
excited gas is particularly strong in the emission-line nebula of A 2597, where there is
a general trend of increasing velocity dispersion in each of the low-ionisation line flux
ratios. This is consistent with the strong interaction between the nebula and the radio
4.3 Discussion 95
RXC J1504.1-0248
ZwCl 348
ZwCl 3146
Figure 4.6 Plots of velocity dispersion versus line flux ratio for, from left to right: [N ii]/Hα,[S ii]/Hα, [O i]/Hα, and [O iii]/Hβ, from the WiFeS observations of R 1504 (top), Z 348 (centre),and Z 3146 (bottom). The colour of each marker indicates the distance of the pixel from thegalaxy nucleus, according to the colour scale to the right of the plots for each galaxy.
lobes that is inferred in this system (as discussed in Section A.2.1.1). The correlation
indicates a significant component of shock excitation in the regions of high dispersion
within a radius of ∼ 5�� (7.5 kpc) from the nucleus, but this also continues to a radius
of nearly 20 kpc, suggesting that shock excitation contributes throughout the nebula.
In A 2597, R 1524, PKS 0745, and Z 348, in addition to the correlation between the
line flux ratios and velocity dispersions, we note that it is the positions closest to the
nucleus that have the highest flux ratios and velocity dispersions. This is likely the
result of higher relative velocities among the clouds of gas with proximity to the nucleus,
resulting in faster shock velocities.
In Figure 4.7, we show similar plots of the [N ii]/Hα and [S ii]/Hα flux ratios
against velocity dispersion, for the observations of 2A 0335+096 and Sersic 159-03 that
were presented in Chapter 3. These galaxies are intermediate between the luminous
systems presented in this chapter and the cases of the low-line-luminosity and nearby
systems of NGC 4696 and A3581, for which little variation in velocity dispersion and
96 4. Shock excitation in line-luminous BCGs
2A0335+096
Sersic 159-03
Figure 4.7 Plots of velocity dispersion versus line flux ratios from the single Gaussian linefits to emission lines in the WiFeS observations of 2A 0335+096 and Sersic 159-03 that werediscussed in Chapter 3. The colours again represent the distances of the pixels from the galaxynucleus.
line flux ratio was observed in the extended filaments. The results for 2A 0335+096 and
Sersic 159-03 do show evidence for weak correlations between the velocity width and the
line flux ratios, though there are large uncertainties in the measurements from the outer
regions of the filaments. This again suggests that shock excitation dominates in these
systems of lower emission-line luminosity.
In the most distant systems, the correlation between the velocity dispersion and
line ratio measurements is less evident in the WiFeS data. For R 1504 and Z 3146, in
particular, the plots in Figure 4.6 do not show correlations in these properties. This
may be because we do not spatially resolve regions dominated by individual emission
mechanisms, or because shocks are a relatively less important source of excitation in
these nebulae. These are both systems that show a significant contribution from star
formation and the spectra are not consistent with purely LINER excitation. However,
the velocity dispersion is high in the emission regions, indicating that the gas is turbulent.
We can examine the variation of the [O iii]/Hβ line flux ratios in the central regions
of R 1504 and Z 3146 using the GMOS IFS data. The velocity dispersions derived from
the line fits to these data are plotted against the [O iii]/Hβ flux ratios in Figure 4.8. For
R 1504, we found that an additional velocity component was needed to fit a blueshifted
4.3 Discussion 97
Figure 4.8 The velocity dispersions measured from the single Gaussian line fits to emissionlines in the GMOS IFS data for R 1504 (left) and Z 3146 (right) are plotted here against the[O iii]/Hβ line flux ratios. Again, the marker colour indicates the distance of the pixel fromthe galaxy nucleus, according to the colour scale shown. In the left panel, dispersions aboveσ � 230 km s−1 are the result of beam-smearing over the steep velocity gradient in the centralnebula. The diamond markers indicate values obtained from the additional velocity componentincluded in the fits, as described in the text.
wing in the profiles over a region in the nebulae close to the nucleus. This is discussed in
Section A.2.4.1 in the appendix. The results derived from the dual-component fits are
plotted in the left panel of Figure 4.8. The values for the fainter, blueshifted velocity
component are distinguished by diamond markers. The high [O iii]/Hβ flux ratios
(log([O iii]/Hβ) � 0.05) that are measured for this component are evident, though the
velocity dispersions are similar to those in the brighter line component.
As seen in Figure A.15, the emission lines across the centre of the nebula are broad-
ened in a manner that is consistent with the effects of beam-smearing where there is a
steep gradient in the line-of-sight gas velocities. The highest velocity dispersions, above
∼ 230 km s−1, are the result of this effect. However, even disregarding this region of the
plot, the distribution shows a complex structure in velocity dispersion. High dispersions
are observed in gas close to the nucleus, which exhibits a range of flux ratio values. The
largest [O iii]/Hβ ratios coincide with the nucleus and may be caused by the influence
of the AGN. Below ∼ 100 km s−1 there is a transition in the structure of the distribu-
tion. The emission from the extended southwest filament, at larger distances from the
nucleus, has low velocity dispersions and produces a narrow range of line ratios. This
suggests that the gas in this filament is dominated by photoionisation, while there is a
larger contribution by shock excitation that increases the gas turbulence in the central
parts of the nebula. The filament may be a cooling, infalling stream of gas, in which
the material has had time to condense and form stars that are now photoionising the
surrounding material. This is consistent with the enhanced ultraviolet and blue con-
tinuum emission observed in the filament (Ogrean et al., 2010). There are also signs
of star formation associated with the core of the galaxy, but the enhanced dispersions
and varied flux ratios suggest that shocks provide an additional ionising source in this
region. We discuss the possibility that the velocity field observed across the centre of
98 4. Shock excitation in line-luminous BCGs
the nebula is produced by a collision of two systems of gas in Section A.2.4.1. This
interaction would be a significant source of turbulence and would drive shocks into the
gas clouds in the collision region. The nearby galaxy to the northwest that is potentially
interacting with the BCG, may also perturb the gas structures in nebula. The multiple
velocity components in the gas surrounding the nucleus suggest that there are complex
gas structures that could include accretion steams into the core. There might also be
interactions with AGN outflows if there are jets forming in the nucleus.
The plot of velocity dispersion against [O iii]/Hβ flux ratio in Z 3146 from the GMOS
data, in the right panel of Figure 4.8, does not exhibit similar evidence for separate
excitation regimes. This may be because these are not resolved in this more distant
system – the photoionisation regions may be mixed with shock excitation on smaller
scales than are resolved by the data. The shock velocities dominate the line widths,
and line ratios are indicative of composite excitation mechanisms (see Figure 4.3). The
[O iii]/Hβ ratio is enhanced in the core of the galaxy (within ∼ 0.2��, or 0.9 kpc).
High gas metallicity or increased starburst activity in the core might contribute to this.
However, we inferred from the BPT diagrams in Figure 4.4 that the effect of the AGN
ionisation field likely contributes to the flux ratios within approximately 1.3�� of the
nucleus in the GMOS data.
4.3.2 Infrared luminosities
A number of previous studies provide evidence that BCG nebulae are powered by the ul-
traviolet emission from young stars, in addition to a secondary mechanism that produces
the characteristic low-ionisation spectrum and strong molecular hydrogen line emission
(e.g., Johnstone & Fabian, 1988; Allen, 1995; Voit & Donahue, 1997; Crawford et al.,
1999; O’Dea et al., 2004; Wilman et al., 2006; Hatch et al., 2007; Donahue et al., 2011;
Mittal et al., 2011; Oonk et al., 2011, and references therein). We briefly discuss here
whether the integrated infrared and optical line luminosities of the nebulae can pro-
vide information about the relative levels of participation of the contributing excitation
mechanisms.
In the left panel of Figure 4.9, the total extinction-corrected Hα fluxes measured
from our WiFeS data are plotted against the integrated infrared flux measurements
that are available in the literature. The right panel shows a plot of the corresponding
total luminosities. The luminosities are also listed in Table 4.4 with the sources of the
infrared data; A 3581, R 1524, and R1504 are excluded because no published infrared
measurements were available for these targets. The infrared fluxes were derived from
fits to the spectral energy distributions, using measurements at mid- to far-infrared
wavelengths as described in the works cited in Table 4.4. The uncertainties in the
photometric calibration of the infrared flux measurements are approximately 15 − 20%
(Egami et al., 2006a; Donahue et al., 2011; Mittal et al., 2011), so the derived fluxes are
associated with relative uncertainties of at least this magnitude. We show representative
4.3 Discussion 99
Table 4.4 Infrared luminosities of the BCGs
BCG DL LHα LIR References(Mpc) (1041 erg s−1) (1043 erg s−1)
NGC 4696 44 0.18 0.29 12A0335+096 147 4.9 2.6 2Sersic 159-03 240 2.3 < 3.0c 3Abell 2597 372 18 18 2PKS 0745-19 472 33 39 2ZwCl 348 1270 63 119 3ZwCl 3146 1490 170 158 4
References: [1] Mittal et al. (2011), [2] Donahue et al. (2011), [3] Quillenet al. (2008), [4] Egami et al. (2006b).c Sersic 159-03 is not detected in 70µm-band observations, so Quillenet al. (2008) obtain an upper limit for the luminosity.
uncertainties of 20% in Figure 4.9.
The infrared and Hα fluxes plotted in the left panel of Figure 4.9 show evidence for a
correlation, with significant scatter. The luminosities appear to be strongly correlated,
but this tight relation between the infrared and Hα luminosities is largely the result of
the scale-dependence that is introduced into the luminosity values by the distances to
the galaxies. This distributes the luminosities over several orders of magnitude, so that
the variability among the different flux measurements becomes second-order to the effect
of the scale-related factor (e.g., Kennicutt, 1990).
From the observational results, the ratio of flux emitted in Hα to that in the infrared
has a mean value of 1.0% (with 1σ variation of 0.5%) for the galaxies in our sample.
This value is represented by the dashed line in the left panel of Figure 4.9. The empirical
relations derived by Kennicutt (1998) indicate that star-forming regions produce slightly
Figure 4.9 Left: The integrated Hα flux versus infrared flux for the BCGs. Right: Theintegrated Hα luminosity versus infrared luminosity for the sample. The solid line shows theKennicutt (1998) relation for starbursts, the dashed line shows the relation for the 200 km s−1
shock models, assuming that 100% of the shock emission is re-radiated at infrared wavelengths,and the dotted line is the case of 10% coupling between the shock models and surrounding dustyenvironment.
100 4. Shock excitation in line-luminous BCGs
more infrared power for a given Hα flux: LHα/LFIR ∼ 0.6%. This is shown as a solid
line in the right panel of Figure 4.9.
O’Dea et al. (2008) also reported a correlation between the integrated infrared and
Hα fluxes among their much larger sample of BCGs, and a significantly stronger corre-
lation in the luminosities, again enhanced by the scaling effect. They suggest that the
correlation between the fluxes indicates that the emission is produced by the same or
a related power source. They note that the observed Hα luminosities for their sample
generally fall below that expected for starbursts, given the measured infrared luminosi-
ties. The authors suggest that this deficit may occur because most of the Hα fluxes were
obtained from slit spectra that do not cover the extent of the emission from the nebulae.
This issue can be addressed by imaging the entire emission region in each system with
an integral-field spectrograph, as we have done for our sample. Though our sample com-
prises a small number of measurements, the results confirm the existence of a correlation
between the fluxes with optical line fluxes that are more accurately constrained.
Models of intermediate-velocity shocks, as discussed in Section 4.3.3, predict that the
fraction of the total shock luminosity that is emitted in the Hα line is LHα/Lshock ∼ 1%,
for shock velocities of 200 km s−1. It is very likely in the gas-rich and dusty regions of
the BGC nebulae, that the shocks are closely coupled to the obscuring dust. In this
case, all of the flux radiated by the shocks is absorbed by dust and re-radiated in the
infrared (LIR = Lshock). This case is indicated by the dashed line in the right panel of
Figure 4.9 and shows good agreement with our empirical data. If the coupling between
the shocks and dusty environment is not as strong, relatively less of the total shock
flux will be re-radiated as infrared emission, for a given shock luminosity. This moves
the relation to smaller LIR relative to LHα. The luminosity relation for 10% coupling
is also plotted on the figure (dotted line). The results presented in Figure 4.9 suggest
that there is close coupling between the shock output and surroundings if the Hα and
infrared luminosities observed are produced entirely by shock excitation. In this case, it
is not possible to distinguish between shock excitation and star formation on the basis
of the relative infrared and Hα luminosities alone. The observed results leave open the
possibility that there is a varying combination of shock and photoionisation excitation
powering the emission from the filaments of the BCGs.
4.3.3 Shock and photoionisation model mixing
As discussed in the previous sections, there is evidence for relatively high rates of star
formation in several of the galaxies. Our results also indicate that intermediate-velocity
shocks are an important source of excitation in the emission-line regions. We therefore
investigate a model of combined shock and photoionisation heating in the gas of the
extended filaments.
We employ a set of shock models, similar to those introduced in Chapter 2, to
investigate the potential role of shock excitation in the BCG nebulae. These models have
4.3 Discussion 101
been provided by M. Dopita (priv. comm.) and are also discussed by Rich et al. (2010)
and Rich et al. (2011). They are one-dimensional shock simulations generated with the
MAPPINGS III shock and photoionisation modelling code (most recently described by
Sutherland & Dopita, 1993). The shock velocities in the models are v = 100−200 km s−1,
which provides a good match to the typical line widths observed in the filaments of the
BCGs.
We also utilise a set of photoionisation models, again provided by M. Dopita, to
simulate the effect of mixing between the spectra produced by shock and photoionisation
excitation. These were generated using MAPPINGS with input radiation fields obtained
from the Starburst99 stellar population synthesis models (Leitherer et al., 1999) for
continuous star formation. The ionisation parameter Q is an independent parameter in
this set of models. This parameter describes the number of ionising photons per atom
and is equal to the ionising photon flux per unit area divided by the hydrogen particle
density. The input ionisation parameter has values of 6.5 ≤ logQ ≤ 8.0 in the set of
photoionisation models.
In Figure 4.10, we plot a sequence of fractional mixing between the photoionisation
and shock models on the optical diagnostic diagrams discussed in Section 4.2.2. The
shock and photoionisation models shown have elemental abundances of twice solar, based
on the abundance set described by Asplund et al. (2009) and Grevesse et al. (2010). This
is typical of abundances inferred for the hot gas in central cluster galaxies (e.g., Sanders
& Fabian, 2006a). In the figure, we show linear combinations of the predicted line flux
ratios from the series of pure shock models (black circles) and the photoionisation-only
models (white diamonds) in fractional steps of 0.2. The degree of mixing is indicated by
the grey level of the markers. The marker size corresponds to the input shock velocity of
the contributing shock model. The shock velocities of the pure shock models are labelled,
and range from 100 km s−1 to 200 km s−1 in intervals of 20 km s−1. The input values
of the ionisation parameter for the photoionisation model sequence are also indicated;
this value has increments of logU = 0.25 between successive models. The measured line
ratios from the WiFeS observations of our sample of BCGs are plotted with the models
in Figure 4.10. The data shown here are the same as those plotted in the panels of
Figure 4.3, and the same symbols are used.
The photoionisation-shock mixing lines provide a good description of the variation
in the flux ratios in the [N ii]/Hα – [O iii]/Hβ plot (top panel). The measured [N ii]/Hα
flux ratios extend across the range of mixing fractions formed from these models, suggest-
ing that there is a significant change in the relative levels of the component mechanisms
among the galaxies. As inferred in other studies, these results suggest that a signifi-
cant fraction of the Hα luminosity of the extended emission may be produced by stellar
photoionisation. This type of mixing could provide an explanation for the sequence of
decreasing [N ii]/Hα flux ratio that is observed in these galaxies. The result is similar
in the [S ii]/Hα – [O iii]/Hβ plot (centre panel), but as discussed previously, the data
102 4. Shock excitation in line-luminous BCGs
Figure 4.10 A mixing sequence between the intermediate-velocity shock and stellar photoionisationmodel predictions is plotted on the optical diagnostic diagrams. The greyscale shade of the valuesrepresents the fraction of shock model line ratios that contribute to the result. The pure shock modelsare indicated with black circle markers and the pure photoionisation models with white diamonds. Theinput shock velocities and ionisation parameters are labelled for the pure shock and photoionisation modelseries, and the marker size corresponds to shock velocity. The BCG data are shown as in Figure 4.3.
4.3 Discussion 103
are less well distinguished by the [S ii]/Hα measurements. This is further seen in the
[O i/Hα – [O iii]/Hβ plot, where the data are offset from the predictions of the models.
The points are closely clustered and do not exhibit evidence for the effects of a mixing
sequence.
The different behaviour of the measured line ratios among the three flux-ratio di-
agnostics indicates a greater complexity in the excitation source than is represented by
the simple mixing model. More extensive modelling is needed to investigate this phe-
nomenon. As mentioned previously, this behaviour could be produced by an ionising
field that heats the partially-ionised regions in which the [O i] emission originates. The
ambient X-ray background of the hot ICM that permeates the BCG might contribute to
such an effect. We investigate this in a preliminary way by recomputing a subset of the
shock models described above with the addition of an input bremsstrahlung radiation
field.
The shock velocity in these models is 200 km s−1, and the gas abundances are twice
solar, as previously discussed. The total emitted surface flux of the bremsstrahlung
field varies between F brem = 0.002 and 0.512 erg s−1 cm−2, and is incremented in steps
of logF brem = 0.30 among the set of models. We assume an effective temperature of
2× 107 K for the X-ray source. Figure 4.11 shows the flux-ratio predictions for this set
of shock+X-ray models plotted on the optical diagnostic diagrams (black markers). We
find that the addition of this photoionising field has the effect of enhancing the relative
line fluxes of the [O i] and [S ii] lines over the [N ii] emission, as required to reproduce
the variation in the line ratios of Figure 4.3. When this sequence of models is mixed
with the photoionisation sequence, the predicted line flux ratio mixing lines traverse
the LINER and LINER/H ii composite regions of the diagrams. In addition, the ratios
extend progressively over the more extreme LINER region in the [S ii]/Hα – [O iii]/Hβ
and [O i]/Hα – [O iii]/Hβ diagrams. More detailed simulations are needed to better
model the effect of including the influence of the X-ray radiation on the emitting gas,
and the geometries of the gas clouds. However, these results suggest that this mechanism
could make a contribution in the manner required.
Figure 4.11 Mixing sequence between a set of 200 km s−1 shock models with an inputbremsstrahlung X-ray ionising field and the photoionisation models. Values of the total emittedflux of the X-ray source logFbrem (erg s−1 cm−2) are labelled.
104 4. Shock excitation in line-luminous BCGs
Voit & Donahue (1990) and Donahue & Voit (1991) have previously investigated
models of photoionisation by powerful extreme-ultraviolet and soft X-ray emission in
the context of exciting the emission from filaments in BCGs. They found that models of
cooled condensations of gas irradiated by the X-ray flux from the surrounding medium
could reproduce the properties of the emission-line spectra in a range of systems, but
required extremely high mass flow rates in the core, which are not consistent with the
more moderate cooling rates that are inferred in BCGs. However, the results support
the possibility that this mechanism operates as a supplementary excitation process.
4.4 Conclusions
We have studied the extended line emission in a small sample of line-luminous BCGs us-
ing optical integral-field and long-slit spectroscopy. We present details of the morpholo-
gies and kinematics of the emission regions in the individual systems in Appendix A.
The nebulae show complex structures and velocity fields. We observed evidence for gas
rotation, galaxy interactions or mergers, and interactions between the radio outflows and
optical nebulae in these galaxies. Improved spatial resolution and sensitivity are needed
to map the structures in more detail and understand the kinematics on smaller scales.
We observe significant variations in the emission-line flux ratios among the galaxies
and have examined these using optical diagnostic flux-ratio diagrams. The emission
properties of the extended nebulae provide strong evidence that shock excitation is
important in these systems. All have velocity dispersions that are higher than expected
in purely star-forming regions and shocks provide a natural explanation for this. In
addition, the velocity dispersions are correlated with the line flux ratios, as expected
in shock-excited gas, in at least three of the observed systems. There is also evidence
for a weak correlation in the emission from Sersic 159-03 and 2A0335+096 (discussed
in Chapter 3). This suggests that shocks are important in these galaxies also, as was
surmised for NGC 4696 in Chapter 2.
The optical line flux ratios measured in the emission nebulae are characteristic of
LINER or composite LINER/H ii-region emission spectra, according to the standard
classification schemes for emission-line galaxies. However, we observe that the distribu-
tion of flux ratios behave differently among the three diagnostics that were compared.
The galaxies exhibit distinct [N ii]/Hα ratios that form a sequence across the diagnostic
diagram. However, this is less evident, though arguably still present, in the [S ii]/Hα –
[O iii]/Hβ diagram, and is absent from the [O i]/Hα – [O iii]/Hβ diagnostic, in which
the measurements for all systems are tightly clustered. This will require further detailed
models to investigate more closely. However, we suggest that the influence of the X-ray
radiation from the hot ICM gas could contribute.
We propose that a mixing sequence between shock and photoionisation excitation can
produce the variation in [N ii]/Hα and [S ii]/Hα flux ratios that is observed among the
4.4 Conclusions 105
BCGs. As has been previously suggested, star formation appears to be increasingly dom-
inant over the secondary heating mechanism responsible for exciting the low-ionisation
spectrum in the most luminous BCG systems.
106 4. Shock excitation in line-luminous BCGs