SPIRAL observations of the radio galaxy MRC B1733—565

Abstract

We present optical integral field spectroscopy of the elliptical host galaxy of MRC B1733—565 taken with SPIRAL on the Anglo-Australian Telescope. Emission-line gas is measured to be rotating about and extended perpendicular to the radio axis. To the south-east, the line emission reaches well beyond the continuum light from the galaxy, ending in a region of enhanced starburst activity, high density and disrupted gas flow. We suggest that this is an interacting gas and dust-rich starburst galaxy from which the AGN host is accreting gas along the extended emission-line region.

galaxies: active, galaxies: elliptical and lenticular, cD, galaxies: individual: MRC B1733—565, galaxies: kinematics and dynamics, galaxies: Seyfert, galaxies: structure

1 Introduction

The hosts of powerful double-lobed radio galaxies are typically giant elliptical galaxies (Carballo et al. 1998). The orientation and axes dimensions of the ellipsoid can be modelled from the radio axis position angle and the rotation axis of the stars and gas. It is not uncommon in elliptical hosts for gas and/or stars to be rotating about an axis perpendicular to the radio axis and decoupled from each other (Van Albada, Kotanyi & Schwarzschild 1982; Bryant & Hunstead 2000). This is thought to be due to the introduction of gas from a merger or interaction with a gas-rich galaxy. To understand the relationship between mergers and stellar/gas dynamics the rotation axes need to be measured in ellipticals that show other signs of mergers. Many elliptical galaxies with radio jets show evidence of mergers through dust lanes, extended emission-line gas and extranuclear starburst regions (Kembhavi & Narlika 1999 and references therein).

Photoionization by the nuclear source is often associated with an ionization cone structure in the emission-line gas (see for example Wilson & Tsvetanov 1994). However, the extent and distribution of the gas in some cases is too large to be photoionized by the nuclear source, and it is not known whether it may instead be ionized by shocks along the radio jet or in the starburst regions. The former case would imply a coincidence between the radio jet structures, and the distribution of ionized gas. However, if either shocks from supernovae in starburst regions or photoionization by young OB stars are the ionization mechanism then the gas distribution should be associated with regions of extranuclear starbursts, identified from line ratios. Mapping the starburst regions in galaxies with radio jets and extended ionized gas may identify the ionization mechanism of that gas.

Integral Field Spectroscopy is an ideal technique to investigate these questions on the dynamics, interactions and morphology of elliptical host galaxies for the following reasons.

• Emission-line images trace the extent and distribution of the ionized gas. The ionization source of the gas may then be deduced from a morphology which is conical, follows the radio axis, or is associated with starburst regions.

• The gas and stellar dynamics can be mapped across the galaxy from the emission and absorption lines respectively. This may identify both disrupted dynamics due to a current merger or decoupling of the gas, as in triaxial systems. It is not possible to measure the rotation axes accurately from long-slit spectroscopy alone.

• The ratio of emission-line images can distinguish AGN and starburst processes, defining the extent of the nuclear AGN excitation and the distribution of extranuclear starbursts.

• A colour image formed from the ratio of red and blue continuum light may show direct evidence of a current merger and highlight starburst regions.

MRC B1733—565 (=PKS B1733—56) is a double-lobed, edge-brightened (FR II) radio galaxy (Sault & Wieringa 1994) with a 17 mag E0 host located at  

The radio axis is at a position angle (p.a.) of 38 ± 1°(Hunstead et al. 1982) and the major axis of the ellipsoid is at a p.a. of 130 ± 10°(Govoni et al. 2000). The redshift of 0.0985 (a distance of 600 Mpc, H₀= 50 km s⁻¹ Mpc⁻¹) was measured from an optical spectrum taken with the IPCS at the Anglo-Australian Telescope (AAT) in 1981 (Hunstead et al. 1982). MRC B1733—565 was chosen as a commissioning target for integral field spectroscopy because the IPCS spectrum shows strong [O iii]λλ4959, 5007 and Hβ emission lines, together with an underlying stellar continuum. The observations are described in Section 2. Section 3 presents the spectral line measurements and emission-line images, while our model of the observed features is discussed in Section 4.

2 Observations and Reduction

SPIRAL is the new optical integral field unit on the Anglo-Australian Telescope (AAT) (Lee & Taylor 2000). It incorporates a lenslet array mounted at the f/8 Cassegrain focus, with optical fibres from each lens feeding a spectrograph located on the dome floor. The field of view is 22 × 11 arcsec², comprised of 512 spectra, one for each 0.7-arcsec spatial element. Wavelengths from 4500—10 000 Å are accessible, with a range of 1350 Å per spectrum with the 316R grating.

During the first commissioning run of SPIRAL, in 2000 March (Lee 2000), MRC B1733—565 was observed in 0.7-arcsec seeing. The total exposure time was 90 min. With the 316R grating the resolution was ∼1700 at 6000 Å and the wavelength coverage was from 5260–6603 Å, which included the [O iii]λλ4959, 5007 and Hβ emission lines. The 22 × 11 arcsec² field covered most of the elliptical galaxy, with 110 of the 512 spectra showing emission lines of sufficient S/N to measure without the need for binning. The composite spectrum in Fig. 1 was formed from the addition of all 110 individual spectra. As no flux standard observation was available the spectra were not flux calibrated.

Figure 1

Composite spectrum of MRC B1733—565, formed from the sum of the spectra in each of the 110 pixels in which spectral lines could be measured. The blue wings apparent on the strongest lines are an instrumental artefact.

The reduction software package, 2dfdr, originally written for 2dF data was adapted for SPIRAL reduction by Jeremy Bailey¹ and used to reduce the MRC B1733—565 data. This involved flat fielding, throughput calibration, sky subtraction then combining the separate data frames. As the brightest galaxy emission extends from the south-east to the north-west, sky subtraction used spectra from the north-east and south-west of the field in order to avoid galaxy emission.

2.1 Images

The 2dfdr software produces emission-line and continuum images by selecting the wavelength range required. Images were formed for the emission lines of [O iii]λ5007 and Hβ by selecting the narrowest wavelength range that included not only the line at the heliocentric redshift, but also when shifted in wavelength by rotation within the galaxy. Regions of the continuum either side of the line were also extracted, scaled, and subtracted from the emission-line images. The resultant line images are shown in Fig. 2.

Figure 2

Emission-line images in (a) [O iii]λ5007 and (b) Hβ lines. Smoothed continuum contours are overlaid, with contour levels of 9, 12, 18, 24, 30, 42, 55, 67, 79, 91 per cent of the peak for (a), but without the three highest contours in (b). The images are 22 × 9.8 arcsec² with north to the right and east at the bottom. The bright object to the north-west of the galaxy is a foreground star. Regions A, B and C are defined to be the nucleus, the region where there is a gap in the Hβ emission, and the south-east patch of Hβ emission respectively. The position angle of the radio axis is marked by the solid line. A black, dotted line outlines the region included in Fig. 3. The position angle of the RGO slit (see Section 3.3 ) is shown by a white dotted line.

Fig. 3 is an approximate R− Vimage formed by dividing the reddest and bluest regions of continuum from the SPIRAL data. The regions were chosen to avoid sky lines, and have a wavelength separation of ∼1250 Å.

Figure 3

An ‘ R − V’ colour image formed from the division of continuum images at either end of the SPIRAL spectrum, separated by ∼1250 Å. Lighter regions correspond to redder colours. The continuum peak (assumed to align with the galaxy nucleus) is marked by a cross. The continuum peak has a real offset of ∼0.5–1 pixels (0.35–0.7 arcsec) to the south of the maximum emission-line intensities. The image has been truncated to 11.9 × 8.4 arcsec with north to the right and east to the bottom. To exclude noisy pixels away from the emission-line regions, only the area marked by the black dotted line in Fig. 2 is shown. A dashed line highlights the halo of the galaxy which is bluer than the nucleus in all directions apart from the south-east. (b) The same as (a) with the smoothed contours of the Hβ emission-line image. Contour levels are 10, 15, 20, 25, 40, 60, 80, 90 per cent of the peak.

2.2 Spectra

Spectral lines at each spatial position were measured with iraf splot. The [O iii]λ5007, [O iii]λ4959 and Hβ lines had sufficient S/N to be measured in 110 individual spectra covering the brightest parts of the galaxy and extending to the north-west and south-east in a well-defined band. The pixels adjacent to these, to the north-east and south-west, were binned 2 × 2 to give 1.4 × 1.4 arcsec² regions, and an attempt was made to measure the emission lines in the binned spectrum. However, the emission-line intensity falls so steeply at the edge of the band of individual pixels that the lines were still too weak to be measured in the binned spectra. This suggests that the emission-line gas is distributed in a disc as discussed later in Section 3. The pixels near the nucleus show both broad- and narrow-line components which were deblended with iraf ngaussfit.

The rotation velocities of the gas, and the ratio of narrow [O iii]λ5007/Hβ was calculated in each pixel. Owing to their proximity in wavelength, the [O iii]λλ4959, 5007 and Hβ lines were not corrected for extinction. The values were then reformed into a rotation image and a line-ratio image respectively. Fig. 4 shows the rotation image from [O iii]λ5007. The line-ratio image could be made by two methods, fitting lines in individual spectra or dividing line images. Hβ was not measurable in as many single-pixel spectra as was [O iii]λ5007. Consequently, while making a line-ratio image from fitting lines in individual spectra gives accurate line-ratio values, the resultant image has blank pixels where either or both lines could not be measured.

Figure 4

(a) Rotation image made from the [O iii]λ5007 emission line. The lightest pixels represent a receding velocity of ∼200 km s⁻¹ while the darkest grey pixels mark approaching velocities of the same magnitude. Regions in which the emission-line could not be measured in a single spectrum are filled by black pixels. The field size is 9.8 × 9.8 arscec² with north to the right and east at the bottom. The line marks the radio axis p.a. of 38°. Smoothed continuum contours are overlaid with the same contour levels as in Fig. 2(a). A, B and C are defined as in Fig. 2(b). (b) Rotation curve along a one-pixel-wide line perpendicular to the radio axis and centred on the nucleus. Pixels are omitted to the north-west, where the foreground star interferes. A curve is fitted to the central 8 pixels, however, velocities are disrupted from the curve further than 4 pixels (∼3 arcsec) to the south-east. Typical errors are indicated by the bar. (c) Rotation curve along a one pixel wide line parallel to the radio axis and centred on the nucleus showing no significant rotation along the radio axis.

An alternative method is to divide the original emission-line images. This method shows the distribution of the ratio well, but is not sufficient to get accurate diagnostic line ratios. Therefore, the divided emission-line images formed the ratio image in Fig. 5, while the emission-line ratio values were obtained from the measured spectra.

Figure 5

[O iii]λ5007/Hβ image in the regions A, B, and C. The image was formed from dividing the emission-line images, while the ratio values were measured from the spectra in regions A and C. Pixels have been blanked out in regions of noise where one or both of the emission lines is too faint to give an accurate indication of the ratio. Darker shades correspond to higher ratios. Values much above ∼5 indicate AGN activity while those below ∼3 are typical of starbursts. The image is 7 × 7.7 arcsec² with the continuum peak marked by the cross. Spectra are shown for the peak pixel at A, and for one pixel coincident with the south-east region of Hβ emission at C. The noise is much higher in the south-east pixel. However, it is clear that the Hβ line is much larger with respect to the [O iii ]λ5007 line than is the case at the nucleus. This is a clear indication of starburst activity to the south-east, where the ratios are around 1–2. The bright pixels near the nucleus have typical values of 8–9, as expected for an AGN. The blue wings on the lines in region A are a product of the SPIRAL optics.

The absorption lines of Na iD and Mg ib did not have sufficient S/N to measure lines in many individual pixels. The errors were large, making it difficult to measure stellar velocities accurately. Therefore, regions of 12 pixels were binned perpendicular and parallel to the radio axis in an attempt to establish the sense of stellar rotation.

3 Results

3.1 Spectra

In the composite spectrum in Fig. 1, the [O iii]λλ4959,5007 and Hβ lines are prominent. While Mg ib and Na iD absorption lines are clear in the combined spectrum, they were difficult to measure in the individual or binned spectra as both are weak, and are affected by nearby emission lines. The emission line to the longer wavelength side of the Mg ib absorption line is [N i]λ5197.9, while He iλ5876 is on the blue edge of the Na iD line. These emission/absorption-line pairs were deblended with iraf ngaussfit. The ratio of the [N i]λ5197.9 and He iλ5876 lines to [O iii]λ5007 in the brightest pixel (assumed to be the nucleus) are given in Table 1 along with Hβ and [O iii]λ4959. The line ratios change substantially in different regions of the galaxy. For comparison, the ratios measured in 10 binned pixels centred ∼5.5 arcsec south-east of the nucleus (p.a. ∼130°) in region C are listed in the same table.

Table 1

The ratio of emission-line fluxes to the [O iii]λ5007 line in the brightest (nuclear) pixel and in region C, ∼5.5 arcsec south-east (p.a. =130°)of the nucleus.

3.2 Images

The continuum-subtracted [O iii]λ5007 emission-line image in Fig. 2(a) reveals that ionized gas extends to >35 kpc (>12 arcsec) at a p.a. of 130°± 5°. Therefore, the gas distribution is perpendicular to the radio axis which sits at 38°± 1°. This excludes shocks along the radio jet as the excitation source of the emission-line gas and does not support photoionization by the nuclear source, typically indicated by an ionization cone morphology. Therefore, an alternative excitation mechanism for the extended emission-line gas must be found. This will be addressed in Section 4. Furthermore, the gas distribution is asymmetric. The ionized gas morphology to the north-west follows the continuum contours, but an extension of the gas to the south-east is not apparent on the continuum image.

The distribution of Hβ emission (in Fig. 2b) is somewhat different from that of [O iii]λ5007. While the morphology follows the continuum contours, the extended south-east gas seen in [O iii] is not clear in Hβ. Instead, there appears to be a gap between the Hβ associated with the galaxy continuum, and a patch of Hβ aligned with the end of the [O iii] extended gas. Two of the possible explanations for the Hβ distribution are first that there is Balmer absorption of the Hβ line in the region along the extended gas where Hβ is weak (defined hereafter as region B), or secondly that there may be two separate sources of Hβ emission, one from the AGN host galaxy and one from another source at region C, defined to be the south-east patch of Hβ aligned with the end of the extended gas. It is difficult to correct the line-ratio measurements for Balmer absorption, as this would require subtraction of a template stellar spectrum. However, as the possible Balmer absorption appears to only affect a small region along the extended gas, this would suggest that the stellar population is different there to elsewhere in the galaxy. As such, a generic template would never adequately model the stellar absorption in the host as well as in this region. While it seems contrived that the stellar population should be different in this small area, the possibility of Balmer absorption cannot be ruled out, and would require integral field spectra covering the Hα line to resolve the issue. The Hβ distribution will be considered further in Section 4 in the light of models for this galaxy.

3.2.1 Rotation images

The rotation image in Fig. 4 shows the pixels which had sufficient S/N to measure the [O iii]λ5007 line without binning. After binning the surrounding pixels by four, the [O iii]λ5007 line was still too faint to measure. This implies an edge to the emission-line gas morphology. The systemic velocity measured at the nucleus is in agreement with the previously published redshift. Gas velocities were measured relative to the systemic velocity and found to be up to ∼±200 km s⁻¹ with a rotation axis at a p.a. of 43 ± 7°. The sense of rotation is such that the gas is receding to the north-west and approaching from the south-east. The gas distribution and rotation imply a partially edge-on disc (<20 kpc thickness) of gas centred on the nuclear source and rotating about the radio axis.

Zero (systemic) gas velocities are found not only at the nucleus, but also along the radio axis. The velocities increase uniformly with distance from the nucleus to the north-west. They increase with same regularity to the south-east but only as far as the continuum emission extends (see Fig. 4). Beyond this distance to the south-east, the gas flow is significantly disrupted, and no longer increases uniformly. Therefore, there is disturbed gas flow in the region of the extended gas (region B) and the separated patch of Hβ emission (at C).

The rotating gas could be measured out to 14.5 kpc (5 arcsec) from the nucleus to the north-west and 20.4 kpc (7 arcsec) to the south-east. Within this distance to the north-west, the rotation velocity continued to rise. This is not surprising as some elliptical galaxies have rotation curves that do not turn over out to 3–4 effective radii (Capaccioli et al. 1993; Bryant & Hunstead 2000). In the case of MRC B1733—565 the gas rotation was measured out to ∼1 effective radius, which is the edge of our observed field of view.

From the Mg ib absorption line in binned regions either side of the radio axis (see Section 2.2), stars were found to be rotating in the same sense as the gas (receding to the north-west, approaching from the south-east) showing no evidence of decoupling between stellar and gas rotation. A much longer integration time would be required to map the stellar rotation in detail.

3.2.2 [O iii]λ5007/Hβ image

The ratio of [O iii]λ5007/Hβ is often used as a diagnostic ratio to discriminate between starburst and AGN processes. Typically values above ∼5 indicate AGN activity, while below ∼3 would suggest a starburst region. The ratio of [O iii]λ5007/Hβ shown in Fig. 5 has values around ∼8–9 at the nucleus (A), indicating AGN activity as expected. Some degree of starburst activity is found throughout the AGN host. This is typical of giant ellipticals which frequently have extranuclear starburst regions surrounding an AGN core. These starbursts are thought to be induced by the introduction of gas, perhaps from mergers.

There are two unusual features in Fig. 5. First, the AGN region defined by the [O iii]λ5007/Hβ ratio appears to be quite extended into region B, to the south-east. This is unusual as the AGN is expected to be compact. Therefore, it is likely that the line ratios in region B are artificially enhanced either by Hβ absorption or by the lack of Hβ in this region owing to two separate Hβ sources, as discussed in Section 3.2. The second unusual feature is that the south-east patch of Hβ emission in region C, exhibits ratio values around 1–2, strongly indicative of a starburst.

3.3 RGO spectrum

Simpson et al. (1996) obtained a long-slit spectrum of MRC B1733—565 in 1992 with the RGO Spectrograph on the AAT. The slit p.a. was 141° (marked by the dotted line in Fig. 2), which is approximately aligned with the south-west edge of the extended gas. The reduced, but not flux calibrated, spectral image was obtained from Simpson so that spectra could be formed from regions along this extended gas. The wavelength coverage, from ∼4000–5300 Å, is blueward of the SPIRAL spectra and includes the redshifted lines of [O ii]λ3727, [Ne iii]λ3869, Hγ, [O iii]λ4363 and He ii λ4686.

In order to form emission-line ratios between the lines observed by SPIRAL and those from the Simpson et al. RGO spectrum, the two sets of line measurements needed to be cross-calibrated. As neither spectrum was flux calibrated, this process relied on scaling the spectra to reproduce the flux ratios measured independently by Simpson et al. (1996) and Hunstead et al. (1982) from spectra extracted to include the total galaxy. However, we need ratios at different positions along the slit rather than total galaxy values. As the line ratios change in different parts of the galaxy, a region of the SPIRAL data was extracted to simulate the slit used in Simpson et al. (1996), and the same size region was extracted from the Simpson et al. RGO spectrum. Therefore, the ratio of SPIRAL-to-RGO lines in these total galaxy spectra were then compared with the ratios found by Simpson et al. (1996) and Hunstead et al. (1982) in order to scale the two spectra. Lines measured in different regions along the Simpson et al. RGO slit could then be divided directly by line fluxes from the SPIRAL data extracted in the same region. This method clearly has uncertainties based on: (i) differences in the ratios found in the two papers that were used as calibrators, (ii) due to the size of SPIRAL pixels it was difficult to select a region of pixels that exactly matched each region extracted along the Simpson et al. RGO slit, and (iii) the errors in line measurements become larger for regions extracted away from the nucleus. The resulting line ratios, given in Table 2, include estimated errors which take into account all of the above factors.

Table 2

Line ratios measured by combining the spectrum obtained from Simpson and the SPIRAL data, for the nucleus (A), along the south-east extended gas where Hβ is absent (B), at the south-east Hβ region (C), and at a region symmetric with B on the north-west side of the nucleus (NW). [O iii] refers to λλ4959 + 5007 except where indicated, while [Ne iii] is λλ3869 + 3968.

The Simpson et al. long-slit RGO spectrum has spatial resolution of ∼0.6 arcsec pixel⁻¹. Regions were extracted to include: (a) four pixels centred on the nucleus in region A, (b) four pixels at the position where the Hβ emission is ‘missing’, from a distance of 4 to 7 pixels to the south-east of the nucleus (region B), and the same to the north-west, and (c) four pixels roughly coincident with the apparent starburst region (at C), from 8 to 11 pixels from the nucleus to the south-east only (the corresponding distance to the north-west is contaminated by the foreground star). The same regions were taken from the SPIRAL data by matching the pixels as accurately as possible to the area in arcsec of the Simpson et al. long-slit regions.

4 Discussion

In Section 3.2 and Fig. 2(a) we found that the continuum contours follow the [O iii]λ5007 distribution to the north-west, while south-east of the nucleus the ionized gas clearly extends well beyond the continuum emission. In the rotation image (see Fig. 4 and Section 3.2.1), the smooth gradient in rotation to the north-west again is aligned with the continuum. However, the rotation is disturbed further to the south-east than the continuum emission which extends only to a distance of 3 arcsec (8.7 kpc) from the nucleus. It is noteworthy that in detailed elliptical isophote fitting of the R-band image of MRC B1733—565, Govoni et al. (2000) found the Fourier coefficient, c4, to be positive until a radius of 3 arcsec, but then to decrease abruptly to negative values beyond this radius. This suggests a disc-like morphology in the inner region, but boxy isophotes (Govoni et al. 2000) outside the distance from the nucleus beyond which we have measured disturbed gas rotation. Furthermore, this same region (region C) is coincident with the proposed starburst position (see Section 3.2.2).

A plausible explanation for the disturbed morphology, disrupted gas rotation, the patch of starburst emission and the extent of the emission-line gas to the south-east, would be an interaction or merger. Two possible merger scenarios may apply. The first would involve a gas rich merging galaxy lying beyond our images to the south-east, from which gas is being accreted towards the nucleus of MRC B1733—565. As this gas settles into rotation about a principal plane the gas flow is disrupted. The injection of gas triggers star formation, which in turn ionizes the gas as it settles into rotation about a principal axis of the AGN host. In this case some level of star formation would be expected not only on the merger side (south-east of the nucleus), but at a similar distance on the opposite side due to rotation of the merging gas. In this picture it is difficult to explain such strong starburst emission in region C compared with B. It is possible that Balmer absorption of Hβ in region B is artificially increasing the [O iii]/Hβ ratio to indicate weaker star formation. In that case, region B would need to have a different stellar population to the other regions of the galaxy and the merging gas, and it is not clear why this should be so. The DSS-II and SuperCOSMOS images show no obvious merging galaxy beyond the SPIRAL field of view. There are, however, some faint structures that would require much deeper optical images to identify.

In the second merger scenario, the patch of strong starbursts (at C) could be the merging gas-rich galaxy itself from which gas and/or dust is being accreted along the south-east gas extension into the AGN host. In this case the region with reduced Hβ emission would not be due to Balmer absorption, but more simply because there are two separate sources of Hβ, one from the AGN host galaxy, and one from the merging starburst galaxy. This would explain more readily the symmetric distribution of star formation within the continuum halo as caused by gas which has merged into an orbit about the AGN host.

A strong argument in favour of the second scenario comes from the colour distribution. In Fig. 3 the nucleus is 0.2–0.3 mag redder than the galaxy halo surrounding the nucleus except to the south-east. In giant ellipticals the halo is typically bluer than the nucleus (Govoni et al. 2000) which is the case to the north-west. However, to the south-east the colours become progressively redder with increasing distance from the nucleus, reaching 1.2 mag redder than the nucleus at the starburst region, C. Fig. 3(b) shows the Hβ contours on the R−V image. The reddest region aligns with the strong Hβ patch (in region C). This is unusual, as the emission-line ratios here clearly indicate starburst activity, yet starbursts are expected to be bluer than typical galaxy colours. Therefore, if this region was a starburst patch induced by a merging galaxy beyond the field, as in the first merger picture, then it would be expected to be bluer. However, if this patch is the merging galaxy it could simply be rich in gas and dust both of which are being accreted on to the AGN host. In order to test this option, the distribution of dust would need to be determined from infrared images.

Further support for the second merger picture comes from testing other starburst/AGN diagnostic ratios that are not affected by Balmer absorption, such as [O ii]λ3727/[O iii]λ5007. While the [O ii] line is not within the wavelength range of the SPIRAL data, it is in the spectrum obtained by Simpson et al. (1996) (see Section 3.3), and therefore the regions A, B and C along the Simpson et al. RGO slit can be tested. Fig. 6 shows the ratio of [O ii]λ3727/[O iii]λ5007 vs [O iii]λ5007/Hβ with regions A and C marked. Typical values for AGNs and starburst/Liners are shown (Terlevich & Melnick 1985). The nuclear region (A) lies comfortably amongst the AGNs as expected. On the other hand, region C is typical of galaxies with a starburst nucleus. Also marked on the plot are region B and the corresponding region to the north-west of the nucleus. Both of the latter points are indicative of some star formation. Dereddening of these values would shift them to the left on Fig. 6. Significant reddening would be required in order for the points B, C and NW not to show some star formation. The optical images have no apparent dust which would be indicative of significant reddening. These line ratios all support a model in which region C is a separate merging starburst galaxy from which gas is being accreted on to the AGN host, inducing some star formation around the AGN host as it settles into a principal orbit.

Figure 6

Log [O iii]λ5007/Hβ vs Log [O ii]λ3727/[O iii]λ5007. The distribution of AGNs and starburst/liners in a sample by Terlevich & Melnick (1985, fig. 5) are marked along with points representing regions A, B, C and a north-west region as defined in the text.

The first merger scenario of an extended starburst region with a separate stellar population giving Balmer absorption, is not ruled out by Fig. 6. However, there is no strong evidence for Balmer absorption. While the region B point is at the limit of values allowed for a starburst region according to the [O iii]/Hβ ratio, it is in the middle of the starburst range allowed by the [O ii]/[O iii] ratio. As both ratios indicate starbursts and the [O ii]/[O iii] ratio is not affected by absorption, it is unlikely that the [O iii]/Hβ ratio is increased sufficiently to indicate Hβ absorption.

4.1 Physical conditions in regions A, B and C

Ionized gas emits a line most efficiently when the gas density is near the critical density for that species. Therefore, the emission lines can be used as tracers of gas density. Furthermore, the temperature can be deduced from the line ratios and densities. In particular, [O iii]λλ4959 + 5007/[O iii]λ4363 (defined as R) is related to temperature and density by  

1

(Seaton 1975). R acts as an excellent thermometer until n_e≥ 10⁵ cm⁻³, where this ratio becomes dependent on density as [O iii]λ4363 has n_e (crit) ∼3 × 10⁷ cm⁻³ compared with ∼7.9 × 10⁵ cm⁻³ for [O iii]λλ4959 + 5007. To identify higher densities an alternate indicator such as [O iii]λλ4959 + 5007/[Ne iii]λλ3869 + 3968 is required. The relationship between these ratios and the electron temperature and density is shown in Fig. 7 (equation 1; also Heckman & Balick 1979). From this figure, the temperature and density at the nucleus (region A) are T_e= 2.0 ± 0.2 × 10⁴ K and n_e∼ 10^5.3±.1 cm⁻³. However, in region C the mean density of n_e= 10^6.8 cm⁻³ is ∼30 times that at A, with a range of 10^6.3≤n_e≤ 10⁸ cm⁻³. The temperature of this region is difficult to determine due to uncertainties in deblending the [O iii]λ4363 line. However, the possible values are shaded in Fig. 7 bounded by the lowest reasonable isotherm of 10⁴ K, the lower limit on the [O iii] ratio, and the measurement errors on the [O iii]/[Ne iii] ratio. The temperature could therefore range from 10⁴ K to 4 × 10⁴ K.

Figure 7

Log [O iii]λ4959+5007/[O iii]λ4363 versus log [O iii]λ4959+ 5007/[Ne iii]λ3869. The temperatures and densities are shown for these ratios (Seaton 1975; Heckman & Balick 1979 ). A point is shown for the ratios in region A, while the limits on the ratios define the shaded areas for regions B (pale grey) and C (dark grey), as explained in the text. Dereddening of the values would shift the points down and to the left of the figure. This would have a marginal affect on the temperature, but may increase the density values of all points. Owing to the uncertainties for each region, the conclusions below would not change.

The significantly higher density in region C supports the model of this region being a merging galaxy which is gas rich and is redder owing to excess dust. If this is the case, and gas is being accreted from C on to the host galaxy then the density in region B may be expected to be intermediate between the two galaxies. From the line uncertainties, only an upper limit to the density could be fixed. The resultant value of n_e≤ 10^6.7 cm⁻³, as shaded on Fig. 7, supports an intermediate density, as expected for gas being accreted from a higher density merging galaxy. Furthermore, gas with n_e∼ 10⁷ cm⁻³ should have Hβ emission enhanced relative to [O iii]λ5007 as n_e is significantly greater than the critical value for [O iii]. At these densities, Balmer emission is proportional to n²_e, but [O iii] intensity is proportional to n_e. Therefore Hβ emission should be stronger relative to [O iii]λ5007 in a high density region compared with those of low densities. It is therefore not surprising that region C has enhanced Hβ, as the density is high at ∼10⁷ cm⁻³. In region B the density is much lower, so that Hβ would not be expected to be enhanced. As the high density at C is likely to be the product of a different environment, this lends support to the merger model in which C is a separate source of Hβ.

The starburst region, C, is likely to be the excitation source for the extended gas to the south-east of the nucleus. This may be due either to photoionization by OB stars, or shock ionization from SNRs. Purely photoionized gas can be driven to a maximum temperature of ∼2 × 10⁴ K while values higher than this require another excitation mechanism. If shocks are the dominant heating mechanism then the temperature would need to be T_e≥ 3 × 10⁴ K (Filippenko & Sargent 1988). Therefore, the temperature in region A is entirely consistent with photoionization, as expected, owing to the contribution from the AGN source. In regions B and C, while photoionization is possible, the ranges include temperatures that cannot eliminate additional shock excitation. Therefore, from temperatures alone it is not possible to distinguish between shock or photoionization in the starburst region, although pure shock excitation is unlikely.

Several authors have modelled shock and photoionization processes based on the line ratios of [O iii]λλ4959 + 5007/Hβ and [O ii]λ3727/[O iii]λλ4959 + 5007. In region A, these line ratios indicate photoionization from the models of Clark et al. (1997, Fig. 14a) and Storchi-Bergmann, Wilson & Baldwin (1992, Fig. 18d). In both models, the region B and C values again are indecisive in choosing between shock and photoionization processes.

5 Conclusion

SPIRAL has been very effective in mapping the emission-line gas distribution, rotation and excitation processes in the host of the powerful radio galaxy MRC B1733—565. A region of emission-line gas extending to >35 kpc has been found to be aligned perpendicular to the radio axis, and rotating about it at up to 200 km s⁻¹. Stellar rotation is in the same sense as that of the gas. While the emission-line morphology follows the continuum light to the north-west of the host galaxy, emission-line gas to the south-east extends well beyond the continuum, and has been found to be associated with disrupted gas flow. Enhanced starburst processes and higher densities are evident at the far end of the extended gas. Two merger models have been proposed to explain these findings. The model which appears to best match the observed data involves a gas- and dust-rich starburst galaxy merging from the south-east. Gas is being accreted from this merging galaxy, and settling into rotation about the AGN host galaxy, while inducing some star formation in the process.

The mapping of starburst regions and ionized gas in MRC B1733—565 has shown that the starbursts are the ionization mechanism of the extended gas, rather than the nuclear source or shocks along the radio axis. A current merger has been shown to be disrupting the gas dynamics within the host galaxy.

Acknowledgments

We thank David Lee, Russell Cannon and Jeremy Bailey for obtaining the SPIRAL data during the first commissioning run. Chris Simpson kindly provided his RGO spectral image.

1

http://www.aao.gov.au/local/www/jab/spiral/spiral reduction.html