High-resolution radio structure and optical kinematics of NGC 7027

Transcription

1 Mon. Not. R. stron. Soc. 34, (23) High-resolution radio structure and optical kinematics of NGC 727 I. ains, 1,2 M. ryce, 2 G. Mellema, 3 M. P. Redman 4 and P. Thomasson 2 1 stronomy Group, University of Hertfordshire, College Lane, Hatfield, Hertfordshire L1 9 2 University of Manchester, Jodrell ank Observatory, Macclesfield, Cheshire SK11 9DL 3 Sterrewacht Leiden, PO ox 913, 23 R Leiden, the Netherlands 4 stronomy Group, University College London, London WC1E 6T ccepted 22 October 22. Received 22 October 1; in original form 22 pril 23 1 INTRODUCTION NGC 727 (PN G , α J2 = 21::9.39, δ J2 = 42:2:3.1) is perhaps the most intensively studied planetary nebula (PN) because of its high surface brightness at all wavelengths. It is a young (Masson 1989), dusty (Woodward et al. 1992), carbonrich (Middlemass 199) PN whose ionized core is surrounded by an extended cloud of neutral, molecular material (e.g. Mufson & Marionni 197; Jaminet et al. 1991; Phillips et al. 1991; ieging, Wilner & Thronson 1991), probably originating in the precursor asymptotic giant branch (G) wind. The extent of CO emission (to zero intensity) in the neutral envelope is 7 arcsec (ieging et al. 1991). The morphology of the ionized gas has been modelled ib@star.herts.ac.uk STRCT We present the results of the highest-resolution 18-cm (L-band) radio continuum and optical kinematic studies to date of the young planetary nebula (PN) NGC 727. We discuss the radio image of NGC 727 made from combined data from the Multi-Element Radio-Linked Interferometer Network and the Very Large rray. This image is used to derive the distribution of the nebular brightness temperature at this frequency. y combining the 18-cm image with the 6-cm (C-band) image of ryce et al., two-dimensional distributions of the spectral index between L- and C-bands, C-band optical depth and emission measure are constructed and analysed. In the optical regime, high dispersion, spatially resolved observations at high spectral resolution have been obtained of this PN. The resulting velocity ellipses confirm the inclination of the north of the PN towards the line of sight. waist feature is seen in the equatorial plane and [S II]6716/6731 Å ratios indicate that a density enhancement exists in this region. The PN equatorial expansion velocity is deprojected and measured as 13 ± 1kms 1 from the [O III]499 Å emission line. This velocity is used to calculate a revised distance estimate of 6 ± 1 pc. Using the electron density derived from the [S II] ratio, an ionized mass of. M is determined. The nature of the bright knot situated to the north-west of the PN is investigated at optical and radio wavelengths. In optical emission lines it is found to be redshifted. It is located in the same region as the position of the peak flux density at 18 cm. It is shown that no evidence of a local density enhancement exists in its environs. It is suggested that the appearance of the knot is due to a temperature effect and its existence is interpreted in terms of a breach in the near side of the PN shell. The existence of a jet, at some stage in the evolution of the PN, is invoked to explain the mechanism by which such a breach may have been created. Key words: circumstellar matter planetary nebulae: individual: NGC 727 (PN G ) radio continuum: stars. as a prolate, ellipsoidal shell inclined at 3 to the line of sight (Scott 1973; Masson 1989; Roelfsema et al. 1991; Walsh & Clegg 1994). Measurements of the electron temperature of the ionized region given in the literature range from to 16 K; see, for example, asart & Daub (1987), Masson (1989) and Roelfsema et al. (1991) and references therein. The size of the emission region at optical and radio wavelengths is 8 arcsec 2. t optical wavelengths, NGC 727 is rectangular in shape and dominated by a bright knot, arcsec in diameter, which is situated to the north-west of the nebular emission. The knot is in a position of low extinction in the differential extinction maps of, for example, therton et al. (1979), Walton et al. (1988) and Woodward et al. (1992). The observations of, for example, Woodward et al. (1992) and Walsh & Clegg (1994) suggest that nebular dust is present both mixed with the ionized gas in the H II region and also in the C 23 RS

2 382 I. ains et al. surrounding neutral halo. Woodward et al. (1992) have found that the overall extinction across the face of the nebula is very high, ranging from 1 to 8 mag. Molecular hydrogen emission at λ = 2.1 µm has been detected emanating from a photodissociation region between the ionized shell and the circumnebular halo of NGC 727 (e.g. Graham et al. 1993; Cox et al. 1997). This has recently been imaged as a quadrupolar biconical region using the Near Infrared Camera and Multi-Object Spectrometer (NICMOS) on the Hubble Space Telescope (HST) (Latter et al. 2). The wispy nature of the emitting region revealed in these new images has led the authors to invoke the presence in the PN, now or at some time in the past, of one or two highly collimated off-axis jets, or possibly bipolar, rotating, episodic jets (RETs; López et al. 1998). They propose that one of these jets may fall along an axis 2 3 to the west of the polar axis, extending through the central star to the opposite side of the PN. This proposed jet would pass through the region of the bright knot, situated in the northwest of the PN. Cox et al. (22) invoke the same jet phenomenon to explain three point-symmetric pairs of cavities that they find in their observations of rγ emission from the PN. Indeed, one of these pairs of cavities is along the same position angle (P) as the proposed jet of Latter et al. (2), and it is along this P that Cox et al. (22) find low-level, high-velocity (± km s 1 )rγ emission. In the previous Multi-Element Radio-Linked Interferometer Network (MERLIN) and Very Large rray (VL) images of NGC 727 at frequencies where it is limb-brightened and optically thin, e.g. at GHz (e.g. asart & Daub 1987; Masson 1989; Hajian, Terzian & ignell 1993; ryce et al. 1997b) and 1 GHz (e.g. Masson 1989), it resembles an elliptical shell of emission. t 2 cm, the VL image of asart & Daub (1987) with a resolution of 1.2 arcsec shows a smooth, rectangular structure with a brighter arc of emission to the north-west. The radio proper motion technique adopted by Masson (1989) to derive the distance to NGC 727 utilized a PN expansion velocity of 17. ± 1. km s 1 measured from the [O III] observations of therton et al. (1979), three epochs of VL observations and a prolate, ellipsoidal shell model of the PN. kinematical age of 6 yr was derived using these results. Note that throughout this paper, wherever L- and C-bands are mentioned, these refer to the radio-frequency bands that are centred on wavelengths of 18 2 and 6 cm, respectively. 1.1 This paper lthough NGC 727 has been so comprehensively studied, to date no high-resolution, spatially resolved, optical spectroscopic study has been carried out and no high-resolution 18-cm radio image exists. This aim of this paper is to provide such a multi-wavelength study. High-resolution kinematics are a valuable tool for deprojecting the object and measuring its distance. The radio data provide a complementary study; an alternative method of studying emission that arises in the same region as that of the optical lines but one that does not suffer from extinction effects. This is a pertinent consideration in the case of NGC 727, because the extinction is spatially varying (Woodward et al. 1992) and so the radio data reveal the intrinsic PN shape. In addition to these observations, archive HST Wide Field Planetary Camera (WFPC) images are presented for purposes of comparison with the new results. The 6-cm MERLIN+VL image of ryce et al. (1997b) is also used to allow temperature, optical depth and emission measure to be analysed. Table 1. Observing details of HST WFPC NGC 727 archive data. Date of obs. Exp. Filter Emission time (s) line 1992 October 28 1 F2N [O III]7 Å 1992 October 28 1 F2N [O III]7 Å 1992 October 3 1 F66N Hα 663 Å 1992 October 3 1 F66N Hα 663 Å 1992 October 3 7 F68N [N II]684 Å 1992 October 3 7 F68N [N II]684 Å 2 OSERVTIONS ND DT REDUCTION 2.1 HST data Observations Table 1 gives the parameters relating to the HST WFPC archive data (Proposal I.D. 3642; Westphal J.). The resolution of the WFPC is.1 arcsec Data reduction The data sets retrieved from the archive had already been calibrated using the standard pipeline. The remaining data reduction was performed using the STSDS package in the IRF software. s two exposures with the same observing parameters were available for each emission line, median filtering was used to remove cosmic ray hits. The data were in group format; each group was mosaiced into a single image of the sky. 2.2 MERLIN data Observations The MERLIN observations were made on 199 December 2 using the full MERLIN array (eight telescopes at that time) at 168. MHz. The usable bandwidth was MHz and the integration time for an individual data point was 16 s. The observations were made over an 18-h period in phase referencing mode, spending 7 min on the target source and 3 min on the phase reference calibrator, The primary amplitude (flux) calibrator was 3C Data reduction The initial editing and calibration of the MERLIN data were carried out using the D-PROGRMME package (a MERLIN-specific package available at Jodrell ank) which enabled the absolute flux scale for the observations to be determined. The flux of the point source baseline calibrator, 2+398, was determined from a comparison of its amplitudes on the shortest baselines with those of 3C 286. The flux of 3C 286 was taken to be Jy (åårs et al. 1977) and, from this, the flux of was found to be 2.38 ±. Jy. The data were input to the MERLIN pipeline where they underwent standard calibration procedures in IPS. The MarkII Lovell baseline was flagged due to confusion. In a preliminary image of NGC 727 resulting from the pipeline, the nebular flux appeared extended and of low surface brightness everywhere, hence it was not possible to self-calibrate the data. Inspection and imaging of the data revealed that the PN was poorly sampled by the MERLIN array; more than per cent of the flux had been resolved out and only the three shortest baselines were C 23 RS, MNRS 34,

3 Radio structure and optical kinematics of NGC constraining the extended nebular structure, and then only for a fraction of the observing time when the source was projected in such a way as to be foreshortened on these baselines. The PN structure therefore could not be accurately reconstructed with the deconvolution algorithms. rchive VL data were obtained to combine with the MERLIN data and provide the required short spacings. 2.3 VL data Observations The VL data selected from the archive (Proposal I.D. M169; Masson C.) were the most recent available at approximately the required frequency and array configuration (-array). The observations took place on 1986 March 6 and the results from them are thought not to have been published elsewhere. -array provides a resolution of 1. arcsec at L-band and data taken in this configuration are most desirable as it has baseline lengths which overlap the shorter MERLIN baselines. The data were observed in two -MHz bands centred on intermediate frequencies (i.f.s) at 64.9 and 1.9 MHz. n integration time of 1 s was used in phase referencing mode, with typically 6 min spent on the source and 3 min spent on the calibrator; in total, 4 min were spent observing NGC 727. The phase calibrator was and the primary flux calibrator was 3C 286. Only 26 of the available 27 antennae were used to collect the data; antenna 24 was down for the duration of the observing session Data reduction The flux of 3C 286 is. Jy at 64.9 MHz and.32 Jy at 1.9 MHz. This was used to scale the flux of using IPS task SETJY. The flux of was.26 ±.4 Jy at 64.9 MHz and.23 ±.1 Jy at 1.9 MHz. The data were calibrated using the standard IPS procedures. 2.4 Combining the MERLIN and VL data s the observations using the two arrays took place at different frequencies within L-band, the flux of one data set needed to be scaled before it could be combined with the other. From the literature, the radio spectrum turnover frequency for NGC 727 is at GHz (e.g. Taylor, Pottasch & Zhang 1987; Terzian 1989; Middlemass 199) and so it is likely that the PN is optically thick at L-band. To verify this, the data taken at the two VL i.f.s were mapped and self-calibrated independently and their respective integrated fluxes were measured and compared. The integrated fluxes are 1.48 ±.7 and 1.6 ±.8 Jy at 64.9 and 1.9 MHz, respectively. ssuming a uniform density shell model, in the optically thick regime the flux density S ν varies with observing frequency ν as S ν ν 2. Using the measured integrated fluxes, the nebular flux is varying as ν 2 to within 2 per cent of this model prediction. Using this model, the VL data were scaled up in flux by factors of 1.28 and 1.2 for the 64.9 and 1.9 MHz data, respectively, to their extrapolated values at 168. MHz. From multi-epoch observations, Masson (1989) has shown that the flux of NGC 727 is decreasing at C-band. The L-band MERLIN and VL data were therefore carefully inspected for any signs of variation in flux at this frequency between the two sets of observations. The PN amplitudes sampled by baselines from the two arrays that overlapped were found to be comparable and so any variation of L-band nebular flux over the period between the VL C 23 RS, MNRS 34, and MERLIN observations appeared negligible. ssuming that the 4.2 mas yr 1 (Masson 1989) angular expansion rate of NGC 727 is constant with time, then over the 1 yr between the VL and MERLIN observations, the PN has expanded by 42 mas, which is only 3 per cent of the VL -array beam size. Therefore, the effect of this expansion on the combination of data from the two epochs is negligible. The three data sets (that from the MERLIN observations and those from the two VL i.f.s) were then concatenated and the data aligned spatially. This was necessary as the data from both arrays were observed with slightly different pointing centres, with the VL data in 19 precessed to coordinates and the MERLIN data in 19 coordinates. oth the VL and the MERLIN data were recorded in dual-polarization (LHC and RHC) mode and, in the initial stages of calibration, these hands were processed separately. The dual-polarization combined data from the two arrays were inverted using the IPS task IMGR, with uniform weighting and a cell size of.4.4 arcsec 2. This process produced a dirty map and dirty beam, which were then used as inputs to the IPS task SDCLN. SDCLN was chosen to implement the deconvolution as it is particularly suitable for CLENing extended structures. bout 73 iterations of SDI CLEN were applied, the first 6 with the nebular emission tightly boxed and the last 17 without boxes, to minimize the spurious, off-source structure. SDCLN was stopped when the residual map on the screen became noise-like. restoring beam of.4.4 arcsec 2 was used to make the final Stokes I (total intensity) image. 2. WHT data 2..1 Observations High-dispersion, spatially resolved, long-slit spectral observations of NGC 727 were made on 1994 ugust 2 and 21 using the William Herschel Telescope (WHT) combined with the Utrecht Echelle Spectrograph (UES) and the 4 4 pixel Tek1 CCD detector. The UES was used in cross-dispersed mode with a grating of 79 grooves mm 1, the maximum slit length (permissible with the use of a cross-disperser) of 1 arcsec and a slit width of 1.11 arcsec (22 µm). The slit width was matched to two detector pixels, each µm 2. This gave a velocity resolution of 6. km s 1. The slit length used meant that a wide inter-order separation was necessary, and so full coverage of the desired spectral orders was not possible with the Tek1 CCD. The solution was to observe each slit position at two grating positions, covering either end of the 31st to 4th orders, corresponding to Å. This then included most of the kinematically and diagnostically useful bright optical emission lines. In the spatial direction, with the derotator, the scale at the detector was.96 arcsec mm 1, giving an angular resolution of.36 arcsec pixel 1. The seeing remained fairly constant at 1 arcsec during the observing session. range of exposures were taken at five slit position angles (see Fig. 1 and Table 2). Slit 3 was aligned with the apparent PN minor axis and slit 4 with the major axis; slits 1 and 2 were respectively parallel to these. Slit passed over both the nebular centre and the bright knot. wavelength calibration spectrum was obtained at each slit and grating position using a thorium argon discharge lamp. wide slit ( arcsec) exposure of the flux standard star SP was also obtained at each grating position for flux calibration purposes. Flatfield frames were obtained using a quartz lamp. series of zeroexposure frames were also taken for de-biasing purposes.

4 384 I. ains et al. Table 2. Optical emission line observational data of NGC 727. Parameters of uncorrupted data files only are given. Slit 1 Slit 2 Slit 3 Slit 4 Slit Slit P Slit right right Nebular Nebular right centred on knot knot centre centre knot Emission lines Exposure times (s) Hα Hβ 1 2, [N II]648 Å 1 2, [N II]684 Å 1 1 1, 1 1 [O III]499 Å 1 1 2, 1, 1, 1 1 [O III]7 Å [S II]6716 Å 1 2, [S II]6731 Å 1 2, D E C LC IL N I TN I O NT I ( O 1N 9( J ) 2 ) E N slit slit 4 slit 3 slit 1 slit RIGHT SCENSION (J2) Figure 1. Illustration of the slit positions used when NGC 727 was observed with the UES on the WHT. The slits have been drawn on to a grey-scale of the HST WFPC image of the nebula in the light of Hα 663 Å (Fig. 2a). The annotations and correspond to those on the plots of the spectra in Figs 9 1 and serve to illustrate the orientation of the spectra Data reduction The initial data reduction stages were performed using IRF. The data frames were de-biased, flat-fielded and combined using a median filter in the usual way. The fanning of the orders, arising from the cross-dispersers within the spectrograph, was corrected in an analogous fashion to the wavelength calibration procedure. First, a standard star frame was read into RC2D, one of the TWODSPEC suite of programs, and the continuum light seen along the orders was fitted with a dispersion relation. This dispersion relation was then applied to the data frames using ISCRUNCH, and hence the orders were straightened; see ppendix of ryce & Mellema (1999) for further details. Finally, the data frames were wavelength and flux calibrated in the usual way. 3 RESULTS 1 THE HST IMGES The HST WFPC images of NGC 727 in the light of [N II]684 Å, [O III]7 Å and Hα663 Å are shown in Fig. 2. The images are shown for purposes of comparison with the results presented in DECLINTION (J2) DECLINTION (J2) DECLINTION (J2) RIGHT SCENSION (J2) RIGHT SCENSION (J2) (c) RIGHT SCENSION (J2) Figure 2. The HST WFPC images of NGC 727 in the light of Hα 663 Å, [O III]7 Å and (c) [N II]684 Å. The images are not flux calibrated. The contours are plotted at 8, 16, 24, 32, 4, 48, 6, 64, 72, 8, 88 and 96 per cent of the respective peak fluxes. In each case, the lowest contour is >36 times the sky background level. C 23 RS, MNRS 34,

5 Radio structure and optical kinematics of NGC Sections 4 and 6. They also serve to show the position, in the northwest of the PN, of the bright knot which comprises several clumpy features. The waist-like feature visible in all three images is spatially coincident with the intersection of the molecular H 2 rings (e.g. Latter et al. 2). It is likely that these rings contain dust, which is obscuring the optical emission. 4 RESULTS 2 THE NEW MERLIN+VL L -ND IMGE OF NGC 727 The new L-band MERLIN+VL image of NGC 727 is shown in Fig. 3. This is the highest-resolution radio image of this object at this frequency to date. The image shows mainly smooth emission in the form of a skewed rectangle. To the north-west and south-east of the PN, two opposing spurs of extended emission can be seen. partial shell of brighter emission exists in an arc to the north-west of the central region of the nebula. This arc is defined by the second highest contour in Fig. 3 and encloses the clumps that comprise the bright knot. The north-eastern and south-western edges of the PN are bounded by steep flux gradients. The peak flux density in the map occurs in the north-west region of the nebula with a value of.99 ±.3 mjy beam 1 and the rms noise level is 7 µjy beam 1. The extent of the nebular emission measured from the 3σ contour is 9 arcsec 2. The extent of the optical emission visible in the HST images in Fig. 2 is comparable to that of the L-band radio emission. The position of peak flux density in the radio image occurs in the same region as the bright knot seen in the HST images; the radio emission is also in the form of bright clumps in this region. The region of the PN that is obscured in the optical images by the dust contained in the DECLINTION (J2) molecular H 2 rings (Latter et al. 2) is seen clearly in the L-band emission. The optical images also show some suggestion of a spur of emission to the north-west of the PN, coincident with the spur seen in the L-band image. However, the counter-directed spur seen in the radio is not visible in the optical emission. We note the appearance in Fig. 3 of a roughly circular contour that is centred on a position of α J2 21:7:1.7, δ J2 +42::9.8 and which indicates a decrease in the emission. This is interesting due to its proximity to the line of sight to the central star: α J2 = 21:7:1.732, δ J2 =+42::9.99 (Monet et al. 1998). However, this is probably a coincidence as the L-band emission is thought to be optically thick (Section 2.4) so the trough is a feature of the surface of the emission and not related to any stellar process. The new L-band image is in marked contrast to the C-band MERLIN+VL map of ryce et al. (1997b), where the nebular emission is resolved into a clumpy, limb-brightened shell. The C- band emission appears to be mainly optically thin apart from within the clumps. t L-band, the more uniform distribution of emission is further verification that the nebula is optically thick (Section 2.4), except possibly in the north-west region where the brighter arc of emission is seen. The extent of the C-band emission measured from the 3σ contour in ryce et al. (1997b) is 1 8 arcsec 2. This is less than the extent of the L-band emission, although the larger beam size at the lower frequency will be a significant factor in this. The brightest nebular emission at C-band occurs in a different region to that at L-band; at C-band the peak flux density is found to the south of the south-west limb, with a value of 679. µjy beam 1. The spur of emission to the north-west of the nebula, which is evident in the L-band map and also in the optical HST images, is RIGHT SCENSION (J2) Figure 3. The new L-band MERLIN+VL image of NGC 727. The contours are plotted at 1, 1, 3, 6, 9,, 1, 18, 21 and 24 times the 3σ level of 22 µjy beam 1 and the grey-scale range is from the rms noise to the peak nebular flux density of.99 mjy beam 1. The dashed contours indicate negative flux. The beam is shown in the bottom left of the image. C 23 RS, MNRS 34,

6 386 I. ains et al DECLINTION (J2) RIGHT SCENSION (J2) Figure 4. The MERLIN+VL image of NGC 727 with L-band contours and C-band grey-scale. The C-band data have been convolved down to the same resolution as the L-band data and transformed to the same spatial grid. The contours are as in Fig. 3; the grey-scale range is from the C-band rms noise of 2.6 mjy beam 1 to the peak nebular flux density of mjy beam 1. The dashed contours indicate negative flux. The beam is shown in the bottom left of the image. partially visible in the C-band image although it shows a lesser degree of extension. The spur seen to the south-east of the PN in the L-band image is absent in the C-band map. These spurs of emission are at the greatest distance from the central star, so it is possible that they are both less dense and at a lower temperature than the rest of the nebular structure and too faint to detect at C-band. The difference between the structures at the two frequencies is illustrated further in Fig. 4, which shows the C-band grey-scale image overlaid with the L-band contours. In order to make this figure, the C-band map was convolved down from a resolution of mas 2 to the same resolution as the L-band map and transformed to the same spatial grid. From an inspection of the nebular limbs visible in Fig. 4, it can be seen that, while in the C-band map the areas of brightest emission occur in the outer regions of the north-east and south-west limbs, in the L-band map it is the inner parts of these limbs which also partially comprise the bright north-western arc of emission that are brighter. The very north-west of the C-band shell structure is almost traced by the extreme north-west of the bright L-band arc of emission, a further indication that at L-band the PN is tending to optical thinness in this region. NLYSIS OF THE RDIO DT.1 Spectral index distribution The spectral index map of NGC 727 between L- and C-bands, shown in Fig., was made with the flux maps of equivalent beam sizes and gridding that are shown in Fig. 4. Clipping was implemented at the 3σ noise levels read off the two constituent flux maps. Fig. shows that the spectral index varies across the face of the nebula from.4 to 1.9. In the regions of highest flux density in the constituent maps, the error propagated to the spectral index is of the order of.2 and in the areas of low flux density, the error is.. spectral index of 2 implies that the region it was measured from is optically thick. The figure illustrates that there is a clumpy shell structure, tending to optical thickness, which encloses the less optically thick central region of the PN. The optical depth of the shell decreases in its north-west segment..2 L-band brightness temperature distribution The brightness temperature T of an object can be derived directly from a flux measurement by application of T = S νc 2 (1) 2kν 2 where S ν is the flux density received per beam area, c is the speed of light, k is oltzmann s constant and is the beam area. T is related to the electron temperature, T e, by the approximation T T e (1 e τν ) (2) where τ ν is the optical depth at frequency ν. map of nebular T distribution at L-band was made by application of equation (1) to the flux map in Fig. 3, and is shown in Fig. 6. If it is assumed that the L-band optical depth τ L 1 (Section 2.4), then this map is equivalent to one of T e distribution (by equation 2). C 23 RS, MNRS 34,

7 Radio structure and optical kinematics of NGC DECLINTION (J2) RIGHT SCENSION (J2) Figure. Spectral index map of MERLIN+VL NGC 727 data between L- and C-bands (18 6 cm). The grey-scale is from a spectral index of.6 to 2.2 and the contours are from a spectral index of.3 to 1.9 in steps of.1. DECLINTION (J2) RIGHT SCENSION (J2) Figure 6. Map of nebular brightness temperature distribution as derived from the L-band flux map. The grey-scale range is from K to the peak temperature of K; the contours start at, and increment in steps of, 2 K. However, if τ L 1, then the derived T e are lower limits to their true values. This is likely to be the case both in the region of the north-western arc of emission and also at the edges of the nebula where the gas is cooler and less dense. Data calibration errors will also introduce errors to the temperatures of the order of per cent. The peak temperature measured is ± 86 K, occurring in the north-western arc and in the region of the bright knot seen in the HST images. ecause the PN is not thought to be completely optically thick here, this is likely to be a lower limit to the actual T e, which is therefore rather higher than the canonical 1 K usually quoted for PN..3 C-band optical depth distribution The convolved-down and regridded C-band flux map (Section 4, Fig. 4) was converted to an image of C-band T by application of equation (1). two-dimensional distribution of optical depth at C-band, τ C, was then derived using this and the L-band T T e map, using equation (2). This map of τ C is shown in Fig. 7. This simple analysis again invokes the assumptions that τ L 1 and the derived T e distribution is constant with ν and constant along the line of sight. In any regions where τ L 1, Fig. 7 represents lower limits to τ C. However, the figure still serves to show the relative distribution of τ C across the face of the PN. C 23 RS, MNRS 34,

8 388 I. ains et al DECLINTION (J2) RIGHT SCENSION (J2) Figure 7. Map of τ C derived from the L- and C-band flux maps. The greyscale is from to 2.48 and the contours are at τ C =.1,.2,.4,.6,.8, 1., 1.2, 1.4 and 1.6. The position at which the peak τ C was measured is indicated by, with a value of 2. s this is close to the black edge effect regions, the second highest measurement of τ C = 1.6 ±.3 is indicated at the positions marked and C. These latter two positions also indicate where asart & Daub (1987) measured their peak τ C. The τ C distribution shows a shell structure, reminiscent of the C- band flux distribution, within which are optically thick clumps. The regions of highest τ C are found where the limbs cross the equatorial plane of the PN. lthough the peak τ C of 2. ±.4 occurs in the clump marked on the image, this measurement is possibly dubious given its proximity to the edge effect regions. The regions marked and C indicate the second-highest measured τ C of 1.6 ±.3. way from the shell, the optical depth drops away to less than 1 and the nebula is mainly optically thin. Within the shell itself, τ C is lowest in the north-west segment. Regions and C are also where both asart & Daub (1987) and ryce et al. (1997b) measured their peak τ C. The former found a τ C of 1 in both these regions while the latter calculated a peak τ C of 1.66 at and 2. at C. The calculations of ryce et al. (1997b) were based on the T e distribution derived by asart & Daub (1987). The discrepancy between the new results and these previous measurements can possibly be explained by the higher resolution that is afforded here compared to that of the observations of asart & Daub (1987), on which both of the previous measurements were based..4 Emission measure distribution The emission measure E is defined as E = n 2 e dl (3) where dl is the differential line-of-sight path length through the emitting region. To a first approximation, E is related to T e, ν and τ ν by E = τ ννghz 2.1 T e C (4) DECLINTION (J2) RIGHT SCENSION (J2) Figure 8. Map of emission measure E derived from the L-band and C-band flux maps. The grey-scale is from E = to cm 6 pc, the peak on the map (marked C); the contours start at cm 6 pc and increment in steps of cm 6 pc. asart & Daub (1987) measured their peak E in the regions marked and C. (Osterbrock 1989). map of E was constructed from the L-band T T e map and the τ C map using equation (4), and the resulting image is shown in Fig. 8. The shell structure seen in the C-band flux map is visible in the structure of E. The values of E seen in the figure are high and this is consistent with NGC 727 being a young PN. The highest values of E are found in the parts of the limbs that are situated in the equatorial region of the PN, in the same regions as the likely peak τ C was measured (Section.3). The peak value of the emission measure occurs towards the middle of the south-western limb, at the point on the figure marked C, at (1.9 ±.) 1 8 cm 6 pc. asart & Daub (1987) found their peak nebular E in the regions marked and C on Fig. 8. Their measured E at both of these points was 1 7 cm 6 pc. Using the T e distribution derived by asart & Daub (1987), ryce et al. (1997b) measured their peak E of cm 6 pc at the position marked C in the south-west limb. The discrepancies between the new results and those of the aforementioned authors can be explained as in Section.3. 6 RESULTS 3 THE OPTICL SPECTR Selected position velocity (p v) arrays of the spatially resolved emission lines observed are shown in Figs These are the highest spectral resolution, spatially resolved spectra observed from this nebula to date. Some of the data frames obtained were contaminated by light scattered from the ends of the slit. full set of [O III]499 Å lines (one from each slit position) was obtained without contamination, and so these have been chosen for display. We also display the [N II]684 Å line except in Fig. 9; at this slit position the frame containing the 684 Å line was corrupted so we show the [N II]648 Å line instead. The spatial axis on each image has been converted to offset in arcsec from the slit centre. The wavelength axis has been converted to Doppler velocity shift v corrected for the Earth s motion. The C C 23 RS, MNRS 34,

9 Radio structure and optical kinematics of NGC velocity axis was not corrected for the nebular systemic velocity V sys in order that a measurement of this quantity could be obtained from the spectra; this is done in Section 7.2. The terms redshifted and blueshifted used in the next sections are relative to this V sys. Each velocity ellipse was examined in terms of its constituent cross-sections along the slit in the spatial direction. Using the LONGSLIT package in the TWODSPEC suite of programs, the crosssections were binned together two at a time and Gaussians were fitted to the profiles obtained. The centroid of each fitted Gaussian gives the wavelength (or velocity) of the component of the profile to which it is fitted. In the succeeding sections, where the relative spectral extents of the p v arrays have been commented on, these have been measured from the maximum peak-to-peak separation given by the cross-section profiles. We note that some of the differences between the profiles of the different species at a given slit position may also be due to the differing extinction between the wavelengths. 6.1 Spectra observed with slit 1 The p v arrays shown in Fig. 9 were observed with slit 1 (Fig. 1). It can be seen that the emission observed from the portion of the Figure 9. Two-dimensional p v arrays of the emission line spectra observed with slit 1: [N II]648 Å( mjy pixel 1 ); [O III]499 Å( mjy pixel 1 ). The scale bar corresponds to the percentage of the flux ranges given (in brackets), the maxima of which are the maxima as measured from each profile. The velocity axis is heliocentric. The annotations and correspond to the slit orientation as given in Fig. 1. C 23 RS, MNRS 34, slit that covered the bright knot has both redshifted and blueshifted components. However, the bright emission with which the knot is associated is mainly redshifted. The [N II]648 Å velocity ellipse is both more spatially and more spectrally extended than that of the [O III]499 Å. Despite these differences, both ellipses show a similar morphology, albeit on differing scales. This slit position is parallel to the suspected PN minor axis, which should result in symmetric velocity ellipses if it is assumed that the emission arises in an ellipsoidal shell. oth of the ellipses shown are fairly symmetric about a central velocity although they are slightly skewed compared to their hydrogen counterparts (not shown), have a more pronounced shell-like structure and are less spectrally extended. They also show a decrease in emission at the eastern edge of the nebula, indicating that in this region possibly there is less gas or it is being obscured. There is also a decrease in the blueshifted emission towards the western edge of the nebula, almost opposite (along the line of sight) the position of the bright knot. In common with the hydrogen lines, both profiles show a greater spectral extent towards the western edge of the nebula than the eastern. 6.2 Spectra observed with slit 2 Examples of the velocity ellipses observed with slit 2 are shown in Fig. 1. This slit position is parallel to the PN major axis. If, as detailed in the literature (Section 1 and references therein), the long Figure 1. s Fig. 9 for observations with slit 2: [N II]684 Å ( mjy pixel 1 ); [O III]499 Å( mjy pixel 1 ).

10 39 I. ains et al. axis of NGC 727 is tilted towards the line of sight, observations with such a slit position should result in velocity ellipses that are tilted with respect to the spectral axis, which is indeed what is observed. Furthermore, the tilt indicates that it is the north of the PN that is inclined towards us. The [N II] emission originates in a shell-like structure. The profile is diminished in emission towards the southern, redshifted part of the nebula. The clump of bright emission that we associate with the bright knot is again mainly redshifted in all the lines apart from the [O III], where it is seen at V sys. The [O III] profile does not form an ellipse and this is possibly symptomatic of the slit clipping the edge of the region containing the [O III]; the region containing the [N II] probably extends further westwards than that of the [O III]. oth profiles show, in the same region, a feature indicative of a waist. 6.3 Spectra observed with slit 3 The [N II]684 Å and [O III]499 Å velocity ellipses observed with slit 3 are shown in Fig. 11. This slit position is aligned with the suspected PN minor axis, which should result in symmetric velocity ellipses if it is assumed that the emission arises in an ellipsoidal shell. Fig. 11 shows that in the regions where there is brighter emission, the [N II] ellipse is indeed fairly spectrally symmetric about a central velocity whereas the [O III] profile shows more deviation from this symmetry and is slightly skewed with respect to the spectral axis Figure 11. s Fig. 9 for observations with slit 3: [N II]684 Å ( 87.7 mjy pixel 1 ); [O III]499 Å( mjy pixel 1 ). The brightest feature in all of the images is apparent towards the western edge of the nebula and is blueshifted in all of the lines apart from those of hydrogen (not shown) in which it extends continuously across blueshifted and redshifted velocities. ll of the other profiles also show an increase in brightness of redshifted emission to the western edge of the nebula, although this does not extend continuously into the blueshifted velocities. Given the proximity of the position of slit 3 to that of slit 1, this redshifted component could be due to the slit position clipping the edge of the bright knot. We feel that the brightness of the knot is insufficient for this redshifted component to be due to instrumental scattering. The [N II] profile has a lesser spectral extent than its hydrogen counterparts (not shown) although its spatial extent is similar to that of the Hα. It shows a decrease in emission in the same redshifted region as Hα, although in the [N II] line it extends further, right to the eastern edge of the shell, making it almost open-ended (a similar effect is seen in the parallel position of slit 1; see Section 6.1). The [N II] ellipse has a hollow centre unlike the hydrogen lines. There is some indication of a waist-like feature. The [O III] profile has a lesser spatial extent than any of the other lines observed at this slit position although its spectral extent is comparable to that of the [N II] profile. It shows the same openended structure towards the eastern edge of the nebula. The pinched feature is more pronounced in this velocity ellipse than any others at this slit position. 6.4 Spectra observed with slit 4 Examples of the velocity ellipses observed with slit 4 are illustrated in Fig.. This slit position is aligned with the PN major axis and, in common with the parallel slit 2 (Section 6.2), the tilt of the spectra again indicates that the north of the PN is inclined towards the line of sight. ll of the ellipses show a bright, blueshifted feature towards the southern edge of the nebula. They also show a bright, redshifted feature near the centre of the nebula, although this is not as distinctive in the [N II]684 Å profile. This redshifted feature could be due to the slit clipping the edge of the bright knot, which is not apparent in the [N II] line because the brighter [N II] emission from the knot does not extend as far eastwards as that of the [O III] (Fig. 2). The slit does not sample the full spatial extent of the profiles; both show some increased brightness and smearing towards the edge of the slit marked and these are probably due to scattering from this part of the slit. This, therefore, makes it difficult to measure the spatial extent of the lines observed at this slit position. oth [N II] and [O III] profiles show a clearly defined shell structure and have a similar spectral extent. The [O III] profile also shows a possible waist-like feature towards the nebular centre, which is not so apparent in the [N II] profile. oth profiles show decreases in redshifted emission in the centre-southern portion of the slit and in the blueshifted emission towards the northern end of the slit; these are also seen in the p v arrays of the other emission lines obtained at this slit position (not shown). 6. Spectra observed with slit Examples of the velocity ellipses observed with slit are shown in Fig. 13. The bright knot is clearly visible in all of the ellipses and is again seen to be mainly redshifted. ll of the profiles are skewed and show a gap in the redshifted emission in the eastern region of the nebula. C 23 RS, MNRS 34,

11 Radio structure and optical kinematics of NGC Figure. s Fig. 9 for observations with slit 4: [N II]684 Å ( mjy pixel 1 ); [O III]499 Å( mjy pixel 1 ). The [N II] profile has the largest spatial extent of all the lines obtained at this slit position and shows a bright blueshifted feature situated westwards of the bright knot which is not apparent in the other lines. It also has a larger spectral extent than that of the [O III]. 7 NLYSIS OF THE OPTICL SPECTR 7.1 [S II]6716/6731 Å ratio plots The [S II]6716/6731 Å ratio (henceforth [S II] ratio) maps shown in Figs and 1 were made in order to assess the density variations in the nebula. These maps were derived from the data obtained from slits 1 and 3 (see Fig. 1) only, as the frames containing the [S II] lines taken at the other slit positions were all contaminated by scattering. Each constituent [S II]6731 Å velocity ellipse is shown above the ratio map it was used to construct. In order to derive the local electron density n e from the line ratios (Osterbrock 1989), an electron temperature of K was adopted, except in the region of the bright knot (observed with slit 1), for which a temperature of 17 K was assumed. These seem to be reasonable temperatures given the measurements quoted in the literature (Section 1) and the results of Section.2. For an [S II] ratio of <.44, only a lower limit to n e can be given as the plot of intensity ratio versus n e flattens out in this region (Osterbrock 1989). C 23 RS, MNRS 34, Figure 13. s Fig. 9 for observations with slit : [N II]684 Å ( mjy pixel 1 ); [O III]499 Å( mjy pixel 1 ). Observations with slit 1 show the [S II] ratio varying from.8 to.3, indicating an n e range of 1 3 to > cm 3.Inthe centre of the ratio ellipse, n e ranges from 1 3 to cm 3 and in the shell-like region it is higher and ranges from 1 4 cm 3 to > cm 3. In the region of the bright knot, with an assumed temperature of 17 K, the derived density is > cm 3. The plot derived from observations with slit 3 shows the [S II] ratio varying from.7 to.4, indicating an n e range of cm 3 to > cm 3. In the centre of the ratio ellipse, n e ranges from cm 3 to 1 4 cm 3. In the bright, shell-like region it is > cm 3. There is also faster moving, less dense material seen on the blueshifted side at velocities up to 3 km s 1. n e in this region varies from cm 3 to 1 4 cm 3.ny corresponding emission from the redshifted portion of the shell may be obscured. oth ratio ellipses show both sides of the dense shell and a real variation in n e around the ellipses. These indicate that there is a significantly less dense region in the centre, enclosed by the shell. In general, the shell region and the centre of the ratio ellipse observed with slit 1 are less dense than the corresponding regions of the ratio ellipse observed with slit 3. Slit 3 was aligned along the suspected PN minor axis; slit 1 was parallel to it but offset northwards (see Fig. 1). Hence, the difference in densities at the two slit positions suggests a possible density enhancement in the equatorial region.

12 392 I. ains et al Figure. s Fig. 9 for observations with slit 1: [S II]6731 Å ( 84. mjy pixel 1 ); [S II] ratio map scaled between.4 (white) and.8 (black). The brighter features around the velocity ellipses in Figs and 1 are not seen to have corresponding variations in n e in the ratio plots. This is also true for the bright redshifted knot that is clearly visible in Fig. ; the ratio ellipse in Fig. shows no remarkable increase in n e with respect to the rest of the shell in the corresponding region. This suggests that the brightness variations seen in the velocity ellipses are possibly due to temperature effects rather than variations in n e. We note that, while the [S II] n e measurements are internally consistent, the relatively low collisional de-excitation density for [S II] means that there can be more material present at a higher density that is not probed by the [S II] ratio. Measurements such as those of Middlemass (199) show n e of the order of 1 cm 3 in NGC 727. Therefore, it is desirable to have observational confirmation of the suggested localized higher temperatures from, for example, the forbidden line ratio involving the [O III]4363Å and 7Å lines, or preferably measurements of optical recombination lines (e.g. Liu et al. 21). In the meantime, the case for the possible increase in temperature in the region of the bright knot is argued further in Section 8.4. The decrease in emission at the eastern edge of the PN seen in Figs and 1 (see also Sections 6.1 and 6.3), which gives the ellipses an almost open-ended appearance, shows no corresponding decrease in density in the same regions in the respective plots in Figs and 1. This suggests that the shell is not open-ended but more likely obscured. However, the decrease in blueshifted Figure 1. s Fig. 9 for observations with slit 3: [S II]6731 Å ( 46.4 mjy pixel 1 ); [S II] ratio map scaled between.4 (white) and.8 (black). emission towards the western edge of the PN, situated almost opposite (along the line of sight) the bright knot, that is seen in Figs 9 and does seem to correspond to a less dense region in Fig.. This ratio ellipse is quite noisy and it is difficult to obtain a reliable measurement of n e in this particular region. The measurement obtained gives n e 1 3 cm 3, which indicates a lower density here than in the rest of the shell-like region of the ellipse. 7.2 Nebular deprojection The skewedness of the line shapes seen from observations with the slit on, or parallel to, the major axis confirms the inclination of the northern region of the nebula towards the line of sight (Figs 1 and ). This means that, in order to measure the nebular equatorial expansion velocity, V eq, it is not sufficient to simply measure the line-of-sight expansion velocity from the velocity ellipses observed with slits that sampled the apparent PN minor axis. Rather, a model that incorporates this inclination should be used to deproject the PN. The model we adopt is illustrated in Fig. 16. For simplicity, we assume the ellipsoidal shell model with the long axis inclined at 3 to the line of sight. If we consider observations made with the slit aligned with the PN major axis, i.e. slit 4, then the redshifted and blueshifted components of V eq (respectively V los(r) and V los() ) are sampled at angular distances ±d from the section of the slit C 23 RS, MNRS 34,

13 Radio structure and optical kinematics of NGC N plane of of sky major axis of PN slit V los(r) V eq θ φ x V eq V los() minor axis of PN line of sight Figure 16. simplified schematic diagram of NGC 727 illustrating the effect of nebular inclination on the measurement of the nebular equatorial expansion velocity V eq. The components of V eq along the line of sight are the redshifted V los(r) and the blueshifted V los() sampled respectively at positions P1 and P2 on the slit. Note that the PN waist (Section 8.2) is not depicted in order to simplify the drawing and analysis. that traversed the nebula centre, at positions P1 and P2 illustrated in Fig. 16. This analysis was applied to the [O III]499 Å velocity ellipse observed using slit 4, the grey-scale of which is shown in Fig.. From the geometry shown in Fig. 16, d = x sin φ, where x is the nebular semiminor axis. Slit 3 sampled the full length of the minor axis; from Figs 2 and 11, the total extent of the [O III] shell along this axis is 8 arcsec. We invoke axial symmetry so that this dimension does not change under the projection shown in Fig. 16 and so x 4 arcsec. d then becomes ±3. arcsec. We measured the velocities at positions P1 and P2 by binning together two at a time the slit cross-sections closest to these positions, and fitting Gaussians to the resulting profiles. Two prominent peaks were visible in these profiles due to emission from the receding and approaching portions of the PN shell. Note, however, that we were only interested in the fits to one of the peaks at each position: at P1, V los(r) was given by the fit to the redshifted peak, and at P2, V los() was given by the fit to the blueshifted peak. The velocity profiles obtained from the cross-sections at the calculated angular positions P1 and P2 on the slit using the [O III]499 Å data are shown in Fig. 17. Note that, in Fig. 17, a middle component is visible in addition to the main redshifted and blueshifted peaks. This is unlikely to be a real component but indicates a non- Gaussian distribution of the emission or possibly extra, low-velocity gas present along the line of sight. ssuming that the expansion velocity is linear along the PN minor axis, V eq and the systemic velocity V sys are then given by V los(r) = V eq cos φ + V sys () V los() = V eq cos φ + V sys. (6) We obtain V eq = 13 ± 1kms 1 and V sys = 7 ± 1kms 1. This value of V eq is lower than the 17. ± 1. km s 1 used by Masson (1989) and taken from therton et al. (1979) in determining the distance to NGC 727. In this paper, he also adopts the prolate, ellipsoidal shell model with the long axis inclined at 3 to the line of sight. He explains that the [O III] velocity was measured d d P1 P2 Flux (mjy) heliocentric velocity ( ) heliocentric velocity ( ) Figure 17. The profiles from the cross-sections of the [O III]499 Å emission line used to measure the nebular V sys and V eq. The dot-dashed lines indicate the Gaussian fits to the profile and the dashed line is the sum of the two fits. Velocity profile measured at position P1 on Fig. 16. The fittothe redshifted peak was used to measure V los(r). Velocity profile measured at position P2 on Fig. 16. The fit to the blueshifted peak was used to measure V los(). across the apparent nebular centre, so probably overestimates the equatorial expansion velocity, although he says that the lower limit given by the error is consistent with the isotropic neutral CO gas velocity of 16. km s 1 measured by Masson et al. (198). However, as the [O III] and CO emission occur in different regions, it is quite possible that the two species have different expansion velocities. It is therefore interesting to attempt to repeat the measurement of therton et al. (1979) and measure the line-of-sight expansion velocity across the apparent nebular centre (i.e. without deprojection) from the new data. From the [O III] data, this velocity is 19 ± 1km s 1, i.e. within the error bars of the velocity measured by therton et al. (1979). The higher value obtained here can be explained by the fact that these new data are of higher spatial and spectral resolution. Cox et al. (22) have measured a line-of-sight expansion velocity of 19. ±.1 km s 1 from their rγ observations. 7.3 Revising the distance measurement Masson (1989) measured the angular expansion rate of 4.2 ±.6 mas yr 1 from the minor axis of NGC 727 using a model of the PN with its major axis inclined by 3 to the line of sight. In order to derive his 88 ± 1 pc PN distance, he combined this expansion measurement made in the plane of the sky with the lineof-sight [O III]499 Å velocity measured by therton et al. (1979). Therefore, for the sake of direct comparison with Masson (1989), we will use the new V eq derived from our deprojected [O III]499 Å data in order to revise the distance to NGC 727. We note that the measurement of angular expansion was made from radio data that traces ionized hydrogen, and so a measurement of V eq from one of the optical hydrogen lines would provide a more internally consistent method. However, this is not feasible as the hydrogen lines are greatly affected by broadening (Section 8.3). [O III]499 Å provides a bright alternative. ecause we are C 23 RS, MNRS 34,