Long-slit optical spectroscopy of Large Magellanic Cloud radio supernova remnants

Transcription

1 Mon. Not. R. Astron. Soc. 383, (2008) doi: /j x Long-slit optical spectroscopy of Large Magellanic Cloud radio supernova remnants J. L. Payne, 1 G. L. White 1 and M. D. Filipović 2,1 1 Centre for Astronomy, James Cook University, Townsville, QLD 4811, Australia 2 University of Western Sydney, Locked Bag 1797, Penrith South, DC, NSW 1797, Australia Accepted 2007 October 18. Received 2007 September 19; in original form 2007 May 23 ABSTRACT We use long-slit spectra from shocked regions of radio supernova remnants (SNRs) in the Large Magellanic Cloud to estimate electron density and derive an average metal abundance of based on diagnostic line ratio plots. These simple diagnostics may be especially useful to determine abundances in more distant galaxies. Abundance values listed in this paper, in units of log(x/h) + 12, for nitrogen (7.3) and oxygen (8.2) agree with those reported in the literature. These estimates which we assume to be dominated by the interstellar medium with little evolutionary interference were obtained from spectral analysis of 50 per cent of known radio SNRs using the double beam spectrograph on the 2.3-m Advanced Technology Telescope at Siding Spring Observatory in Australia and the Cassegrain spectrograph on the 1.9-m Radcliffe telescope at the South African Astronomical Observatory. We also found optical evidence of shocked regions near 2 of 20 radio SNR candidates (J and J ), strengthening their identification as true remnants. Key words: supernova remnants Magellanic Clouds. 1 INTRODUCTION The physics that underlie interactions between supernova remnants (SNRs) and the surrounding interstellar medium (ISM) is best studied through the use of multiwavelength astronomy, especially in the radio, optical and X-ray domains. However, many Galactic SNRs have distance uncertainties of at least a factor of 2, and hence, luminosity and initial supernova (SN) energy calculations are even more uncertain (factors of 4 and 5.5, respectively; Seward et al. 2006). Alves (2004) reviewed distance and structure studies of the Large Magellanic Cloud (LMC), finding that the average of 14 recent measurements converge to a modulus of ± 0.02 mag ( 50.1 kpc). These results demonstrate a high level of consistency and the possibility that a consensus on the LMC distance has been reached. This degree of certainty allows the use of LMC astronomical objects, including SNRs, to determine properties of their surrounding medium. Individual detailed analysis of several remnants can be averaged to estimate the metal abundance of this galaxy. Over the past several years until 2004, but especially during 1994 October and 1995 February, L. Staveley-Smith and some of us (MDF) conducted mosaic observations of the LMC centred at GHz (bandwidth 128 MHz) using the Australia Telescope Compact Array (ATCA). 1 This survey divided the LMC into 12 regions, each containing 112 pointing centres. Each pointing centre was observed approximately 115 times (ranging from 95 to 140) for snova4@msn.com 1 These data are freely available from the ATCA online archive. 15 s. After reduction of data using the MIRIAD software suite (Sault & Killeen 2003), large-scale structure was completed for short spacing using Parkes data (Haynes et al. 1986) from the same region at GHz. The latter was obtained in 1984 September using the Parkes 64-m single-dish radio telescope with a beamwidth of 15 arcmin and an rms noise of 30 mjy beam 1. A complete radio image will be forthcoming in Hughes et al., in preparation. Radio observations are commonly used to discover and characterize SNRs. Verification may be based on a combination of radio spectral index, morphology, co-identifications in other domains (such as optical or X-ray), radio flux density and location within the galaxy under study. Our 21-cm mosaic image of the LMC (resolution 40 arcsec; sensitivity 0.37 mjy beam 1 ) was found to have co-identifications with 56 previously identified radio SNRs from Filipovic et al. (1998). We found an additional 20-candidate radio SNRs in this mosaic based on location, radio intensity, size and morphology. Images of these SNRs and candidate SNRs were used to observe all (76) of these sources using either the DBS or Radcliffe Cassegrain spectrograph. We were able to extract spectra for 25 of the known and two of the candidate SNRs, presented for the first time here. Most of the radio emission from these SNRs is non-thermal, a result of synchrotron radiation created at the shock where relativistic electrons pass through magnetic fields. Shocked material from these remnants also produce strong forbidden optical spectral lines (e.g. [S II], [N II], [O I], [O II] and [O III]). Commonly, the ratio of [S II](λ λ6731)/hα(λ6563) is used to identify these regions (Fesen, Blair & Kirshner 1985). C 2007 The Authors. Journal compilation C 2007 RAS

2 1176 J. L. Payne, G. L. White and M. D. Filipović Only a small percentage of Galactic and Magellanic Cloud optical SNRs have been well studied and SNR emission lines cover an enormous range of ionization stages. Galactic SNRs that have been well studied include the Cygnus Loop, the Crab nebula, Puppis A, Vela, SN 1006 and Cassiopeia A. N49, N63A and N132D are the most studied LMC remnants. 2 In Payne et al. (2007), we reported optical long-slit data for 11 of 16 known radio SNRs in the Small Magellanic Cloud (SMC). We confirmed a previously reported candidate radio SNR, J , based on the presence of shocked material near that source. These shocks showed no evolutionary trends based on abundance or density-sensitive line ratios. Line ratios did allow an estimate of local abundances, with the assumption that these regions are dominated by ISM. These values were averaged to give a reasonable SMC abundance estimate of Here, we present the results of optical spectroscopic observations of known and candidate LMC SNRs based on positions obtained from the above radio observations. These help further characterize remnant shocks and give abundance information about the ISM local to these SNRs. The paper is organized as follows. Section 2 details our observations, spectral extraction and reduction methods. Section 3 discusses our spectral analysis, including line fluxes, line ratios, electron densities and abundances. Section 4 briefly summarizes our results. Spectra for individual radio shocks are given in Appendix A. 2 OBSERVATIONS AND DATA REDUCTION In 2005 August, we obtained long-slit spectra of previously verified radio SNRs and SNR candidates found in the SMC (Payne et al. 2007) and LMC using the double-beam spectrograph (DBS) and 2.3-m Advanced Technology Telescope (ATT) at the Siding Springs Observatory in Australia. The visible waveband ( Å) of this general purpose spectrograph is split by a dichroic at around 6000 Å and feeds two similar spectrographs, with red and blue optimized detectors. For the present study we report values from the red side using the 316R (316 limes mm 1 ) grating. Full width half-maximum (FWHM) measurements of arc lines verify an approximate spectral resolution of 5 Å. We used a slit size 2.5 arcsec 4 arcmin with a 0.96 arcsec pixel 1 spatial resolution. In general, exposure times for each spectrum was limited to 500 s. Positional accuracy is estimated to be better than 12 arcsec in either right ascension (RA) or declination (Dec.). Additional observations of LMC SNR targets were conducted 2006 January, using the 1.9-m telescope and Cassegrain spectrograph at the South African Astronomical Observatory (SAAO). Spectra were obtained using grating number 7 (300 limes mm 1 ) between 4000 and 7000 Å; measurement (FWHM) of arc lines verify a spectral resolution of 5 Å. For these, the slit size was 1.5 arcsec 1.5 arcmin with a spatial resolution of 0.74 arcsec pixel 1. Exposure times were limited to 800 s with a positional accuracy of 10 arcsec. Using the list of 76 target objects created from the GHz radio image, observations were obtained at position angles 3 0 (DBS) or 270 (SAAO). At the time of the observations, we had no access to highly resolved narrow-band [S II], [O III] orhα images of the LMC to better direct placement, so the centre of the slit was 2 See Fesen & Hurford (1996) for a complete discussion and reference lists of well-studied SNRs. 3 Defined as the slit angle east of north. placed near the radio centre coordinate with the expectation that it would cut across the radio shocks of each SNR. In Figs 1 5 we show GHz radio images overlaid by contours of SNRs and candidate SNRs for which spectra were extracted (marked with red slits). Data reduction included bias subtraction and flat-field correction followed by cosmic ray hit cleaning using the IRAF software package (and FIGARO s task bclean ). Extraction (task extractor ), including background sky subtraction, of shocked regions were based on the visual presence of strong [S II] emission lines and location. We created one-dimensional spectra, wavelength calibrated using standard lines from the NeAr/FeAr arcs (DBS) and the CuAr arc (SAAO). Using this visual method, we were able to find regions with [S II]/Hα ratios 0.2; we only include those with (rounded) ratios 0.4. We observed both spectrophotometric standards LTT 7379 and EG 21 each evening, applying these separately to individual spectra. We note that observing conditions were not photometric and there was significant variation in flux densities, up to 50 per cent, between nights. Therefore, we do not report our flux densities as representing absolute photometry. A more important consideration, especially for deriving accurate line ratios, is the comparison of sensitivity versus wavelength. To do this, we examined the sensitivity function of both photometric standards observed during each session. For our working range from 6200 to 6850 Å, we find that sensitivities agree within 1.5 per cent. We also note that for ratios that are close in wavelength, issues of flux calibration are less important. 3 SPECTRAL ANALYSIS We used IRAF s task splot to view and analyse our spectra. In Table 1, we present the result of extractions using the criterion of a [S II]/Hα ratio 0.4. Only fluxes from spectral lines visually distinct from the baseline rms were selected for inclusion. For each extracted shocked region, this table contains the most recent radio name of the remnant, its previous (B1950) name common in the literature, its observed central position, extent and the relative flux density (using LTT 7379) of emission lines including 90 per cent confidence intervals. Listed emission lines include [O I] (λ6300 and λ6364), [N II] (λ6550 and λ6585), [S II] (λ6716 and λ6731) and Hα (λ6563). All values are shown at their rest wavelengths. As can be seen in Figs 1 5, many SNRs have more than one shock region extracted, hence explaining multiple entries in all of the tables presented here. We make no correction for wavelength dependent extinction since we are only interested in line ratios over a relatively small wavelength range ( Å). Extinction curves (Gordon et al. 2003) over that range suggest LMC line ratio corrections would at most (e.g. [O I]/Hα) be 3 per cent and generally much less. The 90 per cent confidence errors reported in Table 1 are based on both line measurement and sensitivity function; we do not account for absolute photometric errors. Line measurement errors were calculated using Monte Carlo simulation techniques found in the task splot for a sample number of 100 and measured rms sensitivity. This 1σ value was multiplied by 1.64 and combined with the sensitivity error to estimate a 90 per cent confidence interval for each flux density. We found that our rms baseline sensitivity measured less than erg cm 2 s 1 in all cases. As pointed out by Haffner, Reynolds & Tufte (1999), areas of faint Hα background emission (brightness 2R) can have [S II]/Hα ratios which approach 0.4 or higher (see their fig. 5 where R = erg cm 2 s 1 sr 1 ) which could be confused with shocked regions. Of course, this would not have been a problem if we were

3 LMC radio SNR optical spectroscopy 1177 Figure GHz radio image and contours of LMC SNRs with slit positions marked. (a) SNR J : contours are 5 and 6 mjy beam 1. (b) SNR J : contours are 2.5, 5 and 10 mjy beam 1. (c) SNR J : contours are 4, 8, 16 and 32 mjy beam 1. (d) SNR J : contours are 2.5, 5 and 10 mjy beam 1. (e) SNR J : contours are 2.5, 5 and 10 mjy beam 1. (f) SNR J : contours are 5, 10, 20, 40 and 80 mjy beam 1. able to place our slit across SNR filaments, properly isolating these regions. We do note, however, that our lowest Hα brightness value is more than 20 times the above value. 4 In addition, inspection of 4 SNR J has a Hα brightness of erg cm 2 s 1 sr 1. Figs 1 5 suggests our selected slit regions are within or near regions of radio emission associated with known or suspected SNRs, so that we still feel confident that we have extracted SNR shocked regions. In some cases, the presence of strong [O I] lines also support this assertion. We found two radio candidate SNRs: J and J ; out of 20, that have shocked regions with

4 1178 J. L. Payne, G. L. White and M. D. Filipović Figure GHz radio images and contours of SNRs in the LMC with slit positions marked. (a) SNR J : contours are 1.5, 3 and 6 mjy beam 1. (b) SNR J : contours are 4, 8, 16, 32 and 64 mjy beam 1. (c) SNR J : contours are 6, 12 and 24 mjy beam 1. (d) SNR J : contours are 7, 10 and 14 mjy beam 1. (e) SNR J : contours are 3.5, 5, 7 and 9 mjy beam 1. (f) SNR J : contours are 20, 40, 80, 160, 320 and 640 mjy beam 1. [S II]/Hα ratios 0.4. As previously mentioned, these radio candidates were initially selected from the GHz image based on their location, size and morphology. Each of these two candidates with shocked regions have a flux density of 70 mjy at GHz. A search of Molonglo Observatory Synthesis Telescope 843-MHz radio images reveal that both are present. However, since these frequencies are adjacent and no other flux densities are currently available, we cannot estimate an accurate spectral index for either candidate. Candidate J is near H II region DEM L 248, and has radio, infrared and optical emission. As Fig. 5(a) suggests, its radio morphology is consistent with that of an SNR. Candidate

5 LMC radio SNR optical spectroscopy 1179 Figure GHz radio images and contours of LMC SNRs with slit positions marked. (a) SNR J : contours are 10, 20, 40 and 80 mjy beam 1. (b) SNR J : contours are 20, 40, 80, 160 and 320 mjy beam 1. (c) SNR J : contours are 4, 6, 8 and 10 mjy beam 1. (d) SNR J : contours are 2, 4 and 8 mjy beam 1. (e) SNR J : contours are 3 and 6 mjy beam 1. (f) SNR J : contours are 1.2, 2.4, 4.8 and 9.6 mjy beam 1. J , shown in Fig. 5(b), is 0.8 arcmin from X-ray source 0344 (Haberl & Pietsch 1999). The latter source may or not be associated, since it is classed with hard X-ray emission (see Haberl & Pietsch 1999 for details) and may represent a background object (e.g. AGN). Our observations did not find discrete shocked regions in 31 of the previously reported radio SNRs identified in the GHz image of the LMC. This is not to suggest they are not present; detection of shocked regions is dependent on instrument sensitivity, positioning of the optical slit over a shocked region, dust obscuration

6 1180 J. L. Payne, G. L. White and M. D. Filipović Figure GHz radio images and contours of SNRs in the LMC with slit positions marked. (a) SNR J : contours are 5, 9 and 13 mjy beam 1. (b) SNR J : contours are 3, 6 and 10 mjy beam 1. (c) SNR J : contours are 3, 6, 12, 24, 48 and 96 mjy beam 1. (d) SNR J : contours are 3, 6 and 9 mjy beam 1. (e) SNR J : contours are 12, 16, 24 and 48 mjy beam 1. (f) SNR J : contours are 20, 30 and 40 mjy beam 1. and atmospheric conditions. It is also significant that we did not have any deep Hα maps available that would have allowed us to align the slit on SNR filaments. New LMC Hα maps (Reid & Parker 2006a,b) as well as Hα, [SII] and [O III] maps from the Magellanic Cloud Emission Line Survey (MCELS; Smith et al. 2005) will greatly improve this situation for further studies. In Table 2, we show for each extracted spectrum, estimated radio diameter (1.377 GHz), selected line ratios ([O I]/Hα, [NII]/Hα, [S II]/Hα, λ6731/hα and λ6716/λ6731) and derived electron density. Individual 90 per cent confidence line ratio errors have been calculated using standard propagation of error techniques; these can also be gauged from inspection of Figs 6 and 7. In many cases,

7 LMC radio SNR optical spectroscopy 1181 Figure GHz radio images and contours of LMC SNRs with slit positions marked. (a) Candidate SNR J : contours are 8, 10, 12 and 14 mjy beam 1. (b) Candidate SNR J : contours are 2, 4 and 6 mjy beam 1. (c) SNR J : contours are 1.2, 2 and 4 mjy beam 1. the line ratio errors are less than 0.1 and hence no error entry is given. In the case of line ratio, it is useful to consider fractional errors since the ratio itself may be small (i.e. 0.1). We find our average 90 per cent fractional errors vary from 0.08 (σ = 0.06) in the case of [S II]/Hα to 0.16 (σ = 0.13) for [N II]/Hα. 3.1 Radio diameter SNR radio diameters were determined using kpvslice from the KARMA software package, paying attention to intensity decreases of 50 per cent near the edges of each SNR. This is an imperfect process since SNRs are often just shells; they are not perfect spheres or Gaussian sources in most cases. The pixel size of the GHz image is arcsec 2. Based on this, we estimate measurement diameter errors of ±3 pc. 3.2 Electron density Electron density and temperature are useful to determine nebulae mass and abundance. Comparison of the intensity of two lines of the same ion, emitted by different levels with similar excitation energies is dependent only on the ratio of collision strengths. This allows an estimate of the average electron density to be made. We make use of these line ratios to test if SNR density is more related to its evolutionary stage or to its swept-up interstellar mass. We estimate the electron density for each of our shocks based on [S II] ratios (λ6716/λ6731) using the Space Telescope Science Data Analysis System task nebular.temden, assuming a temperature of K. This online calculator uses a five-level atom approximation from De Robertis, Dufour & Hunt (1987). Densities for line ratios 1.4 (electron densities <100 cm 3 )or 0.5 (electron densities more than a few thousand cm 3 ) cannot be accurately measured using this method. Values reported in Table 2 cover this entire range with a median of 530 cm ISM abundance of the LMC During the evolution of SNRs, the enriched ejecta is quickly overwhelmed by ISM swept up by the expanding blast wave. The sweptup mass M sw can be expressed as M sw = 4 3 πr3 ρ 0.12n 0 R 3 pc M, (1) where ρ is the mass density of the pre-snr ISM and R is the radius of the SNR (Smith et al. 1993). The ambient density, n 0, of SNRs can vary from 10 3 to 10 3 cm 3. Choosing a nominal ambient density of 1cm 3 implies that all of the remnants in our sample are relatively

8 1182 J. L. Payne, G. L. White and M. D. Filipović Table 1. Selected line fluxes and 1.64σ errors (90 per cent confidence) of LMC SNRs from spectroscopic observations. Errors are based on line measurements and sensitivity function for each night, not on absolute photometry. DBS Object Other Shock location Emission line flux ( erg cm 2 s 1 )} name name RA Dec. Dec. extent 6300 Å 6364 Å 6550 Å 6563 Å 6585 Å 6716 Å 6731 Å (J2000) (B1950) (J2000) (J2000) (arcsec pc 1 ) [OI] [OI] [NII] Hα [N II] [SII] [SII] Confirmed radio SNRs J B / ± ± ± ± ± ± ± 0.2 J B / ± ± ± ± ± ± ± 0.2 J B / ± ± ± ± ± ± ± 0.2 J B / ± ± ± ± ± ± 0.2 J B / ± ± ± ± ± ± ± 0.2 J B / ± ± ± ± ± ± ± 0.3 J B / ± ± ± ± 0.2 J B / ± ± ± ± 0.2 J B / ± ± ± ± ± 0.2 J B / ± ± ± ± ± 0.2 J B / ± ± ± ± 0.2 J B / ± ± ± ± ± 0.2 J B / ± ± ± ± ± ± 0.2 J B / ± ± ± ± ± 0.2 J B / ± ± ± ± ± 0.2 J B / ± ± ± ± 0.2 J B / ± ± ± ± ± 0.2 J B / ± ± ± ± ± ± ± 0.2 J B / ± ± ± ± ± ± ± 0.2 J B / ± ± ± ± ± 0.2 J B / ± ± ± ± ± ± ± 0.2 J B / ± ± ± ± ± ± ± 0.7 J B / ± ± ± ± ± ± ± 1.6 J B / ± ± ± ± ± ± ± 0.2 J B / ± ± ± ± ± 0.2 J B / ± ± ± ± ± 0.2 J B / ± ± ± ± ± 0.2 J B / ± ± ± ± 0.2 J B / ± ± ± ± 0.2 J B / ± ± ± ± ± ± ± 0.2 J B / ± ± ± ± ± 0.2 J B / ± ± ± ± ± ± ± 0.4 J B / ± ± ± ± 0.2 J B / ± ± ± ± ± 0.2 J B / ± ± ± ± ± 0.2

9 LMC radio SNR optical spectroscopy 1183 Table 1 continued DBS Object Other Shock location Emission line flux ( erg cm 2 s 1 )} name name RA Dec. Dec. extent 6300 Å 6364 Å 6550 Å 6563 Å 6585 Å 6716 Å 6731 Å (J2000) (B1950) (J2000) (J2000) (arcsec pc 1 ) [OI] [OI] [NII] Hα [N II] [SII] [SII] Candidate radio SNRs J / ± ± ± ± ± ± ± 0.2 J / ± ± ± ± 0.2 J / ± ± ± ± ± 0.2 SAAO Object Other Shock location Emission line flux ( erg cm 2 s 1 ) Name Name Dec. RA RA extent 6300 Å 6364 Å 6550 Å 6563 Å 6585 Å 6716 Å 6731 Å (J2000) (B1950) (J2000) (J2000) (arcsec pc 1 ) [OI] [OI] [NII] Hα [N II] [SII] [SII] Confirmed radio SNRs J B / ± ± ± ± ± ± 0.4 J B / ± ± ± ± ± 0.3 J B / ± ± ± ± ± ± ± 0.3 mature and have much more interstellar material than ejecta from the original SN explosion (>50 M based on our smallest radius, 7.5 pc). Assuming an ambient density of a few hundred, this sweptup mass is greatly increased. Thus, if one can use line ratios to find gas abundances, the result may reveal much about the abundances of the ISM into which an SNR is expanding. Precise abundance information would require high-resolution deep spectra and need to be based on several faint lines (see discussion on pages in Osterbrock & Ferland 2006 and Russell & Dopita 1990). In an effort to find a simple technique in galaxies too distant to obtain numerous faint complex emission lines, Dopita et al. (1984) pioneered the use of grids based on shock models to determine abundance. This technique was later used by Smith et al. (1993) to estimate abundances in M33. Simple diagnostic line ratio plot estimates may be adversely affected by low shock velocities (Blair & Kirshner 1985). Smith et al. (1993) notes that the [N II] line intensity falls off for shock velocities less than 80 km s 1 and the λ6731 line varies little down to 50 km s 1. Therefore, the use of fig. 8 in Dopita et al. (1984) (which compares the ratios [N II]/Hα and λ6731/hα) should yield abundances which are not susceptible to variations in shock conditions corresponding to a limiting shock velocity of 80 km s 1. This has to be assumed in almost all cases. In Table 3 we list estimated LMC SNR abundances based on Dopita et al. (1984). We adopt the generalized metal abundance formula found there. 5 Smith et al. (1993) notes that while individual abundance estimates may not be very precise using this method, mean abundances can be accurately derived. 6 To account for errors, we determine each abundance based on 90 per cent confidence intervals of both [N II]/Hα and λ6731/hα. 7 Table 3 lists only errors ±0.1 or higher. Although individual errors as high as ±0.4 are present, most are much less. We also determined a mean and standard deviation for each set of values. This allowed us to estimate LMC abundances as a galaxy and gauge abundance variations between individual remnants. Russell & Dopita (1990) made abundance estimates using H II regions and SNRs in the Magellanic Clouds based on the analysis of spectra with available models. Table 12 in that paper give results for SNRs, including oxygen and nitrogen in units of log(x/h) Their oxygen abundance value of 8.25 ± 0.25 does compare reasonably to our value of 8.2 ± 0.3. Our estimate of nitrogen abundance is slightly lower (7.3 ± 0.2 versus 7.45 ± 0.19) but still compares well. Using their abundance values for oxygen, nitrogen and sulphur, we find that their value of log A ( 3.76) is near ours ( 3.9 ± 0.2). 3.4 SNR evolution Systematic variations in the abundance-sensitive [N II]/Hα correlated with the density-sensitive [S II] λ6716/λ6731 ratio in Galactic SNRs have been reported by Daltabuit, D Odorico & Sabbadin (1976). If correlated to each other, this implied they are also correlated with SNR diameters and hence could be used to map evolutionary trends. These relationships have been questioned by Smith et al. (1993) as originating from selection effects instead. In fact, 5 Defined as loga = [ log(n H ) + log( O H ) + log( S H )]. 6 The use of fig. 8 (Dopita et al. 1984) requires an assumed ratio of oxygen to sulphur of Actually, the graph in Dopita et al. (1984) uses the log values of each ratio.

10 1184 J. L. Payne, G. L. White and M. D. Filipović Table 2. Selected line ratios, 90 per cent confidence line ratio intervals (only errors ± 0.1 are shown) and statistics of radio SNR shocks from spectroscopic observations. To calculate electron densities, we used the Space Telescope Science Data Analysis System task nebular.temden, based on the five-level atom approximation of De Robertis et al. (1987) assuming T = K. This method is only valid for [S II] ratios > 0.5 and < 1.4. DBS Object RA Dec GHz [O I]/Hα [N II]/Hα [S II]/Hα 6731/Hα [S II] ratio Electron Density name (J2000) (J2000) Diameter (arcsec pc 1 ) (λ λ6364)/λ6563 (λ λ6585)/λ6563 (λ λ6731)/λ6563 λ6731/λ6563 λ6716/λ6731 (cm 3 ) Confirmed radio SNRs J / ± 0.2 <100 J / ± J / ± 0.1 <100 J / ± J / J / J / ± ± ± ± J / ± ± ± ± J / ± ± ± ± J / ± J / ± ± ± ± J / ± ± ± ± J /73 < J / ± ± ± J / ± ± ± 0.3 <100 J / ± ± ± ± 0.3 <100 J / ± ± ± 0.3 <100 J / J / J / ± ± ± ± J / ± J / J / J / ± ± ± ± ± 0.2 <100 J / ± ± ± ± ± 0.4 <100 J / ± ± ± ± J / ± 0.1 <100 J / ± ± J / ± J / ± ± ± 0.1 <100 J / ± 0.3 <100 J / J / ± ± ± ± 0.5 <100 J / ± 0.2 <100 J / ±

11 LMC radio SNR optical spectroscopy 1185 Table 2 continued DBS Object RA Dec GHz [O I]/Hα [N II]/Hα [S II]/Hα 6731/Hα [S II] ratio Electron Density name (J2000) (J2000) Diameter (arcsec pc 1 ) (λ λ6364)/λ6563 (λ λ6585)/λ6563 (λ λ6731)/λ6563 λ6731/λ6563 λ6716/λ6731 (cm 3 ) Candidate radio SNRs J / ± J / ± ± 0.1 <100 J / ± 0.1 <100 SAAO Object Dec. RA GHz [O I]/Hα [N II]/Hα [S II]/Hα 6731/Hα [S II] ratio Electron density name (J2000) (J2000) Diameter (arcsec pc 1 ) (λ λ6364)/λ6563 (λ λ6585)/λ6563 (λ λ6731)/λ6563 λ6731/λ6563 λ6716/λ6731 (cm 3 ) Confirmed radio SNRs J / ± ± ± ± ± J / ± ± ± ± J / ± ± ± ± ± 0.3 <100 [N II] / H / 6731 Figure 6. Observed LMC λ6716/λ6731 versus [N II]/Hα with 90 per cent confidence limits shown. The dotted line at λ6716/λ6731 = 1.4 and 0.5 denote low- and high-density limits, respectively. Even if one makes use of the relatively large error bars shown, it is not possible to draw a straight line through all of the points. This is similar to the case for M33 presented in Smith et al. (1993). [N II] / H 6716/ Diameter (pc) Diameter (pc) [S II] / H [O I] / H Diameter (pc) Diameter (pc) Figure 7. Comparisons of abundance and density sensitive line ratios with SNR and SNR candidate diameter; 90 per cent confidence limits of ratios are also shown. λ6716/λ6731 density ratios 1.4 and 0.5 have been excluded. We find that no obvious evolutionary trend evident in any of these graphs. the discussion of swept-up mass in the last section argues against such a relationship. Fig. 6 shows a graph of λ6716/λ6731 versus [N II]/Hα for our LMC SNR sample. No apparent trend can be seen. We take this further in Fig. 7 where we show abundance and density-sensitive ratios of SNRs in our sample as a function of SNR diameter. We see no significant linear correlation in any of these four plots. Given our results, we agree with Smith et al. (1993) that abundance and density ratios tell little about SNR evolution. Instead, we believe these ratios are related to the surrounding ISM. 4 SUMMARY The diverse nature of individual SNRs is most likely the result of their unique environment rather than evolutionary stage. Here, we

12 1186 J. L. Payne, G. L. White and M. D. Filipović Table 3. LMC SNR abundances and 90 per cent confidence intervals ( ±0.1) for radio shocks based on the grid shown in fig. 8 of Dopita et al. (1984). DBS Object RA Dec. N abundance O abundance Metal abundance name (J2000) (J2000) log(n/h)+12 log(o/h)+12 log A Confirmed radio SNRs J ± ± ± 0.2 J ± ± ± 0.2 J J ± ± 0.1 J ± ± 0.1 J ± ± 0.1 J ± ± 0.1 J ± ± ± 0.2 J ± ± ± 0.1 J ± ± ± 0.3 J ± ± 0.1 J ± ± 0.1 J ± ± 0.1 J ± ± ± 0.2 J J ± ± ± 0.2 J ± ± 0.1 J ± ± 0.1 J J ± ± 0.1 J J J J ± ± 0.1 J ± ± 0.1 J ± ± 0.1 J J ± ± 0.1 J ± ± 0.1 J J ± ± 0.1 J J ± ± 0.1 J ± ± ± 0.2 J ± ± ± 0.2 Candidate radio SNRs J ± ± ± 0.2 J ± ± 0.1 J ± ± ± 0.2 SAAO Object RA Dec. N abundance O abundance Metal abundance name (J2000) (J2000) log(n/h) + 12 log(o/h) + 12 log A Confirmed radio SNRs J ± ± 0.1 J ± ± ± 0.2 J ± ± ± 0.2 Average abundances SNR sample N abundance O abundance Metal abundance log(n/h) + 12 log(o/h) + 12 log A SMC SNRs based on Payne et al. (2007) Standard deviation Galactic SNRs based on Fesen et al. (1985) Standard deviation LMC SNRs based on Russell & Dopita (1990) Standard deviation LMC SNRs (this work) Standard deviation

13 have estimated an average abundance for the LMC using simple optical line ratios from a sample of mature radio remnants. To do this, we have assumed an average shock velocity 80 km s 1. Despite significant limitations, averaged values are similar to those previously reported in the literature. This helps to verify a relatively simple method for abundance determination in distant galaxies where deeper high-resolution spectral lines cannot be detected. The importance of multiwavelength techniques are also emphasized here since many remnants are initially detected in radio and later confirmed using optical spectroscopy. Future LMC abundance estimates could be improved by the availability of Hα images for direct placement of the spectrograph slit over SNR filaments. A wider optical wavelength range with improved resolution and the use of photometric observations would also allow improved results. Although we found shocked regions near two radio candidate SNRs, many more SNRs should be confirmed by accurate spectrograph slit placement. ACKNOWLEDGMENTS We used the KARMA software package developed by the ATNF. The Australia Telescope Compact Array is part of the Australia Telescope which is founded by the Commonwealth of Australia for operation as a National Facility managed by the CSIRO. We thank the Australian National University Research School of Astronomy and Astrophysics for granting us time on the 2.3-m DBS at Siding Spring Observatory. This especially goes to their kind staff. We also were granted observation time at the South African Astronomical Observatory (SAAO) and wish to thank them for their kind help and accommodations. Travel to the SAAO was funded by Australian Government AINSTO AMNRF grant number 05/06-O- 11. IRAF is distributed by the National Optical Astronomy Observatories, which are operated by the Association of Universities for Research in Astronomy, Inc., under cooperative agreement with the National Science Foundation. LMC radio SNR optical spectroscopy 1187 Finally, we thank our referee for careful attention to detail in this paper. Those suggestions have been most helpful. REFERENCES Alves D. R., 2004, New Astron. Rev., 48, 659 Blair W. P., Kirshner R. P., 1985, ApJ, 289, 582 Daltabuit E., D Odorico S., Sabbadin F., 1976, A&A, 52, 93 De Robertis M. M., Dufour R., Hunt R., 1987, J. R. Astron. Soc. Canada, 81, 195 Dopita M. A., Binette L., D Odorico S., Benvenuti P., 1984, ApJ, 276, 653 Fesen R. A., Hurford A. P., 1996, ApJS, 106, 563 Fesen R. A., Blair W. P., Kirshner R. P., 1985, ApJ, 292, 29 Filipović M. D., Haynes R. F., White G. L., Jones P. A., 1998, A&AS, 130, 421 Griffith M. R., Wright A. E., 1993, AJ, 105, 1666 Gordon K. D., Clayton G. C., Misselt K. A., Landolt A. U., Wolff M. J., 2003, ApJ, 594, 279 Haffner L. M., Reynolds R. J., Tufte S. L., 1999, ApJ, 523, 223 Haberl F., Pietsch W., 1999, A&AS, 139, 277 Haynes R. F., Klein U., Wielebinski R., Murray J. D., 1986, A&A, 159, 22 Kurtz M. J., Mink D. J., 1998, PASP, 110, 934 Osterbrock D. E., Ferland G. J., 2006, Astrophysics of Gaseous Nebulae and Active Galactic Nuclei, 2nd edn. University Science Books, Sausalito Payne J. L., White G. L., Filipović M. D., Pannuti T. G., 2007, MNRAS, 376, 1793 Reid W. A., Parker Q. A., 2006a, MNRAS, 365, 401 Reid W. A., Parker Q. A., 2006b, MNRAS, 373, 521 Russell S. C., Dopita M. A., 1990, ApJS, 74, 93 Sault B., Killeen N., 2003, MIRIAD Users Guide, ATNF Seward F. D., Williams R. M., Chu Y. -H., Dickel J. R., Smith R. C., Points S. D., 2006, ApJ, 640, 327 Smith R. C., Kirshner R. P., Blair W. P., Long K. S., Winkler P. F., 1993, ApJ, 407, 564 Smith R. C., Points S. D., Chu Y.-H., Winkler P. F., Aguilera C., Leiton R., MCELS team, 2005, A&AS, 207, APPENDIX A: INDIVIDUAL SPECTRA In this supplemental section (Figs A1 A7), we show spectra from each extracted region between 6200 and 6850 Å.

14 1188 J. L. Payne, G. L. White and M. D. Filipović (a) (c) (e) Figure A1. Spectra of individual shocks. (a) SNR J (b) SNR J (north). (c) SNR J (south). (d) SNR J (north). (e) SNR J (south). (f) SNR J (north). (b) (d) (f)

15 LMC radio SNR optical spectroscopy 1189 (a) (c) (e) Figure A2. Spectra of individual shocks. (a) SNR J (south). (b) SNR J (north). (c) SNR J (south). (d) SNR J (north). (e) SNR J (south). (f) SNR J (b) (d) (f)

16 1190 J. L. Payne, G. L. White and M. D. Filipović (a) (c) (e) Figure A3. Spectra of individual shocks. (a) SNR J (b) SNR J (c) SNR J (d) SNR J (north). (e) SNR J (south). (f) SNR J (north). (b) (d) (f)

17 LMC radio SNR optical spectroscopy 1191 (a) (c) (e) Figure A4. Spectra of individual shocks. (a) SNR J (south). (b) SNR J (north). (c) SNR J (south). (d) SNR J (north). (e) SNR J (south). (f) SNR J (b) (d) (f)

18 1192 J. L. Payne, G. L. White and M. D. Filipović (a) (c) (e) Figure A5. Spectra of individual shocks. (a) SNR J (b) SNR J (c) SNR J (north). (d) SNR J (south). (e) SNR J (f) SNR J (b) (d) (f)

19 LMC radio SNR optical spectroscopy 1193 (a) (c) (e) Figure A6. Spectra of individual shocks. (a) SNR J (north). (b) SNR J (south). (c) SNR J (d) SNR J (e) SNR J (f) Candidate SNR J (b) (d) (f)

20 1194 J. L. Payne, G. L. White and M. D. Filipović (a) (c) (e) Figure A7. Spectra of individual shocks. (a) Candidate SNR J (north). (b) Candidate SNR J (south). (c) SNR J (west). (d) SNR J (north-east). (e) SNR J (south-east). This paper has been typeset from a TEX/LATEX file prepared by the author. (b) (d)