A MULTI-WAVELENGTH LOOK AT THE YOUNG PLERIONIC SUPERNOVA REMNANT 0540-69.3

Three observations of 0540, with a total integration time of 118 ks, were made with the Chandra ACIS on 2006 February 15, 2006 February 16, and 2006 February 18. This data set was used to analyze the western and southern portions of the shell in Park et al. (2010). These observations were archived as ObsID 5549, 7270, and 7271, respectively. In each of these observations, the entire SNR was imaged on a 1/4 −subarray of the ACIS-S3 back-illuminated chip, which is particularly sensitive to soft (<2 keV) X-rays. The use of a 1/4 −subarray was done in order to mitigate the effects of pileup from the bright central pulsar by reducing the frame time from the standard ACIS value of 3.2 s to approximately 1.0 s (Park et al. 2010). These observations achieved an unbinned spatial resolution ≲ 1'' and a spectral resolution (ΔE/E) of ∼5–20 over the main energy band for the ACIS-S detectors (0.3–8.0 keV). The earlier 28 ks ACIS observation of 0540 (Hwang et al. 2001; Petre et al. 2007) was not included in our survey. This observation was taken in the ACIS FAINT mode (instead of the VFAINT mode used in the three 2006 observations) in order to study the central PWN. However, these data did not provide any noticeable improvement in photon statistics in the shell, so we did not include them in our study.

The level 1 event files were reprocessed using the chandra_repro tool in CIAO 4.4 (Fruscione et al. 2006), with calibration information from CALDB 4.5.1. This tool automatically creates new level 2 event files, accounting for charge transfer inefficiency, good time intervals, and time-dependent gain variations. These level 2 event files were used for all analysis. A false-color image, created by splitting the X-ray emission into soft (0.3–0.8 keV), medium (0.8–2.0 keV), and hard (>2.0 keV) bands, is shown in Figure 1. This image was produced using the CIAO tool reproject_events to reproject the three Chandra observations to a common plane and co-add them. The energy bands are chosen to highlight spectral windows dominated by specific emission lines or processes evident in the spectrum of the entire remnant, also shown in Figure 1. We have chosen the soft band to cover the lowest-energy X-ray lines, principally H and He-like oxygen (O vii at 574 eV and the O viii doublet at 653 and 654 eV), although lines in the iron L-shell such as Fe xvii (725, 727, 739 eV) and Fe xviii (771 eV) are prominent in this band at low temperatures (kT ∼ 0.5–1.0 keV). Fe L shell lines dominate in the medium band, but highly ionized states of silicon (Si xiii at 1865 eV; prominent above kT = 1.0 keV) and magnesium (Mg xi at 1352 eV), are present. Neon (Ne ix at 922 eV and Ne x at 1022 eV) also appears, although Ne ix is only prominent at low temperatures. It is apparent from even a cursory glance at Figure 1 that these lines are minor components in the spectrum, either due to a lack of emitting ejecta or to a low ionization state, and that continuum emission is the dominant component in the spectrum. We expect that if a nonthermal continuum is present, it will be most apparent in the hard band.

Zoom In Zoom Out Reset image size

The X-ray data are complicated somewhat by the fact that the pulsar and the central ∼3'' of the PWN are heavily affected by pileup, despite the use of a subarray to reduce the ACIS frame time. A related effect is the ACIS readout streak, which can be seen prominently in Figure 1. The streak is caused by photons from a bright source being collected while the CCD is reading out, which smears out the source along its chip rows. To prevent the displaced photons from contaminating spectra in the interior, we use the CIAO tool acisreadcorr to block out the contaminated rows.

We use radio-continuum observations from Australian Telescope Compact Array (ATCA) project C014. Details of the observations can be found in Table 1. The radio galaxy PKS 1934-638 was used as a primary flux calibration source for all observations, with the radio sources PKS 0454-810 and PKS 0530-727 used for phase calibration. The miriad³ (Sault et al. 1995) and karma (Gooch 1995) software packages were used for reduction and analysis. Images were formed using miriad multi-frequency synthesis (Sault & Wieringa 1994) and natural weighting. They were deconvolved with primary beam correction applied. The same procedure was used for both U and Q Stokes parameter maps. Figure 2 shows surface brightness maps for the ATCA data.

Zoom In Zoom Out Reset image size

Optical images of 0540 were obtained from the European Southern Observatory (ESO) 30 Doradus/WFI Data Release 1.0. These images were taken with the Wide Field Imager on the 2.2 m telescope at the ESO's La Silla Observatory. For the main band used, [O iii], five exposures with a total exposure time of 1500 s were aligned, sky-subtracted, and co-added by the VOS/ADP team, providing a FWHM of 082. For the Hα band, four exposures with a total exposure time of 4800 s were combined, providing a FWHM of 073. Further details on the filters, data reduction and processing may be found at http://archive.eso.org/archive/adp/ADP/30_Doradus/. We removed bright field stars from the [O iii] and Hα images in order to clarify the faint nebular structure via point-spread function (PSF) subtraction, using standard procedures with the tasks daofind, phot, psf, and allstar in the IRAF DAOPHOT package. The PSF-subtracted [O iii] and Hα images can be seen in Figure 3.

Zoom In Zoom Out Reset image size

Infrared images of 0540 were obtained from the Spitzer Heritage Archive. These images were taken with the Spitzer Infrared Array Camera (IRAC) as part of observing program ID 3680. The observations resulted in a 26.8 s integration time in all four IRAC bandpasses (3.5 μm, 4.5 μm, 5.8 μm, and 8 μm) with a pixel size of ∼12. These observations were retrieved after undergoing standard pipeline processing and are presented here without additional reprocessing beyond minor cropping and rescaling. The shortest-wavelength IR images reveal little of interest and are mostly dominated by field stars. In 5.4 and 8 μm, however, more features become apparent, as shown in Figure 4, which shows a false-color image of 0540 from the IRAC data, with the X-ray contours overlaid.

Zoom In Zoom Out Reset image size

As shown in Figure 1, the X-ray morphology shows a patchy and incomplete shell at a radius of about 30'' from the pulsar, with the bright central PWN surrounded by a "halo" of X-ray emission out to a radius of 9''. This image reveals that the character of the emission varies greatly depending on position. The western half of the SNR is dominated by softer emission with a more obvious shell morphology. The shell is substantially brighter in the southwest than in the northeast. The eastern half consists mostly of diffuse harder emission. The hard and nonthermal "arcs" investigated by Park et al. (2010) are also apparent at the extreme eastern and western boundaries of the SNR shell. The PWN is apparent as a bright and heavily piled-up feature at the center of the shell, elongated slightly toward the NW. The elongation was initially thought (Gotthelf & Wang 2000) to be a jet, but high-resolution optical imaging (Morse et al. 2006) showed a diffuse "breakout" instead of a collimated jet in this region. Other analysis (Sandin et al. 2013) has suggested that this elongation may actually be a torus around the pulsar, similar to that seen around the Crab (Weisskopf et al. 2000), with the jet pointing to the southwest.

The visible radio structure is strongly dependent on frequency, as shown by the maps in Figure 2. The 1513 MHz (20 cm) radio map shows a nearly circular area of diffuse radio emission about 1' in diameter, surrounding the extremely bright (>100 mJy beam⁻¹) PWN. This diffuse emission shows a mostly constant intensity around 8–10 mJy beam⁻¹. The western edge of the shell appears to be somewhat flattened, although given the large beam size at this frequency (97 × 79), determining morphology on such a small scale is difficult. This edge shows a concentration of considerably brighter emission (∼60 mJy beam⁻¹) about 20'' long. The 2290 MHz (13 cm) radio map shows a distinct shell morphology in the northern portion of the shell. The eastern and northern portions of the shell (5–6 mJy beam⁻¹) are much fainter than the western portion of the shell (>15 mJy beam⁻¹), which appears much as it does in the 1513 MHz image. The 5824 MHz (6 cm) image also reveals that the western edge of the SNR appears to be composed of a number of bright (5–10 mJy beam⁻¹) knots. At the highest frequency radio band surveyed, 8895 MHz (3 cm), the shell is mostly below the background noise threshold, and only the PWN and the knots along the western edge of the shell are visible. We find possible structure of the PWN in the 5824 MHz image (Figure 2), where there appears to be elongation of the emission in the NE-SW direction.

At optical wavelengths, the field is very complicated. Data from CO surveys (Ott et al. 2008; Wong et al. 2011) indicates that 0540 is located in a large ridge of molecular clouds to the east of 30 Doradus, and the H ii region N158, containing the OB association LH 104, is located less than 2' to the south (Dunne et al. 2001). In both Hα and [O iii], shell structure is difficult to distinguish from this complex surrounding environment. However, diffuse [O iii] emission appears to roughly trace the SNR shell in the north and south, at a flux level⁴ barely above the background (0.03–0.05 ADU s⁻¹, with a background flux of 0.02–0.03 ADU s⁻¹). The circular pattern becomes lost in diffuse emission at the east and west. A detail of the halo region in Hα and [O iii] is shown in Figure 5. The bright 8'' ring around the PWN is clearly visible (∼0.24 ADU s⁻¹) in the [O iii] images, as is the fainter (0.06–0.07 ADU s⁻¹) extended emission out to about 5'' seen with Hubble Space Telescope (HST) long-slit spectroscopy by Morse et al. (2006). The latter feature is not visible in the HST WFPC2 images. West of the PWN in the shell region, the diffuse [O iii] emission noted by Mathewson et al. (1980) is visible.

Zoom In Zoom Out Reset image size

The unrelated LMC star 2MASS J05400761-6920049 is prominent in the west of the shell in IR, making fine detail in this region difficult to see. (This star has been removed via PSF subtraction in Figure 4.) Bright diffuse emission in the northeast is visible, along with two starlike objects. Neither the diffuse emission or starlike objects appear to be correlated to the SNR shell morphology in radio and X-ray. North of this, when combined with the [O iii] emission, a portion of the shell becomes apparent. A circular structure which appears suggestive of the SNR shell is also visible outside the east limb and in the south. However, Williams et al. (2008) state that this apparent shell structure in the IR is spectroscopically indistinguishable from the background and argue, based on this, that its relation to the SNR may be coincidental. However, Figure 4 shows a portion in the southwest of the shell that closely follows, and is just exterior to, the X-ray contours. This suggests that the IR structure may be a portion of the shell radiating mostly in IR, perhaps caused by interaction with a local molecular cloud. We use this possible association with caution.

For both the optical and IR data, definitively determining whether specific features are associated with the SNR is difficult. Due to these considerations, we only make use of these data in a qualitative manner, to lend further insight to the quantitative conclusions drawn from the radio and X-ray data in the following sections.

Because our radio observations cover several different frequencies, we may calculate the spectral index α, where the flux density S∝ν^α, from the radio flux measurements. We measured the flux density of the core of 0540 at four wavelengths (20 cm, 13 cm, 6 cm, and 3 cm), using the observations listed in Table 1. However, as interferometers such as ATCA lack the zero-spacing measurement responsible for the large-scale emission, we were unable to directly measure the extended emission from the SNR shell. These observations were particularly unsuited for measuring extended emission due to the long baselines used, where the shortest distance between any two antennas ranged from 77 m to 337 m. As an alternative, we measured the integrated flux density for the entire SNR from archival radio continuum observations. We estimate the 36 cm measurements from the Molonglo Synthesis Telescope mosaic image (as described in Mills et al. 1984) and the 36 cm Sydney University Molonglo Sky Survey image (Mauch et al. 2008). The 20 cm flux density was measured from a mosaic image published by Hughes et al. (2007) while the 6 cm and 3 cm flux densities were measured from mosaics published by Dickel et al. (2010).

As we lacked high resolution images with short baselines, we deduce the flux density of the remnant's shell by subtracting the value of the core from the measurement for the entire SNR. The resulting values for the entire core and shell are shown in Table 2 and then used to produce the spectral index plot in Figure 6. Errors in these measurements predominantly arose from defining the edge of the PWN. Using the flux densities in Table 2, we estimate a spectral index for the PWN of α_Core = −0.15 ± 0.03. This is consistent with, but much more refined than, the earlier estimate of the spectral index by Manchester et al. (1993) of α = −0.25  ±  0.10. Our value is also consistent with the typical radio spectral index for a PWN of α ⩾ −0.3 and with spectral index values of other known PWNe such as G326.3-1.8 (α = −0.18; Dickel et al. 2000) and IKT 16 (α = −0.19 ± 0.2; Owen et al. 2011). We also estimate for the entire SNR of α_Total = −0.59  ±  0.01, and a spectral index for the shell alone of α_Shell = −0.65 ± 0.01. Our value for the shell is consistent with the typical value (Green 2009) for synchrotron radiation from an SNR.

Zoom In Zoom Out Reset image size

The fractional polarization, P, was calculated at 6 cm, using the equation

$$\begin{equation} P = \frac{\sqrt{S_Q^2 + S_U^2}}{S_I}, \end{equation} \tag{ 1 }$$

where S_Q, S_U, and S_I are the integrated intensities for the Q, U, and I Stokes parameters. We estimate that the mean fractional polarization of the shell at P = 11% ± 3%. The resulting electric field vectors for the 5824 MHz observations are shown in Figure 7.

Zoom In Zoom Out Reset image size

Polarization position angles were taken from both 6 cm observations (5824 and 4786 MHz) and used to estimate the value of the Faraday rotation measure. The resulting rotation measures are plotted in Figure 8, with the open boxes representing negative values of rotation measure and the filled-in boxes representing positive rotation measure. The average rotation measure for the central plerion was −346 rad m⁻², while the shell averaged a slightly higher value of negative rotation, −369 rad m⁻². The rotation measure was then used to find the zero-wavelength, or intrinsic, values by de-rotating the measured position angles. The vectors in Figure 9 show the intrinsic values and directions of the magnetic field in the remnant.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Figure 10 shows the X-ray emission in grayscale, with radio contours at 1.5 GHz and 5 GHz overlaid. Two areas of coincident emission are immediately apparent: the PWN and the western edge of the shell. The apparent size of the PWN in radio and X-ray is about the same: 5'' across in radio and X-ray, although determining the exact size in radio is difficult, as the radio beam size in our observations is comparable. The equivalent radio and X-ray sizes are at odds with the tendency of most PWNe, such as the Crab (Hester 2008) to decrease in apparent size when seen at increasing energy. A possible cause of this is an unusually low magnetic field, such as that seen in 3C 58 (Green & Scheuer 1992; Slane et al. 2004), which has a magnetic field of ∼80 μG. As discussed below in Section 4.3, we see some evidence of a magnetic field this weak in 0540.

Zoom In Zoom Out Reset image size

The western edge of the shell is another area of coincidence between X-ray and radio emission. In both X-ray and radio, the shell is sharply flattened and brightened on this edge. The knot about 20'' northeast of the PWN is the most prominent brightened area in 5 GHz radio and in X-ray. However, in 1.5 GHz radio, the western edge features two teardrop-shaped knots that have no obvious counterpart in X-ray or in higher-frequency radio observations. This suggests that two different populations of electrons are responsible for the dissimilar areas of emission.

There are a few areas of emission where the radio and X-ray emission do not coincide at all. The noticeable area of radio emission in the southeast portion of the shell, visible as a knot in the 5 GHz map and as an amorphous extension of the shell in 1.5 GHz, has no obvious counterpart in X-ray. The extended area of emission in the northern area of the shell in the 1.5 GHz map also has no counterpart in 5 GHz or in the X-ray. Conversely, the synchrotron-emitting X-ray knot in the east (Park et al. 2010) has no obvious counterpart in any surveyed radio band. One possibility, in this eastern knot, is that the radio synchrotron emission is blocked by the Razin effect (Ginzburg & Syrovatskii 1965; Rybicki & Lightman 1979), where synchrotron emission below a cutoff frequency ν_R is suppressed by thermal plasma in the emitting region. The cutoff frequency is given by (Melrose 1980; Dulk 1985) ν_R = 20n_e/B, where n_e is the electron density in cm⁻³ and B is the magnetic field in Gauss. The highest radio frequency surveyed, 8895 MHz, shows no evidence of this synchrotron knot, so we can use this as a lower limit of the potential Razin cutoff frequency. The magnetic field in this knot is B = 20–140 μG (Park et al. 2010), which means that suppressing synchrotron emission below 8895 MHz would require a minimum electron density in the emitting region of ∼0.01–0.06 cm⁻³. The minimum density found in the east of 0540 by Park et al. is ∼0.17 cm⁻³, so this is not an unreasonable requirement.

The wide frequency range covered by past and present observations of 0540 makes it possible for us to examine the broadband spectrum of the PWN. This spectrum, covering the range 10⁶–10¹⁹ Hz, is shown in Figure 11. It is apparent from Figure 11 that although individual wavebands may be accurately fit with a power law, as evidenced by the radio spectral index calculation in Section 4.1, a single power law does not accurately describe the entire broadband spectrum. The break in the spectrum between the radio and IR bands is attributed (Ginzburg & Syrovatskii 1965) to synchrotron losses from higher energy electrons that have exceeded their synchrotron lifetime.

Zoom In Zoom Out Reset image size

The synchrotron lifetime for an electron radiating at a frequency ν can be approximated (Pacholczyk 1970) as $\tau _s = 6 \times 10^{11} B^{-3/2} \nu _b^{-1/2}$ where τ_s is the synchrotron lifetime in seconds, B is the PWN magnetic field in Gauss, and ν is the frequency in Hertz. At the break frequency, the synchrotron lifetime is equal to the age of the PWN, so we can determine the PWN magnetic field from a determination of the break frequency. Fitting this broadband spectrum to a broken power law (plotted in Figure 11) results in a break frequency of 6 ± 1 × 10¹³ Hz. Assuming a nebular age (Reynolds 1985) of 1100 yr, this results in a magnetic field strength of 250 ± 15 μG. This is consistent with the magnetic field estimate of Manchester et al. (1993), who estimated a magnetic field strength of 250 μG for an age of 1100 yr, and substantially lower than that of Petre et al. (2007), who estimated a PWN equipartition magnetic field of 740 μG based on an RXTE measurement of the pulsar braking index together with models of PWN evolution from Reynolds (1985).

Serafimovich et al. (2004) have pointed out that there appears to be a second break, or "knee" in the spectrum of the pulsar between optical and X-ray wavelengths. This effect, which appears in a few other pulsars such as Vela (Shibanov et al. 2003) and Geminga (Mignani et al. 1998), may be caused by depressed optical emission, perhaps due to unresolved cyclotron absorption lines (Mignani et al. 1998) and propagation effects on the outflow of relativistic particles from the pulsar. Williams et al. (2008) also noted the presence of dust inside the PWN, so the depressed optical flux may also be caused by reddening from dust particles within the PWN itself. Serafimovich et al. also find a knee in the spectrum of the PWN after subtracting the pulsar contribution. We therefore attempted another broken power-law fit to the radio and X-ray data alone. This fit, plotted in blue in Figure 11, results in a break frequency of $2.5^{+56}_{-2.4} \times 10^{14}$ Hz, which in turn gives a PWN magnetic field of $106^{+171}_{-68}\,\mu$G. Although poorly constrained, this raises the possibility that the similar size of the PWN from radio through X-ray could be caused by a very weak magnetic field, leading to correspondingly longer synchrotron lifetimes.

The best-fit models for each shell region are shown in Table 3. Errors are 90% (1.6σ) confidence intervals, calculated using the conf tool in Sherpa. This tool allows all thawed parameters to vary while calculating the confidence interval to find the most general error bounds. We find metal abundances consistent with those of the LMC ISM, although we find slight evidence for depletion in Mg and Si in nearly every region, and a slight enhancement in Ne in Region D. We also find that the surveyed regions are fit by a single-temperature NEI model, with a harder component evident in regions at the edge (A, B, and D). For every shell region surveyed, the equilibrium vapec model resulted in a substantially worse ($\chi ^2_{{\rm red}} = 1.5\hbox{--}2$ or more) fit than an NEI model, required unrealistically high or low values for electron temperature or column density, or both. Extracted spectra for these regions, with best fit models overlaid, are shown in Figure 13. We first discuss the lower temperature thermal component of each shell region individually and then discuss the fits to the higher temperature components in Regions A, B, and D.

Zoom In Zoom Out Reset image size

Table 3. Comparison of vnei and vpshock Fits for Shell Regions

Region	Model	χ2/ν	nH	kT	Mg	Si	Ne	net	norm	Γ	normPL
			(1022 cm−2)	(keV)				(1011 cm−3 s)	10−5Naa		10−5Nbb
A	vpshock+PL	92/92	0.4 ± 0.1	$0.61^{+0.07}_{-0.09}$	0.5 ± 0.2	1.0 ± 0.5	...	<6	$8^{+4}_{-2}$	$2.3^{+0.4}_{-0.3}$	$2.6^{+1.0}_{-0.7}$
	vnei+PL	92/92	0.6 ± 0.2	$0.9^{+0.4}_{-0.3}$	$0.4^{+0.2}_{-0.1}$	$0.7^{+0.4}_{-0.3}$	...	$0.17^{+0.18}_{-0.08}$	$6^{+9}_{-2}$	2.4 ± 0.4	$2.7^{+1.1}_{-0.9}$
B	vpshock+PL	59/54	0.5 ± 0.1	$0.59^{+0.05}_{-0.08}$	0.5 ± 0.2	0.5 ± 0.3	...	<12	$8^{+4}_{-2}$	3.1 ± 0.4	$1.9^{+0.9}_{-0.8}$
	vnei+PL	53/54	0.4 ± 0.1	$0.60^{+0.04}_{-0.06}$	0.6 ± 0.2	0.6 ± 0.3	...	$2.1^{+1.5}_{-0.7}$	$6^{+2}_{-1}$	3.1 ± 0.4	$1.9^{+0.8}_{-0.6}$
C	vpshock	29/36	0.4 ± 0.1	1.1 ± 0.3	0.3 ± 0.1	0.7 ± 0.3	...	$0.5^{0.5}_{-0.2}$	$2.5^{+1.0}_{-0.5}$	...	...
	vnei	33/36	$0.31^{+0.10}_{-0.08}$	$1.36^{+0.08}_{-0.23}$	$0.3^{+0.2}_{-0.1}$	0.7 ± 0.3	...	$0.20^{+0.09}_{-0.10}$	$1.6^{+0.5}_{-0.3}$	...	...
D	vpshock+PL	66/66	$0.41^{+0.19}_{-0.07}$	$0.37^{+0.04}_{-0.09}$	...	...	0.8 ± 0.1	$3^{+7}_{-1}$	$15^{+46}_{-5}$	$2.9^{+0.4}_{-0.3}$	$2.2^{+1.0}_{-0.5}$
	vnei+PL	66/66	$0.4^{+0.2}_{-0.1}$	$0.26^{+0.02}_{-0.03}$	...	...	0.8 ± 0.1	>0.8	$40^{+60}_{-10}$	$3.0^{+0.3}_{-0.3}$	$2.5^{+0.6}_{-0.5}$
E	vpshock	38/38	0.30 ± 0.09	$0.64^{+0.04}_{-0.03}$	$0.4^{+0.2}_{-0.1}$	0.3 ± 0.2	...	<16	6 ± 1	...	...
	vnei	41/38	$0.36^{+0.09}_{-0.20}$	$0.65^{+0.14}_{-0.05}$	$0.4^{+0.1}_{-0.12}$	0.4 ± 0.2	...	$1.2^{+0.6}_{-0.7}$	$5.4^{+2.0}_{-0.8}$	...	...
F	vpshock	54/44	0.6 ± 0.1	$0.63^{+0.06}_{-0.07}$	$0.22^{+0.10}_{-0.09}$	0.4 ± 0.2	...	$1.8^{+2.7}_{-0.8}$	$9^{+4}_{-2}$	...	...
	vnei	53/44	0.6 ± 0.2	$0.67^{+0.16}_{-0.09}$	$0.23^{+0.10}_{-0.09}$	0.4 ± 0.2	...	$0.7^{+0.3}_{-0.2}$	8 ± 3	...	...

Notes. Errors are 90% confidence intervals. Elemental abundances are shown relative to solar values. Abundances not shown are fixed to ISM values. The preferred model for each region is in bold. ^aThermal normalization parameter. N_a = 10⁻¹⁴/4πD²∫n_en_HdV, where D is the distance to the source (50 kpc) and the integral is the volume emission measure. ^bPower-law normalization parameter, N_b = 1 photon keV⁻¹ cm⁻² s⁻¹ at 1 keV.

Download table as:  ASCIITypeset image

Region A covers a large enough area that contamination from the PWN PSF contributes a significant amount of flux, which is accounted for by a fixed power law, shown in orange in part A of Figure 13. This region is chosen to cover the dim northwest portion of the shell, which has not been studied in detail by previous Chandra studies of 0540. The lower-temperature thermal component is best fit by an NEI plasma with an electron temperature of $kT = 0.6^{+0.2}_{-0.1}$ keV. Single-component thermal models did not result in an acceptable fit ($\chi ^2_r \gtrsim 2$), and purely nonthermal models provided a similarly poor fit. The best-fit thermal component requires somewhat depleted abundances of Mg (∼4σ depletion) and Si (3σ depletion) relative to typical LMC values.

Region B covers a small bright feature along the NW edge of the shell. As discussed earlier, this region is similar to the NW region analyzed by Park et al. (2010), with the exception of a point source included in the extraction region by Park et al. As stated in Section 5, based on the starlike appearance of this source in IR through X-ray wavebands, we believe this source to be an unrelated object. The spectrum of the point source is hard, with emission extending out to ∼5 keV, which may have led Park et al. to the conclusion that the spectrum of this region is harder than it actually is. The best fit for the thermal component in Region B is obtained with an NEI plasma with an electron temperature of $kT=0.60^{+0.04}_{-0.06}$ keV and slightly depleted Mg (1σ depletion) and Si (4σ depletion), similar to conditions seen in Region A. The spectrum of the NW region examined by Park et al. is fit by those authors with a two-temperature NEI model, but the harder of the two temperature components in their fit requires an extremely low ionization timescale (n_et < 8 × 10⁸ cm⁻³ s) which is hard to reconcile with the observed density in this region, as discussed in Section 7.1.1 below, and with the age of 0540.

Region C The spectrum in Region C drops off sharply at 2 keV. As in the other shell regions, the best-fit thermal model is an NEI plasma, but the electron temperature is about 2σ higher (kT = 1.1 ± 0.3 keV) than in other shell regions. The unusually high temperature offers the possibility that this region may have two distinct thermal components, but none of the two-temperature models we investigated resulted in an improved fit. A lower-temperature thermal model plus a power law or other nonthermal model resulted in unrealistically steep values for the powerlaw photon index. As with Regions A and B, the abundances of Mg (4σ depletion) and Si (∼3σ depletion) are depleted relative to typical LMC values.

Region D covers a bright linear feature at the western end of the SNR shell, north of the synchrotron-emitting "SW" region analyzed by Park et al. (2010). The most notable feature of this region is the overabundance of Ne (A_Ne = 0.8 ± 0.1, ∼4σ above the typical LMC value). Ne and Fe lines, particularly in low-resolution spectra such as this, can be difficult to distinguish in thermal plasmas, so we generate similar confidence contours (Figure 14) for varying values of the Ne and Fe abundances to confirm that the Ne overabundance is real and not a result of degeneracy in the model parameter space. The contours show that for both vpshock and vnei fits the best-fit region for the neon abundance lies above the typical LMC value of A_Ne = 0.42 and is well bounded on the lower end, while the iron abundance is consistent with the typical LMC value of A_Fe = 0.5.

Zoom In Zoom Out Reset image size

Regions E and F cover separate portions of a bright knot in the southwest, near the "SW" region of Park et al.. Despite their physically distinct appearance, the two regions are both fit by an NEI model similar to that seen elsewhere in the shell, with temperatures of kT ≃ 0.65 keV and depleted abundances of Si (∼6σ below typical LMC value) and Mg (∼3σ below the typical LMC value). As with Region C, the spectra of these regions do not show the higher-energy tail of Regions A, B, and D, and a second component is not required by the best-fit model in either region.

It is interesting to note that fitting the shell regions with vnei and vpshock models results in slightly different fits, as shown in Table 3. Although these two models both simulate NEI conditions, they do so differently; the vnei model implements an NEI plasma with a single, uniform ionization parameter, while the vpshock model implements an NEI plasma with a range of ionization parameters, with an upper bound of the fitted value of τ. In general, vnei and vpshock fits result in almost identical values for the temperature and abundance, which is unsurprising. The exception is Region A, where the two models disagree on the best fit temperature, and the error bars on the vpshock model are unexpectedly small. We investigate these models by generating confidence-contour plots for varying values of temperature and ionization timescale. These plots, shown in Figure 15, reveal that the best-fit kT regions for the two models overlap around kT ∼ 0.6–0.8 keV. The differing best-fit temperatures for the two models are likely a result of the very poorly constrained ionization timescale in the vpshock model, versus the vnei model's relatively well-constrained ionization timescale.

Zoom In Zoom Out Reset image size

The best fit values for ionization timescale differ greatly between vnei and vpshock models, and for most of the regions one of the two models does not have a well-constrained ionization timescale. This suggests that in Regions A, B, E, and F, where the vnei provides a better bound on the ionization timescale, the thermal X-ray emitting material is dominated by one specific density and hence one specific ionization timescale, so a single ionization parameter is a good approximation. For Regions C and D, where a vpshock model provides a better bound on the ionization timescale, we can infer that a wider distribution of densities and thus ionization timescales exists within the thermal plasma.

We attempted to fit the higher-energy tail extending to ∼5 keV in Regions A, B, and D with a second thermal NEI component. For Region A, the second thermal component requires a very high value for the ionization timescale (τ ∼ 5 × 10¹³) which is physically unrealistic for a nonequilibrium plasma (Smith & Hughes 2010), but a second thermal component consisting of an equilibrium plasma results in an inferior fit statistic ($\chi ^2_r=100/92$). For Regions B and D, a two-temperature thermal model results in a fit statistic which is inferior to that of a thermal+power law or srcut fit ($\chi ^2_r=85/52$ for Region B and $\chi ^2_r=69/65$ for Region D). In addition, for all three regions, the srcut model results in a radio spectral index and 1 GHz flux consistent with our radio results reported in Section 4.1. We therefore consider the higher energy emission in all three regions to be nonthermal in character.

The higher energy tails can be well-fit by a power law added to the NEI model, as is shown in Table 4. The power-law indices for all three are consistent with values expected in synchrotron radiation from an SNR shock front (Reynolds 2008). However, a pure power law is a somewhat oversimplified model for synchrotron radiation. We thus attempt an additional fit by replacing the power law component in the fits with a cutoff synchrotron spectrum using the xssrcut model. This model has three parameters: the radio spectral index α, the radio flux density at 1 GHz, and the rolloff frequency. Although we have radio measurements of the flux density and the spectral index integrated over the entire shell, our observations, as detailed in Section 4.1, do not allow us to measure these on spatial scales as small as our extraction regions. We thus allow all three parameters to vary freely when fitting. As shown in Table 4, the radio spectral indices for all three regions are consistent with the integrated value of α = 0.55 found from the radio. The srcut fit demonstrates that the data are consistent with synchrotron radiation from relativistic particles at the SNR shock front.

Table 4. Fits to the Hard Component in Regions A, B, and D

Region	Model	χ2/ν	α	νbreak	S1GHz	Emax	Γ	normPL
				(1017 Hz)	(mJy)	(TeVa)		(10−5Nbb)
A	srcut	92/91	$0.5^{+0.5}_{-0.1}$	$1.4^{+72}_{-1.0}$	<7	$30^{+180}_{-14}$	...	...
A	powerlaw	91/92	...	...	...	...	$2.5^{+0.5}_{-0.4}$	$2.7^{+1.3}_{-0.9}$
B	srcut	58/52	$0.4^{+0.4}_{-0.1}$	$0.18^{+0.33}_{-0.09}$	$0.6^{+2.3}_{-0.5}$	$11^{+7}_{-3}$	...	...
B	powerlaw	53/54	...	...	...	...	3.1 ± 0.4	$1.9^{+0.8}_{-0.6}$
D	srcut	65/65	$0.5^{+0.3}_{-0.1}$	$0.3^{+0.8}_{-0.1}$	<48	$14^{+12}_{-3}$	...	...
D	powerlaw	66/66	...	...	...	...	$2.9^{+0.4}_{-0.3}$	$2.2^{+1.0}_{-0.5}$

Notes. Errors are 90% confidence intervals. ^aCalculated assuming a magnetic field B = 10 μG. ^bPower-law normalization parameter, N_b = 1 photon keV⁻¹ cm⁻² s⁻¹ at 1 keV.

Download table as:  ASCIITypeset image

We can also calculate the maximum particle energy from the rolloff frequency as (Reynolds 2008)

$$\begin{equation} E_{\rm max} = 39 \left(\frac{h\nu _{\rm rolloff}}{1\, {\rm keV}}\right)^{1/2} \left(\frac{B}{10\, {\rm \mu G}}\right)^{1/2}\, {\rm TeV}. \end{equation} \tag{ 2 }$$

As shown in Table 4, assuming that the shell edge has a magnetic field on the order of B = 10 μG, the maximum particle acceleration energies are in the range 10–30 TeV for these regions, although the errors are again large. The values for rolloff frequency and maximum particle energy are comparable to those seen in synchrotron-emitting shocks in other similarly aged SNRs, such as RCW 86 (ν_rolloff = 0.8–1 × 10¹⁶ Hz, which implies a maximum energy E_max = 20–25 TeV; Rho et al. 2002) and SN 1006 (E_max = 20–80 TeV; Rothenflug et al. 2004).

Fit parameters for the wedge regions are shown in Table 5. These regions are all best fit by a vnei+powerlaw model. Although the ionization timescale for most of the fits is large and/or not well constrained, a vnei model results in a superior fit statistic to an equilibrium (vapec) model for all regions. We note that although the power-law continuum is the dominant feature in the spectrum, the thermal component has a minor but crucial contribution to the fit, and pure power law models do not adequately fit the spectra. A representative spectrum for the wedge regions can be seen in Figure 16. Analysis of the spectra from the halo region is affected by severe contamination from the PSF of the pulsar and PWN, which can be seen in the sample spectrum in Figure 16. The best fit values for each wedge have some slight variation across different azimuthal angles, but given the large errors on these fits, are mostly consistent with each other.

Zoom In Zoom Out Reset image size

Table 5. X-Ray Spectral Fits for Halo Regions

Azimuthal Angle	χ2/ν	kT	net	norm	Γ	normPL
		(keV)	(1011 cm−3 s)	10−5Naa		10−5Nbb
0°	130/111	$0.68^{+0.08}_{-0.05}$	>1.6	$3.7^{+0.7}_{-0.8}$	3.4 ± 0.2	$5.5^{+0.5}_{-1.2}$
30°	124/109	$0.62^{+0.08}_{-0.05}$	$2.2^{+4.7}_{-0.6}$	$4.8^{+0.7}_{-1.1}$	3.0 ± 0.2	$4.5^{+0.4}_{-0.9}$
60°	122/110	$0.61^{+0.07}_{-0.05}$	$3^{+10}_{-1}$	$3.9^{+0.7}_{-0.8}$	3.1 ± 0.2	$4.6^{+0.4}_{-0.9}$
90°	122/107	$0.63^{+0.09}_{-0.07}$	$2.5^{+6.5}_{-0.9}$	$3.3^{+0.6}_{-1.1}$	3.2 ± 0.2	$4.7^{+0.4}_{-0.8}$
120°	129/107	$0.60^{+0.09}_{-0.06}$	$2.2^{+3.3}_{-0.6}$	$3.8^{+0.6}_{-1.1}$	2.9 ± 0.2	$4.0^{+0.4}_{-0.9}$
150°	129/107	$0.60^{+0.10}_{-0.07}$	$2.3^{+3.4}_{-0.7}$	$3.8^{+0.7}_{-1.1}$	2.9 ± 0.2	$4.0^{+0.4}_{-0.8}$
180°	127/106	$0.69^{+0.08}_{-0.05}$	>2.5	$3.7^{+0.4}_{-0.8}$	$2.5^{+0.2}_{-0.1}$	3.1 ± 0.3
210°	126/107	$0.78^{+0.07}_{-0.06}$	>7	$3.7^{+0.4}_{-0.8}$	$2.5^{+0.2}_{-0.1}$	3.1 ± 0.3
240°	117/109	$0.74^{+0.06}_{-0.05}$	>9	4.7 ± 0.4	$2.5^{+0.2}_{-0.1}$	3.1 ± 0.3
270°	116/109	$0.74^{+0.06}_{-0.05}$	>7	4.7 ± 0.9	$2.6^{+0.2}_{-0.3}$	3.5 ± 0.7
300°	99/109	$0.75^{+0.06}_{-0.05}$	>9	$4.8^{+0.4}_{-0.6}$	$2.7^{+0.2}_{-0.1}$	3.7 ± 0.3
330°	107/111	$0.72^{+0.06}_{-0.05}$	>9	$4.8^{+0.5}_{-0.4}$	$3.0^{+0.2}_{-0.1}$	4.6 ± 0.3

Notes. Errors are 90% confidence intervals. Elemental abundances are fixed to LMC ISM values. ^aThermal normalization parameter. N_a = 10⁻¹⁴/4πD²∫n_en_HdV, where D is the distance to the source (50 kpc) and the integral is the volume emission measure. ^bPower-law normalization parameter, N_b = 1 photon keV⁻¹ cm⁻² s⁻¹ at 1 keV.

Download table as:  ASCIITypeset image

7.1.1. Density and Shock Velocity

The shell morphology, as discussed in Section 3, is indicative of an interaction between the shell and surrounding denser clouds. We can elucidate this interaction by examining densities and shock velocities calculated from the X-ray spectra. All surveyed regions in the shell show spectra that are best fit by NEI conditions. The regions at the outer extent of the X-ray emission (A, B, and D) have a power-law component added to this soft thermal component. The softer thermal emission is likely caused by the interaction between the expanding supernova blast wave and the surrounding ISM. We can calculate the approximate blast wave speeds in these regions by solving the Rankine–Hugoniot relations in the case of a strong shock, with γ = 5/3, which yields the equation $v=\sqrt{16kT/3\mu m_p}$, with μ = 1.2 for a fully ionized plasma. This relation assumes that the electron and ion temperatures have equilibrated, which is not necessarily the case (e.g., Ghavamian et al. 2007). However, in a fully ionized, shock-heated plasma, the electron and ion temperatures can equilibrate very quickly (within a few hundred years; see Equation (36.37) in Draine 2011) via Coulomb interactions. We may thus use this equation as a conservative lower limit.

The resulting velocities, which can be seen in Table 6, are on the order of a few hundred km s⁻¹. This is extremely slow for a SNR as young as 0540. We may conclude from this that the outward-moving blast wave has interacted with denser surrounding material fairly recently, and that it has quickly decelerated. We calculate the density based on the best-fit ionization timescale τ for each region; the density is thus n_e = τ/t, where t is the time since the region has been shocked. This time, t, is estimated by assuming the initial blast wave speed is ∼10, 000 km s⁻¹ (Chevalier 1977) and that the initial SNR blast wave is spherically symmetric. We assume that the blast wave expands at this constant speed until it reaches the position of the shell region in question. We then calculate t for each region by subtracting the time it takes the blast wave to reach that region from the accepted age of 1100 yr. The densities calculated in this way for each region are listed in Table 6.

Table 6. Shock Speeds and Densities for Shell Regions

Region	v	nea	neb
	(km s−1)	(cm−3 v10)	(cm−3f−1/2)
A	700 ± 100	$1.0^{+1.4}_{-0.5}$	$0.5^{+0.3}_{-0.1}$
B	$510^{+30}_{-20}$	$10^{+7}_{-3}$	$4.0^{+0.6}_{-0.3}$
C	$680^{+90}_{-100}$	3 ± 1	$2.0^{+0.4}_{-0.3}$
D	$400^{+20}_{-50}$	$30^{+60}_{-10}$	$4.4^{+4.4}_{-0.8}$
E	$530^{+60}_{-40}$	3 ± 1	$5.2^{+0.9}_{-0.4}$
F	$530^{+50}_{-20}$	$6^{+3}_{-2}$	7 ± 1

Notes. ^aEstimated density, calculated by dividing the ionization timescale for each region by the approximate time since shock, based on the radial distance of each region. v₁₀ is the initial speed of the SNR shock in multiples of 10,000 km s⁻¹. ^bEstimated density, calculated from the thermal normalization parameter using Equation (3).

Download table as:  ASCIITypeset image

We can check these results of this calculation by calculating the density in a different way, based on the volume of the emitting region and the emission measure (∫n_en_hdV) of the thermal plasma in that region. These quantities can be calculated from the thermal normalization parameter. If we assume that the densities of electrons and protons are roughly constant through the volume in each region and that n_e = 1.2n_h, we can then estimate the density as

$$\begin{equation} n_e = \left(\frac{4 \pi N_a}{1.2 V f}D^2_{{\rm cm}}\right)^{1/2}, \end{equation} \tag{ 3 }$$

where N_a is the thermal normalization parameter, D_cm is the source distance in cm, and V is the volume of the emitting region in cm³. The volume filling factor f is a number between zero and one to account for the percentage of the volume in each region actually taken up by emitting plasma. We assume that the source distance is 50 kpc and that the volume of each region can be approximated by its two-dimensional area on the plane of the sky multiplied by the longest axis of the region.

The results estimated from the normalization parameter, shown in Table 6, are roughly consistent with the values estimated from the ionization timescale, except for the value seen in Region D, where the density calculated from the NEI norm is nearly an entire order of magnitude lower. It is possible that the initial shock speed assumed in the calculation in Table 6 is too high, and that the shock front thus reached this region more recently than assumed in this calculation. It is also possible that the volume filling factor is unusually low in this region. Finally, it is possible that the estimated volume for this region is incorrect. The shape of the western shell in 0540 is complex and very non-spherical, making an accurate estimate of the emitting volume of each region difficult.

We can see that in higher density regions, lower velocity shocks predominate and conversely, in lower density regions, higher velocity shocks predominate. For instance, Region D, with the highest density of any surveyed shell region, shows the lowest temperature X-ray plasma and the slowest shock. This is precisely the situation we would expect to see if the shock was interacting with a dense cloud in the surrounding ISM.

7.1.2. Multi-wavelength Emission

The highly inhomogeneous density of the ISM surrounding 0540 also results in a variety of types of emission around the edge of the shell. As shown in Figure 17, Regions B and D are coincident with an area of diffuse [O iii] emission (Mathewson et al. 1983) and with knots of radio emission. The multi-wavelength emission in the western hemisphere is indicative of a blast wave expanding into a complicated medium, with the radio knots and synchrotron emission seen in the X-ray spectra representing areas of lower density and the softer thermal X-rays and optical emission representing interactions with density enhancements in a clumpy medium. In addition, longer wavelength IR emission in the southwest appears outside of the X-ray emission (see Figure 4), suggesting that a particularly dense cloud has been shock heated with only larger dust grains remaining. The shorter grains have been destroyed by the shock front, or sputtered away in the hot, shocked material, causing the lack of shorter-wavelength diffuse IR emission discussed in Section 2.3. This cloud may be leftover material from the progenitor star's formation, sitting at the edge of a cavity swept out by the progenitor's wind.

Zoom In Zoom Out Reset image size

The western area of the shell also shows strong variations in rotation measure in Figure 8, such as the knots of reversed rotation measure around α = 5^h40^m7^s, δ = −69°19'58'', which suggest that the electron density and magnetic field direction in this region varies on relatively small scales, as we would expect in a highly clumpy medium. On a larger scale, the magnetic field vectors along the entire western edge of the shell, shown in Figure 9, do not appear to be oriented radially, as is typical of young SNRs (Milne 1987). The magnetic field vectors instead appear to be tangential along the edge of the radio emission, a configuration usually seen in older SNRs and attributed to compression in shocks that have decelerated into the radiative regime (Reynolds et al. 2012).

The presence of a hard nonthermal component in Regions A, B, and D, which is evident in Figure 13, is likely due to the presence of synchrotron radiation, which was also seen at the extreme east and west edges of the shell by Park et al. (2010). The X-ray photon indices in these regions of Γ ∼ 3 are within the range of other young SNR with synchrotron emission at their shock fronts (Reynolds 2008), and the results of an srcut fit to the nonthermal component further support the idea of synchrotron emission in these regions. Unfortunately, due to observational constraints detailed in Section 4.1, calculating the radio spectral index for these regions is not feasible. The superimposition of synchrotron emission and thermal emission may be due to line of sight confusion along the clumpy, highly inhomogeneous medium. In lower density areas the shock is collisionless and able to sweep up the ambient magnetic field and accelerate particles to relativistic velocities, thus producing the observed synchrotron emission. Shocks that generate synchrotron emission in this manner are usually very fast (>3000 km s⁻¹ or more; Bell 1978), yet we see the fast, collisionless shocks required for synchrotron emission superimposed on the much slower shocks producing thermal emission. This requires that the slower thermal shocks have decelerated very recently—for instance, for a ∼3000 km s⁻¹ synchrotron producing shock and the 400 km s⁻¹ thermal shock to both appear in Region D, which is 1 pc thick, the slower shock must have decelerated to its current speed within the last ∼350 yr.

7.1.3. Comparison with the Crab Nebula

7.1.4. Cloud/Shock Interaction

7.1.5. Shell Summary

The X-ray observations of the halo region show a mixture of thermal and nonthermal emission. Several explanations have been advanced for the cause of this diffuse halo emission. It is possible that Rayleigh–Taylor (RT) instability between the expanding plerion and the surrounding ejecta may lead to mixing between the synchrotron emitting PWN material with thermal plasma. It is also possible that the diffuse halo emission could be caused by photons from the pulsar and PWN being scattered by foreground dust. A dust halo of this type is seen in the galactic SNRs G21.5-0.9 (Bocchino et al. 2005; Matheson & Safi-Harb 2010) and Kesteven 75 (Helfand et al. 2003). Morse et al. (2006) have suggested, based on optical line profiles, that the diffuse [O iii] emission in the halo is caused by the forward shock from the expanding PWN overtaking slow-moving ejecta in the interior of the SNR. Alternatively, Williams et al. (2008) have suggested, based on IR spectra, that the diffuse halo emission is caused by UV photons from the PWN photoionizing material immediately surrounding the nebula. Finally, Petre et al. (2007) have noted that the halo emission may be simply foreground material from the SNR shell seen in projection.

RT instabilities, arising from the expansion of the pulsar wind into the denser surrounding environment, can cause mixing between thermal and synchrotron emitting material, such as that seen in the older LMC PWN 0453-68.5 (McEntaffer et al. 2012). In that remnant, RT instabilities are caused by repeated compression and expansion of the PWN due to interactions with the SNR reverse shock. However, RT instability can also arise from the expansion of the tenuous synchrotron-emitting PWN material into the denser surrounding ejecta early in the life of a PWN (Jun 1998). Noting that thermal material and power-law emitting material are seen together from radii of 5'' to at least 9'' (1.2–2.2 pc from the pulsar), we can calculate the requirements for RT instability to occur on this scale. Jun et al. (1995) give the most rapidly growing and therefore dominant length scale λ_c for RT instability, given a magnetic field at a set strength B, as

$$\begin{equation} \lambda _c = \frac{2B^2 \cos ^2 \theta }{g(\rho _2 - \rho _1)}, \end{equation} \tag{ 4 }$$

where g is the local acceleration (effective gravity), θ is the angle between the field and the interface surface, and ρ₂ and ρ₁ are the densities of the lighter (PWN) and heavier (SN ejecta) media. The field should be parallel to the interaction surface everywhere (cos θ ∼ 1), as suggested by polarization studies of the nebula (Chanan & Helfand 1990; Lundqvist et al. 2011; Figure 9). We can calculate the required density differential (Hester et al. 1996) as

$$\begin{equation} \delta \rho = \frac{B^2}{\lambda g}. \end{equation} \tag{ 5 }$$

This is a minimum density jump for RT instability to occur at the interface, below which the magnetic field stabilizes the interface and no mixing of any sort occurs. For a density jump ∼δρ, the magnetic field shapes the material at the interface into long fingers (Jun 1998); if the density jump is ≫δρ, the effect of the field is unimportant and we recover the purely hydrodynamic case. The effective gravity of the similar synchrotron nebula in the Crab is ≃ 3.53 × 10⁻³ cm s⁻² (Hester et al. 1996).

Using the PWN magnetic field value of 250 μG calculated in Section 4.3, we get an estimated critical density jump of 3.8 × 10⁻²⁵ g cm⁻³, or about 0.2 cm⁻³ for inner ejecta with an average atomic mass μ = 1.2m_H. Estimating the density of the thermal plasma in the wedge regions leads to a number density in this region of ∼2/f cm⁻³. As this is at least ten times the critical density, and, depending on the value of the filling factor, perhaps well over an order of magnitude higher, it is at least theoretically possible that RT instability could be the cause of the power-law tail seen in the halo spectra.

To check the dust-scattering model, radial brightness profiles for the hard (1.2–8.0 keV) emission were also extracted from eastern and western halves of the interior area between radii of 10 and 60 pixels (5–30'') using the Sherpa tool prof_data. To remove the contribution to this halo from the Chandra PSF, we used our MARX simulations to extract brightness profiles for the PSF from the pulsar and PWN and subtracted these simulated brightness profiles. The PSF-subtracted profiles were fit to a simple dust-scattering halo model (Hayakawa 1970) of the form I∝[j₁(x)/x]², where j₁(x) is the spherical Bessel function of the first kind. The profile for the western half becomes non-monotonic about 35 pixels (18'') from the pulsar, so we cut off the fit at that point. We can confidently reject foreground dust scattering as a primary cause for the halo based on several considerations. In the 1.2–8.0 keV band, where dust scattering would be most noticeable, the brightness profile is not well fit at all ($\chi ^2_r \sim 9$) by the dust scattering model. We may also note, qualitatively, that dust scattering halos of this type are mainly observed around objects in the galactic plane with very high column densities. For instance, G21.5-0.9 has a column density of n_H = 2.2 × 10²² cm⁻² (Matheson & Safi-Harb 2010), four times that of 0540. In addition, the gas-to-dust ratio in the LMC is more than double that of the Milky Way (Meixner et al. 2010), further reducing the scattering contribution from LMC dust. Given the low column density through the Milky Way to 0540, we do not expect any significant amount of scattering from local galactic dust.

The shock (Morse et al. 2006) and photoionization (Williams et al. 2008) scenarios for the halo emission each rely on the presence of ejecta surrounding the PWN. In [O iii] images, the PWN is surrounded by a prominent ring-like structure. A similar [O iii] ring structure exists around the Crab Nebula, which has been interpreted (Sankrit & Hester 1997) as cooling immediately behind a shock being driven into the inner ejecta by the expanding PWN. Morse et al. (2006) have suggested that the halo may be the result of a fast shock from the expanding PWN overtaking and heating inner, slow-moving ejecta. Kirshner et al. (1989) examined the ratio of the [S ii] λλ4069, 4076 doublet to the [S ii] λλ6717, 6731 doublet and found that the ratio was consistent with a temperature of about 10,000 K, characteristic of S⁺ in shock heated gas (Fesen et al. 1982). In the photoionization scenario, supported by Spitzer IR spectra (Williams et al. 2008), a very slow shock of about 20 km s⁻¹ is currently propagating outward inside the 8'' halo, while a faster shock of ∼250 km s⁻¹ is propagating outward outside the halo. The authors imply the low shock velocity from the emissivity of lines such as [O i], [O iv], [Fe ii], and [S ii]. These authors argue that the slow shock is best explained by confinement within an iron-nickel bubble (Woosley 1988; Li et al. 1993). In this phenomenon, which has also been observed in SN1987A (Basko 1994; Wang 2005), radioactive energy deposition by the decay of ⁵⁶Co and ⁵⁶Ni causes an inner bubble of heavy-element ejecta to expand rapidly and sweep up a clumpy shell of inner ejecta. UV photons emitted from the PWN photoionize this surrounding ejecta, which is slow-moving and highly enriched (A_metal ≃ 2) in heavy elements, to produce the observed [O iii] emission. The X-ray spectra for the halo emission are characterized by a plasma temperature of kT ∼ 0.7 keV, which is consistent with a shock moving at ∼550 km s⁻¹, but if this shock were moving through heavy-element ejecta, we would expect the X-ray spectra to show enhanced elemental abundances from this shock-heated ejecta, which they do not.

Neither the shock, photoionization, or RT scenarios present an entirely satisfying explanation of the X-ray emission. The lack of X-ray ejecta emission does not present a problem in the RT scenario, since the analogous RT filaments in the Crab do not emit in the X-ray (Weisskopf et al. 2000). However, we do not see visible RT filaments at any wavelength in the halo region. Filamentary structures that may be due to RT instability are present within the PWN, as shown by HST WFPC observations in [O iii] and [S ii] (Morse et al. 2006), but these filaments do not extend beyond the PWN. These difficulties with explaining the X-ray emission from the halo leave us with the possibility suggested by Petre et al. (2007) that the halo emission may merely be a chance alignment of an area of brighter foreground emission from the shell. Despite the strong contamination, the thermal component best-fit X-ray model is almost identical in temperature and ionization timescale to those seen in the shell. This supports the explanation of line of sight confusion between the shell and the plerion.