INFRARED AND X-RAY SPECTROSCOPY OF THE Kes 75 SUPERNOVA REMNANT SHELL: CHARACTERIZING THE DUST AND GAS PROPERTIES

The Chandra observations of Kes 75 were carried out with the Advanced CCD Imaging Spectrometer on 2006 June 5, 7, 9, and 12, under observation ID numbers 7337, 6686, 7338, and 7339, and an exposure time of 18, 55, 40, and 42 ks, respectively. The data set was previously analyzed by Morton et al. (2007) and Su et al. (2009), but here, we reanalyze the spectra using a different approach to background selection and fitting. The data were reprocessed and cleaned using CIAO v. 4.3, resulting in a total clean exposure time of 152 ks.

Since the observations were carried out close in time and have similar chip orientation, we merged the individual data sets into a single event file. The merged image is shown in Figures 1 and 2. We chose extraction regions for the SE and SW rims similar to those in Morton et al. (2007) and Su et al. (2009), with an additional region covering the inner SNR between the PWN and the southeastern rim of the shell. The extraction regions for the X-ray spectra are shown in Figure 2(a).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Spitzer IR spectroscopy of the shell in Kes 75 was carried out on 2008 June 5 (program ID 50447) using the long–low (LL) and long–high (LH) modules of the Infrared Spectrograph (IRS; Houck et al. 2004) that cover the range from 14–37 μm. The locations of the IRS slits overlaid on the MIPS 24 μm image (Morton et al. 2007) are shown in Figure 3. The spectra were taken at positions 18^h46^m223, −2°59'469 and 18^h46^m3036, −2°58'549, corresponding to the SW and SE portions of the shell, respectively. In order to measure the background emission for the LH module, we also placed a slit at position 18^h46^m2403, −3°01'490, south of the partial shell. All observations were taken with 10 cycles of 6 s exposures and the data were processed with the pipeline version S18.0.2.

Zoom In Zoom Out Reset image size

The LH spectroscopy was cleaned with the IRS Rogue Pixel Mask Editing and Image Cleaning Software (IRSCLEAN1.9). The background spectrum was subtracted from the LH data and the Basic Calibrated Data of each nod position were combined. Spectra for the SW and SE portion of the shell were extracted using the Spitzer IRS Custom Extractor (SPICE) with full slit extractions and extended source calibration. For the LL modules, the spectra were extracted using the CUbe Builder for IRS Spectral Maps (CUBISM; Smith et al. 2007), which allowed us to select sub-slits from the off-source nods for background subtraction. Since the background in the vicinity of Kes 75 is spatially variable (89–102 MJy sr⁻¹ at 24 μm), some off-source nod positions produced negative residuals in the background-subtracted spectra. The final background was produced by averaging spectra extracted from off-source sub-slits that did not result in negative residuals. The source spectra were extracted from sub-slits covering the shell emission and inner region on the SNR, and are shown in Figure 2(b). The rest of the analysis was carried out with the Spectroscopic Modeling, Analysis, and Reduction Tool (SMART; Higdon et al. 2004), including fitting of the emission lines.

The three-color image of Kes 75, displayed in Figure 1, shows the Chandra X-ray emission in blue, MIPS 24 μm IR emission in green, and Very Large Array radio emission in red (Becker & Helfand 1984). The IR image shows a partial SNR shell, with the brightest emission concentrated in the SE and SW rims which coincides spatially with the X-ray and radio emission. The 24 μm image also shows fainter IR emission in the inner region of the SNR, between the PWN and the southern rim, that is more diffuse than the filamentary shell. Extinction-corrected 24 μm fluxes of the bright portions of the shell are measured to be 5.1 and 7.0 Jy in the SE and SW rims, respectively. We also measured the total flux in the shell using the diffuse IR emission inside the SNR as background. Since the IR spectra suggest that the emission from the shell is composed of two distinct dust temperature components, the latter flux densities are estimates for the warm dust component that is concentrated in the shell (see Section 3.3). The measured values are listed in Table 1. As discussed in Section 3.3, the IR emission is dominated by dust continuum, with very little or no contribution from line emission. The radio synchrotron emission from the shell has a spectral index α = 0.7 and a flux density of ∼10 Jy at 1 GHz (Becker & Kundu 1976). When extrapolated to Spitzer IR wavelengths, the contribution from the synchrotron emission is negligible.

The X-ray emission spatially correlates with the IR, as is best seen in Figure 2. The rims of the SNR shell are dominated by thermal emission with a range of temperatures. This is best seen in the X-ray three-color image in Figure 2(a). The colors show that some portions of the shell are dominated by softer X-ray emission, while other regions have a harder X-ray spectrum (blue). The SE portion of the shell appears particularly clumpy, with softer X-ray clumps embedded in more diffuse hard emission. Figure 2 also shows that diffuse X-ray emission fills the inner SNR. Su et al. (2009) show that the spectrum of the diffuse emission north of the PWN is featureless and most likely non-thermal in nature. This emission likely represents the dust-scattered halo from the PWN. However, the emission from the inner SNR region south of the PWN is more enhanced than in the north and shows evidence of thermal X-ray emission.

The Chandra X-ray spectra were extracted from regions shown in Figure 2(a). The best-fit parameters are shown in Table 2. We also refitted the spectrum of the entire PWN, excluding the pulsar, with a power-law model and the Tuebingen–Boulder ISM absorption model XSTBABS. The spectrum contained ∼100,000 counts after subtracting a background outside of the SNR. The absorbed power-law model provided a good fit to the spectrum, giving a χ² of 0.92. Using the XSTBABS absorption model, we find N_H = 3.75 × 10²² cm⁻², a slightly lower value than reported previously, and Γ = 1.98 ± 0.3, consistent with previous results.

Su et al. (2009) modeled the X-ray spectra from the SE and SW rims with a thermal component, plus a non-thermal one to account for possible dust scattering. Instead of using this approach, we assumed that the emission from the dust scattering halo is approximately symmetric about the PWN, and we chose background regions in the northern part of the shell that do not appear to be contaminated by thermal emission to subtract the scattered emission component. The background regions are shown in Figure 2(a). The region labeled "Bkg1" was used as background for the "Inner" region, while the region labeled "Bkg2" was used as background for the SE and SW regions. The source spectra for the SE (SW) rim contained 37,000 (16,000) counts, while the total counts in the background, including emission from the dust scattering halo, was approximately 2900 (2700) counts. The total number of source counts in the "Inner" region was 8900, but the background in this case was significantly higher, approximately 5350 counts. The spectra from the SE and SW regions were grouped by 30 counts bin⁻¹, while the spectrum from the inner region was grouped by 100 counts bin⁻¹.

The X-ray spectra were fit with two different models, a non-equilibrium ionization collisional plasma model (VNEI) and a plane-parallel shock plasma model (VPSHOCK), that also include elemental abundances as free parameters. The models are shown in Figures 4, 5, 6, and 7. A single-temperature VNEI thermal component did not provide a good fit to the SE and SW regions, while these areas were well-fit by the single-temperature VPSHOCK model (see Figures 4 and 7). The best-fit parameters for all models are listed in Table 2. The single-temperature VPSHOCK model implies a temperature of 1.48 keV and 1.93 keV in the SE and SW, respectively, and 2.47 keV in the inner SNR region. The ionization timescales are on the order of $\rm 10^{11}\rm \:s\:cm^{-3}$ in the shell and 10¹⁰ s cm⁻³ in the inner SNR region. We let the hydrogen column density vary and found that it is somewhat lower than for the PWN (N_H = 3 × 10²² cm⁻²), as suggested by Su et al. (2009).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

As will be discussed in the following sections, the IR data imply the presence of an additional, lower temperature plasma component. A somewhat better χ² is achieved using a two-temperature VPSHOCK model, shown in Figure 5. For comparison, we also fit the data with a two-temperature VNEI model (Figure 6), but the fit was slightly worse in this case. The temperature of this second component is approximately 0.2 keV for both models, but the normalization and ionization timescale are not well constrained. A similar χ² value is achieved for a relatively short ionization timescale of ∼10⁹ s cm⁻³ and a very long timescale of ∼10¹³ s cm⁻³. The F-test implies that the presence of this lower temperature component is not statistically significant. However, as will be discussed in the following sections, the IR data imply the presence of an additional, lower temperature plasma component.

The X-ray spectra show clear evidence for Si, S, Ar, Mg, and Fe lines, but in contrast to previous studies (Morton et al. 2007; Su et al. 2009), we find no definitive evidence for enhanced abundances in either of the thermal components. The main difference in our analysis is a more careful approach in background selection that accounted for the emission from the dust scattering halo. After subtracting this emission from the source spectra, the enhanced Fe abundance was no longer required by the fit. While the best-fit residuals in Figures 4, 5, and 6 clearly show that the models do not provide a good fit to some of the emission lines, varying the abundances did not significantly improve the fit, and the abundances remained around the solar value. We believe that the residuals are caused by the variation in temperature in the higher energy component with a shorter ionization timescale. These differences in temperature are clearly seen in Figure 2(a), and indeed, fits to small sub-regions in the SE and SW shells show that the temperature ranges from 0.9 keV for some of the smaller clumps, up to 2 keV for more diffuse regions. The abundances were consistent with solar even for these various sub-regions.

The IRS slits were positioned along the brightest regions of the SW and SE parts of the shell (see Figure 3). The final background-subtracted LH spectra of the shell are shown in Figure 8. The spectra of the shell are dominated by continuum emission from dust with very little contribution from line emission. Emission lines that are present in the background-subtracted LH spectra are listed in Table 3, and they include [O iv] (25.89 μm), [Fe ii] (25.99 μm), [S iii] (33.48 μm), and [Si ii] (34.82 μm). The listed uncertainties in the observed line intensities are statistical uncertainties only, and they do not include the uncertainties caused by the spatially variable background in the vicinity of Kes 75. Since the background for the high-resolution spectra was extracted from a single region south of the SNR (see Figure 3), it is difficult to say how much line emission, if any, is associated with Kes 75. For this reason, the measured line intensities should be treated as upper limits. All of the lines except O iv appear in the local background. In addition, the lines all have a full width at half-maximum (FWHM) that is similar to the expected resolution of the IRS of 500 km s⁻¹, and none show evidence for significant broadening that would be expected from a rapidly expanding blast wave. There are no lines detected in the LL modules after background subtraction. However, the spectra have high uncertainties due to the high local background that varies significantly across the LL slits.

Zoom In Zoom Out Reset image size

Extinction correction was applied to the emission line intensities and the LL spectra used in the dust model fitting. We used the extinction curve of Chiar & Tielens (2006) and the relation N_H/A_K = 1.821 × 10²² cm⁻² from Draine (1989). We set the value of N_H = 3.5 × 10²² cm⁻², an average derived from the fitting of the Chandra data to the SE and SW shell.

The morphological similarities between the X-ray and IR emission in the southern part of the shell suggest that the emitting dust is likely collisionally heated by the shocked, X-ray emitting gas. The lack of strong IR line emission also indicates that no cooler and denser regions are present where strong UV and optical emission might radiatively heat the dust. The dust grain model used to fit the IR spectra is the BARE-GR-FG model from Zubko et al. (2004) which includes polycyclic aromatic hydrocarbons (PAHs), silicate, and graphite grains. Zubko et al. (2004) determined the grain size distribution for the model by simultaneously fitting the average interstellar extinction, emission, and abundances. Our fitting was performed using a dust heating code that calculates spectral models for dust grains immersed in an X-ray plasma (Dwek 1986; Arendt et al. 2010). The model includes stochastic heating of smaller grains and thermal sputtering of dust grains that reduces the size of the initial grain population. The free parameters of this dust model are the electron density n_e, electron temperature T_e, and the change in the size of each grain due to sputtering, Δa, that depends on the age of the shocked gas. We produced a grid of models using a density range n_e (cm⁻³) = [0.35, 0.7, 1.4, 2.8, 5.7, 11, 23, 45, 90, 180], and T_e (K) = [1.0, 1.8, 3.2, 5.6, 10, 18, 32] × 10⁶. The sputtering amount Δa was varied from 0.0–0.01 μm in steps of 0.0005 μm.

We used the above models to fit spectra from three different regions along the Kes 75 shell: the brightest SE and SW rims and the fainter diffuse emission inside of the shell. The LL sub-slits used for spectral extractions are shown in Figure 2(b). We initially used a background from outside of the SNR for all three regions, and we found that the spectra from the SE and SW cannot be well fitted by a single set of n_e and T_e parameters. The emission appears to originate from two distinct dust temperatures, a warmer component and a cooler one that is more spatially extended. The emission from this cooler dust also fills the inside of the SNR ("Inner SNR" region in Figure 2). In order to isolate the warmer dust component that is only present in the rims, we decided to subtract the inner SNR contribution from the SE and SW spectra and then fit the spectra with a single temperature. The final fitting was carried out for regions "SE" and "SW" in Figure 2(b) where the region labeled "Inner" was used as the background. This inner region was then fitted separately using a background region from outside of the SNR shell. The spectra are shown in Figures 9 and 10. Prior to fitting, emission lines were subtracted and the spectra were extinction corrected using the extinction curve from Chiar & Tielens (2006) and N_H = 3.5 × 10²² cm⁻². The best-fit models for the SE and SW regions are shown as solid red curves in Figure 9, while the best-fit model for the inner SNR region is shown in Figure 10. The dotted red curve in Figure 9 represents the spectrum of the grain size distribution prior to any sputtering. The best-fit parameters for these models are listed in Table 4, while the χ² contour levels for these parameters are shown in Figure 11, where the shaded regions represent values of [1.2, 2, 4, 8, 16, 32] × min χ².

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The diffuse IR emission from the inner SNR region seems to originate from a dust population in which the larger grains are characterized by a temperature of 55 K, while the emission from the SNR shell requires the 55 K component in addition to a warm dust component with a temperature of 140 K. We used the inner SNR region as a background for the SE and SW shell spectra, and in that way we have subtracted the contribution from the more diffuse, cooler dust component. The residual IR spectrum from the warm dust in SE and SW rims requires a high plasma temperature and relatively high electron density. The best-fit value for the electron density for both regions is n_e = 90 cm⁻³, with an electron temperature of T_e = (1–2) × 10⁷ K (0.9–1.7 keV). While the fit clearly requires a relatively high density, the temperature is somewhat less constrained (see Figure 11).

The best-fit values for the sputtering parameter Δa are 0.01 (0.002) μm for the SE (SW) spectrum. These values imply that almost all of the smaller PAH grains have been destroyed and approximately 50% (25%) of the mass of silicate and graphite grains. However, since we only have IR spectra longward of 15 μm, this sputtering parameter Δa is not well constrained by the model, even though the lack of emission at IRAC wavelengths does imply that the smaller grains have likely been destroyed in these regions.

While the presence of the cooler plasma component is not conclusive from the fits to the X-ray data, it is implied by the presence of the 55 K dust component. The best-fit model to the IR spectrum of the inner SNR region implies a lower electron temperature of 1.8 × 10⁶ K (0.16 keV) and a much lower electron density, n_e = 2.8 cm⁻³. The minimum χ² for the inner region is actually achieved for a broader density range of 1–15 cm⁻³, but this is still several times lower than what is required for the SE and SW rims. The Δa parameter is better constrained for the inner region. The best-fit value of Δa = 0 implies that very little sputtering has occurred in this region (see Figure 11).

Interestingly, based on the best-fit electron temperature to the SE and SW IR spectra, the dust in these regions appears to be associated with the higher temperature (2 × 10⁷ K) X-ray emitting plasma, rather than the 0.2 keV (2 × 10⁶ K) component as suggested by Morton et al. (2007). The association of the IR emission with this harder X-ray emission is also implied by the spatial correlation of the IR and X-ray emission, as seen in Figure 2(b). The IR emission from the inner SNR region may be associated with the lower temperature plasma component, since the IR and X-ray fits provide independent estimates of the n_e and T_e that are consistent with each other.

The electron density for the two thermal components in the X-ray fits was determined from the normalization of the VPSHOCK model, 10⁻¹⁴n_en_HV/4πd² cm⁻⁵, where V is the volume of the emitting region and d is the distance to the SNR, set to the value of 10.6 kpc. Similar to Morton et al. (2007), we assumed a slab-like volume for the X-ray emitting regions. We took the area of the SE and SW regions shown in Figure 2(a) and assumed that the depth along the line of sight is equal to the long axis of each region. For an assumed distance of 10.6 kpc, the assumed volumes are 15 (26) pc³ for the SE (SW) region. We also assume that n_e = 1.18n_H, as for a fully ionized gas. The estimated electron densities for the X-ray emitting components are 5.6f^−1/2d^−1/2_10.6 cm⁻³ and 3.1f^−1/2d^−1/2_10.6 cm⁻³ for the ∼1.5 keV component for SE and SW rims, respectively (Table 4), where f is the filling factor that accounts for the fact that the emitting gas may occupy a smaller fraction of the chosen volume. The ionization timescale implies a similar density, assuming that the shock age is equal to the estimated SNR age of 880 yr. Since the normalization of the cooler plasma component is not constrained, we can only derive upper limits on the density and gas mass in the ∼0.2 keV component. The upper limits on the density and mass calculated from the normalization of the VPSHOCK model (best-fit value plus 1.6σ) for the same spatial regions (SE/SW regions in Figure 2(a)) are 140d^−1/2_10.6 cm⁻³ and 200d^−1/2_10.6 cm⁻³, and 43d^5/2_10.6 M_☉ and 110d^5/2_10.6 M_☉ for the SE and SW, respectively. These values are listed in Table 4.

The estimate of n_e from the X-ray fits is significantly lower than the high density derived from the fits to the IR spectrum. If the warm dust component is associated with the 1.5 keV plasma, as suggested by implied temperatures, it may be possible that the X-ray emitting gas in the shell is clumpy and concentrated in a smaller volume. If we assume that the density of the clumps is 90 cm⁻³, as derived from the IR data, we estimate a low filling factor that is on the order of a few times 10⁻³. As can be seen from Figure 2(a), the shell does indeed appear to be clumpy, so assuming that higher density clumps are present in the shell may not be unreasonable. However, the large discrepancy between the X-ray and IR derived densities remains a problem.

We estimated the total gas masses in the hot component of the shell to be 1.7d^5/2_10.6 M_☉ in each of the rims. The dust masses were calculated using a grain size distribution from the best-fit models and the total MIPS 24 μm fluxes from the SE and SW rims of the shell. The fluxes used for the total dust estimates are listed in Table 1. For the warmer dust component we used residual fluxes after subtracting the contribution from the inner SNR emission. These values are equal to 2.2 Jy (SE) and 3.4 Jy (SW). We then subtracted these values from the total flux in the shell to estimate the contribution of the cooler dust component and find these values to be 3.3 Jy (SE) and 3.6 Jy (SW). Dust mass estimates are listed in Table 4. We find a total dust mass of 6 × 10⁻³d²_10.6 M_☉ (SE) and 7 × 10⁻³d²_10.6 M_☉ (SW) for the 140 K dust component, and 0.9d^5/2_10.6 M_☉ (SE) and 1.1d^5/2_10.6 M_☉ (SW) for the 55 K component. Since the sputtering amount for the hot dust component is not constrained by our model, the calculated dust masses assume no sputtering and should be treated as upper limits. The best-fit dust mass values that include sputtering are listed in Table 4. The estimate of the mass of the cooler dust component is highly uncertain since we assume that the diffuse IR emission all comes from dust with a temperature of 55 K, even though this may be true only for the region where we have an IR spectrum. A lower (higher) dust temperature would reduce (increase) the mass associated with this component.

We only estimate the dust-to-gas mass ratio for the 140 K dust since the evidence suggests that it is associated with the hot X-ray plasma. The estimates of the dust-to-gas mass ratios are listed in Table 4. Assuming that the warm dust component is associated with the hotter X-ray plasma, we calculate a dust-to-gas mass ratio of 4 × 10⁻³d^−1/2_10.6 for the SE and SW rims. These values are consistent with the dust-to-gas mass ratio of 0.6^+0.4_−0.2% that is typically assumed for the Galaxy (Sodroski et al. 1994), especially considering the uncertainties involved in determining the gas and dust masses and the distance to Kes 75. If dust sputtering has occurred in the shell, the dust-to-gas mass ratio would be a few times lower. Since the sputtering amount is not constrained by our spectrum, we cannot conclude whether any significant dust destruction has occurred in the shell.

Morton et al. (2007) attributed the low-temperature X-ray thermal component to the forward-shocked circumstellar medium (CSM)/ISM, and the high-temperature component to the reverse-shocked ejecta. Our modeling of the IR spectrum implies that the hotter temperature plasma is required to collisionally heat the dust in the shell to the observed temperature of ∼140 K. This association is also implied by the similarities in the morphology between the IR emission and the hotter X-ray gas. If the hot X-ray component does indeed originate from the reverse-shocked ejecta, then the warm dust in the shell may be coming from ejecta dust produced in the SN explosion. The total mass of this dust is approximately 1.3 × 10⁻²d²_10.6 M_☉, not unreasonable for such an interpretation. However, the shape of the IR spectrum shows that the composition of the dust is consistent with that of the ISM, and it does not show any features typical of ejecta dust in other SNRs observed with Spitzer, such as Cas A and G54.1+0.3 (Rho et al. 2008; Temim et al. 2010). Since we see no clear evidence for enhanced abundances in the X-ray spectra, it is possible that the range of different plasma temperatures and densities are due to the interaction of the SNR blast wave with the expansion into a non-uniform ambient medium or clumpy CSM. In this case, the observed dust emission originates from the swept-up CSM/ISM dust. The emission from the inner SNR in this case would be projected emission from the forward-shocked material along the line of sight.

The modeling of the IR spectra from the inner region of the SNR implies that a gas component with a temperature of less than 0.2 keV needs to be present to heat the dust to a temperature of 55 K. This is consistent with the temperature of the second thermal component allowed by the X-ray fits in Table 2, thus favoring the two-component thermal model for the X-ray spectrum. The dust temperature is too warm to be radiatively heated by the PWN, so collisional heating of the dust by the forward-shocked ambient gas seems to be the most plausible explanation. For a typical Galactic dust-to-gas mass ratio, the rough estimate of the cooler dust mass of ∼2d²_10.6 M_☉ implies that the associated gas mass would need to be very high, on the order of ∼280d²_10.6 M_☉. However, if we assume a lower value for the distance, or a higher average dust temperature for the diffuse IR emission, this would significantly reduce the mass estimate of this gas component. The X-ray data do not exclude the possibility of a more massive cooler gas component because its temperature may be too cool to be detected in X-rays due to the high absorbing column density toward Kes 75. There is also a possibility that the more diffuse IR emission from the cooler dust is not associated with the remnant even though it appears to be morphologically contained within the SNR shell.