SPITZER IRS OBSERVATIONS OF THE XA REGION IN THE CYGNUS LOOP SUPERNOVA REMNANT^*

Figure 4 shows the 2D spectrum of the slit lying across the edge shock, in the wavelength region 4900–5050 Å, which includes the [O iii] doublet. The spatial axis is vertical, with east at the bottom and west at the top. The location of the edge shock is between the pair of thin solid lines across the center of the image. The thick solid line about 3/4 the way up demarcates the boundary of the bright radiative filament, about 40'' west of the edge shock. Faint [O iii] emission extends beyond the location of the edge shock. A few features corresponding to places where the slit crosses shock fronts are clearly seen. The edge shock is easily identified, both because it is in the middle of the slit, and from its distance to the bright radiative shock, which is consistent with the separation seen in the [O iii] image (Figure 1).

Zoom In Zoom Out Reset image size

Spectra were extracted over eight spatial pixels (912) corresponding to the edge shock position and an adjacent region of the same width lying beyond the edge. These are shown by pairs of solid and dashed lines, respectively, on the image of the 2D spectrum. The extracted, flux-calibrated spectra are shown in Figure 5. The top plot shows the region around the [O iii] lines, and the excess flux in the edge shock position is clearly seen. The Hα and [N ii] are shown in the bottom plot—they are detected in both edge and background spectra and of equal strengths. (The [S ii] lines, not shown, are likewise detected at both positions with equal strengths.) We therefore ascribe the excess [O iii] flux in the edge shock position to emission from the shock itself and the remaining as part of the background.

Zoom In Zoom Out Reset image size

We obtained the fluxes in the [O iii] λ5007 line in each of the spectra by fitting Gaussians and by integrating under the lines and subtracting the backgrounds. The methods yielded results within 3% of each other. We also found that the strength of the weaker line, [O iii] λ4959 was 1/3 of the stronger line as expected.

The [O iii] λ5007 fluxes are 1.84 × 10⁻¹⁴ erg s⁻¹ cm⁻² for the edge position and 1.52 × 10⁻¹⁴ erg s⁻¹ cm⁻² for the background. The flux due to the edge shock, which we take to be the difference between the two, is 3.2 × 10⁻¹⁵ erg s⁻¹ cm⁻². Assuming that the emission fills the 2'' wide slit and noting the angular region over which the spectrum was extracted, the observed [O iii] λ5007 surface brightness of the edge shock is 1.75 × 10⁻¹⁶ erg s⁻¹ cm⁻² arcsec⁻².

Figure 6 shows the 2D spectrum of the cusp between 6520 Å and 6750 Å, which includes [N ii], Hα, and [S ii] lines (labeled on the image). The spatial axis is vertical with east at the bottom and west at the top. The region between the dashed lines in the figure is approximately that overlapping the IRS apertures. We extracted the 1D spectrum from that region, 1026 wide, and measured the line fluxes, which are reported in Table 2. Selected regions of the spectrum, illustrating the range of line strengths measured, are plotted in Figure 7. The statistical errors are ⩽2% for fluxes ≳ 100 × 10⁻¹⁷ erg s⁻¹ cm⁻² arcsec⁻².

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The Balmer decrement and the [O ii] I₇₃₃₀/I₃₇₂₇ ratio from the optical spectrum of the Cusp indicate that E_{B − V} ∼ 0.16. The usual reddening assumed for the Cygnus Loop is E_{B − V} = 0.08, but Fesen et al. (1982) have measured a range of Balmer decrements in optical spectra; they suggest an additional 0.05–0.10 mag of reddening toward certain filaments. In Table 2, we present the reddening corrected fluxes for E_{B − V} = 0.16.

In the cusp spectrum, the ratio between the components of the [S ii] doublet, F₆₇₁₆/F₆₇₃₁ = 1.24 ± 0.03. For temperatures in the range 5000–20,000 K, this implies electron densities in the range 100–250 cm⁻³. (The calculations reported in this section were done using PyNeb⁸ Luridana et al. (2012).)

[N ii] λ5755 is detected in our cusp spectrum at a level ⩾3σ (Figure 7). The temperature sensitive ratio I_[5755]/I_{[6548 + 6584]} equals (10 ± 3)/429 (Table 2), which implies that the temperature is about 14, 200 ± 2300 K, consistent with the values used in the density calculation above. The temperature-sensitive ratio [O iii] I_[4363]/I_{[5007 + 4949]} is 0.08, which indicates that the [O iii] emission arises at about 80,000 K, a much higher temperature than the typical average value of 25,000 K found in a complete radiative shock. That implies that the shock is incomplete so that the swept-up column density is small enough that radiative cooling (Raymond et al. 1988) has only brought the temperature the gas shocked earliest to 7000–8000 K, and recombination is far from complete. This provides a constraint on the models, discussed below.

Figure 6 shows that emission is detected across the entire slit, and variations in the line intensities are evident. In order to check how sensitive our results were to the exact region of extraction, we compared the cusp spectrum with the spectrum of the brighter knot lying adjacent and to the east of the cusp. The lines are brighter by a factor of ∼1.4, but the relative fluxes are approximately the same. In particular, the [S ii], [N ii] and [O iii] density and temperature sensitive line-ratios are consistent within the error bars. Also, F_[S II]/F_Hα ≈ 0.8 in both spectra.

A far-ultraviolet spectrum of the edge shock, obtained using FUSE, was analyzed by Sankrit et al. (2007). They found that the spectrum could be produced by a 180 km s⁻¹ shock and a swept-up column density 1.66 × 10¹⁸ cm⁻². At this stage of shock-completeness (swept-up column density at a given shock speed), the Ne^{4 +} zone is nearly complete, while the O^{3 +} zone is about 85% complete, and O^{2 +} zone is just starting to develop. Thus, the [O iii] λ5007 to [O iv] ratio, which is independent of the oxygen abundance, is very sensitive to the swept-up column, while the [O iv] to [Ne v] flux ratio is less sensitive.

From Sections 4.1 and 3, the observed [O iii] to [O iv] flux ratio is 1.25. Corrected for interstellar extinction, using E_{B − V} = 0.08 toward the Cygnus Loop (Parker 1967), R_V = 3.1, and the prescription of Cardelli et al. (1989) implemented in the IDL astronomy library of routines, the ratio is ≈1.6. (Note that we have used E_{B − V} = 0.16 for the cusp region—this difference will be discussed further in Section 5.2 below.) The edge shock does not fill the LH aperture, as is evident from the [Ne v] image shown in Figure 1, and so the reported surface brightness is likely to be an underestimate. If we assume that the filling fraction is 0.5, then the resulting [O iii] to [O iv] would be ≈0.8.

In Figure 8, the [O iii] to [O iv] flux ratio is plotted as a function of swept-up column for a 180 km s⁻¹ shock. The dashed lines mark the two points bracketing the observed value—0.8 and 1.6. At these points, for O = 8.70 and Ne = 8.09, the values of the [O iv] to [Ne v] flux ratio are 1.6 and 2.2, respectively. For a given shock velocity and for a particular swept-up column this flux ratio is directly proportional to the abundance ratio, O/Ne. In the model, O/Ne = 4.1, and the observed value of [O iv] to [Ne v] is 3.4. Scaling the model to match the observed values at the two points, we find that the required values of O/Ne are 8.9 and 6.3 for the lower and higher swept-up columns.

Zoom In Zoom Out Reset image size

The [Ne v] emissivity is sensitive to the shock velocity around the 180 km s⁻¹ region, with faster shocks producing brighter [Ne v] emission. The shock velocity is well constrained to be 180 km s⁻¹ by the Ne vi] to Ne v] ratio in the FUV spectrum (Sankrit et al. 2007). The observed Ne vi] emission would not be produced by slower shocks. Therefore, we performed the calculation described above only for a faster, 190 km s⁻¹, shock and found that O/Ne in the range of 9.1–13.2 are required.

Our choice of the oxygen to neon abundance ratio for the original model was based on its success in reproducing the FUV line strengths (Sankrit et al. 2007). However, updates to the atomic rates (see Section 5, above) have resulted in a revision of the ratio of abundances. Furthermore, the forbidden optical and IR lines are not susceptible to uncertainties due to resonance scattering, and therefore the results presented here are more robust than those from the earlier study. The abundance ratio O/Ne = 6.3 is close to the median value of 5.89 found for a sample of 10 Galactic H ii regions (Shaver et al. 1983). The maximum value of the ratio was 8.9 in the sample. Thus, the O/Ne we find for the 180 km s⁻¹ shock are in the normal range for H ii regions, while the high values found for the 190 km s⁻¹ shock are implausible.

To compare the observations of the cusp with theoretical models, we must apply a reddening correction and scale the fluxes from the different apertures to account for the variation in surface brightness within the larger apertures. The value for E_{B − V} obtained from the optical spectrum differs from the usual value for the Cygnus Loop (Section 4.2). As a check, we extracted the flux in the [Ne v] 3425 Å line in the region of the Spitzer SH aperture from the image presented by Szentgyorgyi et al. (2000) and compared it with the [Ne v] 14.33 μm flux; the ratio agrees with the theoretical value and E_{B − V} = 0.08.

The difference in the values of E_{B − V} derived from the Balmer decrement and the [O ii] lines on the one hand, and from the [Ne v] optical and IR lines on the other is consistent with the idea that on average the Balmer lines and low-ionization lines arise from deeper within the interaction region and the high-ionization lines from the outer, less extincted regions. The reddening, E_{B − V} = 0.08 is the normal foreground, and the excess is local. An extended component of high ionization emission has been observed in the XA region. FUSE spectra presented by Sankrit et al. (2007) shows the presence of widespread O vi λλ1032, 1038 emission and strong O vi emission, comparable in strength to the O iv] λ1400 line, was observed in the area by HUT (Danforth et al. 2001). These provide some evidence for a faster shock, or for a transition layer where cool gas evaporates to join the X-ray emitting plasma that envelopes the optical filaments. However, the 56'' HUT aperture encompasses a lot of material outside the region observed by the other instruments considered here, so there is no compelling evidence for a faster shock contributing specifically to the spectra we have observed. An alternative explanation of the discrepancy is that we have underestimated the error in the surface brightness measured from the narrowband image, both due to the calibration and due to the placement of the SH aperture FOV on the image, and the result is consistent with the higher value of reddening.

Faced with these possibly contradictory results, we calculated the corrected spectrum for both values of reddening, using the Cardelli et al. (1989) extinction function with R_V = 3.1. After applying the reddening correction, we accounted for the large variations of intensity within the different apertures by scaling the fluxes to match theoretical intensity ratios. Ideally, intensity ratios that depend only on the atomic rates and are, therefore, insensitive to the model should be used. The ratio of the two IR lines of [Ne v] is one such ratio. To scale the LH aperture to the SH, we use the [Ne v] 14.33 μm to 24.32 μm ratio = 0.82. This means multiplying the LH values by a factor of 1.90, which is plausible because the average flux in the LH aperture is presumably lower than in the SH aperture, which was placed on the intensity peak as judged by the optical images. This scaling greatly improves the agreement of the [S iii] 18.71 μm to 33.47 μm ratio. (We use the [Ne v] ratio rather then the [S iii] ratio, because the latter depends on density in the range indicated by the [S ii] optical lines.)

For the optical and UV fluxes, the only ratios available have some temperature sensitivity, so the scaling is less certain. To scale the optical flux to the SH flux, we use the [Ne iii] λλ3870,3968 to 15.56 μm ratio of 3.87. Since that depends on the reddening correction, the scalings differ for the two reddenings chosen; 1.0 for E_{B − V} = 0.08 and 0.70 for E_{B − V} = 0.16. Next, we scale the UV fluxes by using the O iii] 1665 Å to [O iii] 5007 Å ratio of 0.52. That results in multiplying the UV fluxes by 1.90 for E_{B − V} = 0.08 and 1.0 for E_{B − V} = 0.16. The reddening-corrected [Ne v] λ3425 intensities were not scaled, because the flux was extracted from the region corresponding to the SH aperture. The uncertainty in the placement of the aperture on the image gives a 7% uncertainty, and Szentgyorgyi et al. (2000) quote a photometric uncertainty of 30%. Finally, we normalized the relative fluxes to Hβ = 100, for comparison with the models.

The scaling for the different apertures largely cancels out the difference in reddening correction so that for the two values of E_{B − V}, the scaled, normalized relative fluxes differed by less than 20% for all lines except [Ne v] λ3425. In Table 2, along with the observed fluxes, we present the reddening corrected fluxes only for the case of E_{B − V} = 0.16 because of its better agreement with the model Balmer decrement.

5.2.1. Shock Models

5.2.2. Comparing the IR Spectra with Shock Models

The parameters in the shock model have been constrained largely by the UV and optical spectra. If we disregard the lines strongly affected by resonance scattering, and [Ne v] λ3425, for which the measurement and scaling is discrepant compared to the other lines, then the shock model predictions match the observations to within 50%, except for N iii] λ1750 and [Ar iii] λ7138, and in most cases to within 25%. Reducing the rather uncertain abundance of Ar by a factor of about two would bring the [Ar iii] predictions into agreement with the observed value, without affecting any other line strengths.

In Figure 9, the ratio of the flux predicted by model M150 to the observed, dereddened value is plotted against wavelength for the lines listed in Table 2. The top, middle, and bottom plots show the UV, optical and IR lines, respectively. It should be noted that no distinction is made between strong and weak lines in the plots. So, for instance, the match between models and observation for the weak [Fe iii] lines in the IR spectrum is actually much better than the plot seems to imply, as can be seen by examining Table 2.

Zoom In Zoom Out Reset image size

We now turn to an examination of the IR lines, focusing on models M145, M150, and M155a, since the other two (M155b and M155c) do not predict the observed optical [S ii] doublet ratio. The [Ne v] lines are most sensitive to the shock velocity and are best matched by the 155 km s⁻¹ shock. All models predict [O iv] fluxes that are higher than the observation—a factor of 1.3 for the 145 km s⁻¹ shock and 2.0 for the 155 km s⁻¹ shock. The models reproduce the correct [S iii] line fluxes, with the 150 km s⁻¹ model providing the best match. In contrast, the models consistently predict only about half the observed [S iv] flux.

The [Si ii] fluxes predicted by the model are significantly higher than the observed value. Reducing the Si abundance by a factor of about 2.5, as expected from depletion onto grains, would bring them into agreement. However, the line fluxes predicted by the models are also sensitive to the column density cutoff, since the [Si ii] emission would extend far into the cooling zone.

The ratios among the [Fe ii] lines predicted by the models are in agreement with the observed values, but the absolute fluxes are higher. As with silicon, the match between models and observation would be improved for the [Fe ii] lines with Fe abundances reduced by a factor of about 2. The models correctly predict the [Fe iii] line fluxes, and therefore, there is too low a ratio between [Fe iii] and [Fe ii] line strengths. The ratio is sensitive to the photoionization rate of Fe⁺, the recombination rate of Fe⁺⁺, and especially to the charge transfer rates, which we have taken from Neufeld & Dalgarno (1987) as well as to the cutoff column density.

The models assumed solar abundances for Si and Fe, while those elements should be almost entirely depleted onto dust grains in the ISM. Reductions by factors of 2.5 and 2, respectively, still imply that about 40%–50% of these elements have been returned to the gas phase. The ratios of the UV lines of C iii] and Si iii] to the O iii] line also indicate only modest depletion. This efficient liberation of refractory elements from grains is a little surprising, since only about 20%–40% of the silicon and carbon are returned to the gas phase at a swept-up column of 2 × 10¹⁸ cm⁻² in the faster X-ray producing shock in the NE Cygnus Loop studied by Raymond et al. (2013). The sputtering rates are greatly reduced at the lower temperatures in 150 km s⁻¹ shocks, but grain–grain collisions shatter the dust particles, producing small grains that are more vulnerable to sputtering. Jones et al. (1996) find that 10%–30% of the mass of graphite and silicates in grains is returned to the gas phase for shock speeds below 150 km s⁻¹, so the spectra imply a higher dust destruction efficiency. However, the behavior of grains in high speed collisions is uncertain.

The pair of lines [Ne ii] and [Ne iii] present an interesting puzzle, which may have some important consequences for our understanding of the XA region. Separately considered, the model predictions are in fairly good agreement with the observations. However, the [Ne iii] to [Ne ii] flux ratio is always less than 1. To match the predicted flux ratio of about 1.3, the models would need to be truncated at unacceptably low values of swept-up column density. The observed flux ratio can be explained if an additional source of photoionization is invoked. The extra photoionization would also be helpful in explaining the strength of the [S iv] line and possibly the [Fe iii] to [Fe ii] flux ratio as well.

5.2.3. Shock Driving Pressures

5.2.4. Model Limitations

The set of IR lines detected in the cusp spectra are useful diagnostics of SNR shocks. This is particularly true of the lines from the singly and doubly ionized species, [Ne ii], [Si ii], [Fe ii], [Ne iii], [S iii], and [Fe iii], all of which can be produced in shocks with speeds as low as 80 km s⁻¹ (Hartigan et al. 1987). We compare the IR line fluxes from the cusp region with those measured in other SNRs. In Table 4, we have assembled measurements from the published literature. Our sample consists of Infrared Space Observatory spectra of IC443 (Oliva et al. 1999a) and RCW 103 (Oliva et al. 1999b), and Spitzer spectra of W44, W28, IC443C, and 3C391 (Neufeld et al. 2007) and Kes 69, 3C396, Kes 17, G346.6-0.2, G348.5-0.0, and G349.7+0.2 (Hewitt et al. 2009).

The strongest lines in all the remnants are [Si ii] 34.80 μm and [Ne ii] 12.81 μm. In Table 4, we have arranged the remnants in order of increasing surface brightness of the sum of these two lines. According to this measure, the Cygnus Loop XA region is the faintest of the sample. The remaining objects fall into three groups: at the faint end IC443C and G346.6-0.2 are about three times as bright as XA, the brightest remnants, RCW 103 and G349.7+0.2 are 70 and 180 times as bright as XA, respectively, and the remaining eight objects are between 7 and 30 times as bright.

The ratio of [Ne iii] 15.56 μm to [Ne ii] 12.81 μm in the XA region is 1.3, which is the highest value among all the remnants. The ratio is 1.2 in 3C396, 0.9 in RCW 103, and lies between 0.1 and 0.5 for all other remnants. Shocks with velocities in the 80–120 km s⁻¹ range produce [Ne iii] to [Ne ii] ratios between 0.1 and 0.3 when the pre-shock medium is in photoionization equilibrium, and between 0.5 and 0.9 when it is completely ionized (Hartigan et al. 1987). We have explained the higher than predicted ratio in the cusp spectrum by invoking an incomplete, ∼150 km s⁻¹ shock. However, an alternative explanation is that faster shocks in the vicinity as well as the hot X-ray emitting plasma provide photoionizing radiation that increases the [Ne iii] to [Ne ii] flux ratio above the values expected for only shock excitation.

The [Fe iii] 22.92 μm to [Fe ii] 17.93 μm ratio is another potential diagnostic of the ionization state, though it can also depend on the density. (We do not use the [Fe ii] 25.98 μm flux in the comparison since for the other remnants this line is not well resolved from the [O iv] 25.88 μm line.) The ratio of the intensities in the dereddened and scaled cusp spectrum is 1.60, and in the uncorrected observed spectrum it is 0.91 (Table 2) The [Fe iii] line is detected in four other remnants, RCW 103, W44, W28, and 3C391, and an upper limit is presented for IC443C. The [Fe iii] to [Fe ii] flux ratio ranges between 0.16 and 0.25, significantly lower than in the Cusp and consistent with recombined shocks.

The [Ne iii] and [S iii] zones in the post-shock flow are nearly co-extensive, and therefore, the relative abundances of Ne and S influence the flux ratio of [Ne iii] 15.56 μm and [S iii] 18.71 μm. The ionization potential of S⁺⁺ (34.83 eV) is lower than that of Ne⁺ (40.96 eV), and therefore, photoionization can also play a role in determining the flux ratio of [Ne iii] to [S iii]. This trend is detected in the sample of remnants. For the three objects with the lowest [Ne iii] to [Ne ii] ratios, i.e., <0.16, the [Ne iii] to [S iii] ratios lie between 0.5 and 1.4. For all the remaining objects, except one, where the [Ne iii] to [Ne ii] ratio span the range 0.2–1.3, the [Ne iii] to [S iii] ratio lies between 2.4 and 4.4. The exception is Kes 69, where the ratio is 8.2, indicating a higher Ne to S abundance in this remnant compared with the others.

The [S iv] line is detected in the XA region and in RCW 103 and in none of the other remnants. In RCW 103, the ratio [S iv] 10.51 μm to [S iii] 18.71 μm is about 0.06, which is much lower than for the XA region, where the ratio is about 0.4. Overall, we conclude that the other spectra listed in Table 4 pertain to slower shocks than the one at the Cusp probably because the bright regions chosen for observation are shocks in denser clouds.