SHOCK–CLOUD INTERACTION AND PARTICLE ACCELERATION IN THE SOUTHWESTERN LIMB OF SN 1006

Figure 1 shows the EPIC map of the southwestern quadrant of SN 1006 in the 0.3–2 keV band. We notice a clear indentation in the shock front, corresponding to regions G, H, and I, preceded by a faint "bulge" (red dashed region).

Zoom In Zoom Out Reset image size

D02 carried out an extensive study of the atomic and molecular gas in the environs of SN 1006, concluding that on a large scale the remnant evolves in a very smooth, tenuous environment, as expected at high Galactic latitudes. On a more detailed scale, it can be noted that the presence of two H i structures match the peculiar features observed along the periphery of SN 1006. One of these structures is an elongated cloud abutting the flattened northwestern border of the SNR (i.e., the northwestern cloud), and the other is a cloud located to the southwest of SN 1006 (the southwestern cloud), whose border remarkably corresponds to the concavity of the shock front shown in Figure 1. Both of these features peak at v_LSR ∼ +10 km s⁻¹ and are clearly visible in Figure 3(b) of D02. By assuming that the neutral gas is optically thin (as D02 did), we infer the atomic column density. Figure 1 shows the contours of the column density estimated in the [ + 5.8, +10.7] km s⁻¹ velocity range. In spite of the excellent morphological agreement, D02 had proposed that another H i concentration at v_LSR ∼ −5 km s⁻¹ was a better candidate, although neither the gas at −5 km s⁻¹ nor the one at +10 km s⁻¹ had the expected systemic velocity for Galactic gas placed at ∼2 kpc. Revisiting the same data, it is now evident that the northwestern and southwestern clouds coincide with the northwestern and southwestern lobes of SN 1006. In this picture, the elongated, flat structure of the northwestern cloud naturally appears as the origin of the sharp H_α filaments observed all along the northwestern flank, thus this gas must be in the local environment of SN 1006.

As noted before, these structures have anomalous kinematical velocities. However, a velocity deviation from a Galactic circular rotation model is not surprising for gas placed almost 600 pc above the Galactic plane where the connection with the disk rotation is very unlikely. Also, in this part of the Milky Way, it is known that the disk bends toward more negative Galactic latitudes (Burton 1976), increasing the z distance of these clouds from the Galactic arms. Moreover, by supposing that these structures follow the circular rotation model, we obtain a very unlikely outcome since they would be giant clouds (larger than ∼100 pc), placed as far as ∼15 kpc in the halo of the Milky Way (∼3.7 kpc above the Galactic plane), in perfect agreement with different portions of the border of SN 1006 only by chance. On the other hand, if we consider the hypothesis of cloud disruption as a consequence of the SNR expansion, a departure of ∼40 km s⁻¹ from the systemic velocity would be naturally explained. The X-ray spectral analysis described in the next section will further confirm that the southwestern cloud cannot be located farther than SN 1006. By assuming that the southwestern cloud is at the same distance as SN 1006 (2.2 kpc), we estimate an atomic gas density of about 9–16 cm⁻³, by assuming an extension of the cloud along the line of sight of 85–155 (i.e., the minimum–maximum angular extension in the plane of the sky).

We perform a spatially resolved spectral analysis of the X-ray emission originating from regions A to L of Figure 1 to search for variations along the rim of column density and synchrotron cutoff energy hν_cut. We model the nonthermal emission with the loss-limited model developed by Zirakashvili & Aharonian (2007; highly suited for SN 1006, as shown by Miceli et al. 2013a, 2013b). Spectral analysis was performed in the 0.3–7.5 keV band by using XSPEC V12.8.

Since we are interested in measuring N_H (the interstellar absorption is modeled through the TBABS model), it is important to correctly describe the soft thermal emission (mainly associated with O vii and O viii line complexes). To model the thermal emission, we included a VPSHOCK component (Borkowski et al. 2001), with the same abundances as the ejecta component in region e of Miceli et al. (2012). We first focused on region A of Figure 1, where the thermal emission is the highest, and determined the best-fit values of the temperature, $kT=0.6^{+0.2}_{-0.1}$ keV, and ionization timescale, $\tau =1.5^{+1.0}_{-0.6}\times 10^{10}$ s cm⁻³. We then fixed these values when fitting the other spectra, and let the plasma emission measure as a free parameter. We refer to this procedure as approach 1. We verified that we do not find pronounced variations in the best-fit parameters, even if we model the thermal line emission with two narrow Gaussians (hereafter approach 2; see Miceli et al. 2013b). In particular, the values of N_H (and hν_cut) derived with the two approaches follow the same azimuthal trend and are all consistent within 2σ, except for regions A, B, C, where the N_H derived by adopting approach 2 are unrealistically low (∼5 × 10²⁰ cm⁻²).⁵ Therefore, in the following, we report only the results obtained by adopting approach 1. All the spectra are very well described by our model and the reduced χ² ranges between 1.00 and 1.10 (with 1100–2000 degrees of freedom (dof)).

Figure 2 shows the azimuthal variations of the best-fit values of N_H for regions A–L. We observe a clear increase in the absorbing column in regions G–L, corresponding to the location of the southwestern cloud (see Figure 1). In these regions, N_H is significantly higher than the characteristic value of ∼7 × 10²⁰ cm⁻², estimated by D02 integrating the H i radio emission in the [0, −20] km s⁻¹ range (see the dashed blue curve in Figure 2, derived by sampling the N_H values inferred from the H i observations). If we also include the contribution of the southwestern cloud (derived from H i observations in the [ + 5.8, +10.7] km s⁻¹ velocity range) to the column density, we find that the azimuthal profile of N_H obtained from the radio observation is in amazing agreement with that derived using X-ray spectral analysis. This result further confirms that the southwestern cloud lies at a distance compatible with that of SN 1006 (and the foreground), and not at 15 kpc as estimated from a blindly applied circular rotation model.

Zoom In Zoom Out Reset image size

The morphology of the southwestern cloud, perfectly corresponding to the concave shape of the southwestern shock front, strongly suggests that the back side of the cloud actually interacts with SN 1006. If this is the case, we expect the shock to be slowed down by this interaction. A direct consequence would be a net decrease in the cutoff frequency of the synchrotron emission in the interaction region because in SN 1006 the maximum electron energy is limited by radiative losses and in the loss-limited scenario hν_cut depends on $v_{s}^{2}$ (Zirakashvili & Aharonian 2007). Figure 3 shows the azimuthal profile of the best fit values of hν_cut in regions A–L. Indeed, we find that the cutoff frequency abruptly drops down in regions G, H, I, i.e., exactly where the shock front is indented. Some hints for the presence of this dip in the azimuthal profile of the cutoff energy can be found in Rothenflug et al. (2004) and Miceli et al. (2009), though in these papers the large regions selected for the spatially resolved spectral analysis prevent a significant detection like the one presented here. We conclude that the bulk of the southwestern cloud lies in the foreground (as shown by the N_H variations in Figure 2) and part of it (i.e., regions G, H, I) is physically interacting with SN 1006.

Zoom In Zoom Out Reset image size

We searched for thermal emission from the interstellar medium (ISM) in region H (where the ambient density is expected to be at its maximum), but the spectrum is synchrotron-dominated therein and very well described by our model (reduced χ² = 1.0 with 1203 dof), so an additional thermal component is not statistically needed and the ISM temperature, kT_ISM, is unconstrained. However, we derived some temperature-dependent upper limits for the ISM post-shock density (deduced from the emission measure as in Miceli et al. 2012). We found n_ISM < (12, 3, 0.3) cm⁻³ for kT_ISM = (0.05, 0.25, 2) keV, respectively. The values obtained for low temperatures (expected in the case of propagation in a dense environment) are in agreement with the shock–cloud interaction scenario and similar to those derived from the H i analysis. The shock is probably interacting with the outer border of the cloud (where we expect lower densities) since the amount of the indentation indicates that the interaction is recent and the shock has not yet entered into the bulk of the cloud. We notice that the H_α emission in the southwestern limb is quite faint and noisy (Winkler et al. 2003) and the lack of a clear H_α filament can be due to a heavily ionized preshock medium and/or an unfavorable (not edge-on) orientation of the emitting sheet.

We obtain a heuristic estimate of the value that hν_cut would have had in the case of no interaction by performing a fourth degree polynomial fit to all the points but G, H, I (the best fit curve is shown in Figure 3). We verified that this simple (unphysical) polynomial form also provides a good description of the hν_cut profile in the northeastern limb. The cutoff frequency in region H is reduced by a factor of f ∼ 1.7. Since $h\nu _{{\rm cut}}\propto v_{s}^2\propto n_{{\rm ISM}}^{-1}$, one may suppose that in region H the ambient density is higher by the same factor f and therefore is slightly lower than 0.1 cm⁻³ (assuming n_ISM ∼ 0.03–0.05 cm⁻³ elsewhere). This value is much smaller than the average density of the southwestern cloud estimated by the H i data and seems insufficient to produce the observed indentation in the shock. However, we notice that other (non-interacting) parts of the shell, whose projected location falls within region H, may contribute to the observed synchrotron emission, thus producing a biased overestimation of hν_cut (relatively high hν_cut have been observed well inside the shell by Rothenflug et al. 2004) and an underestimation of n_ISM. In any case, in the concave region of the southwestern shock front, hν_cut is significantly lower than in the adjacent region, in agreement with what we expect in the case of shock–cloud interaction.

The red dashed region in Figure 1 also shows low values of hν_cut and high N_H. It may be associated with a non-interacting region located behind the cloud, or to particles leaving the remnant and diffusing inside the cloud. Further investigation is necessary to ascertain the origin of this "bulge."