SEARCHING FOR DIFFUSE NONTHERMAL X-RAYS FROM THE SUPERBUBBLES N11 AND N51D IN THE LARGE MAGELLANIC CLOUD

N11 (DEM L 34), located at the northwest edge of the LMC (Henize 1956), contains four OB associations, LH9, LH10, LH13, and LH14 (Lucke & Hodge 1970). The Hα emission exhibits a complex structure consisting of several shells and filaments (e.g., Mac Low et al. 1998). The largest Hα shell with a diameter of ∼120 pcis situated around the central cluster LH9. The age of LH9 is estimated to be ∼3.5 Myr (Walborn et al. 1999), the oldest among the four clusters. It is suggested that the stellar activity of LH9 triggered a starburst at the periphery, leading to the formation of the other OB associations (e.g., Walborn & Parker 1992; Hatano et al. 2006).

N51D (DEM L 192) is located at the southeast corner of the supergiant shell LMC4. Two OB associations, LH51 and LH54 (Lucke & Hodge 1970), are surrounded by a filamentary SB shell with a size of about 140 pc× 120 pc. Both the OB clusters are estimated to have an age of ∼3 Myr (Oey & Smedley 1998).

The SBs N11 and N51D were observed with the Suzaku X-ray imaging spectrometer (XIS: Koyama et al. 2007) on 2006 November 7 and 2009 May 17, respectively. The observation IDs and aim points are summarized in Table 1. The XIS consists of four X-ray charge-coupled devices (CCDs). Three of them (XIS0, XIS2, and XIS3) are front-illuminated (FI) and the other (XIS1) is back-illuminated (BI). XIS0, 2, and 3 have a lower and more stable background level for extended emission, while the latter has superior sensitivity in the soft (≲1.0 keV) X-ray band with a significantly improved energy resolution compared with previous X-ray imaging sensors. Combined with X-Ray Telescopes (XRTs; Serlemitsos et al. 2007), the field of view (FOV) of the XIS covers a ∼18'×18' region with a half-power diameter (HPD) of ∼2'. During both the observations, the XIS was operated in the normal full-frame clocking mode with the spaced-row charge injection technique. However, the XIS2 was out of operation during the observation of N51D due to damage, possibly caused by the impact of a micrometeorite.

We reduced the data using the standard tools of HEADAS version 6.7. We first employed the revision 2.0 and 2.4 data products for N11 and N51D, respectively. Both were reprocessed using the xispi software and the makepi files released on 2009 August 13. These files include the latest calibration results for the charge transfer inefficiency and gain. We cleaned the reprocessed data in accordance with the standard screening criteria⁵ and removed flickering pixels. We also excluded time intervals that suffered from telemetry saturation. After the filtering, the effective exposure times were obtained to be ∼31 ks and ∼86 ks for the observations of N11 and N51D, respectively. Only grade 0, 2, 3, 4, and 6 events were used in the following analysis.

Detailed information on the XMM-Newton observations of both SBs is given in Table 1. The European Photon Imaging Camera (EPIC), aboard XMM-Newton, has two metal oxide semiconductor (MOS) CCD arrays (Turner et al. 2001) and one pn CCD array (Strüder et al. 2001). The latter has a higher quantum efficiency than the other CCD arrays. All three cameras have a larger FOV (∼30' diameter circle) than the XIS with a much better HPD of ∼6'' at the optical axis. This makes observations with the EPIC highly sensitive to point-like sources. During both the observations, the MOS and pn CCD arrays were operated in prime full window mode and extended prime full window mode, respectively. Thick filters were used in the observation of N11, while thin ones were used in the observation of N51D.

All the data were processed using version 9.0.0 of the XMM Science Analysis Software (SAS) with the latest calibration files. For the MOS, we selected X-ray events with patterns 0–12, which were passed through the #XMMEA_EM filter. For the pn, only events with patterns 0–4 and flag = 0 were extracted. Light curves in the 10–15 keV band, where the effective area of the telescopes rapidly decreases, were accumulated from the whole FOV. To minimize any uncertainty due to the NXB, we removed the periods when the count rate was larger than 0.3 counts s⁻¹ (MOS) and 1.0 counts s⁻¹ (pn). The resultant exposure times are given in Table 1.

Figures 1(a) and 1(b) show XIS images of the N11 region in the soft (0.5–2.0 keV) and hard (2.0–5.0 keV) X-ray bands, respectively. Data from the four detectors were combined. The images were binned by a factor of 8 and smoothed with a Gaussian kernel of σ = 25''. Extended emission can be clearly seen in the soft-band image, while no significant diffuse structure was found above 2.0 keV.

Zoom In Zoom Out Reset image size

As already mentioned in previous works (e.g., Nazé et al. 2004), the soft emission peaks near the dense stellar cluster HD 32228 and is widely associated with the OB association LH9. This part is confined by the largest Hα shell of the N11 complex (Mac Low et al. 1998), and thus it is considered that the soft X-rays originate from hot shocked gas within the SB shell that is blown by the cluster of stars. The emission is further extended to the north of LH9, the H ii region N11B, and shells 2 and 3 (see also Figure 3 of Mac Low et al. 1998). The X-rays from N11B are probably powered by stellar winds from the central young OB stars of LH10. Shells 2 and 3 lie to the north and northwest of N11B, respectively. These X-ray emissions with relatively low surface brightnesses are confined by the faint Hα filaments.

Maddox et al. (2009) claimed that nonthermal X-rays had been detected from the region around LH9. Therefore, we concentrate on the spectral analysis of the same region, an ellipse containing the entire SB around LH9 with major and minor radii of 45 and 35, respectively—hereafter "SRC." The background (BGD) for the SRC spectra was taken from an off-source region, the rectangle shown in Figure 1, but the CCD corners were excluded to remove X-rays from the calibration sources.

A high-resolution EPIC image in the hard band is shown in Figure 1(c). This image was made by merging two MOS images binned by a factor of 32 and smoothed with a Gaussian profile of σ = 96. Similarly to the XIS image for the same energies (Figure 1(b)), no extended emission was observed in the EPIC FOV. Point-source extraction was performed using the ewavelet task of SAS with a detection threshold of 5σ. The positions of the detected sources are indicated with small circles in the figure. Three discrete sources are located in the SRC region, while one is located in the BGD region. The latter, at (R.A., decl.) = (4^h57^m28^s, −66°22'40''), is known to be associated with the cluster HD 268743, or Sk–66°41 (Nazé et al. 2004). Since this source is also clearly visible in the XIS image, we excluded it from the BGD region for spectral analysis.

Here, we analyze the EPIC data of the resolved point sources to take into account their contribution to the hard X-ray flux from the SRC region. Figure 2 shows the BGD-subtracted EPIC spectra, where the three point sources in the SRC region are merged to improve the statistics. The spectrum of each point source was extracted from the circular region with a radius of 15'', while the background was taken from an annulus with inner and outer radii of 15'' and 60'', respectively.

Zoom In Zoom Out Reset image size

The spectra were phenomenologically fitted with an absorbed power-law model. The absorption cross sections were taken from Morrison & McCammon (1983). The elemental abundances were assumed to be the values in Anders & Grevesse (1989). The fit was acceptable (χ²/dof = 15/19) with N_H = 2.9 (1.1–7.8) × 10²¹ cm⁻² and Γ = 1.5 (1.1–2.2). The observed flux in the 2–10 keV band was obtained to be 3.6 (3.1–4.1) × 10⁻¹⁴ erg cm⁻² s⁻¹.

Since the emission from the SRC region is faint, background reduction should be carefully performed to obtain an accurate result. We first subtract the NXB from the SRC and BGD raw spectra, and examine the difference in the hard X-ray fluxes between them (Section 3.3.1). Then, we subtract the other background components, such as the cosmic X-ray background (CXB) and local emission, and perform model fittings to the SRC spectrum (Section 3.3.2).

3.3.1. Search for Hard X-rays

Figure 3(a) shows the XIS0 raw spectrum of the SRC region. Instrumental emission lines of Mn-Kα, Ni-Kα and Kβ, and Au-Lα can be observed at ∼5.9 keV, ∼7.5 keV, ∼8.3 keV, and ∼9.7 keV, respectively. Using xisnxbgen software, we constructed the NXB spectrum from the night-Earth database. This was extracted from the same detector coordinates to the SRC region and sorted with the geomagnetic cutoff rigidity for reproducibility. The intensities of the Ni-Kα lines were found to be 1.6 (1.1–2.1) ×10⁻³ counts s⁻¹ and 1.6 (1.5–1.7) × 10⁻³ counts s⁻¹ for the raw SRC and NXB spectra, respectively. Since these are highly consistent with each other, we can consider that the NXB spectrum was correctly reproduced. The NXB for the BGD spectrum was also successfully constructed by the same method.

Zoom In Zoom Out Reset image size

After the NXB subtraction, the surface brightness spectra of the SRC and BGD regions were obtained as shown in Figure 3(b) (shown as circles and squares, respectively). For comparison, we also present the spectrum of another source-free region (shown as gray diamonds), taken from the observation of the north ecliptic pole (NEP). All the spectra are similar to each other at energies above ∼3 keV, suggesting that the hard X-rays from the SRC region are dominated by the CXB. Table 2 gives the 2–10 keV surface brightness of each region, determined using the data of all four XISs. No excess SRC flux was found.

Table 2. 2–10 keV Surfaces Brightnesses of N11 and NEP Regions

Region	Solid Angle	Flux (2–10 keV)	Surface Brightness (2–10 keV)
	(arcmin2)	(×10−13 erg cm−2 s−1)	(×10−15 erg cm−2 s−1 arcmin−2)
SRC	49.4	2.2 (2.0–2.4)	4.4 (4.0–4.8)
(point sources)		0.36 (0.31–0.41)	⋅⋅⋅
(subtracted)		1.8 (1.6–2.0)	3.6 (3.2–4.1)
BGD	70.4	2.8 (2.5–3.1)	4.0 (3.6–4.4)
NEP	113.1	5.0 (4.8–5.1)	4.4 (4.2–4.5)

Note. The NXB is subtracted from each region, but the CXB is still included.

Download table as:  ASCIITypeset image

3.3.2. Spectral Fitting

The spectrum of the BGD region was subtracted from the SRC spectrum. However, since the vignetting of the Suzaku XRT significantly decreases the effective area at large off-axis positions (Serlemitsos et al. 2007), the BGD spectrum I_B(E) was corrected before the subtraction as

$$\begin{equation} I_{\rm B}^{\prime }(E) = I_{\rm B}(E) \cdot \frac{\Omega _{\rm S}}{\Omega _{\rm B}} \cdot f(E), \end{equation} \tag{ 1 }$$

where Ω_S and Ω_B are the solid angles of the SRC and BGD regions, respectively. The correction factor f(E) is the energy-dependent effective-area ratio between the SRC and BGD regions. Using xissim software, we estimated the f(E) values for each XRT as given in Figure 4. This estimation also takes into account the contaminating material built up on the optical blocking filters of the XIS (Koyama et al. 2007). The differences in the results among the telescopes are mainly due to the scattering of the optical axes of the XRTs (see Figure 8 of Serlemitsos et al. 2007).

Zoom In Zoom Out Reset image size

The BGD-subtracted SRC spectra are shown in Figure 5. The data from three FIs were combined because their response characteristics are almost identical. No significant signal was detected in the energy band above ∼3 keV, which is consistent with the result in Section 3.3.1. K-shell emission lines of O viii (∼0.65 keV), Ne ix (∼0.91 keV), and Mg xi (∼1.34 keV) were clearly separate, indicating a thermal origin of the soft X-rays. Thus, we fitted the spectra with an optically thin thermal plasma model (APEC; Smith et al. 2001). The electron temperature (kT_e) and emission measure (n_en_pV, where n_e, n_p, and V are the electron and proton densities, and the emitting volume, respectively) were treated as free parameters. The elemental abundances relative to solar values (Anders & Grevesse 1989) were fixed to the mean LMC values of Russell & Dopita (1992), but the Mg and Si abundances were assumed to be those of Hughes et al. (1998) since the values from Russell & Dopita (1992) were more uncertain. The interstellar extinction in the Galaxy and LMC was separately considered. The Galactic absorption column density with the solar abundances was fixed to be N^G_H = 4.3 × 10²⁰ cm⁻² (Dickey & Lockman 1990). The other component (N^LMC_H) was a free parameter, with the assumption of the LMC metal abundances. In addition, the point-source model derived in Section 3.2 was given as a fixed component (with an independent absorption column). This model gave a best fit with kT_e = 0.18 (0.17–0.19) keV and χ²/dof = 144/109.

Zoom In Zoom Out Reset image size

We next allowed the abundances of O, Ne, Mg, and Fe to vary freely (by using a VAPEC model). Then, we obtained a significantly improved fit with χ²/dof = 99/105. The best-fit parameters are given in Table 3. The Fe abundance was found to be slightly lower than the mean LMC value (∼0.3 solar; Russell & Dopita 1992; Hughes et al. 1998). The observed flux of 4.9 × 10⁻¹³ erg cm⁻² s⁻¹ (in the 0.5–2.0 keV band) is comparable with the result of Nazé et al. (2004) but inconsistent with that of Maddox et al. (2009; ∼2.5 × 10⁻¹³ erg cm⁻² s⁻¹), although the same data were analyzed.

Table 3. Best-fit Spectral Parameters for N11

Parameter	Value
NGH (cm−2)a	4.3 × 1020 (fixed)
NLMCH (cm−2)a	<2.3 × 1021
kTe (keV)	0.30 (0.28–0.32)
O (solar)	0.26 (0.15–0.47)
Ne (solar)	0.39 (0.20–0.47)
Mg (solar)	0.43 (0.22–0.89)
Fe (solar)	0.12 (0.08–0.22)
nenpV (cm−3)	5.4 (2.6–19) ×1058
FobsX (erg cm−2 s−1)b	4.9 × 10−13
LX (erg s−1)b	2.2 × 1035
χ2/dof	99/105 = 0.94

Notes. ^aAbsorption column in the Galaxy and LMC, where the solar and LMC abundances are respectively assumed. ^bObserved flux and luminosity in the 0.5–2.0 keV band.

Download table as:  ASCIITypeset image

The derived electron temperature was slightly higher than the values in previous reports (∼0.2 keV; Dunne et al. 2001; Nazé et al. 2004; Maddox et al. 2009). We found, however, that the fit with a fixed kT_e of 0.2 keV also yielded an acceptable χ²/dof value of 111/106. This temperature difference does not severely affect the flux in the hard (>2 keV) X-ray band.

X-ray images of the N51D region are shown in Figure 6 for the XIS soft X-ray band (a), the XIS hard X-ray band (b), and the EPIC hard X-ray band (c). The binning factors and smoothing Gaussian kernels are the same as those used in Figure 1. Similarly to N11, the diffuse structure was only observed in the soft-band image. The bright soft emission is coincident with the OB association LH54 and is surrounded by the N51D Hα shell (see also Figure 1 of Cooper et al. 2004).

Zoom In Zoom Out Reset image size

The SRC region was selected to be the same as that in Cooper et al. (2004): an ellipse with major and minor radii of 58 and 4', respectively, and a 4'×4' square, indicated in Figure 6. Although a small circular region around the Wolf–Rayet star HD 36402 was excluded from the spectral analysis in the previous work, here we included this region because no local structures were resolved in the low-resolution XIS image. The BGD region was selected to be an off-source polygonal region, but the CCD corner irradiated by the calibration sources was excluded.

Using the EPIC image (Figure 6(c)), we searched for hard discrete sources. As a result, 10 sources were detected in the SRC region, while no point source was found in the BGD region. These sources were not clearly resolved in the XIS image. Therefore, we first analyzed the EPIC spectra of the point sources and then performed a model fit to the XIS data in the SRC region by adding the point-source component as a fixed model.

EPIC spectra of the 10 point sources in the SRC region were extracted and integrated. The BGD spectra were extracted from an annular region around each point source. The resultant BGD-subtracted spectra are shown in Figure 7. The spectra were well reproduced with an absorption power-law model (χ²/dof = 81/94). The best-fit absorption column and photon index were N_H = 1.3 (0.87–1.8) ×10²¹ cm⁻² and Γ = 1.7 (1.6–1.8), respectively. The observed flux was obtained to be 2.4 (2.3–2.5) ×10⁻¹³ erg cm⁻² s⁻¹ in the 2–10 keV band. The flux distribution of the 10 sources is given in Table 4.

Zoom In Zoom Out Reset image size

The spectral analysis of the Suzaku data was performed by essentially the same procedure as that described in Section 3.3. An example of a raw spectrum in the SRC region and the corresponding NXB are compared in Figure 8(a). Given that both spectra exhibit the same intensity of each fluorescent line, we considered that the NXB was successfully constructed. The NXB-subtracted spectra and the 2–10 keV surface brightnesses of the SRC and BGD regions were obtained as shown in Figure 8(b) and Table 5, respectively. The SRC spectrum appears to show excess surface brightness in the hard X-ray band. However, upon subtracting the point-source flux from the SRC, we found that the remaining surface brightness is consistent with that in the BGD region.

Zoom In Zoom Out Reset image size

We applied the vignetting correction to the BGD spectra and then subtracted them from the SRC spectra. The correction factors f(E) were obtained as given in Figure 9. The resultant SRC spectra are shown in Figure 10. The spectra of two active FIs were merged to improve the photon statistics. Several emission lines can be observed below ∼1 keV, while the spectrum in the hard band is featureless. Therefore, we fitted the spectrum with a VAPEC model plus the point-source component (a power law) derived in Section 4.2. The elemental abundances except for those of O, Ne, and Fe were fixed to the LMC values (Russell & Dopita 1992; Hughes et al. 1998). We assumed the Galactic absorption column to be N^G_H = 6.0 × 10²⁰ cm⁻², in accordance with Dickey & Lockman (1990). The fit was acceptable with χ²/dof = 92/103. The best-fit parameters and models are, respectively, shown in Table 6 and in Figure 10 with solid and dotted lines. Note that the point-source component was not fitted here. The model successfully reproduced the spectral shape as well as the flux of the hard X-rays, and hence no additional nonthermal component was required to improve the fit.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Cooper et al. (2004) argued that the EPIC spectrum of the diffuse emission exhibited thermal and nonthermal components, and the former accounted for ∼70% of the total observed flux (in the 0.3–3.0 keV band) of 1.1 × 10⁻¹² erg cm⁻² s⁻¹. With the given temperature and absorption column, the flux of the thermal emission was calculated to be ∼6 × 10⁻¹³ erg cm⁻² s⁻¹ in the 0.5–2.0 keV band. This is comparable to our result.

The enhancement of O and/or Ne abundances was reported in the works of Bomans et al. (2003) and Cooper et al. (2004). On the other hand, our spectrum does not show any evidence of a significant deviation from the typical LMC abundances, suggesting that the hot gas mainly originates from ISM with only slight enrichment by SN ejecta.