SPITZER OBSERVATIONS OF MF 16 NEBULA AND THE ASSOCIATED ULTRALUMINOUS X-RAY SOURCE

The data presented here are archival Spitzer data, taken on 2009 February 21. The IRS spectral map is described by 4 parallel pointings and 12 perpendicular pointings at half-slit steps, for the short high (SH; in the range 9.9–19.6 μm) module. The long high (LH) map (in the range 18.7–37.2 μm) has three parallel pointings and seven perpendicular pointings.

The total exposure time was 1.4 ks and 1.3 ks for SH and LH observations, respectively. The data were preprocessed by the IRS pipeline (version 18) at the Spitzer Science Center prior to download. Spectral maps were constructed using "BCD-level" processed data in conjunction with CUbe Builder for IRS Spectra Maps (CUBISM) version 1.5¹ (Smith et al. 2007; Kennicutt et al. 2003). The absolute calibration error on fluxes obtained with CUBISM is ∼25% (Smith et al. 2004). We adopt a distance of 5.1 Mpc to NGC 6946 as in Kaaret et al. (2010; 1'' is 25 pc). The LH maps obtained with CUBISM have intrinsic sizes of 264 by 176 (660 pc by 440 pc), while the SH maps are 23'' by 138 (575 pc by 345 pc). However, in order to avoid aperture effects, the spectra presented here were extracted from a like-sized aperture. Thus, the SH and LH spectral data result from an equally sized square aperture of size 133 (332.5 pc), which is composed of 9 LH pixels centered on the ULX position.

The high-resolution spectra (10–37 μm) are plotted in Figure 1. Since the spectra are extracted from matched apertures, the spectral lines in both the SH and LH modules are properly calibrated and the continuum is properly matched. We used Spectroscopic Modeling Analysis and Reduction Tool (SMART; see Higdon et al. 2004) version 8.1.2 to measure line fluxes. The fluxes of detected lines and upper limits for non-detections are presented in Table 1. We note that the [O iv] line is detected at a signal-to-noise ratio >12. Abolmasov et al. (2008) estimates an extinction of A_V = 1.54 mag in the direction of NGC 6946 X-1. For the infrared lines, we estimate from Draine (1989) that the extinction is A_λ < 0.1, and therefore negligible.

Zoom In Zoom Out Reset image size

The SH and LH maps for the detected lines are presented in Figure 2. They are shown together with two HST images of the MF 16 nebula in UV and Hα emission. A nearby (200 pc away) star-forming region and the associated H ii region can be seen just outside the Spitzer aperture. As can be seen in the Spitzer spectral maps, the low-ionization lines ([Ne ii], [S iii], and [Si ii]) are contaminated by the surrounding star-forming region. In contrast, the higher ionization lines [O iv] and [Ne iii] are clearly concentrated at the NGC 6946 X-1 location, suggesting a source of high-energy photons.

Zoom In Zoom Out Reset image size

2.2.1. X-Ray Spectrum

We use X-ray spectroscopy in conjunction with published optical/UV data to model the X-ray emission from the ULX and to extract the full optical to X-ray spectrum. We also use it to estimate the UV emission by extrapolating the model to lower energies. To model the X-ray spectrum we used the best available data from the longest (∼58 ks) Chandra observation in the archive (ObsID 1043, taken on 2001 September 7). The data were processed by Berghea et al. (2008; U45 in this paper) and the spectrum is plotted in Figure 3.

Zoom In Zoom Out Reset image size

We note that Fridriksson et al. (2008) find long-term X-ray flux variability of at most 30% in NGC 6946 X-1 based on Chandra, XMM-Newton, and ROSAT data. However, for completeness we have chosen the best data available to represent the source (e.g., Chandra data rather than ROSAT data), as well as one in which the luminosity is approximately averaged over the 10 years of observations (i.e., the 2001 data set). This average X-ray spectrum and luminosity best represents the ULX in a state in which the IR-line fluxes can be modeled. In addition, the equilibrium time-based recombination for the high-ionization lines is approximately 3000–100 years for He ii and [O iv], respectively—well above the variability timescales of the X-ray source (Berghea et al. 2010b).

2.2.2. HST UV and Optical Photometry

We used XSPEC version 12.4 to model the spectrum between 0.3 and 8.0 keV. The X-ray spectrum is the most important component of the total SED since it is likely the only observed emission that directly emanates from the ULX and is not contaminated by the surrounding star-forming region. For this reason we have taken many precautions in modeling the X-ray spectrum, however, we fit the X-ray and optical/UV literature data simultaneously. In this section, we describe the models we chose to fit the X-ray spectrum and extrapolate to the unknown UV region not covered by literature data.

Probably the most common and simplest model for ULX X-ray spectra is the two-component model with a multi-color accretion disk component (MCD; Mitsuda et al. 1984) and a power law for the hard tail. However, this model cannot be extrapolated to UV where the source of IR lines originates (e.g., Done & Kubota 2006), and therefore we use the physically motivated Comptonization models described in detail below.

For all models, following Berghea et al. (2008), the Galactic absorption column density was fixed at 2.02 × 10²¹ cm⁻², and an additional local absorption (N_H) model was added. We assume that the metallicity is approximately solar (see García-Benito et al. 2010 and references therein).

DISKIR model. The DISKIR model in XSPEC is a particularly useful Comptonization model to fit both optical and X-ray data because it also includes emission from an irradiated disk (Gierliński et al. 2009). We found that the best-fit model has a disk of radius 500 R_in (the inner disk radius), and the irradiated flux is at 1.5% of the total flux. The fraction of the flux thermalized in the inner disk was fixed at 0.1. In addition, the electron temperature kT_e and the radius of the Compton illuminated disk r_irr were set at 100 keV and 1.1, respectively. These values are typical for this model (e.g., Gierliński et al. 2009, 2008; Grisé et al. 2012).

BMC model. We also used the Bulk Motion Comptonization model (BMC in XSPEC) described by Titarchuk et al. (1997). Gliozzi et al. (2011) have suggested that this model is useful for ULX-like BH accretion scenarios. The best fit is obtained for a fraction of Comptonized flux of 0.12%.

SIRF model. Finally we tried the Self-IrRadiated Funnel model (SIRF in XSPEC), which describes the emission from a "supercritical funnel" (Abolmasov et al. 2009). This is created in objects similar to SS 433, by the wind formed at high accretion rates. Abolmasov et al. (2009) obtained a reasonable fit for XMM-Newton data of NGC 6946 X-1. The funnel opening angle was fixed at 20°, the inclination angle was set to zero, and the accretion rate fixed at an Eddington ratio of 10⁴. These model parameters were set based on the best-fit results to the X-ray data. We obtained similar results for inner temperature and radius as Abolmasov et al. (2009), but unfortunately could not find an acceptable fit using this model as Table 2 shows. As a result, this model has been disregarded as a viable model to fit the NGC 6946 X-1 SED in the discussion that follows.

The results from the X-ray fits are presented in Table 2, and the models are plotted in Figure 3. The first two models, DISKIR and BMC, both provide equally good fits to the X-ray data and will be used to construct the input SED for the CLOUDY modeling. However, as seen in Figure 3, they predict very different UV and optical emission, and therefore should yield very different results in the photoionization simulations. The physical difference between these two models is that the BMC model requires a bright optical companion to fit the optical and UV data. As seen in Figure 3, the DISKIR model fits the optical and UV data equally well with a bright (supergiant) companion (dotted red line) or an irradiated disk with a less luminous companion (solid red line).

The model for the O9 I supergiant shown as the solid line in blue in Figure 3 is a TLUSTY (Hubeny & Lanz 1995) O star model with effective temperature 31,000 K, log g = 3.5, and a luminosity of 3.16 × 10³⁹ erg s⁻¹. This fits the optical HST data (round black circles in Figure 3) very well. We add this star to both DISKIR and BMC models (dotted lines), but we remind our reader that the irradiated disk alone (solid red line) fits the optical–UV data equally well. Finally, we find that the SIRF model peaks just below the [O iv] edge, and is faint in the optical–UV, and therefore does not provide a good fit for these data as described above.

We use DISKIR and BMC SED models shown in Figure 3 as input spectra for CLOUDY. The bolometric luminosities that result from these models are 8.02 and 2.51 × 10⁴⁰ erg s⁻¹, respectively. We assumed a solar metallicity as per García-Benito et al. (2010). Previous optical studies of NGC 6946 X-1 (Dunne et al. 2000; Blair et al. 2001; Abolmasov et al. 2008) indicate that the ionized gas from the nebula was swept inside a thin shell (∼1 pc in width) around the ULX.

Estimates based on line ratios of [S ii] suggest that the density of the gas in the shell is quite high (400–600 cm⁻³; Blair et al. 2001; Abolmasov et al. 2008). If we use the ratio of the [S iii] IR lines to estimate the density (Dudik et al. 2007), we obtain densities lower than 150 cm⁻³ for temperatures in the range 10,000–20,000 K (Abolmasov et al. 2008). However, Figure 2 shows that the [S iii] lines are contaminated by the nearby star-forming region and therefore we prefer the higher value obtained from previous optical line ratio estimates, since these were obtained from higher resolution maps. Thus, we assume a spherical shell geometry, a density of 400 cm⁻³, and a filling factor of 0.2 (Blair et al. 2001).

The inner and outer radii of the shell were varied to obtain the best fit for the higher ionization lines [O iv] and [Ne iii], since these IR lines are less contaminated by surrounding star-forming regions as illustrated in Figure 2. We also use the previously measured He ii line flux (1.3 × 10³⁷ erg s⁻¹) as a constraint (Wong 2010) since this line has an ionization potential similar to [O iv] and [Ne iii] and likely emanates from a similar region to these IR lines. The predicted lines from the CLOUDY simulations are shown in Table 3 and Figure 4.

Zoom In Zoom Out Reset image size

Previous optical observations have shown that radius of the shell (or shells) is 7–10 pc and the thickness is of the order of ∼1 pc (Blair et al. 2001). The DISKIR model is consistent with a shell of radius 20 pc and is thin compared to the findings of Blair et al. (2001) by a factor of ∼10. The BMC model predicts radial dimensions that are closer to those values observed by Blair et al. (2001), but the shell predicted is smaller by a factor of ∼1.5–2. A larger shell than Blair et al. (2001) is expected for the DISKIR model because it provides more ionizing photons. Both models are very good fits to the high-ionization IR lines as shown in Figure 4. However, the BMC model overpredicts most low-ionization lines, while DISKIR underpredicts them.

We finally note that both models underpredict the amount of gas compared to the Hydrogen column densities obtained from spectral fits. This could be explained if part of the absorption takes place very close to the accretion disk and not near the line-emitting region.

Berghea et al. (2010b) discuss in detail possible scenarios for the [O iv] detected in the Holmberg II ULX in relation with AGNs and star-forming regions. The conclusions of that paper apply to NGC 6946 X-1 as well. The [O iv] is clearly associated with the ULX and not with the star formation nearby (Figure 2). Moreover, the hottest stars alone cannot produce the [O iv] measured from the MF 16 nebula. For more details we again direct our reader to Berghea et al. (2010b) and the references therein. An interesting result of that paper was that hot stars can have an important effect if a high-energy source is also present. The detailed CLOUDY simulation in Berghea et al. (2010b) showed that an O star can increase the [O iv] luminosity by a factor of three. This is also true in our case if the ULX companion is an O-star. We draw a similar conclusion based on the He ii line, which has a very similar ionization potential to [O iv], and therefore requires a very high energy source.

The results of our CLOUDY modeling cannot easily discriminate between an O-star plus accretion disk and an irradiated disk (Figure 3). As discussed in Section 3.2, previous optical data suggest that the radius of the shell is 7–10 pc and the thickness is on order 1 pc (Blair et al. 2001). Table 3 shows that the BMC model predicts a radius and a thickness that are much closer to the optical findings than the DISKIR predictions. Finally, Figure 4 shows that the BMC model predictions for He ii and [Ne iii] are very slightly better than the DISKIR model predictions. For all of these reasons, we prefer the BMC model (the O-star plus accretion disk model) to the DISKIR model (an irradiated disk) to describe NGC 6946 X-1.

Berghea et al. (2010b) explored shocks as a possible mechanism for the lines for both NGC 6946 X-1 and Holmberg II ULXs. They found that the He ii line is probably produced by shocks in the case of another famous nebula-surrounded ULX in Holmberg IX. However, for NGC 6946 X-1 and Holmberg II ULXs the He ii emission is consistent with photoionization and not with shocks. To further confirm that the [O iv] alone is not dominated by shocks, we reproduced Figure 10(a) from Berghea et al. (2010b) for NGC 6946 X-1. We used the brightest pixel in the [O iv] spectral map to estimate a peak emission for this line. We note that this is in fact a lower limit for the estimate, since it represents the average flux over the large LH pixel. As shown in Figure 5, the shock velocity required to produce the peak [O iv] in NGC 6946 X-1 is in excess of 300 km s⁻¹. Dunne et al. (2000) find shock velocities from Hα are at most 270 km s⁻¹. This suggests that the shocks needed to produce the high-ionization lines [O iv] and He ii are simply not present, based on the Hα data.

Zoom In Zoom Out Reset image size

The ultimate source of X-ray and high-ionization emission in ULXs has been widely debated. Indeed, multiple mechanisms have been proposed to explain the high-energy emission, including beamed stellar-mass BHs, super-Eddington stellar-mass BHs, and finally, IMBHs. Evidence for one or more of these scenarios has been found for a variety of ULXs. While our observations and modeling of the ULXs in Holmberg II and the M16 nebulae strongly suggest these objects are not SNRs or beamed stellar-mass BHs, this analysis cannot distinguish between a super-Eddington stellar-mass BH or an IMBH as the source of the X-ray, optical, and high-ionization emission. This is due in part because the X-ray continuum is fit equally well with models describing either scenario. We note, on the other hand, that the infamous SS 433 in our Galaxy is a known beamed, super-Eddington stellar-mass BH with a clear host nebula. All evidence provided in the literature points to shocks or winds/outflows as the primary mechanism(s) for ionization in this object and the nebular region immediately surrounding the ULX (e.g., Abolmasov et al. 2010; Begelman et al. 1980, 2006). Because SS 433 is the only super-Eddington stellar-mass BH with a nebula definitively known to date, this is the only such source we have to compare with NGC 6946 X-1. Our analysis suggests that unlike SS 433, NGC 6946 X-1 has high-ionization lines (specifically the Oxygen line here and the He ii from Abolmasov et al. 2008 that are most likely not produced by shocks). In this singular way the famous SS 433 super-Eddington stellar-mass BH is very different from NGC 6946 X-1.

The deficiency of super-Eddington sources in nebular environments precludes a detailed comparison of NGC 6946 X-1 with such objects. However, multiple active BHs in dense environments that produce high-ionization lines via photoionization are known to exist and are well documented in the literature. These sources, while not IMBHs, may prove an excellent comparative resource. For this reason, in the subsequent analysis, we focus attention on this comparison.

The infrared [O iv] line (together with [Ne v]) is a signature of high ionization and excitation usually associated with actively accreting BHs (e.g., Lutz et al. 1998). This line is often used to disentangle the starburst emission in composite type galaxies and is a good indicator of the AGN power (Genzel et al. 1998; Sturm et al. 2002; Satyapal et al. 2004; Smith et al. 2004; Satyapal et al. 2007; Meléndez et al. 2008). Genzel et al. (1998) and Sturm et al. (2002) have used the ratio of [O iv]/[Ne ii] to distinguish between starbursts and AGN-dominated galaxies in the mid-IR. In Figure 6(a), we compare the NGC 6946 X-1 mid-IR line ratios with data published by Genzel et al. (1998). The comparison clearly indicates high excitation for the ULX, similar to those seen in AGNs. The Holmberg II ULX from Berghea et al. (2010a) is also plotted here, though the [Ne ii] flux for this ULX is an upper-limit. Both sources appear as AGN-like when this diagnostic is used.

Zoom In Zoom Out Reset image size

Other infrared line diagnostics have been proposed by Dale et al. (2006, 2009) to disentangle AGN-type ionizing sources from star formation using the ratios [S iii] 33.48 μm/[Si ii] 34.82 μm and [Ne iii]/[Ne ii]. Figure 6(b) shows the four regions defined by Dale et al. (2006) from their Figure 5, and their published data for different sources in the SINGS sample. NGC 6946 X-1 is located well within region II, together with most Seyfert-like nuclei. Holmberg II is also plotted here from Berghea et al. (2010a) and shows ratios similar to low-luminosity nuclei, but we note that both [S iii] and [Ne ii] are upper limits for this object. We also note that the [S iii], [Si ii], and [Ne ii] are heavily contaminated by star formation as is clear in the spectral maps in Figure 2. However, we also note that the AGN and star-forming regions presented by Dale et al. (2006, 2009) are subject to the same contamination biases and aperture effects as the ULXs. We therefore still find it useful to compare the two ULXs with these sources.

The fact that the Dale et al. (2006, 2009) ratios for NGC 6946 X-1 are Seyfert-like and that the [O iv]/[Ne ii] ratio is quite high suggests that NGC 6946 X-1 is very gas rich compared to most ULXs and has a very strong UV source producing the ionizing radiation. We note that Vázquez et al. (2007) found similar results in their Spitzer IRS study of the NGC 4485/90 ULXs. However, they did not find clear evidence for [O iv] in their sample of objects. In examining this data further we find slightly discrepant ratios from Vázquez et al. (2007). From our analysis, only ULX-2 in their sample actually falls marginally inside region I, the other five ULXs being located in regions II and III. In communicating with the authors we find that the discrepancy is likely due to differences in reduction pipeline used for processing or the apertures used in either analysis.

Finally, we compared the mid-IR line ratios for NGC 6946 X-1 (and Holmberg II ULX) with those of other objects presented in the diagnostic diagrams of Weaver et al. (2010). The error bars for the ULXs in this plot are approximately of the same size as the data points and represent 25% absolute calibration error for CUBISM (Smith et al. 2004). Weaver et al. (2010) find that high-ionization AGNs from the Swift Burst Alert Telescope (BAT) sample (Tueller et al. 2008, 2010) distinguish themselves from other less luminous AGNs and starburst galaxies in their [Ne iii]/[Ne ii] and [O iv]/[Ne ii] ratios. A replication of the Weaver et al. (2010) plot is shown in Figure 7. These high-ionization AGNs have ratios very close to or over 1. In Figure 7, we overplotted the NGC 6946 X-1 as well as the Holmberg II ULX—the only two ULXs known to produce prominent [O iv] emission. Interestingly, NGC 6946 X-1 occupies the overlapping region between the high-ionization BAT AGNs and their lower-luminosity counterparts. The results for NGC 6946 X-1 are consistent with photoionization which is similar to that seen in low-ionization nuclear emission line region (LINER) or Seyfert-like nuclei rather than star forming in nature.

Zoom In Zoom Out Reset image size

All three diagnostic diagrams suggest that the [O iv] emission in Holmberg II ULX and in NGC 6946 X-1 is dominated by the ULX and that the surrounding star-forming regions provide little contamination to this line. Both ULXs have been shown to be candidate IMBHs by other authors (Lehmann et al. 2005; Rao et al. 2010). While these findings cannot definitively prove that NGC 6946 X-1 is either an IMBH or a super-Eddington accretor, it does suggest that the environment may be very similar for this source and AGN-like objects. We finally mention that the correlation of NGC 6946 X-1 with ultraluminous infrared galaxies (ULIRGs) in Figure 6(a) and LINERs in Figure 7 may lend some credence the predictions of McKernan et al. (2010, 2011) who suggest that many LINERs might host ULX-like IMBHs. Indeed many ULIRGs are LINERs (Satyapal et al. 2007; Dudik et al. 2007; Carrillo et al. 1999).

Whether NGC 6946 X-1 is a super-Eddington source or an IMBH cannot be definitely determined from the observed and modeled data presented in this paper. However, as discussed in Section 4.2, the ratios from NGC 6946 X-1 look very similar to the ratios of well-known AGNs. We therefore find it useful to estimate the possible BH mass of this ULX based on AGN criteria with the caveat that the criteria assume NGC 6946 X-1 behaves much like an AGN which is not definitely known from the observed and modeled data presented here. The list below shows four BH mass estimates derived for NGC 6946 X-1.

• 1.  
  Gliozzi et al. (2011, 2009) have proposed a new method to estimate BH masses using the BMC model. This method uses the photon index of the Comptonized part of the spectrum and the normalization to scale from well-known stellar-mass BHs with well-defined spectral transitions. Using a scaling method defined by Shaposhnikov & Titarchuk (2009) that is based on a detailed study of 8 different Galactic BH systems with 17 spectral transitions, Gliozzi et al. (2009, 2011) extend these scaling techniques to extragalactic BH sources. Indeed, one of these sources is a newly discovered ULX in NGC 3621. Thus, this technique can be used in the case of NGC 6946 X-1 as well, with the assumption that the states of ULXs mimic those of GBHB. It assumes that BHs of all masses show similar spectral states, at least with respect to Comptonization. If we scale the results from our BMC fits to XTEJ1550-564 (as done by Gliozzi et al. 2011 in their Figure 1) and use M_☉ = 9.1 and d = 4.4 kpc from Orosz et al. (2011), and θ = 72 from Gliozzi et al. (2011), we obtain a BH mass of ≈1800 M_☉.
• 2.  
  Both DISKIR and BMC X-ray models predict a cool disk. Assuming a BH accreting at the Eddington limit, the mass of the BH can be estimated at 1–1.5 × 10⁴ M_☉. However, this estimate assumes that the ULX is in a typical high state. If this is not the case, the inner disk temperature cannot be used reliably to estimate the mass of the BH (Stobbart et al. 2006).
• 3.  
  Another estimate can be made just based on the bolometric luminosity calculated from our models. Using the values listed in Table 2, and assuming again accretion at Eddington limit, we obtain masses of 130 and 450 M_☉.
• 4.  
  Finally, Dasyra et al. (2008) found a good correlation between the luminosity of the [O iv] line and the mass of the central BH in standard AGNs. We estimate a BH mass of 2.7 × 10⁵ M_☉ if this correlation is extrapolated to NGC 6946 X-1.

In conclusion these four methods predict a very wide range of BH masses for NGC 6946 X-1. Indeed the estimates predict masses that span three orders of magnitude: 10²–10⁵ M_☉, so emphasis on any one estimate is difficult to justify. However, it is interesting to note that all four put this object within the IMBH regime for the given criteria (e.g., accretion at the Eddington rate).