MID-IR FORCAST/SOFIA OBSERVATIONS OF M82

Over the last few 10⁸ yr a series of starbursts has been trigg-ered in M82 (de Grijs 2001; Mayya et al. 2006) due to its interaction with M81 and NGC 3077 (e.g., Appleton et al. 1981; Yun et al. 1994; Sun et al. 2005; de Mello et al. 2008). The most recent starburst (≲ 50 Myr, e.g., Rieke et al. 1980; Satyapal et al. 1997) has created a stellar cluster at the center of M82. A ring of ionized gas that is enveloped by a molecular gas ring surrounds this cluster, and the starburst is likely fueled by gas that is funneled toward this region by a ∼1 kpc (∼1') stellar bar (Telesco et al. 1991; Achtermann & Lacy 1995; Wills et al. 2000; Greve et al. 2002). These gas rings, which appear between about 10'' and 30'' from the nucleus, could be gas swept up by a central expanding superbubble (Matsushita et al. 2000, 2005). High external gas pressure imposed on the molecular clouds probably drives the current starburst (Keto et al. 2005). This view is supported by the detection of warm molecular gas that is likely heated by dissipation of turbulence (Panuzzo et al. 2010).

The view to the center of M82 is obscured by patchy and high extinction (Satyapal et al. 1995; Förster Schreiber et al. 2001). The mid-infrared (mid-IR) wavelength regime, where the extinction is modest, is therefore used extensively to study the center of M82 (e.g., Rieke et al. 1980; Telesco et al. 1991; Förster Schreiber et al. 2001; Beirão et al. 2008). Individual star-forming clouds that are located at the inner edge of the molecular gas ring or within the ionized gas ring have been observed in the mid-IR with large ground-based observatories that provide (sub-) arcsecond resolution (Lipscy & Plavchan 2004; Gandhi et al. 2011). Unfortunately, ground-based mid-IR observations are limited to a few telluric windows at ≲ 30 μm.

We have used the mid-IR camera FORCAST on SOFIA to study the central region of M82. FORCAST provides relatively high spatial resolution and also covers the entire 5–40 μm wavelength range. Emission from hot, small grains dominates the continuum at short mid-IR wavelengths while emission from warm, large grains dominates at long mid-IR wavelengths. A total wavelength coverage is therefore important to determine the detailed contribution of each dust component to the continuum emission and to estimate the heating of the dust.

We adopt a distance of 3.5 Mpc for M82 (Dalcanton et al. 2009), so that 1'' corresponds to 17.5 pc.

We observed M82 in the 6.35 μm (Δλ = 0.14 μm), 6.61 μm (Δλ = 0.24 μm), 7.71 μm (Δλ = 0.47 μm), 31.5 μm (Δλ = 5.7 μm), and 37.1 μm (Δλ = 3.3 μm) passband filters using FORCAST (Adams et al. 2010; T. L. Herter et al. 2012, in preparation) on SOFIA (Young et al. 2012).

FORCAST is a dual-channel camera with a short-wavelength channel (5–25 μm) and a long-wavelength channel (25–40 μm), allowing for simultaneous observations at two specific wavelengths by selecting bandpass filters in filter wheels. A dichroic beam-splitter directs the same field of view to both the short- and long-wavelength cameras. To increase sensitivity, the beam-splitter can be either removed or replaced by a mirror during an observation. The effective field of view of the two 256 square arrays is 34 × 32. After distortion correction, the pixel size is 0768.

We observed M82 on 2010 December 1 in the two long-wavelength bands without dichroic at an altitude of about 43,000 feet. We used the on-chip chop-nod observing mode with a chopper throw and nod-distance of 120'' and 90'', respectively, and a chop frequency of 2 Hz. In this mode the source is always on the detector array. For the 37 μm and 31 μm observations we took seven and five integrations, respectively, with an on-source time of 60 s per integration. The observations in the 37 μm band were taken at a zenith angle of about 58°.

On 2010 December 4, we obtained four integrations of M82 in the 6.6 μm and 7.7 μm bands and eight integrations in the 6.4 μm band from an altitude of about 42,000 feet. All bands were observed without a beam-splitter, except for the 7.7 μm band observations, which were observed simultaneously with the 31 μm band. We observed again in the on-chip chop-nod mode, but with a chopper throw and nod-distance of 90'' and a chop frequency of 5 Hz. All bands were observed at a zenith angle of about 53°.

Since M82 is bright in all bands, we spatially registered the individual chop/nod integrations using a two-dimensional Gaussian fit over the entire galaxy before co-adding the individual integrations. These pointing-corrections are about 3 pixels and improve the noise and the final maps. Due to the on-chip chop-nod observing mode the extent of the final images is about 75'' × 75''. Since pointing accuracy and drifts were on the order of several arcseconds, we determined the absolute position of the M82 observation by comparing the 7.7 μm FORCAST map with the 8 μm Spitzer/IRAC map from the Spitzer archive.

The flux densities in all bands were color-corrected. The standard stars βGem, βUMi, and μUMa were used for absolute flux density calibration and to estimate the beam size. Based on the uncertainty of flat fielding and the water vapor burden, as well as the variance of the flux density measurements, the uncertainty of the intrinsic flux density of the standard stars, and the fact that βGem and βUMi are variable stars, we estimate a 3σ uncertainty of ∼20% for the absolute calibration (T. L. Herter et al. 2012, in preparation; Herter et al. 2012). Due to turbulent airflow across the telescope and pointing instabilities the observations are not diffraction limited. For the M82 observations, we estimate a beam size of about 4'' ± 05 for all the bands.

The original maps of the 6.4 μm, 6.6 μm, and 7.7 μm continuum had a lower signal to noise than the two long-wavelength maps and we therefore smoothed them with a 2 pixel FWHM Gaussian kernel. The final flux densities are presented in Table 1.

We include Herschel/PACS 70 μm photometric data in our spectral energy distribution (SED) analysis below. Since processed and published PACS 70 μm photometric data were unavailable, we used level 2.5 data product from the Herschel archive. The observations were carried out in scan-map mode at medium speed (20'' s⁻¹) resulting in a smeared beam of about 546 × 576 (PACS Observer Manual). We extracted background-subtracted flux densities within a 68 aperture (∼3.5 PACS pixels) at three positions (see Table 1 for positions). The flux densities are 115 ± 11 Jy (main peak), 98 ± 10 Jy (secondary peak), and 116 ± 11 Jy (ridge). Within a radius of 35'', similar to the entire region observed with FORCAST, the 70 μm flux density is 1644 ± 41 Jy.

3.1. Morphology

Figure 1 shows the distribution of the mid-IR emission in each of the observed bands. The 6.4 μm, 6.6 μm, and 7.7 μm emission is slightly more extended than the 31 μm and 37 μm emission. Also, the contrast in the morphology is higher in the 31 μm and 37 μm emission than in the short-wavelength bands, with the peaks being more pronounced. The main peak of all mid-IR distributions is 45 west–southwest from the dynamical center. Strong emission extends further to the west–southwest and "peaks" roughly 9'' from the center ("western ridge"). A secondary peak is visible 4'' east–northeast of the center (except at 6.6 μm). In the 7.7 μm band, the emission at the secondary peak appears as two separate components. The positions of the main and secondary peaks coincide roughly with the position of the ionized ring or the inner edge of the molecular ring.

Zoom In Zoom Out Reset image size

The 7.7 μm FORCAST map is in excellent agreement with the Spitzer/IRAC 8 μm map (Figure 1). Also, the FORCAST 31 μm map agrees very well with the 30 μm map obtained by Telesco et al. (1991) using the NASA Infrared Telescope Facility (IRTF), and the flux densities at the main and secondary peaks agree within the calibration uncertainty.

In Figure 2 we show a three-color image of M82, where we have combined the 6.6 μm (blue), 31 μm (green), and 37 μm (red) bands. All colors are scaled linearly with flux density and start at the 3σ level of the statistical background noise in each band. The image shows that the 6.6 μm is more extended and that the secondary peak and western ridge are more pronounced in the two long-wavelength bands.

Zoom In Zoom Out Reset image size

Figure 3 shows the emission profiles along a 60'' cut following the major axis of M82. The length and position of the cut is shown in Figure 1. The profiles are in steps of 1 pixel and the flux densities are summed over 1 × 5 pixel strips perpendicular to the major axis and normalized to the emission peak. They emphasize that in the 31 μm and 37 μm bands the ratio between the main and secondary peak is much stronger than in the three short mid-IR bands. In fact, the 6.6 μm continuum only shows a gradient in the flux density from the main peak toward the position of the secondary peak. The emission from the "western ridge" is very pronounced in the 31 μm and 37 μm bands, less in the 7.7 μm and 6.4 μm maps, and barely noticeable in the 6.6 μm map. The 31 μm and 37 μm emission show a peak at this position that is stronger than the secondary peak.

Zoom In Zoom Out Reset image size

Figure 3 also shows the flux density ratios along the major axis of M82. The profile of the 37 μm/6.6 μm and 37 μm/7.7 μm flux density ratios basically follows the profile of the 37 μm flux density. In contrast, the flux density ratios of the 7.7 μm/6.6 μm and 7.7 μm/6.4 μm bands are fairly uniform along the major axis, with larger variations only at the edge of the profile, where the signal-to-noise becomes small. The profile of the 37 μm/31 μm flux density ratio differs from all others. It is basically flat within ±10'' of the center of M82 and then increases on either side, while the flux density ratios of all other bands either decrease or stay constant with distance from the center. This indicates a higher color temperature of the large grains in the inner region than in the outer region. Also, the difference of the profiles between the three short mid-IR bands and two long mid-IR bands indicate that they trace different dust components with spatially distinct excitation conditions.

3.2. Mid-IR Spectral Energy Distribution

Siebenmorgen & Krügel (2007)⁶ have created template SEDs for starbursts as a function of the total luminosity (L(tot)), optical extinction, contribution to the total luminosity by OB stars, and radius of the nuclear starburst, covering the wavelength range 0.03–2000 μm. We use the FORCAST and PACS 70 μm data points as well as low-resolution Spitzer-IRS spectra between 5.3 μm to 12.8 μm from regions 2 and 3 of Beirão et al. (2008) to find the SED template that fits the data in the two mid-IR emission peaks within a 68 × 68 (∼115 × 115 pc) area, corresponding to ∼9 × 9 FORCAST pixels. The lower size limit of the area is given by the Spitzer observation. Since this area is much smaller than the smallest nuclear radius in the SED template grid (350 pc), we constrained our search algorithm to that template series and allow it to fit an additional scaling factor. Figure 4 shows the data points and the best-fitting SED template at the main and secondary peaks. It includes three data points from Telesco et al. (1991), multiplied by a factor of two, which corresponds to the flux density ratio within a 68 and 45 aperture in a FORCAST band at the peaks. The flux densities at 70 μm and 37 μm at the main and secondary peaks are very similar and the 70 μm flux density is thus crucial to constrain the SED fit.

Zoom In Zoom Out Reset image size

For the main mid-IR emission peak, the best-fitting SED template has a total luminosity of L_MP(tot) = 10^11.2 L_☉, an OB luminosity fraction of 40%, an extinction of A_V = 18, and a hydrogen density of n = 5 × 10³ cm⁻³ in the hot spots. The template requires an additional scaling factor of 0.041, which is of the order of the ratio of observed to the template area, and lowers the total luminosity to L_MP(tot) = 6.7 × 10⁹ L_☉. The best-fitting model for the secondary emission peak has a total luminosity of L_SP(tot) = 10^10.7 L_☉, an OB luminosity fraction of 40%, an extinction of A_V = 9, a hydrogen density of n = 5 × 10³ cm⁻³, and requires an additional scaling factor of 0.11, which reduces the total luminosity to L_SP(tot) = 5.7 × 10⁹ L_☉.

Due to limited resolution of the SED template parameters we estimate the derived extinction to be within a factor of two. Both values are lower than the extinction reported by Förster Schreiber et al. (2001) for the entire central region (A_V = 52 ± 17). This discrepancy could be due to the patchy extinction in the central region (Satyapal et al. 1995; Lipscy & Plavchan 2004; Gandhi et al. 2011).

3.2.1. Extinction

The extinction laws between 3 and 10 μm toward the center of M82 and the Galactic center (GC) are similar (Förster Schreiber et al. 2001), and for the GC it is best modeled with R_V = 5.5 (Draine 2003). To estimate the extinction and emissivity at the FORCAST bands we therefore use the GC extinction law as modeled by Li & Draine (2001) and Weingartner & Draine (2001)⁷ with R_V = 5.5, a gas-to-dust mass ratio of 105, a dust grain mass per hydrogen atom of m_d = 2.2 × 10⁻²⁶ g/H, N(H)/A_V = 1.37 × 10²¹ cm⁻² mag⁻¹ (with $N({\rm H}) =N({\rm H\,\mathsc{i}})+N({\rm H}_{2})$), and an emissivity index of β = 1.79 for the 31 μm and 37 μm bands. The mass absorption coefficient and albedo at 31.62 and 37.15 μm are κ = 385.2 cm² g⁻¹ and 0.0007, and κ = 288.8 cm² g⁻¹ and 0.0006, respectively. Using the relation κ_abs(λ) = 0.4 ln 10 (1 − albedo) × [A(λ)/N_H]/m_dust cm² g⁻¹ (Li & Draine 2001) leads to A(37)/A_V = 9.46 × 10⁻³ and A(31)/A_V = 1.26 × 10⁻², resulting in τ(37 μm) = 0.16 and τ(31 μm) = 0.21 at the main emission peak, respectively. At the secondary emission peak the opacities are τ(37 μm) = 0.08 and τ(31 μm) = 0.11.

3.2.2. Dust Mass

We use three different methods to estimate the dust masses at the main and secondary emission peaks within 68 × 68.

Applying the extinction laws from the previous section we estimate the gas mass using M_g = μ × m_H × N(H) × Area, where μ = 1.4 is the mean atomic mass per hydrogen, m_H is the mass of atomic hydrogen, and N(H) is the total hydrogen column density. We obtain a total gas mass of M_g(MP) = 3.8 × 10⁶ M_☉ at the main peak and M_g(SP) = 1.9 × 10⁶ M_☉ at the secondary peak. Using a gas-to-dust ratio of 105 this gives dust masses of M_d(MP) = 3.6 × 10⁴ M_☉ and M_d(SP) = 1.8 × 10⁴ M_☉ at the main and secondary peak, respectively. Due to the uncertainty of the extinction, the dust masses can vary by a factor of two.

The dust mass can also be estimated by M_d = 1/κ_abs × F_ν(λ) × D²/B(λ, T), where κ_abs is the mass absorption coefficient, F_ν is the flux density, D is the distance to M82, and B(λ, T) is the Planck function at wavelength λ and temperature T. Using the previously estimated opacities for 31 μm and 37 μm and a modified blackbody function, we estimate a color temperature of 68 ± 10 K at both emission peaks. The resulting dust masses are 1.15 × 10⁴ M_☉ and 7.9 × 10³ M_☉ at the main and secondary peaks, respectively. The uncertainty of the masses is within a factor of four.

Sanders et al. (1991) estimate the dust mass using the relation M_d = (L(FIR)/10⁸ L_☉) × (40 K/T_d)⁵ 10⁴ M_☉, where T_d is the dust temperature and L(FIR) is the far-IR luminosity in the range between 40 and 500 μm. From the best-fitting model SED, we determine a L_MP(FIR) = 3.08 × 10⁹ L_☉ at the main peak and L_SP(FIR) = 2.52 × 10⁹ L_☉ at the secondary peak. Using the color temperatures derived above we obtain dust masses of M_d(MP) = 2.2 × 10⁴ M_☉ and M_d(SP) = 1.8 × 10⁴ M_☉ at the main and secondary peak. These dust masses are good to a factor of three.

High-resolution CO measurements with a 42 aperture suggest molecular hydrogen column densities of N(H₂) ≈ (4–10) × 10²² cm⁻² at our main peak and N(H₂) ≈ 4 × 10²² cm⁻² at our secondary peak (Weiß et al. 2001). So, the enclosed gas masses, including He and other heavy elements, are (1.7–4.2) × 10⁶ M_☉. Assuming a gas-to-dust ratio of 105 this yields dust masses between (1.6–4) × 10⁴ M_☉, in agreement with our calculations above.

We have presented the first results of mid-IR observations of M82 obtained with FORCAST on SOFIA. M82 was observed in the 6.4 μm, 6.6 μm, 7.7 μm, 31.5 μm, and 37.1 μm bands. The observations cover a 75'' × 75'' region. All bands show a strong peak located 45 west–southwest of the kinematic center of M82. A secondary peak 4'' east–northeast of the nucleus is seen in the 6.4 μm, 7.7 μm, 31.5 μm, and 37.1 μm bands, but not in the 6.6 μm continuum. The profiles of the flux density ratios over 60'' along the major axis indicates that the emission at the three short mid-IR bands is dominated by a different dust component than the emission at the two long mid-IR bands. We fitted SED templates to the FORCAST data combined with PACS 70 μm and low-resolution Spitzer-IRS spectra and estimated extinctions of A_V = 18 and A_V = 9 toward the main and secondary peaks and a dust color temperature of 68 K in both peaks. The dust masses at the locations of the emission peaks are on the order 10⁴ M_☉.

We thank the SOFIA telescope engineering and operations team, as well as the SOFIA flight crews and mission operations team for their support. This work is based on observations made with FORCAST on the NASA/DLR Stratospheric Observatory for Infrared Astronomy (SOFIA). SOFIA science mission operations are conducted jointly by the Universities Space Research Association, Inc. (USRA), under NASA contract NAS2-97001, and the Deutsches SOFIA Institut (DSI) under DLR contract 50 OK 0901. Financial support for FORCAST was provided to Cornell by NASA through award 8500-98-014 issued by USRA. This work is based in part on observations made with the Spitzer Space Telescope, obtained from the NASA/IPAC Infrared Science Archive, both of which are operated by the Jet Propulsion Laboratory, California Institute of Technology under a contract with the National Aeronautics and Space Administration. This research has made use of NASA's Astrophysics Data System Abstract Service.