THE DISPLACED DUSTY INTERSTELLAR MEDIUM OF NGC 3077: TIDAL STRIPPING IN THE M 81 TRIPLET

Figure 1 presents a multi-wavelength view of the tidal features around NGC 3077. The top left panel shows the SPIRE map at 250 μm (convolved to the 500 μm beam). The brightest source in the field is NGC 3077 itself, but significant emission is detected east of the main body of the galaxy. The nuclear starburst in NGC 3077 is distributed over scales of ∼10'' (Martin 1997, Ott et al. 2003, Walter et al. 2006), which is unresolved at SPIRE's resolution. The extended dust emission toward the east is spatially coincident with H i column densities ⩾10²¹ cm⁻² in the tidal feature (Figure 1, top right; H i data are taken from THINGS, "The H i Nearby Galaxy Survey"; Walter et al. 2008). The apparent correlation of H i emission with dust seen in the SPIRE data (in particular toward high H i column densities) implies that the dust emission is indeed physically coincident with the tidal arm and does not arise from Galactic cirrus which is observed toward the direction of the M 81 group (e.g., Davies et al. 2010; Sollima et al. 2010).

Zoom In Zoom Out Reset image size

To quantify the dust properties, we have selected 25 apertures of radius 1' (∼1 kpc) each from visual inspection of the SPIRE 250 μm map. The locations (and numbers) of the apertures are shown in the bottom right panel of Figure 1 (green circles) and the positions are tabulated in Table 1. The red circle indicates the aperture (no. 1) centered on NGC 3077. For each aperture, we have derived the flux in the SPIRE bands (Table 1). The error in this measurement is derived from the dispersion of 44 background apertures of the same size that are located away from the H i tidal arms (at column densities <2 × 10²⁰ cm⁻²). This dispersion is mainly driven by unresolved background sources at high redshift. The H i column densities (at 40'' resolution) for each aperture are also given in Table 1—all column densities are >4 M_☉ pc⁻² (5 × 10²⁰ cm⁻²). Significant dust emission (here defined as a >3σ detection at 250 μm) is detected in eight apertures, i.e., an area of ∼30 kpc². These apertures are labeled in blue in Figure 1 (bottom right) and marked with a boldface font in Table 1. Averaging the other apertures also yields a statistical detection of dust emission.

Table 1. Aperture Measurements

No.a	R.A.	Decl.	F(250 μm)b	F(350 μm)b	F(500 μm)b	$\Sigma _{{\rm dust}}$c	$\Sigma _{{\rm H}\,{\mathsc {i}}}$d	ΣSFRe
	J2000.0	J2000.0	(MJy sr−1)	(MJy sr−1)	(MJy sr−1)	(M☉ pc−2)	(M☉ pc−2)	(10−4 M☉ yr−1 kpc−2)
1	10 03 17.18	+68 43 53.0	23.61	8.41	3.07	0.054 ± 0.005	12.42	220
2	10 03 05.57	+68 45 49.4	0.73	0.34	0.15	⋅⋅⋅	2.95	⋅⋅⋅
3	10 02 49.53	+68 44 12.2	1.54	0.91	0.48	0.053 ± 0.015	4.88	⋅⋅⋅
4	10 02 51.35	+68 42 15.7	1.71	0.99	0.53	0.058 ± 0.016	7.04	⋅⋅⋅
5	10 02 52.28	+68 40 19.3	0.02	0.11	0.09	⋅⋅⋅	6.04	⋅⋅⋅
6	10 03 10.96	+68 41 12.8	2.63	1.32	0.58	0.075 ± 0.020	10.18	⋅⋅⋅
7	10 03 16.30	+68 39 16.3	−0.10	0.05	0.05	⋅⋅⋅	5.63	⋅⋅⋅
8	10 03 33.21	+68 41 32.2	2.53	1.35	0.65	0.078 ± 0.021	10.69	0.65
9	10 03 38.53	+68 39 40.5	0.29	0.31	0.17	⋅⋅⋅	7.69	⋅⋅⋅
10	10 03 57.30	+68 40 50.1	2.74	1.73	0.95	0.101 ± 0.027	14.78	4.00
11	10 03 50.72	+68 42 51.7	2.65	1.68	0.91	0.097 ± 0.026	16.23	1.23
12	10 03 42.18	+68 44 43.2	0.76	0.57	0.34	⋅⋅⋅	8.16	⋅⋅⋅
13	10 03 27.96	+68 46 15.9	0.84	0.51	0.31	⋅⋅⋅	3.30	⋅⋅⋅
14	10 04 20.44	+68 40 45.2	0.63	0.41	0.21	⋅⋅⋅	6.39	⋅⋅⋅
15	10 04 13.43	+68 42 44.4	1.54	1.02	0.52	0.057 ± 0.016	15.62	0.29
16	10 04 35.72	+68 42 29.4	0.59	0.41	0.21	⋅⋅⋅	6.30	1.09
17	10 04 26.79	+68 44 26.1	0.19	0.16	0.10	⋅⋅⋅	5.23	⋅⋅⋅
18	10 04 04.49	+68 44 38.4	0.63	0.42	0.26	⋅⋅⋅	12.60	⋅⋅⋅
19	10 04 16.16	+68 46 22.8	−0.23	−0.07	0.01	⋅⋅⋅	6.49	⋅⋅⋅
20	10 03 50.70	+68 46 37.3	0.54	0.43	0.31	⋅⋅⋅	8.93	⋅⋅⋅
21	10 04 11.77	+68 48 24.2	0.04	0.11	0.13	⋅⋅⋅	4.71	⋅⋅⋅
22	10 03 49.41	+68 48 48.8	0.89	0.51	0.29	⋅⋅⋅	5.94	⋅⋅⋅
23	10 03 28.82	+68 49 32.7	0.72	0.39	0.19	⋅⋅⋅	4.58	⋅⋅⋅
24	10 03 21.66	+68 51 29.2	0.78	0.42	0.23	⋅⋅⋅	5.35	⋅⋅⋅
25	10 03 14.47	+68 53 25.6	0.78	0.38	0.24	⋅⋅⋅	4.36	⋅⋅⋅

Notes. ^aAperture number in bold indicates regions with significant emission (i.e., >3 σ at 250  μm). Dust mass surface densities have been derived for those regions assuming a temperature of T = 12.6 K (only exception: aperture 1 [NGC 3077, T = 30.6 K]) and β = 2 (see Section 3.4). ^bThe uncertainties in the SPIRE measurements are (0.33, 0.16, and 0.086) MJy sr⁻¹ at 250, 350, and 500 μm, respectively, based on background aperture measurement. These uncertainties do not include the SPIRE calibration uncertainties of ∼15%. ^cThe uncertainty in dust mass surface densities have been derived using different temperatures (±1 K) as the fitting errors give unrealistically small numbers. ^dH i surface density uncertainties are typically <10%. ^eSFR surface densities are derived from the values given in Walter et al. (2006; tidal arm) and Kennicutt et al. (2008; NGC 3077).

Download table as:  ASCIITypeset image

Figure 2 shows the SPIRE measurements for NGC 3077 (aperture 1, red curve) and the sum of all detected tidal regions (green curve, boldface aperture numbers in Table 1). A comparison of the two spectral energy distributions (SEDs) directly shows that the 250 μm/500 μm ratio is significantly lower in the tidal feature compared to the main body of NGC 3077, reflecting a lower temperature in the tidal region. Given the lower temperature in the tidal arm and the comparable total fluxes in the SPIRE bands this immediately implies that the dust mass in the tidal feature is larger than in NGC 3077 itself (see more detailed discussion in Section 3.6).

Zoom In Zoom Out Reset image size

We have attempted to use Spitzer MIPS 160 μm imaging to further constrain the SED toward shorter wavelengths. The main body of NGC 3077 is very bright, and we plot the corresponding flux density in Figure 2. Because of coverage and extended cirrus emission, the background of the MIPS 160 μm is however badly behaved at the flux levels of interest, leading to significant error bars in the flux determination. We nevertheless derived a 160 μm flux density for the brightest aperture (no. 10, the error bar is dominated by the background uncertainties)—the complete SED for this aperture is also plotted in Figure 2 for comparison.

We have used a blackbody fit and the Li & Draine (2001) dust emissivity (equivalent to a modified blackbody with β = 2) to derive temperatures and dust masses. The temperature for NGC 3077 (aperture 1) is 30.6 ± 2 K and is in agreement with high temperatures expected for starbursts, and the central temperatures derived for a sub-sample of the KINGFISH galaxies presented by Engelbracht et al. (2010) of T_center = 25.7 ± 1 K. The temperature in the tidal feature is much lower. Based on the SPIRE data only, we derive an average temperature of T = 12.0 ± 1 K. In the brightest aperture of the tidal feature (no. 10) the addition of the 160 μm flux estimate increases the temperature slightly (12.6 K), but is consistent with the "SPIRE-only" value. Changing β to a value of 1.5 increases the temperature by ∼3 K (see dashed SED fits in Figure 2). In the following, we will adopt β = 2 and T = 12.6 ± 2 for the tidal feature apertures but note that the temperature (and the corresponding masses) depends on the exact choice of dust model.

Finding low dust temperatures may not be unexpected as the intensity of the radiation field in the tidal arm is presumably low: the total star formation in the entire tidal feature is only 2.3 × 10⁻³ M_☉ yr⁻¹ (based on Hα observations; Walter et al. 2006). This SFR is consistent with the rate determined in Weisz et al. (2008) through stellar population studies, and GALEX measurements that cover regions 10 and 11, implying that the SFR in the tidal feature was roughly constant over the recent past (few hundred million years). For those apertures that include H ii regions we sum up their Hα luminosities using the values given in Walter et al. (2006) and give the SFR surface densities (averaged over the size of our apertures) in Table 1. We plot the 250 μm/500 μm ratios for these aperatures as a function of the Hα luminosity surface density in Figure 3. SFRs can be estimated using SFR[M_☉ yr⁻¹] = 7.9 × 10⁻⁴² L(Hα)[erg s⁻¹] (Kennicutt 1998, see top x-axis in Figure 3; choosing a Kroupa IMF would decrease the SFRs by ∼30%) and the brightest region outside of NGC 3077 (aperture 10) accounts for ∼50% of the total SFR in the tidal complex (1.3 × 10⁻³ M_☉ yr⁻¹ or a SFR surface density for this aperture of 4 × 10⁻⁴ M_☉ yr⁻¹ kpc⁻²). For reference, single massive stars create H ii regions with Hα luminosities of L_Hα,O7 ≈ 5 × 10³⁶ erg s⁻¹ and L_Hα,O5 ≈ 5 × 10³⁷ erg s⁻¹ (e.g., Devereux et al. 1997), i.e., the total Hα flux in the tidal feature can in principle be created by a few massive stars. Figure 3 shows that the dust temperature does not change as a function of the SFR surface density over two orders of magnitude of the latter. This suggests that the currently observed star formation does not provide the main heating source for the dust on kpc scales and we speculate that other mechanisms may be responsible for the dust heating.

Zoom In Zoom Out Reset image size

Assuming G₀∼ 4 × 10⁻³ for the intergalactic radiation field (Sternberg et al. 2002, note that G₀ = 1 for the Galactic interstellar radiation field), that the temperature of dust goes as T ∼ G^1/(4+β)₀ and β = 2, and scaling from the results by Li & Draine (2001) for the Galactic interstellar radiation field (16 K for a silicate grain, 20 K for a graphite grain, Figure 3 in Li & Draine 2001), we estimate that the dust in the tidal feature should have a temperature of 6–8 K in the intergalactic field. This suggests that other heating sources (the radiation field of NGC 3077, an older stellar population in the feature as seen in the HST imaging by Weisz et al. 2008 and/or shocks) must be partly responsible for heating the dust to our derived temperature of 12 ± 2 K. The 250 μm/500 μm ratio does not change as a function of projected distance to NGC 3077 within our uncertainties but we note that the measurements only span a small range in distances.

From the SED fits discussed above we derive dust mass surface densities for each of our apertures (Table 1) with an average value of 7.4 × 10⁻² M_☉ pc⁻² (β = 2, T = 12.6 K). We derive a total dust mass in the tidal feature (excluding NGC 3077 and upper limits) of ∼(1.8 ± 0.9) × 10⁶ M_☉ (the large error bar includes uncertainties in dust properties, see below). Our simple dust mass estimate is consistent with results from more detailed modeling using updated Draine & Li (2007) models (G. Aniano et al. 2011, in preparation). The dust mass of NGC 3077 is only 1.9 × 10⁵ M_☉ (β = 2 and T = 30.6 K), in reasonable agreement with the mass estimates by Price & Gullixson (1989) of the absorbing central dust clouds based on NIR observations (they report a mass of ∼10⁵ M_☉). Both mass estimates depend on the choice of β and the temperature. For example, if we chose β = 1.5 for the tidal feature, the dust mass would go down by a factor of ∼2 (as the temperature would increase by ∼3 K; Section 3.4). Likewise, lowering the dust temperature in NGC 3077 by 5 K would increase its dust mass by a factor of ∼2. Given these large systematic uncertainties involved we assign a conservative error of 50% to the dust mass of the tidal feature.

For the average H i surface mass densities of the apertures where dust emission was significantly detected, we derive an average value of 11.3 M_☉ pc⁻² (Table 1), i.e., more than two orders of magnitude larger than the dust surface densities. The average dust-to-H i gas ratio is (6.5 ± 3) × 10⁻³, and the ratio does not vary within the errors between apertures and is not a function of H i surface density (the corresponding value for NGC 3077 (aperture 1) is 0.0043). Within the uncertainties this value is close to the average dust-to-gas ratio of M_dust/M$_{\rm H\,{\mathsc {i}}}\approx 0.005$ derived by Draine et al. (2007) for 12 spiral galaxies of the SINGS sample for which Spitzer and SCUBA measurements were available. As noted by Draine et al. (2007), an M_dust/M$_{\rm H\,{\mathsc {i}}}$ ratio between 0.003 and 0.01 is consistent with galactic metallicities between 0.3 and 1 times solar, if a similar fraction of heavy elements is in the form of dust as in the Milky Way. Our derived dust-to-gas ratio thus indicates that the tidal arm is substantially chemically enriched, in agreement with the H ii region metallicities derived by Croxall et al. (2009). For comparison, the M_dust/M$_{\rm H\,{\mathsc {i}}}$ ratio for metal-poor dwarf irregular galaxies in the (more extended) M 81 group of galaxies is up to one order of magnitude lower than found here (Walter et al. 2007).

So far, we have only considered the H i gas. Molecular gas has been mapped in the CO (1–0) transition in the tidal feature over a restricted area (Walter & Heithausen 1999; Heithausen & Walter 2000; Walter et al. 2006). The two main molecular complexes (region numbers 1 and 2 in Heithausen & Walter 2000) are spatially coincident with aperture numbers 10 and 11 in this study. The implied molecular gas masses are dependent on the choice of the H₂-to-CO conversion factor X_CO. Heithausen & Walter (2000) used a conversion factor of X_CO = 8 × 10²⁰ cm⁻² (K km s⁻¹)⁻¹, a factor of ∼4 larger than the Galactic conversion factor. At the time, their choice was driven by the assumption that the tidal region was metal-poor. Given the abundance of dust and dust-to-gas ratios similar to nearby spiral galaxies, as well as the H ii region metallicity measurements (indicating metallicities as high as Galactic; Croxall et al. 2009), it now seems to be more appropriate to use the Galactic conversion factor in this tidal feature. This implies that the masses given in Heithausen & Walter (2000) should be divided by a factor of four. We thus adopt molecular gas masses of ∼1 × 10⁶ M_☉ and ∼4 × 10⁶ M_☉ for our apertures 10 and 11, respectively (comparable to the H₂ mass present in the center of NGC 3077; Meier et al. 2001; Walter et al. 2002a). Averaged over our r = 1 kpc sized apertures this corresponds to a molecular surface density between 1.1 and 0.3 M_☉ pc⁻² (i.e., the interstellar medium, ISM, in the tidal feature is dominated by the atomic gas phase even in the CO-brightest regions.