EDP Sciences logo
Subscriber Authentication Point
Search
Menu
Home All issues Volume 536 (December 2011) A&A, 536 (2011) A17 Full HTML
Planck early results
Free Access
Issue
A&A
Volume 536, December 2011
Planck early results
Article Number A17
Number of page(s) 17
Section Extragalactic astronomy
DOI https://doi.org/10.1051/0004-6361/201116473
Published online 01 December 2011

© ESO, 2011

1. Introduction

Star formation and the exchange and evolution of materials between the stars and the interstellar medium (ISM) are continuous processes that drive the evolution of galaxies. As stars evolve, die, and renew the life cycle of dust and gas, the amount of dust and its distribution in a galaxy has important consequences for its subsequent star formation. Knowing the dust content throughout cosmic history would thus provide clues to the star formation history of the universe as the metallicities evolve. One of the puzzling results that has emerged from the studies of dust SEDs in the early universe is finding very large dust masses in high-redshift galaxies. For example, recent Herschel observations of submillimetre galaxies (SMGs) find excessively large dust masses, and high dust-to-gas mass ratios (D/G) (Santini et al. 2010). These findings contradict the low metallicities measured in the gas. Large dust masses have also been measured for a range of low-metallicity, local universe dwarf galaxies (Dumke et al. 2003; Galliano et al. 2003, 2005; Bendo et al. 2006; Galametz et al. 2009; Grossi et al. 2010; O’Halloran et al. 2010). Our understanding of how dust masses are estimated and what observational contraints are necessary is called into question by the discovery that the dust masses measured in these galaxies appear significantly larger than expected from their metal content.

Excessively large dust masses only seem to be found in low metallicity galaxies to date. The evidence is found in the behaviour of the submillimetre (submm) emission, beyond about 400 − 500μm. From a study of a broad range of metal-rich and metal-poor galaxies, observed with wide wavelength coverage that included submm observations beyond 500μm, a submm excess, beyond the normal dust SED models, was found only for low metallicity systems (Galametz et al. 2010). Such a submm excess is particularly evident in the Magellanic Clouds (Israel et al. 2010; Bot et al. 2010b). The origin of such an excess is still uncertain but several suggestions have been put forward:

  • The submm excess has been modelled as a cold dust compo-nent with a submm emissivity index (β) of β = 1, which suggestsa low dust temperature of  ~10 K (Gallianoet al. 2003, 2005; Galametzet al. 2009, 2010).There are few observational constraints in the submm-wavelength window, hence this cold component is a rather adhoc solution. This description often results in discrepantly largeD/G ratios being found for the low metallicitygalaxies if the estimates of the total gas reservoir are well known.

  • Meny et al. (2007) show that emission by amorphous grains is expected to produce excess emission in the submm wavelength range. As a result, the spectral shape of emission by amorphous grains cannot be reproduced by a single emissivity index over the whole far-infrared (FIR) to millimetre (mm) range, and is expected to flatten at long wavelengths. The intensity of the excess is also predicted to depend strongly on the temperature of the dust grains. Modifying the optical properties of the dust to incorporate the effects of the disordered structure of the amorphous grains shows that the effective submm emissivity index decreases (thus flattening the submm Rayleigh-Jeans slope) as the dust temperature increases. Similarly, an inverse correlation between the dust temperature and the spectral index has been observed in data from the FIR to the submm (e.g. Dupac et al. 2003; Paradis et al. 2010). However, these authors, as well as Shetty et al. (2009a), advise caution in the interpretation of the observed inverse T–β relationship, which is affected by the natural correlation between these two parameters, and requires careful treatment of observational uncertainties. Shetty et al. (2009b) also studied the effect of temperature variations along the line-of-sight on the observed T–β relation. Paradis et al. (2009), using the DIRBE, Archeops and WMAP data, showed that the FIR-submm SEDs of galactic molecular clouds and their neutral surroundings do indeed show a flattening beyond λ ≃ 500μm.

  • Spinning dust emitted in the ionised gas of galaxies has been suggested as the explanation for the radio emission often seen in galaxies (Ferrara & Dettmar 1994). This idea was further explored by Anderson & Watson (1993) and Draine & Lazarian (1998b,a), who characterised the 10 to 100GHz anomalous foreground emission component as radiation from very small spinning dust particles such as polycyclic aromatic hydrocarbons (PAHs). Recent improvements to the model suggest that the peak frequency, which could occur in the submm, depends on many parameters, including the radiation field intensity, the dust size distribution, the dipole moment distribution, and the physical parameters of the gas phases (Hoang et al. 2010; Silsbee et al. 2011; Ysard et al. 2010; Ysard & Verstraete 2010). Spinning dust has been a preferred explanation for the submm excess observed on the global scale in the Magellanic Clouds (Israel et al. 2010; Bot et al. 2010b).

  • Serra Díaz-Cano & Jones (2008) have suggested hydrogenated amorphous carbon as the most likely carbonaceous grain material, rather than graphite. The advantage of using amorphous carbon, particularly in the cases where a submm excess is found, is that amorphous carbons are more emissive and result in a flatter submm slope and smaller dust mass. This has been used with new Herschel observations to model some low metallicity galaxies (Galametz et al. 2010; O’Halloran et al. 2008; Meixner et al. 2010).

  • It has also been suggested that abundance increase of the hot, very small, stochastically heated grains with a low dust emissivity could be the cause of the submm emission when there is a submm excess present (Lisenfeld et al. 2001; Zhu et al. 2009).

Some of the possible causes of the submm excess could be ruled out or constrained if the gas mass estimates were better established. However, estimates of the total gas reservoirs in low metallicity environments have been uncertain, particularly the molecular gas component. Observations of CO in low metallicity galaxies have been a challenge, and the dearth of detected CO has often been interpreted as meaning there is very little mass in molecular gas present in galaxies which are otherwise actively forming massive stars (Leroy et al. 2009). However, relying on CO alone to trace the H2 mass could potentially miss a large reservoir of molecular gas residing outside the CO-emitting region, as indicated by the FIR fine structure observations in the Magellanic Clouds and other low metallicity galaxies (e.g. Poglitsch et al. 1995; Madden et al. 1997; Madden 2000). The presence of a hidden gas mass in low metallicity galaxies is also suggested by dust observations of the Magellanic Clouds (e.g. Bernard et al. 2008; Dobashi et al. 2008; Leroy et al. 2007; Roman-Duval et al. 2010; Bot et al. 2010a).

Most of these studies which find a submm excess in low-metallicity galaxies are derived from models on global galaxy scales, and use data with a wavelength coverage that is limited to λ < 500μm with Herschel or to λ < 870μm for ground-based observations. The Galametz et al. (2010) study demonstrated that dust masses can differ by an order of magnitude depending on the availability of submm constraints on the SED modelling. These require wider wavelength coverage of the submm and millimetre (mm) region and better spatial information, to map out the submm excess and determine the local physical conditions.

The closest low-metallicity galaxies are our neighbouring Large Magellanic Cloud (LMC) and Small Magellanic Cloud (SMC), which have metallicities of 50% and 20% solar metallicity (Z⊙) respectively. They are ideal laboratories for studying the ISM and star formation in different low metallicity environments. Recent studies with Spitzer (SAGE: Meixner et al. 2006; Bolatto et al. 2007) and Herschel (HERITAGE: Meixner et al. 2010) have mapped out the the temperature and dust mass distribution (Bernard et al. 2008; Gordon et al. 2009, 2010; Leroy et al. 2007). Recent Herschel studies of the LMC already point to a possible submm band excess and an excessive dust mass, which may be suggesting the presence of molecular gas not probed by CO. This could be explained by the presence of amorphous carbon grains (Roman-Duval et al. 2010; Meixner et al. 2010; Gordon et al. 2010). It was difficult to be conclusive about the cause of the excess seen in the Herschel observation because of the lack of longer wavelength coverage. Global excess mm and submm emission in the LMC and SMC was indisputably shown by Bot et al. (2010b) and Israel et al. (2010), using broader wavelength coverage that included the submm to cm observations of the Top-Hat balloon experiment, WMAP and COBE. Their results favour spinning dust as the explanation for the submm and mm excess, but require additional spatial coverage to be conclusive. Planck1 observations have both wide wavelength coverage and the spatial resolution to locate the submm emission within the LMC and SMC, which will help determine the origin of the submm excess. In the framework of the preliminary analysis carried out here, we focus our interest on dust emission from the galaxies themselves, which we isolate by subtracting other unrelated components, instead of carrying out a full multi-component analysis exploiting all spatial information available in the data.

2. Observations

2.1. Planck data

The Planck first mission results are presented in Planck Collaboration (2011a) and the in-flight performances of the two focal plane instruments, the HFI (High Frequency Instrument) and the LFI (Low Frequency Instrument), are given in Planck HFI Core Team (2011a) and Mennella et al. (2011) respectively. The data processing and calibration of the HFI and LFI data used here is described in Planck HFI Core Team (2011b) and Zacchei et al. (2011) respectively.

thumbnail Fig.1

Planck total intensity data for the LMC (left) and SMC (right) at 857 (top), 100 (middle) and 28.5GHz(bottom) at full resolution. The top panels are shown in log scale. The circle in the top panels shows the region used to extract average SEDs.

Figure 1 shows the total intensity maps, observed toward the LMC and SMC at three HFI and LFI frequencies. Both galaxies are well detected at high frequencies. Around 100GHz, their emission can barely be distinguished from CMB fluctuations. At lower frequencies, the contrast between the galaxies’ emission and the CMB fluctuations becomes larger again. Note that the apparent variation of the noise level at a constant right-ascension value across the LMC is real and is due to the LMC being positioned at the edge of the Planck deep field.

The Planck DR2 data have had the CMB fluctuations removed in a way which is inappropriate for detailed examination of foreground sources like the LMC and SMC. We therefore use the original data before CMB subtraction (version DX4 of the Planck-HFI and Planck-LFI data) and perform our own subtraction of the CMB fluctuations, based on a local ILC (see Sect. 2.3.2).

We use the internal variance (σII2\hbox{$\sigII^2$}) provided with the Planck data, which represents the white noise on the intensity. The LMC and SMC have been observed many times, as the Planck scanning strategy (Planck Collaboration 2011a) covers the region close to the ecliptic pole repeatedly. In particular, the LMC is located at the boundary of the Planck deep field. Its eastern half has a much higher signal to noise ratio than its western half. We assume the absolute uncertainties from calibration given in Planck HFI Core Team (2011b) and Zacchei et al. (2011) for HFI and LFI respectively and summarised in Table 1.

Table 1

Characteristics of the FIR/submm data used in this study.

2.2. Ancillary data

2.2.1. Hi LMC/SMC data

To trace the atomic gas in the Magellanic Clouds, we used Hi maps in the 21 cm line, obtained by combining data from the Australia Telescope Compact Array (ATCA) and the Parkes single dish telescope. For the LMC, this data was obtained by Kim et al. (2003) and Staveley-Smith et al. (2003) and covers 11.1° × 12.4° on the sky. The spatial resolution is 1′, corresponding to a physical resolution of about 14.5 pc at the distance of the LMC. For the SMC, the data were obtained by Staveley-Smith et al. (1997) and Stanimirovic et al. (1999). The area covered is a 4.5° × 4.5° region. Hi observations of the SMC tail (7° × 6°) were obtained by Muller et al. (2003) and combined with the observations taken in the direction of the SMC. The spatial resolution is 98″, corresponding to 30 pc at the distance of the SMC.

2.2.2. Hi Galactic data

The LMC and SMC are located at galactic latitudes bII = −34° and bII = −44° respectively. They can therefore suffer from significant contamination by Galactic foreground emission, which has to be removed from the IR data. In order to account for the Galactic foreground emission, we used Galactic Hi column density maps.

For the LMC, the Hi foreground map was constructed by Staveley-Smith et al. (2003) by integrating the Parkes Hi data in the velocity range from  −64 < vhel < 100kms-1, which excludes all LMC and SMC associated gas (v > 100kms-1) but includes essentially all Galactic emission. The spatial resolution is 14′. For the SMC, a map of combined ATCA and Parkes data was build by integrating the Galactic velocities by E. Muller (priv. comm.). The resolution is 98″.

These maps show that the Galactic foreground across the LMC is as strong as NH = 1.3 × 1021H cm-2, with significant variation across the LMC, particularly in a wide filamentary structure oriented southwest to northeast. The Galactic foreground across the SMC is weaker, with values NH ≃ 3.2 × 1020H cm-2 but the associated FIR-submm emission is non-negligible since the SMC emission is weaker than that of the LMC, due to its lower dust and gas content.

These foreground maps are used to subtract the foreground IR emission from the IR maps, using the emissivity SED described in Sect. 2.3.3.

2.2.3. CO data

The 12CO(J = 1 → 0) molecular data used in this work was obtained using the NANTEN telescope, a 4-m radio telescope of Nagoya University at Las Campanas Observatory, Chile (see Fukui et al. 2008). The observed region covers about 30 square degrees where CO clouds were detected in the NANTEN first survey (e.g. Fukui et al. 1999; Mizuno et al. 2001; Fukui et al. 2008). The observed grid spacing was 2′, corresponding to about 30 and 35pc at the distance of the LMC and SMC, while the half-power beam width was 2.6′ at 115GHz.

We used the CO maps of the Magellanic Clouds to construct an integrated intensity map (WCO), integrating over the full vlsr range of the data (100 < vlsr < 400kms-1 for about 80 % of the data, while the remaining 20 % had a velocity range of about 100kms-1 covering the Hi emitting regions.)

2.2.4. Hα data

In order to estimate the free-free contribution to the millimetre fluxes, we used the continuum-subtracted Hα maps from the Southern H-Alpha Sky Survey Atlas (SHASSA, Gaustad et al. 2001) centred on the Magellanic Clouds.

2.2.5. FIR-submm data

thumbnail Fig.2

Error due to the ILC CMB subtraction derived from Monte-Carlo simulations for both LMC and SMC at 100GHz, as a function of the patch size. The values shown are the difference between the recovered and the input CMB divided by the simulated LMC and SMC, both integrated in a 4° ring.

We used the following FIR-submm ancillary data.

  • IRAS–IRIS (improved reprocessing of theIRAS survey) 100μm data, in order to constrain the dusttemperature. The characteristics of these data, including thenoise properties, were taken from Miville-Deschênes &Lagache (2005) and are summarised inTable 1

  • WMAP 7 yr data. The characteristics of these data, including the noise properties, were taken from Bennett et al. (2003) and Jarosik et al. (2011) and are summarised in Table 1.

thumbnail Fig.3

SMC (left) and LMC (right) total intensity maps before (top) and after (bottom) CMB subtraction in the 70.3GHz band at the 13.01 arcmin resolution.

2.3. Additional data processing

2.3.1. Common angular resolution and pixelisation

For data already available in the HEALPix (Górski et al. 2005) format (e.g., WMAP), we obtained the data from the Lambda web site (http://lambda.gsfc.nasa.gov/). For data not originally presented in the HEALPix format, the ancillary data were brought to the HEALPix pixelisation, using a method where the area of the surface of intersection between each HEALPix and FITS pixel of the original data was used as a weight to regrid the data. The HEALPix resolution was chosen so as to match the Shannon sampling of the original data at resolution θ, with a HEALPix resolution set so that the pixel size is  < θ/2.4. The ancillary data and the description of their processing will be presented in Paradis et al. (in prep.).

All ancillary data were then smoothed to an appropriate resolution by convolution with a Gaussian smoothing function with appropriate FWHM, using the smoothing HEALPix function. They were brought to a pixel size matching the Shannon sampling of the final resolution.

2.3.2. CMB subtraction

The upper panel in Fig. 3 shows the total intensity maps observed toward the LMC and SMC before CMB subtraction in the LFI 70.3GHz chanel. The amplitude of CMB fluctuations is of the order of that of the diffuse emission of the galaxies and it is clear that, in order to study the integrated SED of the LMC and SMC galaxies, an efficient CMB subtraction has to be performed.

The standard CMB–subtracted maps produced by the data processing centre (DPC) (Planck HFI Core Team 2011b) were not used in this analysis. They were processed by a needlet internal linear combination method (NILC) (Planck HFI Core Team 2011b) that left a significant amount of foreground emission in the CMB estimate towards the LMC and the SMC.

For this reason we have subtracted an estimate of the CMB optimised locally for the LMC and SMC regions, as described below. This CMB component was reconstructed through a classical ILC by means of Lagrange multipliers (Eriksen et al. 2004). 12° × 12° patches around the LMC and SMC were extracted from the HFI CMB frequency channels maps (100, 143, 217 and 353GHz) reduced to a common resolution (10 arcmin), in units of KCMB. The CMB component obtained on these patches clearly contains less LMC and SMC residual than the standard HFI DPC CMB and therefore will affect the SED determination less. The lower panel in Fig. 3 shows the maps after CMB subtraction in the LFI 70.3GHz channel.

thumbnail Fig.4

Average foreground SED in the direction of the LMC and SMC (lower curve) including IRAS–IRIS (light blue), Planck-LFI (red), Planck-LFI (green) and WMAP data (dark blue), compared to the SED of the high latitude low column-density MW SED (upper curve) derived in Planck Collaboration (2011o), which has been scaled up by a factor of 10 for clarity.

thumbnail Fig.5

Integrated SEDs of the LMC (left) and SMC (right) before and after CMB subtraction. The black points and model are taken from Bot et al. (2010b). The red symbols show the SEDs derived from the DIRBE, IRAS, and WMAP data before CMB subtraction. The blue symbols show the same after CMB subtraction.

We performed Monte-Carlo simulations in order to estimate the error induced on the SED by our CMB removal. The LMC and SMC were simulated as a sum of two correlated components. The spatial template for the first component is the IRAS–IRIS 100μm map. The spatial template for the second component is the 545GHz LMC or SMC Planck map. Their correlation coefficients are 83% for the LMC and 92% for the SMC. We normalised their fluxes inside a 4° ring to the value of a typical LMC or SMC dust component and a typical millimetre excess at each HFI CMB frequency, respectively. 200 independent realisations of a WMAP 7 yr best-fit CMB (Komatsu et al. 2011) and of nominal inhomogeneous HFI white noise were added to the synthetic LMC and SMC on patches of varying sizes. For each of these simulations, for each size, our ILC is performed and compared with the input CMB. We estimate the error due to the CMB subtraction in both the LMC and the SMC by comparing the residual in the CMB map to the emission of the two components at a given frequency. Results at 100GHz are displayed in Fig. 2. For both the LMC and the SMC, the error increases at small and large patch sizes and an optimal patch size with respect to the CMB subtraction can be found at 5° for the LMC and 13° for the SMC. The narrower the patch, the lower is the contribution of the CMB to the total variance which is to be minimised in the ILC. On the other hand, when their sizes increase, patches include foreground emission which is uncorrelated with the galaxies and has a different spectrum, and thus both small and large patches contribute to the total variance which must be minimised.

For the 12° × 12° patches used in the following analysis, we estimate the error due to CMB subtraction as 10.38μKCMB and 28.2μKCMB for the LMC and SMC respectively. In terms of the fraction of the total galaxy brightness, these correspond to 8.1, 6.2, 2.3 and 0.3% for the LMC and 12.1, 10.7, 6.4 and 1.4% for the SMC at 100, 143, 217and 353GHz, respectively.

Table 2

LMC (Cols. 2–4) and SMC (Cols. 5–7) SEDs averaged in a circular region for each galaxy.

2.3.3. Galactic foreground subtraction

The Milky Way (MW) emission is non-negligible compared to the emission of the LMC and SMC. We remove this contribution at all wavelengths using the MW Hi template described in Sect. 2.2.2. We first computed the correlation of all FIR-submm data with this template, in the region where both the MW Hi template and the CMB estimate are available. We excluded a circular region centred on each galaxy (centre coordinates taken as α2000 = 05h18m14.8s, δ2000 = − 68°26′34.5′′ and α2000 = 00h53m59.6s, δ2000 = − 72°40′16.1′′ for the LMC and SMC respectively) with radius 4.09° and 2.38° for the LMC and SMC respectively. The spectral distribution of this correlation factor, taken to represent the SED of the MW foreground is shown in Fig. 4. The SED is compared to that of the high galactic latitude reference region used in Planck Collaboration (2011o) in Fig. 4. It can be seen that the two SEDs are similar, although the MW foreground towards the LMC and SMC appears slightly colder and has a relatively stronger non-thermal component in the millimetre wavelength range. We subtracted the MW foreground from all the data using this SED multiplied by the MW Hi template over the full map extent. Note that the median foreground Hi integrated intensities over the LMC and SMC are  ≃ 297K and  ≃ 172K, which correspond to a brightnesses of 3.0MJysr-1 and 1.7MJysr-1 respectively at 857GHz. This is of the same order as the average brightness of the galaxies at that frequency (see Table 2). However, most of the MW emission is canceled when subtracting a local background around the galaxies and the differential correction due to the spatial structure of the MW foreground then accounts for about 0.4% and 21% of the LMC and the SMC brightness respectively.

3. Integrated SEDs

The integrated SEDs of the LMC and SMC before and after CMB subtraction are shown in Fig. 5 (red and blue diamonds, respectively). They were computed by averaging values in the circular area around each galaxy defined in Sect. 2.3.3, and subtracting an estimate of the sky off the source taken in an annulus with radius 1°. They are compared to the data taken from Israel et al. (2010) and Bot et al. (2010b), which were integrated over the same region. It can be seen that the flux observed in the HFI and LFI bands is consistent with that obtained by previous studies in this wavelength range before subtraction of the CMB fluctuations. A model of thermal dust, free-free and synchrotron emission is fitted to the data as follows. As in Bot et al. (2010b), the thermal dust emission is adjusted according to the Draine & Li (2007) dust model. The free-free emission is deduced from the Hα integrated flux, using the expression from Hunt et al. (2004), assuming an electronic temperature Te = 104 K, the ratio of ionised helium to hydrogen nHe+/nH+=0.087\hbox{$n_{\rm He^+}/n_{\rm H}^+=0.087$}, and no extinction. The synchrotron emission was fitted to the radio data from the literature as in Israel et al. (2010). The combined dust, free-free and synchrotron emission is shown by the blue line in Fig. 5. It can be seen from Fig. 5 that the SED after subtraction of the CMB fluctuations is in fact compatible with no millimetre emission excess for the LMC, for this particular model. However, the CMB-subtracted SED still shows a significant excess for the SMC, with most data points above λ ≃ 1mm being in excess over the model by more than 5σ, leading to an overall significance of the excess of about 50σ.

The CMB subtraction removes part of the millimetre excess in both galaxies. This shows that CMB fluctuations behind the LMC and the SMC average out to a small but positive contribution when integrated over the extent of the galaxies. We note that some excess emission remains in the SMC regardless of the dust model used and the assumptions made on the free-free or synchrotron emission. The shape and intensity of the millimetre excess can change, but we could not find a solution where the SMC SED is explained purely by thermal dust emission, free-free and synchrotron radiation. We emphasise also that the dust model, used to reproduce the dust emission up to the sub-millimetre wavelength range, assumes components heated by a radiation field 10 times lower than the solar neighbourhood radiation field. Compared to other nearby galaxies, this result is rather extreme (Draine et al. 2007; Bot et al. 2010b).

The integrated SEDs of the LMC and SMC after CMB and foreground subtraction are compared in Fig. 6. These SEDs were computed in the same integration region as those in Fig. 5. The comparison in Fig. 6 shows that the SMC SED is flatter than the LMC SED in the submm, while when the SEDs are normalised in the FIR, they are similar above 10 mm, where the emission is presumably dominated by free-free and/or spinning dust (see Sect. 5).

The SEDs at various stages of the background and foreground subtraction are given in Table 2 for the LMC and SMC. The uncertainties given include the contribution from the data variance combined for the integration region, the data variance combined for the background region, and the absolute calibration uncertainties, using the values given in Table 1. They also include the noise resulting from the background (CMB) subtraction and from the foreground MW subtraction and the free-free removal. All uncertainty contributions were added quadratically.

4. Dust temperature and emissivity

4.1. Temperature determination

As shown by several previous studies (e.g. Reach et al. 1995; Finkbeiner et al. 1999; Paradis et al. 2009; Planck Collaboration 2011o,t), the dust emissivity spectrum in our Galaxy cannot be represented by a single dust emissivity index β over the full FIR-submm domain. The data available indicate that β is usually steeper in the FIR and flatter in the submm, with a transition around 500μm (see Paradis et al. 2009; Planck Collaboration 2011t). This is likely to be the case also for the LMC and the SMC. The dust temperature derived will depend on the assumptions made about β, since these two parameters are somewhat degenerate in χ2 space. Note that some of the past attempts at constraining the equilibrium temperature of the large grains in the LMC used the IRAS 60μm emission. Emission at 60μm is highly contaminated by out-of-equilibrium emission from very small grains (VSGs) and this is even more the case in the Magellanic Clouds, due to the presence of the 70μm excess (Bot et al. 2004; Bernard et al. 2008). Combining the IRAS 60μm and 100μm data therefore strongly over-estimates the temperature and accordingly under-estimates the abundances of all types of dust particles. A good sampling at frequencies dominated by big grain (BG) emission became possible using the combination of the IRAS and Spitzer data (Leroy et al. 2007; Bolatto et al. 2007; Bernard et al. 2008; Sandstrom et al. 2010).

As the dust temperature is best derived from the FIR data, we limit the range of frequencies used in the determination to the FIR, which limits the impact of potential changes of the dust emissivity index β with frequency. In the determination of the dust temperature (TD), we used the IRAS–IRIS100μm map and the two highest HFI frequencies at 857and 545GHz. Temperature maps were derived at the common resolution of the 3 bands used (5′) and at lower resolution for further analysis. In each case, the emission was computed in the photometric channels of the instruments used (IRAS, Planck, and WMAP), including the colour corrections, using the actual transmission profiles for each instrument, and following the flux convention description given in the respective explanatory supplements.

In order to derive the thermal dust temperature, we use the same strategy as described in Planck Collaboration (2011o). To minimise computation time, the predictions of the model were tabulated for a large set of parameters (TD, β). For each map pixel, the χ2 was computed for each entry of the table and the shape of the χ2 distribution around the minimum value was used to derive the uncertainty on the free parameters. This includes the effect of the data variance σII2\hbox{$\sigII^2$} and the absolute uncertainties.

We explored several options for deriving maps of the apparent dust temperature, which are summarised in Table 3. We first fitted each pixel of the maps with a modified black body of the form Iν ∝ νβBν(TD) in the above spectral range (method referred to as “free β” in Table 3). The median values of TD and β derived using this method are given in the first line of Table 3. This led to median β values of βLMC ≃ 1.5 and βSMC ≃ 1.2 for the LMC and SMC respectively. Note that the value for the LMC is consistent with that derived using a combination of the IRAS and Herschel data by Gordon et al. (2010). However, inspection of the corresponding maps show correlated variations of the two parameters which were also correlated with the noise level in the maps. This suggests that the spurious values probably originated from the correlation in parameter space and the presence of noise in the data, particularly in low brightness regions of the maps.

Table 3

Dust temperature, β values and fit reduced χ2 for various methods experimented to derive the temperature maps.

thumbnail Fig.6

Integrated SEDs of the LMC (solid) and SMC (dashed) after CMB and galactic foreground subtraction, including data from Planck-HFI (red), Planck-LFI (green), IRAS–IRIS (light blue) and WMAP (dark blue). The uncertainties shown are  ± 3σ. The SMC SED was scaled by a factor 4 providing normalization at 100μm.

We then performed fits of the FIR emission using the fixed β values, with the β values given above for the two galaxies (method referred to as “fixed β” and marked with a in Table 3). Although the median reduced χ2 was slightly higher than for the “free β” method, the temperature maps showed fewer spurious values, in particular in low brightness regions. This resulted in a more coherent distribution of the temperature values than in the “free β” case. Since we later used the temperature maps to investigate the spectral distribution of the dust optical depth, and the dust temperature is a source of uncertainty, we adopted the “fixed β” maps in what follows. The corresponding temperature maps for the LMC and SMC are shown in Fig. 7.

The calculations were also carried out for β = 1.8, which is the average Galactic value (Planck Collaboration 2011o,t), in order to be able to derive dust emissivity values under the same assumption as in the MW. We also experimented using the “standard” β = 2 value.

thumbnail Fig.7

Upper panels: dust temperature maps for the LMC (left) and SMC (right) computed from the foreground subtracted maps using the IRAS–IRIS100μm, HFI857 and 545GHz maps, using a fixed βLMC = 1.5 and βSMC = 1.2. Lower panels: relative uncertainties on the dust temperature at the same resolution, expressed as percentages.

4.2. Angular distribution of dust temperature

4.2.1. LMC

The temperature map derived here for the LMC shows a similar distribution to the one derived from IRAS and Spitzer data by Bernard et al. (2008). The highest temperatures are observed toward 30-Dor and are of the order of 24 K. The large scale distribution shows the existence of an inner warm arm, which follows the distribution of massive star formation as traced by known Hii regions. Figure 8 shows that the warm dust at TD > 20 K in the LMC is reasonably well correlated with Hα emission, indicating that it is heated by the increased radiation field induced by massive star formation. Two regions departing from the correlation are visible to the SW of the 30-Dor at α2000 = 05h40m, δ2000 = −70°30′ and α2000 = 05h40m, δ2000 = −72°30′. The first region was already identified in Bernard et al. (2008) who showed that this low column density region was unexpectedly warm given that no star formation is taking place in this area. They proposed that this could be an artefact of the IRAS–IRIS 100μm used in the analysis, possibly due to IRAS gain variations across the bright 30-Dor source. However, the analysis carried out here and the comparison between temperature maps obtained with and without the MW background subtraction suggests that the temperature measurement has been affected by the low surface brightness. This emphasises the necessity of correcting for temperature bias at low brightness. The second region is close to the edge of the map and is probably also affected by low surface brightness.

thumbnail Fig.8

Comparison between the dust temperature map of the LMC with Hα (top) and CO emission (bottom). The CO contours are at 0.5, 2, 4 and 10 Kkms-1. Hα contours are at 1, 10, 50, 100, 500 and 1000 Rayleigh. The thick line shows the edge of the available CO surveys.

The Planck data however reveals cold areas in the outer regions of the LMC, which were not seen using the Spitzer data. This is essentially due to the large coverage of the Planck data with respect to the limited region which was covered by the Spitzer data. The south of the LMC exhibits a string of cold regions with TD < 20 K. These regions correspond without exception to known molecular clouds when they fall in the region covered by the CO survey. They also correspond to peaks of the dust optical depth as derived in the following sections. Similarly, a set of cold regions exist at the northwest periphery of the LMC. The comparison with the distribution of CO clouds is shown in Fig. 8. As already noticed in Bernard et al. (2008), there is no systematic correlation between cold dust and the presence of molecular material, at least in the inner regions of the LMC. This was confirmed by the statistical characteristics of the IR properties of LMC molecular clouds established by Paradis et al. (2010), which showed no systematic trend for molecular regions to be colder than their surrounding neutral material. However, toward the outer regions of the LMC, molecular clouds systematically appear to show a decrease in the dust temperature. This difference may be due to the absence of star formation activity in the outer regions and/or to less mixing along the line of sight.

Gordon et al. (2010) derived a dust temperature map from the Herschel data (HERITAGE program) for a fraction of the LMC observed during the science demonstration phase (SDP). They used the 100μm to 350μm IRAS and Herschel bands to constrain the temperature. They found that the temperatures derived this way only differ from those derived by Bernard et al. (2008) by up to 10%. The SDP field covered an elongated region across the LMC roughly oriented north–south. This strip crossed the warm inner arm, which is also clearly seen in their map. However, we stress that, in their study, a gradient was removed along the strip based on the values at the edges. The fact that we detect significantly colder than average dust in the outer cold arm of the LMC underlines the need to take those variations into account when subtracting background emission in the Herschel data.

4.2.2. SMC

thumbnail Fig.9

Comparison between the dust temperature map of the SMC with Hα (top) and CO emission (bottom). The CO contours are at 0.5, 1 and 1.5 Kkms-1. Hα contours are at 1, 5, 30 and 100 Rayleigh. The thick line shows the edge of the available CO surveys.

thumbnail Fig.10

Upper panel: map of the dust optical depths of the LMC at HFI 217GHz. Units are 104 × τ. Lower panel: map of the dust optical depth relative uncertainty of the LMC at HFI 217GHz in percent. Black pixels in the maps are masked and have relative uncertainties larger than 50%.

The temperature map derived for the SMC shows more moderate temperature variations than those observed in the LMC but peaks corresponding to well known HII regions are clearly identified, and the overall dust temperature correlates with Hα emission, as shown in Fig. 9. In particular, the massive star forming region SMC-N66 (Henize 1956) (α2000 = 00h59m27.40s, δ2000 = −72°10′11′′) corresponds to the highest temperature (~20 K) in the SMC. Other well known star forming regions like SMC-N83/84 (01h14m21.0s,  −73°17′12′′), N81 (01h09m13.6s,  −73°11′41′′), N88/89/90 (in the wing: 01h24m08.1s,  −73°08′55′′) and DEM S54 (at the centre of the main bar: 00h50m25.9s,  −72°53′10′′) appear as temperature peaks compared to the surroundings. In contrast, the infrared-bright star forming region N76 (located at the northeast of N66: 01h04m01.2s,  −72°01′52′′) and southwest star-forming complex do not stand out. This could be due to the presence of an extended warm component that we observe in the main bar and that seems spatially related to the diffuse Hα emission.

4.3. Optical depth determination

The optical depth is derived from: τ(ν)=IνBν(Td),\begin{equation} \tau(\nu)=\frac{I_\nu}{B_\nu(T_{\rm d})}, \end{equation}(1)where Bν is the Planck function. We used resolution-matched maps of TD and Iν and derived τ maps at the various resolutions of the data used here. The uncertainty on τ (Δτ) is computed as: Δτ(ν)=τ(σII2Iν2+(δBν(TD)δTΔTDBν(TD))2)1/2.\begin{equation} \Delta \tau(\nu)=\tau \left( \frac{\sigII^2}{I_\nu^2} + \left( \frac{\delta B_\nu(\Td)}{\delta T} \frac{\Delta \Td}{B_\nu(\Td)}\right)^2\right)^{1/2}. \end{equation}(2)The optical depth and optical depth uncertainty maps derived at 857GHzare shown in Figs. 10 and 11 for the LMC and the SMC respectively.

thumbnail Fig.11

Upper panel: map of the dust optical depths of the SMC at HFI 217GHz. Units are 104 × τ. Lower panel: map of the dust optical depth relative uncertainty at HFI 217GHz in percent. Black pixels in the maps are masked and have relative uncertainties larger than 50%.

5. Discussion

5.1. The millimetre excess

Bot et al. (2010b) (see Fig. 5) found that the integrated SED of the LMC and more noticeably of the SMC showed excess millimetre emission with respect to a dust and free-free/synchrotron model based respectively on the FIR and the radio data. They investigated several causes for this excess. They noticed that the excess had precisely the colours of CMB fluctuations. They performed simulations by placing the LMC and SMC structure at various positions over a CMB simulated sky, and concluded that a CMB origin was unlikely, but not excluded. Other proposed origins for the excess included the presence of very cold dust, spinning dust or large modifications of the optical properties of thermal dust.

Very cold dust (TD ≃ 5−7 K) has been advocated to explain the flattening of the millimetre emission observed in more distant low metallicity galaxies (e.g. Galliano et al. 2005). However, this has usually led to very large masses, and the existence of such very cold dust remains controversial and difficult to understand in low metallicity systems where the stronger star formation rate and the lower dust abundances prevent efficient screening from UV photons.

Bot et al. (2010b) applied spinning dust models to fit the SMC and LMC SEDs and found a plausible match. However, since the observed excess peaked at a significantly higher frequency than observed for spinning dust in other regions (e.g. Planck Collaboration 2011p), their fit required extreme density and excitation conditions for the small dust particles.

Using the two level system (TLS) model by Meny et al. (2007) for the long wavelength emission of amorphous solids also proved plausible for the LMC, but a convincing fit could not be found for the SMC, essentially because the model could not reproduce the shape of the excess.

The study carried out here, which takes advantage of the Planck measurements to constrain the CMB foreground fluctuations towards the two galaxies, shows that part of the excess observed toward the SMC cannot be accounted for by the fluctuations of the background CMB, as discussed in Sect. 2.3.2, but the intensity of the excess has been greatly reduced compared to that found by Bot et al. (2010b).

To give more insight into this excess, we built a map to trace the spatial distribution of the excess emission in the SMC. To do this, we applied the fitting procedure performed for the integrated SEDs of the Magellanic Cloud, to each point of the SMC at the angular resolution of the LFI lowest frequency channel. The SEDs were build at each point using the CMB and foreground–subtracted data. To model the SEDs, we assumed that, at each point, the SEDs are dominated by free-free and dust emission up to the longest wavelengths covered by Planck, following what is observed for the integrated emission of the SMC (see Fig. 5). The free-free component at each wavelength was extrapolated from the Hα emission (as in Sect. 3), assuming that the extinction in the SMC is negligible. The thermal dust emission was then fitted to the data points at λref < 550μm, using the Draine et al. (2007) dust model. The millimetre excess was then defined as the difference between the data and the dust-and-free-free model. The resulting spatial distribution of the excess emission at 100GHz (3mm) is shown in Fig. 12. This shows that the peak of the excess is located at the southwest tip of the bar. This region also corresponds to the maximum of the optical depth derived from the FIR and shown in Fig. 11, and the overall excess spatial distribution is consistent with being proportional to the dust column density.

5.2. FIR dust emissivity

The wavelength dependence of the average dust optical depth in the LMC and SMC is shown in Fig. 13. The spectral index of the emissivity in the FIR is consistent with β = 1.5 and β = 1.2 for the LMC and the SMC respectively. This value for the LMC is consistent with findings by Gordon et al. (2010) using the Herschel data. We see no hint of a change of the spectral index with wavelength for the LMC. In contrast, the SMC SED clearly flattens at λ > 800μm to reach extremely flat β values around λ = 3 mm. The dust emissivities as interpolated using the power laws shown in Fig. 13 can be compared to the reference value for the solar neighbourhood of τ/NH = 10-25cm2 at 250μm (Boulanger et al. 1996). This comparison indicates lower than solar dust abundances for the two galaxies, by about 1/2.4 and 1/13 respectively. This is in rough agreement with the metallicity for the LMC and significantly lower than the metallicity for the SMC.

thumbnail Fig.12

Map of the millimetre excess in the SMC at 3mm computed at the resolution of the LFI–28.5GHz channel, after free-free subtraction. The contours show the Hα distribution.

thumbnail Fig.13

Average dust optical depth of the LMC (black) and SMC (blue) obtained from the CMB, foreground and free-free subtracted SEDs. The average is taken in the same regions as Fig. 5. The SMC SED was normalised (multiplied by 6.31) to that of the LMC at 100μm. The square symbol (green) shows the reference value for the solar neighbourhood by Boulanger et al. (1996). The dashed lines show τ ∝ ν1.5 and τ ∝ ν1.2 normalised at 100μm. Error bars are  ± 3-σ.

thumbnail Fig.14

Fit of the SMC SED using the Finkbeiner model (upper left), the TLS model (upper right), the spinning dust model (lower left) and a combination of TLS and spinning dust models (lower right). In the lower panels, the squares represent the flux in each band predicted by the best model (the blue line).

Gordon et al. (2010) found hints of such excess emission around 500μm in the LMC Herschel data, but this remained within the current calibration uncertainties of the Herschel Spectral and Photometric Imaging REceiver (SPIRE) instrument. We also observe that the 500μm optical depth of the LMC is slightly higher (by 8.4%) than the power law extrapolation shown in Fig. 13, but this is only marginally larger than the 1σ uncertainty at this wavelength. In any case, data points at longer wavelengths do not support the existence of excess emission in addition to a β = 1.5 power law for the LMC.

5.3. Possible interpretation of the excess

In this section, we attempt to fit the SED of the SMC using various models which have been proposed to explain excess submm emission, in addition to a single dust emissivity power-law for thermal dust. Figure 14 shows several fits to the FIR-Submm SED of the SMC.

The first fit uses the Finkbeiner et al. (1999) model ♯7 (see their Table 3), which was designed to explain the flatter-than-expected emission spectrum of our Galaxy as observed by FIRAS. Here we use this model to assess the possibility that the millimetre excess is due to very cold dust. The model was fitted using the same β values as in Finkbeiner et al. (1999) for the two dust components (βwarm = 2.6, βcold = 1.5). It also assumes the same type of relationship between the cold and warm dust temperatures, which in the model reflects the fact that both dust species are subjected to the same radiation field. However, we allow the IR/optical opacity ratio (qcold/qwarm) to vary, so that the ratio between the warm and cold dust temperatures is free to vary. We also leave the cold dust component abundance (fcold) and the warm dust temperature (Twarm) as free parameters. The best fit values are found for Twarm = 16.4K qcold/qwarm = 168.2, fcold = 8.7 × 10-3. For these parameters, the temperature of the cold component is Tcold = 5.9K. This cold temperature is needed to reproduce part of the millimetre excess but is much colder than that obtained by Finkbeiner et al. (1999) for the Galaxy (Tcold = 9.6K). It is obtained at the expense of strongly increasing the IR/optical opacity ratio for the cold component (qcold/qwarm increased by a factor 15), which controls the Tcold/Twarm ratio. The mass fraction of the cold component derived here, fcold, is about 4 times lower than that derived for the MW in Finkbeiner et al. (1999), which compensates for the increased qcold/qwarm. It is apparent that, despite invoking a much larger IR/optical opacity ratio for the cold particles, such a model has difficulties producing the submm excess above about λ = 2 mm.

The second fit employs the TLS model developed by Meny et al. (2007). The fit was obtained by minimizing χ2 in the range 100μm < λ < 5mm against the following 3 parameters: TD the dust temperature, lcor the correlation length of defects in the material and A the density of TLS sites in the material composing the grains. The values derived for these parameters are TD = 18.9 K, lcor = 12.85 nm and A = 7.678. The reduced χ2 for these parameters is χ2 = 2.56. Compared to the best values found from Paradis (2007) for the MW (TD = 17.9 K, lcor = 12.85 nm, A = 2.42), this indicates dust material with more TLS sites (more mechanical defects) but a similar defect correlation length compared with the dust dominating the MW emission. Note that a grey-body fit of the same SED over the same frequency range leads to TD = 18.9 K and β = 0.74 and a reduced χ2 = 6.12, showing that the TLS model reproduces the data better than a single grey-body fit.

The third fit uses spinning dust. In this case, the thermal dust emission was reproduced using the Draine & Li (2007) dust model. The remaining excess was then fitted with the spinning dust model described in Silsbee et al. (2011). Spinning dust emission has been proved to be sensitive to the neutral and ionised gas densities and to the size distribution of dust grains. The radiation field and the size distributions were taken to be the same as for the Draine & Li (2007) model. The relevant gas parameters were computed with CLOUDY (Ferland et al. 1998): we took the parameters from the optically thin zone of isochoric simulations. In both cases, we took into account the lower metallicity and dust grains abundance with respect to the MW. The electric dipole moment distribution of grains was taken as in Draine & Lazarian (1998b)2. The fit was obtained by adding two components according to the PDR fraction inferred from the thermal emission model: a diffuse medium with nH = 30cm-3 and 100% of the PAH mass expected from the IR modelling, and a denser medium with nH = 5000cm-3 and 82% of the expected PAH mass.

The fourth fit is a combination of the TLS model and the spinning dust model presented above.

It is clear from Fig. 14 that the model combining TLS and spinning dust is the only one of the proposed models giving a satisfactory fit to the millimetre excess observed in the SMC over the whole spectral range. In addition, it alleviates the need for using more PAH than allowed by the NIR emission to produce the required level of spinning dust emission observed. Alternative models, however, cannot be excluded. For instance, very large grains (a > 30μm) that would be efficient radiators (Rowan-Robinson 1992) could create emission in the millimetre wavelengths without too much mass. Exploring this possibility would require specific modelling. This is beyond the scope of the current article but should be explored in future studies.

6. Conclusions

We assessed the existence and investigated the origin of millimetre excess emission in the LMC and the SMC using the Planckdata. In the framework of this preliminary analysis, we focused our interest on the thermal dust emission from the galaxies, which we isolated by subtracting other unrelated components such as galactic foreground, CMB fluctuations and free-free emission. More detailed studies in the future will involve full multi-component analysis or the use of more sophisticated component separation algorithms exploiting all spatial and spectral information available in the data.

The integrated SED of the two galaxies before subtraction of the foreground (Milky Way) and background (CMB fluctuations) emission are in good agreement with previous determinations.

The background CMB contribution was first subtracted using an ILC method performed locally around the two galaxies. The uncertainty of this contribution was measured through a detailed Monte-Carlo simulation. We subtracted the foreground emission from the Milky Way using a Galactic Hi template and the proper dust emissivity derived in a region surrounding the two galaxies and dominated by MW emission. We also subtracted the free-free contribution from ionised gas in the galaxies, using the Hα emission, taking advantage of the low extinction in those galaxies. The remaining emission of both galaxies correlates with the gas emission of the LMC and SMC.

We showed that the excess previously reported in the LMC can be fully explained by CMB fluctuations. For the SMC, subtracting the CMB fluctuations decreases the intensity of the excess but a significant millimetre emission above the expected thermal dust, free-free and synchrotron emission remains.

We combined the Planck and IRAS–IRIS data at 100μm to produce thermal dust temperature and optical depth maps of the two galaxies. The LMC temperature map shows the presence of a warm inner arm already found with the Spitzer data, but also shows the existence of a previously unidentified cold outer arm. Several cold regions were found along this arm, some of which are associated to known molecular clouds.

We used the dust optical depth maps to constrain the thermal dust emissivity spectral index (β). The average spectral index in the FIR (λ < 500μm) is found to be consistent with β = 1.5 and β = 1.2 for the LMC and the SMC respectively. This is significantly flatter than is observed in the Milky Way. The absolute values of the emissivities in the FIR, when compared to that in our solar neighbourhood, are compatible with D/G mass ratios of 1/2.4 and 1/13 for the LMC and SMC. This is compatible with the metallicity of the LMC but significantly lower than the metallicity of the SMC. In the submm, the LMC SED remains consistent with β = 1.5, while the SED of the SMC is even flatter.

The spatial distribution of the mm excess in the SMC appears to follow the general pattern of the gas distribution. It therefore appears unlikely that the excess could originate from very cold dust. Indeed, this is confirmed by attempts to fit the SMC emission SED with models with two dust components, which led to poor fits. Alternative models, such as emission excess due to the amorphous nature of large grains, are likely to provide a natural explanation for the observed SEDs, although spinning dust is needed to explain the SED above λ = 3 mm.

1

Planck (http://www.esa.int/Planck) is a project of the European Space Agency (ESA) with instruments provided by two scientific consortia funded by ESA member states (in particular the lead countries: France and Italy) with contributions from NASA (USA), and telescope reflectors provided in a collaboration between ESA and a scientific consortium led and funded by Denmark.

2

Ysard & Verstraete (2010) showed that it is in good agreement with the anomalous emission extracted from the WMAP data.

Acknowledgments

A description of the Planck Collaboration and a list of its members can be found at http://www.rssd.esa.int/index.php?project=PLANCK&page=Planck_Collaboration

References

  1. Anderson, N., & Watson, W. D. 1993, A&A, 270, 477 [NASA ADS] [Google Scholar]
  2. Bendo, G. J., Dale, D. A., Draine, B. T., et al. 2006, ApJ, 652, 283 [NASA ADS] [CrossRef] [Google Scholar]
  3. Bennett, C. L., Halpern, M., Hinshaw, G., et al. 2003, ApJS, 148, 1 [NASA ADS] [CrossRef] [Google Scholar]
  4. Bernard, J., Reach, W. T., Paradis, D., et al. 2008, AJ, 136, 919 [NASA ADS] [CrossRef] [Google Scholar]
  5. Bolatto, A. D., Simon, J. D., Stanimirović, S., et al. 2007, ApJ, 655, 212 [NASA ADS] [CrossRef] [Google Scholar]
  6. Bot, C., Boulanger, F., Lagache, G., Cambrésy, L., & Egret, D. 2004, A&A, 423, 567 [NASA ADS] [CrossRef] [EDP Sciences] [MathSciNet] [Google Scholar]
  7. Bot, C., Rubio, M., Boulanger, F., et al. 2010a, A&A, 524, A52 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  8. Bot, C., Ysard, N., Paradis, D., et al. 2010b, A&A, 523, A20 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  9. Boulanger, F., Abergel, A., Bernard, J., et al. 1996, A&A, 312, 256 [NASA ADS] [Google Scholar]
  10. Dobashi, K., Bernard, J., Hughes, A., et al. 2008, A&A, 484, 205 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  11. Draine, B. T., & Lazarian, A. 1998a, ApJ, 494, L19 [NASA ADS] [CrossRef] [Google Scholar]
  12. Draine, B. T., & Lazarian, A. 1998b, ApJ, 508, 157 [NASA ADS] [CrossRef] [Google Scholar]
  13. Draine, B. T., & Li, A. 2007, ApJ, 657, 810 [NASA ADS] [CrossRef] [Google Scholar]
  14. Draine, B. T., Dale, D. A., Bendo, G., et al. 2007, ApJ, 663, 866 [NASA ADS] [CrossRef] [Google Scholar]
  15. Dumke, M., Krause, M., & Wielebinski, R. 2003, in Rev. Mex. Astron. Astrof., Conf. Ser., ed. V. Avila-Reese, C. Firmani, C. S. Frenk, & C. Allen, 27, 17, 185 [Google Scholar]
  16. Dupac, X., Bernard, J., Boudet, N., et al. 2003, A&A, 404, L11 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  17. Eriksen, H. K., Banday, A. J., Górski, K. M., & Lilje, P. B. 2004, ApJ, 612, 633 [NASA ADS] [CrossRef] [Google Scholar]
  18. Ferland, G. J., Korista, K. T., Verner, D. A., et al. 1998, PASP, 110, 761 [Google Scholar]
  19. Ferrara, A., & Dettmar, R. 1994, ApJ, 427, 155 [NASA ADS] [CrossRef] [Google Scholar]
  20. Finkbeiner, D. P., Davis, M., & Schlegel, D. J. 1999, ApJ, 524, 867 [NASA ADS] [CrossRef] [Google Scholar]
  21. Fukui, Y., Mizuno, N., Yamaguchi, R., et al. 1999, PASJ, 51, 745 [NASA ADS] [CrossRef] [Google Scholar]
  22. Fukui, Y., Kawamura, A., Minamidani, T., et al. 2008, ApJS, 178, 56 [NASA ADS] [CrossRef] [Google Scholar]
  23. Galametz, M., Madden, S., Galliano, F., et al. 2009, A&A, 508, 645 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  24. Galametz, M., Madden, S. C., Galliano, F., et al. 2010, A&A, 518, L55 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  25. Galliano, F., Madden, S. C., Jones, A. P., et al. 2003, A&A, 407, 159 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  26. Galliano, F., Madden, S. C., Jones, A. P., Wilson, C. D., & Bernard, J. 2005, A&A, 434, 867 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  27. Gaustad, J. E., McCullough, P. R., Rosing, W., & Van Buren, D. 2001, PASP, 113, 1326 [NASA ADS] [CrossRef] [Google Scholar]
  28. Gordon, K. D., Bot, C., Muller, E., et al. 2009, ApJ, 690, L76 [NASA ADS] [CrossRef] [Google Scholar]
  29. Gordon, K. D., Galliano, F., Hony, S., et al. 2010, A&A, 518, L89 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  30. Górski, K. M., Hivon, E., Banday, A. J., et al. 2005, ApJ, 622, 759 [NASA ADS] [CrossRef] [Google Scholar]
  31. Grossi, M., Hunt, L. K., Madden, S., et al. 2010, A&A, 518, L52 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  32. Henize, K. G. 1956, ApJS, 2, 315 [NASA ADS] [CrossRef] [Google Scholar]
  33. Hoang, T., Draine, B. T., & Lazarian, A. 2010, ApJ, 715, 1462 [NASA ADS] [CrossRef] [Google Scholar]
  34. Hunt, L. K., Dyer, K. K., Thuan, T. X., & Ulvestad, J. S. 2004, ApJ, 606, 853 [NASA ADS] [CrossRef] [Google Scholar]
  35. Israel, F. P., Wall, W. F., Raban, D., et al. 2010, A&A, 519, A67 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  36. Jarosik, N., Bennett, C. L., Dunkley, J., et al. 2011, ApJS, 192, 14 [NASA ADS] [CrossRef] [Google Scholar]
  37. Kim, S., Staveley-Smith, L., Dopita, M. A., et al. 2003, ApJS, 148, 473 [NASA ADS] [CrossRef] [Google Scholar]
  38. Komatsu, E., Smith, K. M., Dunkley, J., et al. 2011, ApJS, 192, 18 [NASA ADS] [CrossRef] [Google Scholar]
  39. Leroy, A., Bolatto, A., Stanimirovic, S., et al. 2007, ApJ, 658, 1027 [NASA ADS] [CrossRef] [Google Scholar]
  40. Leroy, A. K., Bolatto, A., Bot, C., et al. 2009, ApJ, 702, 352 [NASA ADS] [CrossRef] [Google Scholar]
  41. Lisenfeld, U., Sievers, A., Israel, F., & Stil, J. 2001, Astrophys. Space Sci. Suppl., 277, 105 [NASA ADS] [CrossRef] [Google Scholar]
  42. Madden, S. C. 2000, New Astron. Rev., 44, 249 [NASA ADS] [CrossRef] [Google Scholar]
  43. Madden, S. C., Poglitsch, A., Geis, N., Stacey, G. J., & Townes, C. H. 1997, ApJ, 483, 200 [NASA ADS] [CrossRef] [MathSciNet] [Google Scholar]
  44. Meixner, M., Gordon, K. D., Indebetouw, R., et al. 2006, AJ, 132, 2268 [NASA ADS] [CrossRef] [Google Scholar]
  45. Meixner, M., Galliano, F., Hony, S., et al. 2010, A&A, 518, L71 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  46. Mennella, A., Bersanelli, M., Butler, R. C., et al. 2011, A&A, 536, A3 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  47. Meny, C., Gromov, V., Boudet, N., et al. 2007, A&A, 468, 171 [Google Scholar]
  48. Miville-Deschênes, M., & Lagache, G. 2005, ApJS, 157, 302 [NASA ADS] [CrossRef] [Google Scholar]
  49. Mizuno, N., Rubio, M., Mizuno, A., et al. 2001, PASJ, 53, L45 [NASA ADS] [Google Scholar]
  50. Muller, E., Staveley-Smith, L., Zealey, W., & Stanimirović, S. 2003, MNRAS, 339, 105 [NASA ADS] [CrossRef] [Google Scholar]
  51. O’Halloran, B., Madden, S. C., & Abel, N. P. 2008, ApJ, 681, 1205 [NASA ADS] [CrossRef] [Google Scholar]
  52. O’Halloran, B., Galametz, M., Madden, S. C., et al. 2010, A&A, 518, L58 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  53. Paradis, D. 2007, Ph.D. Thesis [Google Scholar]
  54. Paradis, D., Bernard, J., & Mény, C. 2009, A&A, 506, 745 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  55. Paradis, D., Veneziani, M., Noriega-Crespo, A., et al. 2010, A&A, 520, L8 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  56. Planck Collaboration 2011a, A&A, 536, A1 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  57. Planck Collaboration 2011b, A&A, 536, A2 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  58. Planck Collaboration 2011c, A&A, 536, A7 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  59. Planck Collaboration 2011d, A&A, 536, A8 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  60. Planck Collaboration 2011e, A&A, 536, A9 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  61. Planck Collaboration 2011f, A&A, 536, A10 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  62. Planck Collaboration 2011g, A&A, 536, A11 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  63. Planck Collaboration 2011h, A&A, 536, A12 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  64. Planck Collaboration 2011i, A&A, 536, A13 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  65. Planck Collaboration 2011j, A&A, 536, A14 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  66. Planck Collaboration 2011k, A&A, 536, A15 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  67. Planck Collaboration 2011l, A&A, 536, A16 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  68. Planck Collaboration 2011m, A&A, 536, A17 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  69. Planck Collaboration 2011n, A&A, 536, A18 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  70. Planck Collaboration 2011o, A&A, 536, A19 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  71. Planck Collaboration 2011p, A&A, 536, A20 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  72. Planck Collaboration 2011q, A&A, 536, A21 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  73. Planck Collaboration 2011r, A&A, 536, A22 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  74. Planck Collaboration 2011s, A&A, 536, A23 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  75. Planck Collaboration 2011t, A&A, 536, A24 [Google Scholar]
  76. Planck Collaboration 2011u, A&A, 536, A25 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  77. Planck Collaboration 2011v, The Explanatory Supplement to the Planck Early Release Compact Source Catalogue (ESA) [Google Scholar]
  78. Planck Collaboration 2011w, A&A, 536, A26 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  79. Planck HFI Core Team 2011a, A&A, 536, A4 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  80. Planck HFI Core Team 2011b, A&A, 536, A6 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  81. Poglitsch, A., Krabbe, A., Madden, S. C., et al. 1995, ApJ, 454, 293 [NASA ADS] [CrossRef] [Google Scholar]
  82. Reach, W. T., Dwek, E., Fixsen, D. J., et al. 1995, ApJ, 451, 188 [NASA ADS] [CrossRef] [Google Scholar]
  83. Roman-Duval, J., Israel, F. P., Bolatto, A., et al. 2010, A&A, 518, L74 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  84. Rowan-Robinson, M. 1992, MNRAS, 258, 787 [NASA ADS] [Google Scholar]
  85. Sandstrom, K. M., Bolatto, A. D., Draine, B. T., Bot, C., & Stanimirović, S. 2010, ApJ, 715, 701 [NASA ADS] [CrossRef] [MathSciNet] [Google Scholar]
  86. Santini, P., Maiolino, R., Magnelli, B., et al. 2010, A&A, 518, L154 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  87. Serra Díaz-Cano, L., & Jones, A. P. 2008, A&A, 492, 127 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  88. Shetty, R., Kauffmann, J., Schnee, S., & Goodman, A. A. 2009a, ApJ, 696, 676 [NASA ADS] [CrossRef] [Google Scholar]
  89. Shetty, R., Kauffmann, J., Schnee, S., Goodman, A. A., & Ercolano, B. 2009b, ApJ, 696, 2234 [NASA ADS] [CrossRef] [Google Scholar]
  90. Silsbee, K., Ali-Haimoud, Y., & Hirata, C. M. 2011, MNRAS, 411, 2750 [NASA ADS] [CrossRef] [Google Scholar]
  91. Stanimirovic, S., Staveley-Smith, L., Dickey, J. M., Sault, R. J., & Snowden, S. L. 1999, MNRAS, 302, 417 [NASA ADS] [CrossRef] [Google Scholar]
  92. Staveley-Smith, L., Sault, R. J., Hatzidimitriou, D., Kesteven, M. J., & McConnell, D. 1997, MNRAS, 289, 225 [NASA ADS] [CrossRef] [Google Scholar]
  93. Staveley-Smith, L., Kim, S., Calabretta, M. R., Haynes, R. F., & Kesteven, M. J. 2003, MNRAS, 339, 87 [NASA ADS] [CrossRef] [Google Scholar]
  94. Ysard, N., & Verstraete, L. 2010, A&A, 509, A12 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  95. Ysard, N., Miville-Deschênes, M. A., & Verstraete, L. 2010, A&A, 509, L1 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  96. Zacchei, A., Maino, D., Baccigalupi, C., et al. 2011, A&A, 536, A5 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  97. Zhu, M., Papadopoulos, P. P., Xilouris, E. M., Kuno, N., & Lisenfeld, U. 2009, ApJ, 706, 941 [NASA ADS] [CrossRef] [Google Scholar]

All Tables

Table 1

Characteristics of the FIR/submm data used in this study.

In the text
Table 2

LMC (Cols. 2–4) and SMC (Cols. 5–7) SEDs averaged in a circular region for each galaxy.

In the text
Table 3

Dust temperature, β values and fit reduced χ2 for various methods experimented to derive the temperature maps.

In the text

All Figures

thumbnail Fig.1

Planck total intensity data for the LMC (left) and SMC (right) at 857 (top), 100 (middle) and 28.5GHz(bottom) at full resolution. The top panels are shown in log scale. The circle in the top panels shows the region used to extract average SEDs.
In the text
thumbnail Fig.2

Error due to the ILC CMB subtraction derived from Monte-Carlo simulations for both LMC and SMC at 100GHz, as a function of the patch size. The values shown are the difference between the recovered and the input CMB divided by the simulated LMC and SMC, both integrated in a 4° ring.
In the text
thumbnail Fig.3

SMC (left) and LMC (right) total intensity maps before (top) and after (bottom) CMB subtraction in the 70.3GHz band at the 13.01 arcmin resolution.
In the text
thumbnail Fig.4

Average foreground SED in the direction of the LMC and SMC (lower curve) including IRAS–IRIS (light blue), Planck-LFI (red), Planck-LFI (green) and WMAP data (dark blue), compared to the SED of the high latitude low column-density MW SED (upper curve) derived in Planck Collaboration (2011o), which has been scaled up by a factor of 10 for clarity.
In the text
thumbnail Fig.5

Integrated SEDs of the LMC (left) and SMC (right) before and after CMB subtraction. The black points and model are taken from Bot et al. (2010b). The red symbols show the SEDs derived from the DIRBE, IRAS, and WMAP data before CMB subtraction. The blue symbols show the same after CMB subtraction.
In the text
thumbnail Fig.6

Integrated SEDs of the LMC (solid) and SMC (dashed) after CMB and galactic foreground subtraction, including data from Planck-HFI (red), Planck-LFI (green), IRAS–IRIS (light blue) and WMAP (dark blue). The uncertainties shown are  ± 3σ. The SMC SED was scaled by a factor 4 providing normalization at 100μm.
In the text
thumbnail Fig.7

Upper panels: dust temperature maps for the LMC (left) and SMC (right) computed from the foreground subtracted maps using the IRAS–IRIS100μm, HFI857 and 545GHz maps, using a fixed βLMC = 1.5 and βSMC = 1.2. Lower panels: relative uncertainties on the dust temperature at the same resolution, expressed as percentages.
In the text
thumbnail Fig.8

Comparison between the dust temperature map of the LMC with Hα (top) and CO emission (bottom). The CO contours are at 0.5, 2, 4 and 10 Kkms-1. Hα contours are at 1, 10, 50, 100, 500 and 1000 Rayleigh. The thick line shows the edge of the available CO surveys.
In the text
thumbnail Fig.9

Comparison between the dust temperature map of the SMC with Hα (top) and CO emission (bottom). The CO contours are at 0.5, 1 and 1.5 Kkms-1. Hα contours are at 1, 5, 30 and 100 Rayleigh. The thick line shows the edge of the available CO surveys.
In the text
thumbnail Fig.10

Upper panel: map of the dust optical depths of the LMC at HFI 217GHz. Units are 104 × τ. Lower panel: map of the dust optical depth relative uncertainty of the LMC at HFI 217GHz in percent. Black pixels in the maps are masked and have relative uncertainties larger than 50%.
In the text
thumbnail Fig.11

Upper panel: map of the dust optical depths of the SMC at HFI 217GHz. Units are 104 × τ. Lower panel: map of the dust optical depth relative uncertainty at HFI 217GHz in percent. Black pixels in the maps are masked and have relative uncertainties larger than 50%.
In the text
thumbnail Fig.12

Map of the millimetre excess in the SMC at 3mm computed at the resolution of the LFI–28.5GHz channel, after free-free subtraction. The contours show the Hα distribution.
In the text
thumbnail Fig.13

Average dust optical depth of the LMC (black) and SMC (blue) obtained from the CMB, foreground and free-free subtracted SEDs. The average is taken in the same regions as Fig. 5. The SMC SED was normalised (multiplied by 6.31) to that of the LMC at 100μm. The square symbol (green) shows the reference value for the solar neighbourhood by Boulanger et al. (1996). The dashed lines show τ ∝ ν1.5 and τ ∝ ν1.2 normalised at 100μm. Error bars are  ± 3-σ.
In the text
thumbnail Fig.14

Fit of the SMC SED using the Finkbeiner model (upper left), the TLS model (upper right), the spinning dust model (lower left) and a combination of TLS and spinning dust models (lower right). In the lower panels, the squares represent the flux in each band predicted by the best model (the blue line).
In the text

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.