Open Access
Issue
A&A
Volume 674, June 2023
Article Number A141
Number of page(s) 29
Section Interstellar and circumstellar matter
DOI https://doi.org/10.1051/0004-6361/202245107
Published online 16 June 2023

© The Authors 2023

Licence Creative CommonsOpen Access article, published by EDP Sciences, under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

This article is published in open access under the Subscribe to Open model. Subscribe to A&A to support open access publication.

1 Introduction

It is currently well accepted that dust emission in the interstellar medium (ISM) can be divided into three domains: the near-infrared (NIR; from 0.7 to 5 μm) to mid-infrared (MIR; from 5 to 40 μm), which is dominated by polycyclic aromatic hydrocarbon (PAH) emission in the 3–20 μm range, the MIR to far-infrared (FIR; from 40 to 350 μm), which is dominated by emission from very small particles or grains (potentially carbon grains, abbreviated as VSGs) in the ~20–100 μm range and the FIR to submillimeter/millimeter (submm/mm) emission dominated by large grains (silicates or a mixture of carbon and silicate grains, abbreviated as BGs) above the range of ~ 100–200 μm. The understanding of the NIR-to-MIR regime has experienced a significant progress over the past 25 yr with Infrared Space Observatory (ISO) spectroscopic data (Trewhella et al. 2000; Boulanger et al. 2000) and then Spitzer Space Telescope data (Meixner et al. 2006; Bernard et al. 2008; Paradis et al. 2011a; Tibbs et al. 2011, e.g.). Based on the Planck1-Herschel2 missions, the FIR, submm and mm regimes have been extensively studied in the past decade for Galactic (see, e.g., Juvela et al. 2011; Paradis et al. 2012, 2014; Planck Collaboration Int. XIV 2014; Planck Collaboration Int. XVII. 2014; Planck Collaboration XI 2014; Meisner and Finkbeiner 2015; Juvela et al. 2018) and extragalactic studies (see for instance Planck Collaboration XVII 2011; Galliano et al. 2011; Dale et al. 2012; Galametz et al. 2012; Chastenet et al. 2017). Nevertheless, the origin of the emission arising in the FIR and submm is still only poorly constrained in terms of grain composition, size and shape. However, the MIR-to-FIR domain has always lacked data. Most of the data in this wavelength domain come from photometric data in a few bands. For instance, the Spitzer telescope provided data at 24 μm and 70 μm, very similar to the photometric data observed with Infared Astronomical Satellite (IRAS, 25 and 60 μm). Herschel/PACS spectroscopic data do not provide data below 55 μm. They do not give any information on the emission of dust between 20 and 55 μm spectral range, which is crucial for constraining the very small particles or grains, however. Spitzer spectroscopic data were mainly centered on the NIR-to-MIR emission. Only very few programs focused on the MIR-to-FIR domain. This was the case of the SAGE-spec Spitzer Legacy Program (Kemper et al. 2010), which was a spectroscopic follow-up to the SAGE-LMC photometric survey of the Large Magellanic Cloud (LMC, Meixner et al. 2006) carried out with the Spitzer Space Telescope. Extended regions in the diffuse medium and HII regions were observed with the use of the Infrared Spectrometer (IRS) staring mode from ~5 to 38 μm, and with the MIPS-SED mode from 52 to 97 μm. These data are crucial for setting constraints on the dust at the origin of the MIR-to-FIR emission. Some IRS observations were also available for the Dwarf Galaxy Survey (DGS) sample. Rémy-Ruyer et al. (2015) analyzed a sample of 98 low-metallicity galaxies (from the Dwarf Galaxy and KINGFISH surveys) and modeled their global spectral energy distributions (SEDs) from the NIR to the submm using the dust model described in Galliano et al. (2011). They merged different data sets, including IRS spectra when available. For 11 sources, they included an additional modified blackbody component at MIR-to-FIR wavelengths with temperatures in the range 80–300 K, which significantly improved the modeling of the entire galaxies. They attributed this component to hot HII regions that were added to the total emission of the galaxies, even though in this type of regions an increase in emission over the entire SED is expected and not specifically in the MIR-to-FIR range.

The advantage of studying the LMC is that the IRS observations spatially resolve different environments of the Galaxy. This was not possible in the DGS survey. The LMC is one of the closest galaxy, at a distance of ~50 kpc, with a complex structure, including HI filaments, arcs, holes and shells (Kim et al. 1998). The small scale of the neutral atomic ISM is dominated by a turbulent and fractal structure due to the dynamical feedback of the star formation processes. The large scale of the HI disk has a symmetrical and rotational field. Most of LMC SED studies were based on the results derived using a single dust model (see for instance Bernard et al. 2008; Paradis et al. 2009, 2011a; Galliano et al. 2011; Galametz et al. 2013; Stephens et al. 2014; Roman-Duval et al. 2017; Chastenet et al. 2019). Only Chastenet et al. (2017) and Paradis et al. (2019) fit the SED of the LMC using two or more dust models from the DustEM package3 (Compiègne et al. 2011). More recently, Chastenet et al. (2021) performed a comparative study of M101 to derive the dust mass and show the dependence of the results on the dust models. None of the previous LMC studies had constraints in the MIR-to-FIR range, more specifically, between 25 and 70 μm. In addition, except for Paradis et al. (2011a), who investigated the impact of another radiation field (RF) template, all analyses considered the standard Mathis RF, adapted to the Milky Way (MW) interstellar RF, and never attempted to modify its shape. The LMC studies showed an abundance of dust half that of the MW, which is explained by the low metallicity of the LMC in comparison with the MW. Its dust emission spectrum is significantly flatter in the submm than in the MW (Planck Collaboration XVII 2011). PAHs appear to be enhanced in molecular clouds and in the old stellar bar, but are potentially destroyed in regions with high RF (Paradis et al. 2011a, Paradis et al. 2019). The PAH production by fragmentation might also be linked to the metallicity of the Galaxy. On the other hand, very small grains might be formed in HII regions (Paradis et al. 2019). Large VSG potentially produced by the erosion of large grains might be responsible for the 70 μm excess reported for the Magellanic Clouds (Paradis et al. 2009).

Using the combination of photometric and spectroscopic data in the NIR-to-FIR domain, we fit the spectral shape of dust emission in different environments of the LMC (two diffuse, two molecular, and six HII regions) with four differents dust models (Jones et al. 2013; Compiègne et al. 2011; Draine and Li 2007; Désert et al. 1990). The modeling was performed with four interstellar radiation field templates, and we allowed different parameters, such as the dust abundances and the dust size distribution, to vary. We also analyzed the extinction curves produced from dust models.

After describing the data sets (Sect. 2), we present the method for extracting the SED in each region in Sect. 3. In Sects. 4 and 5 we briefly describe the targets and the dust models we used as part of the DustEM package. Then we describe the fitting results in Sect. 6. After the discussion in Sect. 7, we provide a summary of the results in Sect. 8.

2 Observations

2.1 Spitzer data

2.1.1 IRS staring and MIPS SED mode

Spectroscopic data were obtained as part of the SAGE-Spec Spitzer Legacy program (PID: 40159), a spectroscopic follow-up to the SAGE-LMC photometric survey of the LMC. Extended regions (atomic, molecular and HII regions) were observed in the IRS staring (between 5 μm and 38 μm) and MIPS SED modes (between 52 μm and 97 μm). We used the latest available data release that was produced by the SAGE-Spec team. The data were in the past reduced by the team using the standard pipeline data as produced by the Spitzer Science Center. The individual observations were combined into a spectral cube using CUBISM (Smith et al. 2007). The MIPS SED extended source observations were reduced using the MIPS DAT v3.10 (Gordon et al. 2005) and were calibrated according to the prescription of Lu et al. (2008).

2.1.2 MIPS photometry

The SAGE-LMC survey (Meixner et al. 2006) observed the entire LMC using the IRAC (Fazio et al. 2004) and MIPS (Rieke et al. 2004) instruments. We combined the spectroscopic data with MIPS photometry at 70 and 160 μm at 18″ and 40″ angular resolution.

2.2 Herschel data

To trace dust in the FIR and submm, we used the Herschel PACS (160 μm, at 13″ angular resolution) and the SPIRE (250, 350, and 500 μm, at 18″, 25″, and 36″ angular resolution, respectively) data as part of the Heritage program (Meixner et al. 2010). We used the latest available version of the data in the ESA Herschel Science Archive4.

2.3 Gas tracers

2.3.1 Atomic hydrogen

We used the 21 cm map of Kim et al. (2003; spatial resolution of 1′) to trace the atomic gas, integrated in the velocity range 190 km s−1< VLSR < 386 km s−1. This map was made by combining interferometric data from the Australia Telescope Compact Array (ATCA; 1’) and the Parkes antenna (15.3′; Staveley-Smith et al. 2003). To derive the HI column density, we applied the standard conversion factor XHI, equal to 1.82 × 1018 H/cm2/(K km s−1) (Spitzer 1978; Lee et al. 2015), such that (1)

where WHI is the integrated intensity map.

2.3.2 Carbon monoxide

To trace the molecular gas, we used data from the 22 m Mopra telescope of the Australia Telescope National Facility at an angular resolution of 45″. This survey of the LMC was made as part of the MAGMA project (Wong et al. 2011). The integrated-intensity map (WCO) was converted into molecular column densities using the relation (2)

where XCO is the CO-to-H2 conversion factor. We used an XCO value equal to 4 × 1020 H/cm2/(K km s−1), as in Paradis et al. (2019). This value is a good compromise that takes the high dispersion of the XCO values into account that were derived by different authors (see for instance Hughes et al. 2010; Leroy et al. 2011; Roman-Duval et al. 2014).

2.3.3 Ionized hydrogen

The ionized gas is usually traced by the Hα recombination line. We therefore used the Southern H-Alpha Sky Survey Atlas (SHASSA; Gaustad et al. 2001) at an angular resolution of 48″. The H+ column density can be derived, assuming a constant electron density ne along each line of sight, by applying the relation (Lagache et al. 1999) (3)

Following Dickinson et al. (2003), 1 Rayleigh=2.25 pc cm−6 for Te = 8000 K. The electron density was derived in Paradis et al. (2011a) for different regimes over the entire LMC, and more recently, in Paradis et al. (2019) for the molecular clouds of the LMC. The electron density can be as high as 3.98 cm−3 for bright HII regions, as low as 0.055 cm−3 for the diffuse ionized gas of the LMC, and close to 1 cm−3 for the molecular clouds. A value of 1.52 cm−3 was determined for typical HII regions. We adopted the appropriate electron density depending on the Hα brigthness, following Paradis et al. (2011a) (see Sect. 4).

2.4 Convolution

All Spitzer and Herschel data were smoothed to a common angular resolution (40″) and were reprojected on the same grid. The smoothing was performed using a Gaussian kernel with where is the original resolution of the maps. For the IRS and MIPS SED data, the smoothing was performed on each plane of the cubes. For the gas tracers, for which the angular resolution is slightly higher than 40″ (from 45″ to the integrated-intensity maps are only reprojected on the same grid as the infrared data. The pixelization is therefore slightly oversampled, but the impact on the gas estimates is very limited. Because the gas tracers are only used to determine the indicative gas column densities, the shape of the SED is not affected at all.

3 Construction of the spectral energy distribution

For the 24 extended regions observed as part of the SAGE-Spec program, we extracted the SED by computing the median brightness at each wavelength in a circular region enlarged by 2 pixels around the central position of the source. We selected ten regions (SSDR1, SSDR5, SSDR7, SSDR8, SSDR9, SSDR10, DEML10, DEML34, DEML86, and DEML323) for which the IRS SS and LL spectra are well overlaid at the same wavelengths and with reasonable dispersion in the data. A short description of the regions is given in Sect. 4.

We then computed the total column density for each region using Eqs. (1), (2), and (3), and we normalized each SED to an HI column density NH equal to 1 × 1020 H/cm2. We considered absolute calibration uncertainties of 20% for the IRS spectra (Protostars and Planets v.5), 15% for the MIPS SED data (MIPS Instrument Handbook5), 10% for the MIPS photometric data (MIPS Instrument Handbook6), and 7% for the Herschel data (Balog et al. 2014, for PACS 160 μm, and the observer manual v2.4 for SPIRE). However, to model the SED (see Sect. 6.1), it is crucial to increase the weight of the FIR to submm data to ensure an equal balance between the large number of spectro-scopic data and the low number of photometric data in each SED. The SEDs account for ~400 spectrocopic data and 5 photometric data (MIPS 70 data were only used to rescale the photometric data, see below). We first tested the fitting procedure without changing the weight of the data. The results were not acceptable because the χ2 of the fits were very good, but the SPIRE data were not well reproduced. We therefore applied different weights in the photometric data to check the reliability of the fitting results. We obtained satisfactory results by increasing the weight of the SPIRE data by a factor of 50 (i.e., a factor of 150 for the three SPIRE data, which corresponds to a similar weight between the spectral data and the photometric fluxes). In this way, we ensured a good representation of the SED over the entire wavelength range.

The SEDs were normalized by computing the ratio of the integrated flux in the MIPS-SED band and the MIPS 70 μm photometric data. The photometric data (from 70 to 500 μm) were rescaled by multiplying them by this factor. In a few cases (mainly DEML34 and DEML86), the difference between the MIPS and PACS 160 μm data is significant. This discrepancy is not the result of a nonlinearity effect between MIPS and PACS because this effect appears for a brightness above 50 MJy sr−1 at 160 μm, which is significantly higher than our values. Recent analyses of the original Heritage maps (Clark et al. 2021) showed missing dust in the periphery of the Magellanic Clouds (mainly at shorter wavelengths), but Herschel PACS maps appear to overestimate the brightness of the large-scale emission by 20–30% compared to absolutely calibrated all-sky survey data. However, when the disagreement between MIPS and PACS 160 μm is visible, our fits tend to reproduce the MIPS 160 μm data. The results are therefore not affected.

Table 1

Characteristics of the studied regions.

4 Target description

The location of each region is given in Kemper et al. (2010). In Table 1, we present the type of environment, that is, diffuse, molecular, or ionized, and the hydrogen column density in each phase of the gas. The adopted value for the electron density is also given in the table, in agreement with considerations presented in Paradis et al. (2011a). While the SSDR designation for the SAGE-Spec diffuse regions should correspond to diffuse regions (atomic and molecular), some of them (SSDR8 and SSDR10) apparently represent ionized regions. Our sample has only two diffuse regions (SSDR7 and SSDR9), two molecular regions (SSDR1 and SSDR5) and six HII regions (SSDR8, SSDR10, and the four DEML regions).

5 Dust models

We used the DustEM wrapper package to model the SEDs of the different regions using the following dust models: THEMIS (Jones et al. 2013, hereafter AJ13), Compiègne et al. (2011, hereafter MC11), Draine and Li (2007, hereafter DL07), and an improved version of the Désert et al. (1990) model (hereafter DBP90). Below is a very brief description of the models (we refer to the original papers corresponding to each model). The THEMIS (AJ13) model considers two dust components covered by an aromatic mantle: a population of carbonaceous grains consisting of large grains (a<200 nm) of amorphous and aliphatic nature (a-C:H), smaller grains (a<20 nm) of a more aromatic nature (a-C), and a population of large amorphous silicate grains (a<200 nm) containing nanometer-scale inclusions of FeS (large a-Sil). MC11 includes PAHs (neutral and ionized), small (a< 10 nm) and large (a> 10 nm) amorphous carbon grains (SamC and LamC), and large amorphous silicates (aSil). The silicate-graphite-PAH model (DL07) assumes a mixture of carbonaceous and amorphous silicates grains, including different amounts of PAH material (neutral and ionized). In the adapted version of the Désert et al. (1990) model we used, we substituted the original PAH component, in which the PAHs bands are incompletely described, with the neutral PAH component taken from MC11. The small (VSG) and large grain (BG) components are made of carbon and silicates grains.

6 Fitting results

We modeled the SEDs of the different regions by first allowing the standard parameters to vary and then changing the parameters of the carbon dust size distribution. In a second step, we injected different radiation fields in the modeling. The dust-to-gas mass ratio is discussed for the different models.

6.1 Standard free parameters

We first fit the observations with the different dust models, allowing standard parameters to vary: the abundances of the different dust components (Ycomponent), the intensity of the solar neigborhood radiation field (Xisrf, Mathis et al. 1983), and the intensity of the NIR continuum (INIRCont). These standard parameters (except for the NIR continuum) were constrained based on Galactic observations at high latitudes. Figure 1 presents the results we obtained with the different models for each SED and for the ten selected regions. Table 2 gives the values of the reduced χ2 for each SED, as well as the values of χ2 divided by the number of data in three wavelength ranges ([5–20]; ]20– 97] and [160–500]). For none of the dust models the standard parameters result in satisfactory modeling. For instance, AJ13 shows an important underestimate of the model in the MIR domain (and more specifically, between 15 and 30 μm) and an overestimate in the MIPS SED range (especially in HII regions). These discrepancies result in high values of the reduced χ2 with AJ13 compared to other models. Even if the other models are significantly better than AJ13, they all describe the data in the MIR domain imperfectly. For instance, MC11 underestimates the model in the 20–50 μm range of diffuse regions slightly, even if this is less pronounced than with AJ13. For the HII regions, the SEDs are reasonably well represented (as opposed to AJ13). DL07 shows the same behavior as AJ13 for the HII regions in particular, but it is significantly less pronounced than with AJ13. In DBP90, either the 20–50 μm range is underestimated or the 60–100 μm range is overestimated.

Regardless of the model we considered, the MIR-to-FIR domain is clearly poorly described, showing the low quality of modeling using the standard parameters. Since the standard parameters are not suitable to reproduce LMC observations, we try to improve the various modeling in the following section by allowing the parameters of the size distribution of the dust (which causes the MIR-to-FIR emission) to vary.

6.2 Changing the parameters of the dust size distribution of carbon grains

Depending on the model and the dust component, this size distribution is governed by a power law or log-normal distribution. Below, we summarize the original dust size distribution of interest for each model:

  • a-C (AJ13): A power-law distribution dn/daaα (α = −5) with an exponential tail, with ac = 50 nm and for aat = 10 nm (D(a) =1 otherwise) for grain sizes between 0.4nm (amin) and 4900 nm (amax).

  • SamC (MC11): A logarithmic normal distribution (with a0 the center radius equal to 2 nm and σ the width of the distribution equal to 0.35) with a grain size between 0.6 nm and 20 nm.

  • Graphite (DL07): A logarithmic normal distribution with a0 = 2 nm and σ = 0.55 with a between 0.31 nm and 40 nm.

  • VSG (DBP90): A power-law distribution with α = −2.6 for grains in the range 1.2 nm and 15 nm.

We first allowed either α in the power law or a0 in the log-normal distribution to vary in order to fit the SED better. In a second step, we also included amin and amax as free parameters for AJ13, MC11, and DL07. In the case of DBP90, changing amin and amax is highly degenerate with changing α. Therefore, when amin and amax are allowed to vary in the range of the original value, the α parameter is fixed to the original value (−2.6). For the other models, the dust size parameters are not anticorrelated, but are not completely independent either. The degeneracy between parameters is common when data are fit with models. When the degeneracy (or anticorrelation) is weak, as is the case with the dust size parameters for AJ13, MC11, and DL07, the parameters can be left as free parameters. Changing one parameter has a very limited impact on the others.

The values of the χ2 of the various fits are given in Table 2. All χ2 are improved for all models when α or a0 are included in the fits. The values of χ2 that are obtained when α/a0 and amin and amax are allowed to vary are only given when the fit is improved. The inclusion of amin and amax in the fits does not improve the modeling, especially in the 160–500 μm range for MC11 or DL07 models. We therefore only adopted free a0 for the VSG size parameters for MC11 and DL07. For DBP90, the fits are almost all better with free sizes (amin and amax) compared to the fits with free α. We therefore consider α = −2.6 and free amin and amax in the following. For AJ13, the fits are significantly improved in the 20–97 μm domain when free α, amin and amax are used (except for SSDR7). We therefore adopt these three free parameters for AJ13.

Figure 2 presents the best fits for each model (free α, amin, and amax for AJ13, free a0 forMC11 andDL07, and free amin and amax for DBP90). The values of the best-fit parameters are given in Tables 3, 4, 5, and 6. A null dust abundance is not possible for computational reasons. In this case, if a null abundance is required in the fit, its value was set to 1.00 × 10−6.

When the VSG size parameters are allowed to vary, the results of the fits in terms of χ2 are good for the diffuse and molecular regions regardless of the model (MC11, DL07, or DBP90). However, the HII regions are only poorly described with the DL07 model compared to models MC11 and DBP90. The model shows a lack of emission in the MIPS SED range. Modeling with AJ13 gives lower-quality fits, especially for diffuse and molecular environments. For instance, AJ13 shows a lack of emission in the IRS spectra between approximately 20 and 40 μm for SSDR1, SSDR5, and SSDR9. For the HII regions, the data are still not well reproduced with AJ13 in comparison with the other models, but the fits are reasonable. For MC11, the a0 parameter always increases (from 2.03 to 4.60) for all regions when compared to the standard value of 2.0, whereas for DL07, this parameter always decreases for the HII regions and increases for the two molecular regions. For the two atomic regions, the trend is unclear. For DBP90, amin is always larger than the standard parameters for all regions, and amax is also larger for six of the ten regions.

In some regions and for some models, we find that the best models do not contain silicates (mainly with MC11 and occasionally with AJ13; see Tables 3, 4, A.1, and A.2). This is possible because we did not impose a lower limit value for the dust abundances (except for reasons of computational limitations), but from a physical point of view, this result is surprising. This absence of silicates is not induced by the choice of the free parameters we adopted and in particular by the VSG size distribution parameters. The lack of a silicate component in some models was already visible with the use of standard parameters alone. Moreover, we performed additional tests with the MC11 model by leaving the power-law parameters free in the silicate grain-size distribution. The absence of silicates is unchanged in some cases, illustrating that the choice of the free parameters does not seem to affect the fitting results. We note that the silicates are absent from the best fits only for models including two populations of large grains that are composed of carbon and silicate. The potential absence of silicates in some models might therefore reflect the predominance of carbon grain emission over silicate component emission. However, in the modeling, the presence or absence of the carbon and/or silicate coarsegrained components is constrained only by the slope of the FIR to submm emission, and this absence might also result from the lack of observational constraints at longer wavelengths, that is, from 500 μm to 1 mm, for instance, combined with the fact that the emission of large grains in the FIR to submm is not very sensitive to the grain composition because the two emissions are somewhat degenerated. For this reason, (sub)millimeter data with small uncertainties, which are not easily obtained with ground-based telescopes, are crucial to constrain the BG component.

thumbnail Fig. 1

Modeling of the SEDs of the ten regions with different dust models and free standard parameters (XISRF and dust abundances), using the Mathis RF. The observations are shown in black (Spitzer IRS SS and LL, MIPS SED, MIPS 160 μm, Herschel Photometric PACS 160 μm, and SPIRE 250 μm, 350 μm, and 500 μm data). The total modeled SED is shown as a red line. The other colored lines correspond to the different dust components of the models. The dashed line represents the additional NIR continuum. Blue asterisks show the color-corrected brightness derived from the models. The orange diamonds in the DBP90 panels show the MIPS 70 μm photometric data normalized to the integrated flux in the MIPS-SED band. Each column shows the fit using different dust models (from left to right: AJ13, MC11, DL07, and DBP90). Each row presents a different region. The figure continues on the next page.

Table 2

Best-fit reduced χ2 obtained for the modeling of the full SED of the ten regions and for the four adopted dust models (between 5 μm and 500 μm), using the Mathis RF.

thumbnail Fig. 2

Modeling of the SEDs of the ten regions with different dust models and free parameters (XISRF, dust abundances, and small-grain dust size distribution) using the Mathis RF. The observations are shown in black (Spitzer IRS SS and LL, MIPS SED, MIPS 160 μm, Herschel Photometric PACS 160 mic, and SPIRE 250 μm, 350 mic, and 500 μm data) are shown in black. The total modeled SED is shown as a red line. The other colored lines correspond to the different dust components of the models. The dashed line represents the additional NIR continuum. Blue asterisks show the color-corrected brightness derived from the models. The orange diamonds in the DBP90 panels show the MIPS 70 μm photometric data normalized to the integrated flux in the MIPS-SED band. Each column shows the fit using different dust models (from left to right: AJ13, MC11, DL07, and DBP90). Each row presents a different region. The figure continues on the next page.

6.3 Changing the radiation field

Dust in galaxies can be illuminated by the radiation field from different stellar populations. The solar neighborhood RF is the standard RF used in most modeling of dust emission. In order to see a possible improvement of our best modeling and to test the robustness of our results, we also performed the modeling using other radiation fields. Bica et al. (1996) made a catalog of 504 star clusters and 120 stellar associations in the LMC using UBV photometry. They studied groups of stellar clusters with ages from 10 Myr to 16 Gyr. Kawamura et al. (2009) estimated that the youngest stellar objects are younger than 10 Myr. Using the code GALEV7, we generated Ultraviolet (UV)/visible spectrum of stellar clusters with various ages: 4-Myr, 60-Myr, and 600-Myr. We then made the format of the output files compatible with the DustEM package (see Fig. 3). Results of the modeling using the 4 Myr RF are given in Tables 3, 4, 5, and 6. Results of the fits using the 60 Myr and 600 Myr RFs and the corresponding figures are given in the appendix. Changing the RF in the modeling directly affects the values of dust abundances, and the intensity of the RF in particular (see the following sections), but affects the reduced χ2 very little. Therefore, changing the RF neither improves nor deteriorates the modeling.

In addition, the impact of the RFs on the VSG dust-size parameters is presented in Fig. 4. The values of a0 or amin, amax can vary slightly, but stay globally stable for MC11, DL07, and DBP90. For AJ13, the impact on amin is negligible, whereas amax is significantly affected for SSDR1, SSDR7, and SSDR9. This means that the amax values are poorly constrained. The value of α does not vary much with the RF, except for SSDR9. However, the modeling of SSDR9 is poor regardless of the RF. Therefore, the analysis of this region using AJ13 should be considered with caution.

6.4 Dust over gas mass ratio

For all models, we compute the dust over gas mass ratio, which corresponds to the total dust abundance (Ydust,tot, see Tables 3 to 6 and Tables A.1 to A.4) to check the reliability of the models. Below are the reference values for our Galaxy: 0.74 × 10−2 (AJ13), 1.02 × 10−2 (MC11), 1.10 × 10−2 (DL07) and 0.73 × 10−2 (DBP90). In our study we do not observe any inconsistent values. The dust over gas mass can be increased by a maximal factor of 3–4 in some HII regions or slightly decreased in some other regions. Statistically, this study shows that the total dust accounts for ~0.2% to ~4% of the total mass of the ISM in the LMC, depending on the region and the model. Roman-Duval (2022) observed variations by a factor of 4 of the dust over gas mass, from low to high column densities, derived from metal depletions in the LMC (see their Table 5). However, their ratio does not exceed 0.34%, with an integrated value over all column densities of 0.23%. The RF again affects the derived total dust abundance only little.

Table 3

Best-fit parameters for the ten regions obtained with AJ13 using the Mathis RF (top table) and 4 Myr RF (bottom table).

thumbnail Fig. 3

Radiation field templates used for the SED modeling: Mathis (solid red line), 4 Myr (short dashed black line), 60 Myr (dotted green line), and 600 Myr stellar clusters (long dashed blue line).

7 Discussion

In this section, we discuss the VSG population in the LMC in terms of its relative abundance, its contribution to the submm emission, and its possible link to the 70 μm excess observed in previous studies. We also show that the behavior of the VSG size distribution is different depending on the environment, which we explain by several possible scenarios of the dust evolution. Different extinction curves calculated for each model and region are presented.

Table 4

Best-fit parameters for the ten regions obtained with MC11 using the Mathis RF (top table) and 4 Myr RF (bottom table).

7.1 Very small grain population in the Large Magellanic Cloud

7.1.1 Increase in the relative abundance

The results of the modeling indicate a clear increase in the VSG abundance relative to the BG component compared to our Galaxy (see the second to last column of Tables 4 to 6). For comparison, we list the mass ratios of the VSGs over the large-grain component in our diffuse Galaxy derived from the original models: 2.02 × 10−2 (MC11), 1.69 × 10−2 (DL07), 7.34 × 10−2 (DBP90), and 1.30 × 10−1 and 1.36 × 10−1 in the MW plane and MW diffuse medium (DBP90) from Bernard et al. (2008).

In our sample of regions, this ratio is significantly increased regardless of the model (with a factor of 1.9–120 with MC11, a factor of 10–416 with DL07, and a factor of 1.2 to 12 with DBP90; see Fig. 5). When the RF changes, the ratio is directly affected and mainly decreases. However, the relative VSG abundance is always higher than the Galactic diffuse value (except for SSDR7 with DBP90 with the use of the 60 and 600 Myr RFs). The SSDR5 (molecular) and SSDR7 (diffuse) regions show the lowest values when compared to the other regions, with all the models and RFs (except with DBP90 when using the 4 Myr RF). The HII regions seem to have an enhanced VSG relative abundance in comparison to diffuse/molecular environments, even if the trend is not significant.

For AJ13, the MIR domain is dominated by the a-C component, which also describes the PAH component. In some cases, this component can also dominate the FIR-to-submm range. Even if it is not possible to directly compare with the other models, we computed the ratio of the abundance of the a-C component and the abundance of the other dust components (olivine, pyroxene, and a-C:H; see Table 3). The ratios are not systematically higher than the Galactic ratio (2.97 × 10−2). However, the trend is similar to that of the other models: higher ratios in the HII regions than in the diffuse and molecular regions.

Lisenfeld et al. (2002) studied the dwarf galaxy NGC1569 using SCUBA and IRAS data and used DBP90 as the model. They observed an increase in the VSG abundance relative to the large grains by a factor of 2 to 7 (depending on the radiation field and the dust size distribution in the modeling) compared to the solar neighborhood. The authors interpreted this as the result of large grain shattering due to shocks (turbulence and supernovae remnants) in the ISM.

Table 5

Best-fit parameters for the ten regions obtained with DL07 using the Mathis RF (top table) and 4 My RF (bottom table).

7.1.2 Contribution to the submillimeter emission

The original dust models that describe the dust emission at high Galactic latitudes contribute little to the VSG component emission at long wavelengths (see Table 7). For instance, depending on the model, the contribution of the VSG emission to the total emission is in the range 1.4—4.3% at 250 μm and between 1.2% and 13.7% at 1.1mm in our Galaxy. DBP90 gives the highest VSG contribution in the submm-mm, with values as high as 13.7% at 1.1mm. Compiègne et al. (2011) studied the diffuse dust emission of the inner Galactic plane using Herschel and Spitzer data. The authors identified large variations in the VSG abundances. The contribution of the VSGs in the FIR does not exceed 7% at 160 μm. They computed the averaged SED into two extreme regions in terms of dust property variations. In their Fig. 2, the maximum contribution of the VSGs at 500 μm is about 3–4%.

Our study brings new constraints on the VSG component. We have identified an increase in the relative VSG abundance (see Sect. 7.1.1) and changes in the dust size distribution (see Sect. 6.2) that result in a significant increase in the VSG contribution in the FIR-submm compared to the values observed in our Galaxy and using the Mathis RF (see Table 7). This result is valid for all models. For instance, depending on the region, at 500 μm, the contribution of the VSGs can be increased by a factor of ~5 to ~28 with MC11, a factor of ~4 to ~46 with DL07, and a factor of ~3 to ~10 with DBP90. In the case of AJ13, except for SSDR1 and SSDR9, which show a decrease in the contribution of the VSGs (represented by ), the increase reaches from a factor of ~1.4 to As a consequence, FIR-to-mm data should be modeled with caution with large grains alone, especially when the dust mass or column density are derived. However, the contribution of the VSG to the SED fit depending on the type of region clearly cannot be predicted. Diffuse and HII regions can have a wide range of values of the VSG contribution in each type of regions.

The values of the VSG contribution are significantly different (mainly lower) when the RFs change (see Tables 7 and A.5). This behavior is expected because the adopted RFs are bluer than the RF of Mathis. As opposed to BGs, which absorb energy over the whole RF spectrum, PAHs and VSGs absorb most of their energy at short wavelengths. As a consequence, PAHs and VSGs absorb and reemit more energy with a bluer RF. To produce the same IR brightness with a bluer RF, the model approximately needs lower PAH and VSG abundances. However, a few cases do not follow this expected behavior because other parameters, such as the VSG size distribution, are at play and are linked to the abundances.

The VSG contributions are nevertheless higher than the Galactic values most of the time. DL07 contributes less, except for SSDR1 and SSDR9, than the other models. These contributions vary from region to region and from model to model, without a clear trend with environment. This prevents any predictions.

Table 6

Best-fit parameters for the ten regions obtained with DBP90 using the Mathis RF (top and middle tables) and 4 My RF (bottom table).

thumbnail Fig. 4

Parameters of the small grain size distribution for each model, each region, and for the different RFs: Mathis (black), 4 Myr (red), 60 Myr (blue) and 600 Myr (green). Panel A: a0 (in nm) for MC11 and DL07; panel B: amin and amax (in nm) for DBP90; panel C: amin and amax (in nm) for AJ13 and panel D: α for AJ13. Arrows indicate high values of the amax parameter.

7.1.3 70 μm excess and submillimeter flattening

Bernard et al. (2008) and Paradis et al. (2011a) reported a 60–70 μm excess mainly in the neutral medium of the LMC and in the diffuse ionized gas with the use of standard parameters in the DBP90 model. Nevertheless, this does not exclude the fact that this excess is sometimes observed in the highly ionized gas or molecular gas of some regions. According to the authors, the OI (63 μm) or OIII (58 and 88 μm) emission lines are not responsible for this excess, even though Oliveira et al. (2019) observed an enhancement of these lines with respect to the dust continuum in photodissociated regions with young stellar objects (YSO) when compared to Galactic YSOs. By changing the slope of the VSG dust size distribution, Bernard et al. (2008) were able to reproduce the data reasonably well with a power-law distribution with an arbitrary value of α=−1 (instead of 2.6). Figure 1 shows the MIPS 70 μm photometric data normalized to the integrated flux in the MIPS-SED band (see Sect. 3) as orange diamonds. The integrated brightness in the MIPS 70 μm band derived from DBP90 is shown with the blue asterisks at the same wavelength. We observe a significant excess at 70 μm in several regions such as SSDR1, SSDR8, and SSDR9, and a softened excess in SSDR5 and SSDR10. This excess tends to disappear in all DEML regions. DBP90 with standard parameters is clearly not able to reproduce the 70 μm data in some cases. In addition, the influence of the oxygen lines in the 70 μm might be negligible for the total flux because the integrated flux in the MIPS-SED band does not deviate from the spectroscopic data even if some gas lines are observed.

When the VSG dust-size distribution is changed (α=−2.6, amin and amax), the excess no longer appears (see Fig. 2). This study reinforces the idea that in the framework of DBP90, the VSG size distribution in some regions of the LMC is different than that in our Galaxy. The analysis of the dust size distribution for all models is further discussed in the following Sect. 7.2.

It is known that the dust emission spectrum of the LMC becomes flatter in the submm than that of our Galaxy. This is even more pronounced in the SMC. At the same time, the SMC also shows a much higher excess of emission at 70 μm than the LMC (Bernard et al. 2008; Bot et al. 2010). We can therefore examine wether there is a possible link between the 70 μm excess and the submm flattening. For instance, in the LMC, this 70 μm excess and the submm flattening tend to disappear in most of the molecular gas phase (Paradis et al. 2019). This is not necessarily the case in molecular regions that may mix HI and CO gas phases. However, changing the VSG size distribution does not seem to affect the submm flattening as the diffuse and HII regions have a distinct size distribution while exhibiting submm flattening (Paradis et al. 2019), (see Sect. 7.2). In the same way, the increase in the VSG relative abundance in the ionized gas of the LMC highlighted by Paradis et al. (2019) might explain the submm flattening observed in the ionized regions, but not in the atomic regions. In conclusion, a change in the size distribution or the relative abundance of VSGs or a combination of both might explain the difference in the submm emission observed in the LMC compared to our Galaxy. Therefore, it might be expected that the adequate change in the VSG size distribution and abundance in the SMC might help reproduce the 70 μm excess identified in Bot et al. (2004) and might result in a large contribution of the VSG emission in the submm-mm.

If dust mainly originates from carbon stars (Boyer et al. 2012), the large amount of carbon dust relative to the silicate dust might explain the emissivity behavior observed in the LMC. In other words, small carbon grains (or a combination of small and large carbon grains) might cause the general behavior of the submm-mm flattening in the emission spectrum. For the SMC, the flattening seems to be too pronounced to be explained by carbon grains alone. In a previous unpublished study, we found a submm emissivity spectral index of 0.9 in the SMC using IRIS (new processing of IRAS data by Miville-Deschênes and Lagache 2005) and Planck data. In addition, variations in the submm flattening have been observed in the diffuse medium of our Galaxy, whereas the number of VSG does not affect the Galactic submm emission because its contribution at long wavelengths is low. The negligible submm emission from VSGs in our Galaxy, which shows a submm excess, indicates that the VSG component alone cannot cause the submm excess observed in our Galaxy. Therefore, other processes might be at play, such as TLS (two-level-system) processes proposed by Mény et al. (2007), describing the amorphous state of large dust grains to explain the submm behavior observed in our Galaxy. The TLS model is able to reproduce the different dust emission behavior observed in our Galaxy (Paradis et al. 2011b, 2012; Planck Collaboration Int. XIV 2014) and in the Magellanic Clouds (Planck Collaboration Int. XVII. 2014). More recently, the TLS model was also fully able to reproduce observations in molecular complexes of our Galaxy, such as the Perseus molecular cloud and W43 (Nashimoto et al. 2020)

To summarize, because in our Galaxy, the VSG component emission is negligible in the submm range, the VSG contribution alone cannot be the origin of the submm excess. It is therefore most likely that the very pronounced submm flattening in the Magellanic Clouds originates from a combination of at least two emission processes: the emission from the VSG component, plus the TLS processes in large grains; whereas only the TLS processes might cause the local variations observed in the diffuse regions of the MW.

thumbnail Fig. 5

Ratio of the small- to large-grain component for the different models and for the ten regions using different RFs (Mathis in black, 4 Myr in red, 60 Myr in blue, and 600-Myr in green). The Galactic value is indicated with the gray line.

Table 7

Ratio of the small-grain component emission to the total dust emission for each best-fit model using the Mathis (top table) and 4 Myr RF (bottom table) at 250, 500, 850, and 1100 μm.

7.2 Very small grain size distribution: Diffuse versus HII regions

First, we compare the results obtained with the different models using the Mathis RF. For the same reason as in Sect. 7.1.1, AJ13 is not discussed here because the a-C component includes both PAHs and small grains. The fits improve significantly when some parameters of the VSG dust size distribution are changed. The values of a0 presented in Tables 4 and 5 show some variations from one model to the next. The values of a0 extend from 2 nm to 4.6 nm for MC11 and from 1 nm to 3.5 nm for DL07. The values for MC11 are close to the Galactic value of 2 nm in HII regions, whereas the values systematically increase in the diffuse regions. For DL07, all the values decrease compared to the Galactic values. The same trend holds for MC11: The values in HII regions are significantly lower than those in diffuse regions. The mean value of a0 in each type of environment is equal to 2.47 nm (1.34 nm) in all HII regions with MC11 (DL07), whereas the value is 4.17 nm (3.11 nm) in diffuse regions with MC11 (DL07). Molecular regions show intermediate values between these two extreme environments (diffuse and HII regions). We caution, however, that the mean value for each type of environment was obtained with only two values for the diffuse and the molecular medium, and with six values for the ionized medium. This shift in the central value of the log-normal VSG size distribution in the different type of environments shows that HII regions contain mostly smaller VSGs (and fewer large grains in this component) and also an increase in the number of small VSGs in comparison with the diffuse regions. For DBP90, the fits show lower values for both amin and amax (see Table 6) in HII regions (mean values amin = 2.93 nm and amax = 13.1 nm) when compared to diffuse regions (mean values amin = 5.65 nm and amax = 20.3 nm), again showing smaller VSGs in HII regions than in the diffuse regions. The modeling with different RFs (Tables A.1 to A.4) does not change these conclusions and confirms the trend observed with the use of the Mathis RF.

To summarize, our results indicate the same trend for all models, regardless of the RF: The size distribution of VSGs is different in HII and diffuse regions with an increase in the quantity of small VSGs (and fewer large VSGs) in HII regions when compared to diffuse LMC regions. In this analysis, the SEDs represent the mixture of all gas phases (except for SSDR7 and SSDR9, which are almost purely atomic regions). For instance, SSDR1 and SSDR5 have a high level of HI emission, and most of the ionized regions of our sample have high column densities in the ionized gas and in other gas phases. We therefore expect a larger dispersion in the VSG size parameters, that is, even more pronounced results, when the dust emission associated with each gas phase is analyzed independently. For instance, in the SSDR8, DEML10, and DEML323 regions, the amount of hydrogen column density in the HI gas phase is higher than in the Hα gas phase. We therefore expect lower values of amin/amax or a0 (depending on the model) in these three regions when the Hα gas phase is considered alone.

7.3 Life cycle of very small grains

The results of this analysis show a significant increase in the relative VSG abundance compared to the MW. They also show an increase in smaller VSGs compared to larger ones in HII regions compared to the atomic medium. Some scenarios emphasize that while Galactic dust might mainly be produced by O-rich AGB stars (67% and 20% from O-rich and C-rich AGBs; Gehrz 1989), most of the dust in the Magellanic Clouds might originate from carbon stars (extreme AGBs, i.e., mostly embedded carbon stars) with a dust production reaching 61% and 66% in the LMC and SMC (Boyer et al. 2012). The dust produced in these environments might consist of small carbon grains. Additional sources of dust production are supernovae, although their dust-destruction rates remain poorly constrained, and dust growth in the ISM. Another explanation of the enhancement of small carbon grains might be the destruction of large grains into smaller grains.

Two options might explain the enhancement of small VSGs in HII regions. First, (small) VSGs might be formed in HII regions rather than in molecular clouds, as proposed by Paradis et al. (2019). These authors reported that the relative abundance of VSGs is enhanced in the ionized phase of the gas, regardless of the nature of the clouds, that is, quiescent or forming stars, and independently of the intensity of the radiation field. These VSGs might be large PAH clusters or cationic PAH clusters (Rapacioli et al. 2005, 2011; Roser and Ricca 2020). Regardless of the nature of these grain species, grain growth might then occur in the atomic and molecular regions via accretion or coagulation, and might explain the presence of larger VSGs in the molecular and diffuse environments. To examine the hypothesis that PAH clusters might cause the VSG increase in the LMC, we present in Fig. 6 the relative PAH abundance for each model (except for AJ13) and region. First, for all models, the PAH relative abundance is lower than the Galactic value, which agrees with the hypothesis of the presence of PAH clusters. However, Fig. 6 does not show any trend with the nature of the region, unlike in the case of the VSGs. This absence of a real trend suggests that other processes might occur. For instance, the BG component might be also affected by strong shocks and turbulence and might contribute to the increase in the relative abundance of VSGs, as observed in the LMC. In this second option, the largest VSGs might be destroyed in HII regions in shocks, resulting in an increase in the number of the smallest VSGs. This effect might be the result of supernova explosions or turbulence. This hypothesis might explain the changes in the size of the VSG population. As these two processes most likely occur together, the VSG component might include two distinct populations: small VSGs originating from large PAH clusters, or cationic PAHs clusters that mainly formed in HII regions, and large VSGs resulting from BG destruction.

Heiles et al. (2000) analyzed the Barnard Loop HII region and reported an increase in the 60 μm emission. They showed that this increase is not the consequence of the presence of warm large grains and proposed that it might be due to an enhancement of the VSG population relative to BGs in the ionized region compared to the global neutral medium. As a consequence, it appears that the increase in the relative VSG abundance in the ionized medium might be a more general result than just an isolated result concerning the LMC. Jones et al. (1996) developed an analytical model to derive the fragment size distribution as well as the final crater mass in grain-grain collisions depending on different parameters (grain properties, sizes, and collision velocity). They found that grain shattering leads to the redistribution of the dust mass from large grains into smaller grains. More recently, Hirashita (2010) theoretically studied interstellar shattering of large grains (a~0.1 μm) to explain the production of small grains. They were able to reproduce the small-grain abundance derived by Mathis et al. (1983) in the warm neutral medium. They also showed that additional shattering in the warm ionized medium could destroy carbonaceous grains with a size of ~0.01 μm and generate smaller grains. Conversely, silicate grains are harder to shatter than graphite. However, in this study (see Table 5), the abundances of small silicates compared to large silicates are high in four regions with DL07. According to the theory of Hirashita (2010), we cannot explain this enhancement of small silicates as caused by large silicate destruction or through another source of production.

A recent analysis of the turbulence in the LMC (Szotkowski et al. 2019) reported spatial variations of HI turbulent properties. Turbulence is often characterized by estimating the spatial power spectrum (SPS) of the intensity fluctuations. The authors pointed out several localized steepenings of the small-scale SPS slope around HII regions and around 30 Doradus in particular, which agrees with numerical simulations (Grisdale et al. 2017, 2019) that suggest a steepening of the SPS slope due to stellar feedback that erodes and destroys small clouds. This study agrees with the possible additional grain shattering in the ionized medium, that is, in HII regions, where we observe the smallest VSG populations compared to the atomic and molecular environments.

thumbnail Fig. 6

Ratio of the PAHs to the large-grain component abundance for the different models and for the ten regions using different RFs (Mathis in black, 4 Myr in red, 60 Myr in blue, and 600 Myr in green). The Galactic value is indicated with the gray line.

7.4 Extinction

For the four tested models, it is always possible to find a set of parameters that correctly fits the emission part of the SED. Fitting the dust emission from the NIR to FIR wavelength therefore does not allow us to unambiguously determine which dust models reproduce the observations best. In this context, it is interesting to determine wether these best-fit models also show similar extinction curves, or if extinction data might help to differentiate between the different grain models.

It is difficult to compare the modeled extinction curves with observations because no extinction curves are available that are associated with the studied regions. In the Milky Way, a wide variety of possible shapes of these curves has been observed (Papaj et al. 1991; Megier et al. 1997; Barbaro et al. 2001; Wegner 2002; Fitzpatrick and Massa 2007; Gordon et al. 2021), and we can similarily expect the extinction curves in the LMC to have different profiles. This was observed by Gordon et al. (2003), who found that the LMC-averaged extinction curve is characterized by a stronger far-UV rise than the curve of the MW, and by a weaker 2175 Å bump (see for instance LMC2-supershell, near the 30 Doradus star-bursting region). The authors argued that the difference between the extinction curves of the Magellanic Clouds and the MW might arise because the observed environments are different. For the Magellanic Clouds, the extinction curves are observed in active star formation regions in which large grains might be destroyed by strong shocks and/or UV photons. Nevertheless, the sample of the LMC extinction curves is quite limited compared to that of our Galaxy.

The extinction curves calculated by DustEM for the different dust best-fit models are shown in Figs. 7 and A.4. For comparison, we also plot the average LMC and LMC2 supershell extinction curves (Gordon et al. 2003) and the average Galactic extinction curve that is representative of the ISM (Cardelli et al. 1989). The predicted extinction curves are highly model dependent, are different from one region to the next, and are sensitive to the RF. According to the model predictions computed using the Mathis RF (Fig. 7), the extinction curves of the DL07 and DBP09 models show a prominent 2175 Å bump followed by a fast far-UV rise in the case of DL07 for all regions. No such strong bumps are observed in our Galaxy or in the Magellanic Clouds. On the other hand, the extinction curves of all regions modeled with MC11 have large and flattened bumps (except for SSDR7, which shows no bump at all). When the DBP90 model is used, the bump in diffuse or molecular increases strongly compared to HII regions. The extinction curves calculated with the AJ13 model show different behaviors depending on the region, with a weaker 2175 Å bump in diffuse or molecular regions than in ionized regions. The UV bumps are significantly stronger in the extinction curves calculated with DL07 and DBP90 than with AJ13, which are closest to the observed extinction curves.

A change in the RF has a strong effect on the extinction curves, and more specifically, on the 2175 Å bump, whose strength decreases when the strength of the RF increases (see Figs. 7 and A.4). However, for some regions and models, the bump remains unchanged, for example, for SSDR9 and DL07. The extinction curves calculated with the 4 Myr RF agree better with the Galactic and LMC behaviors, in contrast with those calculated using Mathis RF and 60–600 Myr RFs. In models using the 4 Myr RF, the extinction curves calculated for DL07 and DBP90 models show the strongest 2175 Å bumps, whereas MC10 shows very flat bumps and a UV rise for almost all the regions. The AJ13 model seems to minimize the discrepancies between predictions and observations, which makes this model more compatible with LMC curves than the other models.

Even if it is hard to interpret these results because we lack an extinction curve associated with the different studied regions, no model seems to give a satisfactory prediction. However, we now know that the dust emission models give very distinct behaviors in terms of extinction predictions. This point is important and should be used to refine the constraints on the dust models themselves. For instance, if models were to consider a two-population VSG component, as discussed in Sect. 7.3, with only one carbonaceous component such as large PAH clusters or cationic PAHs impacting the 2175 Å bump, they might better reconcile with extinction curves. Moreover, it appears that the extinction curves can also play an important role to better constrain the RF of each environments. These preliminary results in terms of the extinction description appear to indicate that the Mathis RF is not best suited for use in the Magellanic Clouds, and they suggest that stronger RFs might be necessary.

Recently, new cosmic dust models have been or will soon be made available (Siebenmorgen 2023; Hensley and Draine 2023; Ysard 2020). It would be interesting to continue this study by using these models to fit the LMC observations, both in emission and extinction. Finally, this work highlights the fact that future studies should, if possible, simultaneously fit dust emission and extinction.

thumbnail Fig. 7

Extinction curves for each region derived from the different dust models with Mathis RF (top) and 4 Myr RF (bottom). The averaged Galactic and LMC extinction curves are given in black diamonds and dark gray crosses for comparison. The LMC2 supershell extinction curve is presented with light gray triangles. We caution that the scales have been chosen to show the difference between the different models in a clear way, and this must be taken into account in the comparisons.

8 Conclusion

Using Spitzer IRS, MIPS SED, and photometric data combined with Berschel data, we modeled the spectra in diffuse, molecular, and HII regions of the LMC. We compared four distinct dust models available in the DustEM package: Jones et al. (2013, AJ13), Compiègne et al. (2011, MC11), Draine and Li (2007, DL07), and an updated version of Désert et al. (1990, DBP90). To verify the robustness of the results, we adopted four different radiation fields (interstellar RF or Mathis, stellar clusters with various ages: 4Myr, 60Myr, and 600 Myr). None of the models is able to reproduce the MIR-to-FIR emission using the Galactic standard parameters even when the abundances of the dust components and the radiation field strength were allowed to vary. Changes in the size distribution and abundances of the dust component that dominates the MIR-to-FIR emission (commonly referred to as very small grains, or VSGs) are needed to reasonably fit the dust emission spectra.

One of the first results of this analysis is the significant increase in the VSG abundance relative to the large-grain (BG) component in the LMC compared to the Milky Way. Changes in the size distribution and abundance of dust have a clear impact on the contribution of the VSG emission in the submm. Depending on the model, the VSG component can strongly dominate the submm emission, especially when the standard Mathis RF is used. Although no correlation could be shown between this strong VSG emission in the submm and the type of environment, we recommend caution when the FIR-to-submm emission in the LMC is analyzed using the large-grain component alone. Small carbon grains might partly cause the global submm-mm flattening observed in the LMC, even if other processes might be at play to explain local variations observed in the LMC and the Milky Way. The 70 μm emission excess reported in previous studies of the Magellanic Clouds might result from distinct VSG properties (size distribution and abundances) compared to our Galaxy.

Another important result is the increase in the number of small VSGs (and a decrease in large VSGs) in HII regions when compared to the diffuse regions of the LMC. In contrast to our Galaxy, where dust is probably mainly be produced by O-rich AGB stars, some dust in the LMC might come from C-rich AGB stars (extreme AGBs, i.e., mosty embedded carbon stars). The presence of small VSGs in HII regions might be explained by i) the formation of small VSGs in HII regions (rather than in molecular clouds); grain growth might occur in the diffuse and molecular medium via accretion or coagulation processes. It might also be explained by ii) the destruction of the largest VSGs and BG component in HII regions by shocks resulting from supernova explosions or turbulence. If these two scenarios take place, the VSG component might include two populations: small VSGs resulting from large PAH clusters or cationic PAH clusters, and large VSGs resulting from BG destruction.

The extinction curves calculated by DustEM show a great diversity of behaviors in the dust models and radiation fields. The AJ13 model shows reasonable predictions compared to the common behaviors. In the LMC, stronger RFs appear to reproduce the shape of the extinction curve better, in particular, by reducing the strong 2175 Åbump predicted by the models. Observations in the LMC are in that sense important to better constrain the dust models (and more specifically, the VSG component), but also to better constrain the RFs. Further studies that simultaneously fit dust emission and extinction and/or use the latest grain models should provide better constraints on the properties of grains in the LMC and other galaxies.

Acknowledgements

We acknowledge the use of the DustEM package. D.P. acknowledges W. Marty, and D. Rambaud for very useful discussions.

Appendix A Additional material

thumbnail Fig. A.1

Modeling of the SEDs of the ten regions with different dust models and free parameters (XISRF, dust abundances, and small-grain dust size distribution) using the 4 Myr RF. The observations are shown in black (Spitzer IRS SS and LL, MIPS SED, MIPS 160 μm, Berschel Photometric PACS 160 mic, and SPIRE 250 μm, 350 μm, and 500 μm data). The total modeled SED is shown as a red line. The other colored lines correspond to the different dust components of the models (see Fig. 1 or 2). The dashed line represents the additional NIR continuum. Blue asterisks show the color-corrected brightness derived from the models. Each column shows the fit using different dust models (from left to right: AJ13, MC11, DL07, and DBP90). Each row presents a different region. The figure continues on the next page.

Table A.1

Same table as Table 3, but using a 60 Myr (top) and 600 Myr RF (bottom).

thumbnail Fig. A.2

Modeling of the SEDs of the ten regions with different dust models and free parameters (XISRF, dust abundances, and small grains dust size distribution) using the 60 Myr RF. The observations are shown in black (Spitzer IRS SS and LL, MIPS SED, MIPS 160 μm, Berschel Photometric PACS 160 mic, and SPIRE 250 μm, 350 μm and 500 μm data). The total modeled SED is shown as a red line. The other colored lines correspond to the different dust components of the models (see Fig. 1 or 2). The dashed line represents the additional NIR continuum. Blue asterisks show the color-corrected brightness derived from the models. Each column shows the fit using different dust models (from left to right: AJ13, MC11, DL07, and DBP90). Each row presents a different region. The figure continues on the next page.

Table A.2

Same table as Table 4, but using a 60 Myr (top) and 600 Myr RF (bottom).

thumbnail Fig. A.3

Modeling of the SEDs of the ten regions with different dust models and free parameters (XISRF, dust abundances, and small grains dust size distribution) using the 600 Myr RF. The observations are shown in black (Spitzer IRS SS and LL, MIPS SED, MIPS 160 μm, Herschel Photometric PACS 160 mic, and SPIRE 250 μm, 350 μm and 500 μm data). The total modeled SED is shown as a red line. The other colored lines correspond to the different dust components of the models (see Fig. 1 or 2). The dashed line represents the additional NIR continuum. Blue asterisks show the color-corrected brightness derived from the models. Each column shows the fit using different dust models (from left to right: AJ13, MC11, DL07, and DBP90). Each row presents a different region. The figure continues on the next page.

Table A.3

Same table as Table 5, but using a 60 Myr (top) and 600 Myr RF (bottom).

Table A.4

Same table as Table 6, but using a 60 Myr (top) and 600 Myr RF (bottom).

thumbnail Fig. A.4

Extinction curves for each region derived from the different dust models with a 60 Myr RF (left) and 600 Myr (right). The averaged Galactic and LMC extinction curves are plotted as black diamonds and dark gray crosses for comparison. The LMC2 supershell extinction curve is presented with light gray triangles. We caution that the scales have been chosen to show the difference between the different models in a clear way, and this must be taken into account in the comparisons.

Table A.5

Ratio of the small-grain component emission to the total dust emission for each best-fit model using the 60 Myr (top table) and 60 -Myr RF (bottom table) at 250, 500, 850, and 1100 μm.

References

  1. Balog, Z., Müller, T., Nielbock, M., et al. 2014, Exp. Astron., 37, 129 [Google Scholar]
  2. Barbaro, G., Mazzei, P., Morbidelli, L., et al. 2001, A&A, 365, 157 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  3. Bernard, J.-P., Reach, W. T., Paradis, D., et al. 2008, AJ, 136, 919 [NASA ADS] [CrossRef] [Google Scholar]
  4. Bica, E., Claria, J. J., Dottori, H., et al. 1996, ApJS, 102, 57 [NASA ADS] [CrossRef] [Google Scholar]
  5. Bot, C., Boulanger, F., Lagache, G., et al. 2004, A&A, 423, 567 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  6. Bot, C., Ysard, N., Paradis, D., et al. 2010, A&A, 523, A20 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  7. Boyer, M. L., Srinivasan, S., Riebel, D., et al. 2012, ApJ, 748, 40 [Google Scholar]
  8. Boulanger, F., Abergel, A., Cesarsky, D., et al. 2000, ESA SP, 455, 91 [NASA ADS] [Google Scholar]
  9. Cardelli, J. A., Clayton, G. C., & Mathis, J. S. 1989, ApJ, 345, 245 [Google Scholar]
  10. Chastenet, J., Bot, C., Gordon, K. D., et al. 2017, A&A, 601, A55 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  11. Chastenet, J., Sandstrom, K., Chiang, I-Da, et al. 2019, ApJ, 876, 62 [NASA ADS] [CrossRef] [Google Scholar]
  12. Chastenet, J., Sandstrom, K., Chiang, I-Da, et al. 2021, ApJ, 912, 103 [NASA ADS] [CrossRef] [Google Scholar]
  13. Clark, C. J. R., Roman-Duval, J. C., Gordon, K. G., Bot, C., & Smith, M. W. L. 2021, ApJ, 921, 35 [CrossRef] [Google Scholar]
  14. Compiègne, M., Verstraete, L., Jones, A., et al. 2011, A&A, 525, A103 [Google Scholar]
  15. Dale, D. A., Aniano, G., Engelbracht, C. W., et al., 2012, ApJ, 745, 95 [NASA ADS] [CrossRef] [Google Scholar]
  16. Désert, F.-X., Boulanger, F., & Puget, J.-L. 1990, A&A, 237, 215 [Google Scholar]
  17. Dickinson, C., Davies, R. D., Davis, R. J. 2003, MNRAS, 341, 369 [NASA ADS] [CrossRef] [Google Scholar]
  18. Draine, B. T., & Li, A. 2001, ApJ, 551, 807 [NASA ADS] [CrossRef] [Google Scholar]
  19. Draine, B. T., & Li, A. 2007, ApJ, 657, 810 [CrossRef] [Google Scholar]
  20. Fazio, G. G., Hora, J. L., Allen, L. E., et al. 2004, ApJS, 154, 10 [Google Scholar]
  21. Fitzpatrick, E. L., & Massa, D. 2007, ApJ, 663, 320 [Google Scholar]
  22. Galametz, M., Kennicutt, R. C., Albrecht, M., et al. 2012, MNRAS, 425, 763 [NASA ADS] [CrossRef] [Google Scholar]
  23. Galametz, M., Hony, S., Galliano, F., et al. 2013, MNRAS, 431, 1596 [NASA ADS] [CrossRef] [Google Scholar]
  24. Galliano, F., Hony, S., Bernard, J.-P., et al. 2011, A&A, 536, A88 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  25. Gaustad, J. E., McCullough, P. R., Rosing, W., & Van Buren, D. 2001, PASP, 113, 1326 [NASA ADS] [CrossRef] [Google Scholar]
  26. Gehrz, R. 1989, IAU Symp., 135, 445 [Google Scholar]
  27. Gordon, K. D., Clayton, G. C., Misselt, K. A., et al. 2003, ApJ, 594, 279 [NASA ADS] [CrossRef] [Google Scholar]
  28. Gordon, K. D., Rieke, G. H., Engelbracht, C. W., et al. 2005, PASP, 117, 503 [NASA ADS] [CrossRef] [Google Scholar]
  29. Gordon, K. D., Misselt, K., Pendleton, Y., et al. 2019, BAAS, 51, 458 [Google Scholar]
  30. Gordon, K. D., Misselt, K. A., Bouwman J., et al. 2021, ApJ, 916, 33 [NASA ADS] [CrossRef] [Google Scholar]
  31. Grisdale, K., Agertz, O., Romeo, A. B., et al. 2017, MNRAS, 466, 1093 [CrossRef] [Google Scholar]
  32. Grisdale, K., Agertz, O., Renaud, F., et al. 2019, MNRAS, 486, 5482 [NASA ADS] [CrossRef] [Google Scholar]
  33. Heiles, C., Haffner, L. M., Reynolds, R. J., & Tufte, S. L. 2000, ApJ, 536, 335 [NASA ADS] [CrossRef] [Google Scholar]
  34. Hensley, B. S., & Hensley, B. S., & Draine, B. T. 2023, ApJ, 948, 55 [Google Scholar]
  35. Hirashita, H. 2010, MNRAS, 407, 49 [Google Scholar]
  36. Hughes, A., Wong, T., Ott, J., et al. 2010, MNRAS, 406, 2065 [NASA ADS] [Google Scholar]
  37. Jones, A. P., Tielens, A. G. G. M., & Hollenbach, D. J. 1996, ApJ, 469, 740 [Google Scholar]
  38. Jones, A. P., Fanciullo, L., Köhler, M., et al. 2013, A&A, 558, A62 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  39. Juvela, M., Ristorcelli, I., Pelkonen, V.-M., et al. 2011, A&A, 527, A111 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  40. Juvela, M., He, J., Pattle, K., et al., 2018, A&A, 612, A71 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  41. Kawamura, A., Mizuno, Y., Minamidani, T., et al. 2009, ApJS, 184, 1 [NASA ADS] [CrossRef] [Google Scholar]
  42. Kemper, F., Woods, P. M., Antoniou, V., et al. 2010, PASJ, 122, 683 [CrossRef] [Google Scholar]
  43. Kim, S., Staveley-Smith, L., Dopita, M. A., et al. 1998, ApJ, 503, 674 [Google Scholar]
  44. Kim, S., Staveley-Smith, L., Dopita, M. A., et al. 2003, ApJS, 148, 473 [Google Scholar]
  45. Kohler, M., Guillet, V., & Jones, A. 2011, A&A, 528, A96 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  46. Kohler, M., Stepnik, B., Jones, A., et al. 2012, A&A, 548, A61 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  47. Lagache, G., Abergel, A., Boulanger, F., et al. 1999, A&A, 344, 322 [Google Scholar]
  48. Lee, M. Y., Stanimirovic, S., Murray, C. E., et al. 2015, AJ, 809, 56 [Google Scholar]
  49. Leroy, A., Bolatto, A., Gordon, K., et al. 2011, ApJ, 737, 12 [CrossRef] [Google Scholar]
  50. Lisenfeld, U., Israël, F. P., Stil, J. M., & Sievers, A. 2002, A&A, 394, 823 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  51. Lu, N., Smith, P. S., Engelbracht, C. W., et al. 2008, PASP, 120, 328 [NASA ADS] [CrossRef] [Google Scholar]
  52. Mathis, J. S., Mezger, P. G., & Panagia, N. 1983, A&A, 128, 212 [NASA ADS] [Google Scholar]
  53. Megier, A., Krelowski, J., Patriarchi, P., & Aiello, S. 1997, MNRAS, 292, 853 [NASA ADS] [CrossRef] [Google Scholar]
  54. Meisner, A. M., & Finkbeiner, D. P. 2015, ApJ, 798, 88 [Google Scholar]
  55. Meixner, M., Gordon, K. D., Indebetouw, R., et al. 2006, ApJ, 132, 2268 [Google Scholar]
  56. Meixner, M., Galliano, F., Hony, S., et al. 2010, A&A, 518, L71 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  57. Mény, C., Gromov, V., Boudet, N., et al. 2007, A&A, 468, A171 [CrossRef] [EDP Sciences] [Google Scholar]
  58. Miville-Deschênes, M.-A., & Lagache, G. 2005, ApJS, 157, 302 [Google Scholar]
  59. Nashimoto, M., Hattori, M., Génova-Santos, R., & Poidevin, F. 2020, PASJ, 72, 6 [NASA ADS] [CrossRef] [Google Scholar]
  60. Oliveira, J. M., van Loon, J. Th., Sewilo, M., et al. 2019, MNRAS, 490, 3909 [NASA ADS] [CrossRef] [Google Scholar]
  61. Papaj, J., Wegner, W. & Krelowski, J. 1991, MNRAS, 252, 403 [NASA ADS] [CrossRef] [Google Scholar]
  62. Paradis, D., Reach, W. T., Bernard, J.-P., et al. 2009, AJ, 138, 196 [Google Scholar]
  63. Paradis, D., Paladini, R., Noriega-Crespo, A., et al. 2011a, ApJ, 735, 6 [NASA ADS] [CrossRef] [Google Scholar]
  64. Paradis, D., Bernard, J.-P., Mény, C., & Gromov, V. 2011b, A&A, 534, A118 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  65. Paradis, D., Paladini, R., Noriega-Crespo, A., et al. 2012, A&A, 537, A113 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  66. Paradis, D., Mény, C., Noriega-Crespo, A., et al. 2014, A&A, 572, A37 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  67. Paradis, D., Mény, C., Juvela, M., et al. 2019, A&A, 627, A15 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  68. Planck Collaboration XVII. 2011, A&A, 536, A17 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  69. Planck Collaboration XI. 2014, A&A, 571, A11 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  70. Planck Collaboration Int. XIV. 2014, A&A, 564, A45 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  71. Planck Collaboration Int. XVII. 2014, A&A, 566, A55 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  72. Rapacioli, M., Joblin, C., & Boissel, P. 2005, A&A, 429, 193 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  73. Rapacioli, M., Spiegelman, F., Joalland, B., et al. 2011, EAS, 46, 223 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  74. Rieke, G. H., Young, E. T., Engelbracht, C. W., et al. 2004, ApJS, 154, 25 [Google Scholar]
  75. Rémy-Ruyer, A., Madden, S. C., Galliano, F., et al. 2015, A&A, 582, A121 [Google Scholar]
  76. Roman-Duval, J., Gordon, K. D., Meixner, M., et al. 2014, ApJ, 797, 86 [NASA ADS] [CrossRef] [Google Scholar]
  77. Roman-Duval, J., Bot, C., Chastenet, J., & Gordon, K. G. 2017, ApJ, 841, 72 [NASA ADS] [CrossRef] [Google Scholar]
  78. Roman-Duval, J., Jenkins, E. B., Tchernyshyov, K., et al. 2022, ApJ, 935, 105 [Google Scholar]
  79. Roser, J. E., & Ricca, A. 2020, IAU Symp., 350, 415 [NASA ADS] [Google Scholar]
  80. Siebenmorgen, R. 2023, A&A, 670, A115 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  81. Smith, J. D. T., Armus, L., Dale, D. A., et al. 2007, PASP, 119, 1133 [NASA ADS] [CrossRef] [Google Scholar]
  82. Spitzer, L. 1978, Physical Processes in the Interstellar Medium (New York: Wiley-Interscience) [Google Scholar]
  83. Staveley-Smith, L., Kim, S., Calabretta, M. R., et al. 2003, MNRAS, 339, 87 [NASA ADS] [CrossRef] [Google Scholar]
  84. Stephens, I. W., Evans, J. M., Xue, R., et al. 2014, ApJ, 784, 147 [NASA ADS] [CrossRef] [Google Scholar]
  85. Szotkowski, S., Yoder, D., Stanimirovic, S., et al. 2019, ApJ, 887, 111 [NASA ADS] [CrossRef] [Google Scholar]
  86. Tibbs, C., Flagey, N., Paladini, R., et al. 2011, MNRAS, 418, 1889 [NASA ADS] [CrossRef] [Google Scholar]
  87. Trewhella, M., Davies, J. I., Alton, A. B., et al. 2000, ApJ, 543, 153 [NASA ADS] [CrossRef] [Google Scholar]
  88. Wegner, W. 2002, Balt. Astron., 11, 1 [NASA ADS] [Google Scholar]
  89. Wong, T., Hughes, A., Ott, J., et al. 2011, ApJS, 197, 16 [NASA ADS] [CrossRef] [Google Scholar]
  90. Ysard, N. IAU Symp., 2020, 350, 53 [NASA ADS] [Google Scholar]

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

Herschel is an ESA space observatory with science instruments provided by European-led Principal Investigator consortia and with important participation from NASA.

All Tables

Table 1

Characteristics of the studied regions.

Table 2

Best-fit reduced χ2 obtained for the modeling of the full SED of the ten regions and for the four adopted dust models (between 5 μm and 500 μm), using the Mathis RF.

Table 3

Best-fit parameters for the ten regions obtained with AJ13 using the Mathis RF (top table) and 4 Myr RF (bottom table).

Table 4

Best-fit parameters for the ten regions obtained with MC11 using the Mathis RF (top table) and 4 Myr RF (bottom table).

Table 5

Best-fit parameters for the ten regions obtained with DL07 using the Mathis RF (top table) and 4 My RF (bottom table).

Table 6

Best-fit parameters for the ten regions obtained with DBP90 using the Mathis RF (top and middle tables) and 4 My RF (bottom table).

Table 7

Ratio of the small-grain component emission to the total dust emission for each best-fit model using the Mathis (top table) and 4 Myr RF (bottom table) at 250, 500, 850, and 1100 μm.

Table A.1

Same table as Table 3, but using a 60 Myr (top) and 600 Myr RF (bottom).

Table A.2

Same table as Table 4, but using a 60 Myr (top) and 600 Myr RF (bottom).

Table A.3

Same table as Table 5, but using a 60 Myr (top) and 600 Myr RF (bottom).

Table A.4

Same table as Table 6, but using a 60 Myr (top) and 600 Myr RF (bottom).

Table A.5

Ratio of the small-grain component emission to the total dust emission for each best-fit model using the 60 Myr (top table) and 60 -Myr RF (bottom table) at 250, 500, 850, and 1100 μm.

All Figures

thumbnail Fig. 1

Modeling of the SEDs of the ten regions with different dust models and free standard parameters (XISRF and dust abundances), using the Mathis RF. The observations are shown in black (Spitzer IRS SS and LL, MIPS SED, MIPS 160 μm, Herschel Photometric PACS 160 μm, and SPIRE 250 μm, 350 μm, and 500 μm data). The total modeled SED is shown as a red line. The other colored lines correspond to the different dust components of the models. The dashed line represents the additional NIR continuum. Blue asterisks show the color-corrected brightness derived from the models. The orange diamonds in the DBP90 panels show the MIPS 70 μm photometric data normalized to the integrated flux in the MIPS-SED band. Each column shows the fit using different dust models (from left to right: AJ13, MC11, DL07, and DBP90). Each row presents a different region. The figure continues on the next page.

In the text
thumbnail Fig. 2

Modeling of the SEDs of the ten regions with different dust models and free parameters (XISRF, dust abundances, and small-grain dust size distribution) using the Mathis RF. The observations are shown in black (Spitzer IRS SS and LL, MIPS SED, MIPS 160 μm, Herschel Photometric PACS 160 mic, and SPIRE 250 μm, 350 mic, and 500 μm data) are shown in black. The total modeled SED is shown as a red line. The other colored lines correspond to the different dust components of the models. The dashed line represents the additional NIR continuum. Blue asterisks show the color-corrected brightness derived from the models. The orange diamonds in the DBP90 panels show the MIPS 70 μm photometric data normalized to the integrated flux in the MIPS-SED band. Each column shows the fit using different dust models (from left to right: AJ13, MC11, DL07, and DBP90). Each row presents a different region. The figure continues on the next page.

In the text
thumbnail Fig. 3

Radiation field templates used for the SED modeling: Mathis (solid red line), 4 Myr (short dashed black line), 60 Myr (dotted green line), and 600 Myr stellar clusters (long dashed blue line).

In the text
thumbnail Fig. 4

Parameters of the small grain size distribution for each model, each region, and for the different RFs: Mathis (black), 4 Myr (red), 60 Myr (blue) and 600 Myr (green). Panel A: a0 (in nm) for MC11 and DL07; panel B: amin and amax (in nm) for DBP90; panel C: amin and amax (in nm) for AJ13 and panel D: α for AJ13. Arrows indicate high values of the amax parameter.

In the text
thumbnail Fig. 5

Ratio of the small- to large-grain component for the different models and for the ten regions using different RFs (Mathis in black, 4 Myr in red, 60 Myr in blue, and 600-Myr in green). The Galactic value is indicated with the gray line.

In the text
thumbnail Fig. 6

Ratio of the PAHs to the large-grain component abundance for the different models and for the ten regions using different RFs (Mathis in black, 4 Myr in red, 60 Myr in blue, and 600 Myr in green). The Galactic value is indicated with the gray line.

In the text
thumbnail Fig. 7

Extinction curves for each region derived from the different dust models with Mathis RF (top) and 4 Myr RF (bottom). The averaged Galactic and LMC extinction curves are given in black diamonds and dark gray crosses for comparison. The LMC2 supershell extinction curve is presented with light gray triangles. We caution that the scales have been chosen to show the difference between the different models in a clear way, and this must be taken into account in the comparisons.

In the text
thumbnail Fig. A.1

Modeling of the SEDs of the ten regions with different dust models and free parameters (XISRF, dust abundances, and small-grain dust size distribution) using the 4 Myr RF. The observations are shown in black (Spitzer IRS SS and LL, MIPS SED, MIPS 160 μm, Berschel Photometric PACS 160 mic, and SPIRE 250 μm, 350 μm, and 500 μm data). The total modeled SED is shown as a red line. The other colored lines correspond to the different dust components of the models (see Fig. 1 or 2). The dashed line represents the additional NIR continuum. Blue asterisks show the color-corrected brightness derived from the models. Each column shows the fit using different dust models (from left to right: AJ13, MC11, DL07, and DBP90). Each row presents a different region. The figure continues on the next page.

In the text
thumbnail Fig. A.2

Modeling of the SEDs of the ten regions with different dust models and free parameters (XISRF, dust abundances, and small grains dust size distribution) using the 60 Myr RF. The observations are shown in black (Spitzer IRS SS and LL, MIPS SED, MIPS 160 μm, Berschel Photometric PACS 160 mic, and SPIRE 250 μm, 350 μm and 500 μm data). The total modeled SED is shown as a red line. The other colored lines correspond to the different dust components of the models (see Fig. 1 or 2). The dashed line represents the additional NIR continuum. Blue asterisks show the color-corrected brightness derived from the models. Each column shows the fit using different dust models (from left to right: AJ13, MC11, DL07, and DBP90). Each row presents a different region. The figure continues on the next page.

In the text
thumbnail Fig. A.3

Modeling of the SEDs of the ten regions with different dust models and free parameters (XISRF, dust abundances, and small grains dust size distribution) using the 600 Myr RF. The observations are shown in black (Spitzer IRS SS and LL, MIPS SED, MIPS 160 μm, Herschel Photometric PACS 160 mic, and SPIRE 250 μm, 350 μm and 500 μm data). The total modeled SED is shown as a red line. The other colored lines correspond to the different dust components of the models (see Fig. 1 or 2). The dashed line represents the additional NIR continuum. Blue asterisks show the color-corrected brightness derived from the models. Each column shows the fit using different dust models (from left to right: AJ13, MC11, DL07, and DBP90). Each row presents a different region. The figure continues on the next page.

In the text
thumbnail Fig. A.4

Extinction curves for each region derived from the different dust models with a 60 Myr RF (left) and 600 Myr (right). The averaged Galactic and LMC extinction curves are plotted as black diamonds and dark gray crosses for comparison. The LMC2 supershell extinction curve is presented with light gray triangles. We caution that the scales have been chosen to show the difference between the different models in a clear way, and this must be taken into account in the comparisons.

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.