The Herschel Exploitation of Local Galaxy Andromeda (HELGA): IV. Dust scaling relations at sub-kpc resolution
S. Viaene, J. Fritz, M. Baes, G.J. Bendo, J.A.D.L. Blommaert, M. Boquien, A. Boselli, L. Ciesla, L. Cortese, I. De Looze, W.K. Gear, G. Gentile, T.M. Hughes, T. Jarrett, O. Ł. Karczewski, M.W.L. Smith, L. Spinoglio, A. Tamm, E. Tempel, D. Thilker, J. Verstappen
AAstronomy & Astrophysics manuscript no. HELGAIV_arxiv c (cid:13)
ESO 2018October 13, 2018
The
Herschel (cid:63)
Exploitation of Local Galaxy Andromeda (HELGA)
IV. Dust scaling relations at sub-kpc resolution
S. Viaene , J. Fritz , M. Baes , G.J. Bendo , J.A.D.L. Blommaert , , M. Boquien , A. Boselli , L. Ciesla , L. Cortese ,I. De Looze , W.K. Gear , G. Gentile , , T.M. Hughes , T. Jarrett , , O. Ł. Karczewski , M.W.L. Smith , L.Spinoglio , A. Tamm , E. Tempel , , D. Thilker , and J. Verstappen Sterrenkundig Observatorium, Universiteit Gent, Krijgslaan 281, B-9000 Gent, Belgiume-mail: [email protected] Jodrell Bank Centre for Astrophysics, School of Physics and Astronomy, University of Manchester, Oxford Road, ManchesterM13 9PL, UK Instituut voor Sterrenkunde, Katholieke Universiteit Leuven, Celestijnenlaan 200D, B-3001 Leuven, Belgium Department of Physics and Astrophysics, Vrije Universiteit Brussel, Pleinlaan 2, 1050 Brussels, Belgium Institute of Astronomy, University of Cambridge, Madingley Road, Cambridge, CB3 0HA, UK Laboratoire d’Astrophysique de Marseille, UMR 6110 CNRS, 38 rue F. Joliot-Curie, F-13388 Marseille, France University of Crete, Department of Physics, Heraklion 71003, Greece Centre for Astrophysics & Supercomputing, Swinburne University of Technology, Mail H30 - PO Box 218, Hawthorn, VIC 3122,Australia School of Physics and Astronomy, Cardi ff University, Queens Buildings, The Parade, Cardi ff CF24 3AA, UK Infrared Processing and Analysis Center, California Institute of Technology, Pasadena, CA 91125, USA Astronomy Department, University of Cape Town, Rondebosch 7701, South Africa Department of Physics and Astronomy, University of Sussex, Brighton, BN1 9QH, UK. Istituto di Astrofisica e Planetologia Spaziali, INAF-IAPS, Via Fosso del Cavaliere 100, I-00133 Roma, Italy Tartu Observatory, Observatooriumi 1, 61602 Tõravere, Estonia National Institute of Chemical Physics and Biophysics, Rävala pst 10, Tallinn 10143, Estonia Department of Physics and Astronomy, Johns Hopkins University, 3701 San Martin Drive, Baltimore, MD 21218, USOctober 13, 2018
ABSTRACT
Context.
Dust and stars play a complex game of interactions in the interstellar medium and around young stars. The imprints of theseprocesses are visible in scaling relations between stellar characteristics, star formation parameters, and dust properties.
Aims.
In the present work, we aim to examine dust scaling relations on a sub-kpc resolution in the Andromeda galaxy (M31). Thegoal is to investigate the properties of M31 on both a global and local scale and compare them to other galaxies of the local universe.
Methods.
New
Herschel observations are combined with available data from GALEX, SDSS, WISE, and
Spitzer to construct adataset covering UV to submm wavelengths. All images were brought to the beam size and pixel grid of the SPIRE 500 µ m frame.This divides M31 in 22437 pixels of 36 arcseconds in size on the sky, corresponding to physical regions of 137 ×
608 pc in the galaxy’sdisk. A panchromatic spectral energy distribution was modelled for each pixel and maps of the physical quantities were constructed.Several scaling relations were investigated, focussing on the interactions of dust with starlight.
Results.
We find, on a sub-kpc scale, strong correlations between M dust / M (cid:63) and NUV–r, and between M dust / M (cid:63) and µ (cid:63) (the stellarmass surface density). Striking similarities with corresponding relations based on integrated galaxies are found. We decompose M31in four macro-regions based on their FIR morphology; the bulge, inner disk, star forming ring, and the outer disk region. In the scalingrelations, all regions closely follow the galaxy-scale average trends and behave like galaxies of di ff erent morphological types. Thespecific star formation characteristics we derive for these macro-regions give strong hints of an inside-out formation of the bulge-diskgeometry, as well as an internal downsizing process. Within each macro-region, however, a great diversity in individual micro-regionsis found, regardless of the properties of the macro-regions. Furthermore, we confirm that dust in the bulge of M31 is heated only bythe old stellar populations. Conclusions.
In general, the local dust scaling relations indicate that the dust content in M31 is maintained by a subtle interplay ofpast and present star formation. The similarity with galaxy-based relations strongly suggests that they are in situ correlations, withunderlying processes that must be local in nature.
Key words. galaxies: individual: M31 - galaxies: ISM - infrared: ISM - galaxies: fundamental: parameters - dust, extinction - methods:observational (cid:63)
Herschel is an ESA space observatory with science instruments pro-vided by European-led Principal Investigator consortia and with impor-tant participation from NASA.
1. Introduction
The interstellar medium (ISM) harbours a rich variety of materi-als that all interact with one another through multiple chemody-namical processes. Hydrogen is by far the most abundant and oc-
Article number, page 1 of 22 a r X i v : . [ a s t r o - ph . GA ] A ug urs primarily in its neutral form, varying from warm ( ∼ ∼
80 K). In the very cold and dense environments, itis converted into molecular hydrogen (H ). The presence of ion-ized hydrogen (H ii ) increases as new stars irradiate the neutralgas. The formation of stars and their associated nucleosynthe-sis are the principal processes driving the chemical enrichmentof the ISM. A good fraction of the produced metals are lockedup in dust grains of various sizes. The micron-sized grains arethought to be silicic and carbonaceous in nature, while there isa population of nanometre-sized polycyclic aromatic hydrocar-bons (PAHs).Although it only accounts for a relatively small fraction ofthe ISM mass, interstellar dust plays a vital role as a catalyst inthe formation of H , which is crucial in the star formation pro-cess. At the same time, dust tends to absorb up to 50% of the op-tical and ultraviolet (UV) light of stars (Popescu & Tu ff s 2002),heavily a ff ecting our view of the universe. It re-emits the ab-sorbed energy at longer wavelengths in the mid-infrared (MIR),far-infrared (FIR), and submillimetre (submm) bands. To studythe ISM in greater detail, spatially resolved FIR observations arecrucial as they constrain the local dust distribution and proper-ties. Recent space missions were able to uncover the MIR tosubmm window in great detail with the Spitzer Space Telescope(Werner et al. 2004), the Herschel
Space Observatory (Pilbrattet al. 2010), and the Widefield Infrared Survey Explorer (WISE,Wright et al. 2010).Correlations between the main properties of dust, gas, andstars, also known as scaling relations , define a tight link betweenthese constituents. In the past, relations between the dust-to-gasratio on the one hand and metallicity or stellar mass on the otherhand were the only notable relations that were being investigated(Issa et al. 1990; Lisenfeld & Ferrara 1998; Popescu et al. 2002;Draine et al. 2007; Galametz et al. 2011). Only recently wereother scaling laws more systematically investigated. Namely,the ratio of dust to stellar mass, the specific dust mass, whichwas found to correlate with the specific star formation rate (i.e.star formation rate divided by the stellar mass, Brinchmann et al.2004; da Cunha et al. 2010; Rowlands et al. 2012; Smith et al.2012a), NUV–r colour (i.e. the di ff erence in absolute magnitudebetween the GALEX NUV and SDSS r band), and stellar masssurface density µ (cid:63) (Cortese et al. 2012; Agius et al. 2013).Each of the above scaling laws was derived on a galaxy-galaxy basis, considering galaxies as independent systems inequilibrium. In order to fully understand the coupling of dustwith stars and the ISM, we must zoom in on individual galax-ies. This is however troublesome at FIR / submm wavelengthsbecause of the limited angular resolution. Only in the last fewyears, we have been able, thanks to Herschel , to observe nearbygalaxies in the FIR and submm spectral domain whilst achiev-ing sub-kpc resolutions for the closest galaxies ( < . µ m band. Dust scaling relations on subgalactic scaleswere thus so far limited to gas-to-dust ratios for a handful of lo-cal galaxies (see e.g. Muñoz-Mateos et al. 2009a; Bendo et al.2010; Magrini et al. 2011; Sandstrom et al. 2012; Parkin et al.2012).Today, exploiting PACS (Poglitsch et al. 2010) and SPIRE(Gri ffi n et al. 2010), plus the 3.5-metre mirror onboard Herschel ,IR astronomy has gone a leap forward. Spatial resolutions andsensitivities have been reached that allow us, for the first time,to accurately characterise the dust emission in distinct regions ofnearby galaxies.The observed spectral energy distribution (SED) at these fre-quencies can be reproduced by means of theoretical models. Inparticular, a modified black body can be fitted to the observed FIR ( λ > µ m) data extracted from sub-kpc regions in galax-ies (see e.g. Smith et al. 2010; Hughes et al. 2014). While stilla step forward with respect to previous works, this approach in-evitably su ff ers from some simplifications. For example, if dustis heated by di ff erent sources, it cannot be truthfully representedby a single temperature component. Other studies make use ofthe physical dust models from Draine & Li (2007), which coverthe 1 − µ m wavelength range (e.g. Foyle et al. 2013 forM83; Aniano et al. 2012 for NGC 626 and NGC 6946; and,more recently, Draine et al. 2014 for M31). These studies usedata at shorter wavelengths as well, being thus able to fully sam-ple the spectral range of dust emission, including emission fromwarmer dust, small transiently heated grains and PAHs. Theymainly focus on the distribution and properties of the dust, notincluding any constraints on the radiation field which heats thedust from for example, UV / optical observations.To properly investigate scaling relations on a local scale, onewould ideally need a self–consistent model to derive the desiredphysical quantities. The complexity of a full stellar and chem-ical evolution model for galaxies, however, requires some sim-plifications. The models should treat both stellar and dust com-ponents, taking into account their influence on each other (theso–called dust energy balance). Panchromatic emission mod-elling of subgalactic regions has been carried out by MentuchCooper et al. (2012) for the Whirlpool galaxy (M51). However,the stellar and dust components were treated separately. Boquienet al. (2012, 2013), have performed panchromatic pixel-by-pixelfits of nearby star forming galaxies using CIGALE (Noll et al.2009), which does include a dust energy balance. They showedthat most of the free parameters could accurately be constrained,given a su ffi ciently large range of priors.We will perform panchromatic SED fitting, using the MAG-PHYS code (da Cunha et al. 2008). The code treats both stellarlight and dust emission at the same time, forcing an energy bal-ance. It has an extended library of theoretical SEDs based on thelatest version of the stellar population models from Bruzual &Charlot (2003) and physically motivated, multi-component dustmodels. Furthermore, it applies a Bayesian fitting method in-cluding a thorough error analysis.The proximity of ISM regions is crucial in order to obtain thedesired, sub-kpc spatial resolution. The closest giant molecularcloud systems are of course in our own Milky Way, but it isnot possible to probe the entire Galaxy. The Magellanic cloudsare the nearest galaxies as they are close satellites of the MilkyWay. These objects are, however, quite irregular and lower inmetallicity and in total mass, hence they do not represent thewell–evolved ISM of virialised large galaxies.Andromeda (M31) is the closest large galaxy at a distance D M31 =
785 kpc (McConnachie et al. 2005), which means everyarcsecond on the sky corresponds to 3 . = . M (cid:12) yr − , Ford et al. 2013) withan inclination of 77 ◦ and a position angle of its major axis of38 ◦ (McConnachie et al. 2005). The gas and dust components ofAndromeda have been extensively studied in the past (e.g. Wal-terbos & Schwering 1987; Montalto et al. 2009; Tabatabaei &Berkhuijsen 2010) using low–resolution data at FIR wavelengthsand simplified models.Although mapped in all wavelengths from UV to the FIRin the past, high–quality submm observations are thus far notavailable, yet these wavelengths are crucial to constraining theproperties of the cold dust. The Herschel
Exploitation of LocalGalaxy Andromeda (Fritz et al. 2012, hereafter paper I) is thefirst programme that mapped M31 from 100 µ m to 500 µ m with Article number, page 2 of 22. Viaene: Dust scaling relations in M31
Herschel , covering a large 5 . ◦ × . ◦ field centred around thegalaxy. Even at the sparsest Herschel resolution (36 (cid:48)(cid:48) at 500 µ m),physical scales of only 140 pc are resolved. Andromeda is con-sequently the best suited object for studying the ISM in great de-tail while allowing at the same time, the comparison with globalproperties.In Smith et al. (2012b) (hereafter paper II), we performed apixel-by-pixel SED fit to the Herschel data and map the maindust properties of Andromeda. Ford et al. (2013) (hereafter pa-per III) investigate the star formation law in M31 on both globaland local scales. A catalogue of giant molecular clouds was re-cently constructed by Kirk et al. (2014) (hereafter paper IV).We aim to expand on this work by carrying out an in-depthinvestigation of the dust scaling relations in Andromeda. Wedo this by fitting panchromatic spectral energy distribution mod-els to each statistically independent 36–arcsecond region in thegalaxy. In this way we have produced the largest and most com-plete view of the stars and ISM dust in a large spiral galaxy.The arrangement of the paper is as follows. In Sect. 2 wegive an overview of the data used and in Sect. 3 we briefly dis-cuss the processing of these data and our SED fitting method.Appendix A goes into more detail on the processing of multi-wavelength data. The results are given in Sect. 3, along withthe parameter maps of Andromeda. We analyse the dust scal-ing relations of Andromeda in Sect. 4. In Sect. 5 we present ourdiscussion and main conclusions.
2. The dataset
Modelling the full spectrum of a galaxy requires a fair num-ber of free parameters and consequently su ffi cient data pointsto sample the problem in a meaningful way. The Andromedagalaxy has been observed by many space borne telescopes suchas the Galaxy Evolution Explorer (GALEX, Martin et al. 2005), Spitzer , and WISE. Recently, the
Herschel
Space Observatorywas added to this list and has been the main drive for this in-vestigation. Ground–based observations from the Sloan DigitalSky Survey (SDSS, York et al. 2000) complete our panchromaticdataset. A detailed account on the data treatment, including un-certainty estimates, for each of the observations is given in Ap-pendix A.2.
Far-Infrared and submillimetre observations with the
Herschel
Space Observatory catch the peak in emission of the di ff use in-terstellar dust. This component plays an essential role in theenergy balance of the SED. Andromeda was observed with bothPACS and SPIRE instruments in parallel mode. Because of thelarge extent of the galaxy, observations were split into two fields.Both fields were combined during data reduction, resulting in ∼ . ◦ × . ◦ maps at 100, 160, 250, 350, and 500 µ m. In theoverlapping area of the two fields, the signal-to-noise ratio isslightly higher. The full width half maximum (FWHM) of pointsources in the final PACS maps are 12 . (cid:48)(cid:48) and 13 . (cid:48)(cid:48) at 100 µ mand 160 µ m, respectively (Lutz 2010). The resulting SPIREmaps are characterised by beams with a FWHM of 18 . (cid:48)(cid:48) , 24 . (cid:48)(cid:48) ,and 36 . (cid:48)(cid:48) at 250, 350, and 500 µ m (Herschel Space Observa-tory 2011). Galactic dusty structures tend to cause foregroundemission when observing nearby galaxies. The north-east partof the M31 disk clearly su ff ers from this kind of cirrus emission.Following a technique devised by Davies et al. (2010), Galacticcirrus emission was disentangled from the light of the M31 disk. A detailed description of the data reduction process, includingcirrus removal, can be found in paper I.The Multiband Imaging Photometre of Spitzer (MIPS; Riekeet al. 2004) observed the mid-infrared and far-infrared light ofM31 in its three bands (24, 70, and 160 µ m). Gordon et al.(2006) made a complete data reduction of the observations, cov-ering a 1 ◦ × ◦ area along the major axis of the galaxy. Theimages have standard MIPS FWHM values of 6 . (cid:48)(cid:48) , 18 . (cid:48)(cid:48) , and38 . (cid:48)(cid:48) at 24, 70, and 160 µ m, respectively (Rieke et al. 2004).Both MIPS and PACS cover a wavelength range around 160 µ m.While this could be used to more accurately estimate the uncer-tainties at this wavelength, it limits our working resolution. BothMIPS and PACS measurements come with a total uncertainty of ∼
10% in their 160 µ m band so they can be considered equallysensitive. We therefore opted to omit the MIPS 160 µ m imagefrom our sample.The same area of M31 was also mapped in all four bandsof the Spitzer
Infrared Array Camera (IRAC; Fazio et al. 2004).The complete data reduction, including background subtraction,was carried out by Barmby et al. (2006). Their final, backgroundsubtracted frames have the standard FWHM values of 1 .
6, 1 . .
8, and 1 . .
6, 4 .
5, 5 .
8, and 8 µ m wavebands,respectively.Complementary to the IRAC / MIPS observations, the mid-infrared part of M31 has been observed by WISE as part of anall-sky survey at 3 .
4, 4 .
6, 12, and 22 µ m. High–quality mo-saics of M31 were provided by the WISE Nearby Galaxy Atlasteam (Jarrett et al. 2013). Recent results from these authors haveproven the possibility of enhancing the resolution of WISE usingdeconvolution techniques. Here, however, we use the mosaicswith the standard beams because we will have to degrade theresolution to the SPIRE 500 µ m beam in order to remain consis-tent. The FWHM of the WISE beams are 6 .
1, 6 .
4, 6 .
5, and 12 . .
4, 4 .
6, 12, and 22 µ m, respectively (Wright et al.2010).Several WISE and Spitzer bands lie close to each other incentral wavelength. This overlap improves our sampling of theambiguous MIR SED and will reduce the dependence of theSED fit on a single data point, which is important in the coarselysampled wavelength ranges, e.g. around 24 µ m. At the sametime, this serves as a sanity check of the measurements of bothinstruments. We found no strong outliers between WISE andSpitzer fluxes.E ff orts to observe M31 in the NIR bands include the 2MASSsurvey (Skrutskie et al. 2006; Beaton et al. 2007) and the ongo-ing ANDROIDS project (Sick et al. 2013), all of them coveringthe J , H , and K bands. The main di ffi culty of NIR imaging is thebrightness of the sky. At these wavelengths the brightness canvary significantly between pointings, making it extremely hardto produce a large–scale mosaic with a uniform background. Tomeet the goals of our paper, it is important to have a reliableand consistent absolute flux calibration over the entire disk ofM31. No JHK bands were included in our dataset for this rea-son. The NIR part of the SED is, however, su ffi ciently coveredby the WISE, IRAC, and SDSS i and z bands. The Sloan Digital Sky Survey mapped M31 at superb resolution(FWHM ∼ . u , g , r , i , and z filters. Back-ground estimation for these observations proved di ffi cult becauseof the great extent of the galaxy and the narrow field of viewof the telescope. Tempel et al. (2011) created detailed mosaicsfrom the separate SDSS tiles, taking special care of background Article number, page 3 of 22 ubtraction and flux preservation. The resulting frames span astunning 2 . ◦ × ◦ field with a pixel scale of 3 .
96 arcseconds.The mosaics are contaminated by several artefacts around thebrightest sources, especially in the u and z bands. They are mostlikely ghost projections as they are slightly smaller and appearon each of the four sides of brightest sources along the pixel grid.After masking (see Sect. A.1.2), the images proved su ffi cientlyreliable for SED fitting at SPIRE resolutions.Unattenuated ultraviolet photons are the main tracers of veryrecent star formation. Most of the emitted UV light, however,is heavily attenuated by interstellar and circumstellar dust andconsequently important to constrain the dust distribution in ourspatially resolved SED. Thilker et al. (2005) created images us-ing separate observations from the Galaxy Evolution Explorer(GALEX) in both near-UV (NUV) and far-UV (FUV) filters.The number of frames has recently been expanded to 80, almostfully covering a 5 ◦ × ◦ field around the centre of M31. Their mo-saics have FWHM values around 5 (cid:48)(cid:48) . Because of the co-addingof separate tiles, background variations were visible at the edgesof each tile. Additionally, the UV sky around M31 is cloudedwith scattered light from Galactic cirrus structures. Both fea-tures will be taken into account as background variations in theuncertainty estimation for the fluxes.
3. Method and results
The data was processed in several steps to create a homogeneousset. We give a brief description here. For a complete account,we refer the reader to appendix A.1.First, the background was subtracted from the images. Thisproved necessary for the GALEX, WISE, and
Herschel frames.The average background level was already zero for the SDSS andSpitzer subsets, hence no further background subtraction wasneeded. Second, foreground stars and background galaxies weremasked and replaced by the local background. The GALEXand SDSS frames were masked using UV colours, while MIRcolours were used to mask the WISE, IRAC, and MIPS images.As a third step, all frames were convolved to the resolution of theSPIRE 500 µ m point spread function (PSF) using the convolu-tion kernels from Aniano et al. (2011). Finally, the pixel scaleswere resized to match the pixel grid of this frame, which wasrebinned to a 36 arcsec / pixel scale.A detailed uncertainty analysis was performed for each pixelas well. Therefore, we did not start from the original errors re-lating to the observations and data reduction, but they were es-timated directly from the convolved and rescaled images. Threesources of uncertainty were considered: background variationsin the frame, calibration uncertainties, and Poisson noise due toincoming photons. The last was only considered in the UV andoptical bands, where they are known to be significant.The above procedures yield a panchromatic SED for thou-sands of pixels, each corresponding to a sub-kpc region in An-dromeda. A complete UV-to-submm spectral energy distributionwill be fitted to each of these regions to investigate their under-lying properties. We make use of the Bayesian SED fitting code MAGPHYS (daCunha et al. 2008) to perform the strenuous task of modellingpanchromatic SEDs. The program determines the best fit froma library of optical and infrared SEDs, taking special care ofthe dust-energy balance when combining the optical and infrared part of the spectrum. This library is derived from one generalmulti-component galaxy-SED model, characterised by a numberof parameters.The stellar emission is computed by assuming a Chabrier(2003) initial mass function (IMF) and evolved in time usingthe latest version of the stellar population synthesis (SPS) modelof Bruzual & Charlot (2003). The obscuring e ff ects of interstel-lar and circumstellar dust are computed using the Charlot & Fall(2000) model.A multi-component dust model is used to calculate the in-frared and submm emission from the reprocessed starlight. Themodel consists of five modified black bodies, three of which havefixed temperatures (850 K, 250 K, and 130 K) representing thehot dust. The other two have variable temperatures and embodythe warm and cold dust components in thermal equilibrium. ThePAHs are modelled using a fixed template based on observationsof the starforming region M17. Although MAGPHYS keeps theemissivity index of the modified black body, β , fixed at 2 for thecoldest dust component, this is partially compensated by addingmultiple dust components at multiple temperatures, broadeningthe FIR-submm peak. The total amount of dust is distributed intwo di ff erent geometries: (1) the di ff use dust in the ISM, whichconsists of all ingredients of the aforementioned dust model, and(2) the circumstellar dust, which resides in the birth clouds ofnew stars and consists of all ingredients except for the cold dustcomponent.The library of template SEDs is derived from this multi-parameter model for the FUV–submm SED. Each free param-eter comes with a physically motivated probability distribution.From these distributions, a random parameter set is drawn to cre-ate a template SED. The standard MAGPHYS library consists of25000 UV–optical templates and 50000 IR–submm templates.When modelling the observed SED of a galaxy, maximum like-lihood distributions are created for each of the free parametersin the model. This is done by weighing the parameters of eachtemplate fit with its respective χ value. The di ff erent outputparameters are summarised in Table 1 and are briefly discussedbelow. – The contribution of the dust component in the di ff use ISM tothe total infrared luminosity f µ = L ISMdust / L Totdust is derived fromboth the absorption of starlight and from infrared emission. – The total stellar mass M ∗ is derived from the population syn-thesis models and is proportional to the flux in the UV toNIR wavebands. – The standard star formation rate (SFR) expresses the num-ber of stars formed per year, averaged over the last 100 Myr.The process of star formation is modelled with an exponen-tially declining SFR law starting from the birth of the galaxy.Superimposed are bursts with a random chance of occurringthroughout its lifetime. – The specific star formation rate (sSFR) is then simply the ra-tio of the SFR and the stellar mass and compares the numberof stars formed during the last 100 Myr with the total numberof stars formed throughout the lifetime of the galaxy. – Dust attenuation is expressed by the optical depth parameter,which is evaluated in the V band: τ V = τ BC V + τ ISM V . Starlightof young stars in their birth clouds (BC) experiences extinc-tion from circumstellar dust and from the di ff use interstellardust. This is parametrised by τ BC V . Most of the stars however,only irradiate the interstellar dust, modelled by τ ISM V . – The bulk of the dust mass is contributed by warm and colddust in the di ff use ISM and by warm circumstellar dust. A Article number, page 4 of 22. Viaene: Dust scaling relations in M31
Table 1:
Overview of the output parameters from a MAGPHYS SEDfit.
Symbol Unit Description f µ ISM dust to total dust luminosity τ V Total V band optical depth τ ISM V ISM dust contribution to τ V SFR M (cid:12) yr − Star formation ratesSFR yr − Specific star formation rate M ∗ M (cid:12) Total stellar mass L dust L (cid:12) Total luminosity of emitting dust M dust M (cid:12) Total dust mass T BCW K Dust temperature in birth clouds T ISMC K Dust temperature in ISM L TotC L (cid:12) Total cold dust luminosity L TotPAH L (cid:12) Total PAH luminosityfactor of 1 . M dust = . M BCW + M ISMW + M ISMC ) . (1) – The equilibrium temperature for the warm circumstellar dustand cold ISM dust is left free for the FIR / submm modifiedblack–body components. They are represented in T BCW and T ISMC , respectively. – Each dust component produces infrared emission, which issummed in the total dust luminosity L dust . The relative con-tributions of the dust components are quantified in fractionsto the total BC or ISM dust luminosity. We refer the readerto Sect. 2.2.1 of da Cunha et al. (2008) for a detailed expla-nation of the infrared emission parameters. Two importantcomponents will be discussed in this work: L TotC , the total lu-minosity of the cold dust in the di ff use ISM and L TotPAH , thetotal luminosity from PAHs in the ISM and around youngstars.The MAGPHYS SED libraries are derived from realistic,galaxy scale parameter values. The parameter space is thus op-timised for objects that are orders of magnitude brighter thanthe sub-kpc regions to be modelled here. Pixel-by-pixel fittingmakes no sense when the physical properties of a single pixel-region are out of the bounds of the MAGPHYS standard param-eter space. We therefore adopted a flux scaling of 10 to obtainfluxes of the order of integrated nearby galaxies and feed thesehigher fluxes to the code for fitting. Most of the output parame-ters will remain una ff ected because of their relative nature. Onlyfour parameters scale with flux and do that linearly: M ∗ , L dust , M dust and the SFR. These parameters were scaled back by thesame factor to obtain their true fitted value.Another limitation of the standard version of MAGPHYS isthe range of cold dust temperatures, which is fixed between 15 Kand 25 K. The boundaries of this interval are encountered inlow (high) FIR surface brightness areas (see e.g. paper II). Thiscauses the peak of the modified black body to be o ff set with re-spect to the observations and influences related parameters suchas star formation and dust mass. We estimate that over 60% ofthe derived temperatures for cold dust lie outside of the standard
10 15 20 25 30 T ISMC . . . . . . . f
30 35 40 45 50 55 60 65 70 T BCW . . . . . . . . . f
10 15 20 25 30 T ISMC . . . . . . . . . f
30 35 40 45 50 55 60 65 70 T BCW . . . . . . . . . . f Fig. 1:
Distribution of the dust temperatures for the individual pixelfits.
Top row ; are the results from the standard MAGPHYS version.
Bottom row ; are the results from the modified version, using broadertemperature ranges for the priors. The red lines indicate the averagesample values. −
25 K interval (see upper panels of Fig. 1). The same is truefor the temperature ranges of the warm dust. Here about 15% ofall regions are estimated to lie outside the 30 −
60 K range. Inorder to execute reliable fits, it is thus mandatory to expand thetemperature intervals and create a custom infrared library.A custom set of infrared SEDs was constructed (da Cunha,private communication), incorporating a wider range in cold andwarm dust temperatures. The new library features cold dust tem-peratures T ISMC ranging from 10 −
30 K and warm dust temper-atures T BCW from 30 −
70 K. With this extended library, the de-rived cold dust temperatures are more spread out over the param-eter space, also populating the coldest ( <
15 K) regions (see thelower-left panel of Fig. 1). The distribution of the warm dusttemperature is also considerably changed (lower–right panel),although the parameter space was only increased by 10 K. Thepeak is also not shifted towards higher temperatures as the dis-tribution derived from the standard library would suggest. Whatwe see here is a manifestation of the shift in cold dust tempera-ture. As both dust components are not independent - the infraredSED is fitted entirely at the same time - the shift to lower colddust temperatures will also cause a decrease in the warm dusttemperature distribution in order to still match the flux in theiroverlapping area (FIR).
We perform panchromatic fits for a total of 22437 pixels withinan ellipse with major axis of 22 kpc and an apparent eccentricityof 0 .
96, covering 2 .
24 square degrees on the sky. The pixelsare the same size as the FWHM of the 500 µ m beam, makingthem statistically independent from each other. The choice ofour aperture is limited to the field of view of the IRAC frames,which cover the main stellar disk of M31, but does not extend Article number, page 5 of 22 p to NGC 205 or to the faint outer dust structures as seen in theSPIRE maps. However, the field is large enough to cover over95% of the total dust emission of the galaxy (Draine et al. 2014).
Instead of eliminating a priori those pixels with a non-optimalspectral coverage, we decided to exploit one of the characteris-tics of MAGPHYS, that is that the final results, i.e. the physi-cal parameters we are looking for, are given as the peak valuesof a probability distribution function (PDF). When a pixel hasa SED which is characterised by a poor spectral sampling, theparameters that are more influenced by the missing data points,whatever they are, will have a flatter PDF, showing the tendencyto assume unrealistic values. For example, output parameters re-lated to the stellar components (stellar mass, SFR, etc.) will bequestionable for pixels in regions of Andromeda where the UVand optical background variations become dominant. Parame-ters related to interstellar or circumstellar dust turn unreliablewhen reaching very low flux density areas in the
Herschel bands.To decide which pixels are to be considered reliable for agiven parameter and which ones should be not considered, weevaluate the mean relative error of each fit. Each PDF comeswith a median value (50th percentile), a lower limit (16th per-centile), and an upper limit (84th percentile). A way to quantifythe uncertainty of the median value is to look at the shape ofthe PDF. Broadly speaking, if the peak is narrow, the di ff erencebetween the 84th and 16th percentile will be small and the cor-responding parameter will be well constrained. For a broad peakor a flat distribution, the opposite is true. We define this meanrelative error as follows: σ rel = . · ( p − p ) / p , (2)where the p x indicate the percentile levels of the PDF. This erroronly reflects the uncertainty on the modelling.A double criterion was needed to filter out unreliable esti-mates for each of the parameters considered in Table 1, becausethe average σ rel is quite di ff erent in each parameter. Further-more, it proved necessary to filter out most of the parameter es-timates related to the outermost pixels in our aperture. These re-gions have FIR emission below the Herschel detection limits. Areliable detection at these wavelengths is crucial in constrainingmost of the dust-related parameters. Pixels with a non-detectionat either PACS 100 µ m or PACS 160 µ m make up almost 40%of the sample. At the same time, these pixels generally havehigher photometric uncertainties at shorter wavelengths as theycorrespond to the faint outskirts of the galaxy. Together withtheir poorly sampled FIR SEDs, their corresponding parameterestimates will have broad PDFs. We rank, for each parameter,all pixels according to increasing relative error for that particu-lar parameter. Then, 40% of the parameter estimates (those withthe highest mean relative errors) were removed from the sample.This corresponds to the exclusion of 8975 pixels per parameter.Secondly, as several parameter estimates with broad PDFswere still present after the first filtering, an optimal cut was foundwhich excludes these estimates. We chose to remove any param-eter estimate with σ rel > .
32. The combination of these filtersexcluded, for each parameter, all pixels with an unreliable esti-mate of this parameter. We note that the excluded pixels them-selves might di ff er for each parameter set. For example, the dustmass of a particular pixel might be poorly constrained and thusremoved. On the other hand, the stellar mass of that same pixel,will be kept in the sample if it meets our filter criteria. Several − . − . − . − . . MAGPHYS (This work) − . − . − . − . . M od BB ( P ape r II ) log(Σ M dust ) (M (cid:12) pc − )
10 15 20 25 30
MAGPHYS (This work) M od BB ( P ape r II ) T ISMC (K) − . − . − . − . − . − . − . MAGPHYS (This work) − . − . − . − . − . − . − . F U V + µ m ( P ape r III ) log(Σ SFR ) (M (cid:12) yr − kpc − ) χ f Fig. 2:
Upper–left: χ distribution of the fits. Other panels: densityplots comparing dust mass surface density, cold dust temperature, andstar formation rate derived from MAGPHYS single pixel fits againstmodified black–body fits to Herschel bands (paper II) and FUV + µ mSFR tracers (paper III). Red indicates a small number of data points,yellow a large number. The black line represents the 1 : 1 relation. pixels (4384 in total) did not meet the requirements in any of theparameters and were thus completely removed from the sample(meaning they were not considered in the χ distribution or inany further analysis). The distribution of the best–fit χ values isshown in the upper–left panel of Fig. 2 and has an average valueof 1 . T ISM C and L dust are the best constrained parameters inthe sample, with median uncertainties of 5% and 6%, respec-tively, and 13462 usable pixels. On the other hand, τ V , T BC W ,and sSFR are the least constrained parameters. However, in thecase of sSFR, this quantity is computed from the SFR and stellarmass, the uncertainty on this parameter takes into account theuncertainty on both of its constituents, hence the relatively highmedian σ rel of 0.18 for 6608 usable pixels. Quite di ff erently,the high mean σ rel (0.21 for 13462 usable pixels) on estimatesfor T BC W stems from the poor spectral coverage in the 30 − µ mregime. Other degeneracies in the fitting procedure will certainlyreflect in the total number of reliable pixels per parameter, aswell as in their median relative error. This is most obvious for τ V , with a median σ rel of 0.14 and only 3753 usable pixels. Thisparameter is not only influenced by the amount of dust, but alsoby its geometry with respect to the stars. This is impossible totake into account without complex radiative transfer modellingand falls beyond the goal of this paper. We compare our results with previously derived values for eachpixel, see Fig. 2. The MAGPHYS dust mass and cold dust tem-perature are compared to the modified black–body fits from pa-
Article number, page 6 of 22. Viaene: Dust scaling relations in M31 per II. Furthermore, our star formation rate is compared to theSFR derived from FUV + µ m fluxes in paper III. Each pixelregion of the paper II and paper III maps corresponds exactly toa pixel region in our sample, hence we are comparing parameterestimates for the exact same physical region.In general, the di ff erent approaches yield consistent results.The dust mass shows the tightest relation ( rms = . ff set ∆ log( Σ M dust ) = .
22. It is therefore important tounderstand what we are comparing here. For each dust compo-nent, the flux S ν is modelled with a modified black–body func-tion, S ν = M dust D κ abs B ν ( T dust ) , (3)where D is the distance to the galaxy, B ν ( T dust ) the Planck func-tion, and κ abs is the dust mass absorption coe ffi cient, modelled as κ abs = κ abs ( λ ) × (cid:18) λ λ (cid:19) β (4)with λ the normalisation wavelength and β the emissivity in-dex. MAGPHYS adopts the dust model from Dunne et al. (2000)(hereafter D00) who normalize the dust mass absorption coe ffi -cient at 850 µ m: κ = .
077 m kg − . At Herschel wavelengthsthis becomes κ = .
454 m kg − assuming a fixed β =
2. In pa-per II, the Draine (2003) (hereafter D03) absorption coe ffi cient,which is κ = .
192 m kg − , and a variable β were adopted.Lower κ abs ( λ ) values are associated with more silicate-rich dustcompositions (Karczewski et al. 2013). They will result in higherdust mass estimates, which is the case for paper II. On the otherhand, a variable β will generally result in higher dust tempera-tures. This will in turn yield lower dust masses. As MAGPHYScomputes the total dust mass from components of various tem-peratures, grain sizes, and compositions, this should result inmore realistic dust mass estimates.A smaller o ff set ( ∆ T d = .
47 K), but larger scatter ( rms = .
93) is seen in the temperature of the cold dust. The emissivityindex β was fixed at 2 in this paper, while left as a free parameterin paper II. As there is a known degeneracy between β and T dust (see e.g. Hughes et al. 2014; Tabatabaei et al. 2014; Galametzet al. 2012; Smith et al. 2012b, and references therein), this couldexplain the scatter in the relation. Furthermore, most of the pix-els in paper II had β <
2. Given this temperature- β degeneracy,smaller β values will yield higher dust temperatures. This prob-ably explains the systematic o ff set from our sample.The SFR shows a less clear deviation from the 1:1-relation.The rms of the scatter in the points is 0 .
28 and they have an o ff setof ∆ log( Σ S FR ) = − .
42. We tend to find systematically lowerSFR values compared to the FUV + µ m tracer used in paperIII. It must be noted that the SFR is derived in a di ff erent wayin both approaches. The FUV + µ m tracer is empirically de-rived from a sample of starforming galaxies (Leroy et al. 2008).Several regions in M31 exhibit only low star forming activity,far from the rates of starforming galaxies. Additionally, this for-malism assumes a stationary star formation rate over timescalesof 100 Myr. In M31, we resolve sub-kpc structures, where starformation may vary on timescales of a few Myr (Boselli et al.2009). MAGPHYS does allow variations in SFR down to starformation timescales of 1 Myr.Most of the outliers in the SFR plot of Fig. 2 correspond topixels surrounding the bulge of M31. It is in these areas that theMIR emission from old stars is modelled quite di ff erently. Over-all, we expect that our SED fits give a more realistic estimate ofthe SFR because they take information from the full spectrumand are derived from local star formation histories. One of the most straightforward checks we can perform betweenour modelling technique and other techniques, is a comparisonbetween the total number of stars and dust in M31. With respectto the first component, we find a value of log( M (cid:63) / M (cid:12) ) = . / L ratio are consistent withmore heavyweight IMFs. Tamm et al. (2012) did exploit SSPmodels to calculate the stellar mass, and found values in thelog( M (cid:63) / M (cid:12) ) = − .
48 range, using the same IMF as wedo. This discrepancy might arise from a number of possiblecauses: first of all, the aperture used to extract the total fluxesis slightly smaller in our case and secondly, a di ff erent maskingroutine was applied to remove the foreground stars. The likelymost e ff ective di ff erence, however, could be due to the fact thatMAGPHYS considers an exponentially declining star formationhistory to which star formation bursts are added at di ff erent agesand with di ff erent intensities. This can cause the M / L to de-crease, and might explain the di ff erence in stellar mass.Instead, when we compare the stellar mass in the inner 1 kpcwith the value calculated with MAGPHYS by Groves et al.(2012), we find a remarkably good agreement (log( M (cid:63) / M (cid:12) ) = .
01 vs log( M (cid:63) / M (cid:12) ) = .
91 in our case).The dwarf elliptical companion of Andromeda, M32, alsofalls in our field of view. We find a total stellar mass oflog( M (cid:63) / M (cid:12) ) = .
77. Of course, as the light of M32 is highlycontaminated by Andromeda itself, mass estimates of this galaxyare highly dependent on the aperture and on the estimation ofthe background flux of M31. Nevertheless, our estimate is onlya factor of 2–3 lower then dynamical mass estimates (e.g. Rich-stone & Sargent 1972). This discrepancy is of the same order asthe di ff erence in total mass we find for M31.As a total dust mass, we find M dust = (2 . + . − . ) · M (cid:12) ,for the sum of all pixel-derived dust masses, using the D00 dustmodel. This estimate is preferred over the dust mass from a fit tothe integrated fluxes as the most recent dust mass estimates forM31 are derived from the sum of pixel masses. Furthermore, it isknown that dust mass estimates from integrated fluxes underes-timate the total dust mass (see also Sect. 3.2.4). We note that theuncertainly on this estimate appears very small. This is becauseof the very narrow PDF for M dust from the global fit of M31. Aspreviously stated, this error only reflects the uncertainty on theSED fitting. The absorption coe ffi cient κ abs ( λ ) from D03 wasused in the next estimates.In paper I we estimated the dust mass from a modified black-body fit to the global flux and found M dust = (5 . ± . · M (cid:12) .Paper II derived a total dust mass from modified black-body fitsto high signal-to-noise pixels (which cover about half of the areaconsidered) and found M dust = . · M (cid:12) . Independent Her-schel observations of M31 (Krause in prep.; Draine et al. 2014)yield M dust = (6 . ± . · M (cid:12) (corrected to the distanceadopted in this paper: D M = .
785 Mpc) as the sum of thedust masses of each pixel.We find a total dust mass of the same order as these previousestimations; however, our result is somewhat lower. The reasonfor this discrepancy is most likely the di ff erence in dust model,as was already clear from Fig. 2. Determining the conversionfactor q between dust models is rather di ffi cult and requires the Article number, page 7 of 22 ssumption of an average emissivity index: q = κ D κ D = κ D κ D · (cid:32) (cid:33) − β . (5)For M31, the mean β was found to be in the range 1 . − . . .
6, or dust masses in the range 5 . − . · M (cid:12) . The D00 dust model thus tends to produce dustmasses that are about half of the D03 masses. Keeping this inmind, all dust mass estimates for Andromeda agree within theiruncertainty ranges. The MAGPHYS code was conceived for galaxy-scale SED fit-ting and it does a good job there (e.g. da Cunha et al. 2008, 2010;Clemens et al. 2013). At these large scales, a forced energy bal-ance is justified because globally, most of the absorbed starlightis re-emitted by dust. When zooming in to sub-kpc regions, thisassumption might not be valid any more. Light of neighbouringregions might be a significant influence on the thermal equilib-rium of a star forming cloud, and if this is true, it may translateinto an o ff set between the local parameters and the global valuefor that galaxy.As a test for our extended library (see Sect. 3.1) we comparethe mean values of the physical parameters to their global coun-terparts (upper part of Table 2). These parameters were derivedfrom a MAGPHYS fit to the integrated fluxes of Andromeda.The fluxes are listed in Table B.1. In the case of additive param-eters (bottom part of Table 2), we compare their sum to the valuederived from a global SED fit. It is important to note that we relyon our filtered set of pixels for each parameter. This means atleast 40% of the pixels are excluded. Most of these badly con-strained pixel values lie in the outskirts of M31. Nevertheless,their exclusion will surely a ff ect the additive parameters.Several parameters mimic their global counterparts quitewell. This is the case for f µ , T ISMC , τ V , and τ ISM V , where theagreement lies within 1 standard deviation. The distribution ofthe pixel-derived sSFRs has a broad shape. In order to comparethe sSFR of the pixels to the global value, we make use of thetotal SFR and stellar mass as derived from the pixels. We thenfind the pixel-sSFR by dividing the total SFR by the total M (cid:63) .This value again lies close to its global counterpart, despite thewide range of sSFRs found on an individual pixel basis.The temperature of the warm circumstellar dust di ff ers sig-nificantly: the average pixel value is 43 K with a standard devia-tion of 7 K, while a fit to the global fluxes reveals T BCW = + − K.Both values still overlap at a 2 σ level, but the relative deviationis much larger than the other parameters. The peak of this dustcomponent lies between 30 µ m and 70 µ m. The MIPS 70 µ mband provides the only data point in this region, making it di ffi -cult to estimate this parameter accurately.Additive parameters will also su ff er because we exclude asignificant number of pixels. Most of them do add up to the sameorder of magnitude as the global values: SFR, M ∗ , L dust , L totPAH ,and L totC . All of them lie within 5 −
20% below the global value.This again indicates that the excluded pixels do not contributesignificantly to the light of M31. The dust mass, however, turnsout to be ∼
7% higher when summing all pixels. It is knownthat SED fitting is not a linear procedure and will depend on theemployed resolution. Aniano et al. (2012) found that modellingon global fluxes yields dust masses that are up to 20% lower thanresolved estimates. Galliano et al. (2011) found discrepancies of up to 50% depending on the applied resolution. Even takinginto account another 10% in flux due to the excluded pixels, ourdi ff erence in dust mass estimates lies within this range.The fact that we reproduce the global properties of M31 fromour local SEDs, boosts confidence that the procedure appliedhere is valid, even though MAGPHYS was conceived for galaxy-scale SED fitting. We construct detailed maps of the SED parameters from the col-lection of pixels with reliable parameter fits in Fig. 3 and brieflydiscuss the morphologies observed.The dust luminosity L dust closely follows the morphologyseen in the PACS images (see also Fig. A.3). Dust emissionin Andromeda is brightest in the bulge and in the 10 kpc ring.Some fainter emission regions are seen in the outer parts of thegalaxy, coinciding with a ring at 15 kpc.As expected, the dust mass M dust map closely resembles theSPIRE images (see also Fig. A.3). Compared to the L dust map,some intriguing distinctions can be noted. There seems to be al-most no dust in the centre of M31, while the bulge is actuallythe brightest in dust luminosity. We hereby confirm earlier state-ments (paper II; Tempel et al. 2010; Groves et al. 2012; Draineet al. 2014) that the bulge of Andromeda holds a small amountof relatively warm ( >
25 K) dust. The south-west side is alsosmoother than in the L dust map, pointing out that the heating ofISM dust and not the mass is crucial to the observed luminosity.The PAH luminosity L totPAH appears relatively weak whencompared to the L dust map. The general morphology, however,is similar. The surface brightness at these wavelengths is thehighest in the bulge of Andromeda. Furthermore, the emissionis mostly concentrated in the 10 kpc ring and in the dusty partsof the inner disk. If reprocessed UV light of recent and ongoingstar formation is the only energy source for MIR emission, nobright MIR and PAH features are expected in the bulge of M31.Emission from PAHs can, however, be enhanced by increases inthe di ff use ISRF (Bendo et al. 2008). This again, suggests thatthe radiation field of older stars in the centre of M31 is quitestrong.A similar morphology is seen when looking at the contribu-tion of the di ff use cold dust to the total dust emission, L totC . Thecold dust, only found in the di ff use ISM, appears significantlymore luminous than the PAH emission (the L dust , L totC and L totPAH maps in Fig. 3 have the same scale). Interestingly, the emissionfrom this component is equally bright in the bulge and in thering, in contrast with the PAH luminosity map.Consequently, the temperature of the ISM dust T ISMC peaksin the centre ( ∼
30 K). It follows a smooth radial decline until itreaches a plateau at 16 K in the ring. Higher values are reachedin the brightest star forming regions. This suggests the cold ISMdust is partially heated by recent and ongoing star formation.On the other hand, older stars can also contribute to the heatingof the dust. In the NIR wavebands, the surface brightness isslightly enhanced in the ring, indicating a higher concentrationof older stars. Outside the star forming ring, the temperaturequickly drops two degrees to 14 K.For the warm dust temperature T BCW , the picture is far lessclear. The map is crowded with blanked pixels due to their highuncertainties. As already mentioned in Sect. 3.2.4, the MIPS70 µ m data point is the only observation in this temperatureregime. Additionally, the emission of the cold dust componentoverlaps greatly with the SED of the warm dust, making it dif-ficult to disentangle both components. We do find significantly Article number, page 8 of 22. Viaene: Dust scaling relations in M31
Table 2:
Comparison of the main properties for M31 as derived from the pixel-by-pixel fitting and from a fit to the global fluxes. N pix denotes thenumber of reliable pixels used for the analysis and σ rel the median relative uncertainty for this sample. The mean value of the relative parametersare given in the upper part of the table, along with the standard deviation on their distribution. The sum of the additive parameters are given in thebottom part of the table, along with their uncertainty. parameter N pix Median σ rel Meanlocal std.dev Global Unit f µ .
85 0 .
08 0 . + . − . – sS FR .
45 0 .
02 3 . ± .
01 10 − yr − T BCW + − KT ISMC . . . + . − . K τ V .
62 0 .
68 0 . ± .
01 – τ ISM V .
20 0 .
13 0 . ± .
01 – χ .
26 – 1 .
35 –parameter N pix
Median σ rel Total local Global Unit M (cid:63) . ± .
02 5 . ± .
01 10 M (cid:12) M dust . ± .
06 2 . + . − . M (cid:12) L dust . ± .
002 4 . + . − . L (cid:12) S FR . ± . . + . − . M (cid:12) yr − L totPAH . ± . . + . − . L (cid:12) L totC . ± .
002 2 . + . − . L (cid:12) higher temperatures in the bulge, where the ISRF is highest, andin the 10 kpc ring, where most of the new stars are being formed.Outside these areas, the warm dust is relatively cold ( <
45 K).The stellar mass M (cid:63) is one of the best constrained parame-ters thanks to the good coverage of the optical and infrared SED.It is highest in the bulge and the central regions around it anddeclines smoothly towards the outskirts of the galaxy. Interest-ingly, a small peak is seen where M32 resides. We do not detectthis dwarf satellite in our Herschel maps. This is caused by theoverlap of emission from M32 and M31’s di ff use dust emissionat this location on the sky. Nevertheless, it is evident that M32does not contain much dust.The star formation rate map of M31 largely coincides withthe dust luminosity (except in the bulge), although the distri-bution is more peaked in the rings. Regions where the dustemission is lower (inter-ring regions and outskirts) are mostlyblanked out because it is hard to constrain very low star forma-tion rates. Some residual star formation is seen in the bulge.However it must be noted that the high dust luminosities in thecentral region might cause a degeneracy between the SFR, whichdirectly heats the dust, and M (cid:63) , representing the number of olderstars that have been proven to strongly contribute to dust heating.The specific star formation rate is obtained by dividing theSFR over the last 100 Myr by the stellar mass and gives a mea-sure of ongoing vs past star formation. This quantity combinesthe uncertainties of both parameters, hence the large number ofblanked pixels. The 10 kpc and 15 kpc rings have the highestsSFR in M31. Interestingly, the inner ring has very low valuesof sSFR and the bulge has close to zero. In general, we can saythat stars are nowadays formed most e ffi ciently in the rings ofthe galaxy.The V –band optical depth is the poorest constrained param-eter of the sample. Accurately estimating this value requires de-tailed knowledge on the dust geometry. This is not available here, so assumptions must be made based on colour criteria andan extinction law. As we are not able to probe individual starformation regions, the optical depth must be seen as an averageover each resolution element. Liu et al. (2013) showed that whenaveraged over scales of ∼ −
200 pc, the dust geometry canbe approximated by a foreground dust screen. Individual starsor star forming regions are, however, likely to experience muchgreater optical depths than the averaged values. In Andromeda,this average varies from 0 . τ ISM V . This parameter closely re-sembles the ring–like structure we also see in the M dust andranges from 0 . − .
1. The ratio of those two parameters τ ISM V /τ V = µ , measures the contribution of di ff use cold ISM dustto the extinction of starlight. We find a median µ = (54 ± ∼ ff use dust is the main contributor to starlight extinction in themore active regions of a galaxy. This is consistent with the re-sults of Keel et al. (2014), who find that a greater fraction ofthe UV extinction is caused by the di ff use dust component. Fur-thermore, detailed radiative transfer models of galaxies show theimportance of this di ff use component in the FIR / submm emis-sion, and thus absorption of starlight (see e.g. Tu ff s et al. 2004;Bianchi 2008; Popescu et al. 2011).The contribution of the ISM dust to the dust luminosity f µ peaks ( ∼ ff use ISM. It linearly decreases along the disk, reaching ∼ Article number, page 9 of 22 . − − . . . − . − . . . − . − − . . . LINEAR − . − . . . − . − . . . − . − . . . − . − . . . − . − . . . × × × ×
12 14 16 18 20 22 24 × × − × − . . . . . . . . .
00 0 .
08 0 .
16 0 .
24 0 .
32 0 .
40 0 .
48 0 . .
40 0 .
48 0 .
56 0 .
64 0 .
72 0 .
80 0 .
88 0 . Offset from centre (degrees) O ff s e tf r o m c en t r e ( deg r ee s ) L dust L (cid:12) M dust M (cid:12) L TotPAH L (cid:12) L TotC L (cid:12) T ISMC
K M ? M (cid:12) sSFR yr − SFR M (cid:12) yr − τ V τ ISMV T BCW
K f µ Fig. 3:
Parameter maps for M31, rotated from a position angle of 38 ◦ . See Table 1 for the meaning of the parameters. Pixels with uncertainties thatwere considered too large (see Sect. 3.2.1) were blanked out. The green ellipses in the upper–left panel represent the apertures of the macro-regionsof M31: the bulge, inner disk region, 10 kpc ring, and the outer disk. As a first step towards a spatially resolved analysis, we decom-pose Andromeda into macro–regions, located at di ff erent galac-tocentric distances: the bulge, the inner disk, the star forming ring centred at a radius of 10 kpc, and the outer part of the disk.We choose to base our definition of these regions on the mor-phology of the L dust map of Fig 3. The advantage is that, in thelight of constructing dust scaling relations, each region corre-sponds to a separate regime in terms of SFR, radiation field, dust Article number, page 10 of 22. Viaene: Dust scaling relations in M31 content, and composition. For example, the bulge is limited tothe inner 1 kpc region and is thus significantly smaller than theoptical / NIR bulge, but it coincides with the zone with the hard-est radiation field. The shapes of the borders between regions areapparent ellipses on the sky and do not necessarily coincide withprojected circles matching the disk of M31 (see also Table 3 fortheir exact definition).We choose a set of individual pixels, selected to representthe typical shape of an SED in these regions. They are shownin Fig. 4, together with the residual values for each wavelengthband. Along with the single pixel SEDs, we also show SED fitsto the integrated fluxes of these macro–regions in Fig. 5. Inte-grated fluxes for the separate regions can be found in Table B.1.The goodness of fit is here expressed by the χ value for the bestfitting template SED. The parameter values from this SED willdi ff er slightly from the peak values in the PDFs, which expresstheir most likely values. Nevertheless, the template SED gives agood indication of how well the observed fluxes can be matched.When comparing the fits in Figs. 4 and 5, it is clear that themacro-region fits have a systematically lower χ than their re-spective single-pixel fits. This is not surprising as one mightexpect greater signal-to-noise variations on smaller scales. Ingeneral, however, most of the observed fluxes are well repro-duced by the best fit. Only the MIPS 24 µ m point seems to besystematically below the theoretical SED. The MIPS 24 µ m ob-servations do come with large error bars, so that is accounted forin the determination of the parameter PDFs.In the bulge, stars are completely dominant over the dustcomponent in terms of mass and luminosity. This is visible inthe ratio of the total dust luminosity to the g –band luminosity( L dust / L g = . ff set between the unattenu-ated (blue) and the attenuated (red) SED. The optical / NIR SEDis much more luminous than the UV part of the spectrum, in-dicating a relatively low SFR and a strong interstellar radiationfield (ISRF), dominated by older stars. Furthermore, the peak ofthe FIR SED lies at relatively short wavelengths caused by hightemperatures of the cold dust. There are only weak PAH featuresvisible in the centre of M31, although the MIR flux is relativelylarge compared to the other regions. Some residual star forma-tion can be found in the bulge of M31, but the contribution to thetotal SFR is negligible.The inner disk is forming stars at a slow pace (2 . × − M (cid:12) yr − ). This is confirmed by a visible o ff set between theunattenuated and attenuated SED in the UV regime. The FIRemission peaks at longer wavelengths than in the bulge, indi-cating a milder ISRF and lower dust temperatures. We conse-quently find a higher L dust / L g ratio of 1 .
50. This less harsh en-vironment allows PAHs to survive longer, giving rise to moreprominent MIR features. The same conditions hold in the outerdisk of M31, although the surface brightness is systematicallylower there. The dust also gains in importance here ( L dust / L g = . ff use dust is colder in the outskirts of the galaxy.The most active star forming region of M31 is unquestion-ably the ring at ∼
10 kpc. This region contains only 20% of thestellar mass and almost half of the total dust mass of M31 at tem-peratures near the galaxy’s average (see also Table 3). The lu-minosity di ff erence between the UV and the optical / NIR SED isthe smallest of all regions, indicating that new stars dominate theradiation field. This also translates in strong PAH features in theMIR. The o ff set between the attenuated and unattenuated SEDis large in the UV and even visible in the optical-NIR regime,indicating significant dust heating. Consequently, we find thehighest dust–to– g -band ratio L dust / L g = .
84 here.
4. Dust scaling relations
In the following we will investigate scaling relations of stellarand dust properties in Andromeda at di ff erent sizes: we first con-sider the galaxy as a whole, and we will then look at its maincomponents as separate regions. Finally, we will push the analy-sis down to the smallest possible scale: the hundred–pc sized re-gions defined by the statistically independent pixels whose SEDwe have modelled as explained above.Our results can then be compared to already known resultsfor similar physical quantities. In this respect, the ideal samplefor comparison is surely the one provided by the local datasetof the Herschel
Reference Survey (HRS; Boselli et al. 2010).The HRS is a volume–limited, K –band selected survey includingmore than 300 galaxies selected to cover both the whole range ofHubble types and di ff erent environments. Cortese et al. (2012)have analysed in detail how the specific dust mass correlates tothe stellar mass surface density ( µ (cid:63) ) and to NUV–r colour (seeFig. 6). Furthermore, da Cunha et al. (2010) also found linksbetween the dust mass and SFR and between f µ and the specificSFR using a sample of low-redshift star forming galaxies fromthe SDSS survey. We will check where Andromeda is locatedwith respect to the above relations and, more importantly, wewill address the issue regarding the physical scales at which theaforementioned relations start to build up. Andromeda is classified as a SA(s)b galaxy (de Vaucouleurset al. 1991) and has a prominent boxy bulge. The disk containstwo conspicuous, concentric dusty rings and two spiral arms (seee.g. paper IV; Gordon et al. 2006).In Table 3 we report a summary of the main physical proper-ties we have derived for Andromeda from integrated fluxes. Thetotal stellar mass was found to be 5 . × M (cid:12) , a typical valuein local starforming galaxies (see e.g. Clemens et al. 2013). Thetotal dust mass was found to be 2 . × M (cid:12) , comparable to theamount of dust in our own Galaxy (e.g. Sodroski et al. 1997).The distribution of dust in the Andromeda galaxy is, however,atypical for an early–type spiral. The infrared emission fromearly-type galaxies is usually quite compact (e.g. Bendo et al.2007; Muñoz-Mateos et al. 2009b). Extended ring structures,such as the ones in M31, appear rather infrequently.As an early–type spiral, Andromeda has a low star formationrate of 0 . M (cid:12) yr − , about ten times smaller than the value mea-sured in the Milky Way (Kennicutt & Evans 2012). In da Cunhaet al. (2010) a relation was derived between the SFR of normal,low redshift (z < .
22) SDSS galaxies and their dust content.Despite a total dust mass which is close to the sample average,the SFR in M31 is about one order of magnitude below the av-erage relation at this dust content. Consequently, Andromeda’sspecific star formation rate (sSFR) of 3 . × − yr − is at thelower end compared to the trend found by da Cunha et al. (2010).If we compare the locus of M31 in the scaling relation plotspresented in Cortese et al. (2012), we find that Andromeda fol-lows precisely the average trends defined by HRS galaxies (seethe cyan dot in the two top panels in Fig. 6). In general, M31has dust and stellar masses that are typical for early–type spiralgalaxies. On the other hand, the star formation activity is un-usually low. This is consistent with a more active star formationhistory, possibly related to an encounter with M32 (Block et al.2006). Article number, page 11 of 22 l og ( λ L λ / L (cid:12) ) χ = − . . . χ = l og ( λ L λ / L (cid:12) ) χ = − λ ( µm ) − . . . χ = − λ ( µm ) Fig. 4:
The panchromatic SED of four representative pixels in the bulge, inner disk, the 10 kpc ring, and the outer disk region. The blue linerepresents the unattenuated SED and the red line the best fit to the observations. Residuals are plotted below each graph.The χ values are thosefor the best fitting template SED. l og ( λ L λ / L (cid:12) ) χ = − . . . χ = l og ( λ L λ / L (cid:12) ) χ = − λ ( µm ) − . . . χ = − λ ( µm ) Fig. 5:
The panchromatic SED of four main apertures: the bulge, the inner disk region, the 10 kpc ring, and the integrated galaxy. The blue linerepresents the unattenuated SED and the red line the best fit to the observations. Residuals are plotted below each graph. The χ values are thosefor the best fitting template SED.Article number, page 12 of 22. Viaene: Dust scaling relations in M31 Table 3:
Summary of the parameters characterising the main apertures of M31. The semi-major axis a and apparent eccentricity (cid:15) describe theannuli on the plane of the sky. The other parameters are physical properties derived from an SED fit to the integrated fluxes inside these apertures. Region RA centre DEC centre a (cid:15) M (cid:63) µ (cid:63) M dust SFR sSFR NUV–rhh:mm:ss dd:mm:ss kpc 10 M (cid:12) M (cid:12) kpc − M (cid:12) − M (cid:12) yr − − yr − MagGlobal 10:39:52.6 41:14:30.1 22.0 0.96 5 .
50 3 .
11 27 . . .
38 4 . .
81 25 . .
11 0 .
18 0 .
27 6 . .
00 4 .
23 3 .
95 2 .
30 1 .
20 5 . .
96 1 .
66 12 . .
06 9 .
55 4 . .
07 1 .
58 12 . .
18 1 .
07 3 . .
06 4 .
65 0 .
08 0 .
03 0 .
60 5 . As outlined in Sect. 3.4, we have grouped our set of pixels in fourmacro-regions based on the morphology of the L dust map. Thesemacro-regions represent physically di ff erent components in thegalaxy. It is well known, for example, that the bulges in spiralgalaxies usually host the oldest stellar populations (e.g. Moor-thy & Holtzman 2006). Bulges are usually devoid of ISM andhave barely any star formation (e.g. Fisher et al. 2009), closelyresembling elliptical galaxies in many respects. However, unlikestand-alone elliptical galaxies, galactic bulges are intersected bygalactic disks. At this intersection, a significant amount of gas,dust, and many young stars are present. In this respect, An-dromeda is again an atypical early-type spiral as none of thesecomponents are prominently visible at the bulge-disk intersec-tion.In the top panels of Fig. 6, we plot the physical quantitiesderived from fitting the integrated fluxes of Andromeda’s fourmacro-regions (blue points), together with the relations for HRSgalaxies (red dots) and their average trend (black line). Quiteremarkably, they all fall on (or very close to) the average HRSrelation, each of those components lying on a specific part of theplot which is typical for a given Hubble (morphological) type.On average, the outer parts of M31 closely resemble the phys-ical properties of late–type HRS galaxies, while its bulge has,instead, characteristics similar to those of elliptical galaxies.In the remainder of the paper, we define objects as early-type when NUV–r > .
2. This is the transition where most ofthe HRS objects are either E or S0 galaxies. Consequently, wedefine objects as late-type bluewards from this line. In Fig. 6,we also indicate the bluest S0 galaxy (NUV–r = .
22) and thereddest Sa galaxy (NUV–r = .
69) of the sample to indicate thespread of the transition zone.In this respect, it is worth noting that Andromeda’s bulge issignificantly redder than any of the submm detected HRS galax-ies because we are picking, by definition, only its very central,hence redder, regions. There is a known colour gradient in el-liptical galaxies, which have bluer outskirts with respect to theirinner parts, and this di ff erence can be as high as ∼ ff set in thebulge colour with respect to the HRS elliptical galaxies, whosecolours are instead calculated from global apertures. For sim-ilar reasons, Andromeda’s bulge is found at larger stellar masssurface densities values if compared to global elliptical galaxies,where also the outer, less dense parts are included in the mea-surements.From a geometrical perspective, these regions follow a pat-tern that is determined by the average galactocentric distance:going from the centre outwards, we find the macro-regions atprogressively bluer colours or, equivalently, lower stellar masssurface density. This reflects a typical inside-out formation of − − − − − − l og ( M du s t / M ? ) M31 BulgeRing Inner diskOuter disk M31BulgeRingInner diskOuter disk log( µ ? ) [ M (cid:12) / kpc ] − − − − − − l og ( M du s t / M ? ) NUV-r
Fig. 6: Top: dust scaling relations from the
Herschel
Reference Sur-vey (Cortese et al. 2012); the specific dust mass M dust / M (cid:63) as a functionof the stellar mass surface density µ (cid:63) (left) and NUV-r colour (right).NUV-r serves as an inverse tracer of the sSFR. Red are Herschel de-tected galaxies, green are upper limits for the undetected ones. Theblue dots represent the macro-regions of M31 and the cyan dot indicatesthe position of the galaxy itself. The thick black line is the HRS meantrend. The vertical solid line indicates our division between late-typeand early-type galaxies. The dashed vertical lines indicate the bluestearly-type galaxy and the reddest late-type galaxy, respectively.
Bot-tom:
Same scaling relations, but now represented in a density plot usingsingle pixel regions from M31. Yellow points indicate a higher densityof points, red a lower density. The black line is the HRS mean trend,the green line is the M31 mean trend. the bulge-disk geometry (White & Frenk 1991; Mo et al. 1998).This is consistent with the results from Pérez et al. (2013) andGonzález Delgado et al. (2013), where they confirm an inside-out growth pattern for a sample of local starforming galaxies.Inside the disk, the situation might be more complex. An in-creasing sSFR for the macro-regions, from the inner disk to the10 kpc ring supports this scenario. However, it does drop downsignificantly in the outer disk (see Table 3).Muñoz-Mateos et al. (2007) have derived sSFR gradientsfor a sample of nearby galaxies. They found a slope that is,on average, positive and constant outwards of one scale length.When considering the very coarse sampling provided by the four
Article number, page 13 of 22 − − − − l og ( M du s t / M ? ) Average HRSBulgeRingInner diskOuter disk log( µ ? ) [ M (cid:12) / kpc ] − − − − − l og ( s S F R ) [ y r − ] Fig. 7:
Average trends of the scaling relations for the stellar mass sur-face density µ (cid:63) , separated by the main morphological regions of M31:the bulge (red), the inner disk (cyan), the 10 kpc ring (blue), and theouter disk (green). Top: µ (cid:63) vs M dust / M (cid:63) . All pixel values are binned in µ (cid:63) ; each point is the average of a bin. Bottom: µ (cid:63) vs sSFR, where allpixels are binned in sSFR. macro-regions, Andromeda shows hints of a similar behaviour,with the sSFR declining in the outer disk with respect to the in-ner one. However, the 10 kpc ring exhibits a clear peak in SFRand sSFR.This can also be seen from the top panel of Fig. 7, where thescaling relations of the individual pixels are separated in the dif-ferent macro-regions, and the pixel-derived properties are binnedwithin each macro-region: the bulge (red), inner disk (cyan), ring(blue), and outer disk (green). In the log( µ (cid:63) ) vs log( M dust / M (cid:63) )plot, all regions but the ring follow a continuous relation whichcoincides with the HRS scaling relation. The ring is clearly o ff -set in this relation, indicating a dustier environment compared tothe average in the disk.The specific dust mass was found to correlate with both thestellar mass surface density and NUV–r colour. It is important toverify whether both relations are not connected by a tighter cor-relation between µ (cid:63) and NUV–r. In the bottom panel of Fig. 7,the relation between µ (cid:63) and sSFR is displayed. We prefer sSFRover NUV–r as the former allows a direct comparison of physicalquantities. We find that both quantities are tightly anti-correlated(see also Salim et al. 2005), so any evident link with one of themdirectly implies a link with the other. It must be noted, however,that sSFR values below 10 − yr − should be interpreted withcare, these values correspond to low star formation rates for anystellar mass and are subject to model degeneracies. It is imme-diately evident, as seen before, that the regions can be separatedin stellar mass surface density. All of them lie within a small interval in µ (cid:63) of about 1 order of magnitude. In sSFR, how-ever, the spread within one region covers over three orders ofmagnitude, except for the bulge. This implies that, no matter thedensity of stars, a wide range of star formation rates are possible.The bulge, which is the region with the lowest SFR, is an obvi-ous exception to this trend. Across the regions, the correlationsSFR and µ (cid:63) is not particularly tight. In comparison, the link be-tween both parameters and the specific dust mass is much morecompact. The correlation between sSFR and µ (cid:63) is therefore notlikely to be the main driver of the other dust scaling relations. We can now go a step further and see if and how the aforemen-tioned scaling relations still hold at a sub-kpc level. We are notyet able to reach the resolution of the molecular clouds, the cra-dles of star formation, but we are instead sampling giant molec-ular cloud aggregates or complexes. Hence, we cannot yet say ifthe scaling relations will eventually break, and at which resolu-tion.In the lower panels of Fig. 6 we show the M dust / M (cid:63) ratioas a function of both the stellar mass surface density and NUV–r colour, for the statistically independent pixels. Regions witha higher number density of pixels in the plot are colour–codedin yellow. Displayed as a green line is the average relation forthe pixels. The relations found between the dust–to–stellar massratio and NUV–r and µ (cid:63) confirm, from a local perspective, thefindings of Cortese et al. (2012). As NUV–r traces sSFR verywell, we also confirm the local nature of the results from daCunha et al. (2010), who use the same spectral fitting tool aswe do, and found a strong correlation between the sSFR and thespecific dust content in galaxies.As already mentioned in Sect. 4.1, M31 has a significantlylower SFR compared to the galaxies in the da Cunha et al. (2010)sample. In the relation between the dust mass and the SFR foreach pixel, we recover an average displacement with respect tothe extrapolation of the empirical law found by da Cunha et al.(2010). Nevertheless, the slope of the average relation definedby individual pixels is remarkably similar. This suggests thatthe intrinsic star formation relation (i.e. more dust equals morestar formation, or vice versa) still holds, but is less e ffi cient andhence scaled down to lower SFR in M31.Following da Cunha et al. (2010) we can compare f µ , thefraction of the total IR luminosity contributed by dust in the dif-fuse ISM, to the specific SFR for each pixel. In Fig. 8, we showhow this relation, found for galaxies as a whole, is recovered ona local basis as well, with two main di ff erences. First, the regioncorresponding to values of f µ close to 1 is more densely popu-lated. It turns out that these regions correspond to the very redbulge of M31. Secondly, systematically lower sSFR values arefound for a given f µ at these local scales. This is likely a man-ifestation of the low star formation activity of Andromeda withrespect to the galaxies in the sample of da Cunha et al. (2010).More specifically, these results combined together show thatthe interplay between the ISM, the radiation field and the stellarcontent (so the SFR as well) takes place on a sub-kpc scale. Thatthe scaling relations were local in nature was already suggestedin the previous section, when we subdivided M31 in morpho-logically distinct regions. We can now state that, globally, themain scaling relations are built up from the properties of the lo-cal galactic environment.Furthermore, even on the smallest scales accessible with ourdata, the inside–out trend is, on average, preserved. The bulgeis dominated by very red sub-kpc regions with very low sSFR Article number, page 14 of 22. Viaene: Dust scaling relations in M31 − − − − − log(sSFR) [yr − ] . . . . f µ Fig. 8:
Density plot of the sSFR against f µ , the luminosity fraction ofISM dust to the total dust. Yellow points indicate a higher density ofpoints, red a lower density. The black line represents the mean trend ofthe points. values and high stellar densities. Pixels in the inner disk regionare on average slightly bluer, but still have low sSFR values. Starforming pixels are mostly found in the 10 kpc ring and also havethe highest specific star formation. In the outer part of M31 thestar forming activity drops again.It must be stressed that the above considerations are the av-erage trends constructed from the large set of pixels within eachmacro-region. It is clear from the sub-kpc regions plotted inFig. 6 and from the trends within each region in Fig. 7, that awide range of physical properties are present within each of themacro-regions. For example, the star forming ring does hold asmall number of early-type pixels, while it is dominated by thestarforming pixels. Vice versa, the inner disk of M31 holds asignificant amount of late-type pixels, while the bulk of its sub-kpc regions are red and have low sSFR values. The exceptionto these findings is the bulge of M31, which consists purely ofearly-type regions with high stellar densities and only a minorSFR. We can now analyse the local characteristics of the radiation fieldin Andromeda and study the dependence of the dust temperatureas a function of the stellar characteristics. This exercise has al-ready been performed in Smith et al. (2012b), but we repeat ithere using a physically consistent model, which simultaneouslytakes into account information from the whole spectral domain.In Fig. 9 we plot the cold dust component temperature as afunction of the stellar mass surface density, for each single in-dependent pixel (upper panel) and for each of the macro-regionsof M31 (lower panel). We observe a clear bimodality in thistrend, which can be characterised by two slopes, with a break atlog( T ) ∼ .
25 or T ∼
18 K. The upper part of the relation, to-wards the higher temperature regime, is entirely and exclusivelypopulated by pixels of the bulge. In the intermediate regime, theinner disk region dominates, while the ring and outer disk corre-spond to the lowest regime of temperatures and stellar densities.Interestingly, the transition takes place at stellar surface densi- . . . . . l og ( T I S M C / K ) BulgeDiskT log( µ ? ) [ M (cid:12) / kpc ] . . . . . . l og ( T I S M C / K ) BulgeInner diskRingOuter disk
Fig. 9: Top:
Density plot of the stellar mass surface density µ (cid:63) vs T ISMC for the individual sub-kpc regions of M31. Yellow points indicate ahigher density of points, red a lower density. A linear function was fitto the bulge (green line) and disk (blue line) regions. The black lineis a theoretical function for pure heating by the old stellar populations.
Bottom:
Same relation, but separated in the macro-regions for M31:the bulge (red), the inner disk (cyan), the 10 kpc ring (blue), and theouter disk (green). All pixel values are binned in µ (cid:63) ; each point is theaverage of a bin. ties of log( µ (cid:63) ) ∼ .
5. In the case of integrated galaxies, this isexactly the value that indicates the transition from disk to bulgedominated systems (Schiminovich et al. 2007).As described in Smith et al. (2012b), we expect a slope of6 − in a T d vs µ (cid:63) plot (as µ (cid:63) ∼ T + from the Stefan–Boltzmannlaw, weighted with the dust emissivity index β =
2) if the heat-ing of di ff use dust is purely due to the ISRF of the old stellarpopulations. The slope of the linear fit to the bulge pixels is6 . − , which compares to the value of 4 . − that was calcu-lated by Smith et al. (2012b), correlating the dust temperaturefrom black–body fitting and the 3 . µ m surface density. Figure 9visually shows how that our value is close to the expected one.The slope of the “low–temperature” regime is instead 11 . − .The bimodal relation we find is clearly indicative of two dif-ferent heating regimes. This duality lies in the line of previousinvestigations of dust heating sources on sub-kpc scales (e.g. Bo-quien et al. 2011; Bendo et al. 2012). In the disk of Andromeda,the dust is heated by both old and new stars, giving rise to arather flat slope and a significant amount of scatter. The bulgeof M31 is instead dominated by the light of old stars, makingthem the dominant dust heaters. This further supports the resultsof Smith et al. (2012b), Groves et al. (2012), and Draine et al.(2014). Article number, page 15 of 22 . Discussion and conclusions
In the present work, we have performed SED fitting of apanchromatic dataset, collected for our neighbour galaxy M31.New
Herschel observations were combined with GALEX,SDSS, WISE, and
Spitzer data, covering UV to submm wave-lengths and allowing us to derive, by exploiting a physicallyself–consistent model, some physical parameters both on aglobal and on a local scale. To create statistically independentregions, all the data were convolved to a resolution matchingthat of the SPIRE 500 µ m waveband, the lowest in our dataset,which allowed us to probe physical scales of ∼ ×
608 pc inthe plane of M31. In this paper we concentrate on the analysisof the scaling relations linking the dust and stellar properties.We have fitted a multi-component theoretical SED to eachpixel which allowed us to estimate several physical properties ofthat region. Every physical parameter for every pixel was givenan uncertainty estimation based on the broadness of its corre-sponding PDF. Physical quantities that could not be su ffi cientlyconstrained were removed from the sample. Furthermore, 2-Dparameter maps are constructed for each physical quantity.Additionally, we have decomposed Andromeda in fourmacro-regions: the bulge, the star forming ring (the so–called10 kpc ring), and the inner and outer disk regions. The same fit-ting routine was applied to the integrated fluxes of each of thesemain components, as well as to the global observed fluxes.From the point of view of the dust scaling relations, M31 isan average galaxy when compared to the local galaxies in theHRS sample. On the other hand, it lies above the average M dust vs SFR relation of da Cunha et al. (2010); despite a dust massclose to the sample average, Andromeda is forming stars signif-icantly less e ffi ciently than the other galaxies.By investigating the properties of the distinct morphologicalcomponents, we find strong hints for an inside–out star forma-tion scenario. In this evolutionary model, the bulk of stars arebeing formed at early epochs in the bulge, and the more recentstar formation happens at larger galactocentric distances (i.e. (cid:38) ff ect of the passage of M32through Andromeda’s disk. Beginning with a model with twospiral arms, they end up with a morphological structure closelymatching the 10 kpc ring and the “hole” which is easily visiblein IR images towards the south. Similarly, Gordon et al. (2006)argued that a head–on encounter with M32, might have resultedin star forming waves propagating through the 10 kpc ring.The bulge and inner disk region have red colours and highstellar mass surface density ( µ (cid:63) ). The star forming ring and outerdisk region are bluer and have lower µ (cid:63) . Each of these regionslies on the average trend of the HRS scaling relations. In termsof NUV–r colour, the bulge of M31 is a remarkable exception,being redder than any of the submm detected HRS galaxies. Themacro-regions thus have characteristics closely resembling thoseof global galaxies, where the bulge and inner disk may be seenas early-type while the ring and outer disk resemble late-typegalaxies.The results for M31 not only support an inside-out forma-tion pattern for the bulge-disk morphology, but they also tell us something about the di ff erences in the local environments. Whenlooking at the dust–to–stellar mass ratio as a function of the stel-lar mass surface density (see Fig. 7), we observe a smooth tran-sition from high stellar mass / low dust regions in the bulge, tothe low stellar mass / high dust content of the outermost regions.Only the starforming ring, containing a significant fraction of thedust of the whole galaxy, is slightly displaced from this relation.On the other hand, the four regions behave quite di ff erentlywhen the sSFR is plotted as a function of µ (cid:63) (see Fig. 7). A ten-dency is found for regions of higher stellar mass surface densityto host less star formation, in a way mimicking an internal down-sizing process in star formation, in which the regions of higheststellar density have already stopped forming stars. Within eachregion, the variation in µ (cid:63) spans only about 1 order of magni-tude, whereas the sSFR always varies by more than 3 (with theexception of the bulge, where there is barely any star formation).These relations seem to suggest that downsizing relationsbreak down when considering the smallest scales. Star formationat these scales has a weak dependence on the stellar mass: small-scale environments characterised by di ff erent stellar masses canbe associated with very di ff erent levels of star formation, at leastas far as a quiescent galaxy like M31 is concerned. What drivesthe general star formation mode, which determines where thegalaxy as a whole places itself on the scaling relations, must in-stead be the total mass of all its constituents.When considering the modelling of the observed SED on thesmallest scales, our main conclusions are as follows.1. The SED of sub-kpc regions can be successfully fitted usinggalaxy-based models, provided that the parameter space isadequately sampled.2. When investigating the dust heating in the bulge, we recoverthe theoretical ( T ISMC ) ∼ µ (cid:63) relation. This indicates that oldstars are the dominant heating source in this region. Thedust heating is more ambiguous in the disk, where both starformation and the di ff use ISRF irradiate the dust.3. We find strong correlations, on a pixel–by–pixel scale, be-tween M dust / M (cid:63) and NUV–r (or, equivalently, sSFR), andbetween M dust / M (cid:63) and µ (cid:63) . These scaling relations, involv-ing the dusty component of the ISM, are remarkably similarto those found for entire local galaxies. This suggests thatthe dust scaling relations are built in situ , with underlyingphysical processes that must be local in nature.4. As already found for other galaxies, M31 seems to have un-dergone an inside–out evolution in its star formation process,possibly influenced by interactions with its satellites.5. When considering the smallest scales, a wide range in dustcontent, sSFR, and M dust / M (cid:63) is found within Andromeda il-lustrating the great diversity of sub-kpc regions. Even withinthe late-type ring and outer disk of M31, early-type micro-regions can be found. Vice versa, the inner part of M31 stillholds a small number of late-type regions.The fact that we are able to reproduce the dust scaling rela-tions on a sub-kpc scale states that these relations are not onlypartially a manifestation of a galaxy-wide equilibrium, but theyalso arise from local scales. Local evolutionary processes in-volving dust creation and destruction lie at the base of these re-lations. The balance between dust depletion and production re-flects the relative presence of old and new stars, the latter beingresponsible for dust generation. This raises the question: at whatscales does the balanced interplay between dust and stars breakdown? Answers to this may be found in similar studies of theGalactic ISM or by future, high–resolution FIR space missions.Zooming into the ISM of our own galaxy can unveil two very Article number, page 16 of 22. Viaene: Dust scaling relations in M31 di ff erent results. If these scaling relations break down at the sizeof individual molecular clouds, it would indicate that non–localscattered light plays an important role in the dust energy balance.Alternatively, if the scaling relations stay intact, non–local lightis negligible at each scale, which would call for a revision of thephysical properties of interstellar dust. Acknowledgements.
We would like to thank Elisabete da Cunha for kindly pro-viding the extra libraries in MAGPHYS.We thank all the people involved in the construction and the launch of Herschel.SPIRE has been developed by a consortium of institutes led by Cardi ff Uni-versity (UK) and including Univ. Lethbridge (Canada); NAOC (China); CEA,LAM (France); IFSI, Univ. Padua (Italy); IAC (Spain); Stockholm Observatory(Sweden); Imperial College London, RAL, UCL-MSSL, UKATC, Univ. Sussex(UK); and Caltech, JPL, NHSC, Univ. Colorado (USA). This development hasbeen supported by national funding agencies: CSA (Canada); NAOC (China);CEA, CNES, CNRS (France); ASI (Italy); MCINN (Spain); SNSB (Sweden);STFC and UKSA (UK); and NASA (USA). HIPE is a joint development (arejoint developments) by the
Herschel
Science Ground Segment Consortium, con-sisting of ESA, the NASA
Herschel
Science Center, and the HIFI, PACS andSPIRE consortia.GALEX is a NASA Small Explorer, launched in 2003 April. We gratefullyacknowledge NASA’s support for construction, operation and science analysisfor the GALEX mission, developed in cooperation with the Centre Nationald’Etudes Spatiales (CNES) of France and the Korean Ministry of Science andTechnology.Funding for the SDSS and SDSS-II has been provided by the Alfred P. SloanFoundation, the Participating Institutions, the National Science Foundation, theUS Department of Energy, the National Aeronautics and Space Administration,the Japanese Monbukagakusho, the Max Planck Society, and the Higher Educa-tion Funding Council for England. The SDSSWeb Site is http: // / .The SDSS is managed by the Astrophysical Research Consortium for the Partic-ipating Institutions. The Participating Institutions are the American Museum ofNatural History, Astrophysical Institute Potsdam, University of Basel, Universityof Cambridge, Case Western Reserve University, University of Chicago, DrexelUniversity, Fermilab, the Institute for Advanced Study, the Japan ParticipationGroup, Johns Hopkins University, the Joint Institute for Nuclear Astrophysics,the Kavli Institute for Particle Astrophysics and Cosmology, the Korean Scien-tist Group, the Chinese Academy of Sciences (LAMOST), Los Alamos NationalLaboratory, the Max-Planck-Institute for Astronomy (MPIA), the Max-Planck-Institute for Astrophysics (MPA), New Mexico State University, Ohio State Uni-versity, University of Pittsburgh, University of Portsmouth, Princeton University,the United States Naval Observatory, and the University of Washington.This work is based [in part] on observations made with the Spitzer
Space Tele-scope, which is operated by the Jet Propulsion Laboratory, California Institute ofTechnology, under NASA contract 1407. We especially thank P. Barmby and K.Gordon for providing the Spitzer data.This publication makes use of data products from the Wide-field Infrared SurveyExplorer, which is a joint project of the University of California, Los Angeles,and the Jet Propulsion Laboratory / California Institute of Technology, funded bythe National Aeronautics and Space Administration.
References
Agius, N. K., Sansom, A. E., Popescu, C. C., et al. 2013, MNRAS, 431, 1929Aniano, G., Draine, B. T., Calzetti, D., et al. 2012, ApJ, 756, 138Aniano, G., Draine, B. T., Gordon, K. D., & Sandstrom, K. 2011, PASP, 123,1218Azimlu, M., Marciniak, R., & Barmby, P. 2011, AJ, 142, 139Barmby, P., Ashby, M. L. N., Bianchi, L., et al. 2006, ApJ, 650, L45Barmby, P., Perina, S., Bellazzini, M., et al. 2009, AJ, 138, 1667Beaton, R. L., Majewski, S. R., Guhathakurta, P., et al. 2007, ApJ, 658, L91Bendo, G. J., Boselli, A., Dariush, A., et al. 2012, MNRAS, 419, 1833Bendo, G. J., Calzetti, D., Engelbracht, C. W., et al. 2007, MNRAS, 380, 1313Bendo, G. J., Draine, B. T., Engelbracht, C. W., et al. 2008, MNRAS, 389, 629Bendo, G. J., Wilson, C. D., Warren, B. E., et al. 2010, MNRAS, 402, 1409Bertin, E. & Arnouts, S. 1996, A&AS, 117, 393Bianchi, S. 2008, A&A, 490, 461Block, D. L., Bournaud, F., Combes, F., et al. 2006, Nature, 443, 832Boquien, M., Boselli, A., Buat, V., et al. 2013, A&A, 554, A14Boquien, M., Buat, V., Boselli, A., et al. 2012, A&A, 539, A145Boquien, M., Calzetti, D., Combes, F., et al. 2011, AJ, 142, 111Boselli, A., Boissier, S., Cortese, L., et al. 2009, ApJ, 706, 1527Boselli, A., Eales, S., Cortese, L., et al. 2010, PASP, 122, 261Brinchmann, J., Charlot, S., White, S. D. M., et al. 2004, MNRAS, 351, 1151Bruzual, G. & Charlot, S. 2003, MNRAS, 344, 1000 Chabrier, G. 2003, PASP, 115, 763Charlot, S. & Fall, S. M. 2000, ApJ, 539, 718Chemin, L., Carignan, C., & Foster, T. 2009, ApJ, 705, 1395Clemens, M. S., Negrello, M., De Zotti, G., et al. 2013, MNRASCorbelli, E., Lorenzoni, S., Walterbos, R., Braun, R., & Thilker, D. 2010, A&A,511, A89Cortese, L., Ciesla, L., Boselli, A., et al. 2012, A&A, 540, A52da Cunha, E., Charlot, S., & Elbaz, D. 2008, MNRAS, 388, 1595da Cunha, E., Eminian, C., Charlot, S., & Blaizot, J. 2010, MNRAS, 403, 1894Davies, J. I., Wilson, C. D., Auld, R., et al. 2010, MNRAS, 409, 102de Vaucouleurs, G., de Vaucouleurs, A., Corwin, Jr., H. G., et al. 1991, ThirdReference Catalogue of Bright Galaxies.Draine, B. T. 2003, ARA&A, 41, 241Draine, B. T., Aniano, G., Krause, O., et al. 2014, ApJ, 780, 172Draine, B. T., Dale, D. A., Bendo, G., et al. 2007, ApJ, 663, 866Draine, B. T. & Li, A. 2007, ApJ, 657, 810Dunne, L., Eales, S., Edmunds, M., et al. 2000, MNRAS, 315, 115Engelbracht, C. W., Blaylock, M., Su, K. Y. L., et al. 2007, PASP, 119, 994Fazio, G. G., Hora, J. L., Allen, L. E., et al. 2004, ApJS, 154, 10Fisher, D. B., Drory, N., & Fabricius, M. H. 2009, ApJ, 697, 630Ford, G. P., Gear, W. K., Smith, M. W. L., et al. 2013, ApJ, 769, 55Foyle, K., Natale, G., Wilson, C. D., et al. 2013, MNRAS, 432, 2182Fritz, J., Gentile, G., Smith, M. W. L., et al. 2012, A&A, 546, A34Galametz, M., Kennicutt, R. C., Albrecht, M., et al. 2012, MNRAS, 425, 763Galametz, M., Madden, S. C., Galliano, F., et al. 2011, A&A, 532, A56Galliano, F., Hony, S., Bernard, J.-P., et al. 2011, A&A, 536, A88Gil de Paz, A., Boissier, S., Madore, B. F., et al. 2007, ApJS, 173, 185González Delgado, R. M., Pérez, E., Cid Fernandes, R., et al. 2013,arXiv:1310.5517Gordon, K. D., Bailin, J., Engelbracht, C. W., et al. 2006, ApJ, 638, L87Gordon, K. D., Engelbracht, C. W., Fadda, D., et al. 2007, PASP, 119, 1019Gri ffi n, M. J., Abergel, A., Abreu, A., et al. 2010, A&A, 518, L3Groves, B., Krause, O., Sandstrom, K., et al. 2012, MNRAS, 426, 892Herschel Space Observatory. 2011, SPIRE Observer’s Manual, http://herschel.esac.esa.int/Docs/SPIRE/html/spire_om.html Hughes, T. M., Baes, M., Fritz, J., et al. 2014, ArXiv e-printsIssa, M. R., MacLaren, I., & Wolfendale, A. W. 1990, A&A, 236, 237Jarrett, T. H., Cohen, M., Masci, F., et al. 2011, ApJ, 735, 112Jarrett, T. H., Masci, F., Tsai, C. W., et al. 2013, AJ, 145, 6Karczewski, O. Ł., Barlow, M. J., Page, M. J., et al. 2013, MNRAS, 431, 2493Keel, W. C., Manning, A. M., Holwerda, B. W., Lintott, C. J., & Schawinski, K.2014, AJ, 147, 44Kennicutt, R. C. & Evans, N. J. 2012, ARA&A, 50, 531Kirk, J. M., Gear, W. K., Fritz, J., et al. 2014, arXiv:1306.2913Krause, O. in prep.Leroy, A. K., Walter, F., Brinks, E., et al. 2008, AJ, 136, 2782Lisenfeld, U. & Ferrara, A. 1998, ApJ, 496, 145Liu, G., Calzetti, D., Hong, S., et al. 2013, ApJ, 778, L41Lutz, D. 2010, PACS photometer PSF, http://herschel.esac.esa.int/twiki/pub/Public/PacsCalibrationWeb/bolopsfv1.01.pdf
Magrini, L., Bianchi, S., Corbelli, E., et al. 2011, A&A, 535, A13Martin, D. C., Fanson, J., Schiminovich, D., et al. 2005, ApJ, 619, L1McConnachie, A. W., Irwin, M. J., Ferguson, A. M. N., et al. 2005, MNRAS,356, 979Mentuch Cooper, E., Wilson, C. D., Foyle, K., et al. 2012, ApJ, 755, 165Mo, H. J., Mao, S., & White, S. D. M. 1998, MNRAS, 295, 319Montalto, M., Seitz, S., Ri ff eser, A., et al. 2009, A&A, 507, 283Moorthy, B. K. & Holtzman, J. A. 2006, MNRAS, 371, 583Morrissey, P., Conrow, T., Barlow, T. A., et al. 2007, ApJS, 173, 682Muñoz-Mateos, J. C., Gil de Paz, A., Boissier, S., et al. 2009a, ApJ, 701, 1965Muñoz-Mateos, J. C., Gil de Paz, A., Boissier, S., et al. 2007, ApJ, 658, 1006Muñoz-Mateos, J. C., Gil de Paz, A., Zamorano, J., et al. 2009b, ApJ, 703, 1569Noll, S., Burgarella, D., Giovannoli, E., et al. 2009, A&A, 507, 1793Padmanabhan, N., Schlegel, D. J., Finkbeiner, D. P., et al. 2008, ApJ, 674, 1217Parkin, T. J., Wilson, C. D., Foyle, K., et al. 2012, MNRAS, 422, 2291Pérez, E., Cid Fernandes, R., González Delgado, R. M., et al. 2013, ApJ, 764,L1Petty, S. M., Neill, J. D., Jarrett, T. H., et al. 2013, AJ, 146, 77Pilbratt, G. L., Riedinger, J. R., Passvogel, T., et al. 2010, A&A, 518, L1Poglitsch, A., Waelkens, C., Geis, N., et al. 2010, A&A, 518, L2Popescu, C. C. & Tu ff s, R. J. 2002, MNRAS, 335, L41Popescu, C. C., Tu ff s, R. J., Dopita, M. A., et al. 2011, A&A, 527, A109Popescu, C. C., Tu ff s, R. J., Völk, H. J., Pierini, D., & Madore, B. F. 2002, ApJ,567, 221Richstone, D. & Sargent, W. L. W. 1972, ApJ, 176, 91Rieke, G. H., Young, E. T., Engelbracht, C. W., et al. 2004, ApJS, 154, 25Rowlands, K., Dunne, L., Maddox, S., et al. 2012, MNRAS, 419, 2545Salim, S., Charlot, S., Rich, R. M., et al. 2005, ApJ, 619, L39Sandstrom, K. M., Leroy, A. K., Walter, F., et al. 2012, arXiv:1212.1208Schiminovich, D., Wyder, T. K., Martin, D. C., et al. 2007, ApJS, 173, 315 Article number, page 17 of 22 ick, J., Courteau, S., Cuillandre, J.-C., et al. 2013, arXiv:1303.6290Skrutskie, M. F., Cutri, R. M., Stiening, R., et al. 2006, AJ, 131, 1163Smith, D. J. B., Dunne, L., da Cunha, E., et al. 2012a, MNRAS, 427, 703Smith, M. W. L., Eales, S. A., Gomez, H. L., et al. 2012b, ApJ, 756, 40Smith, M. W. L., Vlahakis, C., Baes, M., et al. 2010, A&A, 518, L51Sodroski, T. J., Odegard, N., Arendt, R. G., et al. 1997, ApJ, 480, 173Tabatabaei, F. S. & Berkhuijsen, E. M. 2010, A&A, 517, A77Tabatabaei, F. S., Braine, J., Xilouris, E. M., et al. 2014, A&A, 561, A95Tamm, A., Tempel, E., Tenjes, P., Tihhonova, O., & Tuvikene, T. 2012, A&A,546, A4Tempel, E., Tamm, A., & Tenjes, P. 2010, A&A, 509, A91Tempel, E., Tuvikene, T., Tamm, A., & Tenjes, P. 2011, A&A, 526, A155Thilker, D. A., Hoopes, C. G., Bianchi, L., et al. 2005, ApJ, 619, L67Tu ff s, R. J., Popescu, C. C., Völk, H. J., Kylafis, N. D., & Dopita, M. A. 2004,A&A, 419, 821Walterbos, R. A. M. & Schwering, P. B. W. 1987, A&A, 180, 27Werner, M. W., Roellig, T. L., Low, F. J., et al. 2004, ApJS, 154, 1White, S. D. M. & Frenk, C. S. 1991, ApJ, 379, 52Wright, E. L., Eisenhardt, P. R. M., Mainzer, A. K., et al. 2010, AJ, 140, 1868York, D. G., Adelman, J., Anderson, Jr., J. E., et al. 2000, AJ, 120, 1579 Appendix A: Multi-wavelength data processing
In this section we present the road towards spatially resolved,panchromatic SED fitting. The data obtained from varioussources and own
Herschel observations are manipulated in or-der to make a consistent comparison over such a wide wave-length range (Appendix A.1). These manipulations bring withthem a complex uncertainty propagation which is addressed inAppendix A.2.
Appendix A.1: Image manipulations
Appendix A.1.1: Background subtraction
The WISE and GALEX subsets had a non-zero average back-ground value due to emission from unresolved sources. The
Herschel images also come with a flat, non-zero background asa consequence of the data reduction process. The average back-ground for the
Spitzer and SDSS images was already zero, henceno background subtraction was performed for these frames.While global background gradients were significant in theWISE frames, no clear gradients were identified in the GALEXor
Herschel images. We therefore fit and subtract a second–orderpolynomial to the background in the WISE frames using stan-dard
ESO-MIDAS routines. In the other frames, we estimate thebackground as follows.A set of regions was chosen far enough from visible emis-sion from M31 to avoid contamination by the galaxy, but closeenough to make a reliable estimation of the background nearthe M31. Inside these pre-defined regions, a number of aper-ture measurements was made. The number of measurements perregion was set to be proportional to the number of pixels in-side. For PACS and SPIRE, a total of 10000 measurements werespread over 8 regions. In the GALEX fields, 20000 measure-ments were divided over 23 regions. From the set of measuredfluxes, a sigma clipped median was derived as a reliable esti-mate for the background flux. We used 3 σ as a threshold anditerated until convergence. This median background value wasconsequently subtracted from the images. Appendix A.1.2: Masking
Andromeda covers a large part of the sky for a single galaxyand lies close to the Galactic disk (with a Galactic latitude of − . ◦ ). It is consequently contaminated by the light of thou-sands of foreground stars, especially in the UV and optical partof the spectrum. At longer wavelengths, the infrared emission ofbackground galaxies becomes the main source of contaminatingsources. At Herschel wavelengths, however, most of the emis-sion from non-M31 point sources is negligible even at scales ofthe SPIRE 500 µ m beam. As mentioned before, the extendedemission of the Milky Way Galactic Cirrus is prominently visi-ble here. This dust emission can fortunately be associated withHI emission. Using the velocity information of HI maps, theGalactic cirrus can partly be disentangled from the emission ofM31. Paper I goes into more detail about this technique.We made use of SExtractor v2 . . ff use emission from M31. In this way, we could re-place the non-M31 point sources with the local M31 backgroundvalue obtained from these maps. Article number, page 18 of 22. Viaene: Dust scaling relations in M31 h m m m m m m m m m RA (J2000) +42 ◦ D e c ( J ) +42 ◦ D e c ( J ) h m m m m m m m m m m m RA (J2000) +42 ◦ D e c ( J ) +42 ◦ D e c ( J ) .
00 0 .
03 0 .
06 0 .
09 0 .
12 0 . S ( Jy/Sr )0 1 2 3 4 5 S ( Jy/Sr ) uu - masked3.63.6 - masked Fig. A.1:
Masking of point sources that do not belong to M31: beforeand after view of the galaxy in the u band (top) and IRAC 3 . µ m band(bottom). For each source an optimal radius was derived by comparingthe pixel flux with the local background at increasing distancefrom the peak location. Once the pixel-to-background flux ratiodropped below 2, the radius was cut o ff at that distance. Basedon this radius, a total flux was extracted in order to make colourevaluations. We constructed point source masks for the GALEX,SDSS, WISE, and Spitzer subsets based on di ff erent colour cri-teria.The GALEX and SDSS point sources were evaluated basedon their UV colour. This technique was applied by Gil de Pazet al. (2007) for over 1000 galaxies and proved successful. Inpractice, we mask all sources with | FUV-NUV | > .
75 (A.1)if they are detected at the 1 σ level in their particular wavelengthband. SExtractor identified 58330 point sources in the UV fields,of which over 51000 were masked in the FUV and NUV. Manypoint sources from the UV catalogue were not detected at opti-cal bands, hence only 25000 sources were masked in the SDSSbands. Around 7000 sources were identified as extragalactic.They were therefore assumed to belong to M31 and were notmasked.As an example, Fig. A.1 shows the u –band image of M31before and after the mask was applied. The contamination of theimage has been significantly reduced using the above technique. . . . . . F . /F . . . . . . . F . / F Fig. A.2:
Colour-colour plot of the IRAC selected bright sources. Thesources inside the red rectangle are identified as H ii regions belongingto M31. They were consequently not masked. The point sources in the WISE and the
Spitzer
IRAC andMIPS frames were masked analogously, based on their IRACcolours (see below). At these wavelengths, however, the non-M31 point sources are a mix of foreground stars and backgroundgalaxies. Furthermore, some bright sources may be associatedwith H ii regions in M31 and must not be masked. We designed ascheme based on the technique by Muñoz-Mateos et al. (2009b),which was successfully applied to the SINGS galaxies. Fore-ground stars have almost no PAH emission, while the di ff useISM in galaxies shows a roughly constant F . / F ratio (Draine& Li 2007). Background galaxies are redshifted spirals or ellip-ticals and can consequently have a wide range in F . / F . It isthus possible to construct a rough filter relying on the di ff erencein MIR flux ratios. First, it was checked which point source ex-tracted from the IRAC 3 . µ m had a non-detection at 8 µ m. Thiscriterion proved to be su ffi cient to select the foreground starsin the field. A second, colour-based, criterion disentangled thebackground galaxies from the H ii regions:0 . < F . / F < .
85 (A.2) F . / F . < . . (A.3)Figure A.2 shows the colour–colour diagram for these sources.The H ii regions follow a more or less horizontal track at thelower–left part of the plot. The colour criteria for filtering outthese H ii regions were obtained empirically to ensure e ff ectiveidentification. Once identified, these star forming regions wereconsequently not masked. The resulting mask was applied to allIRAC and MIPS bands. Sources that were not detected at longerwavelengths were obviously not masked. Figure A.1 shows theIRAC 3 . µ m image of M31 before and after the mask was ap-plied. SExtractor identified 1933 sources in the IRAC bands and536 in MIPS. From these catalogues, around 1800 sources weremasked in IRAC and 230 in MIPS. All masked sources in IRACwere also masked in the WISE frames as both instruments coverroughly the same wavelength range. No additional masking wasnecessary for the WISE data.As a way to check the reliability of our masks, we comparedthe masked regions with the locations of known Andromeda Article number, page 19 of 22 ources, i.e. H ii regions and planetary Nebula (Azimlu et al.2011) and bright young clusters (Barmby et al. 2009). The over-lap between our masked sources and actual M31 sources provednegligible; 0 . . . . .
5% of the identifiedsources were incorrectly masked in the GALEX, SDSS, WISE,IRAC, and MIPS bands, respectively. The handful of sourcesthat were incorrectly masked were manually restored.
Appendix A.1.3: Convolution and rescaling
The masked images were all brought to the same resolution be-fore extracting the separate pixel values. By doing this, infor-mation is lost because of the significantly lower resolutions ofthe end products. It is, however, a critical step to make a con-sistent comparison of the fluxes over this wide range of wave-lengths. Our working resolution was limited by the SPIRE500 µ m point spread function, which is 36 . . ffi cient IDL rou-tine convolve_image.pro which makes use of their designedkernels and takes NaN values into account when convolving.The kernels for the GALEX, WISE, IRAC, MIPS, PACS, andSPIRE instruments were readily available to make the convolu-tion. For the SDSS images, we used a Gaussian-to-SPIRE 500kernel which assumes an initial FWHM of 4 (cid:48)(cid:48) for the SDSS im-ages (see Sect. 2.2).The SPIRE 500 µ m beam is sampled with a pixel scale of12 (cid:48)(cid:48) which means the PSF covers nine pixels which are not in-dependent. The frame thus had to be rebinned to a pixel scaleof 36 (cid:48)(cid:48) to make each pixel correspond to a statistically indepen-dent region in M31. The convolved frames were consequentlyrescaled to match the pixel grid of the SPIRE 500 µ m rebinnedimage. Our data cube covers the electromagnetic spectrum fromUV to submm wavelengths. Figure A.3 gives an overview of allframes used for the fitting of a panchromatic SED to each pixel.This series of steps results in sets of corresponding pixelswhich each represent a physical region of 136 . × . i = ◦ ). O ff course, it must be noted that the third dimension, the directionalong the line of sight, also contributes to the appearance of eachpixel. Spiral galaxies are, however, known to have relativelythin disks compared to their lengths, so even along this axis, theresolution remains subgalactic. The attenuation e ff ects of thislarger dimension will, however, be treated during the modellingin terms of optical depth parameters (see Sect. 3.1). Appendix A.2: Single–pixel uncertainties
Each pixel comes with several sources of uncertainty, which willhave to be estimated and combined to a total error. Uncertaintypropagation based on initial errors can become complex and haz-ardous after masking, convolutions, rebinning, and rescaling. Intheir Appendix D.1, Aniano et al. (2012) opted to start the un-certainty estimation after all these steps in their resolved analy-sis for NGC 628 and NGC 6946. They postulate two sources ofuncertainties: background variations and calibration errors. Ad-ditionally, we add a third source for the UV and optical subsets:the Poisson error. This term is negligible for infrared and submmobservations because of the large number of incoming photons.
Appendix A.2.1: Background variations
Variations of the background can give rise to errors in the fluxmeasurements. To estimate the impact of this term, we select re-gions around M31 where the background is dominant. A sigmaclipping filter is used on the pixels inside each region. Di ff er-ent methods of sigma clipping were evaluated. They proved notto a ff ect the resulting variance by much as the evaluated regionswere chosen to be free of point sources. In the end, a low–levelclipping was done (10 σ and only two iterations) to filter out anynon-background emission. The variance of the remaining pixelsfrom all of the regions will be a good representation of the back-ground variation error σ background , following Equation D2 fromAniano et al. (2012), σ background = (cid:115) N bg − (cid:88) (x , y) [ I obs ( x , y )] , (A.4)where N bg is the number of background pixels used and I obs theobserved background subtracted flux of the pixel with coordi-nates x and y. For each telescope, a di ff erent set of backgroundregions was used. This was needed because the background andGalactic Cirrus features change in morphology and brightnessalong the electromagnetic spectrum. Appendix A.2.2: Poisson errors
Photon arrivals are considered a random event following a Pois-son distribution; the variability of the counts scales with thesquare root of the number of photons. Infrared and submm ob-servations deal with huge numbers of photons (all of which havelow energy) and consequently have a negligible Poisson errorcompared to calibration uncertainties and background variations.Optical and UV observations, however, do not collect asmany photons and consequently their Poisson-like nature couldstart to play a more prominent role. In the cases of the GALEXand SDSS observations, the images were converted from fluxto actual photon counts using the individual exposure time foreach pixel and then converted to counts / Sr. This surface densityunit was necessary to rebin and rescale the frames to the SPIRE500 µ m pixel grid without disrupting the exposure time infor-mation. We note that convolution did not take place here. Theresulting images were converted back to counts using the newpixel scale and from here the Poisson errors could be computedfor each individual pixel by taking the square root of the counts. Appendix A.2.3: Calibration errors
In higher signal-to-noise areas, the calibration of the instrumen-tation can become a dominant term. We therefore include afixed percentage as calibration uncertainty for each filter (seeTable A.1) in addition to the previous error terms.Finally, all error sources are added in quadrature for eachpixel i and each wavelength band λ to obtain its total photometricuncertainty σ Tot λ, i = (cid:113) ( σ bg λ, i ) + ( σ cal λ, i ) + ( σ Pois λ, i ) . (A.5) Appendix B: Integrated photometry of the separateregions
Article number, page 20 of 22. Viaene: Dust scaling relations in M31 − . − . − . . . . − . − . . . − . − . − . . . . − . − . − . . . . − . − . . . − . − . . . − . − . . . − . − . . . − . − . . . − . − . . . − . − . − . . . . − . − . − . . . . − . − . − . . . . − . − . . . . . . . . . . . . . . Flux (arbitrary units)
Offset from centre (degrees) O ff s e tf r o m c en t r e ( deg r ee s ) FUV NUV ug r iz W1 3.64.5 W2 5.88 W3 W424 70 100160 250 350500
Fig. A.3:
Overview of the FUV to submm dataset at SPIRE 500 resolution, rotated from a position angle of 38 ◦ .Article number, page 21 of 22 able B.1: Overview of the obtained fluxes for the di ff erent regions of Andromeda. All measurements are in units of Jansky. Band Global Bulge Inner disk Ring Outer disk M32FUV 1.483 ± ± ± ± ± ± ± ± ± ± ± ± u ± ± ± ± ± ± g ± ± ± ± ± ± r ± ± ± ± ± ± i ± ± ± ± ± ± z ± ± ± ± ± ± ± ± ± ± ± ± ±
24 63.1 ± ± ± ± ± ±
11 36.7 ± ± ± ± ± ± ± ± ± ± ± ±
49 34.4 ± ±
15 59 ±
13 59 ±
13 1.75 ± ±
35 21.2 ± ± ±
16 44.4 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±
48 8 ± ±
10 56 ±
22 29 ±
12 0.5 ± ±
110 97.9 ± ±
23 492 ±
49 233 ±
23 1.49 ± ±
350 195 ±
19 805 ±
81 1601 ±
160 893 ±
89 8.13 ± ±
750 199 ±
20 1630 ±
160 3588 ±
360 2094 ±
210 18.9 ± ±
420 87.4 ± ±
79 2736 ±
190 1992 ±
140 13.25 ± ±
220 33.8 ± ±
36 1395 ±
98 1168 ±
82 7.29 ± ±
94 11.76 ± ±
14 583 ±
41 551 ±
39 3.37 ± Article number, page 22 of 22 able A.1:
Overview of the adopted relative calibration uncertaintiesfor each pixel.
Filter Error ReferenceGALEX-FUV 5% aGALEX-NUV 3% aSDSS 2% bWISE-W1 2 .
4% cWISE-W2 2 .
8% cWISE-W3 4 .
5% cWISE-W4 5 .
7% cIRAC-3 . .
3% dIRAC-4 . .
1% dIRAC-5 . .
1% dIRAC-8 16 .
7% dMIPS 24 4% eMIPS 70 10% fPACS 10% gSPIRE 7% h