The DustPedia Archive is now available here!

# Scientific Goals

## Introduction

Four hundred thousand years after the Big Bang the Universe had cooled sufficiently to form a gas of atoms. Predominantly these atoms were of the simplest form possible (Hydrogen and Helium) with just a small fraction of heavier elements, which astronomers collectively refer to as ‘metals’. Subsequently clouds of this gas collapsed and merged together to form stars and galaxies. Since then the stars have been enriching the gas in galaxies with metals via the process of nucleosynthesis. This increasing abundance of metals was a crucial step in the path that led to the formation of the Earth and eventually intelligent life.

As a galaxy evolves new stars form from the metal enriched gas, which contains about 60% of its metals directly as atoms or molecules in the gas phase and about 40%, which are contained within larger particles that astronomers refer to as cosmic dust. Cosmic dust forms by nucleation and growth from the vapour phase in the cool atmospheres of low mass stars as they come to the end of their lives and probably also in the gas ejected from supernovae as more massive stars expire. Once deposited into the interstellar medium the dust grains are subject to various physical processes that allow them to grow via the accretion of atoms and molecules and disintegrate in shock heated gas or via high-energy photon or cosmic ray processing. The dust is composed of a mixture of carbonaceous and amorphous silicate grains with a size distribution governed by the growth and destruction mechanisms (sizes of order 0.01-1.0μm).

The cosmic dust reveals itself to us primarily in three ways. Firstly, the dust absorbs and scatters radiation from the stars, which not only causes the extinction of the light, but also causes the stellar spectrum to become ‘redder’. Secondly, because the absorbed stellar light heats the dust it subsequently radiates some of its energy away. Dust temperatures range from about 10-100K and so the energy radiated from dust is emitted predominantly in the far-infrared and sub-mm part of the electromagnetic spectrum (wavelengths of about 10μm to 1mm). Imprinted on this spectrum are signatures of dust composition, structure and chemistry. Finally, the alignment of dust gains in the Galactic magnetic field leads to the polarisation of starlight [1].

One might argue that dust extinction and polarisation are effects imprinted on the radiation from stars while the only direct measure of the dust itself comes from the radiation it emits. Observation of this radiation had to await the arrival of space telescopes because, by far the majority of the cosmic far-infrared and sub-mm radiation is efficiently absorbed by molecules in the Earth’s atmosphere. The first far infrared space telescope (IRAS) was launched in 1984 and it revolutionised our ideas about what the physical properties of cosmic dust are, just how much dust there was and how important the dust is in governing physical processes in the interstellar medium. We now know that for a typical galaxy like the Milky Way a little under half of the radiation produced by stars is subsequently re-processed through cosmic dust. There are other galaxies in which 99% of the stellar radiation is reprocessed in this way. Subsequent space missions (ISO, Spitzer) have greatly extended our understanding of cosmic dust. In addition, considerable steps forward have been gained through observations of dust emission from our own Galaxy by telescopes designed to observe the cosmic microwave background (COBE and WMAP) [2, 3, 4, 5]. Despite these great advances we still do not have good models of the physical properties of cosmic dust, there are serious flaws in our models of how dust is produced, how it is destroyed and how it influences other physical processes in the interstellar medium. However, there is now the huge potential for this to change dramatically as new data from the Herschel Space Observatory and Planck satellite is becoming available.

The Herschel Space Observatory was launched in 2009. It was revolutionary in this field because at 3.5m its collecting mirror is 5-6 times larger than any of the previous far-infrared telescopes, giving both improved sensitivity and spatial resolution. In addition, one of the major discoveries from previous far-infrared missions was that the cosmic dust was somewhat colder than expected, so the Herschel instruments were designed to look at a previously un-explored part of the electromagnetic spectrum between the far-infrared and sub-mm (250-500μm) as well as the regions previously explored with smaller telescopes (70-160μm)1 . Importantly, this wavelength coverage can now be greatly extended to longer wavelengths (out to 1382μm) with observations from the Planck satellite that was launched at the same time as Herschel. Planck was primarily designed to study the cosmic microwave background, but it will also observe many nearby galaxies.

These recent mid-infrared to sub-mm observations of cosmic dust and their interpretation are important for four primary reasons. Firstly, as a depository of metals the dust content of a galaxy at face value is a measure of how far along the evolutionary path a galaxy has progressed. Secondly, cosmic dust plays an important role in many of the physical processes that regulate the evolution of galaxies. For example, it provides opacity so that giant clouds of gas collapsing under gravity can heat up to temperatures sufficient for stars to form and nucleosynthesis to start. It is on the surface of dust grains that molecular hydrogen, which is the crucial gaseous ingredient for star formation, forms. Thirdly, dust traces other physical processes and galaxy constituents that are not so easily measured. For example far infrared emission from galaxies is closely related to the rate at which stars form and the relatively easily measured mass of dust relates closely to the difficult to measure mass of molecular hydrogen. Fourthly, dust can greatly affect what you measure at other wavelengths. The ultra-violet emission of hot young stars for example is greatly attenuated by dust and may lay hidden and the reddening effect can mislead us in our determination of the ages of stellar populations. On cosmological scales the far infrared background can be used as a measure of the star formation history of the Universe. In addition, dust extinction through the Universe may noticeably influence our observations of the most distant objects. It is problems related to and developed from the above that we plan to investigate in this project.

The DustPedia team have considerable previous experience and expertise in using Herschel/Planck data and extracting world class science from it. We lead three substantial projects ‘The Herschel Virgo Cluster Survey’ (HeViCS) [9], ‘The Herschel Fornax Cluster Survey’ (HeFoCS) [10] and ‘The Herschel Dwarf Galaxy Survey’ [11] and are CoIs on many other Herschel surveys (H-ATLAS [93], HRS [12], HELGA [12a], VNGS, NHEMESES [13], HEROES, FRIEDL). In this project we plan to exploit, along with other data (for example WISE, and Spitzer), the extensive Herschel Space Observatory and Planck telescope data archives to make a definitive study (DustPedia) of cosmic dust in galaxies in the nearby Universe.

1. The two Herschel instruments that are the main focus of this project are the Spectral and Photometric Imaging Receiver (SPIRE), which operates at 250, 350 and 500μm and the Photo-detector Array Camera & Spectrometer (PACS) which operates between 70 and 160μm.

## Objectives

Many observations of galaxies over large look back times correspond very well with cosmological models of galaxy and larger scale structure formation [14]. This has led to the wide spread belief that the current cosmological model is broadly correct. Although there is good agreement over large spatial scales there are some challenging disagreements between theory and observation when one looks over smaller scales and particularly at the properties of nearby galaxies. The distributions of galaxy mass and size, their locations within larger scale structures in the Universe and their star formation histories as a function of galactic mass are all examples of disparity with the currently favoured model. In a recent Nature article Peebles and Nusser [14] stress the importance of nearby galaxies if we want to understand the detailed processes of galaxy evolution and hence develop a complete model of how galaxies change with time. They specifically say that ‘….nearby galaxies offer rich and still far from completely explored clues to a better picture of how galaxies form.’. The reason of course is that nearby galaxies can be studied in far greater detail than those that lie at the edge of the Cosmos and importantly that in general cosmological surveys cover such small areas of sky that they in fact do not sample the local population at all. Observations of cosmic dust address many aspects of the current galaxy evolutionary model i.e. star formation rate, growth of the metal abundance, loss of metals in galactic winds, physical processes in the interstellar medium etc. and so offer the potential for a much better understanding. In this proposal we intend to address six specific but quite broad science issues that have a direct bearing on current evolutionary models of galaxies.

Our objectives are:

1. Measure the complete UV-mm spectral energy distribution (SEDs) for a large number (> 1000) of galaxies, and for different environments within individual galaxies.
2. Interpret the galaxy SEDs using radiative transfer and full SED models, to derive stellar, gas and dust properties, star formation rates and histories as a function of morphological type.
3. Determine how the dust NIR-mm/radio SED evolves throughout the Universe and how this is related to the underlying dust properties?
4. Develop a dust evolution model that is consistent with the SEDs of galaxies of different morphological types and determine the primary sources and sinks for cosmic dust.
5. Derive dust mass functions to the lowest-possible luminosities and masses and to compare these with cosmological surveys and the cosmic far infrared background.

## Sample and data reduction

Our proposed programme of work combines the all sky Wide-Field Infrared Survey Explorer (WISE) data [17] with the state-of-the-art Herschel and Planck data so that we can make use of a near infrared (stellar mass) selected sample of galaxies. Our programme is unique because we will combine WISE all sky data with data from a large number of separate Herschel/Planck projects to produce a single unique data set, the whole being greater than its parts. All the data will be reduced using our current state-of-the-art data reduction pipelines that have been optimised for the analysis of either point or extended sources. They will all have a uniform calibration applied. This complete near infrared selected sample will provide unique data on local galaxies that will be used in other research projects.

#### Sample selection

To achieve the objectives set out above we have defined a representative sample of galaxies to study. The Herschel data alone does not provide us with a large completely sampled region of sky, but we can use the WISE data to define a near infrared selected sample. As many galaxy properties correlate first with stellar mass, the near infrared, where for most galaxies the bulk of the stellar radiation emerges, provides the best available choice for galaxy selection. There are 3045 galaxies with a WISE 3.4μm detection (SN>3) within 3000 km s-1 (≈40 Mpc) of the Sun and with a size of D25>1 arc min (≈12 kpc at 40 Mpc) – the W sample. Currently within the Herschel archive there are 613 observations of galaxies from this W sample. The intersection of these two samples we will refer to as the WH sample. We will carry out a similar selection for galaxies detected by Planck once the final data becomes available in late 2013 and the intersection of this sample with the WH sample is the WHP sample. The WH sample consists of 80 early type galaxies (T< 0) and 533 late types, 70 of which may be described as star forming irregulars/dwarfs. Also, within the WH sample we have 94 galaxies that have D25> 5 arc min and so are resolved in the Herschel bands and lend themselves to full radiative transfer and infrared/mm SED modelling (see below).

#### Herschel and Planck data reduction

As part of our previous work on Herschel data we have developed specialist data reduction routines that are optimised to the production of background subtracted and calibrated images of extended sources (BRIGADE, [66]). We can now apply these methods to all the observations of nearby galaxies in the WH and WHP databases to produce a large sample of galaxies with detailed spectral energy distributions in the infrared and sub-mm. We will make careful checks of the Herschel/Planck calibrations and carry out the necessary photometry to make accurate measurements of these galaxies in all Herschel and Planck photometric bands creating a ‘uniform’ sample. Using the intersection of the WISE, Herschel and Planck samples, along with IRAS data, we will look for correlations that will enable us to predict the Herschel/Planck flux densities for those galaxies that are in the WISE sample but not observed, or detected by Herschel/Planck. These correlations will probably be a function of environment and morphology. We will then have a complete stellar mass selected sample of local galaxies with observations across the near and far infrared/sub-mm spectral range. This is our state-of-the-art sample of local galaxies – this sample will form the basis of the work specified in this project. Some gross properties of the sample are shown in Fig. 1.

a) Velocities.

b) B-band magnitude.

c) Diameter

d) Hubble T types.

Fig. 1. Gross properties of the W (black, 3045 galaxies) and WH (red, 613 galaxies) samples. a) Velocities. b) B-band magnitude. c) Diameter d) Hubble T types. From the WH sample there are 94 resolved galaxies with D25>5 arc min and 80 early types with T<0.

## Ancillary data

As well as far infrared/sub-mm data, our programme requires data taken at other wavelengths, this is particularly true for the radiative transfer/SED modelling. There are now vast amounts of data available through databases like the NASA Extra-galactic Database (NED) and the European Virtual Observatory. We will make use of: GALEX UV data, particularly that taken as part of GUViCS an already completed GALEX programme to observe the same region of sky as HeViCS (Davies CoI); SDSS provides in the optical extensive uniform coverage in the northern hemisphere, this is soon to be complemented with the VST ATLAS survey in the south. For the northern targets we will also make use of the 2.3m "Aristarchos" telescope, operated by CoI institute NOA, for any complementary broadband and narrowband optical imaging that may be needed. In the NIR 2MASS will detect most if not all of these far infrared bright local galaxies and where necessary this can be supplemented with UKIDSS, the VISTA hemi-sphere survey (Davies CoI) and the longer wavelength WISE data. All of the galaxies in the WH sample also have 21cm observations of atomic gas. CO and hence molecular gas data is somewhat more limited and will come from us (≈100 galaxies) and where possible from the literature (correlations with other data may allow us to predict molecular gas masses). We have just observed 24 Fornax galaxies in CO using MOPRA (Smith PI, Davies CoI) and have another 80 observed as part of other projects. Our programme of research may lead to applications for substantial amounts of time on telescopes that provide data at wavelengths beyond Herschel and that have better sensitivity and/or spatial resolution than Planck i.e. ALMA, APEX, JCMT, and SMA. All of the dwarfs in our sample observed spectroscopically by Herschel have measures of metallicity and we have recently obtained metallicity measurements for 260 WH galaxies. We will perform more metallicity observations using the facilities at Skinakas Observatory in Crete where members of our team have guaranteed time (at least two weeks/year). We will also make use of the XMM-Newton and Chandra archives where observations of ≈40 early-type galaxies in the WH sample exist.

## Summary of Data Sets

1. The W sample consists of 3045 galaxies selected by their WISE 3.4μm flux, a D25>>1 arcmin, and a v< 3000 km s-1.
2. The WH sample consists of 613 galaxies from the W sample that have also been observed by Herschel.
3. The WHP sample consists of an as yet un-determined number of galaxies selected from the WH sample that have also been detected by Planck.
4. The spiral galaxy sample consists of 463 galaxies selected from the WH sample with types later than Sa (T≥1).
5. The resolved galaxy sample consists of 94 galaxies selected from the WH sample with diameter of D25>5 arc min.
6. The dwarf galaxy sample consists of about 70 galaxies selected from the WH sample.
7. The early type galaxy sub-sample consists of 80 galaxies selected from the WH sample that have types earlier than Sa (T≤0).

## Spectral Energy Distribution (SED) fitting

Many studies measure the integrated properties of dust within galaxies using a modified blackbody approach to fit the far infrared/sub-mm observations. This is straight-forward since there are only 3 free parameters to fit: dust temperature, T, dust abundance, N, and an adopted power-law index, β, for the dust opacity. Often β = 2 is used for galaxies, following the description of the Galaxy: silicate composition or silicate and carbonaceous grain mixture [73]. However, this value of β is likely to vary with environment and there are known degeneracies that lead to noise-induced correlations between physical parameters, the most famous one being the T-β anti-correlation seen in χ2 least squares SED fits. Shetty et al [74] have demonstrated the pitfalls in this approach and the possibly artificial anti-correlations that can emerge. Uncertainties in the data, as low as 10%, can produce false results and lead to biases in interpretation. Recently, Kelly et al [75] have shown how a Bayesian method does help remove these degeneracies.

To go beyond the model limitations intrinsic to fitting only the FIR-submm SED, empirical approaches (e.g. Draine et al. [80], Galliano et al. [47]; see Fig. 2) take advantage of the staggering potential of databases, such as that which DustPedia will produce. These models use realistic grain cross-sections and size distributions. When applied globally, these models use the shape of the observed SED to constrain the distribution of starlight intensity in the studied region, without any assumption on the actual relative distribution of stars and matter (Fig. 2).

Fig 2. An example of the multiphase SED modelling proposed in DustPedia. The data is fitted using the model of Galliano et al. [47, 48]. This example, applied to M82, uses an extensive range of data including Spitzer, IRAS, ISO, WISE, 2MASS (J, H, and K) and Herschel.

Our team has developed a state of the art SED model, widely used to explain the observed UV to sub-mm observations of local galaxies (e.g. [47], [48]). The Galliano SED model is numerically fast and statistically robust. It can be run on large maps or on a large sample of galaxies, together with Monte-Carlo error propagation [47]. We are currently extending the Bayesian hierarchical modelling presented by Kelly et al. [75] to our dust model [76]. We will model DustPedia galaxies, looking for correlations between gas-to-dust mass ratio, metallicity, gas density, PAH fraction, dark gas fraction, sub-mm excess, etc. in order to provide reliable, non-biased and self-consistent observational benchmarks as constraints on dust evolution models.

We will model the DustPedia data incorporating the new dust models of Jones et al. ([70], [71], [72]). These use recently published EUV-cm optical property data and constraints on the evolution of hydrogenated amorphous carbon dust and PAHs ([70], [71], [72], [81], [82], [63]), to allow us to constrain the dust physical property inputs to the full SED and radiative transport (see below) modelling (from UV to mm wavelengths) for a wide range of galactic environments. The new carbonaceous dust data, coupled to that of amorphous silicates, appears to be qualitatively consistent with many dust observables, including: the FUV extinction rise, the UV extinction bump, the NIR-FIR extinction, the IR emission bands, the MIR-mm dust emission and the 3.4 micron aliphatic CH absorption band towards the Galactic Centre. All of these dust observables arise naturally from the new dust model data and are tightly-coupled in a completely self-consistent way, across the entire wavelength range of interest, and allow for very few truly free parameters. Once the dust model size distributions and dust compositions are calibrated on the Milky Way data (in Work Package 6) including spatial and spectral variations, the dust evolution model can be used to explore dust variations within and across galaxies.

Due to today’s faster and more efficient computing capabilities, DustPedia will take the Galliano SED and new dust model to the next step in order to study spatially well-resolved galaxies and to generate galactic SED building blocks. The observed dust SED of a galaxy is effectively the superposition of various dust components associated with HII regions, molecular clouds and neutral atomic media. Thus, a physical model of a galaxy’s SED can be thought of an ensemble of these building blocks.

Our innovative modelling approach begins with the characterization of the building block SEDs, which are chosen to be representative of the dust phases over a broad spectrum of the parameter space (infrared luminosity, star formation activity, metallicity). This step will use a Principal Component Analysis (PCA) to decompose the SEDs of the nearest galaxies into dust components associated with their atomic, molecular and ionized gas phases of the galaxy. We can then quantify how the energy is distributed in an ensemble of galactic dust components. This idea has previously been applied to our Galaxy [3], [83]. The nearest galaxies, affording the best spatial resolution for this first step would be the LMC, SMC, M31, M33, IC10 and other Local Group galaxies, which span an extensive parameter space. Having generated the characteristic SED building blocks, spanning a wide range of physical parameter space, we can then provide the community with unique and valuable analytical tools, a complete grid of SED sub-components over a wide range of extragalactic conditions: metallicities, grain properties, stellar properties/radiation field, geometry, etc.

From this library of SED building blocks we can solve for the overall galaxy SED. To build a galaxy with these characteristic elements, it will be necessary to efficiently explore the model parameter space, which is extremely time consuming. Full-scale solutions will not be possible for a large number of galaxies and so one of the greatest challenges will be to efficiently build such grids. Significant computing resources will therefore be required to generate the grids, and to exploit them efficiently. We will therefore develop and implement "smart" SED-fitting algorithms for galaxies beyond those nearest to us. This requires developing built-in knowledge to determine what parts of the parameter space are incompatible with the observations. Such an approach will help to constrain the large number of parameters and reduce the problem to an ensemble of models that are compatible with the observations.

SED fitting (over the optical to sub-mm part of the spectrum), and in particular spatially resolved pixel-by-pixel SED fitting based on an energy balance technique, is an important step towards understanding the complex interplay between stars and dust in galaxies. This is an approach we will take by necessity for the galaxies in our sample that are unresolved, but this approach does have a number of limitations. In particular, all images must be convolved to the same resolution (the resolution of the band with the coarsest resolution), which effectively implies a degrading of the available information. Moreover, such an analysis is per definition local, whereas non-local radiative transfer effects can play an important role. We therefore intend to apply full panchromatic radiative transfer modelling to this new data set (94 resolved galaxies with D25> 5 arc min). The goal of such a modelling approach is to simultaneously derive the 3D distribution and the spectral properties of the stellar populations and the interstellar dust in a galaxy, fully taking into account the effects of absorption, multiple scattering and thermal emission.

Our team has extensive experience of making and using radiative transfer models of galaxies [18, 19, 20, 21, 22, 23]. We have previously developed a Monte Carlo radiative transfer code called SKIRT [21], which we now intend to use to model these new observations of galaxies. Up until now this code, and others like it, have been used almost exclusively to model edge-on spiral galaxies [22, 23], which have the advantage that the extinction features and emission from dust can be easily distinguished along the line of sight. Now with the development of faster computers and efficient codes along with multi-wavelength data we can apply these models to different classes of galaxies with arbitrary inclinations, Fig. 3. The SKIRT code includes the ability to specify the stellar population mix, the physical properties of the dust and to vary these over different spatial scales and with different mass densities. It produces simulated images of galaxies in any desired wave band with an arbitrary resolution, which has the advantage that we can compare models with data at the best resolution at every wavelength. We have recently developed a genetic-algorithm based fitting code, FitSKIRT, designed to automatically fit radiative transfer models to a set of images [94].

As input to the numerical model we will need to assemble a large database of observations (images) from the UV to the FIR/sub-mm. Although there is not uniform coverage over the whole sky there is sufficient data on these nearby galaxies to carry out this challenging exercise. In particular, most of the galaxies in our resolved sample are part of the S4G survey, and hence have deep IRAC observations (and mass models for the stellar populations) available.

The first objective of this radiative transfer modelling is to investigate the dust energy balance in galaxies of different morphological types [24, 25]. In a radiative transfer modelling approach, the dust is simultaneously studied in two complementary ways: through the extinction it causes in the UV and optical and through its direct thermal emission at FIR/sub-mm wavelengths – the energy removed and emitted must be equal. When these models have been applied to edge-on galaxies there has been an inconsistency in that the optical extinction generally underestimates the observed far infrared/sub-mm emission by about a factor of 3-4 [22, 25]. Two broadly different scenarios have been proposed to explain this difference: either there has been a significant underestimate of the dust emissivity or that the dust is significantly clumped so that it has a small effect on the extinction of the bulk of the starlight. In the past, discriminating between these two scenarios has been a problem because of the limited number of edge-on galaxies that could be studied and because of the lack of observations beyond 100 μm that measure the bulk of the emission from the dust – these problems will all be solved using the data and methods we now have to hand.

Fig 3: Left: a radiative transfer model for the Sombrero galaxy, fitted to an optical V-band image [21, 26]. Right: three-colour image of M81. The top panel is the real image (red=250μm, green=3.6μm, blue=0.15μm), the bottom panel represents a radiative transfer model fit to these data. The model makes use of combinations of stellar templates from standard model libraries to simulate the optical/NIR emission from the stars, which can be distributed in a bulge and disc. Dust is also distributed within a disc separate from the stars. A spiral density wave can be imprinted on the stellar and dust disc and the dust can be ‘clumped’ over various scales [95].

Our second objective is to address the dust heating in galaxies of different morphological types. Different sources can contribute to the heating of the dust in a galaxy, including the diffuse interstellar radiation field from evolved stars, young stars in star formation regions, and/or an active galactic nucleus. Current studies of dust heating are generally based upon correlations between properties of the dust (e.g. far infrared flux, colour ratios) and tracers of the interstellar radiation field (e.g. K band emission) and star formation tracers (e.g. Hα or far UV emission), and cannot take into account the non-local character of the heating. For example we have recently shown that in the Sombrero galaxy (M104) the dust is primarily heated by evolved stars in the massive bulge – these are stars not spatially coincident with the dust (Fig. 3). The main heating source of the dust as well as its dependence on location (i.e. bulge, disc, spiral arm), can be derived at each point in the galaxy, via our Mont Carlo photon tracing model [26]. These results will be used to determine the accuracy of traditional estimators of the star formation rate based on for example, far infrared luminosity. Total far infrared luminosities are often used to estimate star formation rates in high redshift galaxies [27].

For galaxies that are unresolved by Herschel we will carry out a simpler but similar global energy balance (SED decomposition) [28]. These SED decompositions will make use of standard stellar libraries to model the stars and the far infrared/sub-mm data and model templates (see § SED fitting) to model the dust emission. The energy balance method then provides a means of comparing the extinction, as predicted by that observed for the Milky Way, with the predicted far infrared emission. We will then be able to look for changes in dust properties and resulting dust emissivity in galaxies of different morphological types and within different environments. We will also be able to compare results from radiative transfer modelling of the resolved galaxies to global values to judge the accuracy of these methods.

As well as addressing the specific problems discussed above, the model (radiative transfer and global SED fitting) will output considerable data on the properties of the sample galaxies that will be used in other parts of this proposal and by other scientists. The output from the model will include: stellar masses, stellar population mix, stellar population gradients, star formation rates, specific star formation rate (when combined with the gas mass), temperature of the interstellar radiation field, temperature gradients, extinction and extinction gradient, and other galaxy scaling relations (mass, size, surface density).

## A global dust model

In many ways the culmination of our work on the radiative transfer modelling of galaxies and the detailed analysis of their spectral energy distributions is the modelling of the physical processes operating on dust in the interstellar medium, in galactic halos and in the intergalactic medium between and in the vicinity of galaxies [41, 42]. We expect to find significant differences between the far infrared/sub-mm spectral energy distributions of disc, ellipsoidal and dwarf galaxies that are related to their current evolutionary state and to the compositional and size evolution of their constituent dust populations. Early type galaxies are rather quiescent but have a hot, tenuous ISM (similar to the IGM), dwarf galaxies an ISM often dominated by star-formation and disc galaxies lie somewhere in between. Our objective is to find a single unifying dust model that weaves the observations of these disparate (extra-) galactic media together into a consistent picture of dust formation, evolution and destruction [56, 57]. The model will, principally, consider the three key and distinct stages in the life of dust:

1. Dust sources – we will use our data to model the contributions of the major sources of amorphous silicate and amorphous carbonaceous dust, i.e. evolved stars and supernova (their dust mass production per unit time). In addition we will consider the, seemingly inescapable requirement, that a significant fraction of ISM dust must be formed in dense, molecular media by the (re-)accretion of atomic species eroded from dust in regions shocked to high velocities (v > 50 km/s). This will be an individual prescription for each galaxy derived from the radiative transfer modelling that provides stellar populations and star formation rates, and will rely on the derived initial mass functions and the stellar and star-formation histories of each galaxy. We will then be able to assess and compare the expected dust input rates from the evolved stellar population and supernova with that measured to address the important issue of the origin of dust and whether known sources of dust production can account for the dust observed. This will be based on previous work, undertaken by us and by other groups, on the rate at which evolved stars and supernova produce dust.
2. Dust evolution – we will investigate the utility of existing, viable dust models and the likely dust optical properties in the interpretation of observational data [67, 68, 69, 70]. We will, in parallel, re-evaluate the dust model input physics using the recently derived optical properties for amorphous hydrocarbon and silicate dust analogues, e.g. optEC_(s)(a) and a-Sil_Fe [70, 71, 72]. The nature of the dust and its affect on the SED will be investigated with our existing dust SED modelling tool [69]. This tool allows us to quantify the dust extinction and emission, as a function of the input dust physical and optical properties, which can be varied at will. We can compare the SEDs arising from the various dust models mentioned above in order to explore the available dust model parameter space. The aim here is to construct a dust model that is consistent with the various galactic types (elliptical, spiral, starburst, dwarf, ...) and with the expected spectral and spatial evolution of the dust populations within these galaxies. This will then feed back into our fully self-consistent radiative transfer models to predict the extinction, the observed far infrared/sub-mm emission and ultimately the dust mass in galaxies. These models will provide a crucial link between the physical properties of the grains (chemical composition, size distribution, particle structure, temperature) and the observed SEDs of galaxies of different morphological types. An important aspect of this work will be a prediction of the wavelength-dependent emissivity function(s) that needs to be used when modelling galaxies of different morphological types.
3. Dust sinks – dust is removed from the ISM principally through the effects of supernova-generated shock waves that erode it into its atomic constituents. However, it is now becoming clear that intense UV radiation fields, such as in photo-dissociation regions, can significantly affect the carbonaceous dust population. Thus, our models will include dust processing arising from; sputtering, grain-grain collisions, electron, ion and UV irradiation in intense radiation fields, supernova shocks and in hot gas. We will develop and adapt our existing dust evolutionary modelling tools, GRASHex and optEC_(s)(a), which already include dust destruction and evolution as a result of energetic ion, electron and UV irradiation as a function of the given environment (shocked gas, X-ray emitting hot gas, PDRs etc). In conjunction with our new ISM dust model, we will quantify the dust destruction rate (dust mass loss per unit time) for given galaxy types. In addition, we will assess the importance of the environment-dependent process of ram pressure stripping and how the gas and dust are stripped, how the stripped dust responds to the stripping process, how the dust evolves and how long it can survive in the hot tenuous inter-galactic medium.

## Connection between dust emissivity, metallicity, gas, and star formation

The fundamental input into models of dust grain sizes and composition is the shape and total energy content of the far infrared/sub-mm spectral energy distribution. We will use the WH and WHP samples to study the emissivity of dust, which provides the link between observation and the physical properties of cosmic dust grains [30, 31, 32, 33 and 34]. In a galaxy, most of the dust mass emits thermal radiation beyond 100μm and thus Herschel and Planck observations are fundamental for its derivation. In this region of the spectrum the emission is optically thin and only weakly dependent on the temperature. The shape of the SED, however, is not that of the ideal blackbody curve, but it is modified by the poorly known emissivity function. The dust emissivity function is normally written as ; $$\kappa_{\lambda} = \kappa_{0}(\left(\frac{\lambda_{0}}{\lambda}\right)^{\beta}$$ λ 0 is a normalising wavelength, κ0 is the mass emission coefficient and β is the power law index that governs how the emissivity changes with wavelength. The power law index can be derived directly from the spectrum, in a rather straightforward manner. We have already identified systematic changes in β across the largest extra-galactic source in the sky M31 [65], with values peaking (β≈2.5) in the 3 kpc star formation ring. High values of β indicate larger grains, ice mantles and/or very cold dust. Lower values (β ≈1.5) in both the outer and inner regions of M31 indicate freshly formed dust and/or smaller grains. The availability of Planck data also allows us to identify possible (non-dust) contaminating emission (free-free and synchrotron) from sources other than dust, which for a few types of galaxies could contribute to the emission at longer wavelengths.

The value of κλ in the far infrared/sub-mm region is poorly known and, as mentioned above, there is a fundamental problem relating dust mass inferred from extinction with that measured by its emission. There is little or nothing known about how κλ may change within galaxies of different types or in different environments. κλ Is notoriously difficult to measure absolutely because of the difficulty of going from what you measure (extinction or emission) to a value for the dust mass. Fundamentally measurements of dust mass rely on an assumed dust-to-gas mass ratio, measurements of gas phase element depletions into dust or laboratory measurements that relate extinction to mass. We will carry out a number of new investigations into these fundamental measurements that underpin the dust emissivity:

1. A promising method for studying the dust emissivity is to obtain the dust mass from the gas phase mass and metallicity and then use this to predict the far infrared/sub-mm flux densities. We have previously carried out this analysis using 22 galaxies for which there are both gas mass, metallicity and SCUBA 850μm observations available [35]. This gives an 850μm dust emissivity very similar to what has been derived for the dust in the Milky Way. The critical assumption when using this method is that a fixed fraction of metals are contained within cosmic dust, an observation that seems to be confirmed using UV spectroscopy in our Galaxy. We have recently completed spectroscopic observations to determine the metallicity of 260 galaxies observed by Herschel [36]. This sample is ideal for investigating the relationship between metallicity and dust mass over a wide range of galaxy types. If this fixed dust fraction of metals is confirmed in other galaxies we will be able to derive the dust emissivity for this large sample over all the Herschel and Planck bands. In addition the combination of metallicity and dust mass will provide the first comprehensive measure of the total amount of metals in a large sample of galaxies. This is an important parameter in models of the chemical evolution of galaxies. For the larger objects in our sample, resolved observations of gas density and metallicity are either available in the literature, or will become available soon when various observing campaigns, including those carried out by our team, will be completed. We will thus have the unprecedented possibility of studying emissivity variations in and across many objects.
2. An alternative approach is to use our detailed SED modelling and radiative transfer models (§ SED fitting, § radiative transfer modelling). Previously we have modelled optical and near infrared observations of a small number of edge-on galaxies to derive the extinction [37], the extinction has then been related to a gas column density and finally to a dust mass by assuming relations derived from our Galaxy (for example AV/NH and a constant gas-to-dust ratio). Assuming an energy balance, so that energy absorbed equals that emitted in the far infrared/sub-mm, the emissivity and the calculated dust mass produce far infrared/sub-mm flux densities, which fall short of those measured. As mentioned above possible solutions are that the emissivity is incorrect (Fig. 4) or that the dust is clumped on small scales so that the dust causing extinction is not the same as that which is emitting. Previously this work was limited because there were so few observations where extinction features could be matched with emission measurements (primarily SCUBA 850 μm observations of a few nearby edge on galaxies). We now have Herschel resolved observations of nearby galaxies in which we can measure the extinction, assess dust clumping and relate this to the emission at not just one but a number of far infrared/sub-mm wavelengths. We will also be able to study this relation between dust extinction and emission within individual galaxies and in galaxies of different morphology. Hence we will carry out the most extensive study of cosmic dust emissivity .

Fig 4. illustrates previous work by us on the dust emissivity in the far infrared [31]. It is a compilation of previously measured values with our values marked by filled circles. The dashed and dotted lines correspond to a wavelength dependency for the emissivity which varies as β=2.0 and β=1.5. The lines have been set to pass arbitrarily through the frequently-cited model value of [73] at 250µm.

3. A related quantity to the dust emissivity is the dust emissivity per hydrogen atom. This quantity is often measured and quoted for our Galaxy [39]. For direct comparisons we require the surface density of hydrogen gas. All of our sample galaxies have either global or resolved 21cm observations of atomic hydrogen, but we currently only have limited observations of molecular gas (CO) for about 100 galaxies. We will search for correlations within the data to help us predict molecular gas masses and so in combination with the atomic gas derive the dust emissivity per hydrogen atom.
4. Finally, our measures of dust emissivity across the far infrared/sub-mm can be compared with those measured from laboratory experiments and predicted from theoretical models [41, 42 and references therein]. Thus our data will provide primary input into our revised models of cosmic dust and lead to new insights into the physical processes that produce, evolve and destroy the dust (see § A global dust model).

## Spiral galaxies

The determination of the dust content and distribution in spiral galaxies and the study of its effects on their observable properties has a rich and controversial history. Despite the recent progress in the field (to which all members of our team have contributed significantly), the physical properties of dust (structure, composition and optical properties), its total mass and spatial distribution, and its detailed effects on the observable properties of galaxies are still poorly constrained. A better knowledge is crucial both to understand the evolution of the cold interstellar medium and to translate the observed properties of galaxies into physically meaningful quantities such as intrinsic luminosities, stellar spectral energy distributions and star formation rates.

During the past few years, members of our team have made considerable steps forward in modelling the dusty medium in spiral galaxies using radiative transfer models ([19], [20], [21], [22], [23], [24], [26). Among the results, dust disks were found to be spatially thin and radially extended compared to the stellar distribution and only moderately opaque (at least when seen face-on). As described above a strong discrepancy was found though between the predicted and the observed FIR/sub-mm fluxes (e.g. [84], [19], [85], [26]). The optically determined dust mass is about a factor three less than the dust mass determined from the FIR spectrum.

Our new radiative transfer models of galaxies of all inclinations will enable us to investigate more fully the two competing solutions to this energy budget crisis - inhomogeneity in the dust distribution or changes in the dust emissivity [19], [24], [86]. Preliminary results of our 3D clumpy Monte Carlo simulations indicate that a clumpy dust medium can help to restore the energy balance, but it is yet unclear whether it can generate the necessary dust mass to reconcile the independently derived optically and FIR determined values ([19]). The other scenario has been advocated by Alton et al. [87] and Dasyra et al. [85] i.e. that the energy balance in spiral galaxies can be restored if the dust emissivity at sub-mm wavelengths is about a factor four higher than that for diffuse Galactic dust. Indeed, high emissivities are found in models and observations of dust in Galactic dense clouds. However, it is not clear why cold dust (thought to be heated by the diffuse interstellar radiation field) should have properties typical of dense environments. However, dust properties throughout galaxies with widely-varying physical properties have not been studied to the greatest depth possible and in a systematic way using all the data available – something that DustPedia will be able to do.

Here we propose to analyse the resolved spiral galaxies of the WHP sample in the framework of our sophisticated radiative transfer and SED models as described in sections § SED fitting and § radiative transfer modelling. The final result will be, for each galaxy in our sample, an accurate model of the intrinsic stellar and dust distributions, including large-scale spiral structure and strongly clumped ISM. Our study will be the first test of the energy balance in a sizable sample of spiral galaxies and will describe in detail the dust properties of different galactic phases.

Due to a homogeneous analysis of all the resolved spiral galaxies in our sample, we will be able to statistically study the energy balance and the properties of the interstellar dust medium (optical depth, dust mass, dust/stars scale length ratio, dust/stars scale height ratio etc) as a function of basic galaxy properties (Hubble type, luminosity, circular velocity). Combining our WHP observations with HI and CO data, we will study the variation in the gas-to-dust ratio, both as a function of radius within a galaxy and as a function of basic galaxy properties.

## Early type galaxies

Pinpointing the place of early type galaxies in galaxy evolution models remains an outstanding astronomical issue [52]. One cosmological scenario has them forming early in the universe, leaving them as quiescent objects, evolving passively into old age. Other evidence supports the idea that they have formed from more recent galaxy-galaxy merger events and have some current star formation activity. Although early type galaxies were once thought to be devoid of gas and dust, previous far infrared and sub-mm observations have indicated dust masses some 10 - 100 times larger than those derived from the optical extinction alone [53, 54]. This implies that there must be a more diffuse component of cool dust present or again possibly that the far infrared emissivity maybe incorrect. Earlier mid-infrared observations with ISOCAM placed important constraints on the small end of the dust size distribution [55]. Many early types have excess mid-infrared emission that can be attributed to PAHs and very small grains. Recent Herschel observations have produced evidence for the presence of larger colder dust grains in some early types (see Fig. 5) [33]. However, to date conclusive experiments on the range of physical dust properties have been lacking because of the small samples with full spectral coverage from the UV to the far infrared/sub-mm. Determination of the physical properties and the quantity of dust in local early type galaxies, as well as its distribution relative to the gas and the stellar components, provide clues to its origin/fate and a link to the evolutionary history of the galaxy.

Fig 5. Optical (SDSS gri three colour, left) and far infrared (Herschel 250μm, right) images of two early type galaxies taken from our guaranteed time HRS project [33].

We can use the early type galaxy WH sub-sample (≈80 galaxies) to measure the dust abundance, its emission spectrum and trace its spatial distribution. This analysis will allow us to characterise the physical properties of the dust residing in early-type galaxies and to investigate its evolution and survival in the harsh interstellar environment. Specifically, we will address the following questions: i) What is the total inventory of dust in early-type galaxies, what is the fraction of young stars (hence the amount of star formation) in these galaxies and their contribution to dust heating and how does this reconcile with dust evolution models?

The source of the dust in early type galaxies is a subject of much debate, with two major competing hypotheses [33]:

1. Dust is produced by evolved stars: Over the course of their evolution, high redshift proto-elliptical galaxies process their gas and dust reservoirs into stars, polluting the Inter-Galactic Medium with enriched gas. After many Gyrs, the cold gas is exhausted, star formation all but ceases, the stellar populations fade away and the removal of dust from the interstellar medium through star formation ceases. The remaining dust is modified or destroyed via sputtering by the hot X-ray gas. Thus in early type galaxies, the creation of dust in supernovae and the destruction of grains in supernova-induced shocks is no longer taking place. As long as dust is not introduced from elsewhere, we essentially have a stellar system that currently produces dust in the atmospheres of its mass-losing evolved (AGB) stars and destroys it via sputtering in the hot gas. If the timescale for dust destruction is short (such that early type galaxies are now only left with dust produced by evolved stars), we only require the interpretation of the effects of sputtering to determine the elusive properties of dust produced in this way. If true the dust life-cycle is simpler and more easily understood than in an actively star forming galaxy. This is something we will be able to measure with the DustPedia data base and use these measurements to refine our models of dust input and output. If the dust has an internal origin, then its spatial distribution should also follow that of the stars a scenario we can also test using this sample.
2. Dust is accreted during a galaxy merger: To try to identify those galaxies within our sample that may have undergone a recent merger, we will use images of optical obscuration as well as the dust emission because the merging process is expected to leave dust ‘trails’ across the galaxy. DustPedia data provides us with the appropriate resolution and wavelength coverage to look for the association of dust emission with these prominent extinction features. Sputtering timescales are typically 106 to 108 years while the dynamical timescale is greater than 108 years - accreted dust should be sputtered before it is dispersed. Recent mergers should reveal themselves as strong extended dust features and not as the very small dust features often seen close to the centres of early type galaxies. Given the relatively uniform stellar population in early type galaxies one might expect that there would be small scatter in the dust-to-stars mass ratio if the dust was predominantly produced in evolved stars rather than deposited by the more stochastic merging process. Our studies will provide constraints to distinguish between these two scenarios because we can use the UV/optical emission to study the star and dust formation rates (input), the far infrared/sub-mm properties to measure the current dust mass (accumulation) and the x-ray emission to assess the strength of dust destruction via sputtering (output)4.

Given the low level of ongoing star formation early type galaxies can provide a new and unique environment for the study of dust production, evolution and destruction. The properties of the dust can be compared with dust from other environments to give us a better understanding of processes in the interstellar medium of galaxies.

4. There are ≈40 galaxies in the WH early type galaxy subsample that have been observed with the Chandra and XMM space observatories

## Dwarf Galaxies

Only in the local Universe can we detect far infrared emission from dwarf galaxies [43, 44]. Their importance is that they provide a unique opportunity to study star formation and processes in the interstellar medium in galaxies of very low metallicity, similar conditions to those that other galaxies would have experienced during their early history. One of the most interesting conundrums to recently come to light is the behaviour of their dust-to-gas mass ratios as a function of metallicity [45, 46]. On the one hand the high dust-to-gas ratios determined for some of these star forming dwarf galaxies (using the usual graphite dust to explain the emission properties) results in a dust-to-gas ratio too high considering the metallicities and the limit of the amount of metals possible to be incorporated into dust. On the other hand, for metallicity values lower than about 0.1 solar, the dust masses are too low compared to the observed gas mass. For galaxies with dust-to-gas ratios higher than reconcilable, the dust masses may be reduced by replacing the commonly used graphite with other more emissive grains, such as amorphous carbon – a new solution currently suggested for the low metallicity dwarfs [47]. Hence, these galaxies offer the possibility of studying cosmic dust that may be quite different to that in our own Galaxy, thus opening up new avenues of investigation with regard to processes in the interstellar medium. From the WH/WHP samples we can extract a sample (≈70 galaxies) of star forming dwarfs to investigate this problem further. Fitting the spectral energy distribution will provide us with the dust mass and data from the literature and our own correlations will provide us with the gas mass. There are then three questions we can ask:

1. Is this dust mass correct? - Is there evidence within the spectral energy distribution that the constituents and physical make up of the dust is different to that in other galaxies so that the dust emissivity is different? For example, there is evidence that some of these star forming dwarf galaxies show excess emission at 500μm (above that expected from a modified blackbody, Fig. 6) that is not seen in more massive galaxies [43]. We will also combine the Herschel and Planck observations with those at shorter wavelengths from Spitzer and WISE to investigate properties of the grains emitting at MIR wavelengths, where emission from PAHs and very small warm grains dominate [48].
2. How do metals become incorporated into dust? How efficient is this process? Are silicate dust grains and carbon dust incorporated into dust similarly? Is there differential depletion between the different dust components? One of the most fundamentally important measurable quantities in galaxies that can be studied to this end is the behaviour dust-to-gas ratio as a function of a galaxy’s metallicity. DustPedia will contribute to this plaguing issue with the critical low metallicity component. Our progress in dust modelling and the detailed SED modelling in DustPedia will quantify accurately the reservoirs of the elements and the dust budget in low metallicity galaxies – a step only recently possible with the large mid-infrared to mm dust tracers now available. Since gas masses in dwarf galaxies are vastly dominated by the atomic phase, the available HI masses will provide the bulk of the total gas reservoir for this study. Molecular hydrogen is difficult to measure in these low mass galaxies. In these dwarf galaxies, the deficit of CO line emission, which we take for granted as a viable tracer for molecular hydrogen, is reduced by the low metallicity and is also dissociated by the radiation field produced by these actively star forming galaxies. How much molecular gas remains undetected in the low metallicity (low Av) ISM has been difficult to quantify for many galaxies using the observations. To date more than half of the DustPedia low metallicity sample has already been or is currently being observed with ground-based telescopes (e.g. [88], [89]) or with the Spitzer and Herschel spectrometers [90]. Thus, studies using DustPedia with detailed modelling of the gas tracers, as is being attempted for a handful of low metallicity galaxies (e.g., [91], [92]) will quantify and characterise this elusive "dark" molecular component.

3. Fig. 6. Left - The far infrared/sub-mm spectral energy distribution of three dwarf galaxies taken from HeViCs [43]. Right - we model the excess 500μm emission using a cold dust component.

## Cosmological implications

It is probably true to say that we know more about the dust properties of distant galaxies than we do about those local to us. Surveys such as IRAS and those undertaken in the sub-mm using, for example SCUBA and more recently Herschel, identified populations of dust enshrouded galaxies producing most of their stellar radiation in the far infrared/sub-mm [77, 78]. These galaxies must have produced enormous amounts of cosmic dust over rather short periods of time and there is good evidence for an increasing co-moving dust mass density as we look back through cosmic history [60]. Today we observe locally the result of this evolution and to assess just how strong it has been we need to carry out a detailed analysis of the properties of local galaxies.

Our radiative transfer and SED fitting methods will provide us with a wealth of data on the properties of our sample galaxies. We will obtain for each galaxy: a stellar mass, a measure of the mass contained within young and old stars, the distribution of young and old stars (resolved galaxies), a dust mass, a dust distribution (hot and cold components), and dust temperature (spatially varying temperatures for resolved galaxies). These values and their distributions provide the zero redshift benchmark for more distant surveys and galaxy scaling relations for comparison with theoretical models of galaxy formation and evolution. They also provide the data for an in depth study of how both morphology and environment affect both the stars and the dust. As stated above we have a diverse range of morphology within the sample and our survey region covers diverse environments from the Virgo cluster to diffuse groups like Sculpter. So, we will be able to answer questions such as; what is the relationship between the mass of old and young stars as a function of environment and is there evidence for dust as well as gas being stripped from galaxies in dense cluster regions? Using standard calibrations we can use the UV and/or the far infrared to measure the star formation rates and hence the local star formation rate density and how it varies throughout the local volume. There are literature values for atomic hydrogen mass for almost all of the sample galaxies, in some cases resolved, which can be used to give the specific star formation rate and specific star formation rate density of the local Universe. In this way we can carry out the most detailed analysis of the properties of galaxies.

As well as trying to quantify and understand dust/gas/stars in individual galaxies we can also consider the distribution of these properties amongst galaxies and how this relates to broader cosmological issues. The basic questions we will address are:

i) What is the far infrared luminosity density of the local Universe and how does this density compare to that produced at other wavelengths? To answer this question we first need to construct luminosity functions (Fig. 7) [58]. Integration of the luminosity function over flux density gives the total flux density per unit volume in the Universe. The first far infrared luminosity functions were produced from IRAS observations (λ≤100μm) [59] - this has essentially remained the state-of-the-art until the advent of Herschel. The reason for this is that neither ISO or Spitzer observed the large numbers of complete samples of galaxies that IRAS did or surveyed large areas like Herschel and Planck have done. In addition Herschel/Planck observations at wavelengths 200μm<λ<1382μm are effectively unique and sample the region where the majority of the dust is radiating. Our selection from WISE provides a stellar mass selected sample for which we will have measured and predicted Herschel/Planck observations at wavelengths from 70-1382μm. We can measure for the first time the local (within 3000 km s-1) total far-infrared luminosity density of the Universe due to cosmic dust in galaxies. This value can be compared directly with the luminosity density emitted by stars taken from our SED and radiative transfer models. This will directly provide a ‘global’ value for the fraction of radiation absorbed by dust and a ‘typical’ optical depth for stellar photons.

Fig 7. Far infrared luminosity functions for Virgo cluster galaxies. The red dot-dashed line is that measured by IRAS at 100μm for galaxies in all environments [59]. Will the luminosity functions measured over the local volume correspond better with that measured by IRAS?

ii) What is the cosmic dust mass density of the local Universe and how does this density compare with the density of gas and stars? The dust mass function is the function that fits the observations of the numbers of galaxies per unit volume in each interval of dust mass. Integrating this function over mass gives the total dust mass in galaxies per unit volume in the local Universe. Values for the dust mass density over much larger distance scales have previously been done in a statistical way (photometric redshifts) [60], but never in the definitive way we are proposing here and not to the low galactic dust masses that we can achieve. The derived dust mass density can be compared with observations of the stars and gas in these galaxies to give for the first time global dust-to-gas and dust-to-stars ratios for the local Universe. As stated in the introduction the stellar, gas and metals mass fractions are linked through a chemical evolution model of galaxies. We can for the first time apply a chemical evolution model like this to both local galaxies and the local volume as a whole and make inferences about, for example, how much gas and/or metals have been lost by galaxies to the local inter-galactic medium. Our local dust mass density can also be compared with values derived by the cosmological surveys that look at unresolved galaxies at large distances over relatively small areas of sky.

iii) Is the local far infrared luminosity density consistent with the measured far infrared background? If not what evolution in the luminosity density is required? Having made observations of the local luminosity density we can use a cosmological model and integrate along a line of sight to predict the flux density at a given wavelength of the cosmic far infrared background [61]. We will be able to do this for the first time using the longer wavelengths observed by Herschel and Planck. This is important because as part of the Planck data analysis they will make an accurate measurement of the far infrared background, which is one of the many ‘backgrounds’ that need to be subtracted before the cosmic microwave radiation is revealed. This then provides the two crucial parts of a model of how the dust mass in galaxies evolves with time because deviations of the background from the predictions of a model based on the local luminosity density reveal evolution in the luminosity density.

iv) What is the extinction cross section of a galaxy? Recently Menard et al [29] have calculated the extent of dust around galaxies and the implication this has with regard to the dust optical depth of the Universe. Menard et al. did this by considering the extinction of quasars by galaxies. Using our radiative transfer models of resolved galaxies and extrapolating the derived dust density profiles we can repeat this measurement, but now using dust emission rather than optical extinction.

ID Reference