Hi-GAL: The Herschel Infrared Galactic Plane Survey

The SPIRE digital readout electronics impose a limitation on the brightest sources that can be observed for a given offset setting (DC voltage removal) before digitization (SPIRE Instrument Users Manual, 2007). This problem can be alleviated to some extent by choosing a bias setting that gives the largest dynamic range per offset range. Simulations of the effect of bias variation show that setting a bias higher (∼3x) than the predicted nominal value will approximately double the dynamic range for most detectors under the conditions likely to be found in orbit (telescope temperature and emissivity and sky background). The same simulations show that a significant (>10%) fraction of the SPIRE 250 μm array detectors will saturate on sources greater than 500 Jy. The situation is slightly more relaxed for the 350 and 500 μm arrays. We take the upper limit of detectable sources in the SPIRE bands as 500 Jy beam^-1, assuming that a strong source instrument setting is used. This setting is required for all observations of bright regions/sources with SPIRE and is not a special Hi-GAL configuration. The saturation limits for PACS should be around 2000 Jy at nominal bias, which will be used for the Hi-GAL survey.

The 1σ sensitivities provided by HSPOT, for a single Astronomical Observation Request (AOR), are 17.6 and 26.8 mJy in the two PACS 70 and 170 μm bands, and 12.8, 17.6, and 14.9 mJy for the SPIRE bands; co-addition of the orthogonal scanning patterns will provide better figures. These sensitivities result from the adopted scanning strategy designed to maximize redundancy and map fidelity especially for large-scale diffuse structures. However, the limiting factor for the detectability of sources and clouds will likely be cirrus confusion. Estimates based on recent BLAST measurements (Roy et al. 2010) suggest values on the order of 75, 140, and 160 mJy in the 170, 250, and 500 μm Herschel bands for a representative region of the Galactic plane at l = 45°; these values are greater than the detector sensitivities. Figure 4 shows the expected flux from a 20 M_⊙ envelope as a function of distance (in kpc) in each of the three above-mentioned bands, and for three different dust temperatures; we adopted β = 2 and the dust opacity from Preibisch et al. (1993). The horizontal dashed lines (color coded with wavelength; see online article for color figures) represent the predicted confusion noises based on BLAST images. Figure 4 shows that we will detect the representative 20 M_⊙ core everywhere in the Galaxy except for very cold dust (T ≤ 10 K), for which detectability is predicted to be limited within a distance of about 5 kpc. We may then conclude that cirrus confusion is not going to be a problem for the investigations of the intermediate and high-mass star-formation studies that are the "core science" of this project (see § 3.3 and 3.4).

Zoom In Zoom Out Reset image size

The diffraction-limited instrument beams at all wavelengths can be used to translate the confusion noises reported in the previous section into brightness units to investigate the detectability limits expected for diffuse extended structures. Calculations of the expected brightness levels from optically thin dust as a function of temperature and H column density (assuming gas/dust = 100) are reported in Figure 5a and 5b for λ = 170 and 500 μm, respectively, where the expected confusion noise levels are also reported in red lines.

Zoom In Zoom Out Reset image size

The figures show that with a typical confusion noise of about 10 MJy sr^-1 at 500 μm, it will be possible to detect most Infrared Dark Clouds (IRDCs, see § 3.2), where column densities are in the range 10²³–10²⁵ cm^-2. Less dense clouds with column densities of the order of few 10²¹ cm^-2 should be easily detectable at levels of tens of MJy sr^-1 at 170 μm at temperatures as low as T ∼ 20 K. The situation may be less simple in the regions closer to the Galactic center. However, our broad spectral coverage provides an important advantage for measuring the temperature accurately, and for isolating structures and sources with temperature different from the standard diffuse ISM cirrus (∼20 K). Besides, all of these quantities are based on extrapolated estimates and we will provide the definitive measurement of the cirrus confusion at Herschel wavelengths and resolution.

Virtually every block of the Hi-GAL survey will contain secondary calibrators, either stars or other well-known objects, ensuring accurate checks of the flux calibration of Hi-GAL data. Our baseline calibration targets will be the 400 stars used as calibrators for the Spitzer GLIMPSE-I/II surveys. These are mostly A0-5V or K0-M0III stars, although they also include ∼60 calibrators of hot or warm dwarfs (B-G). In order to remove sources with nonphotospheric FIR emission, we will make predictions for the 24 μm fluxes and then test for excesses in the MIPSGAL data first. Once anomalous sources are excluded we can extrapolate to the FIR and create an initial set of calibrators.

We will also be able to obtain a reliable calibration for extended sources, which is one of the more difficult parts of the nominal instrumental calibration activity. As part of the calibration scheme we will compare fluxes in the SPIRE 500 μm band with fluxes from the same band of Planck-HFI in suitable locations.⁵⁷

At near-infrared wavelengths, the emission from dust is produced by small particles (size 1.2–15 μm), whose abundance varies significantly, being strongly depleted by coagulation processes in the dense ISM (e.g., Bernard et al. 1993; Abergel et al. 1994; Stepnik et al. 2003). Far-infrared (FIR) emission is produced by larger grains (size 15–110 μm), which are more stable and dominate the total dust mass, and trace all phases of the ISM. The ISM dust spectrum peaks in the FIR where the Galaxy is transparent. FIR emission is therefore a reliable tracer of the overall ISM column density structure in our Galaxy. Other phase-independent tracers include dust absorption and gamma-ray production; however, the former can be used to sample only the nearest 1 kpc, and gamma-ray surveys currently lack sensitivity and angular resolution.

Variations in the FIR emissivity (the ratio of surface brightness to column density) are dominated by the nonlinear effects of dust temperature through the Planck function. Fortunately, the shape of the dust SED as measured by PACS and SPIRE will be most sensitive to temperature variations as the spectral bands sample the peak of the Big Grain emission, and the contribution of Very Small Grains can be estimated from the Hi-GAL data at 70 μm and MIPSGAL at 24 μm. The dust temperature (T_d) and its spatial variations will therefore be measured precisely. This important parameter can be used, in conjunction with complementary data from Planck, HI, CO, Hα, and gamma-ray surveys, to estimate the strength and spectral shape of the InterStellar Radiation Field (ISRF), which is set by the stellar content in a given region. So far, the dust temperature in the Galactic plane has been mapped over limited regions using IRAS (Kim et al. 1999, Douglas & Taylor 2007, and over the entire plane at a resolution of 40' (Lagache et al. 1998) using DIRBE. Hi-GAL will improve with respect to the latter by a factor of about 100 in linear scales. It will trace the local radiation field on scales relevant to star formation, and provide mass estimates even at large distances. In the case of the dense medium, determining the 3D distribution of the ISRF strength and spectral shape in a given cloud will require radiative transfer modeling. This is possible, even for complex geometries, using Monte Carlo codes (e.g., Juvela & Padoan 2003). Using such codes, the equilibrium dust temperature T_d and the dust emission can be predicted at any 3D location in the cloud. Integration along the line of sight in turn allows us to predict 2D emission maps.

About a quarter of the mass in the ISM is in molecular form (Blitz 1997) and most of that material resides in giant molecular clouds (GMCs). Since GMCs are also the dominant sites of star formation, understanding their origins and evolution is essential to our understanding of the Galactic environment.

In combination with molecular line surveys, Hi-GAL will provide the data needed to derive basic physical properties of GMCs. We will detect and characterize cold structures in the inner GP and classify them based on star-formation activity. Detection statistics for clouds with different temperatures and degree of star-formation activity will provide the fraction of quiescent versus star-forming clouds. It will thus be possible to constrain the properties and lifetimes of GMCs in our Galaxy and to compare with the predictions of fast evolution of molecular clouds (Hartmann et al. 2001) or a more traditional slow evolution of star formation in our Galaxy (Shu et al. 1987). Variations with galactocentric radius will determine if the slow/fast scenarios are mutually exclusive or reflect different initial/environmental conditions. A large-area survey like Hi-GAL will provide the needed statistical significance in all mass bins, especially at the high-mass end, and in a variety of Galactic environments.

Direct detection of cold (i.e., T < 20 K) dust, which could be the quiescent counterparts to GMCs, has been difficult in the past (Sodroski et al. 1994; Reach et al. 1995; Lagache et al. 1998; Kim et al. 1999; Douglas & Taylor 2007) either because of insufficient wavelength coverage (e.g., IRAS) or inadequate spatial resolution (DIRBE, FIRAS). CO observations are problematic due to molecular freeze-out onto grains (Flower et al. 2005), or photochemical effects in low-metallicity environments (Bot et al. 2007). The recent detection of very cold clumps in the GP with Archeops (Désert et al. 2008) and BLAST (Olmi et al. 2009), confirms the FIR and submillimeter continuum as the best tool to trace cold ISM components. Notable examples are Infrared Dark Clouds (IRDCs) and Hi Self-Absorption (HISA) clouds.

IRDCs are structures initially discovered as extinction features against the bright mid-IR Galactic background, and soon verified to exhibit properties similar to molecular clouds. Their properties (n > 10⁵ cm^-3, N_H ∼ 10²²–10²⁴ cm^-2, and T < 25 K —Egan et al. 1998; Carey et al. 2000, R ∼ 5 pc, and M ∼ 10³ M_⊙ —Simon et al. 2006; Rathborne et al. 2006) suggest that they are the precursors of cluster-forming molecular clumps like Orion. Thus, IRDCs are ideal for the study of the pristine, undisturbed physical conditions that may produce massive stars and clusters. IRDCs have only been detected against the bright Galactic mid-IR background (mostly for |l| ≤ 30°); their true Galactic distribution is unknown. Modeling of IRDCs in the IR and submillimeter (Fig. 6, Stamatellos et al. 2004) proves Herschel's unique ability to detect them and to measure their SEDs. IRDCs, with τ_200 μm≥1, are not detectable by either IRAS or Spitzer. Hi-GAL will provide a definitive inventory of cold dust and potential sites of massive star cluster formation everywhere in the inner GP.

Zoom In Zoom Out Reset image size

HISAs are traced by cold H i gas seen in absorption against a background of warm H i emission (Gibson et al. 2000, 2005; Goldsmith & Li 2005); they may provide additional clues to the formation of molecular clouds. When compared with molecular tracers, these cold (relative to the ambient neutral medium) and relatively quiescent (Δv ∼ 1 - 3 km s) clouds show a wide range of H I/H₂ ratios (Li & Goldsmith 2003; Klaassen et al. 2005) which suggests that they might be H I → H₂ conversion sites. Detailed studies of their FIR → mm SED shapes can help clarify this issue, providing evidence for the grain types necessary for a reasonable H I/H₂ formation timescale (Goldsmith & Li 2005).

The paradigm for the formation of solar-type stars via accretion through a circumstellar disk (Shu et al. 1987) predicts an evolution from cores to protostars and, finally, pre–main-sequence stars that is well matched with distinctive characteristics of their SEDs (Lada & Wilking 1984; Andre et al. 1993). The empirical classification of the SED of low-mass YSOs has thus been used as a powerful tool to constrain theoretical models.

Higher mass stars reach the conditions for H-burning faster than the time required to assemble them, so that winds and radiative feedback will strongly influence accretion and may limit the final mass of the star (Zinnecker & Yorke 2007). However, since massive stars exist, several theories have been proposed to solve this puzzle, including accretion from a disk, a very high-pressure ambient medium, the "flashlight effect" (Yorke & Sonnhalter 2002; McKee & Tan 2003), competitive accretion, and coalescence (e.g., Stahler et al. 2000). Application of SED-based classification tools, and evolutionary diagnostics like the M_env - L_bol diagram which relates the bolometric luminosity of a YSO to the mass of its envelope (Molinari et al. 2008), to a large sample of luminous protostar candidates in the GP will define a timeline for the various phases of massive star formation that will constrain the theories and lead to new estimates of the SFR. Clearly the source distance is the crucial parameter here; we are collecting the information from the major molecular line surveys over the inner Galactic plane, while planning to undertake additional surveying activities at a variety of facilities in several high-density tracers to get additional data (see § 4 for more detail).

An evolutionary sequence for massive YSOs has been proposed in which cold massive cloud cores evolve into Hot Molecular Cores with outflow, IR-bright massive YSO, and finally into ultracompact (UC) H ii regions (e.g., Evans et al. 2002; Kurtz et al. 2000), but it is qualitative and based on small and possibly incomplete samples. Samples of bright and massive YSOs (Molinari et al. 1996; Sridharan et al. 2002; Hoare et al. 2004) are IRAS or MSX selected and tend to suffer from age biases and confusion which prevent firm quantitative conclusions. A phase of intense and accelerating accretion prior to H-burning ignition, that may be observable (e.g., Molinari et al. 1998) in the form of dense condensations devoid of IR as well as radio continuum emission, seems confirmed by recent large millimeter surveys (Beltrán et al. 2006; Hill et al. 2005). Millimeter continuum alone, however, cannot distinguish between precollapse condensations and rapidly accreting cores; Hi-GAL will use the full potential of Herschel wavelength coverage and spatial resolution to trace the SED peak of dust envelopes in all phases, from massive prestellar condensations to ultracompact (UC) H ii regions. An angular resolution of 30'' or less, typical of Hi-GAL, has been proven (e.g., Molinari et al. 2008) to be the key to building accurate SEDs, deriving reliable luminosities for massive YSOs, and distinguishing embedded UC H iis strongly emitting in the mid-IR and radio from pre-UC H iis (Fig. 7).

Zoom In Zoom Out Reset image size

The abundance of high-mass YSOs per mass bin in the various evolutionary phases will provide an estimate of the duration of each phase. This timeline can be directly compared with the predictions of various models, and together with the YSO mass function, will be used to infer the SFR. As an example, using current estimates for SFR and IMF (McKee & Williams 1997; Kroupa 2001), and a 10⁵ yr period to assemble a massive star (McKee & Tan 2003), we expect ≥1000 FIR/submillimeter objects with M > 15 M_⊙ in the Galaxy, and only ∼180 objects with M > 50 M_⊙. Recent millimeter surveys in the Cyg X region (Motte et al. 2007) confirm the rarity of such massive precursors and strengthen the need for a systematic unbiased search.

Galactic phenomenology currently invokes an indeterminate mixture of spontaneous and triggered star formation. Triggering agents include radiation pressure from OB stars (Sugitani et al. 1989), compression by expanding H ii regions (Elmegreen & Lada 1977; Deharveng et al. 2005), or fragmentation of supershells by multiple supernovae in OB associations (McCray & Kafatos 1987). On larger scales we still do not know if spiral density waves actively induce star formation (e.g., Elmegreen 2002) or simply assemble star-forming regions, with local feedback and triggering becoming more important within the arms (Sleath & Alexander 1996). The mean SFE of a galaxy can increase (up to 50 times) in starbursts (Sanders et al. 1991) and galaxy mergers due to strong feedback effects, a process observed in miniature in Galactic star-forming regions (e.g., Moore et al. 2007). Whether the IMF depends on local triggering and other environmental factors is unclear.

Hi-GAL will enable quantitative analysis based on basic observables—the luminosity functions of YSOs, the mass function of dense star-forming structures and quiescent clouds. Hi-GAL will provide the essential context of high-mass star formation, as it relates to molecular gas, H i gas, stars, H ii regions, OB associations, supernova remnants (SNRs), and spiral arms. Theoretical models and numerical simulations will be tested in multiple ways. We will discover whether a local triggering agent is necessary for high-mass star formation or if a spiral arm is sufficient, clarifying the differences between spontaneous and triggered star formation. We will quantify the relationship between the interaction strength (estimated using available data from ancillary surveys) and the resulting increase in SFE above the spontaneous rate. By locally relating the SFR to the properties of the ISM we will probe star-formation thresholds as a function of environment and spatial scale, and possibly unveil the mechanism giving rise to global Schmidt-like scaling laws. We will determine the dominant physical process underlying triggering.

The 5 band FIR images and source catalogs provided by Hi-GAL will allow research in many fields that can only be partially anticipated. A detailed description of the specific outcomes in all these fields goes beyond the scope of the present paper. These aspects will of course be the object of dedicated publications to be released in due time. In the following we list some of the possible byproducts of Hi-GAL:

• 1.  
  An input catalog for ALMA: we expect to detect some 200–400 objects per tile, most of which will mark very cold objects to be studied at all possible wavelengths;
• 2.  
  Complete characterization of the Galactic foreground in the far-IR and submillimeter, critical for the correct interpretation and modeling of cosmological backgrounds;
• 3.  
  Dust formation and destruction in SNRs;
• 4.  
  Debris dust disks around main-sequence stars, with unbiased statistics on frequency and mass as a function of star age;
• 5.  
  Evolution of dust properties, especially around AGB stars, the factories of cosmic dust;
• 6.  
  Detection of detached dust shells around first ascent giant stars to investigate missing mass in AGB envelopes;
• 7.  
  Detection of multiple shells around AGB stars, post-AGB objects, and planetary nebulae, as well as around various classes of interacting binaries;
• 8.  
  Detection of ejecta shells and swept-up ISM bubbles around massive stars, providing a complete census of WR and LBV stars;
• 9.  
  Extinction maps to aid in correcting near-IR galactic star counts;
• 10.  
  Detection of Solar System objects via comparison of cross-linked rasters: in particular the detection of asteroids will be very interesting for studies of the albedo;
• 11.  
  Nearby low-mass SFRs in the GP: Herschel will detect many nearby star-forming regions and individual YSOs.