1. Introduction

Emission from galaxies at wavelengths longer than a few micrometers is produced mainly by dust. It is often assumed that dust emission is associated predominantly with star formation (e.g., Dwek, 1998; Pipino et al., 2011; Asano et al., 2012). Indeed, as is widely known, star-formation activity is always accompanied by dust production on a timescale of 106 yr, very probably through dust production in supernova explosion (e.g., Kozasa and Hasegawa, 1987; Kozasa et al., 1991; Nozawa et al., 2003, 2007, 2008, 2010), though the actual mechanism of dust supply has not been fully understood yet. In such a case, the energy emitted by massive stars in the ultraviolet (UV) is scattered efficiently and finally absorbed by dust grains. The heated grains re-emit the energy in the far-infrared (FIR). Consequently, radiative processes in FIR can be related both to the composition and the amount of dust in galaxies and to the properties of their stellar population, especially the fraction of newly-formed massive stars.

Many studies have shown that a significant amount of star formation in galaxies is obscured by dust (Le Floc’h et al., 2005; Caputi et al., 2007; Buat et al., 2007a, b, 2008; Reddy et al., 2008). It has also been found that such “hidden” star formation becomes more and more important with increasing redshift from z = 0 to 1 (Takeuchi et al., 2005). Therefore, FIR observations are of crucial importance to understand the true star-formation activity and history in the Universe.

In the context of star-formation activity, supernovae (SNe) are often discussed as a source of dust supply, because SNe are the final stage of the evolution of massive (M > 8M⊙) stars that are the tracers of star-formation history, since their lifetime is very short (a few Myr). Up to now, the quantity of dust produced by a supernova is a matter of debate, because theories tend to predict a high efficiency of SN dust production (0.1-1M⊙ per core-collapse SN: e.g., Kozasa et al., 1991; Clayton et al., 2001; Nozawa et al., 2003) while a smaller amount is inferred from observations (e.g., Rho et al., 2008, 2009), but the discrepancy between theory and observation is now diminishing.

By contrast, in present-day evolved galaxies, the major sources of dust are stars in the asymptotic giant branch (AGB), and post-AGB, phases. The AGB is the final phase of evolved intermediate-mass (0.8M⊙ < M < 8M⊙) stars, expelling dust into the interstellar medium (ISM) via planetary nebulae. In the context of this work, since both AGB, and post-AGB, stars expel gas and associated dust by super-wind phenomena, we say they are a source of dust, though a detailed mechanism is still not perfectly clarified. Further, the mass contribution of dust from AGB-related phenomena has been proved to be significant even ~500 Myr after the birth of a galaxy (Valiante et al., 2009; Asano et al., 2010). Some recent studies claim that the contribution to the mid–far-IR luminosity is also very important (Kelson and Holden, 2010; Takeuchi et al., 2010). Thus, a proper understanding of dust produced by AGB stars, as well as SN-related dust, is crucial for interpreting dust emission at any age of the Universe.

In order to estimate the contribution from each component of dust emission, all-sky far-IR survey data are desirable. The Infrared Astronomical Satellite (IRAS; Neugebauer et al., 1984) has brought a vast amount of statistics and very efficient methods of analysis have been devised from the IRAS Point Source Catalog (PSC). The four bands of IRAS have enabled us even to perform a very detailed classification of extragalactic and various galactic objects, like blue and red galaxies, Seyferts and QSOs, carbon stars, Hii regions, reflection nebulae, planetary nebulae, T Tauri stars, etc. Hacking et al. (1985) classified the stars on the IRAS color-color plane, and found that C- and O-rich stars occupy a separated region different from that for normal stars. Thronson et al. (1987) showed a distinction between C- and O-rich stars based on IRAS colors. Van der Veen and Habing (1988) introduced ten zones for the detailed classification of IRAS-detected stars, which was later extensively used by Busso et al. (1996). Walker and Cohen (1988) also defined occupation zones for stars. A thorough description of the IRAS color-color classification method including extragalactic objects and star-forming regions can be found in, e.g., Walker et al. (1989).

After many years since IRAS, the advent of AKARI (ASTRO-F) opened a new window to explore the Universe, as a survey-oriented space telescope at MIR and FIR (Murakami et al., 2007). The primary purpose of the AKARI mission is to provide second-generation infrared (IR) catalogs to obtain a better spatial resolution and a wider spectral coverage than the IRAS catalog. Using the AKARI FIS All-Sky Survey Bright Source Catalog and existing databases (NED and SIMBAD), we have constructed a method to classify the sources on the color-color diagram only from FIS bands (Pollo et al., 2010a). In Pollo et al. (2010a), we found that it is quite plausible to select AGB stars from the FIS catalog exclusively. The method itself remains to be improved by a more sophisticated examination, but this is to be done in our future work (Rybka et al., 2011, in preparation).

As an interesting comparison, we note that Ita et al. (2010) showed color-color diagrams of nearby stars from the AKARI IRC All-Sky Survey. Also, recently, Phillips and Marquez-Lugo (2011) presented IRC-FIS-based color-color diagrams of planetary nebulae, and Cox et al. (2011) have shown AKARI images of 13 post-AGBs and (proto)planetary nebulae.

In this work, we make an attempt to quantify the contribution from AGB, and post-AGB, stars to the FIR flux related to point sources from the outer regions of the Milky Way, using data gathered by the satellite AKARI in the FIS All Sky Survey. We expect that our conclusions on the Milky Way can be also applied—with due attention—to other similar galaxies.

The article is organized as follows: in Section 2, we present the data used for this analysis and the statistics of AGB stars found in them. In Section 3, we present and discuss the location of different types of AGB stars in the FIR color-color plots. In Section 4, we discuss the contribution from AGB, and post-AGB, stars to the FIR flux of the Milky Way. Section 5 presents a summary.

2. The Data

2.1 AKARI

AKARI is a Japanese astronomical satellite whose purpose is to perform various large-area surveys at the IR from NIR to FIR with a wavelength coverage of 2-160 μm, as well as pointed observations. 1 Footnote 1 AKARI is equipped with a cryogenically-cooled telescope of 68.5 cm aperture diameter and two scientific instruments, the Far-Infrared Surveyor (FIS; Kawada et al., 2007) and the Infrared Camera (IRC; Onaka et al., 2007). Among various astronomical observations performed by AKARI, an all sky survey with FIS and IRC has been carried out; it is referred to as the AKARI All-Sky Survey.

Since FIS is an instrument dedicated to FIR, λ = 50–180 μ m, all the AKARI FIS bands are in the FIR wavelengths: N60 (65 μ m), WIDE-S (90 μ m), WIDE-L (140 μ m), and N160 (160 μ m) (Kawada et al., 2007). Hereafter, we use a notation S65, S90, S140 and S160 for flux densities in these bands.

Especially, since FIS has a better sensitivity and reaches longer wavelengths than IRAS, we can expect a different composition of sources: we should see objects with cool dust which were difficult to detect by IRAS bands.

2.2 Catalogs

As mentioned above, one of the main goals of AKARI was to carry out the All-Sky Survey in the four photometric bands in the far-infrared wavelength range, 50–180 μ m, with FIS (Kawada et al., 2007). The FIS scanned 96% of the entire sky more than twice in the 16 months of the cryogenic mission phase.

In March, 2010, the AKARI/FIS Bright Source Catalogue V. 1.0 was released to the scientific community. It contains in total 427 071 point sources measured at 65, 90, 140, 160 μ m. The position accuracy of the FIS sources is 8″, since the source extraction is made with grids of this size.

Effective size of the point spread function of AKARI FIS in FWHM is estimated to be 37±1″, 39±1″, 58±3″, and 61±4″ at N60, WIDE-S, WIDE-L, and N160, respectively (Kawada et al., 2007). Errors are not estimated for each individual source, but instead they are, in total, estimated to be 35%, 30%, 60%, and 60% at N60, WIDE-S, WIDE-L, and N160, respectively (Yamamura et al., 2009). For the examination of properties of objects on color-color planes, the effect of photometric errors is a matter of concern. In Pollo et al. (2010a), we have examined the effect of photometric errors on the boundary of the star-galaxy classification, and found that this affects the position of the boundary on the color-color plane, but not so significantly that it would change quantitative conclusions. It should be noted that photometric errors do not affect the classification of sources used for the purpose of this work: they are all identified by a direct cross-match with public databases, as explained in the latter part of this section.

In addition to the error estimation, quality flags are attributed to all the detected sources. Their values range from 0 (no detection), through 1 (source not confirmed), 2 (detection confirmed, flux not reliable) to 3 (source confirmed and flux reliable). In practice, only flags 1 and 3 dominate. It would seem reasonable to narrow down the analysis only to the flag 3 sources; however, the total number of such sources is small and as a result the statistics becomes poor. For that reason, for the subsequent analysis we decided to use all the detected sources, with due attention. However, we use the safe measurements with flag 3 for a quality check of our results.

Among all the AKARI FIS All-Sky Survey sources, we found 43342 sources with fluxes measured at all four FIR bands.

In order to identify and classify AKARI FIS sources, we searched for counterparts of sources with a complete FIR photometric information in public databases: NED and SIMBAD. The details of this procedure were discussed in Pollo et al. (2010a). We stress that, for the purpose of this work, we use only identifications based on a direct cross-match with public databases. The classification of sources (including stars used in this work) is, in most cases, consistent with the information in the SIMBAD and NED databases. However, since the information given there is not always the most recent, in case of all stars used for this work we searched the available literature and in dubious cases we used the classification given in the most recent publications. The following papers were extremely helpful in this task: Deguchi et al. (1990); Lewis et al. (1995); Young (1995); Josselin et al. (1996); Sloan and Price (1998); Percy et al. (2001); Couch et al. (2003); Jura (2003); Maas et al. (2003); Winters et al. (2003); Whitelock et al. (2006); Szczerba et al. (2007); Mesler and Henson (2008); Ita et al. (2010).

It should be noted here that the results presented in Pollo et al. (2010a) were based on the β-1 version of the AKARI All-Sky Survey (Yamamura et al., 2008), which contained less sources than were included in the published version of the survey, and preliminary measurements of positions and flux densities. Using the published AKARI FIS Bright Source Catalogue Version 1.0 (hereafter: BSC) data we verified our identifications, which, in most cases, were proved to be correct, i.e. the previously found identification was the closest-found counterpart in the search range of 8″ (smaller than the previously assumed 40″), while the median distance between the counterpart and the source was ~4″. Also, most often in the search range of 8″, the chosen counterpart was the only possibility. The risk of miss-identification still exists, when such a procedure is applied, but since we are dealing with relatively bright sources, it is small. Also, since our work is mainly statistical, a few possible mis-identifications should not affect the final result.

Using the improved flux densities, as well as only measurements with the best flag 3, from the BSC, we also confirmed the validity of the division lines presented in Pollo et al. (2010a) (Rybka et al., 2011, in preparation).

For the purpose of this work, we use the sources identified in the work of Pollo et al. (2010a), but with confirmed identifications and flux densities from the BSC. Identification of Galactic objects, which are the center of interest of this paper, was confirmed in all cases. However, for some flux densities (mostly at 160, but, in a few cases, 140 μm) zeroes were given. To keep the condition that only sources with complete FIR photometric information are used, we removed these sources with zero flux densities from our sample. This procedure decreased the number of Galactic sources in our sample to ~80% and is responsible for inconsistency between the numbers presented in this work and the tables presented in Pollo et al. (2010a).

2.3 Point sources outside the Galactic plane

The area of low emission of Galactic dust (ɤ 10 MJy sr−1) in practice corresponds to the sky area outside the Galactic plane and Magellanic Clouds. In this area, we expected the best efficiency of the identification.

Indeed, for 5176 AKARI sources with the full four-band FIR photometry in the β-1 version of the All-Sky Survey, detected in this area, we could identify all but 22 sources.

Verified with BSC, all these 5176 sources have a flag 3 measurement in the S90 band, which implies that they are all confirmed sources and their flux densities at this band should be treated as reliable. However, only 1380 of them have a flag 3 measurement at S140 and 405 at the S65 band. The number of objects having a flag 3, reliable, measurement at all four FIS bands is only 142. For that reason, for the analysis of colors we decided to use all the available data, which, as shown in Pollo et al. (2010a, b), still gives statistically-meaningful results. However, when discussing particular cases, we should be aware of possible biases in color measurements. These issues will be discussed in detail in Rybka et al. (2011, in preparation).

Most of these sources discussed in Pollo et al. (2010a) are nearby galaxies (4272). We also found 349 stars and 48 “other” (extended or line emitters) Galactic sources which mostly are related to stars. As mentioned above, identification of all these 397 Galactic sources has been confirmed in the published BSC. The condition of the full photometric information in FIR was kept in the case of 330 of these sources, i.e. 83% of them.

The properties of these 330 sources are summarized in Table 1. Most of the Galactic point sources presented there (270 from 330, i.e. 82%) are related either to AGB, or post-AGB, stars (this category includes mainly planetary and protoplanetary nebulae, PNe and PPNe). In case of some of them the classification is tentative: they are pulsating red giants possibly being AGBs or close to the AGB phase. Since the primary purpose of this work is to make a first-step estimate of the contribution from the evolved intermediate mass stars to the dust supply and the FIR flux of galaxies, we include also these tentatively-classified objects to our sample: if they are not exactly AGBs, they are pulsating red giants not so far from AGBs in evolutionary stage and physical properties. It should also be noted that, among the remaining sources, there is a certain number of other evolved stars, such as supergiants.

Table 1. Classification of 330 Milky Way sources with a complete photometry, located in the sky regions of low 100-μm emissivity (I100 ≤ 10 MJy sr−1).
Full size table

3. AGB and Post-AGB Stars in the FIR Color-Color Diagrams

Color-color diagrams of the AGB, and post-AGB, stars from our sample are shown in Fig. 1 and Figs. A.1–A.14. In all these figures, the left panel presents the diagram using all the data, while, in the right panel, only the sources with the secure flag 3 for all the flux densities used in a diagram are shown. Different types of AGB stars, viz., carbon stars, OH/IR stars, AGB stars without classification, and post-AGB stars, are shown as small symbols, while the corresponding large symbols mark the corresponding median values.

Fig. 1.
figure 1

Positions of different types of AGB stars in the FIR color-color diagram S65 / S90 vs. S65/S140. Mira-type variables are shown as open circles, carbon stars—as open squares, OH/IR stars—as full triangles, post-AGB stars—as full squares, and AGB stars without detailed classification—as open triangles. Individual sources are marked by small symbols, while large symbols denote corresponding median values. In the left panel, all sources are presented. In the right panel, only sources with the highest quality flags 3 are shown (note difference in scale).

Full size image

In most of the color-color diagrams, the AGB stars display a significant scatter. This is probably mainly due to the errors in the flux density. This conclusion is supported by the fact that this scatter is much lower in the color using shorter, most secure, wavelengths: (S65/S90). The scatter is also significantly reduced if we use, in the color-color diagrams, only objects with secure measurements (flag 3).

The median colors of all the sub-types of AGB stars shown in the diagrams are usually very similar. In some of the diagrams (Figs. A.1, A.3, A.5, A.7, A.9, A.10, A.11, A.12, A.14) the OH/IR stars seem to be much bluer than the rest. However, first of all, this is the smallest subcategory of our AGB stars. Second, it should be noted that all these diagrams where the deviation appears make use of the S160 flux density, which is not reliable for all but one of these stars (the rest of them received flag 1). Moreover, this single OH/IR star with a good S160 measurement does not have a reliable measurement of the S140 flux density. As a result, only one OH/IR star is shown in only some of the “high quality” diagrams using the S160 flux density, and, of course, without a corresponding median. However, in the “high quality” diagrams using only shorter wavelengths these stars do not seem to differ from the rest of the AGB stars. For that reason, we do not regard this deviation as being significant—it is probably due to the combined effect of small statistics and large errors.

The same problem of a small number of high-quality measurements in S160 applies to carbon stars which are the second-smallest subcategory of our sample of AGB stars. Consequently, their corresponding medians are also not shown in the “high quality” versions of the diagrams. However, in the remaining diagrams, they also seem to be placed very similarly to other AGB stars.

At the same time, the only group which appears to be separated in most of the diagrams are post-AGB stars, or, more precisely, a subsample of post-AGB stars. As shown in the S65/S90 vs. S65/S140 diagram, presented in Fig. 1, these objects are significantly redder and this effect remains visible also in the diagram using only the secure flag 3. Especially in the right panel of Fig. 1, which presents the diagram using only “secure” measurements, it is seen that some post-AGB objects (all of them are planetary nebulae) form a clearly-separated “cloud” in the red part of the diagram. Another group of post-AGB objects is mixed with the main AGB cloud, but also seems to remain in its redder part. Consequently, the median value for the post-AGB stars is shifted closer to the main cloud, but the difference with respect to the AGB stars is still evident. The same effect is visible in practically all the other diagrams, presented in Figs. A.1–A.14.

As an attempt to interpret this behavior, we present the identifications of these “red” and “blue” clouds of post-AGB objects, respectively, in Tables 2 and 3. All of these sources are well-known PNe, with the exception of one PPN. In the first and second column of Tables 2 and 3, we give the most commonly-used names of these nebulae, while, in the third column, we put the comments about the morphologies and estimation of their kinematic ages, taken from the literature. Since the estimation of the ages of planetary nebulae depends on the estimation of their distances, very often highly uncertain, there are many contradictory estimations found in the literature and the numbers given in the tables should be treated as indicative. The evolutionary stage of planetary nebulae does not depend only on their age, but also on other factors, such as the mass of the exciting star (see, e.g. Blöcker, 1995). Given a small number of objects, we do not attempt to formulate any strong conclusions about the relation between the properties of these objects and their position in the FIR color-color diagrams. We observe, however, that most of the members of the bluer cloud of post-AGB stars are young PNe, and their morphological structure is usually not very well developed. The PPN Frosty Leo also belongs to this group. Even older PNe in this group, like the Saturn Nebula, are not strongly evolved yet, either. The remarkable difference is a particular born-again PN Abell 78.

Table 2. Identifications and properties of the separated red cloud of post-AGB stars selected in the high-quality color-color diagram S65/S90 vs. S65/S160, presented in the right panel of Fig. 1.
Full size table
Table 3. Identifications and properties of the bluer cloud of post-AGB stars, mixed with the AGB stars, selected in the high-quality color-color diagram S65/S90 vs. S65/S160, presented in the right panel of Fig. 1.
Full size table

In contrast, the members of the red cloud of post-AGB objects usually have well-developed morphological structures. They are very often bipolar, with strong asymmetrical features. Their estimated ages are usually older.

Given the above observations, we can formulate the careful speculation that PNe move towards the redder part of the FIR color-color diagram (i.e. become more luminous in the S140 and S160 bands) as a result of their evolution. This is related not simply to their age, but rather to the evolutionary status and stage of development of the nebula. If so, this reddening may be simply interpreted as a result of the cooling of grains (a similar interpretation is given by Phillips and Marquez-Lugo, 2011, who analysed a much larger sample of planetary nebulae observed by AKARI). It is also possible that the very often seen complex bipolar structure of these redder nebulae is related to their increased luminosity in the longer IR wavelengths, and that both these features have a common origin. This issue undoubtedly will be a subject of further studies.

The consequences of this effect for the observed flux of the outer part of the Milky Way are discussed in the next section.

4. FIR Flux from AGB and Post-AGB Stars—Contribution to the Total FIR Flux Related to Point Sources outside the Galactic Plane

As seen from Table 1, AGB, or plausible AGB, stars are more than 80% of Galactic FIR point sources detected outside the Galactic plane. Planetary nebulae are 10% of these sources. These percentages are also presented in Fig. 2 as vertical lines.

Fig. 2.
figure 2

Contribution from AGB stars to the total number of point sources detected in FIR (solid horizontal line for all sources and dashed horizontal line for planetary nebulae) and to the total FIR flux of the Milky Way outside of the Galactic plane, coming from point sources (full circles for all sources and empty circles for planetary nebulae only) in different FIR wavelengths from 65 to 160 μm. Indicated errors are rms errors, related to the number statistics of sources.

Full size image

At the same time, as shown in Fig. 2, the contribution from AGB stars to the total FIR flux from Milky Way sources located outside the Galactic plane corresponds to their number percentage but changes with wavelength. It is high in all the range between 65 and 160 μm, above 85% at 65 μm, and dropping above 100 μm, but at 160 μm still remaining at a level of more than 60%.

More than 20% of this contribution to the FIR flux comes from post-AGB stars (mainly planetary nebulae), while, in number, they only account for 10% of the sources. Their contribution becomes more significant at longer wavelengths, since a significant fraction of them does not become fainter at wavelengths longer than 100 μm, contrary to the other stars. A subsample of post-AGB stars which form a separate redder group on the color-color diagrams is responsible for this effect.

5. Summary

We conclude that Galactic objects luminous in the FIR outside the Galactic plane in more than 80% of cases are red giants, mainly AGB, and post-AGB, stars. They dominate not only in number, but also their contribution to the FIR flux associated with point sources in the outer part of the Galaxy is dominant. While it has already been established that a significant mass fraction of Galactic dust is related to AGB stars, now we show that they also contribute significantly to the FIR flux of the outer parts of our Galaxy, thanks to the new data from AKARI.

This issue needs to be studied further but we may expect a similar behavior for other evolved Milky Way-type galaxies. Further exploration of AKARI All Sky Survey data should provide a more precise estimation of the role of the AGB component in the FIR spectral-energy distribution of the Milky Way and other galaxies. In particular, our measurement implies that the contribution from stars in late phases of evolution should probably not be neglected when modeling the FIR emission of Milky Way-type galaxies.

Additionally, we point out that the FIR color-color diagrams only provide a possibility for the successful selection of a sample of post-AGB stars, in particular of planetary nebulae.