MERGING GALAXY CLUSTER A2255 IN MID-INFRARED

CLEVL (CLusters of galaxies EVoLution studies) is one of the Mission Programs of the AKARI IR telescope (Murakami et al. 2007) designed to understand the formation and the evolution of galaxies in cluster environments. The program is divided into three major components according to the redshifts of the targets—low-redshift, intermediate-redshift, and high-redshift clusters of galaxies. A2255 is one of the target galaxy clusters for the low-redshift CLEVL program (Im et al. 2008; Lee et al. 2009).

A2255 is observed by the AKARI InfraRed Camera (IRC; Onaka et al. 2007) over eight fields (Figure 1), with each field covering an area of 10 × 10 arcmin². The IRC consists of three cameras—the NIR band camera [N2, N3, N4], the MIR-S band camera [S7, S9W, S11], and the MIR-L band camera [L15, L18W, L24], where the number next to each alphabet denotes the central wavelength of the filter. Among these, the NIR and MIR-S band cameras are placed to observe the same field of view (FOV) on the sky, while the MIR-L band camera points to a different field separated by ∼10 arcmin from the NIR and MIR-S FOV. The wide FOV (10 × 10 arcmin²) and the continuous wavelength coverage at 3–24 μm are two main advantages of AKARI with respect to Spitzer, in terms of the study of galaxy clusters. For the CLEVL low-redshift cluster programs, we used the IRC02 Astronomical Observation Template (AOT; see AKARI Observer's Manual version 1.2⁸) mode. This AOT takes moderate-length exposure by using two filters for each camera. The filter composition we selected is [N3, N4, S7, S11, L15 & L24]. With this observational design, we obtain all-filter coverage for the central four fields (∼400 arcmin²). Four fields in the south are covered only in L15 and L24 (dashed line in Figure 1), while four fields in the north are covered in N3/N4/S7/S11 (solid line in Figure 1). Thus, the final data coverage with at least one AKARI filter is 1200 arcmin². The on-source exposure time for each filter is roughly ∼140 s. The summary of the observation and the characteristics of the reduced images are specified in Table 1.

Zoom In Zoom Out Reset image size

The images are reduced using the IRC pipeline version 070104⁹ (provided as a form of the IRAF external package). The IRC pipeline consists of sky subtraction, astrometric calibration, and final coaddition of individual frames. For the MIR-L images whose astrometric calibration within the pipeline is relatively poor, we derive an astrometric solution using the IRAF task ccmap using Two Micron All Sky Survey sources or S7 images as a reference. The full-width-at-half-maximum (FWHM) of the reduced images is between 40–60, depending on the filter. The astrometric accuracy is 1''–2'' rms for N3/N4/S7/S11 images and 3''–4'' rms for L15/L24 images.

To measure the MIR fluxes, we use SExtractor (Bertin & Arnouts 1996). In order to include all objects detected in at least one band, we make a reference image for source detection by combining images of different bands: for the photometry of N3 through S11 images, we use an N3/N4/S7/S11 combined image as a reference image. Since the sensitivities in the L15 and L24 images are relatively poor compared to the case of N3–S11, we only combine L15 and L24 images to make a reference image for the photometry in L15/L24 bands. We perform dual-mode photometry using the reference images, with SExtractor configuration of DETECT_THRESH = 3.0, DETECT_MINAREA = 2.0, and BACKGROUND mesh size of 32 pixels. As a measure of the total flux of each galaxy, we use FLUX_AUTO from the SExtractor output, i.e., the flux within a Kron elliptical aperture. The fluxes with a unit of ADU are converted to f_ν in μJy, using the IRC flux calibration table in the AKARI IRC data users' manual version 1.4 (08/06/03; Lorente et al. 2007)¹⁰. The "dual-mode" photometry we used in this paper is consistent with "single-mode" photometry, which is photometry performed on each image independently. The difference between dual-mode photometry and single-mode photometry is at most <5% for each object.

After the construction of the AKARI band-merged catalog, we matched the AKARI photometry catalog with catalogs of previously known cluster member galaxies. In order to identify the membership for A2255, we use the spectroscopic redshifts from the SDSS DR2 (Abazajian et al. 2004) and BATC photometric redshifts by Yuan et al. (2003). Yuan et al. (2003) constructed a cluster member catalog that consists of 214 spectroscopic members and 313 photometric members out to ∼3 Mpc from the cluster center. Since all the AKARI sources fall within the coverage of this catalog, we use the Yuan et al. (2003) cluster member catalog (hereafter the Y03 catalog) as a basis of our analysis in the following sections.

The SDSS spectroscopic targets are selected down to r ≲ 17.77 mag after correcting for Galactic extinction (Strauss et al. 2002), therefore all members with spectroscopic redshifts are brighter than 17.77 mag in the r band. The completeness of the cluster member selection with spectroscopic redshifts at this magnitude range is on average ∼85%, although it depends on the magnitudes and the clustercentric distances of the galaxies (see Figure 1 in Park & Hwang 2009). The addition of cluster members with photometric redshifts corrects this incompleteness. Note that the number of member galaxies with photometric redshifts brighter than r = 17.77 mag is 40. The magnitude limit for the BATC photometry is ∼20 mag in each BATC filter, and the BATC photometric redshift identification is nearly complete for galaxies brighter than V < 19 mag (Yuan et al. 2003). This magnitude corresponds to r < 18.6–18.9 mag for different types of galaxies (c.f. galaxy colors in various photometric band systems; Fukugita et al. 1995). Therefore, the cluster member selection is complete down to r < 18.6–18.9 mag in terms of both the redshift identification and the member identification. Above this limit, there are more than ∼100 member galaxies with 18.9 mag < r < 20 mag. We expect a significant incompleteness in cluster member selection for this magnitude range. Although it is difficult to estimate the exact incompleteness at this faint end, this sample incompleteness would not cause much problem considering the relatively high MIR detection limit compared to the optical limit. We describe the effect of the incompleteness on our analysis in more detail in Section 4 (also, please refer to Figure 8). The redshift cut for A2255 member galaxies is 0.068 < z < 0.090, and the median redshift of A2255 member galaxies is 〈z〉 = 0.081.

The matching between the Y03 catalog and the AKARI IR source catalog is done using a matching radius of 5'' (roughly ∼1.5× FWHM radius of the AKARI N3 image, close to FWHM in other bands). We checked the optical and AKARI images of individual member galaxy to confirm that the matched object is not blended or mis-matched with neighboring sources. Most (∼95%) of the member galaxies are not blended in MIR. For 5% of galaxies that are suspected to be blended in MIR, we assign the "−1.00" value in the final photometry table (Table 2). Galaxies outside the survey coverage are also assigned "−1.00," and these are not included in the following analysis. If the galaxy is within the survey coverage and has the flux below the detection limit, we assign "99.00." We identify 122 spectroscopically confirmed member galaxies and 170 member galaxies selected by photometric redshifts (292 in total) in the Y03 catalog which have AKARI MIR counterparts, over ∼1200 arcmin² of the AKARI observation.

A2255 is observed by Spitzer MIPS 24 μm and 70 μm at 5σ flux limits of 250 μJy and 5 mJy, respectively (PID 40562, PI: G. Rieke). Since the coverage of MIPS 24 μm is ∼1500 arcmin², and it includes the region not covered by AKARI (see Figure 11(b) for comparison of the AKARI IRC and the Spitzer MIPS coverages), we derive the MIPS 24/70 μm photometry for cluster member galaxies within the Spitzer MIPS coverage. The inclusion of 24/70 μm-detected galaxies increases the number of member galaxies detected in MIR.

We run SExtractor over pbcd (post-Basic Calibrated Data) MIPS images to measure 24/70 μm fluxes. We use FLUX_AUTO of each object as total flux like we did in the case of the AKARI IRC bands. The FLUX_AUTO is measured at the previously defined optical coordinates of galaxies (i.e., the Y03 catalog) using ASSOC parameters. All AKARI-Y03 matched members have either 24/70 μm flux or upper limits, and there are 134 additional member galaxies in the Y03 catalog that lie within Spitzer MIPS coverage and outside the AKARI coverage.

In Table 2, we present MIR photometry of A2255 member galaxies from the AKARI and Spitzer observations. The table contains photometry for 426 member galaxies from the Y03 catalog: 292 galaxies within AKARI IRC fields, 134 galaxies within Spitzer MIPS coverage but outside the AKARI IRC coverage. The columns in the table are galaxy id (with coordinate information), spectroscopic or photometric redshift, AKARI IRC flux (3, 4, 7, 11, 15, and 24 μm), and MIPS 24 and 70 μm flux. As mentioned in Section 2.2, the value of −1.00 in the magnitude column indicates that the object is not covered by the corresponding filter, and 99.00 indicates that the flux of the object is below the detection limit.

In this section, we describe other wavelength data used in this study: optical imaging and spectroscopy, UV imaging, radio, and X-ray imaging.

In the optical wavelengths, we use the ugriz SDSS photometry as well as the 13-band photometry (from UV to i band) in Yuan et al. (2003). These optical photometric points are used to construct the spectral energy distributions (SEDs) of cluster member galaxies (see Section 3.3 for details). We also use the spectroscopic redshift from the SDSS database for membership identification. The line equivalent widths and the stellar metallicities are taken from the SDSS-MPA catalog¹¹ and from Gallazzi et al. (2005).

We also used the UV flux information which provides another measure of SFRs (see Section 4.1 for details). We queried the Galaxy Evolution Explorer (GALEX) source catalog from the all sky survey and the nearby galaxies survey¹² (tile UVE_A2255) over ∼1 deg² of A2255. We found 313 matches within 10'' from the optical coordinates of A2255 member galaxies within the MIR coverage (Note that the total number of member galaxies with the MIR data is 426; Section 2.3). For 76 galaxies that are used in the comparison between UV- and IR-derived SFRs (Section 4.1), we visually inspected the UV images to confirm whether the cross-identification is correct and the UV flux is not contaminated by nearby sources. The depths in FUV (1500 Å) and NUV (2300 Å) wavelengths are roughly ∼20 μJy at the 5σ flux limit.

The radio flux is also used in the comparison between the different SFR indicators (Section 4.1). We use the 1.4 GHz radio source catalog of Miller & Owen (2003) and cross-identify the sources with IR-detected cluster members using a matching radius of 5''. The number of matched members within MIR FOV is 33. Among these, 20 galaxies are classified as star-forming galaxies, 3 as Seyfert galaxies, and the remaining 10 galaxies as AGN candidates with old stellar population (Miller & Owen 2003; the classification is based on the optical spectra of galaxies).

Finally, we use the X-ray point source catalog obtained with Chandra (Davis et al. 2003) and cross-identify eight sources using a matching radius of 5''. These sources overlap with radio sources (Miller & Owen 2003) described above. We present a separate section about AGNs in Section 3.5.

It is well known that cluster member galaxies show a clear red sequence in their color–magnitude relation defined by elliptical galaxies, reflecting the similar age and/or metallicity of galaxies (e.g., Gladders et al. 1998). We investigate the color–magnitude relation of A2255 member galaxies at different wavelengths—optical, NIR, and MIR. Figure 2 represents the color–magnitude diagram of A2255 member galaxies in g−r, N3 − N4, and N3 − S11, from top to bottom.

Zoom In Zoom Out Reset image size

In the optical and NIR (Figures 2(a) and (b)), there is a tight sequence of galaxies in the color–magnitude diagram. In the optical color–magnitude diagram, a tight sequence made by bright, passively evolving member galaxies can be seen. The same sequence appears in r versus N3 − N4 color–magnitude relation, but in a reversed way that the brighter galaxies have the bluer N3 − N4 colors (Figure 2(b)). We define the "red sequence" in the optical color–magnitude diagram to divide the member galaxies into two groups: (1) red-sequence galaxies and (2) non red-sequence galaxies. All the member galaxies with spectroscopic redshifts in the non-red-sequence category show nebular emission lines indicative of ongoing star formation, thus non-red-sequence galaxies are thought to be star-forming galaxies while red-sequence galaxies are not. The color–magnitude relation is derived using a linear fit to r versus the g−r relation for r < 17.5 mag galaxies (see the solid line in Figure 2(a)), by excluding the outliers iteratively based on the bi-weight estimator. The fitted color–magnitude relation is as follows:

$$\begin{equation} g - r = -0.037\times r + 1.53. \end{equation} \tag{ 1 }$$

The standard deviation of residuals to this fit is σ_rms = 0.08 mag, indicating the tightness of the optical red sequence. This value for scatter includes the photometric errors, and the slope in color–magnitude relation is consistent with that derived in other studies (e.g., Gallazzi et al. 2006). We consider galaxies to belong to the red sequence if they lie within Δ(g − r) < 2σ_rms from the red sequence, where Δ(g − r) is an offset of color from the fitted relation. In Figure 2, circles indicate optical red-sequence galaxies and clovers indicate non red-sequence galaxies. While the optical red sequence is produced by galaxies with similar ages and metallicities, the same objects form an NIR "blue" sequence since we are sampling the Rayleigh–Jeans tail of the black body radiation with the N3 − N4 colors (e.g., Lacy et al. 2004; Stern et al. 2005). In NIR, the non-red-sequence galaxies in the optical have the redder N3 − N4 colors than the N3 − N4 blue sequence—possibly due to the dust emission arising from star formation.

In contrast, the tight sequence deteriorates in the color–magnitude relation in MIR, e.g., in the r versus N3–S11 color–magnitude diagram (Figure 2(c)). The spread in N3–S11 colors is much larger than those in optical or NIR, even if the outliers (clovers) from the optical red sequence are excluded. The linear relation derived from the r versus N3–S11 diagram (solid line in Figure 2(c)) using the same method as the optical is

$$\begin{equation} \begin{array}{l} N3-S11 = 0.106\times r - 3.18. \end{array} \end{equation} \tag{ 2 }$$

The scatter around this relation is σ_rms = 0.44 mag, roughly six times larger than that in the optical. The large scatter of σ_rms = 0.44 mag in the MIR blue sequence cannot be fully accounted for even if a maximum photometric error of ∼0.3 mag in the S11 band is assumed.

In addition to the large scatter, there are two interesting characteristics in the N3 − S11 color–magnitude diagram. The first is that there is a weak "blue" sequence of galaxies, although the scatter is very large as described above. Contrary to the case of optical color–magnitude relation, the brighter galaxies have bluer MIR color. The second is that most of the galaxies detected in the S11 band show redder N3 − S11 colors than expected from stellar radiation alone (N3 − S11 ∼ −2.0; dashed line in Figure 2(c)). The value of N3 − S11 ∼ −2.0 is calculated using a model of Piovan et al. (2003), where only stellar photospheric emission is considered—i.e., no dust continuum is taken into account. Not only galaxies that are outliers of N3 − S11 blue sequence, but also galaxies in the blue sequence are considered to have "MIR-excess" compared to the stellar continuum. The possible origins of the MIR-excess in these galaxies are the dust emission related to the star formation, AGN activity, and the circumstellar dust shells around AGB stars (e.g., Ko et al. 2009), etc. We investigate these possibilities by comparing MIR colors with galaxy model expectations in the following section.

In order to understand origins of the MIR excess in more detail, we present an MIR color–color diagram of A2255 member galaxies in Figure 3(a). The MIR colors used here are N3 − S7 versus N3 − S11, and the expected colors from various model galaxy templates are overplotted. The overplotted color tracks are calculated using models of elliptical galaxies with different ages and metallicities which include the dust emission from circumstellar dust around AGB stars (Piovan et al. 2003) or SED templates of local star-forming galaxies and IR luminous galaxies (Chary & Elbaz 2001).

Zoom In Zoom Out Reset image size

The plot shows that galaxies in the N3 − S11 blue sequence (〈N3 − S11〉 ∼ −1.7) are likely to be passively evolving galaxies with stellar population age around 10 Gyr or larger. The reddest in both MIR colors (N3 − S11>0.2 mag and N3 − S7>0 mag) are star-forming galaxies. Galaxies between two populations are dominated by young, passively evolving galaxies with stellar ages between 1 and 10 Gyr, yet these could also be galaxies with a small amount of star formation.

Based on the color tracks of model galaxy templates (Figure 3(a)), we classify MIR-excess galaxies into three classes according to the amount of MIR excess. The reddest galaxies, having N3 − S11>0.2 mag are defined as "strong MIR-excess," where the MIR color corresponds to that of an actively star-forming galaxy calculated using SED templates of Chary & Elbaz (2001). Galaxies at the N3 − S11 blue sequence are defined as "weak MIR-excess" galaxies (N3 − S11 < −1.2 mag), which have MIR colors of passively evolving galaxies with old ages. The objects between the two populations are tagged as "intermediate MIR-excess" galaxies whose MIR colors can have multiple origins such as circumstellar dust emission from intermediate age stars, residual star formation, and AGN activity. These three classifications based on the MIR color allow us to investigate galaxy populations at different evolutionary stages. The environmental dependence of galaxies is discussed in the following sections based on these classifications.

These MIR-excess terms are defined using N3 − S11 colors and thus limited by N3 and S11 coverage (Figure 1) and depth. Therefore, we develop another criteria to define MIR excess using 24 μm flux from Spitzer MIPS, which covers a larger area than the S11 coverage. Figure 3(b) shows the correlation between N3 − S11 and mag(z)−mag(24 μm) for the observed galaxies, in addition to the expected model colors. Two MIR colors, N3 − S11 and mag(z)−mag(24 μm) correlate reasonably well, and the locations of model tracks are consistent with the case of N3 − S7 versus N3 − S11. Therefore we use mag(z)−mag(24 μm) to define MIR-excess galaxies in addition to N3 − S11. The corresponding criteria for strong/intermediate/weak MIR excess are, mag(z)−mag(24 μm) > − 0.5, −2.0< mag(z)−mag(24 μm) <−0.5, and −3.5< mag(z)−mag(24 μm) <−2.0, respectively.

In Table 3, we compare the number of cluster member galaxies with or without MIR detection, and MIR excess. The fractions of MIR-excess galaxies among cluster members at the same r-band magnitudes are different for the AKARI S11 and Spitzer 24 μm fields, e.g., 74% for S11 and 37% for 24 μm at r < 17.5. This difference is due to the shallower depth of the 24 μm image compared to the S11 image. As we showed in Figure 2(c), S11 flux limits or 24 μm flux limits place limits on the MIR color that can be considered as MIR excess. With 5σ flux limit of 80 μJy for S11 band, weak MIR excess galaxies are complete only at r < 17.5, while intermediate and strong MIR-excess galaxies are complete down to r ∼ 18.5 mag. On the other hand at 24 μm with flux limit of 250 μJy, the complete limit for weak MIR-excess galaxies is r < 15 mag, and the limit for intermediate MIR-excess galaxies is r < 16.5 mag. Since our survey consists of fields covered with different filters having different depths, we take this different MIR-excess fraction into account when discussing the spatial distribution of MIR-excess galaxies (Section 5).

Table 3 as well as Figure 2 shows that the fraction of MIR-excess galaxies is very high (>70% at r < 17.5 mag, using S11), contrary to the general belief that there are little dust and gas among cluster member galaxies. Many of these MIR-excess galaxies are on the optical red sequence (weak and intermediate MIR-excess galaxies; see Figure 2(c)), so that it is essential to include AGB circumstellar dust emission when describe the SED of red cluster galaxies from optical through MIR (Bressan et al. 2007). This is also consistent with the trend in other galaxy cluster, A2218, in which a significant MIR-excess is seen in fainter early-type member galaxies (Ko et al. 2009).

For the MIR-excess galaxies defined in Section 3.2, we calculate either total IR luminosity (i.e., IR-derived SFR) or stellar population age based on the SED fitting. The derivation of these quantities allows us to investigate star formation activities/quenching sequence and its relation to the environment in A2255. The SED fitting follows the procedure described in Shim et al. (2007) except that we include the photometric data from optical (SDSS ugriz) to MIR (AKARI IRC points and Spitzer MIPS 24/70 μm points if available). The template library we used is IR galaxy templates with different IR luminosities (Chary & Elbaz 2001), and the early-type galaxy templates from Piovan et al. (2003). The IR galaxy templates are constructed through empirical interpolation between the observed local IR galaxies. As it is mentioned in Chary & Elbaz (2001), the optical–NIR part of the SEDs are arbitrarily determined to match the IR versus optical luminosity ratio, and thus we used MIR (3–70 μm) data points only to fit the IR SEDs. The early-type galaxy templates are constructed considering the effect of circumstellar dust around AGB stars to the integrated spectrum in addition to the photospheric emission. The best-fit template is found using χ² minimization between the model fluxes and the observed fluxes.

Figure 4 shows the examples of the SED fitting for 34 relatively bright galaxies in optical (r < 16.5 mag, i.e., M_r < −21.0 mag). These SED panels are arranged in descending order of N3 − S11 values. Since these galaxies are sufficiently bright and large, we also present their color-composite stamp images in Figure 5 as a guide to their morphologies. The images are color composites of SDSS g (blue), r (green), and i (red) band inverse images. The size of each cutout is 30'' × 30'', and the images are displayed in logarithmic scale. The objects in Figure 5 are aligned with the same order as Figure 4. At r > 16.5 mag, it becomes difficult to investigate the morphologies using the SDSS images unless they are sufficiently extended. The last object in Figures 4 and 5 does not fall in the magnitude cut for optically bright objects, yet the object is included as an example of member galaxies with known AGNs showing MIR-excess.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The MIR (N3 − S11) colors are marked in the lower right of each cutout image in Figure 5. The first seven galaxies are strong MIR-excess galaxies, and the eighth galaxy is an intermediate MIR-excess galaxy. These eight galaxies are best-fitted with IR galaxy templates (Chary & Elbaz 2001) and show late-type morphology or disky features. The ninth galaxy is an intermediate MIR-excess galaxy, best-fitted with an early-type galaxy template. All the remaining galaxies are weak MIR-excess galaxies, which are best-fitted by early-type galaxy templates of various ages and metallicities. Although there are two free parameters—metallicity and age—for early-type galaxy templates, we only use age as a meaningful parameter from fitting since the choice of metallicity is very limited (Z = 0.004, 0.008, and 0.02). Again, while the age itself is a model-dependent parameter (thus, there exists an "unphysical" age which is larger than the age of universe), we see that the amount of MIR excess is mainly affected by the age of a galaxy.

In Section 3.1 (Figure 2), we show that there are a large number of galaxies that show MIR excess while lying on the tight optical red sequence at the same time. They fall on to the "intermediate" (−1.2 < N3 − S11 < 0.2) or "weak" (−2.0 < N3 − S11 < −1.2) MIR-excess categories (Figure 3). To check the origin of the weak/intermediate MIR excess, and provide a physical meaning of the intermediate MIR-excess galaxies, we investigate various properties of this population.

First, we checked the optical spectra of the intermediate MIR-excess galaxies with r-band magnitude brighter than 17.77 mag. We find no sign of emission lines, which indicates that the MIR excess of most of these galaxies is not due to the star formation activity (at least within the fiber aperture).

We also examined the relation between MIR excess and other spectral properties of red-sequence galaxies (including weak MIR-excess galaxies). The top panel of Figure 6 shows the N3 − S11 colors versus the luminosity-weighted mean stellar ages derived from the SED fitting (see Section 3.3). The second panel from the top illustrates the N3 − S11 colors versus the metallicities measured from the SDSS spectra (Gallazzi et al. 2005), next is the N3 − S11 colors versus the Hβ absorption line equivalent widths (Gallazzi et al. 2005), and the final panel shows the N3 − S11 colors versus D_n4000 (Kauffmann et al. 2003). The SED-fitted ages are derived by fixing metallicity to the solar value, which should be a good approximation judging from Figure 6(b). Due to the limitation in modeling, there are "unphysical" ages that are larger than the age of the universe (∼17 Gyr). The absolute age of the SEDs should not be taken too seriously, as these models are not meant to provide absolute ages (Piovan et al. 2003). These model fit parameters are meant to provide the relative age scales represented by different SED shapes. We find that the ages of the intermediate/weak MIR-excess galaxies derived by the SED fitting correlate with the N3 − S11 colors (Figure 6(a)), with Spearman's rank correlation coefficient of −0.71. Considering the degree of freedom at 38 (N−2, the number of points used is 40), the correlation is reliable by more than 99.9% (Zar 1972). The correlation is already expected by the model color tracks overplotted in Figure 6(a). On the other hand, there is little correlation between metallicity and the N3 − S11 color (Figure 6(b); r_s = 0.008, consistent with null hypothesis). Since the N3 − S11 colors are more sensitive to age than metallicity, the MIR excess can be used to break the old age–metallicity degeneracies (e.g., Ko et al. 2009). Figures 6(c) and (d) show the relation between MIR excess and other well-known age indicators, the equivalent widths of Balmer absorption line (Hβ) and D_n(4000), a measure of the strength of the 4000 Å break (Kauffmann et al. 2003). Spearman's rank correlation coefficient in this case is r_s = −0.07 and −0.14, i.e., significance level (of the rejection of null hypothesis) greater than 50% (degree of freedom at 38) and 40% (degree of freedom at 18), respectively. Thus, the correlation between the MIR color and Hβ, D_n(4000) is relatively weak.

Zoom In Zoom Out Reset image size

Figure 7 illustrates the discrepancy between the different age indicators more clearly. The correlation coefficients between the SED-fitted age and age indicators are r_s = 0.14, − 0.70, and −0.18 for D_n(4000), N3 − S11, and the Hβ equivalent width, respectively. These represent significance levels smaller than 40%, larger than 99%, and smaller than 70%, respectively. Only N3 − S11 shows reliable correlation with the stellar age derived through SED fitting. This implies that the estimation of mean stellar age is not an easy task. Each age indicator has its pros and cons: Hβ is known to be less sensitive to metallicity compared to a simple color index such as B−V as viewed from the stellar population synthesis modeling, yet it is also easily affected by emission line component. D_n(4000) is known to be a good tracer for young stellar age (<1 Gyr, D_n(4000) < 1.5; Balogh et al. 1999), yet highly affected by metallicity for older age (Kauffmann et al. 2003). Moreover, both Hβ and D_n(4000) is limited by the finite fiber size used for taking the SDSS spectra (3'' diameter) that only samples the central part of a galaxy if it is extended. We show morphologies of five galaxies with the mean stellar ages from the SED fitting younger than 5 Gyr (Figure 7). Most of the galaxies have outer disks and late-type morphology, and finite SDSS fiber size sample only the central regions indicated as a 3'' bar in Figure 7. Therefore, we address that MIR-excess is a relatively good age indicator that is free of finite spectroscopic aperture and age–metallicity degeneracy.

Zoom In Zoom Out Reset image size

As the morphologies of five young galaxies (<5 Gyr) suggest (Figure 7), the fraction of late-type morphologies is large for "intermediate" MIR-excess galaxies (−1.2 < N3 − S11 < 0.2). We examine the morphologies of A2255 member galaxies at the bright end (r < 16.5 mag; Figure 5), since the visual morphological classification becomes difficult at the fainter magnitudes. While seven out of seven strong MIR-excess galaxies display late-type galaxy morphology, only five out of 25 (20%) weak MIR-excess galaxies show late-type morphology which can easily be understood as a consequence of the well-known correlation between color and morphology of galaxies. In the case of the intermediate MIR-excess galaxies (S171251.20+640423.0 and S171225.70+641946.0 in Figure 5 and galaxies in Figure 7), they show late-type morphology such as disks despite being at the optical red sequence. The natural explanation of the colors and the morphologies of the intermediate MIR-excess galaxies is that these galaxies are late-type galaxies with small amount of star formation, very possibly in the process of quenching star formation. Studying E+A galaxies (i.e., characterized by old stellar population and strong Balmer absorption) would be another way to see the truncation of star formation, yet we did not find any E+A galaxies using a criteria of EW(Hδ) >5 Å (Goto 2007, in addition to the criteria for no detectable [O ii] and Hα emission lines) among the member galaxies with spectroscopic information. The average EW(Hδ) of the intermediate MIR-excess galaxies is ∼1 Å, with a typical measurement uncertainty of 0.5 Å. Only ∼50% of the intermediate MIR excess galaxies are spectroscopically observed, thus it is not clear whether the remaining ∼50% would be classified as E+A galaxies or not. This result demonstrates an advantage of using MIR excess in studying the transition population over optical spectra. The "weakened" star formation activity in intermediate MIR-excess galaxies, regardless of its origin (weak star formation or the AGB-dust), is naturally explained by the idea that these galaxies are in the process of transformation from star-forming galaxies (strong MIR-excess galaxies) to quiescent galaxies (weak MIR-excess galaxies) at optical red sequence. The relation with these transformation and the cluster environment will be described in more detail in Section 5.3.

AGNs cause MIR excess in galaxies (Quillen et al. 1999) as well as ongoing star formation, and thus alter the derived SFRs of galaxies. The different diagnostics for AGNs, such as infrared color and optical line ratio, produce a significantly different result for demarcating AGNs and star-forming galaxies among MIR-excess galaxies (Brand et al. 2009). Therefore, we do not force the differentiation of AGNs from our MIR-excess galaxies in this paper. Instead, we briefly discuss how many AGNs are included in MIR-excess galaxies and how they affect our analysis and conclusion, using the previously known AGNs detected in either X-ray or radio, classified as AGNs using optical line diagnostics (Miller & Owen 2003; Davis et al. 2003).

In the AKARI/Spitzer FOV, there lie 13 previously known AGNs. Among the 13 AGNs, 10 are detected in either 11 μm (S11) or 24 μm (L24/MIPS 24 μm) images, and 3 are not. Ten AGNs detected in 11 μm or 24 μm are radio sources with 1.4 GHz flux larger than 0.35 mJy, while three are Seyferts and seven are objects with old stellar population either with weak [N ii] or [S ii] emission lines, or without any signs of emission lines (Miller & Owen 2003). By number, AGNs contribute 3% of the star-forming galaxies, and 15% of the weak MIR-excess galaxies. No AGN cross-identification is found in intermediate MIR-excess galaxies. The contribution from three Seyfert galaxies to the total SFR is less than 5 M_☉ yr⁻¹, i.e., less than 4% of $\sum _{>0.25 \,M_\odot \,{\rm yr}^{-1}} \mbox{SFR}$ (see Section 4.2 for details). With this little contribution, we conclude that our main analysis and conclusion is not affected by the presence of AGNs among MIR-excess galaxies especially in the sense that we mainly discuss star-forming galaxies and intermediate MIR-excess galaxies in A2255.

We calculate SFRs of member galaxies from the derived IR luminosity through the SED fitting (Section 3.3, Figure 4), and compare the SFR with those from other wavelengths. The total IR luminosities of A2255 member galaxies range from 1.0 × 10⁸ to 3.2 × 10¹⁰ L_☉, while the median value is 〈L_IR〉 ≃ 2.8 × 10⁹ L_☉. These values correspond to ∼0.1 M_☉ yr⁻¹ < SFR(IR) ≲ 7.0 M_☉ yr⁻¹ when converted using the Kennicutt (1998) relation between L_IR and SFR. However, considering the flux limits in MIR and cluster member completeness limit in optical, the conservative limit in L_IR in our survey is L_IR > 1.5 × 10⁹ L_☉ (Figure 8). We do not find any luminous IR galaxy (L_IR > 10¹¹ L_☉) candidates in A2255.

Zoom In Zoom Out Reset image size

We compare the "IR-derived SFR" with SFRs derived using other star formation indicators (Figure 9) at different wavelengths. Figure 9(a) shows the comparison between SFR_IR and SFR_Hα(cor.), i.e., IR-derived SFR and the Hα-derived SFR corrected for dust extinction ("cor." means the luminosity is corrected for internal dust extinction). In order to derive SFR_Hα(cor.), we first derive SFR_Hα(uncor.), SFR based on Hα flux not corrected for the internal dust extinction, using Hα fluxes from the SDSS-MPA catalog (see Section 2.4) and the Hα SFR conversion formula of Kennicutt (1998). Since the SDSS spectra are obtained using fibers with a fixed size of 3'' diameter, we adopt the aperture correction for each object from Brinchmann et al. (2004), i.e., f_total/f_fiber.¹³ The "extinction" in Hα flux—A(Hα)—is estimated from the Balmer decrement (Hα/Hβ flux ratio), by adopting a case B recombination at T = 10, 000K (Osterbrock 1989) and the assumption of the Calzetti et al. (2000) extinction law for starburst galaxies. The median amount of extinction in Hα is 〈A(Hα)〉 = 0.78 mag (Figure 9(b)). There is a weak correlation between SFRs and A(Hα), i.e., galaxies with larger SFRs having larger A(Hα).

Zoom In Zoom Out Reset image size

We also compare the UV-derived SFR with the IR-derived SFR (Figure 9(c)), using the GALEX NUV flux as a measure of SFR_UV. As in the case of Hα, we derive the extinction correction in UV wavelengths from the UV slope β inferred by FUV − NUV colors (Meurer et al. 1999). Star-forming galaxies in A2255 show relatively blue FUV − NUV colors of 〈FUV − NUV〉 = 0.55 mag (Figure 9(d)). The mean extinction correction derived is 〈A(UV)〉 ∼ 1.2 mag. When this correction is applied (SFR_IR versus SFR_UV(cor.); Figure 9(c)), the two SFRs are comparable with each other. Both the Hα- and the UV-derived SFRs agree with the IR-derived SFR when corrected properly for the internal dust extinction, albeit with a large scatter (> × 2).

Finally, in Figure 9(e), we present the comparison between SFR_IR and SFR_radio using 1.4 GHz flux in Miller & Owen (2003). The conversion formula from 1.4GHz flux to SFR_radio is adopted from Bell (2003). Although the number of the matched member galaxies is small, SFRs from the two indicators differ by about a factor of two, with the SFR_IR being systematically larger than SFR_radio.

Overall, we find a conventional extinction correction to the dust extinction works roughly well for estimating SFRs in individual galaxies (e.g., Choi et al. 2006), but such correlations accompany large scatters. Therefore, we consider the usage of the IR data is a robust way to derive the star formation compared to UV and the optical data.

In this section, we derive the global properties of star formation activity in A2255, in order to examine if the merging activity enhanced the star formation activity of A2255 in particular. This is done by deriving the "total" SFR in A2255 by constructing the IR luminosity function and comparing the total SFR of A2255 with other galaxy clusters. The construction of the luminosity function is done using the following equation:

$$\begin{equation} \phi (\log \, L_i) = \frac{1}{A} \frac{n}{\Delta (\log \, L)}, \end{equation} \tag{ 3 }$$

where A is the surveyed area, n is the number of galaxies whose luminosity falls within the ith bin, and Δ(log  L) is the luminosity bin size. We used the survey coverage of ∼2000 arcmin², i.e., ∼16 Mpc² in physical scale.

We present the derived IR luminosity function with the Poisson error bars in Figure 10(a). The overplotted solid line is the best-fit Schechter luminosity function for Coma and A3266 composite (Bai et al. 2009), with parameters α = −1.41 and log L*_IR(L_☉) = 10.49. The IR luminosity function of A2255 is consistent with those of Coma and A3266 in terms of both the shape and the normalization above the completeness limit of IR luminosity (L_IR ∼ 1.5 × 10⁹ L_☉; Figure 8).

Zoom In Zoom Out Reset image size

Comparing the IR luminosity functions of Coma and A3266, Bai et al. (2009) suggested a possibility that local galaxy clusters share a universal IR luminosity function. A2255 is close to the Coma cluster in terms of cluster mass (∼10¹⁵ M_☉; Burns et al. 1995; Lokas & Mamon 2003), but has a slightly lower mass than A3266 (∼3.3 × 10¹⁵ M_☉; Bai et al. 2009). In terms of the dynamical status, A2255, A3266, and the Coma cluster all share similar properties—these galaxy clusters are suggested to be in the late phase of the cluster–cluster merger (e.g., Watanabe et al. 1999; Sauvageot et al. 2005). The consistency of the A2255 IR luminosity function with those of Coma and A3266 supports the idea of universal IR luminosity function in local galaxy clusters (e.g., Bai et al. 2009), at least when they are in the late stage of cluster-scale merger. Yet it should be noted that the number of galaxy clusters studied is small, and most galaxy clusters studied to date are biased to a dynamically unrelaxed system.

With the derived IR luminosity function, we calculate the "total" SFR in A2255 by integrating the IR luminosity function. In the luminosity range of L_IR > 1.5 × 10⁹ L_☉, i.e., SFR(IR)>0.25 M_☉ yr⁻¹, the integration of the IR luminosity function yields $\sum _{>0.25 \,M_{\odot } \,{\rm yr}^{-1}}^{\mbox{LF}} {\rm SFR} = 115 \,M_{\odot }$ yr⁻¹ for A2255. This is consistent with the summation of SFRs of "individual" member galaxies, $\sum _{>0.25 \,M_{\odot } \,\rm yr^{-1}} {\rm SFR} = 131 \,M_{\odot }$ yr⁻¹.

In the SFR summation of individual star-forming galaxies ($\sum _{>0.25 \,M_{\odot } \,\rm yr^{-1}} {\rm SFR} = 131 \,M_{\odot }$ yr⁻¹), we split the contribution to the SFR from spectroscopic member galaxies and photometric member galaxies in order to see the effects from the cluster membership determination. Among 131 M_☉ yr⁻¹, the member galaxies with spectroscopic redshifts contribute 103 M_☉ yr⁻¹, and the member galaxies with photometric redshifts contribute 28 M_☉ yr⁻¹. The contribution from galaxies with photometric redshifts (i.e., optically faint member galaxies) is about 20%: therefore, even at the very unlikely case that all photometric members turn out to be non-members, it does not have a strong effect on the derived total SFR.

In order to compare the star formation activity of A2255 with the other galaxy clusters, we use a "specific SFR" (SSFR) of the galaxy cluster—the total SFR normalized by the cluster virial mass (∑SFR/M_cl). The result is marked as a filled star in Figure 10(b). The ∑SFR/M_cl value is calculated by integrating SFRs of galaxies with SFRs larger than 2 M_☉ yr⁻¹ and within a 0.5r₂₀₀ radius, in order to facilitate the comparison with the mass-normalized SFR of other clusters (other points; Bai et al. 2007, 2009). The points for Coma, A3266, MS1054-03, and RX J0152 (open triangles in Figure 10(b)) are from the Spitzer MIPS 24 μm studies (Bai et al. 2007, 2009), while the points for A2218, A1689, A2219, and Cl0024+16 are from ISO observation (Bai et al. 2007). The sum of the A2255 SFR is ∑SFR_IR(>2 M_☉ yr⁻¹, within 0.5 r₂₀₀) = 15.5 M_☉ yr⁻¹. We use r₂₀₀ = 2.10 Mpc and M_cl = 0.45–1.3 × 10¹⁵ M_☉ from previous studies (Neumann 2005; Burns et al. 1995; Feretti et al. 1997). The derived SSFR of A2255 is comparable to those of four other clusters at z < 0.2. On the other hand, compared to two intermediate-redshift clusters, A2219 and Cl0024+16, the SSFR of A2255 is more than an order of magnitude lower. From the study of the total SFR in A2255, we find no evidence that the overall star formation activity is enhanced in A2255 compared to other clusters at low to intermediate redshifts.

We use the local surface density of galaxies and clustercentric radius as indicators of environment. The local surface density of galaxies is expressed as Σ_5th, the density of member galaxies (with either spectroscopic or photometric redshifts) within a circle whose radius is the distance to the fifth nearest galaxy in comoving scale. Galaxies brighter than r ∼ 19 mag are used for calculating the local density. If a galaxy cluster is simply a relaxed system, the local surface density would monotonically decrease as the clustercentric distance increases. However, for clusters like A2255, this is not the case. We identify substructures with densities higher than the local average densities to be areas that deviates from a smooth projected number density represented by the Navarro–Frenk–White (NFW) function (Navarro et al. 1997). To fit the relation between the clustercentric distance versus the observed Σ_5th, we use a two-dimensional projected form of the NFW function (Elíasdóttir & Möller 2007; see the solid line in Figure 11(a)):

$$\begin{equation} \begin{array}{l} \Sigma _{\rm nfw} = \dsty\frac{2 r_s \delta _c \rho _c}{1 - X^2} \left(1 - \dsty\frac{2}{\sqrt{1-X^2}} \mbox{arctanh} \sqrt{\frac{1-X}{1+X}} \right), X<1 \\ \ms \Sigma _{\rm nfw} = \dsty\frac{2 r_s \delta _c \rho _c}{X^2 - 1} \left(1 - \dsty\frac{2}{\sqrt{X^2-1}} \mbox{arctan} \sqrt{\frac{X-1}{1+X}} \right), X>1 \end{array} \end{equation} \tag{ 4 }$$

Zoom In Zoom Out Reset image size

In this equation, X indicates R_cl/r_s, where R_cl is the clustercentric radius. The parameters r_s, δ_c, and ρ_c represent the scale radius, the characteristic overdensity, and the critical density. Our best-fitted parameters are r_s = 0.68 Mpc and δ_cρ_c = 127 Mpc⁻³. The galaxies above σ_rms from the fitted line are considered as galaxies at substructures with a locally high density of galaxies. The coordinates of these galaxies are plotted as open circles in Figure 11(b), over the distribution of all member galaxies in the Y03 catalog. The overplotted thick lines are contours of "galaxies at high density environment," and we define these peaks as substructures. This means that it is not necessary that all substructure galaxies have high Σ_5th, although most of them do. The locations of these structures are consistent with galaxy number density contour. The coordinates of the substructures marked in Figure 11(b) are (17:13:44, 64:14:42) for substructure "V," (17:14:48, 64:10:00) for "W," (17:13:10, 64:05:00) for "X," (17:12:20, 64:00:00) for "Y," and (17:11:40, 64:08:00) for "Z." Substructures A, B, C, D, and E are consistent with the substructures defined in Yuan et al. (2003) with the same alphabets. They first defined the substructures from the peak of number density contour (>0.15 galaxies arcmin⁻²; see Figure 5 of Yuan et al. 2003). By studying the velocity distribution of galaxies in the substructures, these authors confirm that substructures A, B, and C have different velocity components compared to the main cluster, while substructures D and E might be a result of a projection effect. Our MIR observation is limited to within the central ∼2 Mpc radius of A2255. Consequently, the substructures are all defined within this ∼2 Mpc radius circle (about virial radius of the cluster).

Besides the substructures, we also investigate the large-scale distribution of galaxies around A2255, since A2255 is known to be a member of the North Ecliptic Pole (NEP) supercluster (e.g., Mullis et al. 2001). Understanding the large-scale distribution of galaxies will be helpful to see how the distribution of galaxies in A2255 is connected to the features such as walls and filaments. Figure 12(a) shows the redshift distribution of galaxies located in a 3 × 3 deg² region surrounding A2255. The spectroscopic redshifts are obtained from SDSS DR7 (Abazajian et al. 2009). While the solid histogram shows the redshift distribution of all the galaxies within a circle with 10 Mpc radius centered on A2255, the filled histogram indicates the redshift distribution of galaxies within 2 Mpc distance from the cluster center. The large-scale analysis shows strong redshift peak at z ∼ 0.08 which can be attributed to the presence of A2255 and the NEP supercluster. On the other hand, in the central region within the clustercentric distance of 2 Mpc, the redshift distribution is far different from a single Gaussian distribution: the possibility that the distribution of galaxies located within the central 2 Mpc follows a single Gaussian function is only 40% by the Kolmogorov–Smirnov test. Instead, the redshift distribution is well described by the sum of two redshift components with z₁ = 0.077 ± 0.0027 and z₂ = 0.082 ± 0.0017, with a lack of galaxies at z ∼ 0.08. Figures 12(b)–(e) show the spatial distribution of galaxies at different redshift bins. The galaxies at the blue velocity component (b; 0.07 < z < 0.078) and the red velocity component (d; 0.081 < z < 0.084) are concentrated in the cluster center. The redshift range of 0.078 < z < 0.081, where only few of the A2255 cluster members reside in, consists of galaxies that are distributed like a flat sheet at z ∼ 0.08 (Figure 12(c)). At the reddest velocity tail (e; 0.084 < z < 0.09), galaxies are distributed along a filamentary structure that pass through A2255. The two velocity groups (0.07 < z < 0.078 and 0.081 < z < 0.084) are virtually superimposed in the sky. The velocity dispersion of two groups are 800 km s⁻¹ (blue) and 500 km s⁻¹ (red). Using simple virial theorem, the masses of each component are 1.7 × 10¹⁵ M_☉ (blue) and 6.7 × 10¹⁴ M_☉ (red) when virial radius is 2 Mpc for each group. According to the criteria for gravitationally bounded two-body problem (Beers et al. 1982; Tran et al. 2005), the probability that these two components are gravitationally bounded is when the following equation is satisfied:

$$\begin{equation} V_r^2 R_p \le 2 G M \sin ^2\alpha \cos \alpha. \end{equation} \tag{ 5 }$$

Zoom In Zoom Out Reset image size

In this equation, V_r is the relative velocity between the two groups (1500 km s⁻¹), R_p is the projected separation (less than ∼100 kpc since two components are nearly superimposed), M is the total mass of the system and α is the projected angle with respect to the plane of the sky. The equation is valid for the projection angle range of 6° < α < 89°, and thus the probability that the two groups are bounded is more than 90%. Therefore, A2255 is clearly thought to be a "merging" galaxy cluster with two different velocity peaks.

Figures 13(a) and (b) show the projected two-dimensional spatial distribution of star-forming galaxies in A2255. The color of each point indicates the redshift, and the size of each point is proportional to the SSFR. The figures show that galaxies with high SSFR are located preferentially in the outer region of the cluster (Figures 13(a) and (b)), reflecting the well-known morphology-density relation. There are a few galaxies with high SSFRs near the cluster center, but they have mostly redshifts much lower than the mean redshift of cluster (blue points in the center of Figure 13(a)). The R_cl–v plot of star-forming galaxies (Figures 13(c) and (d)) clearly confirms the trend that galaxies with high SSFR are either located at the outer region of A2255 or at the blue velocity peak.

Zoom In Zoom Out Reset image size

In order to analyze the distribution of galaxies with high SSFR in more detail, we investigate SSFR as a function of environmental parameters in Figure 14. Figure 14(a) shows R_cl versus SSFR plot. The average SSFR does not change much as a function of R_cl, i.e., there is no clear sign of enhancement in SSFR as a function of R_cl. Figure 14(b) shows how SSFR changes as a function of local galaxy surface density, Σ_5th. Previous works of intermediate redshift clusters suggested that SSFR is enhanced at the intermediate density region of clusters and the enhanced activity is related to the infall of galaxies into the gravitational potential of clusters (e.g., Koyama et al. 2010). For A2255, we find that there is no clear enhancement of SSFR from surface density of Σ_5th = 0.5–2.5 Mpc⁻². There is a slight increase in the lower density region, at Σ_5th = 0.5 Mpc⁻², yet the amount of difference is less than a factor of 1.5. Therefore, we conclude that star formation is not enhanced in dense regions of A2255.

Zoom In Zoom Out Reset image size

We also examine if star formation activities are enhanced at substructures as suggested from studies of other clusters (e.g., Koyama et al. 2010). Figure 14(c) shows the average SSFR for the substructures identified in Section 5.1, in comparison to galaxies in the other areas of the cluster. Errors for the average SSFR are the standard deviations of SSFR are derived through bootstrapping. We find that there is little difference between the average SSFR of galaxies in substructures (Sub) and those not in substructures (N). Several substructures (A, W, and Y) contain only one or no star-forming galaxy. We conclude that SSFR is not particularly enhanced in substructures even if some galaxies in substructures show high SSFR compared to global average.

Although we find no convincing evidence for the enhancement of SSFR at a particular environment (except that SSFRs are higher at the outer region), we find an interesting trend in the distribution of star-forming galaxies in the inner region. As seen in Figure 13(c), the redshift distribution of star-forming galaxies (filled histogram) is bimodal: a blue velocity peak at 〈z〉 = 0.077 and a red velocity peak at 〈z〉 = 0.082. (Note that this analysis is only possible for samples with spectroscopic redshifts since the accuracy of photometric redshifts is too low.) At R_cl < 0.5 Mpc, the majority of galaxies with high SSFR belong to the blue velocity peak. We show the average SSFR of galaxies at the blue velocity peak and at the red velocity peak separately in Figure 14(a). Figure 14(a) indicates that the average SSFR of galaxies at the blue velocity peak (z < 0.08) is higher than that of galaxies at the red velocity peak (z > 0.08) by a factor of ∼3, at the innermost region with R_cl < 0.5 Mpc. By randomly selecting and assigning galaxies at R_cl < 0.5 Mpc to either blue or red peaks, we find that the chance probability of finding this observed difference in the average SSFR is less than 5%. This trend appears to continue out to R_cl ∼ 1.0 Mpc, but the trend is not statistically significant compared to what we found at R_cl < 0.5 Mpc.

As we discussed in Section 3.4, intermediate MIR-excess galaxies are probes of galaxy transformation from star-forming galaxies (strong MIR excess) to quiescent galaxies (weak MIR excess). By studying the environment of intermediate MIR-excess galaxies, we can gain insight into where in the cluster the galaxy transformation is taking place and what physical mechanisms govern the transformation. The two-dimensional spatial distribution of weak/intermediate MIR-excess galaxies is plotted in Figures 15(a) and (b). The plots show that weak MIR-excess galaxies populate near the cluster center, while intermediate MIR-excess galaxies prefer R_cl > 0.5 Mpc. We show the trend in a more quantitative way in Figure 16, by presenting the number ratios of galaxies with different levels of MIR-excess as a function of the environmental parameters. Figure 16(a) illustrates the ratio of intermediate MIR-excess galaxies to weak MIR-excess galaxies as a function of the clustercentric radius. Figure 16(b) illustrates the same number ratio as a function of the local galaxy surface density. These are a clearer form of Figure 17, showing the relative fraction of strong/intermediate/weak MIR-excess galaxies as a function of environment, with emphasis on the weak and intermediate MIR-excess galaxies. Figures 17(b) and (d) clearly show that the weak MIR-excess galaxies dominate at the highest densities, and the strong MIR-excess galaxies dominate at the lowest densities. The intermediate MIR-excess galaxies populate at around Σ_5th ∼ 2.0 Mpc⁻² (R_cl ∼ 0.5 Mpc), but their number disappears quickly at Σ_5th < 1.5 Mpc⁻² (R_cl ∼ 1.6 Mpc). The relative fraction of intermediate MIR-excess to weak MIR-excess galaxies is the highest at the similar density/R_cl, Σ_5th ∼ 1.5 Mpc⁻² and R_cl ∼ 1.3 Mpc (Figures 16(a) and (b)). From this, we conclude that galaxies in transformation (as represented by intermediate MIR-excess galaxies) populate the intermediate density regions (or the clustercentric distance of R_cl = 0.6–1.6 Mpc). This clustercentric radius corresponds to 0.3–0.8R_vir, which exactly coincides with the location in clusters where the ram-pressure stripping is known to be efficient in quenching the star formation in galaxies by stripping their gas reservoir based on the sphere of influence argument (Treu et al. 2003). This is also consistent with the result found in other galaxy clusters (A3112; Braglia et al. 2010).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

As in the case of star-forming galaxies, a significant number of intermediate MIR-excess galaxies have redshifts at blue velocity component (Figure 15). The difference in the case of star-forming galaxies is that intermediate MIR-excess galaxies at blue velocity peaks are located at R_cl > 1 Mpc. We also examine the correlation between substructures and the fraction of intermediate MIR-excess galaxies (Figure 16(c)). As in the case of star-forming galaxies, there is no clear evidence that intermediate MIR-excess galaxies are preferentially located in the substructures. Note that, however, as we mentioned in Section 3.2, the observation of MIR-excess galaxies is limited by the depth of MIR images—especially at weak/intermediate MIR excess for regions without S11-band coverage. The substructures B, C, D, and E are such substructures, and thus the reason we do not find many weak/intermediate MIR-excess galaxies in these substructures may be due to the shallow depth of the 24 μm image. Most of the galaxies in substructures B, C, D, and E have r-band magnitudes of 16.5 < r < 18 mag, therefore only objects with z − 24 μm > − 1, i.e., close to strong MIR-excess galaxies can be detected as intermediate MIR-excess galaxies.