Efficient Detection of Emission-line Galaxies in the Cl0016+1609 and MACSJ1621.4+3810 Supercluster Filaments Using SITELLE^*

Both clusters were observed with the same configuration. In order to achieve a signal-to-noise ratio (S/N) of 5 in the emission lines, each field was observed for 4 hr. ⁶ A resolution of 600 allows for a separation of the sky lines from the redshifted [O ii] 3727 Å lines of interest. The sources were observed using 1 × 1 binning in order to avoid saturation.

The data are flux calibrated with standard stars, wavelength calibrated with the HeNe lamp, and phase calibrated using flat fields.

To capture the [O ii]3726-3729 line emission, the C2 filter is used for Cl0016+1609 at z = 0.546 and the C3 filter for MACSJ1621.4+3810 at z = 0.465.

Reconstructed deep images and detection maps are created with the SITELLE data using the ORCS package (Martin et al. 2015). Sources are then extracted using SExtractor (Bertin & Arnouts 1996). The deep image includes the sum of all emission in the filter, including the continuum emission, and so produces an image in which the galaxies can be easily identified. The detection map includes the addition of flux only from the emission lines within the filter's window, thus the morphology reveals the structure of the line-emitting gas, rather than the shape of the galaxy. The output catalogs from both images are matched by R.A. and decl. to generate a list of emission-line sources that have an optical galaxy host. For Cl0016+1609, there are 371 matched sources of the 2082 deep-detected galaxies with r < 21.5 in the five detection frames. For MACSJ1621.4+3810, there are 304.

For each source, a Python script subtracts strong sky lines by finding a nearby patch of sky, and a sky-subtracted spectrum for each object is generated. The generated object spectra are constructed by binning 3 × 3 pixels. This method includes the bright line emission from the target galaxies and avoids any line emission from nearby sources, which are always more than 3 pixels from the source of interest. The resulting spectra are examined by eye by two of us (L.E. and K.Z.), and those sources with lines near the expected peak of the [O ii]3726–3729 doublet are discussed below.

For Cl0016+1609, there are 23 emission-line sources within the expected redshift range of the supercluster, plus an additional four interloping systems identified. The source ID numbers and positions are listed in columns 1–3, respectively, of Table 1 (top) and their spectra plotted in Figure 3. Table 1 also shows the results of the best-fit Gaussian functions used to model the lines and continuum noise. This includes the peak wavelength in column 4 and the standard deviation in column 5. The S/N level of the emission line is given in column 6. The noise is measured in the sky-subtracted continuum over in 50 Å bins that flank the location of the line peak. This error is illustrated in Figure 3 as a gray shaded region around the spectrum, for clarity, only the first and last spectra are marked. Where sample sizes are small, errors on percentages are derived from the posterior probability, assuming a binomial distribution and the Jeffreys interval using astropy (Astropy Collaboration et al. 2018). For the 23 emission-line systems within the expected location of the supercluster, 13 (56 ± 10%) have an S/N greater than 3.0, and 10 (43${}_{-9}^{+10}$%) have low S/N. There are four high-S/N sources that have velocities that fall above or below 3σ_cl: ID number 973 appears to be a foreground source identified in NED as SDSS J001858.83+160149.7, with a redshift of z = 0.1161 Tanaka et al. (2007). ID 1570 is identified as SDSSJ001900.37+160223.3 in NED and is also a foreground source. There is no redshift included in NED, but using the Hβ–[O iii] complex that appears in the SITELLE data, its redshift is calculated to be z = 0.2137. ID 1376, identified in NED as WISEA J001902.06+160350.0, is a background source. ID 280 has a strong emission line >3σ_cl, thus is included as a background object in this study.

Zoom In Zoom Out Reset image size

The 10 identified emission-line sources for MACSJ1621.4+3810 are shown in Table 1 (bottom) and plotted in Figure 3. The identified emission-line sources all have wavelengths within that expected for a cluster with a velocity dispersion σ_cl < 1100 km s⁻¹(Edge et al. 2003). All (${100}_{-15}^{+0}$%) have S/N > 3.

A search using NED finds small levels of Galactic extinction for both sources, with E(B − V) = 0.05 for Cl0016+1609, and E(B − V) = 0.01 for MACSJ1621.4+3810. Similar to other SITELLE studies, we have applied no Galactic extinction correction (Duarte Puertas et al. 2019).

MegaCam images from the CFHT are used to determine the galaxy density maps for Cl0016+1609, and Figure 5 (top), a color–magnitude diagram for red-sequence galaxies (crosses). The emission-line galaxies detected in this study (circles) are overplotted. All but two (85${}_{-15}^{+5}$%) of the high-S/N emission-line sources of Cl0016+1609 are blue galaxies, according to the red-sequence definition of Durret et al. (2016). The mean ($g^{\prime} -i^{\prime}$) color of the emission-line sources is 0.4 magnitudes below the mean color of the red-sequence galaxies. One source (ID = 625) is very red with $g^{\prime} -i^{\prime} =2.6$, it has a very faint $g^{\prime}$ magnitude. This source may be a dusty red star-forming galaxy (Fadda et al. 2008; Gallazzi et al. 2009), or an obscured AGN (Lacy et al. 2004; Imanishi et al. 2007).

Zoom In Zoom Out Reset image size

For MACSJ1621+3810, deep V and I imaging from the Subaru Telescope provide the color data. All matches are within 1'' except for ID = 213, whose closest match is an I < 25 source at a distance of 4''. For this system, the separation from the red sequence is even more extreme (Figure 5, bottom), with ${100}_{-15}^{+0}$% of the emission-line sources matched to blue galaxies. Even the BCG is blue, which is rare but not unique (Chu et al. 2021). The mean (V − I) colors of the emission lines are ∼1 mag bluer than the red sequence.

The emission-line sources detected in this way show the expected blue color of strongly star-forming galaxies (Blanton et al. 2003; Baldry et al. 2004); this gives confidence in the use of SITELLE for identifying star-forming galaxies in superclusters.

The k-corrected (Chilingarian & Zolotukhin 2012) galaxy colors are used to calculate stellar masses, using the conversions in Bell & de Jong (2001) for the V and I colors and in Bell et al. (2003) for $g^{\prime}$ and $i^{\prime}$. They are listed in Table 1.

Cosmic SFRs have been plummeting since their peak around redshift 2 (Madau et al. 1996; Ly et al. 2007). By z ∼ 0.5, median SFRs of 3.2h⁻² M_⊙ yr⁻¹ are measured for blue galaxies, in cluster cores as well as the field (Cooper et al. 2008).

Assuming the emission lines from these blue moderate-density region sources are coming from star-forming regions, the SFRs can be calculated by calibrating the measured [O ii] 3727 Å luminosity to an expected Hα value and applying the relation between SFR and Hα luminosity (Kennicutt 1998). This requires knowing or assuming the extinction value at Hα and the natural line ratio between the two. Kennicutt (1998) accomplished this by deriving a relationship for extinction-corrected [O ii] 3727 Å luminosity, calibrated to line ratios of local blue-irregular, normal, and peculiar star-forming galaxies (Gallagher et al. 1989; Kennicutt 1992). Given the mass–metallicity relationship, higher-metallicity galaxies are expected to have more dust, and thus more reddening. Concerned that high-SFR systems would typically be underestimated, and low-SFR systems overestimated, Kewley et al. (2004) derived a relationship for SFR based on [O ii] 3727 Å luminosity that does not assume local reddening so that one can apply a proper reddening correction. For cases where the reddening is unknown, the approach taken by Gilbank et al. (2010, 2011) is useful. These authors use the galaxy stellar mass to empirically correct for reddening. The SFRs calculated in this paper use their Equation (8), given as Equation (1), below:

$$\begin{eqnarray}&&{\mathrm{SFR}}_{\mathrm{corr}}=\displaystyle \frac{{\mathrm{SFR}}_{0}}{a\,\tanh [(x-b)/c]+d},\end{eqnarray} \tag{ 1 }$$

where $x={\rm{log}}({M}_{\ast }/{M}_{\odot })$ and a = −1.424, b = 9.827, c = 0.572, and d = 1.700. For sources with unknown mass, the calculations are performed assuming ${\rm{log}}({M}_{\ast }/{M}_{\odot })=10$. SFR₀ is the nominal SFR in M_⊙ yr⁻¹ before the mass correction is applied, as defined in Equation (2), below.

$$\begin{eqnarray}&&{\mathrm{SFR}}_{0}=\displaystyle \frac{{10}^{0.4{A}_{{\rm{H}}\alpha }}}{1.5{r}_{\mathrm{lines}}}\displaystyle \frac{L([O\,{II}])}{1.27\times {10}^{41}},\end{eqnarray} \tag{ 2 }$$

where L([O ii]) is the [O ii] 3727 Å luminosity in erg s⁻¹, r_lines is the ratio between the [O ii] 3727 Å and Hα emission lines, taken to be 0.5, and A_Hα is the extinction at Hα. A value of A_Hα = 1.0 mag is adopted to calculate the nominal rate, SFR₀, and then the correction is applied (following Gilbank et al. 2010).

There are several important caveats worth mentioning here. First, AGNs are not removed. If there is AGN contamination, the values quoted here would be overestimates (Zhuang & Ho 2019). However, AGN are thought to be most important if found along the red sequence (Yan et al. 2006), whereas most of the sources in this study avoid the red sequence. Second, the calibration of the relationship is based on galaxies between 0.03 < z < 0.20. Several studies have found that SFRs increase with redshift at fixed stellar mass. For example, total SFRs calculated using UV and FIR data increase with redshift at fixed stellar mass (Tomczak et al. 2016), so mass-calibrated SFRs may still be underestimated by a factor of ∼5 given the higher redshift of the clusters studied in this paper. Third, there is a huge variation source to source in the extinction levels at a given stellar mass. Moustakas et al. (2006) calculate that the scatter in SFR is a factor of 2.5. Column 7 of Table 1 gives the calculated SFR and its error. The error quoted is calculated from the 1σ uncertainties in the [O ii] 3727 Å-derived SFR at each mass, as compared to the corrected Hα-derived rates (see Figure 3 of Gilbank et al. 2010).

For Cl0016+1609, the SFRs range from 0.2 to 14 M_⊙ yr⁻¹, with a median SFR of 2 M_⊙ yr⁻¹ as can be seen in the histogram of Figure 6. The star-forming galaxies are blue when compared to the cluster members (Figure 5), with one exception (ID = 625), which is perhaps an AGN or an extremely dusty starburst.

Zoom In Zoom Out Reset image size

The two highest-SFR galaxies have peak wavelengths close to the expected location for the cluster, bright $i^{\prime}$ magnitudes, small isophotal areas and radii, and thus bright $i^{\prime}$ surface brightnesses (see Table 1). Their luminosity is emitted from a concentrated region. No evidence of a relationship with the galaxy position angle, ellipticity, or elongation is identified. The set of star-forming galaxies as a whole shows no strong trend with many of the galaxy properties measured (LGD, velocity difference between the galaxy and cluster, surface brightness and radius). But the fainter $i^{\prime}$ magnitude galaxies and those with bluer ($g^{\prime} -i^{\prime}$) colors have lower SFRs. The SFRs and sSFRs show the expected trend, with higher SFRs for high-mass galaxies, and higher sSFRs for low-mass galaxies (see Figure 7, left panels). The R² statistic for the fits is R² = 0.62 and R² = 0.58, respectively). The sSFR appears evenly distributed across values of LGD.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

For MACSJ1621.4+3810, the SFRs range from 0.32 to 22.8 M_⊙ yr⁻¹, with a median value of 0.8 M_⊙ yr⁻¹ (Figure 6, bottom). In this cluster, the source with the strongest emission lines and highest calculated SFR is the BCG (ID = 1875). This source is the reddest of the emission-line sources, though still 0.3 mag bluer than the red sequence in (V − I) (see Figures 5 and 7). BCG's emission lines are usually associated with the cluster's cool core, so this source is removed when looking at possible trends with SFR. With only nine data points for the emission-line sources left, it is difficult to assert strong trends in the data. Nonetheless, a least-squares fit to LGD and mass with SFR and sSFR was calculated. There is no trend with LGD, but a weak trend with SFR increasing with mass is found (R² = 0.34), and the sSFR steadily decreases with mass (R² = 0.96; see Figure 7, right). The results are in line with the strong correlation found between SFR and galaxy stellar mass for local galaxies (Brinchmann et al. 2004; Lara-López et al. 2010; Rodrigues et al. 2008; Vulcani et al. 2019).

At lower redshift, Coma, Abell 1763, and Abell 901 (Edwards et al. 2010; Bianconi et al. 2016; Gallazzi et al. 2009) also host the highest sSFRs in the filaments. Moran et al. (2005) find a small set of [O ii] 3727 Å-emitting galaxies surrounding the main core of Cl0024+16 at z = 0.4. Those galaxies host a spread of [O ii] 3727 Å luminosity, and, as in these superclusters, no trend with LGD is measured.

If the star formation activity of galaxies residing in filaments is indeed a result of galaxy–galaxy interactions, they might be expected to show tell-tale signs of merging like tidal tails, trainwreck shapes, or nearby neighbors.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

In our two superclusters, ${39}_{-8}^{+9}$% of emission-line sources have nearby neighbors, this fraction is similar to that for luminous infrared galaxies (LIRGs; Shi et al. 2009), and significantly greater than 4%–20% usually found at z ∼ 0.5 (Lotz et al. 2011) either by close-pair analysis (Lin et al. 2008), the Gini coefficient (Lotz et al. 2008, explored below), or asymmetry (Shi et al. 2009). The high fraction of nearby neighbors is also in line with the studies of Guo et al. (2015) who find that galaxies in filaments have up to a factor of 2 more satellites than those outside of filaments.

The Python program statmorph (Rodriguez-Gomez et al. 2019) is used to further quantify the galaxy morphology, for all sources with emission-line S/N > 3 and within the supercluster redshifts (0.53–0.55 for Cl0016+1609 and 0.46–0.47 for MACSJ1621.4+3810). The statmorph code provides three shape parameters explored in this study. The code is run on deep continuum images constructed from the SITELLE data (left panel of Figures 9 and 10 for Cl0016+1609 and left panel of Figures 11 and 12 for MACSJ1621.4+3810). The parameters are the Gini coefficient (G), the M₂₀ parameter (M₂₀), and the merger statistic. As discussed in Lotz et al. (2004, 2008) and Snyder et al. (2015), G is a measure of how the flux is distributed over the galaxy pixels. With G = 1, all the flux is concentrated in a single pixel, whereas with G = 0, the flux is equally distributed over all galaxy pixels. M₂₀ is the normalized second-order moment of the brightest 20% of the galaxy's flux. It traces bright features such as arms, bars, nuclei, and bright star formation regions. The merger statistic separates merging galaxies in G–M₂₀ space. It is defined by the gray solid line in the left-hand panels of Figure 8. Table 1 and Figure 8 show the results for the star-forming galaxies in the Cl0016+1609 and MACSJ1621.4+3810 superclusters.

For Cl0016+1609, the majority of emission-line galaxies are disky, with ${69}_{-15}^{+10}$% as Sb/Sc/Sd/Irr systems, which can be seen in Figure 8 (top), with one system (8${}_{-3}^{+14}$%) characterized as E/SO/Sa, and three (23${}_{-8}^{+15}$%) merging systems. Of the merging systems, shown as triangles, ${100}_{-37}^{+0}$% are found in moderate-density environments (LGD = 3000–8000), where the galaxy density is high enough that encounters would be expected. ID 1411 is on the border of a positive merger statistic, it is identified as E/SO/Sa, thus it could be an emission-line elliptical; however, with its blue color, S0/Sa is a more likely identification. No trend with orientation, position angle, or galaxy size is found with either SFR or with LGD (not shown).

For MACSJ1621.4+3810, the emission-line galaxies are fairly evenly split between disky systems (circles, ${40}_{-12}^{+16}$%) and mergers (triangles, ${60}_{-16}^{+12}$%), with ${0}_{-0}^{+15}$% bulge-dominated systems, according to the G–M₂₀ classification scheme (see Figure 8, bottom). Even the BCG (ID 1875, with G = 0.52 and M₂₀ = −1.9) is classified as disky, though it is not far from the E/S0/Sa border. And, recall, that deep SITELLE images are constructed from the fairly narrow C3 filter, which has a bandwidth of 500 Å, which includes the flux of the line emission and could be dominating the luminosity. Of the merging systems, ${83}_{-23}^{+7}$% are found in moderate-density environments (between ∼9000 and 25,000) and all have nearby (projected) companions as seen in the deep images of Figures 11 and 12. Five of six (${83}_{-23}^{+7}$%) of the merging systems appear large, as their best-fit model includes the neighboring galaxy. The merging galaxies (ID: 213, 1895, 3066, 3283, 2832, 3072) have low to moderate SFRs (0.3–2.7 M_⊙ yr⁻¹). No trend with orientation or galaxy size is found for either the SFR or local density. The star-forming mergers with merging characteristics are not separated in color, environment, size, or magnitude from the nonmerging star-forming galaxies (not shown).

Eight of the 23 emission-line galaxies for Cl0016+1609 (${35}_{-9}^{+11}$%) show nearby projected companions within 50 kpc (IDs: 2381, 2249, 1984, 2687, 297, 636, 1411, and 1211). For MACSJ1621.4+3810, half of the 10 emission-line systems (50 ± 14%) host nearby projected companions (IDs: 213, 1875, 3022, 3066, and 3072). A few also display asymmetric shapes (Cl0016+1609: 2381, 376; MACSJ1621.4+3810: 213, 3022, 3699), emission-line images revealing multiple cores (MACSJ1621.4+3810:3022), and material connecting to nearby systems (e.g., MACSJ1621.4+3810:213, 1875).

In our two superclusters, ${39}_{-8}^{+9}$ 39+8−9% of emission-line sources have nearby neighbors, this fraction is similar to that for luminous infrared galaxies (LIRGs; Shi et al. 2009), and significantly greater than 4%–20% usually found at z ∼ 0.5 (Lotz et al. 2011) either by close-pair analysis (Lin et al. 2008), the Gini coefficient (Lotz et al. 2008, explored below), or asymmetry (Shi et al. 2009). The high fraction of nearby neighbors is also in line with the studies of Guo et al. (2015) who find that galaxies in filaments have up to a factor of 2 more satellites than those outside of filaments.

Several studies have shown that the disks of star-forming galaxies become truncated as the galaxies move through the intracluster medium (e.g., Gullieuszik et al. 2020), caught by the cluster potential. Such environmentally driven gas depletion is expected to preferentially remove gas from the outskirts (Gunn et al. 1972; Bekki 2009; Vulcani et al. 2020). In this case, one would expect to see the galaxies in more dense environments have a smaller extent of emission-line gas relative to their host galaxy's size. To investigate, the Petrosian radius for both the deep galaxy image (R_gal) and emission-line image (R_em) from SITELLE are measured using SExtractor. The ratio of the two is shown in the middle panels of Figure 8. Values of R_gal/R_em are also listed in Table 1 and range from ∼0.5 to ∼1.5. These values are similar to the range found in Vulcani et al. (2019), who examine four star-forming galaxies inside of local filaments (Tempel et al. 2014). They find ratios between 0.5 and 2, where the emission-line image is of Hα emission from MUSE observations. No strong trend with LGD is found for Cl0016+1609, but a least-squares fit to the data for MACSJ1621.4+3810 finds a trend with an R² statistic of 0.65.