Detection of Intracluster Globular Clusters in the First JWST Images of the Gravitational Lens Cluster SMACS J0723.3-7327 at $z=0.39$

First we obtain the photometry of the sources (including both extended and point sources) detected with 10$\sigma$ threshold in the images using Source Extractor (Bertin & Arnouts, 1996). We carry out dual mode photometry with the F277W image as a reference image for source detection. As the pixel scales for LW and SW detector are different, the SW images were reprojected and aligned with respect to the F277W image using the wregister task in IRAF (Tody, 1986; Science Software Branch at STScI, 2012) and astroalign (Beroiz et al., 2020), respectively. For aligning the images, we use the positions of several stars. We use LW images for the analysis of galaxies because the normal red sequence of galaxies in galaxy clusters is better distinguishable in the LW color-magnitude diagrams (CMDs), than in the SW CMDs, as shown in the following.

Figure 2 display effective radius $r_{h}$ vs. F277W (AUTO) magnitudes of the sources detected with the Source Extractor. We also mark the sources with small stellarity $<0.4$ (red circles), and the sources with large stellarity $>0.8$ (blue circles). The prominent narrow slanted sequence dominated by blue circles represent the point sources with small effective radii. Some of the bright sources with large effective radii are found to have a large value of stellarity $>0.8$ so we consider that effective radius is a better indicator to distinguish extended sources than stellarity. In this study we select extended sources using the effective radius criterion: $r_{h}>1.8$ pixels in the F277W image.

The extended sources include galaxies and gravitationally-lensed sources. We adopt AUTO magnitudes for the extended sources provided by the Source Extractor, and use them for the analysis of photometric properties of the galaxies in the following.

For better detection of the point sources in the images that include a number of bright galaxies, we subtract the diffuse background light from the original images. First, we derive a smooth background image of the original F200W image, using a ring median filter with $r_{in}=18$ pixel and $r_{out}=20$ pixel. Then we subtract this smooth background image from the original image, and use the background-subtracted image for the point source detection.

In Figure 4 we display the LW CMD (F277W vs (F277W–F356W)) and SW CMD (F200W vs (F150W–F200W)) (in AUTO magnitudes) of the extended sources in the central field of SMACS 0723. We also mark the cluster member galaxies (N=26) confirmed with spectroscopic observations (Mahler et al., 2022), the selected bright galaxies (including the BCG), and the known gravitational lens image sources in Mahler et al. (2022).

In the F277W vs (F277W–F356W) CMD, two prominent features are noted: an LW-blue vertical sequence with $(F277W-F356W)_{0}\approx-0.6$, and an LW-red slanted sequence with $(F277W-F356W)_{0}\approx+0.1$. The known cluster members are mostly located in the bright part of the LW-blue sequence. The known cluster members are mostly early-type galaxies, showing that this sequence corresponds to the normal red sequence in the optical CMDs of galaxy clusters. So these member galaxies were plotted in red circles. The selected bright galaxies including the BCG are also located in the bright part of this sequence. The presence of a large number of galaxies located in this LW-blue sequence shows that SMACS 0723 is indeed a rich cluster.

In contrast, the known gravitational lens image sources are mostly located along the LW-red sequence, but with a larger scatter than the LW-blue sequence. They look red in the released color images based on the combination of all SW and LW filter images. The gravitational lens image sources have much higher redshifts than that of SMACS 0723 so they are not the cluster members. They are background galaxies most of which show some star formation activity (i.e., late type galaxies having different colors compared with early-type galaxies).

However, the F200W vs (F150W–F200W) CMD shows only one notable slanted sequence at (F150W–F200W)${}_{0}\approx+0.2$. Both the cluster member galaxies and the lens image sources are located mostly in the same sequence, so they are hardly distinguished in this sequence.

Therefore, as the red sequence galaxies which are mostly early-type cluster members, we select those located along the LW-blue sequence as marked in the LW CMD:$-0.65<(F277W-F356W)_{0}<-0.45$. We use them to check any correlation between the GCs and the galaxies in the following analysis.

Figure 5 displays F200W vs (F150W–F200W) CMDs of the point sources in the central and parallel fields. The CMD of the central field shows clearly a strong concentration of the point sources with GC colors ($-0.2<(F150W-F200W)_{0}<0.6$) in the magnitude range of $26.5<F200W_{0}<30$ mag. In contrast, a much smaller number of the sources are located in the same region of the CMD of the parallel field. This indicates that most of these sources in the central field belong to SMACS 0723.

We plot also the GCs and other compact source candidates in the Image 2 field of Sparkler galaxy ($z=1.38$) located at $R=15\arcsec$ (82 kpc) east of the BCG in Mowla et al. (2022). We cross-matched the compact sources in Mowla et al. (2022) with our photometry and found that all the sources (N=12) were successfully recovered. However, only five of them passed our point source selection criteria. This indicates that a significant fraction of the compact sources in Mowla et al. (2022) are at least marginally resolved or do not follow the clean JWST/NIRCam PSFs. These sources are located in the similar GC color range, and the matched GCs occupy the brightest part of the GCs ($27<F200W_{0}<28$ mag). Note that there are dozens more sources in the similar CMD region ($26.5<F200W_{0}<28$ mag) of the central field, but few are in the CMD of the parallel field. They are probably UCDs, but some of them may be candidates for background GCs that were gravitationally lensed like the Sparkler galaxy GCs. The brightest of them are as bright as F200W${}_{0}\approx 26.5$ mag.

In Figure 5 we also plot the color histograms of the point sources with $27.5<F200W_{0}<29.5$ mag in the central (thin red line) and parallel (thick black line) fields. Assuming the parallel field as the background, we subtract the contribution of the background from the color histogram of the central field. The resulting net color histogram of the central field (thick red line) shows a significant excess component that is close to Gaussian with a peak at $(F150W-F200W)_{0}\approx 0.2$. This indicates that these point sources are mostly genuine GCs which belong to SMACS 0723.

We select the GC candidate sample using the cyan box marked in the figure ($-0.2<(F150W-F200W)_{0}<+0.6$ and $27.5<F200W_{0}<29.5$ mag) for the following analysis. We conservatively selected the GC candidates with photometric completeness higher than  70% ($F200W_{0}<29.5$ mag). The number of the selected GC candidates in the two fields is about 2500. Figure 1(c,d, and e) display the zoomed-in images of the sample section of the BCG region that show these GCs better after subtracting the smooth background image from the original image.

The apparent magnitude range ($27.5<F200W_{0}<29.5$ mag) of these sources at the cluster distance corresponds to the absolute magnitude range of $-14.2<M_{F200W}<-12.2$ mag, so these GCs belong to the bright end of the GCs, which include the UCDs. Considering the redshift of SMACS 0723, the ages of these GCs are younger than 9.5 Gyr. The ages of most Milky Way GCs range from 11 Gyr to 13 Gyr, with a mean value of $12.3\pm 0.4$ Gyrs (Oliveira et al., 2020; Kang & Lee, 2021)). Thus the SMACS 0723 GCs may be about 8 Gyrs old, assuming that they formed as the Milky Way GCs.

We select two representative regions to check any difference between galaxy GCs and intracluster GCs: the BCG region (at $R<15\arcsec$ from the BCG center), and the intracluster GC region (with a radius of $15\arcsec$) that covers the western loop (Mahler et al., 2022; Pascale et al., 2022; Montes & Trujillo, 2022) and no bright galaxies, as marked in Figure 1(a). In Figure 6(a,b) we display F200W vs (F150W–F200W) CMDs of the point sources in the BCG region and the intracluster GC region. We also plot the color histograms of both regions in Figure 6(c). We derived the color histograms for the BCG and intracluster GC regions after subtracting the background contribution based on the parallel field. Note that the color distribution of the intracluster GC region shows a strong peak at $(F150W-F200W)_{0}=0.05$ and a weaker component at $(F150W-F200W)_{0}\approx 0.4$. In contrast the color distribution of the BCG region shows a strong peak at $(F150W-F200W)_{0}=0.15$, which is 0.1 redder than the peak in the intracluster GC region.

We calculate the blue GC fractions, obtaining f(blue GC)=65% for the intracluster GC region and 54% for the BCG region. Thus the intracluster region has a higher fraction of blue GCs than the BCG region.

The color distribution for the intracluster GC region is broad across the peak color, showing that there are not only blue GCs but also red GCs, although the blue GC fraction is higher. The presence of both subpopulations in this region implies that the origin of the intracluster GCs is dual: the dominant progenitors of the intracluster GCs are low mass dwarf galaxies, and the minor progenitors are massive galaxies, as discussed in the case of Virgo intracluster GCs by Lee et al. (2010).

Considering these, we divide the GCs in the entire field into two subpopulations according to their colors: the blue (metal-poor) GCs with $-0.2<(F150W-F200W)_{0}<+0.2$ and the red (metal-rich) GCs with $+0.2<(F150W-F200W)_{0}<+0.6$. We use these blue and red GCs for the following analysis to study any difference between the two subpopulations.

Figure 7(a) shows the spatial distribution of the GC candidates selected in the central field. In Figure 7(b) we plot the number density contours and color gradient map of these sources on the gray scale map of the F200W image. The contours were smoothed with a Gaussian smoothing scale, $\sigma_{G}\simeq 3\arcsec$. The lowest level in the contours starts with $2\sigma_{bg}$ where $\sigma_{bg}$ is estimated from the background fluctuation of the selected GC candidates in the low density region of the parallel field: $\sigma_{bg}=0.00052\pm 0.000029$ per kpc². The step between contours is $4\sigma_{bg}$. We also mark the locations of the known cluster members, the selected brightest galaxies (including the BCG), the gravitational lens image sources (Mahler et al., 2022), and the red sequence galaxies selected in this study, to check any spatial correlation with the GCs.

Several features are noted in this figure. First, the overall distribution of the GC candidates shows clearly a central concentration around the BCG that is located close to the galaxy cluster center. This implies that most of the selected GC candidates are the members of SMACS 0723, indeed. The strongest peak of the GC number density is slightly offset to the east direction from the BCG center, and this can be explained as follows. There are two galaxies close to the BCG center in the west direction so it is difficult to subtract completely the stellar light of these galaxies in the central region. This decreases the detection rate of the GCs in that region, causing the slight offset from the BCG center. However the center of the contours at $R>10\arcsec$ is consistent with the BCG center.

Second, the overall distribution of the GCs is elongated along the major axis of the BCG. Third, the spatial distribution of the GCs is much more extended than that of the stellar light of the BCG.

Fourth, strong peaks are found at the position of some brightest galaxies. The second strongest peak of the GC number density is seen at the position of G2, a bright elliptical member galaxy ($z=0.3841$) at $R\approx 190$ kpc in the northwest of the BCG. Another weaker peak is seen at the position of G3, a bright galaxy close to G2. A recognizable clumping of GCs is seen also at the position of G4, another elliptical member galaxy ($z=0.3845$). There is an isolated weak GC clump at $R\approx 300$ kpc south of the BCG, at the center of which one bright red sequence galaxy is located. The rest of the known cluster member galaxies do not show any strong clumps of GCs.

Fifth, there is a strong concentration of GCs at $R\approx 200$ kpc in the east of the BCG (called as the GC Clump 1 (GCC 1)). Interestingly there are no recognizable bright galaxies which can be a host of these GCs. Similarly there is a weak concentration of GCs at $R\approx 100$ kpc in the west of the BCG (called as the GC Clump 2 (GCC 2)). There are no recognizable bright galaxies which can be a host of these GCs.

Sixth, there is seen a concentration of GCs extended to the southwest from the second strongest G2 peak. This can be a part of the large western loop-like structure, which will be described in detail later. We selected the intracluster GC region to cover this substructure, as marked by the large circle in the figure.

Finally, the GCs located outside the few bright red sequence galaxies (including the BCG) are considered to be mostly intracluster GCs.

In Figure 8 we display the spatial distributions of the blue GC and red GC candidates selected in the central field. This figure shows that the central concentration of the red GCs around the BCG is more prominent than that of the blue GCs, which will be also shown by their radial number density profiles in the following. In addition, these maps show a loop-like structure with a radius of 100 kpc at $R\approx 200$ kpc west of the BCG. The structure is seen better in the red GC map. This will be compared with the substructure in the ICL map in the following.

Adopting the BCG center as the cluster center we derive the radial number density profile of the GCs, displaying it in Figure 9(a). The number density profiles of the GCs derived before background subtraction show a flattening at $R>90\arcsec$ in the parallel field. We estimate the background level from the sources at $R>90\arcsec$ in the parallel field, obtaining $\Sigma_{bg}=0.000831\pm 0.000075$ per kpc². In the figure the number density of the GCs deceases as the cluster-centric distance increases, confirming again that these GCs mostly belong to the galaxy cluster. The radial number density of the GCs keeps decreasing until $R\approx 600$ kpc. The central excess component represents the BCG GCs, while the outer part is dominated by the intracluster GCs.

In Figure 9(b) we also plot the radial number density profiles of the blue and red GCs. The radial profile of the red GCs shows a stronger excess in the central region than the blue GCs, and it shows an opposite trend in the outer region.

We fit the radial number density profiles of the GCs with the two component Sérsic law (the BCG component plus the intracluster GC component): $\Sigma_{GC}(R)=\Sigma_{\rm eff}\exp{\{-b_{n}[(R/R_{\rm eff})^{1/n}-1]\}}$ and $(b_{n}=2n-0.3271)$ (Sérsic, 1963; Graham & Driver, 2005). We also fit the data with the power law for the outer region of the intracluster GC component ($158<R<631$): $\Sigma_{GC}(R)\propto R^{\alpha}$. Table 2 lists a summary of the fitting results. The radial profile of all GCs is fit well with the Sérsic parameters, $R_{\rm eff,1}=28.5\pm 3.2$ pc and $n_{1}=0.1\pm 0.1$ for the BCG component, and $R_{\rm eff,2}=224.5\pm 22.1$ pc, $n_{2}=1.3\pm 0.3$ for the intracluster component. The outer region is also fit well with a power law index $\alpha=-2.0\pm 0.2$. The radial profiles of the blue GCs and red GCs are also fit with similar parameters: in the case of the blue GC, $R_{\rm eff,1}=29.5\pm 6.5$ pc, $n_{1}=0.1\pm 0.2$ for the BCG component, and $R_{\rm eff,2}=213.7\pm 22.1$ pc, $n_{2}=1.1\pm 0.3$ for the intracluster GC component, and $\alpha=-2.2\pm 0.2$ for the outer region; in the case of the red GCs, $R_{\rm eff,1}=26.7\pm 3.4$ pc, $n_{1}=0.1\pm 0.1$ for the BCG component $R_{\rm eff,2}=241.7\pm 48.6$ pc, $n_{2}=1.5\pm 0.6$ for the intracluster GC component, and $\alpha=-1.9\pm 0.3$ for the outer region.

Thus the effective radius of the intracluster GC system is about eight times larger than that of the BCG GC system. The outermost region of the intracluster GC system is also represented well by the power law with an index of $\alpha=-2.0\pm 0.2$. This profile is slightly steeper than that of the intracluster GC system in the Virgo cluster, $\alpha=-1.5\pm 0.1$(Lee et al., 2010).

From the JWST/NIRCam images of SMACS 0723, Pascale et al. (2022) and Mahler et al. (2022) derived a smooth map of the diffuse light using a median box car filter with $21\times 21$ pixels and $100\times 100$ pixels, respectively, showing the distribution of the ICL. They noted that the diffuse ICL is elongated along the major axis of the BCG, but in a much more extended way. Also they found two interesting substructures of the ICL: a large scale loop in the west of the BCG and a large lobe-like feature in the east component.

To compare the distributions of the GCs and ICL, we derive similarly a smooth map of the diffuse light applying a median boxcar smoothing with $51\times 51$ pixels to the F277W image, and display it in Figure 10(a). We overlay the GC number density contour maps (all GCs, blue GCs, and red GCs) to the smooth map of the ICL in Figure 10(b,c, and d). We also mark the positions of bright stars, known cluster member galaxies, and gravitational lens image sources (Mahler et al., 2022) in the figure.

Figure 10 shows several distinguishable features. First, the spatial distributions of the GCs and ICL are consistent in general. However, the GCs are found farther than the ICL boundary. Second, the western large scale loop structure seen in the GC number density map is consistent with that of the ICL. A number of GCs are located along the loop of the diffuse ICL. Third, the GC number density map shows also an extended structure in the east of the BCG, which is consistent with the lobe-like diffuse component of the ICL. However, the GC Clump 1 (the strongest clump of GCs at $R\approx 200$ kpc east of the BCG) does not have a bright counterpart in the ICL map, although it is close to the boundary of the east loop. The GC Clump 1 is better seen in the blue GC map, indicating that it is composed mainly of the blue GCs.

The existence of a number of substructures with varying scales (in addition to the BCG) in the GC map and ICL map indicates that a large number of galaxy groups with diverse masses are falling to the cluster center, and they are in the middle of merging process now. The elongated structure seen in the maps of the GCs and ICL indicates that most galaxies are falling to the cluster center along the east-west direction. These maps illustrate the distribution of the stars and GCs that are stripped off from their host galaxies or are the remnant of disrupted galaxies.

Considering the distributions of the GCs and ICL, G2 is probably a main host of a galaxy group that includes G3 (called the G2 group). Figure 10 shows that the G2 group and the BCG are connected not only in the GC map but also in the ICL map. The existence of the GCs and diffuse light between the G2 group and the BCG implies that both systems are physically interacting. This is similar to the case of the NGC 4839 group that is known to be falling to the Coma center, as shown in the observations of galaxies and various ICM including the GCs (Briel et al., 1992; White et al., 1993; Colless & Dunn, 1996; Lyskova et al., 2019; Churazov et al., 2021; Oh et al., 2022). The loop structure might have formed when the G2 group is falling to the cluster center, which needs to be investigated further in the future.

Chandra X-ray brightness map of SMACS 0723 is given in Mahler et al. (2022) (their Figs. 1 and 9). Comparing the GC number density map in this study with the X-ray map of SMACS 0723 given in Mahler et al. (2022), we find both maps display generally similar structures in the sense that both show a large scale elongated distribution. Interestingly the X-ray contour map of the central region of SMACS 0723 (Fig. 1 in Mahler et al. (2022)) shows a distinguishable component at the position of the western loop structure in addition to the strongest peak at the BCG center. The location of this component is consistent with the center of the western loop in the GC number density map and ICL map.

Dark matter mass density maps (and radial profiles) of SMACS 0723 derived from the strong lensing analysis were presented by Mahler et al. (2022); Pascale et al. (2022); Caminha et al. (2022); Golubchik et al. (2022) who used the JWST/NIRCam images and HST/ACS/WFC3 images. In Figure 11 we overlay the GC number density contour maps on the dark matter surface mass density map (the convergence ($\kappa$=surface mass density normalized to the critical mass density) map) provided by (Mahler et al., 2022) (pseudo color maps). The GC maps and dark matter mass maps show remarkably similar structures in the sense that both show a large scale structure that is elongated along the major axis of the BCG, but much more extended than the BCG. They show also similar substructures.

Mahler et al. (2022) present a comparison of the dark matter radial density profiles of SMACS 0723 they derived from the JWST/NIRCam images with those derived using different lensing models in the literature: RELICS-lenstool (Coe et al., 2019), RELICS-GLAFIC (Coe et al., 2019), LTM (Golubchik et al., 2022) (their Fig. 6). In this comparison, the first three models produce similar profiles extending out to $R\approx 1$ Mpc, while the LTM profile is much more steeply declining in the outer region ($R\approx 500$ kpc).

In Figure 12 we present a comparison of the radial number density profiles of the GCs with the dark matter mass density profiles based on the lensing analysis (Mahler et al., 2022). This figure shows that the radial density profile of the GCs is very similar to those of the dark matter in the inner range of $R=50-200$ kpc. However it becomes steeper in the outer region at $R=200-600$ kpc, deviating from the three lens models (RELICS-lenstool, RELICS-GLAFIC, and Mahler et al. (2022)) and getting closer to the LTM model. However, the outermost GC profile is between the two model groups.

These results are consistent with those for the GCs found in the deep HST images of Abell 2744, another massive cluster at $z=0.308$ that is known as one of the most GC-abundant galaxy clusters (Lee & Jang, 2016a). Similar trends are also seen in the radial profiles of the intracluster GC number density and the dark matter mass derived from the simulated galaxy clusters selected from the cosmological simulations (Ramos-Almendares et al., 2018, 2020).