New KiDS in town Sextans II: a new stellar system in the outskirts of the Milky Way

We report on the discovery of a significant and compact over-density of old and metal-poor stars in the KiDS survey (data release 4). The discovery is confirmed by deeper HSC-SSC data revealing the old Main Sequence Turn-O ff of a stellar system located at a distance from the sun of D ⊙ = 145 + 14 − 13 kpc in the direction of the Sextans constellation. The system has absolute integrated magnitude ( M V = − 3 . 9 ± 0 . 4), half-light radius ( r h = 193 + 61 − 46 pc), and ellipticity ( e = 0 . 46 + 0 . 11 − 0 . 15 ) typical of Ultra Faint Dwarf galaxies (UFDs). The central surface brightness is near the lower limits of known local dwarf galaxies of similar integrated luminosity, as expected for stellar systems that escaped detection until now. The distance of the newly found system suggests that it is likely a satellite of our own Milky Way, consequently, we tentatively baptise it Sextans II (KiDS-UFD-1).


Introduction
The past two decades have seen a surge of interest in the research of Milky Way (MW) satellites thanks to the advent of digital wide-field deep panchromatic photometric surveys.The application of high-performance over-density detection techniques to large data sets has led to a substantial increase in the number of known stellar systems inhabiting the MW halo (e.g.Belokurov et al. 2007;Walsh et al. 2009;Koposov et al. 2015;Torrealba et al. 2016;Cerny et al. 2021, and references therein).The quest for MW satellite galaxies carried out with modern facilities has opened up the study of the faint end of the galaxy luminosity function to so-called ultra-faint dwarf (UFD) galaxies (see Belokurov 2013, for discussion and references).These objects provide valuable insights into the mass assembly history of the MW.In addition, the UFDs are the most dark matterdominated objects in the Universe, as well as the faintest, oldest, and least chemically evolved galaxies known so far (see e.g. the detailed review by Simon 2019, and references therein).Therefore, they represent fossil evidence that is useful for probing the very early stages of the Universe, such as the epoch of the reionisation (e.g.Bovill & Ricotti 2011;Wheeler et al. 2015).Because of their pristine nature, UFDs are also ideal systems for testing and improving our understanding of the sub-branches of the stellar evolutionary models, such as the synthesis of heavy elements (e.g.Ji et al. 2016).
KiDS is an extensive multi-band photometric survey conducted using the VLT Survey Telescope (VST; Capaccioli & Schipani 2011;Capaccioli et al. 2012).It has observed about 1350 square degrees of the sky in the u, g, r, and i bands.KiDS was conceived as a cosmological survey, mainly with an aim to use weak gravitational lensing of galaxies to study the assembly of large-scale structures.However, as a side-product of extra-galactic studies, KiDS provides a deep and accurate catalogue of stars, making it suitable in the search for low-surface brightness stellar systems in the Local Group.In this Letter, we present a new UFD galaxy uncovered in a large-scale search for low-luminosity stellar over-densities over the whole KiDS Data Release 4 (DR4) area.

KiDS data
We used the most recent KiDS data release (DR4, Kuijken et al. 2019), which includes 1006 of imaging and catalogue data with limiting 5σ AB magnitudes of 24.23 ± 0.12, 25.12 ± 0.14, 25.02 ± 0.13, and 23.68 ± 0.27 mag in the u, g, r, and i filters, respectively.These data are 2 mag deeper than SDSS and PanSTARRS DR1.The data also include photometry in the Z, Y, J, H, K s infrared (IR) bands obtained with the VISTA Kilodegree INfrared Galaxy survey (VIKING; Edge et al. 2013) carried out on the VISTA telescope.KiDS covers two different stripes in the sky, namely, a northern and a southern stripe (KiDS-N and KiDS-S, hereafter; see Table 2 in Kuijken et al. 2019, for the exact boundaries).
From the KiDS DR4 catalogue, we selected sources identified as stars, namely objects that are flagged as 1, 4, or 5 in the sg2dphot column (see Appendix A.1.2 in Kuijken et al. 2019).In this way, we obtained a list of 4 422 730 stars in KiDS-N and 3 116 050 stars in KiDS-S 1 .Successively, we filtered out sources with low-quality photometry by imposing a maximum tolerated photometric uncertainty in both the g and r bands of 0.2 mag.Nonetheless, the catalogues at our disposal still include distant galaxies that have been erroneously classified as stars.To enhance the purity of our catalogues, we took advantage of the multi-band photometry provided by KiDS, utilising colour-colour diagrams to further purge unresolved galaxies.Specifically, we focused on the g − r versus g − i plot, where stars occupy a distinctive, narrow sequence (see for example Fig. 18 in Kuijken et al. 2019).To this end, we selected only sources falling within the boundaries of a polygon defined by the following coordinates (expressed in mag): (g − r, r − i) = (1.25,0.25); (1.25, 0.8); (−0.65, −0.05); (−0.65, −0.6).
It is important to note that we also opted to exclude red stars, characterised by g − r > 1.25 mag, as this colour range is dominated by M-type dwarf stars located within the MW disk.After the colour-colour selection, we are left with 3 558 843 stars, 2 127 949 of which are located within the footprint of KiDS-N, and the remaining 1 430 894 stars are in KiDS-S.

Method
has been widely used in recent years for the discovery of lowluminosity galaxies and globular clusters in the Milky Way halo (e.g.Bechtol et al. 2015;Koposov et al. 2015;Torrealba et al. 2016;Homma et al. 2019;Cerny et al. 2021, and references therein).This method consists of a preliminary isochrone-based matched filter on the catalogue of stars.In particular, prior to running the searching algorithm, we selected stars whose positions in the g−r, r colour-magnitude diagram (CMD) were within 0.05 mag in colour from a given isochrone representative of an old and metal-poor stellar population2 , which is characteristic of the faint galaxies expected to inhabit the MW halo.We opted to work with the suite of PARSEC isochrones (Bressan et al. 2012) and the isochrone filter was carried out by adopting three different ages, namely, log(t/yr) = 10.0, 10.06, 10.12 dex, and three different metallicities, or [Fe/H] ∼ −2.2, −1.5, −1.2 dex.Additionally, we varied the distance modulus of the isochrone in a range between 16 and 23.6 mag (corresponding to a range of physical distances between ∼16 and ∼520 kpc) with steps of 0.4 mag.Therefore, we scanned the KiDS catalogue with 180 different isochrone filters.
For each isochrone, we analysed the resulting filtered map as follows.We divided the map into 4 • × 4 • regions, and then build density maps using the star coordinates as input parameters, moothed through a kernel density estimation (KDE) technique adopting a Gaussian kernel.Subsequently, the algorithm counts the number of stars (N stars ) in each pixel of the sub-field and measures the number of standard deviations above a local mean: where N bkg and σ bkg represent the average of the local stellar background and its standard deviation, respectively3 .In this work, we use a pixel size of 30 × 30 and a bandwidth of the Gaussian kernel function of 14 .As a sanity check, we conducted a validation of the algorithm by assessing its capability to accurately identify known stellar systems residing within the KiDS footprint, including Galactic globular clusters and dwarf galaxies.In this way, our algorithm recovered Leo V with S pxl = 21.5, by applying an isochrone filter5 with log t = 10.06 dex, [Fe/H] ∼ −2.2, and a distance modulus = 21.2 mag, corresponding to a distance of approximately 175 kpc.This determination is closely aligned with the previously estimated distance to the dwarf, reported as 173 ± 5 kpc by Medina et al. (2017).

Results
Utilising the methodology outlined in the previous section, we identified a highly promising over-density in the Sextans constellation.Our detection revealed a S pxl = 10.6 stellar over-density by applying an isochrone filter with log(t/yr) = 10.12 dex, [Fe/H] ∼ −1.5, and a distance modulus (m − M) 0 = 20.4magnitudes.The left panel of Fig. 1 shows the density map of stars in the region of the newfound galaxy candidate, for the aforementioned isochrone selection filter.The over-density stands .Unfortunately, the photometric depth of the KiDS data is not sufficient to reach the main sequence turn-off (MSTO) of the candidate7 , whose detection would be crucial in order to confirm it as a genuine stellar system and to better assess its properties.In the next section, we describe the successful analysis of the candidate with deeper photometry.

Deeper photometry in HSC-SSP
The Hyper Suprime Cam Subaru Strategic Program (HSC-SSP; Aihara et al. 2018a,b) is a wide-field optical-near infrared survey conducted using the 8.2-meter Subaru Telescope.The HSC-SSP is designed to cover approximately 1400 square degrees of the Northern sky across five distinct filters, namely, g, r, i, z, y.
The survey 5σ magnitude limits for point-sources are impressively deep: 26.5 mag in the g band, 26.1 mag in the r band, and 25.9 mag in the i band (Aihara et al. 2018b), surpassing the limiting magnitude of the KiDS survey by more than one mag in each passband.HSC-SSP has already demonstrated its power for the detection of low-surface-brightness stellar systems (Homma et al. 2016(Homma et al. , 2018(Homma et al. , 2019)).In particular, Homma et al. (2019) reported the discovery of Bootes IV, one of the faintest UFDs identified to date, characterised by surface brightness levels of µ ≥ 32 mag arcsec −2 .Since our candidate lies within the HSC-SSP footprint, this survey provides an excellent means to follow it up.
From the HSC-SSP catalogue provided by the public third data release (Aihara et al. 2022), which included data for slightly less than 700 square degrees, we selected point-like sources within the region of the newly discovered over-density by means of the 'extendedness' parameter in i-band, which is the filter with the best median seeing (i.e.∼0.6 , Aihara et al. 2022).
To further refine the star-galaxy separation, we used a colourcolour diagram, as done above for the KiDS data.In this case, we adopted (g − r, z − y) diagram that is particularly efficient in this sense.Consequently, we selected all point-like sources located within the polygon defined by the criteria: (g − r, z − y) = (−0.3,−0.12); (−0.3, 0.08); (1.2, 0.25); (1.2, 0).The following sections of this manuscript are based on the analysis of this catalogue.

Analysis of the CMD
In Fig. 2, we show the results of the analysis of the deeper HSC-SSP catalogue.The top-left panel shows the density map of the stars filtered by the best-fit isochrone (indicated by a red solid line in the bottom row of the same figure), fully confirming the detection of a strong and compact stellar over-density at the expected location.In contrast, the right panel of the same figure illustrates the density map for sources catalogued as galaxies based on the 'extendedness' parameter and, subsequently, filtered by the same isochrone.Notably, no discernible overdensity of galaxies in the same region is observed.
Panel a of the second row of Fig. 2 displays the CMD8 of all stars located within twice the half-light radius of the structure (for the determination of r h , please refer to Sect.4.3).The sub-giant branch (SGB) and the MSTO associated with the RGB already detected in KiDS data are unequivocally revealed.Notably, these features are not present in the CMD of the control field (same area, located 20 from the centre of the overdensity; see also the Hess diagram obtained by subtracting the two former CMDs).This compelling evidence strongly suggests that our candidate is a genuine stellar system, likely a faint local dwarf galaxy, which has hitherto gone unreported in the existing literature.In what follows, we refer to it as Sextans II (KiDS-UFD-1) 9 .
An in-depth analysis of the CMD also discloses the presence of four stars likely belonging to the HB evolutionary phase.Specifically, two stars with colours of −0.06 mag and −0.08 mag reside within the half-light radius of the galaxy.The third star, exceeding the r h , exhibits g − r = −0.09mag, while the fourth one exhibits g − r = 0.03 and is located at ∼8.3 from the centre of the system ( 1.8r h , see Sect.4.3).The absolute mag-9 Concurrently and independently to our research, Homma et al. (2023) also reported the discovery of Sextans II by using the HSC-SSP catalogue.
nitude of the HB is a valuable distance indicator and we used it as the strongest constraint on the distance modulus.Through a visual isochrone fitting, we re-determined the parameters that provide the best match to the observed CMD: log(t/yr) 10.12 dex; [Fe/H] −2.2 dex; and m − M = 20.8 ± 0.2 mag.This is the red solid line superimposed on all panels in the second row of Fig. 2.
The resulting distance modulus corresponds to a physical distance of approximately 150 kpc, well within the range of Galactic satellites.The green isochrone overlaid on panel a in the second row of Fig. 2 deviates from the best-matching isochrone only in terms of age.Specifically, it is characterised by log(t/yr) = 10.00 dex, showing that decreasing the age of the L13, page 4 of 7 stellar model by ∼ ± 2−3 Gyr does not significantly affect the absolute magnitude of the HB (hence, our distance estimate), as expected.
Finally, panel d in the second row of Fig. 2 presents the CMD of Sextans II, including all stars confined within 4 × r h .This larger region enables a more comprehensive search for additional HB stars that might be associated with the system, albeit at the cost of higher contamination from MW stars within the CMD.We identified a potential fifth HB member at ∼15.9 from the galaxy centre, corresponding to 3.4 × r h .This star has a colour of −0.09 mag, similar to the other three HB stars, and indeed it is barely discernible in the main panel.To enhance the visibility of these four closely colour-matched HB stars, we provide a zoomed-in view of the HB evolutionary phase in an inset at the top of panel d, confined to a narrow range of colours encompassing those of the four stars.

Structural parameters
To estimate the structural parameters, such as the centre, halflight radius, ellipticity, and position angle of the newfound stellar system, we adopted the maximum likelihood approach presented in Martin et al. (2008).In particular, Martin et al. (2008) model the two-dimensional (2D) position of the stars in the sky in the surroundings of the galaxy with the combination of the exponential profile of the galaxy and of a uniform background.
We adopted a Markov chain Monte Carlo (MCMC) technique to contemporaneously estimate the free parameters of our model: x 0 ; y 0 ; r h ; e; θ; N stars , employing only the stars located within the best-fit isochrone mask.Our MCMC implementation utilised 1000 different walkers, each executing a chain composed of 10000 steps.We discarded the first 1000 steps as burn-in, resulting in a robust estimation of the model parameters.The MCMC analysis was conducted using the emcee (Foreman-Mackey et al. 2013) Python library.The bestfit parameters and their uncertainties were determined from the median, 16th, and 84th quantiles of the posterior distributions.In Fig. 3, we present a corner plot displaying the posterior distribution of the six free parameters, while Table 1 lists the median values and uncertainties of these parameters.
The left panel of Fig. 4 illustrates the radial density profile (RDP) of the system, where r represents the elliptical radius, as defined in Eq. ( 4) in Martin et al. (2008).The red solid line denotes the best fit of the model for the 2D position of the stars in the sky derived by employing the median values of the six parameters obtained through MCMC analysis.It is interesting to note that, both the derived half-light radius, r h = 194 +61 −45 pc (assuming D = 145 kpc), and ellipticity e = 0.46 +0.11  −0.15 are well within the range typical for local UFDs (see the latest version of the McConnachie 2012, catalogue).The right panel of Fig. 4 displays the 2D relative position of the filtered stars with respect to the galaxy centre, along with ellipses aligned according to the position angle determined by the MCMC procedure.These ellipses are defined by semi-major axes of 2r h (dashed ellipse) and 4r h (solid ellipse).Additionally, we indicate the relative position of the five stars in the HB evolutionary stage as cyan diamonds.As already stated, four HB stars are situated within 2r h , while the fifth one is at almost 4r h .

Integrated magnitude and stellar mass
In this section, we outline the procedure employed to estimate the absolute magnitude in the V-band for the purpose of comparing the properties of Sextans II with other recently discov- −3.9 ± 0.4 mag ered UFDs.Our approach follows the methodology reported in Martin et al. (2008), which is based on Monte Carlo sampling of a synthetic CMD representative of Sextans II.Specifically, we built a synthetic stellar population with log(t/yr) = 10.12 dex; [Fe/H] ∼ −2.2 dex, placed at a distance modulus 20.8 mag.We considered a total of N stars = 52 ± 12, that is the value obtained from the MCMC approach by adopting a cut at r = 24 mag, which is a conservative threshold adopted to mitigate potential issues of completeness in the central regions of the galaxy that might bias the estimation of its absolute magnitude.We generated 1000 realisations of the CMD, populating it with a Kroupa (2002) initial mass function (IMF), and subsequently computed the average absolute magnitude in the g and r SDSS filters, as L13, page 5 of 7 well as in the Johnson V band10 .Additionally, we determined the stellar mass of the galaxy.The median values, along with the 16th and 84th quantiles, were adopted to represent the best values and associated uncertainties, respectively.These results are presented in the final rows of Table 1.
Figure 5 depicts the M V -r h plane for galaxies and candidate galaxies reported in catalog by McConnachie (2012;version 2021), as well as for galactic globular clusters included in the catalog by Harris (1996;version 2010).The position of Sextans II within this plot clearly establishes it as one of the UFDs with the lowest surface brightness discovered to date, although it is important to note that some extremely faint low-surface brightness galaxies identified in recent years may not be represented in this figure.

Summary and conclusions
In this Letter we report on the discovery of an old and metal-poor stellar system located in the outskirts of the Milky Way, Sextans II (KiDS-UFD-1).This discovery resulted from an extensive search for low-surface brightness stellar systems in the MW halo within the KiDS survey catalogues.The KiDS detection, limited by photometric depth to the RGB and HB of the system, was confirmed with deeper HSC-SSP photometry which clearly unveiled the old MS of Sextans II down to about 1 mag below the MSTO.The isochrone fitting of the CMD revealed that the newly discovered system is ancient (t ≥ 10 Gyr) and metal-poor ([Fe/H] ≤ −1.5) and is located at a distance of D 145 kpc from the Sun and R GC 128.8 kpc from the center of the Galaxy.The structural parameters of Sextans II are typical of UFDs in the Local Group, strongly suggesting that the new system is a faint spheroidal satellite of the MW at the low end of the surface brightness distribution of local dwarfs.The most plausible alternative hypothesis is a disrupting globular cluster.However, even the most extended known globular clusters in the MW have r h ≤ 30.0 pc, more than five times smaller than what we determined for Sextans II; hence the UDF L13, page 6 of 7 hypothesis is, by far, the most likely.The final word on the nature of the system can be provided only by a proper spectroscopic follow up of a reasonable sample of member stars, which may be challenging, given the magnitude range spanned by candidate RGB members 11 .The chemical composition can reveal the presence of a metallicity spread, while the kinematics can reveal the amount of non-baryonic dark matter required to keep the system at the dynamical equilibrium, both features that are generally used to discriminate between dwarf galaxies and star clusters (Willman & Strader 2012).
The research of unknown stellar systems is still ongoing, and Sextans II might not be the only newfound stellar system detected in the KiDS DR4 catalog.Moreover, this work underscores the importance of extending this research to the forthcoming fifth release of the KiDS survey, that will provide multiband photometric data covering the entire 1350 square degrees of the KiDS footprint.This discovery serves as a reminder that the census of the Milky Way's satellite galaxy population is far from complete.The upcoming wide and deep surveys, such as the Legacy Survey of Space and Time (LSST) set to be conducted at the imminent Vera Rubin Observatory, will likely contribute significantly to expanding our knowledge of Milky Way satellites, much like the impactful contributions of the SDSS and PanSTARRS surveys before it. 11In the recently published FASHI catalogue of HI sources (Zhang et al. 2024) we found a source at cz = 10 443.8 km s −1 (FASHI 20230004672, estimated distance ∼140 Mpc) that coincides in the sky with Sextans II, has a characteristic size (3.9 ) comparable with the r h of Sextans II, and a mass compatible with a galaxy group.While we are confident that our colour cuts get rid of most of the contamination by unresolved galaxies, only spectroscopic follow-up will ascertain if there is significant contamination of Sextans II from galaxies possibly associated with this distant HI source.

Fig. 1 .
Fig. 1.Density map of a region with radius 1 • centered on the galaxy candidate.The density map is obtained using stars from the KiDS catalog, and filtered by an isochrone with log(t/yr) = 10.12 dex, [Fe/H] ∼ −1.5 and a distance modulus = 20.4mag (first panel).We adopted a Gaussian kernel with bandwidth = 2.0 to smooth the figure.CMD of stars located within 5 from the over-density center.A red isochrone with log(t/yr) = 10.12 dex; [Fe/H] ∼ −1.5 dex, and m − M = 20.4mag is superimposed on the figure (second panel).Cyan squares indicate the position of stars compatible with the HB evolutionary stage.CMD of stars within a local field having the same area covered by the CMD built for the dwarf galaxy candidate (third panel).Hess diagram obtained by subtracting the CMDs displayed in the second and third panels (fourth panel).

Fig. 2 .
Fig. 2. Top row: density map of stars included in the HSC-SSP catalog, and filtered by an isochrone characterised by log(t/yr) = 10.12 dex; [Fe/H] ∼ −2.2 dex, and m − M = 20.8mag (left).The depicted region encompasses 80 in radius centered on the dwarf galaxy.The figure has been smoothed with a Gaussian kernel with bandwidth = 2 .Density map of sources included in the HSC-SSP catalog and classified as galaxies, filtered by the same isochrone parameters adopted for the stars (right).Bottom row: CMD of the stars located within 2 × r h from the galaxy center (panel a).A red (green) isochrone with log(t/yr) = 10.00 (log(t/yr) = 10.12);[Fe/H] ∼ −2.2 dex, and m − M = 20.8mag is overlaid to the figure.CMD of stars within a local field of the same area as that used to construct the CMD in the panel to the left (panel b).Hess diagram obtained by subtracting the CMDs displayed in panels a and b (panel c).CMD of the stars located within 4 × r h from the galaxy center (panel d).An inset positioned at the top of the panel offers a zoomed-in view of the HB stage, capturing a narrow colour interval and highlighting the presence of four closely-aligned HB stars.

Fig. 3 .
Fig. 3. Posterior probability distributions for each of the six free parameters of the model, derived from the MCMC approach described in Sect.4.3.

Fig. 4 .
Fig.4.Radial density profile of KiDS-UFD-1 (left).Here, r ell is the elliptical radius defined in Eq. (4) inMartin et al. (2008).The red solid line refers to the best-fit model with the parameters derived via the MCMC approach described in Sect.4.3, while the gray lines represent 1000 random extractions from the posterior distributions of the free parameters of the model.Relative position of all stars filtered by the best-fit isochrone with respect to the galaxy center (right).The red dashed and solid ellipses have ellipticity and position angle derived from the MCMC procedure and semi-major axes of 2 × r h and 4 × r h , respectively.Cyan diamonds indicate the relative position of the five HB stars.

Fig. 5 .
Fig. 5. Absolute integrated V-band magnitude as a function of the log of the half-light radius for local low-luminosity stellar systems.Galactic globular clusters, represented as open grey triangles, are from the 2010 version of the Harris (1996) catalogue.Confirmed and candidate dwarf galaxies, represented as filled circles colour-coded according to their heliocentric distance, are taken from the 2021 version of the catalog by McConnachie (2012).The same colour-coding is applied to the filled square marking the position of Sextans II, that is also highlighted with a red contour.The stellar systems with M V and r h values that are most similar to Sextans II have been labelled.The diagonal lines are loci of constant surface brightness, from at µ = 24.0mag arcsec −2 (top) down to µ = 34.0mag arcsec −2 (bottom), in steps of µ = 2.0 mag arcsec −2 .
Forschungsgemeinschaft (DFG) SFB1491.N.R.N. acknowledges financial support from the National Science Foundation of China, Research Fund for Excellent International Scholars (grant n. 12150710511), and from the research grant from China Manned Space Project n.CMS-CSST-2021-A01.