A Massive, Dusty HI-Absorption-Selected Galaxy at $z \approx 2.46$ Identified in a CO Emission Survey

We report a NOrthern Extended Millimeter Array (NOEMA) and Atacama Large Millimeter/submillimeter Array (ALMA) search for redshifted CO emission from the galaxies associated with seven high-metallicity ([M/H] $\geq -1.03$) damped Lyman-$\alpha$ absorbers (DLAs) at $z\approx1.64-2.51$. Our observations yielded one new detection of CO(3-2) emission from a galaxy at $z=2.4604$ using NOEMA, associated with the $z=2.4628$ DLA towards QSO B0201+365. Including previous searches, our search results in detection rates of CO emission of $\approx56^{+38}_{-24}$ % and $\approx11^{+26}_{-9}$ %, respectively, in the fields of DLAs with ${\rm [M/H]}>-0.3$ and ${\rm [M/H]}<-0.3$. Further, the HI-selected galaxies associated with five DLAs with [M/H] $>-0.3$ all have high molecular gas masses, $\gtrsim5\times10^{10}\ {\rm M}_\odot$. This indicates that the highest-metallicity DLAs at $z\approx2$ are associated with the most massive galaxies. The newly-identified $z\approx2.4604$ HI-selected galaxy, DLA0201+365g, has an impact parameter of $\approx7$ kpc to the QSO sightline, and an implied molecular gas mass of $(5.04\pm0.78) \times10^{10}\times(\alpha_{\rm CO}/4.36)\times(r_{31}/0.55)\ {\rm M}_\odot$. Archival Hubble Space Telescope Wide Field and Planetary Camera 2 imaging covering the rest-frame near-ultraviolet (NUV) and far-ultraviolet (FUV) emission from this galaxy yield non-detections of rest-frame NUV and FUV emission, and a $5\sigma$ upper limit of 2.3 M$_\odot$ yr$^{-1}$ on the unobscured star formation rate (SFR). The low NUV-based SFR estimate, despite the very high molecular gas mass, indicates that DLA0201+365g either is a very dusty galaxy, or has a molecular gas depletion time that is around two orders of magnitude larger than that of star-forming galaxies at similar redshifts.


INTRODUCTION
The highest Hi column density absorbers in QSO absorption spectra, the damped Lyman-α absorbers (DLAs, with N(Hi) ≥ 2 × 10 20 cm −2 ; Wolfe et al. 1986) are expected to arise in gas associated with galaxies (e.g. Wolfe et al. 2005). The presence of a DLA in a quasar spectrum thus allows one to identify high-z galaxies via their Lyman-α absorption signature, rather than via their luminosity. Such Hi-absorption-selected galaxies do not have the luminosity bias that affects emission-selected samples, making them complementary probes of galaxy evolution that are sensitive to galaxies with effectively any mass.
Many attempts have been made, mostly at optical and near-IR wavelengths, to identify and characterize high-z Hiabsorption-selected galaxies, and thereby establish their connection to the emission-selected population (e.g. Warren et al. 2001;Kulkarni et al. 2006;Péroux et al. 2011;Fumagalli et al. 2015;Krogager et al. 2017;Mackenzie et al. 2019). A number of these studies (e.g. Møller et al. 2004;Fynbo et al. 2010;Wang et al. 2015) have focussed on DLAs with high absorption metallicities, as the correlation between DLA metallicities and the velocity widths of the low-ionization metal lines suggest the existence of a mass-metallicity relation in DLAs (e.g. Ledoux et al. 2006;Prochaska et al. 2008;Neeleman et al. 2013). However, the presence of the background quasar, which is very bright at these wavelengths relative to the foreground galaxy, has been a significant hindrance to these efforts. As a result, only ≈ 20 Hi-selected galaxies have been identified via such studies at high redshifts, z 2 (e.g. Møller et al. 2004;Fynbo et al. 2013;Krogager et al. 2017;Mackenzie et al. 2019;Ranjan et al. 2020).
In the last few years, with the advent of the Atacama Large Millimeter/submillimeter Array (ALMA), redshifted CO rotational and [Cii] 158µm fine-structure emission lines have emerged as a promising tool to identify and characterize Hi-selected galaxies at high redshifts (Neeleman et al. 2016(Neeleman et al. , 2017(Neeleman et al. , 2019Péroux et al. 2019;Klitsch et al. 2019;Kanekar et al. 2018Kanekar et al. , 2020Kaur et al. 2021), yielding a number of surprises. For example, Hi-selected galaxies at z ≈ 4 identified by their [Cii] 158µm emission have large impact parameters, b ≈ 16 − 45 kpc, from the QSO sightline (Neeleman et al. 2017(Neeleman et al. , 2019, far larger than expected for sightlines that pass through the interstellar medium of galaxies. This indicates that the DLA sightlines significantly traverse the circumgalactic medium (CGM), with high Hi column densities present in the CGM of high-z galaxies. One of the [Cii] 158µm emitters at z ≈ 4.26 has been shown to be a massive, rotating disk galaxy, the first such cold disk to be identified at z 4 . At z ≈ 2, ALMA CO searches have found evidence that a number of Hi-selected galaxies associated with high-metallicity DLAs have both remarkably large molecular gas masses, 5 × 10 10 M , and a range of impact parameters, b ≈ 6 − 100 kpc Fynbo et al. 2018;Kanekar et al. 2020). Follow-up multi-line CO studies have found evidence for highly-excited mid-J CO rotational levels in three of the CO-emitting, Hi-selected galaxies at z ≈ 2 (Klitsch et al. 2022;Kaur et al. 2022).
The very large molecular gas masses of the Hi-selected galaxies at z ≈ 2 is an especially surprising result of the ALMA CO and [Cii] 158µm studies. However, the number of absorber fields at z ≈ 2 that have been searched for CO emission remains quite small, with 5 CO detections in 12 searches so far (Kanekar et al. 2020). The recent upgrade of the NOrthern Extended Millimetre Array (NOEMA) has opened the possibility of carrying out such searches for redshifted CO emission from Hi-selected galaxies at northern declinations, i.e. those inaccessible to ALMA. In this paper, we report a NOEMA and ALMA search for redshifted CO(2-1) or CO(3-2) emission from the fields of seven high-metallicity DLAs at z ≈ 2, with absorption metallicity [M/H] −1.0, which has resulted in one new CO detection, in the field of the z ≈ 2.4628 DLA towards QSO B0201+365. We also report a search for rest-frame far-ultraviolet (FUV) and NUV emission from the new Hi-selected galaxy, using Hubble Space Telescope (HST) Wide Field and Planetary Camera 2 (WFPC2) archival data.

NOEMA and ALMA observations
We used NOEMA and ALMA to carry out a search for redshifted CO(2-1) or CO(3-2) emission in the fields of seven DLAs at z ≈ 1.64 − 2.51. The targetted DLAs have an absorption metallicity, [M/H] ≥ −1.03, higher than the median DLA metallicity (≈ −1.3; Rafelski et al. 2014) in this redshift range. Six of the DLAs were observed with the NOEMA Band-1 receivers between December 2017 and April 2018 (project: W17DG; PI: M. Neeleman), in the D-configuration. The PolyFix correlator was used as the backend, with an instantaneous bandwidth of 8 GHz centred on the redshifted CO(2-1) or CO(3-2) frequency of the target DLA, and sub-divided into 4000 channels, giving a frequency resolution of 2 MHz. The seventh DLA was observed with the ALMA Band-4 receivers (proposal: 2016.1.00628.S; PI: J. X. Prochaska) in January 2017. One of the 1.875 MHz intermediate frequency (IF) bands was used in FDM mode to cover the redshifted CO(3-2) line frequency, with 480 channels and a spectral resolution of 3.9 MHz, while the remaining three IF bands used the TDM mode, with a 2 GHz bandwidth and 128 channels, to cover the continuum emission. The on-source times were 4 − 9 hours (NOEMA) and 1.5 hours (ALMA). We note that the NOEMA and ALMA fields of view (full width at half maximum, FWHM) at the redshifted CO line frequencies are 50 , corresponding to a spatial field of view of 400 kpc at the DLA redshifts. 1 The observational details are summarized in Table 1.
The initial editing and calibration were carried out in gildas 2 for the NOEMA data, and with the standard ALMA pipeline in the Common Astronomy Software Applications package (casa version 5.6; McMullin et al. 2007) for the ALMA data. Only one of the seven QSOs, QSO B0201+365, had a sufficient flux density (≈ 31 mJy) for self-calibration; this was carried out in the Astronomical Image Processing System package ("classic" aips; Greisen 2003), following standard procedures. Any detected continuum emission in each field was then subtracted out from the calibrated spectral-line visibilities, after Fourier-transforming the continuum image to the visibility plane. The residual spectralline visibilities were then imaged in casa, using the task tclean to produce spectral cubes at velocity resolutions of 25 − 200 km s −1 . The cubes were then visually inspected to search for redshifted CO(2-1) or CO(3-2) emission.
No statistically significant (≥ 5σ) emission was detected in the fields of the remaining six DLAs, at or near the redshifted CO line frequency. The 3σ upper limits on the velocity-integrated CO(3-2) or CO(2-1) line flux densities are (0.05 − 0.14) Jy km s −1 , assuming a line FWHM of 200 km s −1 , all below the detection for DLA0201+365g. The corresponding 3σ upper limits on the CO(2-1) or CO(3-2) line luminosities are listed in the last column of Table 1.

HST-WFPC2 data
The field of QSO B0201+365 was observed with HST-WFPC2 in October 1999 with the F450W and F814W filters (program ID: 8085; PI: S. Beckwith). The two filters cover, respectively, the rest-frame FUV and NUV emission from the z = 2.4604 Hi-selected galaxy, with two exposures in a single orbit in each filter. After removing cosmic rays with a custom application 3 of LACOSMIC (van Dokkum 2001), the individual exposures in each filter were combined with DrizzlePac (Hack et al. 2020). Specifically, the HST images were aligned with TweakReg using the Gaia DR3 catalogue as the astrometric reference (Gaia Collaboration 2021). The routine AstroDrizzle was used to drizzle the combined images, with a final pixel fraction of 1 at a pixel scale of 0.0996 /pixel. Since DLA0201+365g is located close to one of the diffraction spikes of the QSO point-spread function (PSF), subtraction of the QSO PSF is critical to detect the faint galaxy. The QSO PSF was subtracted using a PSF model obtained from stacking four stars (using a custom code hst wfc3 psf modeling 4 ) on the same detector chip as the QSO, and separately, from a PSF model obtained by rotating the QSO itself by 180 • ; similar results were obtained from the two approaches. In the following, we use results from the latter approach. Fig. 2 shows overlays of the NOEMA CO(3-2) emission from DLA0201+365g on the HST-WFPC2 F450W and F814W images. We detect no statistically significant emission in the HST images at the location of DLA0201+365g, with 5σ upper limits of m AB < 24.9 in the F450W filter and m AB < 25.7 in the F814W filter, corresponding to the rest-frame FUV and NUV wavelengths for DLA0201+365g, respectively. Assuming a Kroupa initial mass function (Kroupa 2001) and the SFR calibration of Kennicutt & Evans (2012), we obtain 5σ upper limits on the unobscured SFR of < 4.9 M yr −1 (FUV) and < 2.3 M yr −1 (NUV).

DISCUSSION
Our new NOEMA and ALMA CO searches have yielded one detection of redshifted CO emission from the Hi-selected galaxy DLA0201+365g in the field of the z = 2.4628 DLA towards QSO B0201+365, and six non-detections of CO emission in the fields of the other DLAs. The measured CO line luminosities of Table 1 can be converted to molecular gas masses via the relation, M mol = α CO ×r J1 ×L CO(J→J−1) , where α CO is the CO-to-H 2 conversion factor (e.g. Bolatto et al. 2013) and r J1 ≡ L CO(J→J−1) /L CO(1−0) is the ratio of CO(J → J − 1) and CO(1 − 0) line luminosities (e.g. Carilli & Walter 2013). The value of of α CO depends on physical conditions in the gas, including temperature, density, and metallicity (Bolatto et al. 2013;Carilli & Walter 2013;Tacconi et al. 2020). Following Kanekar et al. (2020), we assume α CO = 4.36 M (K km s −1 pc 2 ) −1 for all targets, a reasonable assumption for non-starburst galaxies with near-solar metallicity. We further assume sub-thermal excitation of the J = 2 and J = 3 rotational levels, with r 21 = 0.77 and r 31 = 0.55, applicable to high-z main-sequence galaxies (Daddi et al. 2015;Tacconi et al. 2020). In the case of DLA0201+365g, we obtain a high molecular gas mass of M mol = (5.04 ± 0.78) × (r 31 /0.55) × (α CO /4.36) × 10 10 M .    (1) the QSO name, (2) the QSO redshift, (3) the DLA redshift, (4) the observed CO rotational transition, (5) the redshifted CO line frequency, in GHz, (6) the on-source time, in hours, (7) the synthesized beam, in arcseconds, (8) the velocity resolution of the CO spectral cube used for the search, in km s −1 , (9) the RMS noise on the cube, in µJy/Bm, (10) the integrated CO line flux density, or the 3σ upper limit to this quantity, assuming a boxcar profile with a width of 200 km s −1 , and (11) the CO(2-1) or CO(3-2) line luminosity, L CO , or the 3σ upper limit to this quantity.  Table 2. Absorption properties of the seven DLAs, and emission properties of the associated Hi-selected galaxies. The columns are (1) the QSO name, (2) the DLA redshift, (3) the logarithm of the Hi column density (in cm −2 ), (4) the absorption metallicity, [M/H], (5) the velocity spread of the low-ionization metal lines, ∆V90, in km s −1 , (6) the CO(1-0) line luminosity or the 3σ upper limit to this quantity, in 10 9 K km s −1 pc 2 , assuming r21 = 0.77 and r31 = 0.55 (Tacconi et al. 2020), (7) the molecular gas mass of the Hi-selected galaxy, or the 3σ upper limit to this quantity, in 10 10 M , and (8)  For the CO non-detections in the six DLA fields, our upper limits on the CO(2-1) or CO(3-2) line luminosity yield 3σ upper limits of M mol < (0.6 − 3.3) × 10 10 M on the molecular gas mass of the Hi-selected galaxies associated with the DLAs. Table 2 summarizes the absorption and emission properties of the DLAs and the Hi-selected galaxies.
We note that Klitsch et al. (2022) find tentative evidence for a lower α CO value in two of the z ≈ 2 Hi-selected galaxies of the sample of Kanekar et al. (2020), at z ≈ 2.1929 towards PKS B1228-113 and z ≈ 2.5832 towards QSO J0918+1636. These two galaxies also show near-thermal excitation of the J = 3 rotational level (Kaur et al. 2022). However, these galaxies have the highest measured CO(3-2) line luminosities of the sample, and it is not clear whether their properties are representative of the full sample. Larger samples are needed to explore the excitation conditions of high-z Hi-selected galaxies.
The newly-detected Hi-selected galaxy DLA0201+365g has a high inferred molecular gas mass, ≈ 5 × 10 10 M . Remarkably, the CO line FWHM is only ≈ 125 km s −1 , significantly smaller than the CO FWHMs ( 300 km s −1 ; Kanekar et al. 2020) of the other Hi-selected galaxies at z ≈ 2 with CO detections. This suggests that DLA0201+365g has a low inclination to our line of sight. Conversely, the low-ionization metal absorption lines are both wide, ∆V 90 ≈ 200 km s −1 (Prochaska & Wolfe 1996;Neeleman et al. 2013), and asymmetric, with the strongest Siiiλ1808Å absorption of Fig. 1[C] seen to lie at one edge of the line, ≈ 200 km s −1 redward of the CO emission. It is not straightforward to reconcile the narrow observed CO emission with the wide metal absorption lines. One possibility is that the CO emission is detected only from the central regions of the galaxy, which may have higher excitation of the mid-J CO rotational levels, yielding stronger CO(3-2) emission. Alternatively, part of the lowionization metal line absorption may arise in dense clumps in the CGM of the Hi-selected galaxy, or possibly in a companion galaxy not detected in CO emission.
Our non-detection of DLA0201+365g in the HST-WFPC2 F814W image of Fig. 2, covering its rest-frame NUV emission, implies an upper limit of 2.3 M yr −1 on the unobscured SFR of the Hi-selected galaxy. Wang et al. (2015) used the Gemini Near-Infrared Integral Field Spectrometer (NIFS) to search for redshifted Hα emission from galaxies in the field of the 2.4628 DLA, reporting a non-detection of Hα emission. We re-inspected the Gemini-NIFS Hα cube at the spatial position of the CO emission, and found no evidence for Hα emission at the position of the CO emission. The Hα non-detection places the 3σ upper limit of ≈ 2.2 M yr −1 on the SFR of DLA0201+365g (Wang et al. 2015), similar to the SFR limit from the non-detection of rest-frame NUV emission. We emphasize that these limits are on the dust-unobscured SFR, as any dust obscuration would attenuate the rest-frame NUV and the Hα emission.
Combining our molecular gas mass estimate with the upper limit on the rest-frame NUV SFR allows us to estimate the molecular gas depletion timescale, t dep ≡ M mol /SFR, of DLA0201+365g. This is the timescale on which all the molecular gas of a galaxy would be converted to stars at its current SFR, assuming no replenishment of the gas reservoir. We obtain t dep > 22 Gyr for DLA0201+365g (> 15 Gyr, on including the 2σ uncertainty in the molecular gas mass), more than an order of magnitude larger than the molecular gas depletion timescales in star-forming galaxies at z ≈ 2.5 (e.g. Tacconi et al. 2020) or in the local Universe (e.g. Saintonge et al. 2017).
Large molecular gas depletion timescales (3 − 120 Gyr) have also been found in Hi-selected galaxies at intermediate redshifts, z ≈ 0.5 − 0.8 (Møller et al. 2018;Kanekar et al. 2018). However, in at least one case here, the z ≈ 0.7163 Hi-selected galaxy J1323-0021g, the dust-obscured SFR, estimated from the Herschel image, is ≈ 3 times higher than the SFR estimate from the Hα line (Møller et al. 2018). This implies a significantly lower molecular gas depletion time, ≈ 5 Gyr, when obscuration effects are taken into account.
Similar to the z ≈ 0.7163 Hi-selected galaxy, J1323-0021g, a more plausible explanation of the low NUV and Hα SFR estimates in DLA0201+365g is dust obscuration of the NUV and Hα emission. An independent estimate of the total SFR may be obtained from the typical molecular gas depletion timescales of star-forming galaxies at the DLA redshift. We estimate t dep ≈ 0.22 Gyr for DLA0201+365g from the median of measurements of t dep in star-forming galaxies at z ≈ 2.46 (Tacconi et al. 2020). Combining this with the molecular gas mass estimate of 5.04 × 10 10 M gives a total SFR of ≈ 230 M yr −1 , two orders of magnitude larger than the upper limits on the SFR from the rest-frame NUV and Hα non-detections. We note that even a molecular gas depletion time of ≈ 1 − 2 Gyr, similar to that of galaxies in the local Universe (e.g. Leroy et al. 2013;Saintonge et al. 2017), would yield a total SFR of ≈ 50 M yr −1 . This suggests that DLA0201 is likely to be a very dusty galaxy.
The ratio of the Zn abundance to the Fe (or Cr) abundance in DLAs has been used to quantify the extent of depletion of refractory elements onto dust grains (e.g. Pettini et al. 1994Pettini et al. , 1997Hou et al. 2001;Prochaska & Wolfe 2002;Ledoux et al. 2003;De Cia et al. 2016), although challenges exist such as observations showing that Zn traces S (Fenner et al. 2004;Rafelski et al. 2012;Roman-Duval et al. 2022). Specifically, we note that there are caveats in using [Zn/Fe] as a dust indicator, as the origins of Zn are uncertain and the Zn abundance also depends on the star-formation history of a galaxy (e.g. Fenner et al. 2004 (Prochaska et al. 2007), indicating that the DLA sightline is dusty, with substantial Fe depletion. This is broadly consistent with the result that DLA0201+365g is a dusty galaxy. The depletion in DLA0201+365 is similar to that in the other z ≈ 2 DLAs that show CO emission from the associated galaxies (Kanekar et al. 2020 here, the background QSO shows significant dust reddening (Krogager et al. 2016).
The sample of z ≈ 2 DLAs with searches for redshifted CO emission consists of 19 systems, with six CO detections and 13 non-detections Fynbo et al. 2018;Kanekar et al. 2020, this work), i.e. a CO detection rate of ≈ 32%. All 19 DLAs have absorption metallicities ≥ −1.03, i.e. higher the median DLA metallicity (≈ −1.3; Rafelski et al. 2014) in this redshift range. Fig. 3 plots the measured or inferred CO(1-0) line luminosity of the Hi-selected galaxies against the absorption metallicity of the corresponding DLAs. The figure shows that a transition from a high CO detection rate to a low detection rate appears to take place at the metallicity 3 shows CO emission, yielding a CO detection rate of ≈ 11 +26 −9 %. Thus, although the sample remains small, we find evidence for a higher detection rate of CO emission from galaxies associated with high-metallicity DLAs.
Further, all five CO emitters associated with the highest-metallicity DLAs, with [M/H] > −0.3, have high CO line luminosities, L CO(1−0) > 10 10 K km s −1 pc 2 . This suggests that the CO line luminosities of Hi-selected galaxies may depend on the absorption metallicity. We tested this hypothesis by carrying out Peto-Prentice and Gehan generalized Wilcoxon tests, using the asurv package (Feigelson & Nelson 1985), comparing the CO line luminosity distributions of galaxies associated with two sub-samples of DLAs, with metallicities above and below [M/H]= −0.3. We find that the null hypothesis, that the two samples are drawn from the same CO line luminosity distribution, is ruled out at ≈ 2.3σ significance. We thus find evidence that both the CO detection rate and the CO line luminosity are higher in Hi-selected galaxies associated with the highest-metallicity DLAs, with [M/H] > −0.3.

SUMMARY
We have used NOEMA and ALMA to search for redshifted CO(2-1) or CO(3-2) emission in the fields of seven highmetallicity ([M/H] ≥ −1.03) DLAs at z ≈ 1.64 − 2.51. Our searches yielded one CO detection and six non-detections of CO emission, the latter providing 3σ upper limits of (0.6 − 3.3) × 10 10 M on the molecular gas masses of the associated galaxies. Thus far, including the work presented here, the fields of 19 high-metallicity DLAs at z ≈ 2 have been searched for CO emission from the associated Hi-selected galaxies, with 6 detections of CO emission, i.e. a CO detection fraction of ≈ 32%. We find evidence for both a higher CO detection fraction and higher CO line luminosities in Hi-selected galaxies associated with the highest-metallicity DLAs, with [M/H] > −0.3. The new NOEMA detection of redshifted CO(3-2) emission is from DLA0201+365g, at z = 2.4604, associated with the z = 2.4628 DLA towards QSO B0201+365. We obtain a high molecular gas mass, M mol = (5.04 ± 0.78) × 10 10 × (α CO /4.36) × (r 31 /0.55) M for this galaxy. The non-detection of rest-frame FUV and NUV emission from DLA0201+365g in HST-WFPC2 images of the field places an upper limit of ≈ 2.3 M yr −1 on its dust-unobscured SFR, similar to the limit from the earlier nondetection of Hα emission. This would imply that the galaxy has an extremely large molecular gas depletion timescale, 22 Gyr, more than an order of magnitude larger than that of star-forming galaxies at similar redshifts or in the local Universe. More plausibly, if the galaxy has a molecular gas depletion timescale similar to that of other galaxies at z ≈ 2.46, we obtain a total SFR of ≈ 230 M yr −1 , two orders of magnitude higher than the upper limits from the NUV and Hα non-detections, implying that the object would be a very dusty galaxy. The CO emission is significantly narrower than the spread of the low-ionization metal absorption lines suggesting that either the CO emission arises from only the central regions of the galaxy, or a significant fraction of the absorption arises in the CGM.
BK and NK acknowledge the Department of Atomic Energy for funding support, under project 12-R&D-TFR-5.02-0700. MN acknowledges support from ERC advanced grant 740246 (Cosmic Gas). We thank an anonymous referee for comments on an earlier version of the manuscript, Nichol Cunningham for much help with calibrating the NOEMA data, and Wei-Hao Wang for discussions and for providing us with the analysed Gemini-NIFS spectral cube. This work is based on observations carried out under project number W17DG  with the IRAM NOEMA Interferometer. IRAM is supported by INSU/CNRS (France), MPG (Germany) and IGN (Spain). This paper makes use of the following ALMA data: ADS/JAO.ALMA#2016.1.00628.S. ALMA is a partnership of ESO (representing its member states), NSF (USA) and NINS (Japan), together with NRC (Canada), MOST and ASIAA (Taiwan), and KASI (Republic of Korea), in cooperation with the Republic of Chile. The Joint ALMA Observatory is operated by ESO, AUI/NRAO and NAOJ. This research is based on observations made with the NASA/ESA Hubble Space Telescope obtained from the Space Telescope Science Institute, which is operated by the Association of Universities for Research in Astronomy, Inc., under NASA contract NAS 5-26555. These observations are associated with program 8085. This material is based upon work supported by the National Science Foundation under Grant No. 2107989.