A Survey of Ly α Emission around Damped Ly α Absorbers at z ≈ 2 with the Keck Cosmic Web Imager

We present Keck Cosmic Web Imager Ly α integral ﬁ eld spectroscopy of the ﬁ elds surrounding 14 damped Ly α absorbers ( DLAs ) at z ≈ 2. Of these 14 DLAs, nine have high metallicities ( [ M / H ] > − 0.3 ) , and four of those nine feature a CO-emitting galaxy at an impact parameter  30 kpc. Our search reaches median Ly α line ﬂ ux sensitivities of ∼ 2 × 10 − 17 erg s − 1 cm − 2 over apertures of ∼ 6 kpc and out to impact parameters of ∼ 50 kpc. We recover the Ly α ﬂ ux of three known Ly α -emitting H I -selected galaxies in our sample. In addition, we ﬁ nd two Ly α emitters at impact parameters of ≈ 50 – 70 kpc from the high-metallicity DLA at z ≈ 1.96 toward QSO B0551-366. This ﬁ eld also contains a massive CO-emitting galaxy at an impact parameter of ≈ 15 kpc. Apart from the ﬁ eld with QSO B0551-366, we do not detect signi ﬁ cant Ly α emission in any of the remaining eight high-metallicity DLA ﬁ elds. Considering the depth of our observations and our ability to recover previously known Ly α emitters, we conclude that H I -selected galaxies associated with high-metallicity DLAs at z ≈ 2 are dusty and therefore might feature low Ly α escape fractions. Our results indicate that complementary approaches — using Ly α , CO, H α , and [ C II ] 158 μ m emission — are necessary to identify the wide range of galaxy types associated with z ≈ 2 DLAs.


INTRODUCTION
The H I cycle within and around galaxies is a critical component in our models of galaxy formation and evolution.We know that galaxies must acquire H I from the intergalactic medium in order to sustain their starformation (Prochaska et al. 2005;Kereš et al. 2005;Walter et al. 2020).The inflow of H I is counteracted by outflows of metal-enriched gas powered by active galactic nuclei and/or the late stages of stellar evolution, thereby regulating the rate at which galaxies form their stars (e.g.Kereš et al. 2005;Tumlinson et al. 2017).Furthermore, the removal of H I from galaxies through environment-driven processes (Gunn & Gott 1972;Kawata & Mulchaey 2008;Cortese et al. 2021) is often invoked to explain the quenched fractions and assembly histories of satellite galaxies (e.g.Pasquali et al. 2010;Wetzel et al. 2013;Gallazzi et al. 2021;Trussler et al. 2021;Werle et al. 2022;Oyarzún et al. 2023).

Oyarzún et al.
To characterize the H I content in and around galaxies at low redshift, we often turn to studies of 21 cm emission (e.g.Verheijen 2001;Walter et al. 2008;Begum et al. 2008;Chung et al. 2009;Heald et al. 2011;Catinella et al. 2018).Single-dish 21 cm studies have yielded H I emission-line detections for thousands of galaxies at low redshifts (e.g.Zwaan et al. 2005;Haynes et al. 2018), while 21 cm mapping studies have been used to, for example, quantify the star-formation efficiency in nearby galaxies (Leroy et al. 2008), measure the sizes of H I disks (e.g.Wang et al. 2016), determine galaxy rotation curves (e.g.Walter et al. 2008;Begum et al. 2008), and search for extra-planar gas (e.g.Heald et al. 2011).However, the faintness of the 21 cm transition has so far prevented us from detecting H I in emission from individual galaxies beyond z ≈ 0.4 (Fernández et al. 2016) or in stacked imaging beyond z ≈ 1.4 (Chowdhury et al. 2020(Chowdhury et al. , 2021(Chowdhury et al. , 2022a)).
Alternatively, absorption signatures in the spectra of background quasars (QSOs) produced by high H I column density gas (Damped Lyα absorbers, or DLAs; Wolfe et al. 2005) remain the quintessential technique for studying H I at high redshift.Through DLA characterization, we have been able to constrain the column densities, metallicities, kinematics, dust depletion, molecular fractions, and gas temperatures of H I reservoirs up to z ∼ 5.5 (e.g.Balashev et al. 2017;Prochaska & Wolfe 1997;Kanekar et al. 2014;Noterdaeme et al. 2008;Neeleman et al. 2013Neeleman et al. , 2015;;Klimenko et al. 2020).Moreover, DLAs have been instrumental in determining the evolution of the metal enrichment and cosmic H I mass density since z ≈ 5 (e.g.Noterdaeme et al. 2012;Rafelski et al. 2012Rafelski et al. , 2014;;Jorgenson et al. 2013;Prochaska et al. 2013;Crighton et al. 2015;Rao et al. 2017), connecting with estimates at lower redshifts from 21 cm observations (e.g.Jones et al. 2018;Bera et al. 2019).
On the other hand, it has been challenging to associate the properties of H I measured in absorption with the properties of galaxies measured in emission.DLA galaxies are typically much fainter than the background QSOs, and are found over a wide range of impact parameter (b), which makes the identification of the galaxy through standard optical imaging and spectroscopy very challenging.Despite many searches, only about a dozen galaxies associated with DLAs at z ≳ 2 were detected in over a quarter of a century (e.g.Møller & Warren 1993;Fumagalli et al. 2015).Fortunately, the detection rate has since increased.After it was realized that the stellar mass and gas-phase metallicity relation (e.g.Tremonti et al. 2004) also holds for absorption-selected systems (e.g.Møller et al. 2004;Ledoux et al. 2006;Fynbo et al. 2008), studies in the rest-frame UV/optical have started to target high-metallicity DLAs ([M/H]≳ −1.3) with great success (e.g.Krogager et al. 2017).
At the same time, the advent of the Atacama Large Millimeter/sub-millimeter Array (ALMA) has enabled a search for DLA galaxies at millimeter and submillimeter wavelengths, where the QSOs are much fainter and line emission from cool or cold gas can be luminous.ALMA and Northern Extended Millimetre Array (NOEMA) images have been used to identify a further dozen star-forming counterparts of high-metallicity ([M/H]> −1.3) DLAs through their CO emission at z ≈ 2 or their [C ii] 158µm emission at z ≈ 4 (Neeleman et al. 2017(Neeleman et al. , 2018;;Fynbo et al. 2018;Neeleman et al. 2019;Kanekar et al. 2020;Kaur et al. 2022b).
Although searches for DLA galaxies in the CO and [C ii] 158µm transitions have shown to be quite efficient, the downside is that they are bound to miss galaxies with high CO-to-H 2 conversion factors or with relatively low molecular gas masses (≲ 1 − 5 × 10 10 M ⊙ ; Neeleman et al. 2019;Kanekar et al. 2020;Kaur et al. 2022b).Such undetected galaxies could reside at lower impact parameters than mm-detected galaxies, perhaps explaining why the impact parameters between mm-detected galaxies and DLA sightlines -b < 30 kpc at z ≈ 2 and b ≳ 25 kpc at z ≈ 4 (Neeleman et al. 2017(Neeleman et al. , 2018(Neeleman et al. , 2019;;Kanekar et al. 2020) -can exceed the sizes of H I gas reservoirs at z ≈ 2 − 4 in simulations (< 30 kpc; Rhodin et al. 2019;Stern et al. 2021).
To search for galaxies with high CO-to-H 2 conversion factors and/or low molecular gas masses, we can turn to rest-frame UV/optical emission.Among the standout emission lines at these wavelengths is Lyα, which is produced in H II regions by recombining H I gas.However, only ≈ 20 − 25% of Lyman-break galaxies (LBGs; Steidel et al. 1996Steidel et al. , 1999Steidel et al. , 2003;;Shapley et al. 2003;Stark et al. 2009) at z ≈ 2 show Lyα emission (e.g.Cassata et al. 2015).This is presumably due to the ease with which Lyα is scattered by H I and absorbed by dust grains (e.g.Dijkstra & Kramer 2012;Duval et al. 2014;Rivera-Thorsen et al. 2015;Gronke & Dijkstra 2016;Gronke et al. 2016).As a result, the equivalent width of the Lyα line is particularly high in galaxies with low H I gas covering fractions and low dust extinctions, i.e., low stellar-mass (M * ≲ 10 10 M ⊙ ) galaxies (e.g.Oyarzún et al. 2016Oyarzún et al. , 2017)).
Motivated by the success of recent searches in Lyα with IFUs, in this work we searched for Lyα emission in the fields of 14 DLAs -nine of which have been previously studied in mm-wave CO emission -with the Keck Cosmic Web Imager Integral Field Spectrograph (KCWI; Morrissey et al. 2018) on the Keck II telescope.KCWI is an outstanding instrument for this search because of its high sensitivity at the observedframe wavelength of z ≈ 2 Lyα emission (λ ≈ 4500 Å).Moreover, the effective field-of-view of ∼ 100 kpc enables us to cover the impact parameter range expected for the primary emission counterparts of high H I column density absorbers (≲ 30 kpc; e.g.Rahmati & Schaye 2014;Rhodin et al. 2019).
The paper is structured as follows.We define our target sample of DLAs and describe the observations in Sections 2 and 3. We detail our methodology to identify and characterize Lyα emission in Section 4. We present our results in Section 5, discuss their interpretation in Section 6, and provide a summary of the paper in Section 7. Throughout the paper, we assume H 0 = 70 km s −1 Mpc −1 .All magnitudes are reported in the AB system (Oke & Gunn 1983).

[M/H]
A third DLA in our sample, this time without ancillary CO observations, has also been detected in Lyα.This galaxy is at z = 2.0395 towards B0458-020 and at an impact parameter of b ≲ 3 kpc (Møller et al. 2004;Krogager et al. 2017).Together, the three known Lyα emitters in the fields of J2222-0946, J2206-1958, and B0458-020 constitute our control sample.They were used to assess the efficacy of our survey and to search for additional Lyα associations at larger impact parameters.

The "Blind" subsample
The remaining four DLA fields (Table 1) compose the "blind" sample, i.e. absorbers without earlier searches for the associated galaxies.One of these four DLAs lies towards J2206-1958, a sightline containing one of our control DLAs.The remaining three blind-sample DLAs were observed due to their convenient celestial coordinates in the context of our observational strategy.They will be the targets of future CO observations with ALMA , NOEMA, and/or the JVLA.

OBSERVATIONS
The 14 DLA fields in this work were observed with the Keck Cosmic Web Imager Integral Field Spectrograph (KCWI; Morrissey et al. 2018), on the Keck II telescope.KCWI is optimized for observations in the 3500−5600 Å spectral range with resolution R = 1000 − 20, 000.The size of a spaxel is ∼ 0. ′′ 7, which corresponds to a physical size of ∼ 6 kpc at z ∼ 2. We opted for the configuration with a 20 ′′ × 16 ′′ field of view at a spatial resolution of 0. ′′ 7 and with a spectral resolution of R ∼ 2000 or R ∼ 4000, depending on the target.At z ∼ 2, this configuration corresponds to a field-of-view of ≈ 170 kpc ×130 kpc and a spatial resolution of ≈ 6 kpc.
The first set of KCWI observations was conducted in 2019 September and October, with later observing runs in 2021 February and April.The on-target exposure time varied between 0.5 hr and 1.5 hr, with DLAs with ancillary CO observations given priority.We obtained line flux sensitivities of ≈ 0.1 − 5 × 10 −16 erg s −1 cm −2 at S/N = 5, depending on the exposure time and the DLA redshift (i.e. the redshifted Lyα wavelength).Note-There are two DLAs along the sightline of J2206-1958 ( †).
The KCWI data were analysed with PypeIt1 , a Python package designed for the semi-automated reduction of astronomical spectroscopic data2 .We used the built-in routines to perform bias subtraction, dark correction, and trace pattern identification.For wavelength calibration, we used FeAr lamps.The flat-fielding of the data accounted for pixel-by-pixel variations and the dome illumination pattern.The faintness of our targets allowed PypeIt to use the science frames themselves to perform sky subtraction.Bright objects, such as the QSO in each field, were detected and masked in the computation of the sky model.The astrometric solution of the data cubes was revised using the coordinates of the QSOs.

Flux calibration
Observations of standard stars (m ∼ 10) prior to or after the science exposures were used for flux calibration.We used PypeIt to compute the flux sensitivity curves from these standard stars.The first step of the process was to flux calibrate the data cubes of the standard stars.We verified that this step was performed appropriately by comparing the flux-calibrated standard data cubes with the well-characterized standard spectra.Then, the flux-calibrated standard star data cubes were used as the input sensitivity curves in the co-addition of all the data cubes (i.e., exposures) for each target.
To quantify the accuracy of our flux calibration, we searched for publicly available observations of the QSOs (m ∼ 18; Table 1) in our sample.Four of the QSOs -J2222-0946, J1305+0924, J1709+3258, and J1013+5615 -were observed in the Baryon Oscillation Spectroscopic Survey (BOSS) of the Sloan Digital Sky Survey (SDSS; Alam et al. 2015).We found that our flux-calibration is consistent with that of SDSS/BOSS within a factor of ∼ 2. To maximize the accuracy of our flux calibration, we estimated a flux-correction factor by maximizing the likelihood between the KCWI and the SDSS/BOSS spectra.An optimal flux correction factor of 1.8 was found and used to re-scale our flux-calibrated spectra.The dispersion among these measurements indicates that our flux calibration has a standard error of 12%.All errors on the line fluxes reported in this paper include this factor of 0.12.Finally, all fluxes were corrected for Milky Way extinction using the Fitzpatrick (1999) extinction curve, the dust reddening maps from Schlegel et al. (1998), and the reddening to A V conversion tabulated in Schlafly & Finkbeiner (2011). .Spatially integrated KCWI spectra for the 14 QSOs after continuum normalization.Every panel shows the spectrum (continuous black lines) and the error (dashed green lines) at the wavelength of the DLA.The red lines and shaded regions show the best solution and the error on the DLA fits that were used to determine the redshifts and H I column densities of the absorbers (Rafelski et al. 2012(Rafelski et al. , 2014)).The symbols underneath the name of the target denote the sample the DLAs belong to.Note that the DLA along the J1305+0924 sightline is proximate, i.e., the DLA is within 5000 km s −1 of the Lyα emission from the QSO.
4. METHODOLOGY 4.1.Subtraction of the sky background and the QSO continuum Figure 2 shows the fully-reduced KCWI spectra towards each target QSO centered at, and zoomed in on, the redshifted Lyα absorption line of each of the 14 DLAs.Before searching for emission lines at the DLA redshift, characterization of the QSO emission was required.We started by constructing a QSO continuum model for the data cube of every target.To do this, we computed ϕ qso (λ), a least-squares Chebyshev series fit to the QSO continuum in the brightest spaxel.Then, ϕ qso (λ) was scaled throughout the data cube to produce the initial continuum model cube where A(x, y) is the spaxel-dependent QSO continuum scaling factor.The spaxel-dependent error in the spec-tra -σ(x, y, λ) -was then used to estimate i.e., the mean signal-to-noise of the QSO continuum throughout the data cube.Thresholds in SN qso (x, y) were imposed to identify spaxels affected by the QSO continuum.Depending on the target, we found the optimal value for the threshold -determined through visual inspection -to vary between 5 and 14.
In order to estimate the background emission throughout each cube, all spaxels affected by QSO emission were masked.After masking, the median spectrum across each data cube was fitted with a Chebyshev series to obtain the background model C sky (λ).Implementation of the median instead of the mean ensured that no serendipitous background sources affected our estimate of C sky (λ).For each field, a background-subtracted data cube was then computed, which then was used to measure C qso (x, y, λ) again; this time without contributions from the sky background.Finally, the QSO continuum model was subtracted from this residual data cube.

Lyα line measurements
We searched for emission lines in the processed data cubes in 1000 km s −1 velocity slices centered at the wavelength of the DLA.To avoid the wings of the DLA, narrower velocity slices had to be used at the position of the QSO.Depending on the target, values between 500 km s −1 and 1000 km s −1 were used.Then, gaussian emission lines were fitted in these velocity slices for every spaxel.The free parameters of the fit were the line centroid, the peak flux density, and the line width.This step was repeated 500 times on 500 error-perturbed spectra for every spaxel, which yielded a line flux distribution for every spaxel around the expected redshifted Lyα wavelength.The mean and the standard deviation of each distribution were taken as the measured line flux and line flux error (F Lyα and eF Lyα ) at every spaxel.
To associate a S/N to an emission line, we quantified the line flux distribution across the data cube around the expected redshifted Lyα wavelength.To do this, we first masked all emission lines with ≥ 2σ significance, and then added back the background and the QSO emission.We then produced 200 error-perturbed data cubes for every target and carried out backgroundand QSO-subtraction for each such cube.Line fluxes were then measured across these data cubes, yielding a false positive line flux distribution for every spaxel around the redshifted Lyα wavelength (see Figure 3).The line fluxes of the error-perturbed data cubes were then rank-ordered to obtain the correspondence between line flux and percentile.This correspondence was used to assign a S/N to every Lyα line flux measurement in the original data cube.
We found that S/N Lyα = 5 effectively separates significant and spurious detections in the KCWI data cubes.For all significant detections, we recomputed the line fluxes to ensure that we account for spatially extended sources.To this end, we coadded the spectra of all the spaxels adjacent to the spaxel containing the detection.Our line flux fitting algorithm (see above) was then used to estimate the total line flux in the coadded spectrum.The Lyα line fluxes and errors reported throughout the rest of the paper correspond to the values measured in the coadded spectra.The S/N was not recalculated, i.e., the values of S/N Lyα reported throughout were measured in the spaxel showing the highest significance.
For targets with no significant detections, we characterized our line flux sensitivity within the KCWI fieldof-view.To this end, we inserted 10,000 emission lines with varying line fluxes and profile shapes in randomized locations within each data cube.The shapes of the simulated emission lines were chosen to be Gaussian, with 1σ widths between 150 km s −1 and 225 km s −1 .The line flux at which we were able to recover 95% of the lines with at least S/N Lyα = 5 was defined as our Lyα sensitivity limit (Figure 3).
Inspection of the data cubes revealed 14 sources of continuum emission.With no line emission at the expected redshifted Lyα wavelength, these sources are most likely foreground interlopers at redshifts lower than that of the DLA.Of the 14 sources, 7 are bright at the redshifted Lyα wavelength, and we therefore highlight them in white boxes in figures throughout the paper.
We also performed a search for Lyα nebulae around the QSOs of our sample.This search could not be performed for four of the 13 QSOs (J2222-0946, B1228-113, B0201+365, and J1013+5615) because their Lyα emission was redshifted out of the spectral coverage of KCWI.Out of the remaining nine QSOs, we found evidence for spatially extended Lyα emission in two cases.The emission extends over at least 70 kpc around QSO J2225+0527 and over ≈ 40 kpc around QSO J0453-1305.The two Lyα nebulae are shown in Figure 4.In passing, we note that the spatial resolution of our observations at the redshift of the QSO is typically ≈ 14 kpc FWHM, implying that the Lyα emission extends over at least two spatial beams in both cases.Similarly to Herenz et al. (2015), we find that the Lyα nebulae incidence rate (≈ 20%) is lower than what is typically obtained in dedicated searches (50 − 70%; e.g.Roche et al. 2014;Arrigoni Battaia et al. 2016, 2019;Farina et al. 2017Farina et al. , 2019;;Cai et al. 2019;O'Sullivan et al. 2020), although we are limited by small-number statistics.
The fourth and fifth Lyα detections are towards B0551-366.This DLA is part of the CO sample, featuring a CO detection.The Lyα emitters are at impact parameters of 53 kpc and 70 kpc (see Figure 5), which are much larger than the impact parameter of the CO emitter in this system (15 kpc).Remarkably, the Lyα redshift of the galaxy that is ≈ 53 kpc away from the DLA sightline is in excellent agreement with the DLA redshift (≲ 100 km s −1 ; Figure 5).
None of the 6 DLAs with significant or tentative CO detections show any evidence for Lyα emission within 50 kpc of the DLA (see Figures 5 and 6).In passing, we note that the tentative CO detection at z = 1.83 towards J1709+3258 is just outside the field-of-view of our KCWI coverage.Finally, of the 4 DLAs in the "blind" sample, we obtained a tentative (S/N Lyα ≈ 4) detection of Lyα emission at z ≈ 2.067 towards J0453-1305.The Lyα emission spectrum and image for this tentative detection are shown in the last column of Figure 5.Our detections and nondetections of Lyα emission are summarized in Table 2.
For the new Lyα detections, the Lyα emitters along the sightline towards B0551-366 have fluxes of (12±5)× 10 −17 erg s −1 cm −2 for the source at an impact parameter of ≈ 53 kpc, and of (17 ± 6) × 10 −17 erg s −1 cm −2 for the source at b ≈ 70 kpc.For the tentative Lyα detection associated with the z ≈ 2.0666 DLA towards J0453-1305, we obtain a line flux of (4.2 ± 3) × 10 −17 erg s −1 cm −2 at an impact parameter of 6±6 kpc.We obtain S/N Lyα ≈ 4 for this emitter, which was determined via simulations of the data cubes.Note that the error on the flux is dominated by the error in the flux scale.

QSO
Sample B0458-020 control 2.0396 21.65 ± 0.1 -1.12 ± 0.1 2.041 6.4 ± 3 (6.3σ) 2 ± 1 0 ± 6 J2206-1958a † control/CO(✗) 1.9200 20.65 ± 0.1 -0.60 ± 0.1 1.923 14 ± 6 (6.5σ)  as filled black circles and Lyα nondetections (and 5σ upper limits on the Lyα luminosity) as downward-pointing arrows.The gray shaded region shows the expected Lyα luminosity for a given galaxy metallicity, assuming the observed stellar mass-metallicity relation for starforming galaxies at z ≈ 2 (Erb et al. 2006)  It is clear from Figure 7[A] that most of our Lyα detections arise in the fields of DLAs with relatively low absorption metallicities.Indeed, only one of the nine DLAs with [M/H] > −0.3 shows a detection of Lyα emission, while four of the five DLAs with [M/H] ≤ −0.5 show either clear or tentative detections of Lyα emission.Further, the figure shows that many of the Lyα detections of our survey lie close to the grey shaded region, i.e. are in reasonable agreement with the expected Lyα line luminosity, despite the strong assumptions of (1) a dust-free Lyα to Hα luminosity ratio and (2) absorption metallicity equal to emission metallicity.This is seen to be the case for all DLAs with [M/H] ≲ −0.5.However, for the higher-metallicity DLAs, with [M/H] > −0.3, most of the upper limits on the Lyα line luminosity are in clear conflict with the expected line luminosity, typically lying more than an order of magnitude below the expected values.The most likely cause of this discrepancy is significant dust extinction of the Lyα line in the galaxies in the fields of the highest-metallicity DLAs at z ≈ 2 (more in Section 6).
While no clear trend is apparent in Figure 7[B], which plots the Lyα line luminosity versus DLA H I column density, we cannot yet conclude that Lyα luminosity is not correlated with N H I for DLAs at z ≈ 2. Apparent in this figure is that the coverage toward N H I > 10 21 cm −2 in current surveys is limited.While the survey by Krogager et al. (2017) included some high N H I DLAs, the associated galaxies are biased toward low impact parameters due to the slit-based nature of their search.Therefore, more data is needed to determine how N H I and Lyα luminosity are related at these redshifts.
The two panels of Figure 8 show [A] the DLA metallicity [M/H], and [B] the DLA H I column density, plotted against galaxy impact parameter for our six detections of Lyα emission at z ≈ 2.Besides these Lyα detections, we have included the 6 DLA galaxies identified in ALMA and NOEMA CO searches at z ≈ 1.8 − 2.6 (Kanekar et al. 2020;Kaur et al. 2022a) and the five Lyα detections at similar redshifts obtained via slit spectroscopy in the literature (Krogager et al. 2017).
Figure 8[A] shows that there are galaxies over a wide range of impact parameters (≈ 6 − 100 kpc) in the fields of high-metallicity ([M/H] ≳ −0.7) DLAs.This sug-gests that high-metallicity DLAs at z ≈ 2 may arise from both galaxy disks and extended gas in the environment of massive galaxies, with the circumgalactic medium (CGM) being enriched due to galactic outflows.Conversely, the DLA galaxies associated with low metallicity ([M/H] ≲ −1) DLAs are seen to have low impact parameters (b ≲ 10 kpc; although we note that there are only four galaxies in this category).Figure 8[B] plots the H I column density against impact parameter for the above sample of Lyα-detected and CO-detected galaxies.It is clear that similar H I column densities, ≈ 10 20.5 − 10 21 cm −2 are found in DLAs over a wide range of galaxy impact parameters, ≈ 5 − 100 kpc.
Finally, Figure 9 (Whitaker et al. 2014) with the observed stellar mass-metallicity relation for emission-selected galaxies at z ≈ 2 (Erb et al. 2006).The orange shaded region in this panel shows the expected Lyα luminosity as a function of DLA metallicity by combining the same star-forming main sequence relation at z ≈ 2 with the mass-metallicity relation for H I-selected galaxies at z ≈ 2 (Christensen et al. 2014).While a few of the Lyα luminosities measured in our survey lie on or close to the grey band at intermediate metallicities, it is clear that the upper limits to the Lyα luminosity for the highest-metallicity DLAs are more than an order of magnitude below this band.range of impact parameters, and that these previous observations have shown that DLAs can arise from galaxy groups (e.g., Fynbo et al. 2018).It is therefore likely that some of the galaxies at large impact parameter are companion galaxies of the galaxy that gives rise to the DLA absorption (e.g.Mackenzie et al. 2019;Lofthouse et al. 2023).This is consistent with the results from hydrodynamical simulations (e.g., Rahmati & Schaye 2014;Rhodin et al. 2019).
The different panels in Figure 9 show how the Lyα luminosity depends on [A] DLA metallicity, [B] H I column density, [C] redshift, and [D] impact parameter.We note that the DLAs with searches for Lyα emission at z ≳ 3 typically have low metallicities, [M/H] ≲ −1, while most of our KCWI searches are in the fields of high-metallicity DLAs.The luminosity of our upper limits decreases with redshift, reflecting how the spectral signal-to-noise decreases substantially for z < 2 as a result of the degradation in sensitivity of KCWI blueward of 3700 Å.
Figure 8[D] might hint that DLAs at z ≳ 3 are more often associated with galaxy groups (with ≥ 3 galaxies) than DLAs at z ≈ 2. Four out of 13 DLAs at z ≳ 3 feature at least 3 Lyα emitters in their fields (Mackenzie et al. 2019;Lofthouse et al. 2023), while only a single DLA at z ≈ 2 (out of 15) has ≥ 3 associated Lyα emitters (Nielsen et al. 2022).However, this apparent difference could be driven by variations in survey design.There is a significant difference between the field-of-views of KCWI (≈ 170 kpc × 277 kpc) and VLT-MUSE (≈ 450 kpc × 450 kpc), the instrument used by the surveys of Mackenzie et al. (2019) and Lofthouse et al. (2023).This difference becomes even greater when we consider that our effective field-of-view is ≈ 170 kpc × 130 kpc (Section 3), which can be even smaller if the QSO is not perfectly centered (Figures 5  and 6).Difference in sensitivities between the surveys could also play a role, with Mackenzie et al. (2019) and Lofthouse et al. (2023) achieving lower Lyα luminosities than our analysis (Figure 9).Thus, it is clear that searches with wider fields of view and higher sensitivities at z ≈ 2 are needed to test for possible redshift evolution in DLA environments.

DISCUSSION
Our KCWI survey has yielded new detections of Lyα emission along the DLA sightline towards B0551-366, a tentative detection from the DLA field towards J0453-1305, and recoveries of all three known Lyα detections from the control sample.Restricting to the new (i.e.non-control) searches, we have obtained definite Lyα detections along only a single DLA sightline out of 11, with nine of target DLAs having high metallicities ([M/H] > −0.3).We also do not detect Lyα emission from any of the 4 CO-emitting galaxies in the sample, although one of these sightlines (B0551-366) shows Lyα emission from two other galaxies within ±500 km s −1 of the CO and DLA redshifts.We discuss in this section the implications of these results, in conjunction with those from surveys from the literature, to gain insights into the nature of the galaxies associated with highmetallicity DLAs at z ≈ 2.
6.1.The galaxies associated with high-metallicity DLAs at z ≈ 2 are not typical Lyα emitters The left panel of Figure 7 shows that the Lyα luminosity of galaxies in the field of high-metallicity ([M/H] ≥ −0.3) DLAs at z ≈ 2 lies well below the predicted Lyα luminosity of emission-selected galaxies at similar redshifts.We emphasize that the predicted Lyα luminosity has been obtained with the assumption of a dust-free Lyα-to-Hα ratio.There are two possibilities to explain this discrepancy: (1) the galaxies associated with highmetallicity DLAs have low stellar masses, and hence intrinsic Lyα luminosities below our detection threshold, or (2) the assumption of a dust-free Lyα-to-Hα ratio breaks down, with the galaxies associated in highmetallicity DLA fields being typically massive, dusty galaxies with high production, high absorption of Lyα photons.In this section, we consider the first possibility by turning to two different methods of estimating the luminosities of low mass (M * ≈ 10 8 M ⊙ ) galaxies.As shown in Oyarzún et al. (2016Oyarzún et al. ( , 2017)), galaxies of this stellar mass have Lyα equivalent widths (50−200 Å; Charlot & Fall 1993) and escape fractions (≈ 1; Laursen et al. 2009) that are typical of dust-free interstellar media.
In the first approach, we will assume that any undetected galaxies belong to the star-formation main sequence at z ≈ 2. The slope and scatter of the main sequence at z ≈ 2 has been measured down to a stellar mass of M * ≈ 10 8 M ⊙ by Mérida et al. (2023).At this mass, the SFR distribution has an average of log(SFR/M ⊙ yr −1 ) ∼ 0 and a scatter of ∆log(SFR/M ⊙ yr −1 ) ∼ 0.35 dex.Following the approach of Section 5, this SFR can be converted into a Lyα luminosity L Lyα by assuming a dust-free Lyα-to-Hα ratio and the Hα SFR calibration for a Chabrier IMF.In the resulting L Lyα distribution, the probability that a galaxy has a Lyα luminosity lower than our upper limits is p ≈ 20% (on average) for the 8 high-metallicity DLAs with tight upper limits on the Lyα luminosity (i.e.J1305+0924, B1228-113, B0201+365, J2225+0527, B1230-101, Q1755+578, and J1013+5615).Thus, the probability that all of these DLAs feature typical Lyα   emitters with M * ≈ 10 8 M ⊙ within ≈ 50 kpc of the QSO sightline is p ≲ 10 −5 .
Alternatively, we can turn to the rest-frame Lyα equivalent width (EW Lyα ) instead of the Lyα luminosity.This quantity is defined as where F Lyα is the Lyα line flux and f λ is the restframe flux density of the galaxy in the near-UV.We note that the EW Lyα is a convenient metric because the EW Lyα distribution of high-redshift galaxies has been thoroughly quantified in the literature (e.g.Gronwall et al. 2007;Treu et al. 2012;Jiang et al. 2013).
The expected Lyα detection rate of our observations can now be estimated from the known EW Lyα distribution of galaxies at z ≈ 2. A number of authors have concluded that the EW Lyα distribution at this redshift is well fitted by an exponential profile with a scale length of EW 0 ≈ 50 Å (e.g.Nilsson et al. 2009;Guaita et al. 2010;Mawatari et al. 2012;Ciardullo et al. 2014).Knowing that the EW Lyα distribution accounts for ≈ 80% of the galaxy population at M * ≲ 10 9 M ⊙ (z ∼ 4; Oyarzún et al. 2016), integration of this probability distribution from EW Lyα = 0 up to our EW Lyα limits yields ≈ 80% of the probability that a M * ≈ 10 8 M ⊙ galaxy was not detected by our survey.This results in total probabilities of ≈ 40% for each of our 8 nondetections, yielding a probability of p ≈ 2 × 10 −4 that all of these 8 highmetallicity DLAs have low stellar mass galaxies within ≈ 50 kpc of the QSO sightline.
These results indicate that the majority of the galaxies associated with high-metallicity DLAs at z ≈ 2 are not unobscured galaxies with low stellar masses, i.e., M * ≈ 10 8 M ⊙ .In fact, characterization of the galaxies associated with high-metallicity DLAs at z ≈ 2 has yielded stellar masses exceeding 10 10 M ⊙ (e.g.Fynbo et al. 2013).

The galaxies associated with high-metallicity
DLAs at z ≈ 2 are probably massive and dusty A galaxy can also remain undetected in Lyα emission if it has a low Lyα escape fraction.This would typically arise because of Lyα photon absorption by dust grains, as indicated by several radiative transfer simulations (e.g.Verhamme et al. 2008;Laursen et al. 2009).This has been argued to be the reason for why highredshift galaxies with redder UV slopes show lower Lyα emission equivalent widths (e.g.Blanc et al. 2011;Hayes et al. 2011;Atek et al. 2014;Oyarzún et al. 2017).At the same time, the escape of Lyα photons is affected by the scattering of radiation by neutral gas, such that the structure and kinematics of the interstellar and circumgalactic media of galaxies can shape the Lyα line profile (e.g.Verhamme et al. 2006).
The dependence of the Lyα escape fraction on the neutral gas covering fraction and dust extinction has implications for the use of Lyα as a galaxy tracer.Star-forming galaxies with higher neutral gas masses and dust extinctions tend to have higher stellar masses (e.g.Reddy et al. 2006;Finlator et al. 2007;Tacconi et al. 2020), implying that Lyα emission might not be a good tracer of the high stellar mass end of the starforming galaxy population.In agreement with this picture, the equivalent width of the Lyα emission line in high-redshift galaxies has been found to anti-correlate with the stellar mass (e.g.Oyarzún et al. 2016).
Instead, massive galaxies tend to be bright in mm or sub-mm emission lines (e.g.CO or [C ii] 158µm), whose luminosity correlates with the total molecular gas mass (i.e. with the stellar mass).Thus, it is not surprising that the CO rotational lines have been particularly useful in identifying the galaxies associated with the highest-metallicity DLAs at z ≈ 2 (Kanekar et al. 2020;Kaur et al. 2022b).Kaur et al. (2022b) find that the detection rate of CO emission is strongly dependent on DLA metallicity, with a CO detection rate of ≈ 50% for [M/H]> −0.3, and a far lower CO detection rate at lower metallicities.
The suggestion that Lyα and CO emission are efficient at identifying the galaxies associated with DLAs of different metallicities is supported by Figures 8[A] and 9. Galaxies detected through Lyα emission tend to be brighter for DLAs with [M/H] ≲ −1.0.This is likely to arise due to dust obscuration effects, with highermetallicity galaxies also having high dust contents that impede the escape of Lyα photons.Conversely, the requirements of both a high molecular gas mass and a low CO-to-H 2 conversion factor imply that searches for CO emission favour the identification of the high-mass galaxies associated with high-metallicity DLAs.In line with the detection of dusty galaxies in association with high-metallicity DLAs, it is plausible that the majority of these absorbers arise from massive, dusty galaxies that are faint in Lyα emission (e.g.Fynbo et al. 2013).
In this context, it is interesting to consider why some of the highest-metallicity DLAs show no evidence of CO emission from associated galaxies (Kanekar et al. 2020;Kaur et al. 2022b).This is likely a result of the sensitivities of the current ALMA and NOEMA searches, which typically yield upper limits of ≈ 1 − 5 × 10 10 M ⊙ on the molecular gas masses of these galaxies (Kanekar et al. 2020;Kaur et al. 2022b).These upper limits are not necessarily constraining, and thus current CO nondetections do not rule out the presence of massive galaxies in the DLA fields.
Given that high-mass galaxies associated with highmetallicity DLAs may be missed in searches for both Lyα and CO emission, turning to different diagnostics emerges as a valid strategy.The best probes to identify such galaxies are likely to be Hα, [O iii]λ5007, and [C ii] 158µm emission.The Hα and [O iii]λ5007 lines are less affected by dust obscuration than Lyα, and can be used to probe galaxies in the stellar mass range M * = 10 8 − 10 10 M ⊙ (e.g.Péroux et al. 2011Péroux et al. , 2012;;Jorgenson & Wolfe 2014;Wang et al. 2015).While this line is difficult to access from the ground for galaxies at z > 2, the James Webb Space Telescope should allow the Hα-based identification of large samples of highmass DLA galaxies out to z ∼ 4. Similarly, while the [C ii] 158µm emission from galaxies at z ≈ 2 lies at very high frequencies (> 600 GHz), it should be possible to use ALMA [C ii] 158µm searches to identify such galaxies.6.3.Implications for the nature of H I-selected galaxies at high redshift Searches for the galaxies associated with DLAs at high redshift have made remarkable progress over the last few years, with more than 40 galaxies identified via Lyα, CO, or [C ii] 158µm searches (e.g.Krogager et al. 2017;Fumagalli et al. 2017;Neeleman et al. 2017Neeleman et al. , 2018Neeleman et al. , 2019;;Mackenzie et al. 2019;Kanekar et al. 2020;Kaur et al. 2022b;Lofthouse et al. 2023).While we now have a large sample of H I-selected galaxies, the wide range of DLA redshifts (z ≈ 2 − 4.5), DLA metallicities ([M/H] −2.5 − 0), and different selection techniques imply that it is not straightforward to draw conclusions about their nature.Here, we briefly summarize the current observational view of this population in order of increasing redshift.
First, the bulk of the H I-selected galaxies identified in slit-based Lyα searches at z ≈ 2 lie at low impact parameters to the QSO sightline (e.g.Fynbo et al. 2013;Krogager et al. 2016Krogager et al. , 2017;;Joshi et al. 2021).This is not surprising, given that such searches are only sensitive towards galaxies at low impact parameters (≲ 15 kpc).By design, these DLAs have high metallicities, i.e., between [M/H] ≈ −1.5 and [M/H] ≈ −0.5 (Krogager et al. 2017).The fraction of undetected galaxies and their H I metallicity distribution remains unclear.
At z ≈ 2, our KCWI Lyα survey has revealed only 2-3 new H I-selected galaxies from a search in 11 DLA fields (excluding the control sample).Nine of our DLA targets have [M/H] > −0.3, of which only one showed a detection of Lyα emission.The KCWI survey is sensitive to galaxies with impact parameters ≲ 50 kpc.
At z ≈ 2, ALMA and NOEMA CO searches have yielded six definite detections of H I-selected galaxies, all in the fields of DLAs with a metallicity of [M/H] > −0.3.(Kanekar et al. 2020;Kaur et al. 2022b).Five of the CO detections are at impact parameters of ≲ 30 kpc.The sixth system, towards Q0918+1636, has an impact parameter of ≈ 100 kpc; however, there is a galaxy at lower impact parameter (≈ 16 kpc) to the QSO sightline that was identified in the optical (Fynbo et al. 2018).While the field of Q0918+1636 was not targeted by our survey, the DLA galaxy at an impact parameter of ≈ 16 kpc has been extensively characterized by Fynbo et al. (2011).The nondetection of Lyα emission from this galaxy down to fluxes of ∼ 5 × 10 −18 erg s −1 cm −2 is indicative of dust supression, especially given the high luminosity of its [O ii] and [O iii] emission (Fynbo et al. 2011).
At z ≈ 3−3.5, Lyα spectroscopy with VLT/MUSE has yielded ≈ 28 Lyα detections of galaxies in DLA fields (Fumagalli et al. 2017;Mackenzie et al. 2019;Lofthouse et al. 2023).These DLAs have metallicities between [M/H] ≈ −2.5 and [M/H] ≈ −1.Almost all of the Lyα detections are at high impact parameters (≳ 100 kpc), and there are, in some cases, more than 5 galaxies identified in a single DLA field.Such galaxies would not have been detected either in the slit-based Lyα surveys (because of the field of view) or in the present KCWI survey (because of the sensitivity and/or field of view).
Regarding the sizes of H I gas reservoirs, evidence indicates that they are quite extended, especially in lowredshift massive galaxies.Towards z ≈ 1.3, Chowdhury et al. (2022a) have found that a spatial resolution of sion of massive galaxies, indicating that the H I sizes of these galaxies are ≳ 50 kpc at these redshifts.In the local Universe, the H I size has been found to correlate with the H I mass, with the diameter of H I disks exceeding 40 kpc in galaxies with H I masses > 10 10 M ⊙ (at an H I surface density of 1M ⊙ /pc 2 , i.e., similar to the DLA column density threshold; e.g.Broeils & Rhee 1997;Wang et al. 2016).Thus, both direct measurements of H I 21 cm emission at z ≈ 1.3 and the H I masssize relation at z ≈ 0 suggest that massive galaxies have large spatial extents in H I. Chowdhury et al. (2022b) have concluded that H I dominates the baryonic mass of the disks of galaxies, with the H I mass exceeding the stellar mass and the molecular gas mass by factors of ≈ 4 − 5. Drawing a comparison with the CO detections of Kanekar et al. (2020) and Kaur et al. (2022b), the measurements by Chowdhury et al. (2022b) would imply an H I diameter of ≳ 100 kpc at z ≈ 2, even if we assume that the H I mass is only comparable to the molecular gas mass, (≳ 5 × 10 10 M ⊙ ).With the H I-selected galaxies associated with high-metallicity ([M/H] ≳ −0.5) DLAs at z ≈ 2 typically found at relatively low impact parameters (≲ 30 kpc), it is plausible that high-metallicity DLAs at these redshifts arise from the disks of massive galaxies.
As concluded in Section 6.2, the low detection rate of these high-metallicity DLA fields in Lyα is likely due to dust obscuration effects in massive galaxies.Instead, less affected by dust obscuration (and brighter in Lyα) are expected to be galaxies of low-intermediate stellar masses (Section 6.1).Because of this, it is noteworthy how small the number of companion galaxies identified in Lyα is at z ≈ 2. Only one of the H I-selected galaxies at low impact parameters (B00551-366) has been found to have companion galaxies within ≈ 50 kpc.This is likely to be due to relatively small fields of view of the current Lyα searches at z ≈ 2.
The situation is somewhat different at z ≈ 3 − 3.5, where Lyα searches have mostly identified galaxies at very large impact parameters (≳ 100 kpc) in the fields of DLAs with mostly low metallicities (between [M/H] ≈ −2.5 and [M/H] ≈ −1; Mackenzie et al. 2019;Lofthouse et al. 2023).If we assume that the low metallicity of the DLA is indication of low dust obscuration, galaxies at low impact parameters would have only been missed if they were intrinsically under-luminous.Under this assumption, DLAs at z ≈ 3 − 3.5 with [M/H] ≲ −1 arise from either low stellar mass galaxies at low impact parameters or from massive, dusty galaxies at large impact parameters.The large number of Lyα-emitting compan-ions at distances of ≈ 100 − 200 kpc in a number of the fields is interesting, and suggests that the DLAs probed by these studies arise mostly in galaxy groups.
Finally, the impact parameters of the H I-selected galaxies identified in [C ii] 158µm searches at z ≳ 4 are ≈ 15 − 50 kpc (Neeleman et al. 2017(Neeleman et al. , 2019)).These [C ii] 158µm emitters are all associated with DLAs that have a metallicities of [M/H] ≥ −1.35, and while multiple galaxies have been identified in some fields, these are all at similar impact parameters (< 50 kpc).Together, the high H I column densities (≳ 10 21 cm −2 ) and relatively large impact parameters hint that highmetallicity, high N H I DLAs at z ≳ 4 likely arise from H I clumps in the CGM of massive galaxies.This is consistent with numerical simulations that predict greater amounts of H I in the CGM of galaxies at z ≳ 4 than at lower redshifts (e.g.Stern et al. 2021).

SUMMARY
We used the integral field spectrograph KCWI on the Keck II telescope to carry out a search for Lyα emission in the fields of 14 DLAs at z ≈ 2. Nine of the 14 targets have high metallicities ([M/H] > −0.3).Seven of the 14 fields have been searched for CO emission with ALMA or NOEMA, with 4 confirmed detections.Finally, three of the 14 DLAs are known Lyα emitters from the literature that were observed to quantify our Lyα detection capability (i.e., the control sample).
We detected Lyα emission with the expected strength from the three control-sample DLAs.For the remaining 11 targets, Lyα emission was detected from two galaxies in the field of the z ≈ 1.9622 DLA towards B0551-366 at impact parameters of ≈ 50 − 70 kpc.Also in the field of B0551-366 is a massive CO-detected galaxy at an impact parameter of ≈ 15 kpc from the QSO sightline.This indicates that the Lyα emitters do not directly give rise to the DLA absorption.We find that the low Lyα detection rate in the fields of high-metallicity DLAs is likely a result of Lyα photon absorption by dust produced in massive and dusty galaxies.
We compared the results of our Lyα searches in DLA fields at z ≈ 2 with those of CO searches at z ≈ 2, Lyα searches at z ≈ 3 − 3.5, and [C ii] 158µm searches at z ≈ 4. The impact parameters of the galaxies associated with high-metallicity DLAs at z ≈ 2 are typically ≲ 30 kpc.We argue that high-metallicity galaxies are likely to have a large H I mass, and hence a large H I spatial extent.High-metallicity DLAs at z ≈ 2 are thus likely to arise in the H I reservoirs of massive galaxies.
Finally, we found that galaxies associated with highmetallicity DLAs ([M/H] > −0.3) may remain unidentified in both CO searches (if they do not have extreme molecular gas masses) and Lyα searches (due to high dust obscuration).Searches in the Hα, [O iii]λ5007, and [C ii] 158µm lines will likely be necessary to identify this population.
Figure2.Spatially integrated KCWI spectra for the 14 QSOs after continuum normalization.Every panel shows the spectrum (continuous black lines) and the error (dashed green lines) at the wavelength of the DLA.The red lines and shaded regions show the best solution and the error on the DLA fits that were used to determine the redshifts and H I column densities of the absorbers(Rafelski et al. 2012(Rafelski et al. , 2014)).The symbols underneath the name of the target denote the sample the DLAs belong to.Note that the DLA along the J1305+0924 sightline is proximate, i.e., the DLA is within 5000 km s −1 of the Lyα emission from the QSO.

Figure 3 .
Figure 3. Example of the output from our line detection routine for the DLA along QSO J2225+0527.Plotted are all false positives (black histogram, left y-axis) and our completeness (blue filled line, right y-axis).These quantities are shown at the wavelength of the DLA absorption and after collapsing over the spatial component.The false positive histogram was obtained by running our line detection code on error perturbed spectra.The completeness curve corresponds to the fraction of planted lines that we were able to measure with at least S/NLyα = 5 as a function of line flux.The detection threshold and nondetection upper limit (completeness ∼ 95%) are shown as red dashed lines.

SBFigure 4 .
Figure 4. Two spatially extended Lyα nebulae around QSO J2225+0527 and QSO J0453-1305.Plotted is the surface brightness in Lyα after subtraction of the QSO continuum.The sizes of the point spread functions (1σ and 2σ contours) are shown in the top left corners.The spatial extension of the emission extends over at least two spatial beams, with the structure in QSO J2225+0527 extending over at least 70 kpc and the nebulae around QSO J0453-1305 apparent out to at least 40 kpc.
The symbols in the second column denote significant detections (✓), tentative detections (∼), and nondetections (✗) in the CO imaging.There are two DLAs along the sightline ofJ2206-1958 ( †).The DLA along B0551-366 has two detections in Lyα.The DLA along J0453-1305 is a tentative Lyα detection ( * ).The uncertainty on the Lyα line flux includes the error on the absolute flux calibration, and thus it should not be used to estimate the significance of the line.The signal-to-noise ratios of the individual detections are listed in parenthesis, next to the Lyα flux measurements.

Figure 5 .
Figure 5. Visualization of our line search for the 6 Lyα detections.The six columns correspond to the six Lyα emission lines.Top: spatially integrated Lyα emission after subtraction of the QSO continuum.The red dashed line shows the Lyα absorption wavelength.The Lyα emitters along B0458-020, J2206-1958, and J2222-0946 are part of our control sample, i.e., they were originally detected in searches for Lyα emitters around DLAs at low impact parameters.Two more Lyα emitters at b ≈ 53 kpc and b ≈ 70 kpc away from DLA B0551-366 were also found.This DLA also has a CO-emitter at b ≈ 15 kpc.The field showing a tentative detection -J0453-1305 -belongs in the blind sample, i.e., it has not been a target in any CO or Lyα searches.Bottom: integrated flux of the spectrum in the KCWI cubes centered at the redshifted Lyα absorption and within a window of width equal to the emission line.The symbols show the position of the QSO (red cross), Lyα emission (circles), CO-emitter (yellow star), and interlopers (white squares).

Figure 6 .
Figure 6.KCWI cubes for all the Lyα nondetections.Shown is the integrated flux of the spectrum in the KCWI cubes centered at the redshifted Lyα absorption and within a window of width 350 km s −1 .The position of the QSOs are shown with a red cross and the location of the CO-emitters is shown with yellow stars.Some fields show significant interloper emission that is spatially offset from the QSO (white squares).

Figure 7 .
Figure 7.The dependence of the Lyα luminosity of the galaxies in the fields of DLAs at z ≈ 2 on [A] the DLA metallicity and [B] the DLA H I column density.Detections of Lyα emission with ≥ 5σ significance are plotted as black circles, tentative (4−5σ) detections as open circles, and Lyα nondetections (i.e., 5σ upper limits to the Lyα line luminosity) as downward-pointing arrows.The gray shaded region in panel [A] indicates the expected Lyα luminosity as a function of galaxy metallicity, which was obtained by combining the star-forming main sequence relation at z ≈ 2(Whitaker et al. 2014) with the observed stellar mass-metallicity relation for emission-selected galaxies at z ≈ 2(Erb et al. 2006).The orange shaded region in this panel shows the expected Lyα luminosity as a function of DLA metallicity by combining the same star-forming main sequence relation at z ≈ 2 with the mass-metallicity relation for H I-selected galaxies at z ≈ 2(Christensen et al. 2014).While a few of the Lyα luminosities measured in our survey lie on or close to the grey band at intermediate metallicities, it is clear that the upper limits to the Lyα luminosity for the highest-metallicity DLAs are more than an order of magnitude below this band.

Figure 8 .
Figure 8.[A] DLA metallicity and [B] DLA H I column density plotted against galaxy impact parameter for DLA galaxies at z ≈ 2. As in Figure7, plotted are detections (black circles) and tentative detections (open circles).Also included are the slit spectroscopy measurements fromKrogager et al. (2017) and the CO emission measurements fromKanekar et al. (2020) andKaur et al. (2022a,b).

Figure 9 .
Figure 9.The results of our survey in the context of searches for Lyα emission from DLA fields at z ≳ 2 in the literature.The panels show the Lyα luminosity as a function of [A] DLA metallicity, [B] H I column density, [C] redshift, and [D] impact parameter.The results of our survey are shown as filled (detections) and open (tentative detections) circles, with upper limits to the Lyα luminosity indicated by the downward-pointing arrows.Plotted from the literature are the measurements of Krogager et al. (2017), Fumagalli et al. (2017), Mackenzie et al. (2019), Joshi et al. (2021), Nielsen et al. (2022), and Lofthouse et al. (2023).The corresponding symbols are indicated in [D].
The sample.Left: Distribution of the DLAs in metallicity-NH I space.Control DLA galaxies (i.e., known Lyα emitters) are plotted as large gray circles.DLAs previously targeted for CO observations are denoted with the small blue markers.Nondetections in CO are plotted as circles, potential CO detections as squares, and CO detections as triangles.DLAs without mm-observations are plotted as red stars.Right: Histograms of the metallicities and H I column densities of our 14 DLAs.The full sample is plotted in black, control DLA galaxies in gray, DLAs with CO observations in blue, and DLAs without known prior Lyα or CO observations in red.