Pair Lines of Sight Observations of Multiphase Gas Bearing O VI in a Galaxy Environment

Using $HST$/COS observations of the twin quasar lines of sight Q$0107-025$A $\&$ Q$0107-025$B, we report on the physical properties, chemical abundances and transverse sizes of gas in a multiple galaxy environment at $z = 0.399$ across a transverse separation of $520$ kpc. The absorber towards Q$0107-025$B has $\log N(H I)/cm^{-2} \approx 16.8$ (partial Lyman limit) while the absorber towards the other sightline has $N(H I) \approx 2$ dex lower. The O VI along both sightlines have comparable column densities and broad $b$-values, whereas the low ionization lines are considerably narrower. The low ionization gas is inconsistent with the O VI when modelled assuming photoionization in a single phase. Along both the lines-of-sight, O VI and coinciding broad H I are best explained through collisional ionization in a cooling plasma with solar metallicity. Ionization models infer $1/10$-th solar metallicity for the pLLS and solar metallicity for the lower column density absorber along the other sightline. Within $\pm~250~km~s^{-1}$ and $2$ Mpc of projected distance from the sightlines 12 galaxies are identified, of which 3 are within $300$ kpc. One of them is a dwarf galaxy while the other two are intermediate mass systems at impact parameters of $\rho \sim (1-4)R_{vir}$. The O VI along both lines-of-sight could be either tracing narrow transition temperature zones at the interface of low ionization gas and the hot halo of nearest galaxy, or a more spread-out warm gas bound to the circumgalactic halo/intragroup medium. This latter scenario leads to a warm gas mass limit of $M \gtrsim 4.5 \times 10^{9}$ M$_\odot$.


INTRODUCTION
Our understanding of the distribution and physical properties of diffuse baryons in the universe has primarily come from quasar absorption line studies and deep galaxy redshift surveys. The technique remains one of the most sensitive ways to gain insights on gas in a wide variety of environments, from the dense interstellar regions in galaxies to the diffuse filamentary and sheet like structures that form the intergalactic cosmic web (Bond et al. 1996). However, such studies are routinely based on single line of sight observations, with the physical description of the absorber drawn solely from the one dimensional information along the line of sight. This can be circumvented partially by collecting a statistically large sample of similar absorption systems along several random lines of sight (e.g., Rudie et al. 2012;Tumlinson et al. 2013). But even there, the spatial information on the different gas phases, their volume filling fraction within the virial radius, and the baryonic mass they encompass are all contingent on some assumed geometry based on theoretical considerations. ⋆ E-mail: anshulpratyush7@gmail.com An approach to overcome this limitation is to observe multiple closely separated lines of sight to background quasars through the same foreground absorber (e.g., Bechtold et al. 1994;Dinshaw et al. 1995Dinshaw et al. , 1997Lopez et al. 1999Lopez et al. , 2000Rauch et al. 2001;Petry et al. 2006;Crighton et al. 2010;Rudie et al. 2019). Spatially resolved spectroscopy of gravitationally lensed quasar images or binary quasars are ideal for such close-by transverse lines of sight studies. By resolving the transverse sizes, density, temperature and metallicities at kiloparsec and smaller scales, these observations have resulted in direct insights into the geometrical distribution of gas of various ionization states in a wide variety of absorbing environments (Crotts & Fang 1998;Rauch et al. 1999;Lopez et al. 2005;Misawa et al. 2013;Muzahid 2014;Lehner et al. 2020).
For example, comparisons of H I absorption profiles across pairs of lines of sight, and also cross-correlation measures of transmitted fluxes between multiple closely separated lines of sight have shown Lyα absorbers of the IGM to be coherent gas structures with sizes ranging from 200 kpc to 1 Mpc (e.g., Petitjean et al. 1998;Aracil et al. 2002;Becker et al. 2004). Such estimates have served as key constraints in developing a three dimensional structure of the cosmic web (Cappetta et al. 2010) and in its compar- Table 1. Details of the triplet quasars taken from Crighton et al. (2010) and Bourda et al. (2010 ison with theoretical expectations. Similarly, the structural properties of low and high ionization gas phases in metal absorbers have also been investigated through transverse lines of sight studies. Ionization studies of Mg II absorption systems had long suggested that they constitute a two-phase medium with the Mg II tracing dense (n H > 10 −3 cm −3 ) clouds with sizes of a few hundred parsec (assuming spherical geometry), and the C IV that is often coincident in redshift with the Mg II, tracing a more diffuse (n H ∼ 10 −4 cm −3 ) higher ionization envelope surrounding the low ionization Mg II gas (Bergeron & Stasińska 1986;Steidel 1990;Churchill et al. 1999Churchill et al. , 2000. Through spectroscopic observations of gravitationally lensed quasar images, Rauch et al. (1999Rauch et al. ( , 2001Rauch et al. ( , 2002 and Kobayashi et al. (2002) were able to corroborate these claims by showing that the Mg II absorption remains coherent over transverse scales of 200 − 400 pc or smaller whereas the associated C IV can extend over scales of 10 kpc, but with variations at sub-kiloparsec scales (Rubin et al. 2018). The compact nature of low ionization gas thus established also explains why the redshift number densities of Mg II are lower compared to C IV systems in spectral line surveys of comparable sensitivity.
Similar studies involving O VI systems have helped to constrain the distribution of higher ionization warm/hot gas in the CGM of galaxies. The O VI absorption is found to be uniform over larger transverse scales than C IV suggesting that the highly ionized halos of galaxies are more extended than the photoionized C IV reveals Lopez et al. 2007). By following O VI absorption along and across closely separated lines of sight, Muzahid (2014) was able to estimate a size of R ∼ 330 kpc for the ionized CGM surrounding a 1.2L * galaxy, with a baryonic mass of M ∼ 1.2 × 10 11 M⊙. In the absence of information on the transverse sizes, it is difficult to arrive at such estimations without assuming a morphology for the absorbing medium. More recently, using 43 background QSO lines of sight, Lehner et al. (2020) were able to derive detailed information on the distribution of multiphase gas in the extended halo of M31 from close to the galactic disk (R = 25 kpc) to nearly twice its virial radius (R = 570 kpc). The absorption along multiple sightlines showed a nearly 100% covering fraction for O VI in comparison to all other ionic species, out to ∼ 1.9Rvir, revealing the wide extend of the diffuse warm-hot CGM around M31 (see also Lehner et al. 2015).
In this work, we present analysis on intervening O VI absorption detected along the lines of sight to the quasars Q0107 − 025A, and Q0107 − 025B, separated by 1.29 ′ in the plane of the sky. The HST /COS spectra of these twin lines of sight show O VI absorbers at z = 0.227, and z = 0.399. The former absorber was analyzed by Muzahid (2014) who found identical ionization and chemical properties for the O VI bearing high ionization gas, indicating that the lines of sight are probing a uniform large-scale structure, corresponding to the highly ionized CGM of a luminous galaxy. Here we present a similar analysis for the z = 0.399 O VI absorber with information on the distribution of galaxies in the extended environment surrounding the absorber. The paper is organized as follows: the details on the COS spectra, and galaxy MOS data are described in Section 2 and Section 3, respectively. The spectral line analysis and ionization models are dealt with in Section 4. The association of the absorbers with galaxies, and the summary discussion are given in Section 5 and 6, respectively. Throughout the paper, we adopt the cosmology with H0 = 69.6 km s −1 M pc −1 , Ωm = 0.286 and ΩΛ = 0.714 from Bennett et al. (2014). All the logarithmic values mentioned are in base-10. The distances mentioned in this paper are physical unless stated otherwise.

COS OBSERVATIONS OF THE QUASAR PAIR
The quasars Q0107 − 025A (zem = 0.960) & Q0107 − 025B (zem = 0.956) (hereafter sightlines A & B) were discovered as triplet of quasars by Surdej et al. (1983Surdej et al. ( , 1986. Their coordinates, spectroscopic emission redshifts, and apparent brightness in the Vband are given in Table 1. The sight lines have a projected angular separation of 1.29 ′ in the plane of the sky. The HST /COS G130M and G160M grating observations of these targets were carried out under the Program ID: 11585 (PI: Neil Crighton), with a total integration time of ≈ 11.6 ks on each quasar. The final co-added spectra of A & B sight lines were downloaded from the HST Spectroscopic Legacy Archive . The spectra span the far-UV wavelength range from 1150 − 1775 Å with a spectral resolution of ∼ 17 − 20 km s −1 per resolution element of 0.06 Å. Continuum normalization was done by fitting lower order polynomials to the line free regions. The two quasars are part of a triplet system that includes the quasar Q0107 − 0232 at zem = 0.726 with transverse lines of sight separations of 2.94 ′ and 1.94 ′ from A & B respectively. However, for this quasar only G160M grating observations are available, which covers a few of the significant metal lines like C II, O VI, N V, in addition to H I. These lines suffer from different levels of contamination primarily from absorption associated with a sub-DLA at a higher redshift along the same line of sight (Crighton et al. 2013), rendering line measurements difficult. The third sightline is hence excluded from our main analysis, with details deferred to the Appendix A.

SPECTRAL ANALYSIS & IONIZATION MODELLING
The angular separation of ∆θ = 1.29 ′ between the sightlines A & B corresponds to a transverse separation of 520 kpc at the absorber redshift of z = 0.399. The redshift is defined based on the pixel showing maximum optical depth in the narrow C III 977 Å profile along sightline A. The line measurements were carried out using both the integrated apparent optical depth (AOD) method of Savage & Sembach (1991) and by modelling the absorption lines using Voigt profiles. The AOD method relies on converting the flux profiles of unsaturated lines into apparent column densities which are then integrated over the velocity spread of the absorption. For saturated lines, this approach offers a lower limit on the column density. Voigt profile fits were performed on the detected lines using the VPFIT routine (ver.10.4, Carswell & Webb 2014) 1 after convolving Figure 1. Velocity plots of the key transitions towards sightline A (black) and sightline B (green). The Y-axis is continuum normalized flux, and the X-axis is velocity along the line-of-sight in the rest frame of the absorber. Zero velocity corresponds to z = 0.39949. The sightline B spectrum is arbitrarily shifted along the vertical axis for clarity. Observed wavelengths of the different transitions are indicated in the different panels. Voigt profile fits to the detected lines along the sightline A is shown in Fig. 2 and along the sighline B is shown in Fig. 5 and Fig. 6. Note. The columns show, from left to right: name of the absorption line with their rest wavelength mentioned, equivalent width of the absorption line profile with 1σ error, column density of the fitted Voigt profile with 1σ error, Doppler b-parameter of the fitted Voigt profile with 1σ error and the centroid or the velocity range of the profile. Ly β is contaminated over 25 < v < 90 km s −1 . The velocity range used for estimating the equivalent width and apparent column density are truncated before that.
the model profiles with the appropriate line spread functions 2 . For undetected lines, column density upper limits are derived by converting the 3σ equivalent width limit into column density assuming the linear relationship of the Curve of Growth. Photoionization equilibrium (PIE) models were generated using Cloudy (ver C17.01, Ferland et al. 2017), assuming the absorption to be from uniform density isothermal clouds of plane-parallel geometry. We adopt the solar relative elemental abundance pattern of Asplund et al. (2009). The photoionization in the cloud is assumed to be regulated by the extragalactic background radiation (EBR) for the absorber redshifts as modelled by Khaire & Srianand (2019) which incorporates the most recent measurements of the quasar luminosity function (Croom et al. 2009;Palanque-Delabrouille et al. 2013) and star formation rate densities (Khaire & Srianand 2015). To include ionizations resulting from the collisions of energetic free electrons with atoms and ions, we use the equilibrium and nonequilibrium collisional ionization (CIE, and NECI) models given by Gnat & Sternberg (2007). The modelling results for the absorber at z = 0.399 along sightlines A & B are summarized in Table 5.
In Fig. 1 we compare certain key absorption lines along both sightlines centered on z = 0.399, the adopted rest-frame of the ab-2 http://www.stsci.edu/hst/instrumentation/cos/performance/spectralresolution sorber complex. From the relative strengths of the lines, it is evident that sightline B is intercepting higher column density of neutral gas compared to sightline A, with significant differences in strength of low ionization (C II, O II, C III, O III, N III, Si III) absorption between the sightlines. However, the O VI doublets have comparable strengths along both the lines-of-sight, implying that the O VI is either tracing a gas phase that is more widely distributed than the patchier low ionization phase, or is produced in similar zones of ionization present along both sightlines. We now discuss the individual absorption systems and their ionization models in greater detail.
3.1 System towards Q0107 − 025A at z abs = 0.39949 The absorption at z = 0.399 towards sightline A is seen in Lyα, Lyβ, Lyγ and C III 977 at ≥ 3σ significance, in addition to the O VI 1031, 1037. The absorption has two well resolved components at v = 0 km s −1 (central) and v = −70 km s −1 (offset). The system plot with detected lines is shown in Fig. 2 and the corresponding line measurements are listed in Table 2. We determine the physical and chemical conditions in the two components separately by modelling them individually. The column density of H I in the central component is constrained by simultaneous Voigt profile fit to the Lyα, Lyβ and Lyγ lines. The Lyα is saturated at the line core, whereas the Lyβ and Lyγ are unsaturated. The broad O VI profile with b(O VI) > b(H I) > b(C III) clearly suggests that the line of sight is probing a multi-phased medium, with bulk of the H I possibly associated with a lower ionization gas phase traced by C III. The different b-values of H I and C III solve for a temperature of T = 2.4 +0.6 −0.6 × 10 4 K for this phase. The inferred temperature is consistent with photoionized intergalactic/circumgalactic gas. The PIE model for this phase is shown in Fig. 3(left panel) where the column density for C IV from Crighton et al. (2010), measured from HST /FOS G190H spectra, is also included. The creation and ionization energies of C IV (48 eV and 64 eV) are less than half that of O VI (113 eV and 138 eV). Multiple lines-of-sight observations often find different transverse scales for the absorption from C IV and O VI suggesting that the two ions could be tracing differently distributed gas phases (Lopez et al. 1999(Lopez et al. , 2007Lehner et al. 2020). Substantial fraction of the observed C IV is likely to be from the same phase as C III. We, therefore, associate the C IV with the C III phase rather than the O VI phase. The observed column density ratio of C III to C IV is recovered by the PIE models at n H = 2.0 × 10 −4 cm −3 . At this density, the observed N (C III) and N (C IV) require the carbon abundance to be [C/H] = 0. Such a PIE model predicts a total hydrogen column density of log [N (H )/cm −2 ] ≈ 17.6, a gas pressure of p/K ≈ 4.8 cm −3 K, and a path length along the line of sight of L ≈ 0.7 kpc. The metallicity estimated carries an uncertainty of ≈ 0.2 dex from the cumulative uncertainties in the H I and metal lines. The intrinsic shape of the QSO SED contributing to the ionizing background can introduce an additional uncertainty. Different EBR models generated using a range of observationally consistent QSO SEDs in KS19 yield metallicities that differ by ≈ 0.2 dex (Acharya & Khaire, in preparation). For solar [C/O], the O VI predicted from this phase is a factor of 50 less, thus requiring a separate phase to explain its origin.
Explaining the origin of O VI in the central component through an additional phase would require information on the associated H I. The b(O VI) sets the temperature in such a phase to T ≤ 1.4 × 10 6 K. At the upper limits of these temperatures the neutral hydrogen fraction f H I ∼ N (H I)/N (H ) ∼ 1.5 × 10 −7 (Gnat & Sternberg 2007) is such that the associated H I will be shallow, and thermally broadened. The kinematic superposition of such −0.09 ). The temperature predicted from the line widths spans the range where ionization fractions of C III and O VI peak under collisional ionization. Higher metallicities would lead to enhanced cooling of the collisionally ionized gas by means of metal line emissions (Boehringer & Hensler 1989). This leads to a rapid decline in the temperature of the plasma with electron-ion recombination reactions lagging behind. The delayed recombination would result in the collisional ionization fractions departing from their equilibrium value, particularly for T < 5 × 10 6 K. We therefore explore both the equilibrium and non-equilibrium collisional ionization models of Gnat & Sternberg (2007) to explain the observed column densities.
As shown in Fig. 3 (middle panel), the observed N (C III) and N (O VI) can be simultaneously recovered from such a NECI model for [C/H] = [O/H] = 0.0 and T ∼ 1.6 × 10 5 K. The metallicity is identical to that estimated for the photoionized, low ionization gas in the central component. A more realistic collisional model for the offset component should also include the additional ionization brought about by the extragalactic background radiation. Implementing photoionization and non-equilibrium collisional ionization simultaneously is beyond the scope of our current work. Photoionization from the extragalactic background radiation can play a key role in reducing the cooling of the metal enriched gas (Wiersma et al. 2009;Tepper-García et al. 2011).
We instead resort to hybrid models using Cloudy by fixing the temperature of the plasma to T ∼ 1.6 × 10 5 K at which collisions of electrons with atoms and ions induce ionization, while simultaneously allowing for photoionization by the extragalactic background radiation. In Fig. 3(right panel) we show such a hybrid model. There is no single density or temperature at which C III and O VI are simultaneously explained. Instead, the observed column densities of C III and O VI are recovered to within their 1σ values for Fig. 3), we get the mean value of n H ∼ 0.6 × 10 −4 cm −3 from the hybrid model. The model yields a total hydrogen column density of log [N (H )/cm −2 ] ≈ 18.5, a gas pressure of p/K ≈ 5.4 cm −3 K, and a path length along the line of sight of L ≈ 19.5 kpc for this medium that is ionized by photons and collisions with energetic electrons. Such a model also predicts log[N (C IV)/cm −2 ] = 13.48, indicating that there could be some contribution to the C IV absorption seen in the low resolution FOS data from the offset component also. The ionization models are simplistic. Instead of the absorbing medium having an uniform temperature throughout, the line of sight could very well be probing slightly different temperature -density regions across narrow physical scales that are kinematically undistinguished at the resolution and sensitivity of the data.
To summarize, for the absorption seen in along sightline A, the low and intermediate ionization metal lines and bulk of the H I in the central component is tracing photoionized gas of solar metallicity and T ∼ 10 4 K. Such a phase significantly under-produces the O VI that is coincident in velocity. In the kinematically offset component, the ionization model independent gas temperature given by the line widths of H I and O VI favor collisional ionization. It is plausible that the O VI in the offset component is part of the same warm (T ∼ 10 5 K) collisionally ionized structure.
3.2 System towards Q0107 − 025B at z abs = 0.39916 The absorber at z abs =0.399 along the Q0107 − 025B sightline is partially optically thick in H I. The optical depth estimated from the partial Lyman break is τ912 = 0.48 +0.11 −0.13 corresponding to log [N (H I)/cm −2 ] = 16.88 +0.09 −0.14 , similar to that obtained from simultaneously fitting the Lyman series lines from H I 1215 − 915.8 Å. The uncertainty in optical depth was derived by varying Adopting the same metallicity as the central photoionized gas, the curve encompass the column density predictions for C III and O VI by adopting mean temperature of T = 1 × 10 5 K for the temperature range of T = (0.5 − 1.5) × 10 5 K, for which the models are consistent with the data. In middle and right panels, predicted column density of C IV is also shown.   the continuum slightly within the limits of the uncertainty in the flux values. The partial Lyman limit break is shown in Fig. 4. Applying a single component Voigt profile simultaneously to the Lyman transitions results in a fit that does not explain the broad absorption wings in Lyα and Lyβ (see Fig. 5). The core component is primarily governed by the higher orders of the Lyman series which have narrow profiles, are less saturated, and therefore better constrained. The Lyα, and to a lesser extent Lyβ, suggest the presence of an additional component. The significance of this is explained and discussed later in a subsequent paragraph in this section. Concurrent with the H I absorption are lines from C II, O II, C III, N III, O III, Si III and O VI that are detected at > 3σ (see Fig. 5 and Fig. 6). Prominent non-detections at this redshift include N II, Si II, N V and S VI. The velocity profiles of all prominent lines are shown in Fig. 5 and Fig. 6 and the line measurements in  −0.28 . The uncertainties refer to the range of feasible column densities these ions can take when having b-parameter between aforementioned limits. Also, the uncertainties are higher compared to the values obtained from a free profile fit to these saturated lines.
We also attempt a two component model to C We find that the O VI along this sightline is consistent with both photoionization (PIE) in highly diffuse gas as well non-equilibrium collisional ionization (NECI). As the PIE models shown in Fig. 7  (second panel)  is explained by the PIE models at n H = 3.2 × 10 −5 cm −3 , with an equilibrium temperature of T = 4 × 10 4 K. The photoionized O VI is thus tracing gas that is more diffuse (two orders of magnitude lower in density) than the gas phase responsible for the low and intermediate ions, and the pLLS. Such a photoionized O VI phase  Table 3. would contribute little C II, C III, or O III (see middle left panel Fig. 7). The PIE model for this high ionization phase corresponds to a total hydrogen column density N (H ) ≈ 2.18 × 10 19 cm −2 , a gas pressure of p/K ≈ 1.28 cm −3 K −1 , and a line of sight thickness of L ≈ 221 kpc.
Collisional ionization in a cooling plasma is also a plausible mechanism for the O VI to be produced, even at T 10 5 K implied by the line widths. The column density prediction from the NECI models of Gnat & Sternberg (2007) for gas at solar metallicity is shown in Fig. 7 (third panel). Such a model shows that for [O/H] = 0, the observed N (O VI) is explained at T = 8.3 × 10 4 K (log[T /K] = 4.92). The hybrid models involving both photoionization and collisional ionization were generated for T = 0.1 × 10 5 K and T = 1.5 × 10 5 K set by the 1σ range for the temperature from the two and three component H I fitting models of Table 3 (fourth panel of Fig. 7). This is a conservative range on the temperature that the BLA -O VI bearing gas can have. Within these limits, the hybrid models are consistent with the observed N (O VI) for densities in the range of n H = (0.1 − 2.5) × 10 −3 cm −3 . In these models, a metallicity of solar, similar to the metallicity determined for the BLA -O VI gas along sightline A, is assumed. The [O/H] in the O VI phase along sightline B cannot be constrained to a unique value due to the large temperature uncertainty from the inadequately constrained BLA b-parameter. A hybrid model with the 1σ lower limit on temperature of T = 10 4 K puts a conservative lower limit of [O/H] −1.5 while reproducing the observed N (O VI) at n H = 10 −5.2 cm −3 at which the ionization fraction of O VI peaks. In Table 5, we summarize the ionization modelling results, where we have listed the NECI and hybrid model parameters, with uncertainties, for the two possible temperatures predicted for the BLA-O VI phase from the two component and three component profile fits to the H I lines.

THE GALAXY ENVIRONMENT & ORIGINS OF ABSORPTION
We extracted information on galaxies in the extended environment near to the absorbers from the Tejos et al. (2014) catalogue, which is a composite multi-object spectroscopic survey of galaxies using VLT/VIMOS, Keck/DEIMOS and Gemini/GMOS in the foreground fields of quasars (see Table 4). The details of those observations are summarized in Sec 3 and Table 4 of Tejos et al. (2014). The spectrum from the VIMOS (R ≡ λ/∆λ ≈ 200) and DEIMOS (R ≡ λ/∆λ ≈ 5000) has a wavelength coverage of 5500 − 9500 Å  Table 3. The part of the spectra which is contaminated is shown in purple.
and 6400 − 9100 Å, respectively and for GMOS (R ≡ λ/∆λ ≈ 640), R400 grating centred on a wavelength of 7000 Å is used. We supplement these with spectroscopic data from VLT/MUSE, GMOS, and VIMOS for G1, G2, and G3 respectively (see Table 4) (Beckett et al. , in preparation), which are galaxies closest to the absorber in impact parameter. The MUSE observation of G1 is in the public archive (Program ID : 094.A-0131, PI: Joop Schaye). The MUSE has the wavelength range of 4800 − 9300 Å with resolving power of 1740 − 3590. The projected separations to all these galaxies along with their luminosities, virial radii, stellar and halo masses, and star formation rates are listed in Table 4. A brief overview of how these parameters are estimated is given in tablenotes of Table 4. An HST /ACS image of the field is shown in Fig. 8, that includes these three nearest galaxies. The medium resolution spectra of these galaxies are shown in Fig. 11.
Within a projected separation of ρ < 2 Mpc and a velocity offset of |∆v| < 250 km s −1 between the galaxy's systematic redshift and z = 0.399, 12 galaxies are identified with stellar masses in the range of M * ∼ 10 8.7 − 10 11.5 M⊙. The radial velocity and projected separation cut-offs correspond to the values that are typical of virialized galaxy concentrations such as groups and poor clusters (e.g., Bahcall 1999). Four more galaxies are identified within a wider projected separation of 5 Mpc. The distribution of these galaxies towards both sightlines are shown in Fig. 9 and Fig. 10.
Three of these galaxies (G1, G2, and G3) are within 300 kpc of both sightline, and the remaining at 1 Mpc. The presence of several galaxies within narrow projected positions and radial velocities indicates that the lines-of-sight are probing different regions of a galaxy over-density such as a group or a cluster, with the absorption possibly arising in the CGM or the intragroup/cluster medium, consistent with what has been found for optically thick gas Manuwal et al. 2019;Pradeep et al. 2019;Chen et al. 2020).
3.6, 1.7, 0.8 with sightline B. Conventionally, gas within Rvir is considered as belonging to the halo and those at larger impact parameters have a higher probability of being unbound to that galaxy, and likely related to the surrounding large-scale structures. The virial radius however provides only a characteristic value for the halo size. Environmental influences on the distribution of gas are known to extend beyond Rvir, especially when the galaxies are part of dense groups or clusters (e.g., Moore et al. 1999;McCarthy et al. 2008;Bahé et al. 2013). Given that the galaxies close to the lines of sight are low-and intermediate-mass systems, the absorbers could be tracing interstellar gas displaced through stellar/AGN feedback or tidal interactions even when the impact parameters are not strictly within the virial halo. We now proceed to consider these possibilities.
Due to impact parameters similar to virial radii of G2 and G3 with respect to sightline A and B, the sightlines could potentially be probing gas that has originated in either or both of these galaxies. The CGM of even low-luminosity galaxies < 0.1L * are known to possess substantial mass in the form of relatively cool gas (T ∼ 10 4 − 10 5 K) in the range of M cool ∼ 10 7 − 10 8 M⊙ (Stocke et al. 2013). The circumgalactic gas is a heterogeneous population comprising of gas accreted from the IGM, outflows fueled by starformation or nuclear activity, and interstellar gas of satellite galaxies displaced in tidal interactions (Tumlinson et al. 2017). The metallicity of the gas is a means to distinguish these different processes, with high-metallicity absorbers originating in metal-rich outflows (e.g., Tripp et al. 2011;Muzahid et al. 2015), and low-metallicity absorbers tracing gas getting accreted from the intergalactic medium (e.g., Ribaudo et al. 2011;Bouché et al. 2013). Another discriminating factor is the azimuthal angle, defined as the angle between the QSO line-of-sight and the projected major axis of the galaxy hosting the absorber. Galaxy-quasar pair studies find biconical outflows preferentially directed along the galaxy minor axis with significantly strong warm absorption appearing along extensions closer to the minor axis (Bordoloi et al. 2011;Schroetter et al. 2019). In contrast, accretion of cold streams of gas happens predominantly along the galaxy major axis (Kacprzak et al. 2012;Nielsen et al. 2015;Zabl et al. 2020;Ho & Martin 2020).
Relative to G2, sightlines A and B are aligned closer to the galaxy's projected minor axis at opposite sides with azimuthal angles of Φ ∼ 86 • and Φ ∼ 72 • respectively (see Fig. 8). In comparison, the sightlines are at azimuthal orientations of Φ ∼ 68 • and Φ ∼ 3 • (aligned with projected major axis) with respect to G3. Thus, from the point of view of tracing metal-rich outflows, galaxy G2 is favorably oriented relative to the sightlines. Moreover, G2 has a higher current star-formation rate marked by the prominent emission lines in its spectrum whereas G3 is absorption line dominated (see Fig. 11). For characteristic outflow velocities of 500 km s −1 , the time-scale for transporting metals to distances of ∼ 200 kpc (average impact parameter of either sightline to galaxy G2) is of the order of 600 Myr (e.g., Heckman et al. 2001;Weiner et al. 2009), which is an order of magnitude longer than the timescale of active outflows dictated by short-lived massive stars. The twin lines Figure 8. An HST /ACS image encompassing sightline A, sightline B and three of the nearest galaxies coincident in redshift with the absorbers. The diagonal lines are artifacts of the instrument. The grids (shown in blue) are lines of constant RA, and Dec. The closest galaxies to either sight lines are G1 (i = 35.37 ± 35.81 • , α = −11.13 ± 90.00 • ), G2 (i = 72.5 ± 3.3 • , α = −46.60 ± 11.87 • ) and G3 (i = 81.84 ± 1.07 • , α = −8.88 ± 0.79 • ). Quasars A and B are marked with green circles while G1, G2 and G3 are marked with red circles. The third sightline with (α, δ) = (17.5583 • , −2.2828 • ), discussed in Appendix A is outside this ACS field. The inset image contains the zoomed images of these galaxies. The axes show the Right Ascension (in degrees) and Declination (in degrees). of sight are likely to be therefore probing gas enriched by biconical flows from past rather than on-going star formation in G2. The present SFR could be the culmination of the galaxy's past peak in star-formation.
The partial Lyman limit column of H I along sightline B is consistent with enriched gas from G2 moving into the circumgalactic zone of galaxy G3 (ρ ∼ 0.8Rvir relative to G3), whereas the low column density gas along sightline A can be tracing either intragroup gas or the outskirts of the gaseous halo of G2 itself (ρ ∼ 1.2Rvir). The metals dispersed from G2 can mix with the ambient neutral gas through diffusion or turbulence. They can also remain confined to patchy zones that are either transient (Schaye et al. 2007) or pressure bound within the hot CGM or the group environment. The characteristic temperature and density of T ∼ 1.5 × 10 6 K and n H ∼ 5×10 −5 cm −3 for the hot galactic coronae (assuming Milky Way halo mass) (Faerman et al. 2017) are adequate to keep the low ionization cloud with T ∼ 1.6×10 4 K and n H ∼ 2.5×10 −3 cm −3 in pressure equilibrium. This view is also consistent with the O VI tracing transition temperature gas at the interface of the pLLS cloud and the galaxy's hot corona.
Since the abundance levels are inferred based on column densities integrated along a sightline, a larger column of metal-poor intergalactic gas along sightline B would naturally lead to an inferred metal concentration that is diluted in comparison to the lower H I column density probed by sightline A. The probability of finding such a larger column of H I gas along sightline B is higher as it penetrates close to galaxy G3 (ρ ∼ 0.8Rvir), consistent with the anticorrelation trend seen in absorber-galaxy surveys between the dis- tribution of H I column density and impact parameter to the nearest galaxy (Lanzetta & Bowen 1990;Steidel et al. 1994;Kacprzak et al. 2008;Chen & Tinker 2008;Chen et al. 2010;Bordoloi et al. 2011).
It has also been found that in dense environments, cool gas has a wider spread than around isolated galaxies (e.g., Bordoloi et al. 2011). Nielsen et al. (2018 found significantly larger column densities of gas in pixels offset from galaxy systemic redshifts in group environments compared to Mg II absorption associated with field galaxies. They also found cool gas in groups having kinematics similar to that of outflows along the minor axis of galaxies. Together, it implies that group environments have substantial covering frac-tion of low-ionization gas mixed with metals mainly contributed by outflows and tidal interactions with member galaxies. The higher covering fraction increases the odds of multiple lines of sight intercepting such gas, compared to their smaller extensions around isolated galaxies (see Dutta et al. 2020). An alternative explanation for the pLLS is for sightline B to be piercing an inflowing cold stream of low-metallicity gas from the intragroup medium into galaxy G3. The impact parameter of ρ/Rvir ∼ 0.8 with G3, the inferred metallicity of ∼ 1/10-th solar, and the alignment of the sightline closer to the galaxy's major axis are all consistent with this alternate scenario.
In contrast to the low ionization gas, the O VI along both the lines of sight has comparable column densities of log N ∼ 14.1 − 14.3, broad b-values of ∼ 33 − 52 km s −1 and line of sight velocity separation of |∆v| ≈ 64 km s −1 , consistent with tracing warm gas that is either part of a uniform large-scale medium such as the warm CGM, or restricted to a much narrower thickness transition temperature zone resulting in similar column densities. It is well known from observations and simulations that as many as 50% of the O VI absorbers at z 1 are tracers of collisionally ionized plasma with temperatures of T ∼ 10 5 K (Tripp et al. 2008;Smith et al. 2011;Savage et al. 2014). At both low and high redshifts (z ∼ 2 − 3), the properties of O VI associated with optically thick H I absorbers concur with the physical properties of gas cooling in outflows in the CGM of star-forming galaxies (Grimes et al. 2009;Lehner et al. 2014). In these environments the O VI can form at the narrow interface layers between cold clouds and a hotter phase of gas belonging to the halo, or the intragroup/intracluster medium (Begelman & Fabian 1990;Sembach et al. 2003;Narayanan et al. 2018). In the modelling of O VI along both the sightlines, we find that collisional ionization in cooling plasma can explain the observed O VI at solar metallicities. The O VI and the BLA along both sightlines can be probing a warm (T ∼ 10 5 K) transition temperature layer between cool gas of different column densities entrenched in the hot halo of galaxy G2 or G3 or hot intra-group gas. The chances of finding galaxies within impact parameter of 250 kpc of O VI system of equivalent width ≤ 0.2 Å in a group environment is high as compared to isolated galaxies (Pointon et al. 2017).
If instead, the O VI along both sightlines are from the diffuse warm phase of the CGM of the nearest galaxy (which is G3), or constitutes warm intragroup gas, then the ionization fraction of O VI given by the hybrid model, and the physical scales along and across the lines of sight can be used to determine the mass of this phase. Using equation 2 of Tumlinson et al. (2011) for estimating the gas mass for the mean O VI column density, and CGM gas radius of log[ N (O VI) /cm −2 ] = 14.2, and R = 260 kpc, 4 respectively, and an O VI covering fraction of 0.8, we estimate a CGM or warm intragroup gas mass of 4.5 × 10 9 × Z⊙/Z M⊙. The estimate is a conservative lower limit based on the 1σ lower limit of T = 0.5 × 10 5 K in the hybrid models for the BLA -O VI gas. The metallicity of the O VI gas is poorly constrained due to the uncertainty in the temperature given by the BLA line widths. The mass estimate will be an order of magnitude higher, if the metallicity of the O VI -BLA gas is one-tenth of solar, instead of solar. Similarly, at T > 0.5 × 10 5 K, the observed O VI will be recovered by the hybrid models at lower ionization parameters in which the ionization fraction of O VI would be lower, thus raising the CGM/warm intragroup gas mass estimate.

CONCLUSIONS
In this work, we have presented the metallicity and ionization analysis, and information on the extended galaxy environment for an absorber at z abs = 0.399 seen along two closely separated lines of sight towards the background quasars Q 0107 − 025A and Q 0107 − 025B . The key results are as follows: The O VI from this phase is severely underproduced. A BLA with log N [H I/cm −2 ] = 14.84 ± 0.16 and b(H ) = 64 ± 9 km s −1 is required to explain the broad wings seen in the Lyα and Lyβ profile. The BLA and O VI indicate the presence of gas with T = 8.9 +5.6 −5.6 × 10 4 K. The O VI in the pLLS can be reproduced by photoionization in a more diffuse (n H ∼ 3.2 × 10 −5 cm −3 ), higher ionization gas of similar metallicity (1/10-th solar) as the low ionization phase. Alternatively, the O VI could be from a radiatively cooling collisionally ionized plasma with T 10 5 K and [O/H] = 0, similar to the metallicity in the O VI gas along sightline A.
(v) The sightlines are intercepting a galaxy group environment with 16 galaxies within a projected separation of ρ < 5 Mpc and velocity offset of ∆v < 500 km s −1 from the mean absorber redshift. The sightlines are at projected separations of ρ/Rvir ∼ 1.1 and ρ/Rvir ∼ 0.8 from a galaxy (G3) of stellar mass M * ∼ 10 10.7 M⊙, and luminosity LR ∼ 0.07L * R . The pLLS absorber along sightline B with an inferred metallicity of 1/10-th solar happens close to projected major axis of this galaxy (azimuthal angle, Φ ∼ 3 • ) consistent with the large H I column a result of the line of sight piercing a cold accretion stream from the intragroup medium. The SFR < 0.1 M⊙ yr −1 of this galaxy is however inconsistent with the solar metallicity we derive for the low ionization gas along sightline A and for the O VI along both sightlines. An alternate possibility is for the absorption along both sightlines to be tracing metals dispersed in a past bipolar outflow (∼ 600 Myr) from another galaxy (G2) at ρ/Rvir ∼ 1.2 and ρ/Rvir ∼ 1.7 with similar stellar mass and luminosity, but with a SFR ∼ 3 M⊙ yr −1 .
(vi) The O VI along both sightlines are consistent with tracing radiatively cooling plasma in a collisionally ionized interface layer between a low ionization gas cloud and the hotter phase of the CGM of the galaxy G3, which is nearest in normalized impact parameter (ρ/Rvir), or the larger intragroup gas. Collisional ionization models of such a non-equilibrium scenario predict [O/H] ∼ 0. Alternatives are for the O VI to be tracing the warm diffuse CGM of the nearest galaxy (G3), or warm intragroup gas, in which case, the physical scales of L ∼ 35 kpc and l ∼ 520 kpc along and across the lines   of sight of this gas phase yield an ionized gas mass lower limit of 4.5 × 10 9 × Z⊙/Z M⊙. (1) Label of the 16 galaxies labelled as G1, G2 and so on. Information about G1 is from our analysis of the MUSE observations (Beckett et al. , in preparation) while rest are taken from Tejos et al. (2014) which are MOS observations.
(2) RA of the galaxies in degrees.
(3) DEC of the galaxies in degrees.
(4) The apparent magnitude of the galaxies in R-band is taken from (Beckett et al. , in preparation) and Tejos et al. (2014).
(5) Redshift of the galaxies. (6) Impact parameters in kpc of galaxies with respect to the quasar line of sight A. (7) Impact parameters in kpc of galaxies with respect to the quasar line of sight B. (8) Velocity offset between the galaxies and the absorber at z abs = 0.39949 towards line of sight A. (9) Velocity offset between the galaxies and the absorber at z abs = 0.39916 towards line of sight B. (10 By using cosmology of Bennett et al. (2014), we calculated absolute magnitude of the galaxies in the R-band. (11) The Schechter absolute magnitude of the L * galaxy (Schechter 1976) in the R-band (M* R ) for the galaxies were calculated using relation in Dahlen et al. (2005). This in turn gave the values of (L/L * ).
(12) Stellar masses were estimated using the mass-to-light ratio tables as a function of colour, given in Table A7 of Bell et al. (2003). These relationships have a scatter of 0.1-0.2 dex, with systematic shifts of up to 0.4 dex dependent on the assumed initial mass function. (13) The halo mass was converted into r 200 (kpc) using the halo mass by adopting the cosmology of Bennett et al. (2014). (14) The ratio of impact parameter of the galaxies with respect to sightline A and their virial radii. (15) The ratio of impact parameter of the galaxies with respect to sightline B and their virial radii. (16) Halo masses are estimated using the stellar-to-halo-mass relation given in Behroozi et al. (2010). The uncertainty in halo mass is dominated by a ∼ 0.25 dex uncertainty in stellar mass, which corresponds to ∼ 25 % in R vir . (17) & (18) Star-formation rates (SFR) were estimated by converting H α or [O II] emission line luminosity to SFR with the formulae given in Kennicutt (1998), after applying 1 magnitude of extinction (approximately the mean for low-redshift galaxies as found by Charlot et al. (2002)). (19) This column mentions the spectrographs used for detecting galaxies and obtaining their spectra. Note. The columns show, from left to right: name of the quasar line of sight, the redshift of the absorber towards the corresponding line of sight, different phases probed by the absorber by single/multi-components, the equilibrium model adopted for ionization modelling, neutral hydrogen column density with 1σ error, temperature due to the different phases present, gas density of the absorber, metallicity of the absorber, total column density of the absorber and line of sight thickness of the absorber. The temperature listed is from the combined b-values of H I and metal ions. For the high ionization phase in the central component towards sightline A, the associated H I column density cannot be determined from the data. The upper limit on temperature is from the b-value of O VI. We have adopted two models to