Probing the Extended Gaseous Regions of M31 with Quasar Absorption Lines††thanks: Based on observations made with the NOAO 2.1-m and the NASA/ESA Hubble Space Telescope.
We present Hubble Space Telescope - Cosmic Origins Spectrograph spectra of ten quasars located behind M31, selected to investigate the properties of gas associated with its extended disk and high velocity clouds (HVCs). The sightlines have impact parameters ranging between kpc and kpc. No absorption is detected in the four sightlines selected to sample any extended disk (or halo) gas that might be present in the outer regions of M31 beyond an impact parameter of kpc. Of the six remaining sightlines, all of which lie at kpc and within the cm boundary of the H i disk of M31, we detect low-ionization absorption at M31 velocities along four of them (three of which include Mg ii absorption). We also detect Mg ii absorption from an HVC. This HVC sightline does not pass through the 21 cm disk of M31, but we detect additional Mg ii absorption at velocities distinct from the HVC that presumably arises in the halo. We find that along sightlines where both are detected, the velocity location of the low-ion gas tracks the peak in 21 cm emission. High-ionization absorption is detected along the three inner sightlines, but not along the three outer sightlines, for which C iv data exist.
As inferred from high-resolution 21 cm emission line maps of M31’s disk and extended regions, only one of the sightlines may be capable of harboring a damped Ly system, i.e., with cm. This sightline has impact parameter kpc, and we detect both low- and high-ion absorption lines associated with it.
The impact parameters of our observed sightlines through M31 are similar to the impact parameters of galaxies identified with Mg ii absorbers at redshifts in a 2011 study by Rao et al. However, even if we only count cases where absorption due to M31 is detected, the Mg ii 2796 rest equivalent width values are significantly smaller. In comparison, moderate-to-strong Mg ii absorption from Milky Way gas is detected along all ten sightlines. Thus, this study indicates that M31 does not present itself as an absorbing galaxy which is of higher-redshift galaxies inferred to give rise to moderate-strength quasar absorption lines. M31 also appears not to possess an extensive large gaseous cross section, at least not along the direction of its major axis.
keywords:galaxies: individual: M31 - quasars: absorption lines
The standard paradigm for metal-line absorption systems in quasar spectra is that they arise in the extended gaseous halos/disks of galaxies well beyond their observable optical radii. However, with the exceptions afforded by gravitationally-lensed quasars, rarely is there more than one sightline passing in the vicinity of a galaxy. As such, the study of quasar absorption lines arising in extended gas associated with the great spiral galaxy in Andromeda (M31) represents a unique opportunity. M31’s large extent on the sky means that many quasar sightlines should intercept its extended gas. For example, the atoms cm 21 cm emission contour around M31, as derived from the data discussed by Thilker et al. (2004), is approximately square degrees on the sky (see Figure 1). We list some properties of M31 in Table 1. Quasar surveys have shown that there are as many as 18 quasars per square degree brighter than at (Richards et al. 2009, Abraham et al. 2012). Thus, there are likely to be on the order of 135 such quasars behind M31 within the boundaries of its observed 21 cm emission, and a factor of several more in its extended gaseous disk and halo regions. However, until now, quasar absorption lines have never been successfully used to study the extended gas of M31 because of the lack of sufficiently-bright, identified quasars.
Two of the most recognizable signatures of metal lines in quasar spectra are the Mg ii 2796,2803 and C iv 1548,1550 doublets, which have been studied in numerous quasar absorption-line surveys. The first comprehensive study which demonstrated that galaxies at large impact parameters exist along the sightlines to low-redshift Mg ii absorbers was by Bergeron and collaborators, e.g., Bergeron & Boissé (1991). They estimated that the average Mg ii radius of a spherical gaseous envelope surrounding an galaxy is 3.5 to 5.0 ( 55 to 80 kpc) at for rest equivalent widths Å, where is the Holmberg radius. Others had made similar estimates (e.g., Lanzetta et al. 1987, Lanzetta & Bowen 1990, Steidel 1993). The recent survey of galaxies associated with Mg ii absorbers at by Rao et al. (2011) showed that the gaseous extent of Mg ii-selected absorbing galaxies could be as large as 100 kpc. At , Chen et al. (2010) find that the mean covering fraction for Mg ii absorbers with Å within 130 kpc of a 2 galaxy (for ) is 70%. Therefore, if cross sections have remained constant since , then we might expect that gas giving rise to Mg ii is likely to be present in the extended gaseous regions of M31 out to a radius of kpc or more, assuming it is a typical absorbing galaxy.
|Property||Value||ReferenceaaReferences: 1. Evans et al. (2010); 2. Riess et al. (2012); 3. Corbelli et al. (2010); 4. de Vaucouleurs et al. (1991); 5. Cool et al. (2012); 6. Tamm et al. (2012)|
|bbOptical radius at -band surface brightness magnitudes per square arcsec.||22.3 kpc||4|
|ccAssuming (Cool et al. 2012).||4,5|
|ddFrom the atoms cm contour (Figure 1).||33 kpc||3|
As described in §2, we obtained Hubble Space Telescope (HST) - Cosmic Origins Spectrograph (COS) spectra of ten quasars located behind M31 in order to investigate the properties of the gas in its extended disk and high velocity clouds (HVCs). We searched for Mg ii, C iv, and other absorption lines to do this. In §3 we describe the results obtained from each spectrum. We discuss the results in §4 and end with a summary and conclusions in §5. This study indicates that M31 does not present itself as an absorbing galaxy which is of the higher-redshift galaxies inferred to give rise to moderate-strength quasar absorption lines.
2.1 Existing H i 21 cm Emission Observations
Since M31 is the nearest large spiral galaxy close to the Milky Way, it has been the subject of many observational studies. Specifically for this work, we will make reference to several results over the past decade from radio observational studies of M31’s H i 21 cm emission. These are: the Green Bank Telescope (GBT) study of Thilker et al. (2004), which identified high-velocity clouds (HVCs) but at lower spatial resolution than later studies; the Westerbork Synthesis Radio Telescope (WSRT) study of Braun and Thilker (2004) which discovered the M31-M33 H i bridge, and of Westmeier et al. (2005), which focused on obtaining higher spatial resolution observations of HVCs; the WSRT study of Braun et al. (2009), which obtained observations over a wide field at high spatial resolution; and the study of Corbelli et al. (2010), which smoothed the data to lower spatial resolution in order to fit a tilted-ring model to M31’s warped disk and study its rotation curve. At some level, all of this work was collaborative by various members of the same group, and in later studies they made use of results that could be derived from earlier data sets.
H i emission spectra were extracted from the Thilker et al. (2004) and Braun et al. (2009) datacubes along the sightlines toward our target quasars. These data, originally in units of Jy/beam, were converted to under the assumption of negligible H i opacity. This conventional assumption, while recently questioned by Braun et al. (2009) and Braun (2012) in the dense gaseous environment of the traditional optical disk and slightly beyond, is expected to be satisfied in the outer disk and halo environment. A more significant concern regarding the from observations of emission is the vastly different scale probed by the GBT and WSRT relative to COS. The maximum linear spatial resolution of the high resolution 21 cm observations is pc at the distance of M31. This scale is of order times larger than the linear spatial scale sampled in quasar “pencil-beam” absorption-line observations, where the pencil-beam has the scale of the UV continuum emitting region of the background quasar. Thus, values derived from 21 cm emission observations are averaged over a much larger spatial scale in comparison to those derived from quasar absorption-line spectra. Nevertheless, using 21 cm observations to derive average values along our sightlines, and noting the velocity range of detected emission, provides some important information.
As an aside, we note that it would be interesting if results derived from M31’s 21 cm emission data could someday be compared with determinations from Lyman series absorption seen in the UV spectra of background quasars. One could then get an H i column density measurement averaged over less than a milli-parsec region in M31, in comparison to the 50 pc linear spatial scale offered by the radio observations. This would provide information on the homogeneity and size scale of H i absorbing regions in M31.
2.2 Optical Discovery Spectra of Quasars behind M31
We started this project by developing a list of quasars in especially desirable locations (see below) relative to M31. These were initially quasar candidates, since existing catalogs generally did not include quasars behind M31. The quasar candidates were selected from special plates of the SDSS, which were obtained specifically to find quasars behind the extended regions of M31 (Adelman-McCarthy et al. 2006). Of the 219 candidates, 108 were confirmed as quasars. Twenty-three of the 108 were spectroscopically confirmed during our October 2003 NOAO 2.1 m Gold Camera run at Kitt Peak. To make follow-up observations with HST-COS (§2.3 and §3) more feasible, we concentrated on finding brighter quasars. We also focused our search behind M31’s extended major axis to probe possible disk gas that could sample its outer rotation curve. See Figures 1 and 2, and Tables 1 and 2, for information on their locations relative to M31 and the discovery spectra. Quasars labeled 1 through 4 would sample any extended disk gas (or possibly halo gas) that is undetected in 21 cm emission; quasars 5, 6, 8, and 9 lie near the edge of detected 21 cm emission; the sightline towards quasar 7 passes through a high velocity cloud (HVC) in the circumgalactic environment of M31; and quasar 10 lies behind the 21 cm emission H i disk as well as two other HVCs. Importantly, owing to M31’s systemic velocity of km s (Corbelli et al. 2010) and its direction of rotation, absorption originating on the southwest side of M31 will not be confused with Galactic absorption. Consequently, quasars 1 through 4 offer the best opportunities for observing extended disk gas and measuring M31’s rotation curve much farther out than possible with 21 cm emission observations. Unfortunately, while obtaining information on M31’s rotation curve at large galactocentric distance was one of the primary motivations for observing quasars 1 through 4, no M31 absorption near the expected velocity was detected in their UV spectra (§2.3 and §3). We note, however, that higher quality observations might yet be able to detect gas at these locations. Observing quasars on the extended northeast side of M31 was avoided because of potential confusion with Galactic absorption.
|Quasar||RA (2000)||Dec (2000)||Map||SDSS||b||COS G140L||COS G230L||Sightline notesccSee Figure 1.|
|h m s||IDaaQuasar IDs in order of increasing RA.||mag||(kpc)bbProjected distance from galactic center assuming that the center of M31 is at (004244, +411608) and that the distance to M31 is 752 kpc. See Table 1.||Exp time (s)||Exp time (s)|
|0018+3412||00 18 47.45||+34 12 09.6||1||0.447||17.7||111.9||3446||5646||extended disk|
|0024+3439||00 24 50.05||+34 39 42.8||2||0.822||18.1||98.6||5092||extended disk|
|0030+3700||00 30 17.43||+37 00 54.3||3||1.237||17.5||64.4||4054||3191||extended disk|
|0031+3727||00 31 32.37||+37 27 51.8||4||1.300||18.5||57.6||4455||2246||extended disk|
|0032+3946||00 32 55.70||+39 46 19.3||5||1.138||18.6||31.5||10324||edge of 21 cm disk|
|0037+3908||00 37 48.00||+39 08 58.7||6||1.130||18.4||30.5||10354||edge of 21 cm disk|
|0040+3915||00 40 59.03||+39 15 12.3||7||1.099||18.5||26.9||6293||11101||HVC + 21 cm disk edge|
|0043+4016||00 43 52.45||+40 16 29.4||8||1.093||18.2||13.4||7073||5537||edge of 21 cm disk|
|0043+4234||00 43 54.98||+42 34 30.4||9||0.191||18.0||17.4||3555||6137||edge of 21 cm disk|
|0046+4220||00 46 55.52||+42 20 50.1||10||0.306||18.1||17.5||2572||5217||2 HVCs + 21 cm disk|
2.3 Hst-Cos UV Spectroscopy
The HST-COS spectroscopic data were obtained during the period July-October 2010. Table 2 gives details of the quasars and the HST-COS observations. We decided to make a broad initial absorption-line survey in order to maximize the observed number of metal-line transitions we could reasonably cover within our allocation of 39 HST orbits.111Parallel imaging data were also obtained. These will be discussed in Thilker et al. (in prep.). The aim was to reach a signal-to-noise ratio which would enable us to detect Mg ii and C iv absorption rest equivalent widths commonly seen in prior, large moderate-resolution quasar absorption-line surveys. Therefore, we did not use higher-resolution COS gratings. However, it would indeed be worthwhile to perform follow-up spectroscopy of a number of our detections at higher spectral resolution and signal-to-noise ratios.
The COS gratings used in this study along each sightline are specified in Table 2. The near ultraviolet (NUV) G230L grating has a resolution of 2 pixels or Å at the wavelength of the Mg ii 2796,2803 doublet, which corresponds to km s on a velocity scale. The far ultraviolet (FUV) G140L grating has a resolution of 7 pixels or Å at the wavelength of the C iv 1548,1550 doublet, which corresponds to km s. Given the redshifts of the quasars, we should note that in certain wavelength regions there is the possibility of contamination by Ly forest absorption. For example, Ly forest absorption would potentially be visible near any Galactic or M31 Mg ii absorption when the quasar’s redshift is higher than (i.e., in quasar 4) and near any Galactic or M31 C iv absorption when the quasar’s redshift is higher than (i.e., in all quasars except quasar 9). However, according to Weymann et al. (1998), the incidence of Ly forest absorption lines with rest equivalent widths 0.24 Å at these relatively low redshifts is typically only about one line per Å (about one line per km s), so we did not necessarily anticipate too much confusion due to overlapping Ly forest absorption. There might also be overlapping absorption due to unidentified metal-line systems. In §3 we note instances where Ly forest absorption or other overlapping unidentified absorption appears to be a confusing factor.
Seven quasars were observed with both the NUV and FUV gratings, while three were targeted with the NUV grating alone. These three had low FUV fluxes based on the GALaxy Evolution eXplorer (GALEX) telescope measurements, and so they were not observed. The NUV grating covers Fe ii, Mn ii, Mg ii and Mg i transitions, while the FUV grating covers C iv, Si iv and several lower-ion transitions, as described in §3.
Pipeline flux-calibrated and wavelength-calibrated spectra were used for all the measurements, and no additional calibrations or re-calibrations were carried out. The wavelength scale is heliocentric, and measured velocity offsets relative to a transition of interest are made on this scale. Before making absorption-line measurements, the FUV spectra were re-binned to two pixels per resolution element and all spectra were normalized using an interactive algorithm which fitted splines to a quasar’s observed continuum plus broad emission lines to derive a pseudo-continuum. We used the pipeline-provided standard deviation in flux to calculate the 1 error in the normalized flux. When reporting errors in equivalent width measurements, we do not include (propagate) any errors that might arise during the process of defining a pseudo-continuum.
Figures show the pseudo-continuum-normalized spectra near the predicted locations of metal lines along the ten sightlines, and Table 3 gives the measured metal-line absorption rest equivalent widths or upper limits for both M31 and Galactic lines. To make these measurements, heliocentric velocity locations for the absorbing gas had to be determined. The procedure for this is discussed below and the results on velocity offsets are given in Table 4.
For the low-ion transitions, the narrow Mn ii lines (when present) allow for a more accurate determination of the velocity centroid of Galactic gas since they are well-fitted by single Gaussians. Therefore, the velocity offsets of low-ion Galactic absorption lines are defined by the centroids of Galactic Mn ii absorption for sightlines 2, 8, 9, and 10222The sightline 10 Galactic component is heavily blended with the M31 disk component, as described in the discussion of sightline 10.. The centroids of Galactic Mg ii are used to define the velocity offsets of absorption along other sightlines. The wavelength interval covered by the COS-FUV spectra includes transitions due to Si ii, O i, C ii, C ii, Fe ii, and Al ii. The centroids of these low-ion lines were fixed at the velocities determined from either the Mn ii line or Mg ii as indicated above. In the panels for each figure, dash-dot vertical lines are drawn at the determined velocity offsets of M31 and Galactic gas.
The only high-ion transitions detected in our spectra are due to C and Si. The velocity centroids of Gaussians fitted to the C iv lines were allowed to vary since low-ion and high-ion absorption lines are not a priori required to have the same velocity centroids or line widths. The C iv line and Si iv lines were then constrained to have the same velocity locations and widths as the C iv line, within the uncertainties and resolution of the data. Inspection of the final fits suggests that this was a reasonable constraint.
The 1 error in the normalized flux is shown in the figures as a black dotted line. M31 and Galactic absorption transitions that are identified at a level of significance are indicated in the figures by red profiles. A rest equivalent width detection threshold is an appropriate criterion for identifying absorption because we already know the approximate velocity location of M31 absorption (e.g., from M31’s 21 cm emission). We also searched for significant absorption in a wider velocity window. Gaussian profiles are fitted to detected absorption. If more than one Gaussian is required to fit the data, we show the individual Gaussians as red dashed profiles, visible above the solid red profile. In the absence of multiple Gaussians, the red solid profile will lie on top of the red dashed profile, and the red dashed profile will not be visible. However, the measurements indicated by the red dashed profiles are what we report in Tables 3 and 4. As noted earlier, the positions of most low-ion lines are fixed by the centroid of either the Mn ii or the Mg ii line; however, their widths are allowed to vary in order to obtain the best fit. In a few cases, even the velocity offsets had to be allowed to vary up to one resolution element in order to obtain a satisfactory fit. Also, while performing the fits, we identified some absorption in the spectra which were likely blends resulting from a real M31 or Galactic absorption line plus overlapping or nearby absorption due to, for example, Ly forest absorption, some other unrelated absorption, or even related absorption such as C ii 1334.5 and C ii 1335.7. When this happened, we fitted Gaussians to these nearby absorption components in order to better isolate the M31 and Galactic absorption transition of interest. We refer to this as deblending. However, when we report results in Tables 3 and 4, as noted earlier, only absorption taken to be due to the designated transition of interest in M31 or the Galaxy is reported and shown on the figures. Other nearby absorption lines which were fitted in order to isolate M31 and Galactic gas are shown as green dashed Gaussian profiles. The identifications and measurements of M31 and Galactic lines in the presence of confusing overlapping or nearby absorption should be considered less secure.
When a line is not detected (i.e., the detection is ) at its expected velocity offset, or nearby absorption not due to the transition of interest appears to be present, a red dotted Gaussian profile with FWHM equal to the spectrograph resolution (i.e., Å or km s for the NUV lines and Å or km s for the FUV lines) is shown on the figures to indicate the reported upper limit. If no overlapping or nearby confusing absorption is present, this is just the upper limit on equivalent width generated from the error in normalized flux. However, if overlapping or nearby absorption is present, the upper limit is determined from the strength of this overlapping or nearby absorption. Lacking evidence that a low-oscillator-strength transition should be present along a particular sightline, we would attribute any significant detected absorption as due to overlapping absorption, and list it as an upper limit.
In cases where the velocity of an M31 absorption line overlaps with the velocity of a different Galactic absorption line, for example, the M31 C iv and the Galactic C iv lines, or the M31 Si ii and the Galactic O i lines along sightlines 1, 3, and 4 (Figures 3, 5, and 6), we assign the absorption to the Galactic absorption system. The measurement is listed in Table 3 only for the Galactic absorption line.
The bottom panels for sightlines 5 through 10 (Figures ) show H i 21 cm emission profiles extracted from the GBT data of Thilker et al. (2004). The intensities are scaled to accentuate the very weak emission signal from M31. The dash-dot horizontal line drawn in each 21 cm panel marks the location of zero intensity. The H i 21 cm emission disk of M31 extends out to kpc as determined from the cm column density contour (Figure 1), and no H i 21 cm measurements exist at the positions of quasars 1 through 4. Therefore, to estimate equivalent width upper limits for these four sightlines, we have assumed that M31’s 21 cm rotation curve is flat at large galactocentric distances and we extrapolate the sightline 21 cm emission velocity out to the positions of quasars 1 through 4 to predict a probable velocity location of absorbing gas. Note that M31 is nearly edge-on and inclined on the plane of the sky. Thus a very small inclination correction is needed since . Then the assumption of a flat rotation curve suggests that if metal-line absorption is present in M31’s outer regions, we might find it near a heliocentric velocity location of km s. This is where we determine M31 equivalent width upper limits for sightlines 1 through 4. We note that the choice of where to measure potential absorption in the four outer sightlines is purely an algorithmic decision given that flat rotation curves exist. We also considered the Tamm et al. (2012) study which derives a rotation curve out to a galactocentric radius of kpc. They employ, among other diagnostics, observations of stellar streams (Fardal et al. 2006) and satellite galaxies (Tollerud et al. 2012) which yield rotational velocities of km s near the position of our outermost sightline. This translates to a heliocentric velocity of km s since our outer sightlines lie on the approaching, SW, side of M31. This is well within one resolution element (§2.3) of our assumed velocity location of km s. Therefore, we are confident that we have not missed any absorption from gas in M31 along the outer four sightlines that is above our detection limits.
Tables 3 and 4 summarize all of the measurements and upper limits, both for M31 and the Milky Way Galaxy. A discussion of individual sightlines follows (see Figures ), with emphasis on what they reveal about M31 gas. The discussions are presented in order of increasing sightline right ascensions. This ordering generally follows decreasing impact parameter, , except for the last three sightlines which all have kpc. At the beginning of each discussion we indicate the maximum wavelength at which Ly forest absorption might cause blending and confusion, Å.
- 1. 0018+3412 ( kpc, Å, Fig. 3):
No significant M31 absorption is detected along this sightline, and H i 21 cm emission maps of M31 do not extend this far out. Therefore, rest equivalent width upper limits on absorption were measured at km s as described earlier. At this velocity location, the red dotted Gaussian lines show the velocity positions and rest equivalent widths of hypothetical unresolved absorption lines with 2 levels of significance, and these are the upper limits reported in Table 3. Galactic absorption is clearly present. Suspected confusion (blending) due to overlapping Ly forest absorption is apparent for the Si ii 1260, Si ii 1304, O i 1302, C ii 1334, and C iv 1550 Galactic absorption lines. The method we used to measure such cases was discussed above.
Figure 3: Normalized spectra versus velocity for the labeled transitions in the spectra of 0018+3412. The black dotted line is the error spectrum. All velocities are heliocentric. The vertical dot-dashed lines are Milky Way (near 0 km s) and M31 detected or assumed velocities. See Table 4. Fits to M31 and Galactic absorption lines detected at a significance are shown as heavy dashed (if part of a blend) or solid red lines. See text. Dotted red lines indicate upper limits. Green dashed lines are components within a blend that are unrelated to M31 or Galactic absorption.
- 2. 0024+3439 ( kpc, Å, Fig. 4):
As in the previous sightline, no significant M31 absorption is detected, and H i 21 cm emission maps do not extend this far out, so upper limits were measured at a velocity location of km s. Only NUV spectra of this quasar were obtained. Therefore, for example, the C iv region was not observed. A Galactic MnII 2576 line is detected at a level of significance of , however the two weaker members of the triplet are not detected at . Galactic Mg ii and Fe ii absorption are clearly detected.
- 3. 0030+3700 ( kpc, Å, Fig. 5):
Again, no significant absorption lines from M31 are detected at or near km s, and the 21 cm emission maps do not extend out this far. Among the significant Galactic absorption lines that are detected, the measurements of Si ii 1260, Si iv 1393, C iv 1548 and Fe ii 2586 were made in the presence of overlapping unrelated absorption using the method described earlier. While only the stronger members of the Galactic Si iv and C iv doublets are detected, the rest equivalent width upper limits of the weaker members of these doublets are consistent with their expected strengths based on values. In addition to the detected Galactic metal absorption lines, at least two partial Lyman limit absorption systems are present in the spectrum. One at is clearly visible in the FUV observation (not shown). Based on the difference in flux level between the FUV and NUV observations, and the presence of some strong Ly forest absorption near and just shortward of the Ly broad emission line, at least one other Lyman limit absorption system is likely to be present at . However, it is not directly visible in our observations because it falls in the wavelength gap between the FUV and NUV spectra.
Figure 5: Same as Figure 3, but for 0030+3700.
- 4. 0031+3727 ( kpc, Å, Fig. 6):
As with the first three sightlines, no significant absorption lines from M31 gas are seen, and the 21 cm emission map does not extend out this far. M31 upper limits were estimated at km s for both the high and low ions. Galactic absorption is clearly detected for some transitions, but the measurements of Si ii 1260, C ii 1334, Si ii 1526, Fe ii 1608, and Fe ii 2600 required deblending due to the presence of unrelated overlapping absorption.
Figure 6: Same as Figure 3, but for 0031+3727.
- 5. 0032+3946 ( kpc, Å, Fig. 7):
Only NUV spectra were obtained for this quasar. An M31 Mg ii absorption line with a significance of at a heliocentric velocity of km s appears to be present (see Table 3), however a corresponding 2-pixel-wide absorption feature near the expected position of Mg ii has a significance . If present, this absorption may originate at the southwest edge of M31’s disk (see Figure 1). Apart from strong Galactic emission, the GBT 21 cm data along this sightline (bottom panel of Figure 7) shows evidence for M31 emission between and km s. Although the resolution of the NUV spectrum is Å ( km s) at the position of Mg ii, the centroid of the absorption line can be estimated with an uncertainty of km s (see §4). Thus, the identified Mg ii feature at km s is near the maximum velocity of observed 21 cm emission. Keeping in mind the limitations of using H i 21 cm emission observations to determine H i column densities (§2.1), we find atoms cm along this sightline. Very significant Galactic Mg ii and Fe ii absorption is detected along this sightline, but the Galactic Fe ii line was deblended to separate it from unrelated nearby absorption.
- 6. 0037+3908 ( kpc, Å, Fig. 8):
Only NUV spectra were obtained for this quasar. Absorption from M31 gas is not detected. However, apart from the strong Galactic emission, the GBT data along this sightline reveal M31 21 cm emission between and km s (bottom panel of Figure 8), with an integrated column density of atoms cm (see §2.1). The upper limits on M31 absorption are made at the central velocities predicted by the observed M31 21 cm emission. Very significant Galactic Mg ii and Fe ii absorption is detected along this sightline. The Galactic Fe ii and Fe ii lines were deblended to separate them out from unrelated nearby absorption.
- 7. 0040+3915 ( kpc, Å, Fig. 9):
Only the velocity profiles in the vicinity of Mg ii and Mg i are visible in our observations for two reasons. First, the quasar spectrum exhibits intrinsic broad absorption lines (BALs) and the N v BAL trough overlaps the Mn ii and Fe ii absorption-line regions. This prevents useful measurements of M31 and Galactic lines in those regions. Second, the FUV spectrum shows no useful continuum flux, possibly due to strong shorter-wavelength BALs and/or overlapping intervening Lyman limit absorption. Mg ii due to M31 gas appears as two absorption components in the NUV spectrum. The noise characteristics of the spectrum are worse in the corresponding Mg ii region, and two absorption components are not seen (a single Gaussian was fitted to the absorption), but we give this lower weight due to the higher noise. The two vertical dash-dot lines at km s and km s mark the velocity positions of the two M31 Mg ii absorption components. The sightline passes through an HVC (see Figure 1) , whose 21 cm emission profile can clearly be seen in the bottom panel of the figure peaking at km s. The GBT data reveal that this 21 cm emission extends between and km s. Thus, the two Mg ii absorption components at km s and km s may correspond to M31 halo gas and HVC gas, respectively, with the halo component showing no apparent 21 cm emission. From the WSRT 21 cm emission data, the integrated H i column density in the HVC is estimated to be atoms cm (see §2.1). Very significant Galactic Mg ii absorption is present along this sightline.
- 8. 0043+4016 ( kpc, Å, Fig. 10):
This is the lowest impact parameter sightline. M31 low-ion absorption from Si ii and C ii is detected, and high-ion absorption from C iv is detected, but the Si ii and C iv lines had to be deblended from overlapping unrelated absorption. Given that the 21 cm emission extends over a large range in velocity, we cannot rule out that all the absorption features within the C iv blend are due to C iv absorption over a wide velocity range. Confirmation would require a higher signal-to-noise spectrum; here, we identify the lowest velocity component with the M31 C iv absorption line. The low-ions are centered at km s and the high-ions are centered at km s. However, Mg ii and Fe ii absorption lines from M31 gas were not detected. The GBT data show that 21 cm emission from M31 exists along this sightline between km s and km s, with a total integrated column density of atoms cm. We note that the absorption-line velocities are coincident with the peak in the 21 cm emission-line spectrum (bottom panels of Figure 10). Many significant Galactic absorption lines are present. Galactic Si ii and C iv had to be deblended to separate them out from unrelated overlapping absorption.
Figure 10: Same as Figure 7, but for 0043+4016. FUV data obtained for this quasar are also shown.
- 9. 0043+4234 ( kpc, Å, Fig. 11):
The sightline to this quasar passes “above” the H i 21 cm emission disk of M31 on the receding, northwest, side (see Figure 1). Due to the location of the sightline, the detected M31 and Galactic absorption lines needed to be deblended from each other. We used two-component Gaussian fits with fixed velocity components to do this. Measurements of the Si ii 1260, C ii 1334.5 (and C ii 1335.7), Si iv , and Al ii absorption lines are further complicated by other overlapping absorption. For low-ion absorption the velocity centroid for the Galactic lines was fixed using Mn ii , while allowing the position of the M31 low-ion velocity centroid to vary until the best least-squares solution was found. Deblending indicates that the detected M31 low-ion gas, which gives rise to transitions of Si ii, C ii, Al ii, Fe ii and Mg ii, is located at km s, and the Galactic low-ion gas is located at km s. It is notable that along this sightline there is a detection of Galactic C ii 1335.7. The M31 high-ion gas, which gives rise to transitions of Si iv 1393 and C iv 1548, are also members of a multi-component blend with Galactic lines. Using a procedure similar to the one used for the low-ions, we find that the M31 high-ion gas is at km s and the Galactic high-ion gas is at km s. GBT data reveal M31 21 cm emission between km s and km s, with a total integrated column density of cm (see §2.1). The higher velocity edge of the M31 21 cm emission is uncertain since it may overlap with Galactic 21 cm emission.
Figure 11: Same as Figure 7, but for 0043+4234. FUV data obtained for this quasar are also shown.
- 10. 0046+4220 ( kpc, Å, Fig. 12):
To infer what gaseous structures exist along this sightline we are guided by the observed GBT H i 21 cm emission velocity profile, which is shown in the bottom panels of Figure 12. An inset in the bottom left panel shows the entire 21 cm profile. Most notably, the weaker 21 cm peak near km s represents Galactic emission, while the stronger 21 cm emission peak near km s represents M31’s disk; however, such a velocity separation cannot be distinguished in the COS G140L and G230L absorption-line spectra. More generally, the entire 21 cm velocity profile and the detected low-ion absorption lines have allowed us to infer the existence of five gaseous structures along this sightline: one near km s (Galactic gas), one near km s (M31 disk gas), one near km s (M31 halo gas), one near km s (a M31 HVC), and one near km s (a second M31 HVC). As with sightline 9, detected Galactic and M31 absorption lines needed to be deblended from each other, but the blending along this sightline is more severe. In particular, the low-ion absorption detected near km s must be a blend of Galactic gas and M31 disk gas, with most of the absorption being due to M31 disk gas. This Galactic+M31 blended component is included under the heading of ”Milky Way Absorption Lines” in Table 3 but with a footnote. Absorption due to C ii 1335.7 is among the many transitions detected in this component (see Table 3). A weak (barely significant) blended Galactic and M31 high-ion absorption component is located near km s. The 21 cm emission profile allows us to estimate that the Galactic component peaking near km s has atoms cm and the M31 disk component peaking near km s has atoms cm. Aside from this first blended absorption component, a second low-ion absorption component is seen displaced toward lower velocities by km s, close to the edge of the H i 21 cm emission profile, which we take as evidence for halo gas. However, measurements of the Si ii 1260, O i 1302, Si ii 1304, C ii 1334.5 (and C ii 1335.7), Si ii , C iv , and Al ii absorption lines are complicated by overlapping or nearby absorption. Deblending indicates that the M31 low-ion halo gas component is near km s, and this gives rise to absorption due to Si ii, O i, C ii, Al ii, Fe ii, and Mg ii. Deblending also indicates that a high-ion absorption component is located near km s; it is clearly present in C iv but possibly not Si iv. The 21 cm emission allows us to estimate that the M31 halo component peaking near km s has atoms cm. The velocity locations of the above described absorption components for the low ions and high ions are shown as vertical dot-dashed lines in the panels, including the lower left inset panel. In addition, the GBT 21 cm observations also reveal gas from two M31 HVCs near km s (between km s and km s), and near km s (between km s and km s). The total integrated column densities along the sightlines to these HVCs are atoms cm and atoms cm, respectively. We do not detect metal-line absorption near the velocities of these HVCs, so we have not used vertical lines to mark their velocity locations in Figure 12. Thus, using standard quasar absorption line jargon, we conclude that, given the estimated values for the four detected M31 velocity components, we have detected a DLA system (M31 disk gas), a sub-DLA system (M31 halo gas), and two Lyman limit systems (two M31 HVCs).
Finally, we point out that the blended low-ion absorption near km s in sightline 10 is the only system which reaches DLA H i column densities (i.e., atoms cm). As described above, it is due to a blend of Galactic gas and M31 disk gas. DLAs are the quasar absorption-line systems used to track the evolution of neutral gas in the Universe at low (Rao et al. 2006) and high (e.g., Noterdaeme et al. 2012) redshift. The strength of the Mg ii and Fe ii absorption lines in this component are consistent with criteria used in Mg ii-selected DLA searches (Rao et al. 2006).
Figure 12: Same as Figure 7, but for 0046+4220. FUV data obtained for this quasar are also shown.
|REW (Å)||REW (Å)||REW (Å)||REW (Å)||REW (Å)||REW (Å)||REW (Å)||REW (Å)||REW (Å)||REW (Å)bbIn sightline 10, Milky Way absorption lines are blended with M31 disk gas. See the description in §3.|
|M31 Absorption Lines|
|MgII2796||ccThe two measurements are M31 HVC and disk components, respectively. See Figure 9.|
|ccThe two measurements are M31 HVC and disk components, respectively. See Figure 9.|
|Milky Way Absorption Lines|
4 Summary and Discussion of Results for M31
4.1 Overview on the Detection of Low-Ion and High-Ion Absorption Lines
The detections of M31 gas presented in the previous section and reported in Tables 3 and 4 can be summarized as follows.
Low-ion Mg ii absorption due to M31 gas is detected along four of the 10 observed sightlines (5, 7, 9, and 10). These sightlines have impact parameters ranging between 17 and 32 kpc. We also detect other low-ion gas (e.g., due to Si ii, O i, C ii, Fe ii, or Al ii) along three of the four sightlines with Mg ii detections; sightline 5 was not observed in the FUV, where most of these transitions occur. In addition, we detect C ii absorption at M31 velocities along sightline 8 ( kpc). Sightline 6 ( kpc) is the only “inner” sightline which does not show evidence for M31 low-ion absorption ( Å); however, no FUV spectra were obtained along this sightline. Among these “inner” sightlines, except for the blended Galactic and M31 line in sightline 10, the Mg ii rest equivalent widths ranged between 0.34 and 0.71 Å, with the strongest detection being a two-component absorber with 0.30 and 0.41 Å. The four outer sightlines (1 through 4), with impact parameters 57 to 112 kpc, do not show Mg ii absorption down to rest equivalent upper limits ranging between 0.21 and 0.46 Å.
High-ion C iv absorption due to M31 gas is detected along three of six sightlines (8, 9, and 10) which have usable FUV spectra. These three detections are all in “inner” sightlines, with impact parameters ranging between 13 and 18 kpc, and rest equivalent widths ranging between 0.17 and 0.65 Å. Some Si iv absorption and low-ion absorption is also detected along these three “inner” sightlines. The three C iv non-detections are in outer sightlines (1, 3, and 4), with impact parameters ranging between 57 and 112 kpc, and with rest equivalent width upper limits ranging between 0.18 and 0.30 Å.
We should point out that many of the detections summarized above were near the limit of our sensitivity threshold, despite the fact that our rest equivalent width upper limits are typical of those in large optical quasar absorption-line surveys. Another concern is confusion from overlapping or nearby absorption, but we believe we have dealt with this appropriately.
Also, Rich et al. (private communication) has observed three sightlines in the halo of M31 with COS. They do not cover the Mg ii region, but detect C iv from M31 in some of these sightlines. There are other HST archival observations in the M31 halo, but these do not show any detections.
As elaborated further in §4.3, a clear picture does emerge. The absorption lines that arise in M31 gas are found to be relatively weak in comparison to those often identified in optical quasar absorption-line surveys, and even more so in comparison to absorption lines which arise in the ISM of the Milky Way Galaxy (e.g., Table 3). Moreover, none of the detected M31 absorption lines are found at large impact parameters. This could also be viewed as unexpected since the bulk of intervening low- to moderate-redshift metal-line absorbers seen in quasar spectra are identified with large-impact-parameter galaxies in followup imaging studies (e.g., Rao et al. 2011, Chen et al. 2010). However, all of the large-impact-parameter sightlines we observed were generally along M31’s major axis, so one scenario which might explain the lack of absorption in those cases would be to hypothesize that extended gaseous absorption originates in galactic fountains and preferentially avoids extended regions along the direction of the disk (e.g., Bordoloi et al. 2011, Bouché et al. 2012). Using the observed distribution of HVCs around the Milky Way and M31, Richter (2012) finds an exponential decline in the mean filling factor of HVCs with a characteristic radial extent of kpc. If HVCs alone are responsible for absorption lines, then one would not expect to find any absorption along our four outer sightlines. Alternatively, M31 may simply be typical of a class of luminous galaxies that don’t possess large gaseous cross sections which are capable of giving rise to moderate-strength quasar absorption lines. Our findings for M31 may in some way be connected to the observed relative decrease in the incidence of stronger Mg ii systems with decreasing redshift (e.g., Nestor et al. 2005).
In the past several years there has been speculation that M31 is a galaxy that lies in the “green valley¡Ç¡Ç (e.g., Mutch et al. 2011, Davidge et al. 2012). The idea is that it exhibits properties that put it between the red cloud and blue cloud populations that have been identified in large galaxy surveys. Such galaxies may be in a stage of transition and their star formation may nearly cease in less than 5 Gyrs. While this may be the case for M31, we note that the data we have discussed here should not be taken to offer any clues about this. For example, our data do not allow us to draw any conclusions about the strength of star formation or even the column densities of metal-line absorption. This is because the lines we have identified are likely to be mostly saturated. Thus, the weakness of the metal-line absorption in M31 most likely indicates that the effective gas velocity spread is low; it may either be truly low relative to the spectral resolution and/or there may be a small number of velocity components within the spectral resolution element.
4.3 Mg ii Rest Equivalent Width () versus Impact Parameter ()
Figure 13 is a plot of M31 Mg ii rest equivalent width () detections (or upper limits) versus sightline impact parameter (). The measurement shown for sightline 7, which has kpc, was made by fitting a single Gaussian to both absorption components reported in Table 3, i.e., it is not a simple sum of the results from the two individual Gaussian fits reported in Table 3. Since the impact parameters of sightlines 9 and 10 are very similar, they are displaced from each other in the figure for clarity. Note that the upper limits are upper limits, while the error bars are the uncertainties. The four outermost data points are suggestive of an overall decrease of with increasing impact parameter. Quasar absorption line studies of large samples of absorber-galaxy pairs have shown this to be true as well (Chen et al. 2010; Rao et al. 2011).
For comparison, Figure 14 includes results from the Rao et al. (2011) sample of absorbing galaxies which have been identified for Mg ii-selected DLAs, subDLAs, and Lyman limit systems (LLSs). The mean redshift of the Rao et al. sample is , with redshifts in the range . The identified absorbing galaxies in the Rao et al. sample also have a range of luminosities, mostly , but there is not a significant correlation between luminosity and impact parameter. Rao et al. found only a marginal (1.8) correlation between and . The solid black circles in Figure 14 are DLAs and the open circles are subDLAs and LLSs. The data from this current M31 study are in red. Sightlines 5 through 10 have averaged integrated 21 cm emission H i column densities in the subDLA regime, with the exception of the Galactic and M31 blended component along sightline 10. (See §3.) H i 21 cm emission maps are not available as far out as the four outermost sightlines, but since the cm edge of the H i disk of M31 is at kpc (Figure 1), these sightlines are not expected to have averaged integrated H i column densities in the DLA or subDLA regime.
Thus, as noted in §4.1 and §4.2, it is clear that the sightlines passing near M31, or through its gaseous disk seen in 21 cm emission, do not give rise to the moderate-to-strong Mg ii absorption lines which are often identified in moderate- to high-redshift quasar absorption-line surveys. For comparison, all of the Galactic detections reported in Table 3 have Å, and 4 of the Galactic sightlines have Å (sightline 10 is a blend of Galactic and M31 gas). In the HST Key Project sample of Galactic sightlines (Savage et al. 2000) the median value is Å, and the strongest line has Å.
Of course, our sightlines through M31 are biased sightlines in the context of traditional absorption line surveys, in that the galaxy was pre-selected in order to study the properties of its low-ion and high-ion gas. Therefore, for M31 the probability of occurrence of Mg ii absorption as a function of is not properly estimated from the observed incidence of Mg ii absorption in unbiased quasar absorption-line surveys. Instead, however, this experiment does show that a gas-rich, , spiral galaxy like M31 need not give rise to moderate-to-strong Mg ii absorption along sightlines which pass through its H i 21 cm emission disk, or even through a putative extended gaseous halo.
4.4 Comparison of 21 cm Emission and Absorption-Line Velocities
The range of velocities that exhibit 21 cm emission for sightlines 5 through 10 are shown as cyan and orange vertical bars as a function of impact parameter in Figure 15. Cyan bars correspond to 21 cm emission velocities from M31 gas and orange bars represent HVC velocities. Also plotted are the velocities of the low-ion (red stars) and high-ion (blue triangles) absorption lines from Table 4. The Galactic and M31 blended low-ion absorption line along sightline 10 is shown as the encircled star. For the two inner disk sightlines (9 and 10 at kpc), it appears that the velocity of the high-ion absorption is better correlated with the 21 cm emission velocity range. Along sightline 8, the low and high-ion absorption lie at an outer velocity edge of where 21 cm emission is detected. As noted in §3, this velocity corresponds to the peak in 21 cm emission along this sightline. For sightlines 7 (at kpc) and 5 (at kpc), the low-ion gas again coincides with the peak of 21 cm emission which is near near the edge of the 21 cm profile (see Figures 7 and 9). Thus, in nearly all cases, the low ions occur near the edge of the 21 cm profiles (two are near the low velocity edge and three are near the high velocity edge), and for sightlines 5, 7, and 8, are coincident with the peak in 21 cm emission.
|Low ion||High ion||Low ion||High ion|
|(km s)||(km s)||(km s)||(km s)|
|7. 0040+3915||,bbThe two measurements are M31 HVC and halo components, respectively. See Figure 9.|
|10. 0046+4220||ccThe two measurements are M31 halo and disk components, respectively. The disk component is blended with the Milky Way line.||ccThe two measurements are M31 halo and disk components, respectively. The disk component is blended with the Milky Way line.|
Given the resolution of the NUV and FUV data ( km s at Å and km s at Å), one might question if these differences are significant. However, it is well-known that in data with sufficient signal-to-noise, a Gaussian fit to an absorption line can be used to determine the centroid location of the line to an accuracy much better than the line’s FWHM. In order to determine how accurately absorption-line locations can be determined, we ran 10,000 realizations of lines with equivalent widths drawn from the data. Figure 16 shows the distributions of equivalent widths. Noise was added to the Gaussian profiles generated with these equivalent widths so that the resulting signal-to-noise ratios matched the data. Line centroids were then estimated by refitting Gaussian profiles to the noised-up absorption lines. The resulting distributions of centroid velocities relative to the input values are shown in Figure 17. For both the original as well as the simulated data, the spectra were rebinned to two pixels per resolution element before measurements were made. The signal-to-noise ratios of NUV spectra were, in general, higher than in FUV spectra. Thus, the accuracy with which the line centroids can be measured is higher for the Mg ii lines. Specifically, the centroid standard deviation of the Mg ii distribution is km s compared to km s for C iv. These uncertainties indicate that the separations in velocities of the low and high ions are significant towards sightlines 9 and 10 at approximately the level.
The 21 cm emission studies of M31 (e.g., §2.1) show that for this nearly edge-on galaxy, the sightline velocities of gas giving rise to 21 cm emission can span a large range (e.g., see the lower panels in Figures 7 - 12). Corbelli et al. (2010) fitted a tilted ring model to M31’s H i 21 cm emission data from 8 to 37 kpc to study the details of its rotation, finding that M31’s disk warps beyond galactocentric distances of kpc and that it becomes more inclined with respect to our sightline. As we have shown above, the Mg ii absorption regions are almost always at the peak of the 21 cm emission profile, which occurs near the edge of the 21 cm emission velocity range. Thus, when detected, the low-ion gas appears to trace the 21 cm gas. Interestingly, neither low- nor high-ion absorption lines are detected at the 21 cm velocity locations of the HVCs along sightline 10 ( kpc). Low-ion absorption is also not detected at the 21 cm disk velocity location towards sightline 6 ( kpc). However, the observed low-ion absorption along sightline 7 originates in the HVC detected in 21 cm emission, but at the velocity location of the other absorption component, there is no detected 21 cm emission. This component, at km s, is likely to be M31 halo gas. Thus, it appears that the sightlines through M31 are passing through very different physical and kinematic conditions within its ISM.
A conventional study relating quasar absorption-lines to the galaxies that cause them begins with the detection of an intervening absorption-line system in a spectrum followed by imaging work to identify the galaxy. The experiment with M31 described here is a quasar absorption-line survey conducted in reverse. We probed ten sightlines with vastly different impact parameters through a single spiral galaxy with a luminosity of . As summarized in §4.1, we detected some type of absorption from M31 gas in five of the six inner sightlines ( kpc), but no absorption in any of the four outer sightlines ( kpc). We also reported the first detection of metals in a M31 HVC.
In §4.3 we compared our M31 results to the findings in the conventional Rao et al. (2011) survey. Rao et al. found only a marginal anticorrelation between and , and indeed we find the same qualitative trend in M31, but the values of are far smaller in M31 (Figure 14). And while Rao et al. found that there were fewer systems with moderate-to-strong at large- ( kpc), we found none arising in M31. In §4.4 we compared the velocity locations of low-ion and high-ion gas in M31 to that of M31’s 21 cm emission and found that the high-ion gas is better aligned with the velocities of observed 21 cm emission along two of three sightlines where it is detected. The velocity of the low-ion gas is correlated with the peak of 21 cm emission and is often near the edge of the 21 cm emission velocity range. In one case Mg ii is detected at a velocity location that shows no 21 cm emission.
Broadly, our results indicate that:
Despite the fact that M31 is a gas-rich, spiral galaxy, it produces relatively weak Mg ii and C iv absorption lines in comparison to those found in moderate-to-high redshift quasar absorption-line surveys. For Mg ii, this may indicate that M31 is typical of a class of luminous galaxies that don’t possess gaseous cross sections capable of giving rise to moderate-strength quasar absorption lines even at impact parameters kpc. This finding might also be related to the observed relative decrease in the incidence of stronger Mg ii systems with decreasing redshift.
M31 appears not to possess an extensive large gaseous cross section at impact parameters kpc that is capable of giving rise to moderate-strength quasar absorption lines (e.g., with Å or Å), at least not along the direction of its major axis.
For the relatively weak absorption that we did detect at kpc, we found the low-ion gas to be associated with the peak in the 21 cm emission profile, near one edge of the 21 cm emission velocity range. Two of three sightlines showed high-ion gas to be centrally located within the 21 cm emission profile, with the third being coincident with an edge. It is also likely that we have detected low-ion halo gas through two of the sightlines.
Future UV spectroscopy of quasars behind M31 can build on these findings by: (1) acquiring higher signal-to-noise data to probe down to weaker rest equivalent width values, (2) acquiring higher resolution data to better study the velocity locations of the gas relative to 21 cm emission velocities, and/or (3) probing a larger number of sightlines including ones in M31’s extended halo region.
It would be interesting if results derived from M31’s 21 cm emission data could be compared with determinations from Lyman series absorption seen in the UV spectra of background quasars. One could then get an H i column density measurement averaged over less than a milli-parsec region in M31, in comparison to the 50 pc linear spatial scale offered by the radio observations. This would provide information on the homogeneity and size scale of H i absorbing regions in M31.
We are grateful for the referee’s encouraging remarks and comments which improved and clarified the paper. SMR, DAT, RW, DT, and DVB acknowledge support from HST grant GO-11658. GMS acknowledges support from a Zaccheus Daniel Fellowship and a Dietrich School of Arts and Sciences Graduate Fellowship from the University of Pittsburgh. We are grateful for the help and support provided by the NOAO staff. We thank Michéle Belfort-Mihalyi for assisting on the NOAO observing run.
- [Abraham et al.(2012)] Abraham, S., Philip, N. S., Kembhavi, A., Wadadekar, Y. G., & Sinha, R. 2012, MNRAS, 419, 80
- [Adelman-McCarthy et al.(2006)] Adelman-McCarthy, J. K., et al. 2006, ApJS, 162, 38
- [Bergeron & Boissé(1991)] Bergeron, J., & Boisse, P. 1991, A&A, 243, 344
- [Bordoloi(2011)] Bordoloi, R. et al. 2011, ApJ, 743, 10
- [Bouche(2012)] Bouché, N., Hohensee, W., Vargas, R., Kacprzak, G. G., Martin, C. L., Cooke, J., & Churchill, C. W. 2012, MNRAS, 426, 801
- [Braun et al.(2009)] Braun, R., Thilker, D. A., Walterbos, R. A, M., & Corbelli E. 2009, ApJ, 695, 937
- [Chen et al.(2010)] Chen, H.-W., Helsby, J. E., Gauthier, J.-R., Schectman, S. A., Thompson, I. B., & Tinker, J. L. 2010, ApJ, 714, 1521
- [Cool et al.(2012)] Cool, R. J., et al. 2012, ApJ, 748, 10
- [Corbelli et al.(2010)] Corbelli, E., Lorenzoni, S., Walterbos, R., Braun, R., & Thilker, D. 2010, A&A, 511, 89
- [Davidge et al. (2012)] Davidge, T., et al. 2012, ApJ, 751, 74
- [de Vaucoulers et al.(1991)] de Vaucouleurs, G., de Vaucouleurs, A., Corwin, H. G., Jr., Buta, R. J., Paturel, G., & Fouqué, P. 1991, Third Reference Catalogue of Bright Galaxies (New York: Springer)
- [Evans et al.(2010)] Evans, I. N., et al. 2010, ApJS, 189, 37
- [Fardal et al.(2006)] Fardal, M. A., Babul, A., Geehan, J. J., & Guhathakurta, R. 2006, MNRAS, 366, 1012
- [Lanzetta, Turnshek & Wolfe(1987)] Lanzetta, K., Turnshek, D., & Wolfe, A. 1987, ApJ, 322, 739
- [Lanzetta & Bowen(1990)] Lanzetta, K., & Bowen, D. 1990, ApJ, 357, 321
- [Mutch et al.(2011)] Mutch, S. J., Croton, D. J., & Poole, G. B. 2011, ApJ, 736, 84
- [Nestor, Turnshek, & Rao(2005)] Nestor, D., Turnshek, D., & Rao, S. 2005, ApJ, 628, 637
- [Nestor, Turnshek, & Rao(2006)] Nestor, D., Turnshek, D., & Rao, S. 2006, ApJ, 643, 75
- [Richards et al.(2009)] Richards, G. T., Myers, A. D., Gray, A. G., Riegel, R. N., Nichol, R. C., Brunner, R. J., Szalay, A. S., Schneider, D. P., & Anderson, S. F. 2009, ApJS, 180, 67
- [Rao, Turnshek, & Nestor(2006)] Rao, S., Turnshek, D., & Nestor, D. 2006, ApJ, 636, 610
- [Rao et al.(2011)] Rao, S. M., Belfort-Mihalyi, M., Turnshek, D. A., Monier, E. M., Nestor, D. B., & Quider, A. M. 2011, MNRAS, 416, 1215
- [Richter(2012)] Richter, P. 2012, ApJ, 750, 165
- [Riess, Fliri, & Valls-Gabaud(2011)] Riess, A., Fliri, J., & Valls-Gabaud, D. 2012, ApJ, 745, 156
- [Savage et al.(2000)] Savage, B. D. et al. 2000, ApJS, 129, 563
- [Steidel(1993)] Steidel, C. 1993, The Environment and Evolution of Galaxies, Astrophysics and Space Sciences Library Vol 188 (Kluwer), eds. M. Shull and H. Thronson, p263
- [Tamm et al.(2012)] Tamm, A., Tempel, E., Tenjes, P., Tihhonova, O., & Tuvikene, T. 2012, A&A, 546, 4
- [Thilker et al.(2004)] Thilker, D., Braun, R., Walterbos, R., Corbelli, E., Lockman, F., Murphy, E., & Maddalena, R. 2004, ApJ, 601, L39
- [Tollerud et al.(2012)] Tollerud, E. J., Beaton, R. L., Geha, M. C., et al. 2012, ApJ, 752, 45
- [Westmeier et al.(2005)] Westmeier, T., Braun, R., & Thilker, D. 2005, A&A, 436, 101
- [Weyman et al.(1998)] Weymann, R. J., et al. 1998, ApJ, 506, 1