# Recovering the systemic redshift of galaxies from their Lyman alpha line profile

Recovering the systemic redshift of galaxies from their Lyman alpha line profile Abstract The Lyman alpha (Ly α) line of Hydrogen is a prominent feature in the spectra of star-forming galaxies, usually redshifted by a few hundreds of km s−1 compared to the systemic redshift. This large offset hampers follow-up surveys, galaxy pair statistics, and correlations with quasar absorption lines when only Ly α is available. We propose diagnostics that can be used to recover the systemic redshift directly from the properties of the Ly α line profile. We use spectroscopic observations of Ly α emitters for which a precise measurement of the systemic redshift is available. Our sample contains 13 sources detected between z ≈ 3 and z ≈ 6 as part of various multi-unit spectroscopic explorer guaranteed time observations. We also include a compilation of spectroscopic Ly α data from the literature spanning a wide redshift range (z ≈ 0–8). First, restricting our analysis to double-peaked Ly α spectra, we find a tight correlation between the velocity offset of the red peak with respect to the systemic redshift, $$V_{\rm peak}^{\rm red}$$, and the separation of the peaks. Secondly, we find a correlation between $$V_{\rm peak}^{\rm red}$$ and the full width at half-maximum of the Ly α line. Fitting formulas to estimate systemic redshifts of galaxies with an accuracy of ≤100 km s−1, when only the Ly α emission line is available, are given for the two methods. galaxies: high-redshift, galaxies: starburst, galaxies: statistics, ultraviolet: galaxies 1 INTRODUCTION In the last few decades, large samples of high-redshift galaxies (z > 2) have been assembled from deep photometric surveys based on broad/narrow-band selection techniques (Steidel et al. 2003; Ouchi et al. 2008; Bouwens et al. 2015; Finkelstein et al. 2015; Sobral et al. 2017, and references therein). In parallel, blind spectroscopic searches commonly rely on the Lyman alpha (Ly α) line redshifted to the optical or the near-infrared to identify or confirm sources at z ≥ 2 (e.g. Blanc et al. 2011; Bielby et al. 2011; Le Fèvre et al. 2015). The number of spectroscopic detections of Ly α emitters (LAE) is now being increased dramatically with ongoing observational campaigns with the multi-unit spectroscopic explorer (MUSE, Bacon et al. 2010) on European Southern Obseratory's Very Large Telescope (ESO VLT), allowing us to study galaxy formation and evolution with a homogeneous sample of sources over a large redshift range (2.8 ≲ z ≲ 6.7, e.g. Bacon et al. 2015, 2017; Drake et al. 2016; Herenz et al. 2017; Mahler et al. 2017; Caruana et al. 2018). Several studies have demonstrated that the Ly α emission line is not exactly tracing systemic redshift (e.g. Shapley et al. 2003; McLinden et al. 2011; Rakic et al. 2011; Song et al. 2014; Hashimoto et al. 2015). Instead, the line profiles often show a complex structure that arguably originates from the propagation of resonant Ly α photons in neutral gas within the interstellar medium and/or in the vicinity of galaxies. Among the broad diversity of Ly α profiles in Ly α emitting galaxies, we identify the most common two categories: (i) spectra with a redshifted single peak [∼2/3 of Ly α emitting Lyman Break Galaxies, called LBGs, from Kulas et al. (2012)], and (ii) double-peaked profiles, with a prominent red peak and a smaller blue bump [∼2/3 of the remaining 1/3 of Ly α emitting LBGs that are multiple peaked, from Kulas et al. (2012); 40 per cent of the LAEs observed by Yamada et al. (2012)]. We will refer to the latter as blue bump LAEs in the remainder of this paper. Understanding the nature of blue bump LAEs and studying their occurrence and evolution with redshift will be the goal of a forthcoming study. The vast majority of objects display a red peak shifted by a variable amount peaking around ∼400  km s−1 for LBGs (e.g. Shapley et al. 2003; Kulas et al. 2012), ∼200 km s−1 for LAEs (e.g. Hashimoto et al. 2013; Erb et al. 2014; Song et al. 2014; Hashimoto et al. 2015; Henry et al. 2015; Trainor et al. 2015), and less than ∼150 km s−1 for a small sample of five local Lyman Continuum Emitters (Verhamme et al. 2017). If not accounted for, this offset with respect to the systemic redshift can be problematic when addressing astrophysical issues which require accurate systemic redshift measurements (e.g. galaxy interactions, gas kinematics, baryonic acoustic oscillations, intergalactic medium-galaxy emission/absorption correlations). The scope of this paper is to investigate whether the Ly α profile shape can be used to determine the systemic redshift of galaxies. The outline of this Letter is as follows: in Section2, we gather recent spectroscopic data from MUSE Guaranteed Time Observations (GTO) surveys and from the literature that have sufficient spectral resolution to investigate the Ly α line properties, as well as reliable systemic redshift measurements. In Section 3, we present two diagnostics that can be used to recover the systemic redshift from Ly α, which we compare to models in Section 4. Section 5 summarizes our findings. 2 A SAMPLE OF LAES WITH KNOWN SYSTEMIC REDSHIFT In order to investigate the link between the shape of the Ly α line and the systemic redshift, we collect a diverse sample of LAEs with a precise measure of the systemic redshift. Our sample consists of high-redshift (z > 2) LAEs with detected C iii]λλ1907, 1909, [O iii]λλ4959, 5007 or H α λ6563 emission, and low redshift (z < 0.4) LAEs with Ly α observations in the UV rest-frame obtained with the Cosmic Origins Spectrograph onboard HST, and ancillary optical spectra from the sDSS data base containing several nebular emission lines from which the redshift is determined with great accuracy. We present these data in the following paragraphs. For each Ly α spectrum, we measure $$V_{\rm peak}^{\rm red}$$ as the location of the maximum of the Ly α flux redwards of the systemic redshift, and full width at half-maximum (FWHM) as the width of the part of the spectrum uncorrected for instrumental broadening with flux above half of the maximum, both directly on the data, without any modelling. 2.1 LAEs from MUSE GTO data Stark et al. (2014) reported the detection of C iii]λλ1907, 1909 emission from low mass star-forming galaxies. When observed, this doublet is the strongest UV emission line after Ly α, and, in contrast to Ly α, it is an optically thin nebular line, tracing the systemic redshift1 of the Ly α production site. The redshift window where Ly α and C iii] are both observable within the VLT/MUSE spectral range is 2.9 < z < 3.8. MUSE is an optical Integral Field Unit (IFU) spectrograph with medium spectral resolution (from R ∼ 2000 in the blue to R ∼ 4000 in the red). Within several projects in the MUSE consortium using GTO (Bacon et al. 2017; Brinchmann et al. 2017; Herenz et al. 2017; Inami et al. 2017; Mahler et al. 2017; Maseda et al. 2017; Caruana et al. 2018) and Science Verification (SV) or commissioning data (Patrício et al. 2016), we find 13 LAEs with reliable C iii] detections, that is, non-contaminated by sky lines and with a S/N > 3. We list these objects in Table 1 (see Maseda et al. 2017, for a systematic study of ciii emitters in the MUSE GTO data from the Hubble Ultra Deep Field). For each of these LAEs we measure the shift of the Ly α emission compared to C iii], $$V_{\rm peak}^{\rm red}$$, the observed FWHM, and the separation of the Ly α peaks for the eight blue bump LAEs among them. Table 1. MUSE Ly α+C iii] emitters. The sixth column indicates the separation of the peaks (i.e. 2 × ΔV1/2, in km s−1) for blue bump LAEs, and is left empty for single-peaked profiles. ID RA DEC EW (Å) $$V_{\rm peak}^{\rm red}$$(km s−1) FWHM (km s−1) ΔV (km s−1) zsys, CIII] Observations sys 1a 307.97040 −40.625694 32 176 ± 11 248 ± 9 – 3.5062 Commissioning mul 11b 342.175042 −44.541031 222 215 ± 35 150 ± 35 375 ± 35 3.1163 AS1063 mul 14b 342.178833 −44.535869 29 385 ± 35 300 ± 35 – 3.1150 AS1063 sys 44b 64.0415559 −24.0599916 57 303 ± 35 360 ± 35 570 ± 35 3.2886 MACS0416 sys 132b 64.0400838 −24.0667408 62 331 ± 35 288 ± 35 510 ± 35 3.2882 MACS0416 106c 53.163726 −27.7790755 72 379 ± 13 414 ± 13 828 ± 35 3.2767 udf-10 118c 53.157088 −27.7802688 65 301 ± 28 284 ± 28 568 ± 35 3.0173 udf-10 1180c 53.195735 −27.7827171 80 220 ± 23 348 ± 32 – 3.3228 udf mosaic 6298c 53.169249 −27.7812550 83 582 ± 38 512 ± 56 – 3.1287 udf-10 6666c 53.159576 −27.7767193 52 284 ± 13 377 ± 11 754 ± 35 3.4349 udf-10 50d 150.149656 2.061272 50 431 ± 42 268 ± 39 – 3.8237 GR30 48d 149.852989 2.488099 68 294 ± 35 214 ± 35 705 ± 35. 3.3280 GR34 102d 150.050268 2.600025 76 299 ± 15 229 ± 15 385 ± 35. 3.0400 GR84 ID RA DEC EW (Å) $$V_{\rm peak}^{\rm red}$$(km s−1) FWHM (km s−1) ΔV (km s−1) zsys, CIII] Observations sys 1a 307.97040 −40.625694 32 176 ± 11 248 ± 9 – 3.5062 Commissioning mul 11b 342.175042 −44.541031 222 215 ± 35 150 ± 35 375 ± 35 3.1163 AS1063 mul 14b 342.178833 −44.535869 29 385 ± 35 300 ± 35 – 3.1150 AS1063 sys 44b 64.0415559 −24.0599916 57 303 ± 35 360 ± 35 570 ± 35 3.2886 MACS0416 sys 132b 64.0400838 −24.0667408 62 331 ± 35 288 ± 35 510 ± 35 3.2882 MACS0416 106c 53.163726 −27.7790755 72 379 ± 13 414 ± 13 828 ± 35 3.2767 udf-10 118c 53.157088 −27.7802688 65 301 ± 28 284 ± 28 568 ± 35 3.0173 udf-10 1180c 53.195735 −27.7827171 80 220 ± 23 348 ± 32 – 3.3228 udf mosaic 6298c 53.169249 −27.7812550 83 582 ± 38 512 ± 56 – 3.1287 udf-10 6666c 53.159576 −27.7767193 52 284 ± 13 377 ± 11 754 ± 35 3.4349 udf-10 50d 150.149656 2.061272 50 431 ± 42 268 ± 39 – 3.8237 GR30 48d 149.852989 2.488099 68 294 ± 35 214 ± 35 705 ± 35. 3.3280 GR34 102d 150.050268 2.600025 76 299 ± 15 229 ± 15 385 ± 35. 3.0400 GR84 aPatricio et al. 2016; bRichard et al., in preparation; cBacon et al. 2017, Inami et al. 2017, Maseda et al. 2017; dContini et al., in preparation. View Large Table 1. MUSE Ly α+C iii] emitters. The sixth column indicates the separation of the peaks (i.e. 2 × ΔV1/2, in km s−1) for blue bump LAEs, and is left empty for single-peaked profiles. ID RA DEC EW (Å) $$V_{\rm peak}^{\rm red}$$(km s−1) FWHM (km s−1) ΔV (km s−1) zsys, CIII] Observations sys 1a 307.97040 −40.625694 32 176 ± 11 248 ± 9 – 3.5062 Commissioning mul 11b 342.175042 −44.541031 222 215 ± 35 150 ± 35 375 ± 35 3.1163 AS1063 mul 14b 342.178833 −44.535869 29 385 ± 35 300 ± 35 – 3.1150 AS1063 sys 44b 64.0415559 −24.0599916 57 303 ± 35 360 ± 35 570 ± 35 3.2886 MACS0416 sys 132b 64.0400838 −24.0667408 62 331 ± 35 288 ± 35 510 ± 35 3.2882 MACS0416 106c 53.163726 −27.7790755 72 379 ± 13 414 ± 13 828 ± 35 3.2767 udf-10 118c 53.157088 −27.7802688 65 301 ± 28 284 ± 28 568 ± 35 3.0173 udf-10 1180c 53.195735 −27.7827171 80 220 ± 23 348 ± 32 – 3.3228 udf mosaic 6298c 53.169249 −27.7812550 83 582 ± 38 512 ± 56 – 3.1287 udf-10 6666c 53.159576 −27.7767193 52 284 ± 13 377 ± 11 754 ± 35 3.4349 udf-10 50d 150.149656 2.061272 50 431 ± 42 268 ± 39 – 3.8237 GR30 48d 149.852989 2.488099 68 294 ± 35 214 ± 35 705 ± 35. 3.3280 GR34 102d 150.050268 2.600025 76 299 ± 15 229 ± 15 385 ± 35. 3.0400 GR84 ID RA DEC EW (Å) $$V_{\rm peak}^{\rm red}$$(km s−1) FWHM (km s−1) ΔV (km s−1) zsys, CIII] Observations sys 1a 307.97040 −40.625694 32 176 ± 11 248 ± 9 – 3.5062 Commissioning mul 11b 342.175042 −44.541031 222 215 ± 35 150 ± 35 375 ± 35 3.1163 AS1063 mul 14b 342.178833 −44.535869 29 385 ± 35 300 ± 35 – 3.1150 AS1063 sys 44b 64.0415559 −24.0599916 57 303 ± 35 360 ± 35 570 ± 35 3.2886 MACS0416 sys 132b 64.0400838 −24.0667408 62 331 ± 35 288 ± 35 510 ± 35 3.2882 MACS0416 106c 53.163726 −27.7790755 72 379 ± 13 414 ± 13 828 ± 35 3.2767 udf-10 118c 53.157088 −27.7802688 65 301 ± 28 284 ± 28 568 ± 35 3.0173 udf-10 1180c 53.195735 −27.7827171 80 220 ± 23 348 ± 32 – 3.3228 udf mosaic 6298c 53.169249 −27.7812550 83 582 ± 38 512 ± 56 – 3.1287 udf-10 6666c 53.159576 −27.7767193 52 284 ± 13 377 ± 11 754 ± 35 3.4349 udf-10 50d 150.149656 2.061272 50 431 ± 42 268 ± 39 – 3.8237 GR30 48d 149.852989 2.488099 68 294 ± 35 214 ± 35 705 ± 35. 3.3280 GR34 102d 150.050268 2.600025 76 299 ± 15 229 ± 15 385 ± 35. 3.0400 GR84 aPatricio et al. 2016; bRichard et al., in preparation; cBacon et al. 2017, Inami et al. 2017, Maseda et al. 2017; dContini et al., in preparation. View Large 2.2 High-z data from the literature Stark et al. (2017) reported the detection of C iii] from one of the highest redshift Ly α emitters ever observed (z ∼ 7.730) with $$V_{\rm peak}^{\rm red}$$ ∼ 340 km s−1; the Ly α FWHM $$=360^{+90}_{-70}$$ km s−1 is measured by Oesch et al. (2015). Vanzella et al. (2016) report a narrow Ly α line observed at medium spectral resolution using VLT-Xshooter of a magnified star-forming galaxy at z = 3.1169, with $$V_{\rm peak}^{\rm red}$$(Ly α) ∼ 100 km s−1 and FWHM(Ly  α) ∼104 km s−1. From Hashimoto et al. (2015, 2017), we select the six LAEs observed with MagE (R ∼ 4100). Their systemic redshifts have been obtained with either H α or [O iii] lines. Three of these objects are blue-bump LAEs, for which we also measure the separation of the peaks. Kulas et al. (2012) reported that a significant fraction (∼30  per cent) of their Ly α emitting LBGs show a complex Ly α profile, with at least one secondary peak. Blue bump objects (their Group I) represent the majority of their profiles (11 out of 18 objects). We add these 11 objects to our sample of blue bumps LAEs. 2.3 Low-z data from the literature Green Pea galaxies (hereafter GPs) are LAEs in the local Universe (z ∼ 0.1 to 0.3; Jaskot & Oey 2014; Henry et al. 2015; Verhamme et al. 2017; Yang et al. 2017). The systemic redshift of these objects was compiled from the several nebular lines contained in their SDSS optical spectrum (e.g Izotov, Guseva & Thuan 2011). Note that the C iii] emission line is out of the UV spectral range probed by the available HST-COS observations. For a sample of 17 GPs from Jaskot & Oey (2014), Henry et al. (2015); Verhamme et al. (2017), we measure $$V_{\rm peak}^{\rm red}$$ and FWHM on the data. For 21 new GP observations, we use the $$V_{\rm peak}^{\rm red}$$ and FWHM values computed by Yang et al. (2017) given in their table 2. GPs nearly always exhibit blue bump Ly α profiles (Jaskot & Oey 2014; Henry et al. 2015; Verhamme et al. 2017). For the blue bump GPs, we also measure the separation of the peaks. 3 DERIVING SYSTEMIC REDSHIFT FROM LYMAN-ALPHA 3.1 Method 1: systemic redshift of blue bump LAEs In this section, only blue bump spectra, i.e. double peaks with a red peak higher than the blue peak, are considered. We note that, for all blue bump LAEs studied here, the systemic redshift always falls in-between the Ly α peaks, as illustrated in Fig. 1 for three blue-bump MUSE Ly α+C iii] emitters (see also Kulas et al. 2012; Erb et al. 2014; Yang et al. 2016). Fig. 2, left-hand panel, shows a positive empirical correlation between $$V_{\rm peak}^{\rm red}$$ and half of the separation of the peaks, ΔV1/2, for blue bump LAEs with known systemic redshift. We fit the data using the LTS_LINEFIT program described in Cappellari et al. (2013), which combines the Least Trimmed Squares robust technique of Rousseeuw & van Driessen (2006) into a least-squares fitting algorithm which allows for errors in both variables and intrinsic scatter.2 The best fit, shown by the red line in Fig. 2, is given by \begin{equation*} V_{\mathrm{peak}}^{\mathrm{red}}= 1.05 (\pm 0.11) \times \Delta V_{1/2} - 12 (\pm 37) \,\,{\rm km\,\, s}^{-1} \end{equation*} (1) This relation is so close to the one-to-one relation that we assume from now that the underlying ‘true’ relation between $$V_{\rm peak}^{\rm red}$$ and ΔV1/2 is one-to-one, as expected from radiation transfer modelling (see Section 4 below). The intrinsic scatter estimated from the linear regression is 53 (± 9) km s−1. Figure 1. View largeDownload slide Three examples of rest-frame spectra of LAEs with detected C iii]λλ1907, 1909 doublet probing the systemic redshift in the MUSE-GTO observations. In each panel, the velocity shifts of the Ly α line are shown relative to C iii]1907Å. For all blue bump LAEs in our sample, the systemic redshift falls in between the blue and red peaks of the Ly α emission. Figure 1. View largeDownload slide Three examples of rest-frame spectra of LAEs with detected C iii]λλ1907, 1909 doublet probing the systemic redshift in the MUSE-GTO observations. In each panel, the velocity shifts of the Ly α line are shown relative to C iii]1907Å. For all blue bump LAEs in our sample, the systemic redshift falls in between the blue and red peaks of the Ly α emission. Figure 2. View largeDownload slide Empirical relations to determine systemic redshift from the shape of the Ly α emission. Left: correlation between the shift of the Ly α red peak, ($$V_{\rm peak}^{\rm red}$$) and half of the separation of the peaks (ΔV1/2) for a sample of LAEs with a known systemic redshift: 7 Ly α+C iii] emitters with blue bump Ly α spectra from the MUSE GTO data (red stars), blue bump LAEs among the Yang et al. (2017) GP sample (black dots), blue bump LAEs among the Hashimoto et al. (2015) MagE sample and Group I LBGs from Kulas et al. (2012) (black triangles). Right: correlation between $$V_{\rm peak}^{\rm red}$$ and FWHM among Ly α+ C iii], H α or [O iii] emitters. The black dashed line is the one-to-one relation. We checked that the correlation remains even discarding the two most upper left points. On both sides, the red curve is our best fit to the data, described by equations (1) and (2). The Pearson coefficient and the probability of the null hypothesis are shown on each panel. Figure 2. View largeDownload slide Empirical relations to determine systemic redshift from the shape of the Ly α emission. Left: correlation between the shift of the Ly α red peak, ($$V_{\rm peak}^{\rm red}$$) and half of the separation of the peaks (ΔV1/2) for a sample of LAEs with a known systemic redshift: 7 Ly α+C iii] emitters with blue bump Ly α spectra from the MUSE GTO data (red stars), blue bump LAEs among the Yang et al. (2017) GP sample (black dots), blue bump LAEs among the Hashimoto et al. (2015) MagE sample and Group I LBGs from Kulas et al. (2012) (black triangles). Right: correlation between $$V_{\rm peak}^{\rm red}$$ and FWHM among Ly α+ C iii], H α or [O iii] emitters. The black dashed line is the one-to-one relation. We checked that the correlation remains even discarding the two most upper left points. On both sides, the red curve is our best fit to the data, described by equations (1) and (2). The Pearson coefficient and the probability of the null hypothesis are shown on each panel. 3.2 Method 2: an empirical correlation between FWHM and systemic redshift In this section, both single- and double-peaked profiles are considered. The measurements are always done on the red peak, and the red peak only. In the right-hand panel of Fig. 2 we plot $$V_{\rm peak}^{\rm red}$$ versus FWHM for the full sample of LAEs presented in Section 2 (new MUSE LAEs measurements are reported in Table 1). There is a correlation between $$V_{\rm peak}^{\rm red}$$ and FWHM, although less significant than for Method 1 (see the Pearson coefficients on each panel of Fig. 2). We use the same method (Cappellari et al. 2013) to determine the empirical relation, which can be used to retrieve the systemic redshift of a galaxy: \begin{equation*} V_{\mathrm{peak}}^{\mathrm{red}} = 0.9 (\pm 0.14) \times {\rm FWHM({\rm Ly}\alpha )} -34 (\pm 60)\, {\rm km\,\, s}^{-1} \end{equation*} (2) This relation is also compatible with the one-to-one relation, given the uncertainties in the fit parameters. The intrinsic scatter estimated from the linear regression is 72(±12) km s−1, slightly larger than with Method 1. 3.3 Comparison of the methods We check that the corrected redshifts from both methods give results that are closer to the systemic redshift of the objects than the ‘Ly α redshifts’, i.e. taking $$V_{\rm peak}^{\rm red}$$ as the systemic redshift, as usually done (Fig. 3). The standard deviation of the red histograms (corrected redshifts), reflecting both the intrinsic scatter and measurement errors, are comparable for the two methods, though slightly better for the blue bump method. We therefore propose to use half of the separation of the peaks as a proxy for the red peak shift of blue bump LAEs, and the Ly α FWHM for single-peaked spectra.3 They allow to recover the systemic redshift from the Ly α line, with an uncertainty lower than ±100 km s−1 from z ≈ 0 to 7. This suggests that the same scattering processes, linking the line shift and the line width, are at play at every redshift, and that the effect of the IGM does not erase this correlation. Figure 3. View largeDownload slide Comparison of the distributions of Ly α redshift errors (=zLy α − zsys, in black) with redshift distributions corrected with Method 1 (in red, top panel) and with Method 2 (in red, bottom panel). Figure 3. View largeDownload slide Comparison of the distributions of Ly α redshift errors (=zLy α − zsys, in black) with redshift distributions corrected with Method 1 (in red, top panel) and with Method 2 (in red, bottom panel). 4 DISCUSSION 4.1 Effect of the spectral resolution These two methods to retrieve the systemic redshift of a LAE from the shape of its Ly α profile rely on measurements of either the positions of the blue and red Ly α emission peaks or the (red peak) FWHM. Both of these measures are affected by the spectral resolution. Although the data points presented in Section 3 were collected from the literature and MUSE surveys and span a range of spectral resolutions from R ∼ 1000 (LRIS) to R ∼ 5000 (X-Shooter, HST-COS), they all seem to follow the same relation. We investigated the effect of spectral resolution on synthetic spectra constructed from Ly α radiation transfer simulations. Poorer spectral resolution broadens the peaks, and since Ly α profiles are often asymmetric, it also has the effect of shifting the peak towards longer wavelengths. The latter effect is weaker than the broadening. As a consequence, the effect of spectral resolution may flatten the slope but seems not to break the correlation. 4.2 Comparison with models We now compare our results with numerical simulations of Ly α radiative transfer in expanding shells performed with the MCLya code (Schaerer et al. 2011; Verhamme, Schaerer & Maselli 2006). These models describe in a simple, idealized, way the propagation of Ly α photons emitted in HII regions through gas outflows which seem ubiquitous in star-forming galaxies, especially at high redshift (Shapley et al. 2003; Steidel et al. 2010; Hashimoto et al. 2015). Assuming a central point-source surrounded by an expanding shell of gas with varying HI column density (NHI), speed (Vexp), dust opacity (τd) and temperature (described by the Doppler parameter $$b \propto \sqrt{T}$$), shell models have proven very successful in reproducing a large diversity of Ly α line profiles. Here, we use simulations with different intrinsic Gaussian line widths (σi) and various shell parameter values (NHI, Vexp, τd, b), degraded to mimic the MUSE spectral resolution. We measure FWHM, $$V_{\rm peak}^{\rm red}$$, and ΔV1/2 the same way as for the data. We compare the observed correlation between $$V_{\rm peak}^{\rm red}$$ and the separation between the peaks of blue-bump LAEs (ΔV1/2) with results from models that produce double-peak profiles (Fig. 4, left-hand panel). Predictions from expanding shell models lie very close to the one-to-one relation and reproduce nicely the observed properties of the Ly α lines. Objects with increasing $$V_{\rm peak}^{\rm red}$$ and ΔV1/2 correspond to expanding shells with larger HI column densities. This echoes the analytical solutions for Ly α RT in static homogeneous media (Neufeld 1990; Dijkstra, Haiman & Spaans 2006) that yield profiles with symmetric peaks around the line centre, whose positions are primarily set by the HI opacity and correspond to $$V_{\rm peak}^{\rm red}$$$$\propto \tau _{\rm HI}^{1/3}$$. Figure 4. View largeDownload slide Points show the relationship between half of the separation of the peaks and the shift of the Ly α line, and between the FWHM and the shift of the Ly α line, for synthetic spectra from expanding shells, spheres or bi-conical outflows (Schaerer et al. 2011; Zheng & Wallace 2014). The trend is driven by the column density of the scattering medium, but holds for the different idealized geometries. The symbol colours scale with the column density (in cm−2) of the shells and symbol sizes scale with the radial expansion velocity (from 0 to 400 km s−1). The red line and dashed black line are identical as in Fig. 2. Figure 4. View largeDownload slide Points show the relationship between half of the separation of the peaks and the shift of the Ly α line, and between the FWHM and the shift of the Ly α line, for synthetic spectra from expanding shells, spheres or bi-conical outflows (Schaerer et al. 2011; Zheng & Wallace 2014). The trend is driven by the column density of the scattering medium, but holds for the different idealized geometries. The symbol colours scale with the column density (in cm−2) of the shells and symbol sizes scale with the radial expansion velocity (from 0 to 400 km s−1). The red line and dashed black line are identical as in Fig. 2. As shown in Fig. 4 (right-hand panel), the correlation between the shift of the red peak and the FWHM of the Ly α line naturally arises from scattering processes. The slope predicted by the models is close to one whereas the relation derived from observations in Section 3 is shallower (≈0.9; red curve in the right-hand panel of Figs 2, 4). However, it is worth pointing out that we explore a much larger range of FWHMs in the right-hand panel of Fig. 4 (from 0 to 1200 km s−1) compared to Fig. 2 where observed FWHMs vary from 214 to 512 km s−1. For FWHM values less than 600 km s−1, the model predictions lie close to the FWHM–$$V_{\rm peak}^{\rm red}$$ relation derived in Section 3. Although the exact location of each simulated object in the FWHM–$$V_{\rm peak}^{\rm red}$$ plane seems to depend on each parameter, we see that models with higher HI column densities lead to broader lines and larger shifts of the peak (colour-coded circles). A similar trend is found by Zheng & Wallace (2014) who performed Ly α radiation transfer simulations in anisotropic configurations (bipolar outflows) and inhomogeneous media (i.e. HI distributions with velocity or density gradients). Overall, this may suggest that the FWHM–$$V_{\rm peak}^{\rm red}$$ correlation holds regardless of the assumed geometry and kinematics of the outflows, and that the HI opacity of the ISM and/or the medium surrounding galaxies (i.e. the CGM) is the main driver that shapes the observed Ly α line profiles. 5 CONCLUSIONS The recent increase in the number of LAEs with detected nebular lines allows to calibrate empirical methods to retrieve the systemic redshift from the shape of the Ly α line. In addition to measurements from the literature, we report 13 new detections from several MUSE GTO programs. We searched for Ly α+C iii] emitters in the MUSE-Deep survey (Bacon et al. 2017), behind z ∼ 0.7 galaxy groups (Contini et al., in preparation), and lensed by three clusters (SMACSJ2031.8-4036 in Patrício et al. 2016, AS1063, MACS0416 in Richard et al., in preparation). We find a robust correlation between the shift of the Ly α peak with respect to systemic redshift ($$V_{\rm peak}^{\rm red}$$) and half of the separation of the peaks (ΔV1/2) for LAEs with blue bump spectra. The intrinsic scatter around the relation is ±53 km s−1. We also find a correlation between the shift of the Ly α peak with respect to systemic redshift ($$V_{\rm peak}^{\rm red}$$) and its FWHM, for LAEs with known systemic redshift. The intrinsic scatter is of the same order (±73 km s−1). These two relations have been approximated by linear fitting formulas as given in equations (1) and (2). These formulae have been derived for data with spectral resolution 1000 < R < 5000, they should be used on data with similar spectral resolution. The relative redshift error if estimated from Ly α with $$V_{\rm peak}^{\rm red}$$ = 300 km s−1 is (Δz /z)(Ly α) = ((1+z) × ($$V_{\rm peak}^{\rm red}$$/ c))/z ∼10−3 at z = 3. The two methods presented in this letter can therefore help reduce systematic errors on distance measures. This is of great importance for redshift surveys at z ≳ 3, where spectroscopic redshifts often rely on the Ly α emission line. Futures observations with better spectral resolution should allow to refine the proposed relations. Footnotes 1 CIII] can be used as a redshift indicator when the two components of the doublet are well resolved, i.e. when R ∼ 2000, because their relative strength depends on the density. 2 http://www-astro.physics.ox.ac.u/∼mxc/software/#lts 3 We have also tested the relation between Ly α EWs and $$V_{\rm peak}^{\rm red}$$, but did not find any significant correlation. ACKNOWLEDGEMENTS We thank the anonymous referee for her/his helpful report. AV is supported by a Marie Heim Vögtlin fellowship of the Swiss National Foundation. TG is grateful to the LABEX Lyon Institute of Origins (ANR-10-LABX-0066) of the Université de Lyon for its financial support within the program 'Investissements d’Avenir' (ANR-11-IDEX-0007) of the French government operated by the National Research Agency (ANR). TC, EV, JZ acknowledge support of the ANR FOGHAR (ANR-13-BS05-0010-02), the OCEVU Labex (ANR-11- LABX-0060) and the A*MIDEX project (ANR-11-IDEX-0001-02) funded by the 'Investissements d’Avenir' French government program managed by the ANR. RB and FL acknowledges support from the ERC advanced grant 339659-MUSICOS. JR and VP acknowledge support from the ERC starting grant 336736-CALENDS. RAM acknowledges support by the Swiss National Science Foundation. JS acknowledges support from the ERC grant 278594-GasAroundGalaxies. JB acknowledges support by Fundação para a Ciência e a Tecnologia (FCT) through national funds (UID/FIS/04434/2013) and by FEDER through COMPETE2020 (POCI-01-0145-FEDER-007672) and Investigador FCT contract IF/01654/2014/CP1215/CT0003. REFERENCES Bacon R. , et al. , 2010 , The MUSE second-generation VLT instrument . Ground-based and Airborne Instrumentation for Astronomy III . p. 773508 , SPIE series . Bacon R. et al. , 2015 , A&A , 575 , A75 CrossRef Search ADS Bacon R. , et al. , 2017 , A&A , 608 , 1 Bielby R. M. et al. , 2011 , MNRAS , 414 , 2 https://doi.org/10.1111/j.1365-2966.2011.18338.x CrossRef Search ADS Blanc G. A. et al. , 2011 , ApJ , 736 , 31 https://doi.org/10.1088/0004-637X/736/1/31 CrossRef Search ADS Bouwens R. J. et al. , 2015 , ApJ , 803 , 34 https://doi.org/10.1088/0004-637X/803/1/34 CrossRef Search ADS Brinchmann J. et al. , 2017 , A&A , 608 , 3 , Cappellari M. et al. , 2013 , MNRAS , 432 , 1709 https://doi.org/10.1093/mnras/stt562 CrossRef Search ADS Caruana J. et al. , 2018 , MNRAS , 473 , 30 https://doi.org/10.1093/mnras/stx2307 CrossRef Search ADS Dijkstra M. , Haiman Z. , Spaans M. , 2006 , ApJ , 649 , 14 https://doi.org/10.1086/506243 CrossRef Search ADS Drake A. B. et al. , 2017 , A&A , 608 , 6 , Erb D. K. et al. , 2014 , ApJ , 795 , 33 https://doi.org/10.1088/0004-637X/795/1/33 CrossRef Search ADS Finkelstein S. L. et al. , 2015 , ApJ , 810 , 71 https://doi.org/10.1088/0004-637X/810/1/71 CrossRef Search ADS Hashimoto T. , Ouchi M. , Shimasaku K. , Ono Y. , Nakajima K. , Rauch M. , Lee J. , Okamura S. , 2013 , ApJ , 765 , 70 https://doi.org/10.1088/0004-637X/765/1/70 CrossRef Search ADS Hashimoto T. et al. , 2015 , ApJ , 812 , 157 https://doi.org/10.1088/0004-637X/812/2/157 CrossRef Search ADS Hashimoto T. et al. , 2017 , MNRAS , 465 , 1543 https://doi.org/10.1093/mnras/stw2834 CrossRef Search ADS Henry A. , Scarlata C. , Martin C. L. , Erb D. , 2015 , ApJ , 809 , 19 https://doi.org/10.1088/0004-637X/809/1/19 CrossRef Search ADS Herenz E. C. et al. , 2017 , A&A , 606 , 12 , Inami H. et al. , 2017 , A&A , 608 , 2 , Izotov Y. I. , Guseva N. G. , Thuan T. X. , 2011 , ApJ , 728 , 161 https://doi.org/10.1088/0004-637X/728/2/161 CrossRef Search ADS Jaskot A. E. , Oey M. S. , 2014 , ApJ , 791 , L19 https://doi.org/10.1088/2041-8205/791/2/L19 CrossRef Search ADS Kulas K. R. , Shapley A. E. , Kollmeier J. A. , Zheng Z. , Steidel C. C. , Hainline K. N. , 2012 , ApJ , 745 , 33 https://doi.org/10.1088/0004-637X/745/1/33 CrossRef Search ADS Le Fèvre O. et al. , 2015 , A&A , 576 , A79 CrossRef Search ADS Mahler G. et al. , 2017 , MNRAS , 473 , 663 , Maseda M. V. et al. , 2017 , A&A , 608 , 4 , McLinden E. M. et al. , 2011 , ApJ , 730 , 136 https://doi.org/10.1088/0004-637X/730/2/136 CrossRef Search ADS Neufeld D. A. , 1990 , ApJ , 350 , 216 https://doi.org/10.1086/168375 CrossRef Search ADS Oesch P. A. et al. , 2015 , ApJ , 804 , L30 https://doi.org/10.1088/2041-8205/804/2/L30 CrossRef Search ADS Ouchi M. et al. , 2008 , ApJS , 176 , 301 https://doi.org/10.1086/527673 CrossRef Search ADS Patrício V. et al. , 2016 , MNRAS , 456 , 4191 https://doi.org/10.1093/mnras/stv2859 CrossRef Search ADS Rakic O. , Schaye J. , Steidel C. C. , Rudie G. C. , 2011 , MNRAS , 414 , 3265 https://doi.org/10.1111/j.1365-2966.2011.18624.x CrossRef Search ADS Schaerer D. , Hayes M. , Verhamme A. , Teyssier R. , 2011 , A&A , 531 , A12 CrossRef Search ADS Shapley A. E. , Steidel C. C. , Pettini M. , Adelberger K. L. , 2003 , ApJ , 588 , 65 https://doi.org/10.1086/373922 CrossRef Search ADS Sobral D. et al. , 2017 , MNRAS , 466 , 1242 https://doi.org/10.1093/mnras/stw3090 CrossRef Search ADS Song M. et al. , 2014 , ApJ , 791 , 3 https://doi.org/10.1088/0004-637X/791/1/3 CrossRef Search ADS Stark D. P. et al. , 2014 , MNRAS , 445 , 3200 https://doi.org/10.1093/mnras/stu1618 CrossRef Search ADS Stark D. P. et al. , 2017 , MNRAS , 464 , 469 https://doi.org/10.1093/mnras/stw2233 CrossRef Search ADS Steidel C. C. , Adelberger K. L. , Shapley A. E. , Pettini M. , Dickinson M. , Giavalisco M. , 2003 , ApJ , 592 , 728 https://doi.org/10.1086/375772 CrossRef Search ADS Steidel C. C. , Erb D. K. , Shapley A. E. , Pettini M. , Reddy N. , Bogosavljević M. , Rudie G. C. , Rakic O. , 2010 , ApJ , 717 , 289 https://doi.org/10.1088/0004-637X/717/1/289 CrossRef Search ADS Trainor R. F. , Steidel C. C. , Strom A. L. , Rudie G. C. , 2015 , ApJ , 809 , 89 https://doi.org/10.1088/0004-637X/809/1/89 CrossRef Search ADS Vanzella E. et al. , 2016 , ApJ , 821 , L27 https://doi.org/10.3847/2041-8205/821/2/L27 CrossRef Search ADS Verhamme A. , Schaerer D. , Maselli A. , 2006 , A&A , 460 , 397 CrossRef Search ADS Verhamme A. , Orlitová I. , Schaerer D. , Izotov Y. , Worseck G. , Thuan T. X. , Guseva N. , 2017 , A&A , 597 , A13 CrossRef Search ADS Yamada T. , Matsuda Y. , Kousai K. , Hayashino T. , Morimoto N. , Umemura M. , 2012 , ApJ , 751 , 29 https://doi.org/10.1088/0004-637X/751/1/29 CrossRef Search ADS Yang H. , Malhotra S. , Gronke M. , Rhoads J. E. , Dijkstra M. , Jaskot A. , Zheng Z. , Wang J. , 2016 , ApJ , 820 , 130 https://doi.org/10.3847/0004-637X/820/2/130 CrossRef Search ADS Yang H. et al. , 2017 , ApJ , 844 , 171 , Zheng Z. , Wallace J. , 2014 , ApJ , 794 , 116 https://doi.org/10.1088/0004-637X/794/2/116 CrossRef Search ADS © 2018 The Author(s) Published by Oxford University Press on behalf of the Royal Astronomical Society This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Monthly Notices of the Royal Astronomical Society: Letters Oxford University Press

# Recovering the systemic redshift of galaxies from their Lyman alpha line profile

, Volume Advance Article (1) – Apr 6, 2018
6 pages

/lp/ou_press/recovering-the-systemic-redshift-of-galaxies-from-their-lyman-alpha-uzFnmahW3q
Publisher
Oxford University Press
ISSN
1745-3925
eISSN
1745-3933
D.O.I.
10.1093/mnrasl/sly058
Publisher site
See Article on Publisher Site

### Abstract

Abstract The Lyman alpha (Ly α) line of Hydrogen is a prominent feature in the spectra of star-forming galaxies, usually redshifted by a few hundreds of km s−1 compared to the systemic redshift. This large offset hampers follow-up surveys, galaxy pair statistics, and correlations with quasar absorption lines when only Ly α is available. We propose diagnostics that can be used to recover the systemic redshift directly from the properties of the Ly α line profile. We use spectroscopic observations of Ly α emitters for which a precise measurement of the systemic redshift is available. Our sample contains 13 sources detected between z ≈ 3 and z ≈ 6 as part of various multi-unit spectroscopic explorer guaranteed time observations. We also include a compilation of spectroscopic Ly α data from the literature spanning a wide redshift range (z ≈ 0–8). First, restricting our analysis to double-peaked Ly α spectra, we find a tight correlation between the velocity offset of the red peak with respect to the systemic redshift, $$V_{\rm peak}^{\rm red}$$, and the separation of the peaks. Secondly, we find a correlation between $$V_{\rm peak}^{\rm red}$$ and the full width at half-maximum of the Ly α line. Fitting formulas to estimate systemic redshifts of galaxies with an accuracy of ≤100 km s−1, when only the Ly α emission line is available, are given for the two methods. galaxies: high-redshift, galaxies: starburst, galaxies: statistics, ultraviolet: galaxies 1 INTRODUCTION In the last few decades, large samples of high-redshift galaxies (z > 2) have been assembled from deep photometric surveys based on broad/narrow-band selection techniques (Steidel et al. 2003; Ouchi et al. 2008; Bouwens et al. 2015; Finkelstein et al. 2015; Sobral et al. 2017, and references therein). In parallel, blind spectroscopic searches commonly rely on the Lyman alpha (Ly α) line redshifted to the optical or the near-infrared to identify or confirm sources at z ≥ 2 (e.g. Blanc et al. 2011; Bielby et al. 2011; Le Fèvre et al. 2015). The number of spectroscopic detections of Ly α emitters (LAE) is now being increased dramatically with ongoing observational campaigns with the multi-unit spectroscopic explorer (MUSE, Bacon et al. 2010) on European Southern Obseratory's Very Large Telescope (ESO VLT), allowing us to study galaxy formation and evolution with a homogeneous sample of sources over a large redshift range (2.8 ≲ z ≲ 6.7, e.g. Bacon et al. 2015, 2017; Drake et al. 2016; Herenz et al. 2017; Mahler et al. 2017; Caruana et al. 2018). Several studies have demonstrated that the Ly α emission line is not exactly tracing systemic redshift (e.g. Shapley et al. 2003; McLinden et al. 2011; Rakic et al. 2011; Song et al. 2014; Hashimoto et al. 2015). Instead, the line profiles often show a complex structure that arguably originates from the propagation of resonant Ly α photons in neutral gas within the interstellar medium and/or in the vicinity of galaxies. Among the broad diversity of Ly α profiles in Ly α emitting galaxies, we identify the most common two categories: (i) spectra with a redshifted single peak [∼2/3 of Ly α emitting Lyman Break Galaxies, called LBGs, from Kulas et al. (2012)], and (ii) double-peaked profiles, with a prominent red peak and a smaller blue bump [∼2/3 of the remaining 1/3 of Ly α emitting LBGs that are multiple peaked, from Kulas et al. (2012); 40 per cent of the LAEs observed by Yamada et al. (2012)]. We will refer to the latter as blue bump LAEs in the remainder of this paper. Understanding the nature of blue bump LAEs and studying their occurrence and evolution with redshift will be the goal of a forthcoming study. The vast majority of objects display a red peak shifted by a variable amount peaking around ∼400  km s−1 for LBGs (e.g. Shapley et al. 2003; Kulas et al. 2012), ∼200 km s−1 for LAEs (e.g. Hashimoto et al. 2013; Erb et al. 2014; Song et al. 2014; Hashimoto et al. 2015; Henry et al. 2015; Trainor et al. 2015), and less than ∼150 km s−1 for a small sample of five local Lyman Continuum Emitters (Verhamme et al. 2017). If not accounted for, this offset with respect to the systemic redshift can be problematic when addressing astrophysical issues which require accurate systemic redshift measurements (e.g. galaxy interactions, gas kinematics, baryonic acoustic oscillations, intergalactic medium-galaxy emission/absorption correlations). The scope of this paper is to investigate whether the Ly α profile shape can be used to determine the systemic redshift of galaxies. The outline of this Letter is as follows: in Section2, we gather recent spectroscopic data from MUSE Guaranteed Time Observations (GTO) surveys and from the literature that have sufficient spectral resolution to investigate the Ly α line properties, as well as reliable systemic redshift measurements. In Section 3, we present two diagnostics that can be used to recover the systemic redshift from Ly α, which we compare to models in Section 4. Section 5 summarizes our findings. 2 A SAMPLE OF LAES WITH KNOWN SYSTEMIC REDSHIFT In order to investigate the link between the shape of the Ly α line and the systemic redshift, we collect a diverse sample of LAEs with a precise measure of the systemic redshift. Our sample consists of high-redshift (z > 2) LAEs with detected C iii]λλ1907, 1909, [O iii]λλ4959, 5007 or H α λ6563 emission, and low redshift (z < 0.4) LAEs with Ly α observations in the UV rest-frame obtained with the Cosmic Origins Spectrograph onboard HST, and ancillary optical spectra from the sDSS data base containing several nebular emission lines from which the redshift is determined with great accuracy. We present these data in the following paragraphs. For each Ly α spectrum, we measure $$V_{\rm peak}^{\rm red}$$ as the location of the maximum of the Ly α flux redwards of the systemic redshift, and full width at half-maximum (FWHM) as the width of the part of the spectrum uncorrected for instrumental broadening with flux above half of the maximum, both directly on the data, without any modelling. 2.1 LAEs from MUSE GTO data Stark et al. (2014) reported the detection of C iii]λλ1907, 1909 emission from low mass star-forming galaxies. When observed, this doublet is the strongest UV emission line after Ly α, and, in contrast to Ly α, it is an optically thin nebular line, tracing the systemic redshift1 of the Ly α production site. The redshift window where Ly α and C iii] are both observable within the VLT/MUSE spectral range is 2.9 < z < 3.8. MUSE is an optical Integral Field Unit (IFU) spectrograph with medium spectral resolution (from R ∼ 2000 in the blue to R ∼ 4000 in the red). Within several projects in the MUSE consortium using GTO (Bacon et al. 2017; Brinchmann et al. 2017; Herenz et al. 2017; Inami et al. 2017; Mahler et al. 2017; Maseda et al. 2017; Caruana et al. 2018) and Science Verification (SV) or commissioning data (Patrício et al. 2016), we find 13 LAEs with reliable C iii] detections, that is, non-contaminated by sky lines and with a S/N > 3. We list these objects in Table 1 (see Maseda et al. 2017, for a systematic study of ciii emitters in the MUSE GTO data from the Hubble Ultra Deep Field). For each of these LAEs we measure the shift of the Ly α emission compared to C iii], $$V_{\rm peak}^{\rm red}$$, the observed FWHM, and the separation of the Ly α peaks for the eight blue bump LAEs among them. Table 1. MUSE Ly α+C iii] emitters. The sixth column indicates the separation of the peaks (i.e. 2 × ΔV1/2, in km s−1) for blue bump LAEs, and is left empty for single-peaked profiles. ID RA DEC EW (Å) $$V_{\rm peak}^{\rm red}$$(km s−1) FWHM (km s−1) ΔV (km s−1) zsys, CIII] Observations sys 1a 307.97040 −40.625694 32 176 ± 11 248 ± 9 – 3.5062 Commissioning mul 11b 342.175042 −44.541031 222 215 ± 35 150 ± 35 375 ± 35 3.1163 AS1063 mul 14b 342.178833 −44.535869 29 385 ± 35 300 ± 35 – 3.1150 AS1063 sys 44b 64.0415559 −24.0599916 57 303 ± 35 360 ± 35 570 ± 35 3.2886 MACS0416 sys 132b 64.0400838 −24.0667408 62 331 ± 35 288 ± 35 510 ± 35 3.2882 MACS0416 106c 53.163726 −27.7790755 72 379 ± 13 414 ± 13 828 ± 35 3.2767 udf-10 118c 53.157088 −27.7802688 65 301 ± 28 284 ± 28 568 ± 35 3.0173 udf-10 1180c 53.195735 −27.7827171 80 220 ± 23 348 ± 32 – 3.3228 udf mosaic 6298c 53.169249 −27.7812550 83 582 ± 38 512 ± 56 – 3.1287 udf-10 6666c 53.159576 −27.7767193 52 284 ± 13 377 ± 11 754 ± 35 3.4349 udf-10 50d 150.149656 2.061272 50 431 ± 42 268 ± 39 – 3.8237 GR30 48d 149.852989 2.488099 68 294 ± 35 214 ± 35 705 ± 35. 3.3280 GR34 102d 150.050268 2.600025 76 299 ± 15 229 ± 15 385 ± 35. 3.0400 GR84 ID RA DEC EW (Å) $$V_{\rm peak}^{\rm red}$$(km s−1) FWHM (km s−1) ΔV (km s−1) zsys, CIII] Observations sys 1a 307.97040 −40.625694 32 176 ± 11 248 ± 9 – 3.5062 Commissioning mul 11b 342.175042 −44.541031 222 215 ± 35 150 ± 35 375 ± 35 3.1163 AS1063 mul 14b 342.178833 −44.535869 29 385 ± 35 300 ± 35 – 3.1150 AS1063 sys 44b 64.0415559 −24.0599916 57 303 ± 35 360 ± 35 570 ± 35 3.2886 MACS0416 sys 132b 64.0400838 −24.0667408 62 331 ± 35 288 ± 35 510 ± 35 3.2882 MACS0416 106c 53.163726 −27.7790755 72 379 ± 13 414 ± 13 828 ± 35 3.2767 udf-10 118c 53.157088 −27.7802688 65 301 ± 28 284 ± 28 568 ± 35 3.0173 udf-10 1180c 53.195735 −27.7827171 80 220 ± 23 348 ± 32 – 3.3228 udf mosaic 6298c 53.169249 −27.7812550 83 582 ± 38 512 ± 56 – 3.1287 udf-10 6666c 53.159576 −27.7767193 52 284 ± 13 377 ± 11 754 ± 35 3.4349 udf-10 50d 150.149656 2.061272 50 431 ± 42 268 ± 39 – 3.8237 GR30 48d 149.852989 2.488099 68 294 ± 35 214 ± 35 705 ± 35. 3.3280 GR34 102d 150.050268 2.600025 76 299 ± 15 229 ± 15 385 ± 35. 3.0400 GR84 aPatricio et al. 2016; bRichard et al., in preparation; cBacon et al. 2017, Inami et al. 2017, Maseda et al. 2017; dContini et al., in preparation. View Large Table 1. MUSE Ly α+C iii] emitters. The sixth column indicates the separation of the peaks (i.e. 2 × ΔV1/2, in km s−1) for blue bump LAEs, and is left empty for single-peaked profiles. ID RA DEC EW (Å) $$V_{\rm peak}^{\rm red}$$(km s−1) FWHM (km s−1) ΔV (km s−1) zsys, CIII] Observations sys 1a 307.97040 −40.625694 32 176 ± 11 248 ± 9 – 3.5062 Commissioning mul 11b 342.175042 −44.541031 222 215 ± 35 150 ± 35 375 ± 35 3.1163 AS1063 mul 14b 342.178833 −44.535869 29 385 ± 35 300 ± 35 – 3.1150 AS1063 sys 44b 64.0415559 −24.0599916 57 303 ± 35 360 ± 35 570 ± 35 3.2886 MACS0416 sys 132b 64.0400838 −24.0667408 62 331 ± 35 288 ± 35 510 ± 35 3.2882 MACS0416 106c 53.163726 −27.7790755 72 379 ± 13 414 ± 13 828 ± 35 3.2767 udf-10 118c 53.157088 −27.7802688 65 301 ± 28 284 ± 28 568 ± 35 3.0173 udf-10 1180c 53.195735 −27.7827171 80 220 ± 23 348 ± 32 – 3.3228 udf mosaic 6298c 53.169249 −27.7812550 83 582 ± 38 512 ± 56 – 3.1287 udf-10 6666c 53.159576 −27.7767193 52 284 ± 13 377 ± 11 754 ± 35 3.4349 udf-10 50d 150.149656 2.061272 50 431 ± 42 268 ± 39 – 3.8237 GR30 48d 149.852989 2.488099 68 294 ± 35 214 ± 35 705 ± 35. 3.3280 GR34 102d 150.050268 2.600025 76 299 ± 15 229 ± 15 385 ± 35. 3.0400 GR84 ID RA DEC EW (Å) $$V_{\rm peak}^{\rm red}$$(km s−1) FWHM (km s−1) ΔV (km s−1) zsys, CIII] Observations sys 1a 307.97040 −40.625694 32 176 ± 11 248 ± 9 – 3.5062 Commissioning mul 11b 342.175042 −44.541031 222 215 ± 35 150 ± 35 375 ± 35 3.1163 AS1063 mul 14b 342.178833 −44.535869 29 385 ± 35 300 ± 35 – 3.1150 AS1063 sys 44b 64.0415559 −24.0599916 57 303 ± 35 360 ± 35 570 ± 35 3.2886 MACS0416 sys 132b 64.0400838 −24.0667408 62 331 ± 35 288 ± 35 510 ± 35 3.2882 MACS0416 106c 53.163726 −27.7790755 72 379 ± 13 414 ± 13 828 ± 35 3.2767 udf-10 118c 53.157088 −27.7802688 65 301 ± 28 284 ± 28 568 ± 35 3.0173 udf-10 1180c 53.195735 −27.7827171 80 220 ± 23 348 ± 32 – 3.3228 udf mosaic 6298c 53.169249 −27.7812550 83 582 ± 38 512 ± 56 – 3.1287 udf-10 6666c 53.159576 −27.7767193 52 284 ± 13 377 ± 11 754 ± 35 3.4349 udf-10 50d 150.149656 2.061272 50 431 ± 42 268 ± 39 – 3.8237 GR30 48d 149.852989 2.488099 68 294 ± 35 214 ± 35 705 ± 35. 3.3280 GR34 102d 150.050268 2.600025 76 299 ± 15 229 ± 15 385 ± 35. 3.0400 GR84 aPatricio et al. 2016; bRichard et al., in preparation; cBacon et al. 2017, Inami et al. 2017, Maseda et al. 2017; dContini et al., in preparation. View Large 2.2 High-z data from the literature Stark et al. (2017) reported the detection of C iii] from one of the highest redshift Ly α emitters ever observed (z ∼ 7.730) with $$V_{\rm peak}^{\rm red}$$ ∼ 340 km s−1; the Ly α FWHM $$=360^{+90}_{-70}$$ km s−1 is measured by Oesch et al. (2015). Vanzella et al. (2016) report a narrow Ly α line observed at medium spectral resolution using VLT-Xshooter of a magnified star-forming galaxy at z = 3.1169, with $$V_{\rm peak}^{\rm red}$$(Ly α) ∼ 100 km s−1 and FWHM(Ly  α) ∼104 km s−1. From Hashimoto et al. (2015, 2017), we select the six LAEs observed with MagE (R ∼ 4100). Their systemic redshifts have been obtained with either H α or [O iii] lines. Three of these objects are blue-bump LAEs, for which we also measure the separation of the peaks. Kulas et al. (2012) reported that a significant fraction (∼30  per cent) of their Ly α emitting LBGs show a complex Ly α profile, with at least one secondary peak. Blue bump objects (their Group I) represent the majority of their profiles (11 out of 18 objects). We add these 11 objects to our sample of blue bumps LAEs. 2.3 Low-z data from the literature Green Pea galaxies (hereafter GPs) are LAEs in the local Universe (z ∼ 0.1 to 0.3; Jaskot & Oey 2014; Henry et al. 2015; Verhamme et al. 2017; Yang et al. 2017). The systemic redshift of these objects was compiled from the several nebular lines contained in their SDSS optical spectrum (e.g Izotov, Guseva & Thuan 2011). Note that the C iii] emission line is out of the UV spectral range probed by the available HST-COS observations. For a sample of 17 GPs from Jaskot & Oey (2014), Henry et al. (2015); Verhamme et al. (2017), we measure $$V_{\rm peak}^{\rm red}$$ and FWHM on the data. For 21 new GP observations, we use the $$V_{\rm peak}^{\rm red}$$ and FWHM values computed by Yang et al. (2017) given in their table 2. GPs nearly always exhibit blue bump Ly α profiles (Jaskot & Oey 2014; Henry et al. 2015; Verhamme et al. 2017). For the blue bump GPs, we also measure the separation of the peaks. 3 DERIVING SYSTEMIC REDSHIFT FROM LYMAN-ALPHA 3.1 Method 1: systemic redshift of blue bump LAEs In this section, only blue bump spectra, i.e. double peaks with a red peak higher than the blue peak, are considered. We note that, for all blue bump LAEs studied here, the systemic redshift always falls in-between the Ly α peaks, as illustrated in Fig. 1 for three blue-bump MUSE Ly α+C iii] emitters (see also Kulas et al. 2012; Erb et al. 2014; Yang et al. 2016). Fig. 2, left-hand panel, shows a positive empirical correlation between $$V_{\rm peak}^{\rm red}$$ and half of the separation of the peaks, ΔV1/2, for blue bump LAEs with known systemic redshift. We fit the data using the LTS_LINEFIT program described in Cappellari et al. (2013), which combines the Least Trimmed Squares robust technique of Rousseeuw & van Driessen (2006) into a least-squares fitting algorithm which allows for errors in both variables and intrinsic scatter.2 The best fit, shown by the red line in Fig. 2, is given by \begin{equation*} V_{\mathrm{peak}}^{\mathrm{red}}= 1.05 (\pm 0.11) \times \Delta V_{1/2} - 12 (\pm 37) \,\,{\rm km\,\, s}^{-1} \end{equation*} (1) This relation is so close to the one-to-one relation that we assume from now that the underlying ‘true’ relation between $$V_{\rm peak}^{\rm red}$$ and ΔV1/2 is one-to-one, as expected from radiation transfer modelling (see Section 4 below). The intrinsic scatter estimated from the linear regression is 53 (± 9) km s−1. Figure 1. View largeDownload slide Three examples of rest-frame spectra of LAEs with detected C iii]λλ1907, 1909 doublet probing the systemic redshift in the MUSE-GTO observations. In each panel, the velocity shifts of the Ly α line are shown relative to C iii]1907Å. For all blue bump LAEs in our sample, the systemic redshift falls in between the blue and red peaks of the Ly α emission. Figure 1. View largeDownload slide Three examples of rest-frame spectra of LAEs with detected C iii]λλ1907, 1909 doublet probing the systemic redshift in the MUSE-GTO observations. In each panel, the velocity shifts of the Ly α line are shown relative to C iii]1907Å. For all blue bump LAEs in our sample, the systemic redshift falls in between the blue and red peaks of the Ly α emission. Figure 2. View largeDownload slide Empirical relations to determine systemic redshift from the shape of the Ly α emission. Left: correlation between the shift of the Ly α red peak, ($$V_{\rm peak}^{\rm red}$$) and half of the separation of the peaks (ΔV1/2) for a sample of LAEs with a known systemic redshift: 7 Ly α+C iii] emitters with blue bump Ly α spectra from the MUSE GTO data (red stars), blue bump LAEs among the Yang et al. (2017) GP sample (black dots), blue bump LAEs among the Hashimoto et al. (2015) MagE sample and Group I LBGs from Kulas et al. (2012) (black triangles). Right: correlation between $$V_{\rm peak}^{\rm red}$$ and FWHM among Ly α+ C iii], H α or [O iii] emitters. The black dashed line is the one-to-one relation. We checked that the correlation remains even discarding the two most upper left points. On both sides, the red curve is our best fit to the data, described by equations (1) and (2). The Pearson coefficient and the probability of the null hypothesis are shown on each panel. Figure 2. View largeDownload slide Empirical relations to determine systemic redshift from the shape of the Ly α emission. Left: correlation between the shift of the Ly α red peak, ($$V_{\rm peak}^{\rm red}$$) and half of the separation of the peaks (ΔV1/2) for a sample of LAEs with a known systemic redshift: 7 Ly α+C iii] emitters with blue bump Ly α spectra from the MUSE GTO data (red stars), blue bump LAEs among the Yang et al. (2017) GP sample (black dots), blue bump LAEs among the Hashimoto et al. (2015) MagE sample and Group I LBGs from Kulas et al. (2012) (black triangles). Right: correlation between $$V_{\rm peak}^{\rm red}$$ and FWHM among Ly α+ C iii], H α or [O iii] emitters. The black dashed line is the one-to-one relation. We checked that the correlation remains even discarding the two most upper left points. On both sides, the red curve is our best fit to the data, described by equations (1) and (2). The Pearson coefficient and the probability of the null hypothesis are shown on each panel. 3.2 Method 2: an empirical correlation between FWHM and systemic redshift In this section, both single- and double-peaked profiles are considered. The measurements are always done on the red peak, and the red peak only. In the right-hand panel of Fig. 2 we plot $$V_{\rm peak}^{\rm red}$$ versus FWHM for the full sample of LAEs presented in Section 2 (new MUSE LAEs measurements are reported in Table 1). There is a correlation between $$V_{\rm peak}^{\rm red}$$ and FWHM, although less significant than for Method 1 (see the Pearson coefficients on each panel of Fig. 2). We use the same method (Cappellari et al. 2013) to determine the empirical relation, which can be used to retrieve the systemic redshift of a galaxy: \begin{equation*} V_{\mathrm{peak}}^{\mathrm{red}} = 0.9 (\pm 0.14) \times {\rm FWHM({\rm Ly}\alpha )} -34 (\pm 60)\, {\rm km\,\, s}^{-1} \end{equation*} (2) This relation is also compatible with the one-to-one relation, given the uncertainties in the fit parameters. The intrinsic scatter estimated from the linear regression is 72(±12) km s−1, slightly larger than with Method 1. 3.3 Comparison of the methods We check that the corrected redshifts from both methods give results that are closer to the systemic redshift of the objects than the ‘Ly α redshifts’, i.e. taking $$V_{\rm peak}^{\rm red}$$ as the systemic redshift, as usually done (Fig. 3). The standard deviation of the red histograms (corrected redshifts), reflecting both the intrinsic scatter and measurement errors, are comparable for the two methods, though slightly better for the blue bump method. We therefore propose to use half of the separation of the peaks as a proxy for the red peak shift of blue bump LAEs, and the Ly α FWHM for single-peaked spectra.3 They allow to recover the systemic redshift from the Ly α line, with an uncertainty lower than ±100 km s−1 from z ≈ 0 to 7. This suggests that the same scattering processes, linking the line shift and the line width, are at play at every redshift, and that the effect of the IGM does not erase this correlation. Figure 3. View largeDownload slide Comparison of the distributions of Ly α redshift errors (=zLy α − zsys, in black) with redshift distributions corrected with Method 1 (in red, top panel) and with Method 2 (in red, bottom panel). Figure 3. View largeDownload slide Comparison of the distributions of Ly α redshift errors (=zLy α − zsys, in black) with redshift distributions corrected with Method 1 (in red, top panel) and with Method 2 (in red, bottom panel). 4 DISCUSSION 4.1 Effect of the spectral resolution These two methods to retrieve the systemic redshift of a LAE from the shape of its Ly α profile rely on measurements of either the positions of the blue and red Ly α emission peaks or the (red peak) FWHM. Both of these measures are affected by the spectral resolution. Although the data points presented in Section 3 were collected from the literature and MUSE surveys and span a range of spectral resolutions from R ∼ 1000 (LRIS) to R ∼ 5000 (X-Shooter, HST-COS), they all seem to follow the same relation. We investigated the effect of spectral resolution on synthetic spectra constructed from Ly α radiation transfer simulations. Poorer spectral resolution broadens the peaks, and since Ly α profiles are often asymmetric, it also has the effect of shifting the peak towards longer wavelengths. The latter effect is weaker than the broadening. As a consequence, the effect of spectral resolution may flatten the slope but seems not to break the correlation. 4.2 Comparison with models We now compare our results with numerical simulations of Ly α radiative transfer in expanding shells performed with the MCLya code (Schaerer et al. 2011; Verhamme, Schaerer & Maselli 2006). These models describe in a simple, idealized, way the propagation of Ly α photons emitted in HII regions through gas outflows which seem ubiquitous in star-forming galaxies, especially at high redshift (Shapley et al. 2003; Steidel et al. 2010; Hashimoto et al. 2015). Assuming a central point-source surrounded by an expanding shell of gas with varying HI column density (NHI), speed (Vexp), dust opacity (τd) and temperature (described by the Doppler parameter $$b \propto \sqrt{T}$$), shell models have proven very successful in reproducing a large diversity of Ly α line profiles. Here, we use simulations with different intrinsic Gaussian line widths (σi) and various shell parameter values (NHI, Vexp, τd, b), degraded to mimic the MUSE spectral resolution. We measure FWHM, $$V_{\rm peak}^{\rm red}$$, and ΔV1/2 the same way as for the data. We compare the observed correlation between $$V_{\rm peak}^{\rm red}$$ and the separation between the peaks of blue-bump LAEs (ΔV1/2) with results from models that produce double-peak profiles (Fig. 4, left-hand panel). Predictions from expanding shell models lie very close to the one-to-one relation and reproduce nicely the observed properties of the Ly α lines. Objects with increasing $$V_{\rm peak}^{\rm red}$$ and ΔV1/2 correspond to expanding shells with larger HI column densities. This echoes the analytical solutions for Ly α RT in static homogeneous media (Neufeld 1990; Dijkstra, Haiman & Spaans 2006) that yield profiles with symmetric peaks around the line centre, whose positions are primarily set by the HI opacity and correspond to $$V_{\rm peak}^{\rm red}$$$$\propto \tau _{\rm HI}^{1/3}$$. Figure 4. View largeDownload slide Points show the relationship between half of the separation of the peaks and the shift of the Ly α line, and between the FWHM and the shift of the Ly α line, for synthetic spectra from expanding shells, spheres or bi-conical outflows (Schaerer et al. 2011; Zheng & Wallace 2014). The trend is driven by the column density of the scattering medium, but holds for the different idealized geometries. The symbol colours scale with the column density (in cm−2) of the shells and symbol sizes scale with the radial expansion velocity (from 0 to 400 km s−1). The red line and dashed black line are identical as in Fig. 2. Figure 4. View largeDownload slide Points show the relationship between half of the separation of the peaks and the shift of the Ly α line, and between the FWHM and the shift of the Ly α line, for synthetic spectra from expanding shells, spheres or bi-conical outflows (Schaerer et al. 2011; Zheng & Wallace 2014). The trend is driven by the column density of the scattering medium, but holds for the different idealized geometries. The symbol colours scale with the column density (in cm−2) of the shells and symbol sizes scale with the radial expansion velocity (from 0 to 400 km s−1). The red line and dashed black line are identical as in Fig. 2. As shown in Fig. 4 (right-hand panel), the correlation between the shift of the red peak and the FWHM of the Ly α line naturally arises from scattering processes. The slope predicted by the models is close to one whereas the relation derived from observations in Section 3 is shallower (≈0.9; red curve in the right-hand panel of Figs 2, 4). However, it is worth pointing out that we explore a much larger range of FWHMs in the right-hand panel of Fig. 4 (from 0 to 1200 km s−1) compared to Fig. 2 where observed FWHMs vary from 214 to 512 km s−1. For FWHM values less than 600 km s−1, the model predictions lie close to the FWHM–$$V_{\rm peak}^{\rm red}$$ relation derived in Section 3. Although the exact location of each simulated object in the FWHM–$$V_{\rm peak}^{\rm red}$$ plane seems to depend on each parameter, we see that models with higher HI column densities lead to broader lines and larger shifts of the peak (colour-coded circles). A similar trend is found by Zheng & Wallace (2014) who performed Ly α radiation transfer simulations in anisotropic configurations (bipolar outflows) and inhomogeneous media (i.e. HI distributions with velocity or density gradients). Overall, this may suggest that the FWHM–$$V_{\rm peak}^{\rm red}$$ correlation holds regardless of the assumed geometry and kinematics of the outflows, and that the HI opacity of the ISM and/or the medium surrounding galaxies (i.e. the CGM) is the main driver that shapes the observed Ly α line profiles. 5 CONCLUSIONS The recent increase in the number of LAEs with detected nebular lines allows to calibrate empirical methods to retrieve the systemic redshift from the shape of the Ly α line. In addition to measurements from the literature, we report 13 new detections from several MUSE GTO programs. We searched for Ly α+C iii] emitters in the MUSE-Deep survey (Bacon et al. 2017), behind z ∼ 0.7 galaxy groups (Contini et al., in preparation), and lensed by three clusters (SMACSJ2031.8-4036 in Patrício et al. 2016, AS1063, MACS0416 in Richard et al., in preparation). We find a robust correlation between the shift of the Ly α peak with respect to systemic redshift ($$V_{\rm peak}^{\rm red}$$) and half of the separation of the peaks (ΔV1/2) for LAEs with blue bump spectra. The intrinsic scatter around the relation is ±53 km s−1. We also find a correlation between the shift of the Ly α peak with respect to systemic redshift ($$V_{\rm peak}^{\rm red}$$) and its FWHM, for LAEs with known systemic redshift. The intrinsic scatter is of the same order (±73 km s−1). These two relations have been approximated by linear fitting formulas as given in equations (1) and (2). These formulae have been derived for data with spectral resolution 1000 < R < 5000, they should be used on data with similar spectral resolution. The relative redshift error if estimated from Ly α with $$V_{\rm peak}^{\rm red}$$ = 300 km s−1 is (Δz /z)(Ly α) = ((1+z) × ($$V_{\rm peak}^{\rm red}$$/ c))/z ∼10−3 at z = 3. The two methods presented in this letter can therefore help reduce systematic errors on distance measures. This is of great importance for redshift surveys at z ≳ 3, where spectroscopic redshifts often rely on the Ly α emission line. Futures observations with better spectral resolution should allow to refine the proposed relations. Footnotes 1 CIII] can be used as a redshift indicator when the two components of the doublet are well resolved, i.e. when R ∼ 2000, because their relative strength depends on the density. 2 http://www-astro.physics.ox.ac.u/∼mxc/software/#lts 3 We have also tested the relation between Ly α EWs and $$V_{\rm peak}^{\rm red}$$, but did not find any significant correlation. ACKNOWLEDGEMENTS We thank the anonymous referee for her/his helpful report. AV is supported by a Marie Heim Vögtlin fellowship of the Swiss National Foundation. TG is grateful to the LABEX Lyon Institute of Origins (ANR-10-LABX-0066) of the Université de Lyon for its financial support within the program 'Investissements d’Avenir' (ANR-11-IDEX-0007) of the French government operated by the National Research Agency (ANR). TC, EV, JZ acknowledge support of the ANR FOGHAR (ANR-13-BS05-0010-02), the OCEVU Labex (ANR-11- LABX-0060) and the A*MIDEX project (ANR-11-IDEX-0001-02) funded by the 'Investissements d’Avenir' French government program managed by the ANR. RB and FL acknowledges support from the ERC advanced grant 339659-MUSICOS. JR and VP acknowledge support from the ERC starting grant 336736-CALENDS. RAM acknowledges support by the Swiss National Science Foundation. JS acknowledges support from the ERC grant 278594-GasAroundGalaxies. JB acknowledges support by Fundação para a Ciência e a Tecnologia (FCT) through national funds (UID/FIS/04434/2013) and by FEDER through COMPETE2020 (POCI-01-0145-FEDER-007672) and Investigador FCT contract IF/01654/2014/CP1215/CT0003. REFERENCES Bacon R. , et al. , 2010 , The MUSE second-generation VLT instrument . Ground-based and Airborne Instrumentation for Astronomy III . p. 773508 , SPIE series . Bacon R. et al. , 2015 , A&A , 575 , A75 CrossRef Search ADS Bacon R. , et al. , 2017 , A&A , 608 , 1 Bielby R. M. et al. , 2011 , MNRAS , 414 , 2 https://doi.org/10.1111/j.1365-2966.2011.18338.x CrossRef Search ADS Blanc G. A. et al. , 2011 , ApJ , 736 , 31 https://doi.org/10.1088/0004-637X/736/1/31 CrossRef Search ADS Bouwens R. J. et al. , 2015 , ApJ , 803 , 34 https://doi.org/10.1088/0004-637X/803/1/34 CrossRef Search ADS Brinchmann J. et al. , 2017 , A&A , 608 , 3 , Cappellari M. et al. , 2013 , MNRAS , 432 , 1709 https://doi.org/10.1093/mnras/stt562 CrossRef Search ADS Caruana J. et al. , 2018 , MNRAS , 473 , 30 https://doi.org/10.1093/mnras/stx2307 CrossRef Search ADS Dijkstra M. , Haiman Z. , Spaans M. , 2006 , ApJ , 649 , 14 https://doi.org/10.1086/506243 CrossRef Search ADS Drake A. B. et al. , 2017 , A&A , 608 , 6 , Erb D. K. et al. , 2014 , ApJ , 795 , 33 https://doi.org/10.1088/0004-637X/795/1/33 CrossRef Search ADS Finkelstein S. L. et al. , 2015 , ApJ , 810 , 71 https://doi.org/10.1088/0004-637X/810/1/71 CrossRef Search ADS Hashimoto T. , Ouchi M. , Shimasaku K. , Ono Y. , Nakajima K. , Rauch M. , Lee J. , Okamura S. , 2013 , ApJ , 765 , 70 https://doi.org/10.1088/0004-637X/765/1/70 CrossRef Search ADS Hashimoto T. et al. , 2015 , ApJ , 812 , 157 https://doi.org/10.1088/0004-637X/812/2/157 CrossRef Search ADS Hashimoto T. et al. , 2017 , MNRAS , 465 , 1543 https://doi.org/10.1093/mnras/stw2834 CrossRef Search ADS Henry A. , Scarlata C. , Martin C. L. , Erb D. , 2015 , ApJ , 809 , 19 https://doi.org/10.1088/0004-637X/809/1/19 CrossRef Search ADS Herenz E. C. et al. , 2017 , A&A , 606 , 12 , Inami H. et al. , 2017 , A&A , 608 , 2 , Izotov Y. I. , Guseva N. G. , Thuan T. X. , 2011 , ApJ , 728 , 161 https://doi.org/10.1088/0004-637X/728/2/161 CrossRef Search ADS Jaskot A. E. , Oey M. S. , 2014 , ApJ , 791 , L19 https://doi.org/10.1088/2041-8205/791/2/L19 CrossRef Search ADS Kulas K. R. , Shapley A. E. , Kollmeier J. A. , Zheng Z. , Steidel C. C. , Hainline K. N. , 2012 , ApJ , 745 , 33 https://doi.org/10.1088/0004-637X/745/1/33 CrossRef Search ADS Le Fèvre O. et al. , 2015 , A&A , 576 , A79 CrossRef Search ADS Mahler G. et al. , 2017 , MNRAS , 473 , 663 , Maseda M. V. et al. , 2017 , A&A , 608 , 4 , McLinden E. M. et al. , 2011 , ApJ , 730 , 136 https://doi.org/10.1088/0004-637X/730/2/136 CrossRef Search ADS Neufeld D. A. , 1990 , ApJ , 350 , 216 https://doi.org/10.1086/168375 CrossRef Search ADS Oesch P. A. et al. , 2015 , ApJ , 804 , L30 https://doi.org/10.1088/2041-8205/804/2/L30 CrossRef Search ADS Ouchi M. et al. , 2008 , ApJS , 176 , 301 https://doi.org/10.1086/527673 CrossRef Search ADS Patrício V. et al. , 2016 , MNRAS , 456 , 4191 https://doi.org/10.1093/mnras/stv2859 CrossRef Search ADS Rakic O. , Schaye J. , Steidel C. C. , Rudie G. C. , 2011 , MNRAS , 414 , 3265 https://doi.org/10.1111/j.1365-2966.2011.18624.x CrossRef Search ADS Schaerer D. , Hayes M. , Verhamme A. , Teyssier R. , 2011 , A&A , 531 , A12 CrossRef Search ADS Shapley A. E. , Steidel C. C. , Pettini M. , Adelberger K. L. , 2003 , ApJ , 588 , 65 https://doi.org/10.1086/373922 CrossRef Search ADS Sobral D. et al. , 2017 , MNRAS , 466 , 1242 https://doi.org/10.1093/mnras/stw3090 CrossRef Search ADS Song M. et al. , 2014 , ApJ , 791 , 3 https://doi.org/10.1088/0004-637X/791/1/3 CrossRef Search ADS Stark D. P. et al. , 2014 , MNRAS , 445 , 3200 https://doi.org/10.1093/mnras/stu1618 CrossRef Search ADS Stark D. P. et al. , 2017 , MNRAS , 464 , 469 https://doi.org/10.1093/mnras/stw2233 CrossRef Search ADS Steidel C. C. , Adelberger K. L. , Shapley A. E. , Pettini M. , Dickinson M. , Giavalisco M. , 2003 , ApJ , 592 , 728 https://doi.org/10.1086/375772 CrossRef Search ADS Steidel C. C. , Erb D. K. , Shapley A. E. , Pettini M. , Reddy N. , Bogosavljević M. , Rudie G. C. , Rakic O. , 2010 , ApJ , 717 , 289 https://doi.org/10.1088/0004-637X/717/1/289 CrossRef Search ADS Trainor R. F. , Steidel C. C. , Strom A. L. , Rudie G. C. , 2015 , ApJ , 809 , 89 https://doi.org/10.1088/0004-637X/809/1/89 CrossRef Search ADS Vanzella E. et al. , 2016 , ApJ , 821 , L27 https://doi.org/10.3847/2041-8205/821/2/L27 CrossRef Search ADS Verhamme A. , Schaerer D. , Maselli A. , 2006 , A&A , 460 , 397 CrossRef Search ADS Verhamme A. , Orlitová I. , Schaerer D. , Izotov Y. , Worseck G. , Thuan T. X. , Guseva N. , 2017 , A&A , 597 , A13 CrossRef Search ADS Yamada T. , Matsuda Y. , Kousai K. , Hayashino T. , Morimoto N. , Umemura M. , 2012 , ApJ , 751 , 29 https://doi.org/10.1088/0004-637X/751/1/29 CrossRef Search ADS Yang H. , Malhotra S. , Gronke M. , Rhoads J. E. , Dijkstra M. , Jaskot A. , Zheng Z. , Wang J. , 2016 , ApJ , 820 , 130 https://doi.org/10.3847/0004-637X/820/2/130 CrossRef Search ADS Yang H. et al. , 2017 , ApJ , 844 , 171 , Zheng Z. , Wallace J. , 2014 , ApJ , 794 , 116 https://doi.org/10.1088/0004-637X/794/2/116 CrossRef Search ADS © 2018 The Author(s) Published by Oxford University Press on behalf of the Royal Astronomical Society This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

### Journal

Monthly Notices of the Royal Astronomical Society: LettersOxford University Press

Published: Apr 6, 2018

## You’re reading a free preview. Subscribe to read the entire article.

### DeepDyve is your personal research library

It’s your single place to instantly
that matters to you.

over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month ### Explore the DeepDyve Library ### Search Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly ### Organize Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place. ### Access Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals. ### Your journals are on DeepDyve Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more. All the latest content is available, no embargo periods. DeepDyve ### Freelancer DeepDyve ### Pro Price FREE$49/month
\$360/year

Save searches from
PubMed

Create lists to

Export lists, citations