Discovery of a Very Bright and Intrinsically Very Luminous, Strongly Lensed Lyα Emitting Galaxy at z = 2.82 in the BOSS Emission-Line Lens Survey^*

The rest-frame UV spectrum of the brightest lensed images, A and B, is shown in Figure 2. Despite the relatively short exposure time, the achieved high S/N GTC/OSIRIS spectrum (∼25 in the continuum) shows a strong Lyα emission, and a series of strong absorption lines, similar to those seen in the composite spectrum of hundreds of $z\sim 3$ Lyman break galaxies (LBGs) (Shapley et al. 2003). The strongest absorption features are associated with the interstellar medium and stellar winds in a variety of ionization states, from neutral (O i) to highly ionized species (Si iv or C iv). High ionization features with a strong P Cygni profile are seen both in C iv $\lambda 1548$, 1550 and Al iii $\lambda 1854$, 1862 doublets, which is indicative of stellar winds from very young massive stars. However, we identify additional absorption features unrelated to BG1429+1202 (nine absorption features associated with an intervening metal-line system at ${z}_{\mathrm{abs}}=2.179\pm 0.001$, and one broad absorption line at $5183\,\mathring{\rm A}$ with an observed equivalent width ${W}_{\mathrm{obs}}=6.95\pm 0.2$ Å remains unidentified). Analysis of the spectra for the individual lensed images A and B shows no differences in the profiles of the absorption features and no evidence for velocity offsets between them, as expected if they are both images of the same background source. The same happens to their rest-frame UV slope, β, which is essentially flat in ${F}_{\nu }$. Adopting a simple power-law approximation for the UV spectral range ${F}_{\lambda }\propto {\lambda }^{\beta }$ and using the observed r and z magnitudes (∼1600–2400 Å rest-frame), we derive $\beta =-2.1\pm 0.1$, which implies an effective UV extinction ${A}_{1600}\sim 0.6$ or $E(B-V)\sim 0.13$, reflecting modest dust content (Calzetti et al. 2000).

Zoom In Zoom Out Reset image size

We identified several photospheric absorption features, from which we derived the systemic redshift ${z}_{\mathrm{sys}}=2.8224\,\pm \,0.0013$ using the cleanest among them: O iv $\lambda 1343$, and a close blend of C ii and N iii at $\lambda 1324$.

The Lyα emission line has an observed (A + B) flux of ${F}_{\mathrm{Ly}\alpha }=(2.1\pm 0.3)\times {10}^{-15}$ erg s⁻¹ cm⁻², much higher than the Lyα flux measured within the $1^{\prime\prime}$-radius BOSS fiber (${F}_{\mathrm{Ly}\alpha }^{\mathrm{BOSS}}=0.256\times {10}^{-15}$ erg s⁻¹ cm⁻²). The measured rest-frame equivalent width of Lyα is ${W}_{0}^{\mathrm{Ly}\alpha }=39\pm 15$ Å. The errors reflect the uncertainty in the determination of the stellar continuum redward of Lyα. Although the line appears unresolved in our low-resolution spectrum ($\mathrm{FWHM}\simeq 500$ km s⁻¹), it is resolved in the BOSS spectrum. Fitting a Gaussian to the BOSS Lyα line, we measured a FWHM of 382 ± 50 km s⁻¹, after accounting for the instrumental broadening. The nebular C iii] $\lambda 1906$, 1908 doublet is also detected in emission but with low significance ($4\sigma$) and is not resolved in our GTC/OSIRIS spectrum. This galaxy can be classified as an LAE, given the rest-frame equivalent width and the velocity width of the Lyα line (e.g., Ouchi et al. 2008). From our spectrum and available photometric data we do not detect evidence of any AGN contribution. However, a more extensive characterization awaits higher spatial resolution imaging and multi-wavelength photometry.

A careful analysis of the kinematics goes beyond the present Letter given the low spectral resolution of these data. However, the high S/N of the GTC/OSIRIS spectrum allows us to detect differences in the kinematics of the Lyα emission and interstellar features. Figure 3 (left panel) shows velocity plots of the Lyα emission line and several normalized interstellar absorption lines, relative to ${z}_{\mathrm{sys}}$. The Lyα emission has its peak redshifted with a velocity ${v}_{\mathrm{Ly}\alpha }\sim +200$ km s⁻¹, while the interstellar absorption lines appear blueshifted by $\sim -150$ km s⁻¹, which is consistent with galaxy-scale outflows of material from the galaxy in the form of a wind, similar to those seen in other star-forming galaxies at $z\sim 3$ (Shapley et al. 2003).

Zoom In Zoom Out Reset image size

We also notice that a spatial gradient of the Lyα velocity is present in the 2D spectrum (Figure 3, right upper panel). This velocity structure is seen both in A and B but appears mirrored, as expected for those images in this system, as A and B images straddle the fold critical curve (see Figure 1). We carefully checked the 2D spectrum to see if this pattern is present in other lines, but we found none with this signature. The 1D spatial distribution of Lyα is also compared with that of the rest-frame UV continuum (Figure 3, right lower panel). The Lyα emission appears more extended in the inner region between A and B, than the UV continuum. Deeper and higher spatial resolution observations are needed to constrain the Lyα and UV continuum spatial distributions.

To interpret the properties of the lensed system in more detail, we used the 330 s GTC/OSIRIS g-band image obtained in $\simeq 0\buildrel{\prime\prime}\over{.} 75$ FWHM seeing for accurate lens modeling. As shown in Figure 1, the lens system comprises four images forming a so-called ${\text{}}{fold}$ configuration, when the source lies very close to a ${\text{}}{fold}$ caustic. Similar to previous works (Bolton et al. 2008; Brownstein et al. 2012; Shu et al. 2015, 2016b, 2016c), we have developed a lens model using a nonlinear optimizer, consisting of minimizing a ${\chi }^{2}$ function using the Levenberg–Marquardt algorithm with the LMFIT package (Newville et al. 2014), where the observational data are compared to the model (see Shu et al. 2016b for details). The foreground-light is modeled using the elliptical Sérsic profile and similar to Shu et al. (2015, 2016b, 2016c), its subtraction is performed jointly with the lens modeling. The lens model includes a mass distribution of the foreground lens parameterized as a singular isothermal ellipsoid (SIE), and an additional external shear is included to model the higher-order effect from the environment. The surface brightness distribution of the background source is reconstructed parametrically using an elliptical Sérsic model. The foreground-light model is combined with the predicted lensed images and convolved with the point-spread function that was modeled with a star in the GTC/OSIRIS g-band field of view.

For the lens, the best-fit lens model (${\chi }^{2}/\mathrm{dof}=3494/3709$) predicts an Einstein radius ${b}_{\mathrm{SIE}}=2\buildrel{\prime\prime}\over{.} 95\,\pm \,0\buildrel{\prime\prime}\over{.} 10$, a minor-to-major axis ratio $q=0.34$, and a position angle ${\rm{P}}.{\rm{A}}.\,=\,165\buildrel{\circ}\over{.} 5$. The strength and position angle of the external shear are $\gamma =0.059\pm 0.008$ and ${\phi }_{\gamma }=89\buildrel{\circ}\over{.} 9$, respectively. The characteristic lensing velocity dispersion defined as ${\sigma }_{\mathrm{SIE}}=c\sqrt{\tfrac{{b}_{\mathrm{SIE}}}{4\pi }\tfrac{{D}_{{\rm{L}}}}{{D}_{\mathrm{LS}}}}$ is 390 ± 6 km s⁻¹, where ${D}_{\mathrm{LS}}$ and ${D}_{{\rm{S}}}$ are the angular diameter distances from the lens and the observer to the source, respectively. We also find a minor-to-major axis ratio of the SIE component (0.34), smaller than that of the light distribution suggested by the g-band model result (0.85). The lensing velocity dispersion suggests that cluster or line of sight structures also contribute a substantial fraction of convergence. There is no clear evidence of a crowded environment around the ETG, either by visual inspection of color images or in the SDSS photometric redshifts, but $\sim 1^{\prime}$ to the north there is a galaxy, SDSS J142953.71+120333.9, with a BOSS spectroscopic redshift $z=0.5527\pm 0.0002$, very close to the lensing ETG, indicating that a cluster or group of galaxies at $z\simeq 0.55$ may be present, as suggested by the external shear field.

For the source, the lens model gives local magnifications of 3.2, 3.1, 1.8, and 0.7 for images A, B, C, and D, respectively, which means that the total magnification is 8.8 ± 0.4. The source has an effective radius ${R}_{\mathrm{eff}}=0\buildrel{\prime\prime}\over{.} 159\pm 0\buildrel{\prime\prime}\over{.} 007$, which corresponds to ${R}_{\mathrm{eff}}=1.28\pm 0.06\,\mathrm{kpc}$ for the adopted cosmology. The source has a minor-to-major axis ratio q = 0.56 and a Sérsic index $n=3.9$. It is centered at ${\rm{\Delta }}{\rm{R}}.{\rm{A}}.\,=\,-0\buildrel{\prime\prime}\over{.} 43$ and ${\rm{\Delta }}\mathrm{decl}.\,=\,0\buildrel{\prime\prime}\over{.} 56$ relative to the center of the lens galaxy.

Having determined the magnification of the LAE we can estimate its intrinsic properties. From the total DECaLS DR2 r-band magnitude, we determine a rest-frame 1600 Å luminosity ${L}_{1600}=(6.12\pm 0.48)\times {10}^{30}$ erg s⁻¹ Hz⁻¹. Using Kennicutt's conversion (Kennicutt 1998), this rest-frame UV luminosity translates into an intrinsic star formation rate (SFR) of ≃90 M_⊙ yr⁻¹ when corrected for magnification, reddening, and the lower proportion of low-mass stars in the Chabrier (2003) stellar IMF relative to the standard Salpeter (1955) adopted by Kennicutt (a factor of 1/1.8). Turning to Lyα, assuming the space distribution of Lyα in the source plane and its lensing magnification are similar to that of the rest-frame UV continuum, and applying the correction for the magnification and the slit losses (the GTC/OSIRIS slit captured a fraction ≃0.60 of the total light of BG1429+1202), we derive an intrinsic Lyα luminosity of ${L}_{\mathrm{Ly}\alpha }=(2.80\pm 0.39)\,\times {10}^{43}$ erg s⁻¹. Assuming case-B recombination and Kennicutt's conversion (Kennicutt 1998), this luminosity results in  $\mathrm{SFR}(\mathrm{Ly}\alpha )\simeq 25$ M_⊙ yr⁻¹. Comparing the estimates of SFR from the rest-frame UV and Lyα, we measure ${f}_{\mathrm{esc}}^{\mathrm{Ly}\alpha }\sim 0.30$, consistent with that estimated from LAEs at $z\sim 3$ (e.g., Verhamme et al. 2008; Zheng et al. 2016). However, we should note that the measured ${f}_{\mathrm{esc}}^{\mathrm{Ly}\alpha }$ results from the assumption that the spatial distribution of Lyα (and its magnification) follows the rest-frame UV continuum. Lyα halos are hard to resolve from individual LAEs, and have been studied mainly by using stacking techniques (Steidel et al. 2011; Momose et al. 2016), or in nearby high-redshift analogs (e.g., Yang et al. 2016). However, for a few cases, strong gravitational lensing allows spatially resolved studies of high-redshift galaxies (e.g., Patrício et al. 2016). A more detailed analysis will be possible with high-resolution narrowband imaging and integral field spectroscopy.

In order to establish how typical the intrinsic properties of this galaxy are, we compare it with other UV-selected $z\sim 3$ LBGs and LAEs. BG1429+1202 is intrinsically more luminous in the rest-frame UV by factors of 7 and 19, relative to L^* from the luminosity functions of Reddy & Steidel (2009) for $z\sim 3$ LBGs and $z\sim 3.1$ LAEs selected by narrowband imaging by Ouchi et al. (2008), respectively. It is also intrinsically very luminous in $\mathrm{Ly}\alpha$ emission, when compared with ${L}_{\mathrm{Ly}\alpha }^{* }$ from the luminosity functions of LAEs at $z\sim 3.1$ (factor of ∼5; Ouchi et al. 2008) and LAEs at $z=2.8$ in CDFS (factor of $\sim 9-5;$ Zheng et al. 2016).

Table 2 presents a comparison with other well known, exceptionally bright in the optical, galaxy–galaxy $z\sim 3$ lenses: the Cosmic Eye (Smail et al. 2007), MS 1512-cB58 (Yee et al. 1996), the 8 o'clock (Allam et al. 2007), and at lower redshift, the Cosmic Horseshoe (Belokurov et al. 2007). BG1429+1202 has similar brightness but is intrinsically very luminous in the rest-frame UV continuum and $\mathrm{Ly}\alpha$. It is also the only $\mathrm{Ly}\alpha$ emitter (the $\mathrm{Ly}\alpha$ line of the Cosmic Horseshoe galaxy shares many of the properties of the $\mathrm{Ly}\alpha$ emitters, but its ${W}_{0}^{\mathrm{Ly}\alpha }$ is below the threshold generally adopted to define it as $\mathrm{Ly}\alpha$ emitter; Quider et al. 2009). This puts BG1429+1202 in the small group of very bright $z\sim 3$ galaxies that, due to high magnification and high intrinsic luminosity, have brightnesses that provide the unique opportunity to obtain high S/N spectroscopy, which allows us to study its physical properties in detail.

Table 2.  Intrinsic Properties of BG1429+1202 and Other Bright Galaxy–Galaxy Lenses

Object	z	${m}_{\mathrm{UV}}$ a	$\mu$ b	${L}_{\mathrm{UV}}$ c	${L}_{\mathrm{Ly}\alpha }$ c	References
		($\simeq 1500\mbox{--}1700\,\mathring{\rm A}$)		($\times {10}^{29}$ erg s−1 Hz−1)	($\times {10}^{42}$ erg s−1)
BG1429+1202	2.822	20.16	8.8	6.99	28.0	⋯
MS 1512-cB58	2.726	20.64	30	1.16	⋯	1, 2
Cosmic Eye	3.073	20.30	28	2.08	⋯	3
8 o'clock	2.735	19.22	12.3	10.55	⋯	4
Cosmic Horseshoe	2.381	19.70	24	2.74	3.3	5, 6
LBGs	∼3	24.61	⋯	1.06	⋯	7
LAEs	∼3.1	25.84	⋯	0.36	5.8	8

Notes.

^aRest-frame UV apparent magnitudes from the r- or i-bands, depending on the redshift. ^bTotal magnification factor. ^cIntrinsic rest-frame UV and $\mathrm{Ly}\alpha$ luminosity, respectively, corrected from the lensing magnification.

References. (1) Ellingson et al. (1996), (2) Seitz et al. (1998), (3) Smail et al. (2007), (4) Allam et al. (2007), (5) Belokurov et al. (2007), (6) Quider et al. (2009), (7) Reddy & Steidel (2009), (8) Ouchi et al. (2008).

Download table as:  ASCIITypeset image