# Direct Lyman continuum and Ly α escape observed at redshift 4

Direct Lyman continuum and Ly α escape observed at redshift 4 Abstract We report on the serendipitous discovery of a z = 4.0, M1500 = −22.20 star-forming galaxy (Ion3) showing copious Lyman continuum (LyC) leakage (∼60 per cent escaping), a remarkable multiple peaked Ly α emission, and significant Ly α radiation directly emerging at the resonance frequency. This is the highest redshift confirmed LyC emitter in which the ionizing and Ly α radiation possibly share a common ionized channel (with NH I < 1017.2 cm−2). Ion3 is spatially resolved, it shows clear stellar winds signatures like the P-Cygni N vλ1240 profile, and has blue ultraviolet continuum (β = −2.5 ± 0.25, Fλ ∼ λβ) with weak low-ionization interstellar metal lines. Deep VLT/HAWKI Ks and Spitzer/IRAC 3.6 and 4.5μm imaging show a clear photometric signature of the H α line with equivalent width of 1000 Å rest-frame emerging over a flat continuum (Ks − 4.5μm ≃ 0). From the SED fitting, we derive a stellar mass of 1.5 × 109 M⊙, SFR of 140 M⊙ yr−1 and age of ∼10 Myr, with a low dust extinction, E(B − V) ≲ 0.1, placing the source in the starburst region of the SFR−M* plane. Ion3 shows similar properties of another LyC emitter previously discovered (z = 3.21, Ion2, Vanzella et al. 2016). Ion3 (and Ion2) represents ideal high-redshift reference cases to guide the search for reionizing sources at z > 6.5 with JWST. gravitational lensing: strong, galaxies: formation, galaxies: starburst, ultraviolet: general 1 INTRODUCTION The definition of a reference sample of Lyman continuum (LyC) emitters at z ≲ 4.5 is crucial to guide the identification of the sources that reionized the Universe at z > 6.5, an epoch when the LyC is not directly observable (Worseck et al. 2014). Several LyC leakers showing escape fraction of $$f_{esc}^{abs} \sim 4{\rm -}15\,\,\rm{per\,\,cent}$$ have been confirmed in the nearby universe, z ∼ 0–0.3 (see Izotov et al. 2016a,b and references therein). Recently, Izotov et al. (2017) identified another LyC emitter with fesc ≃ 46 per cent, the highest value currently measured for the local sample. Most searches at high redshift have proved unsuccessful, due to contamination by foreground sources (Siana et al. 2015), or have yielded fairly stringent limits of fesc < 0.1 (e.g. Vanzella et al. 2012; Grazian et al. 2017). However, three LyC emitters have been confirmed at cosmological distances, 2.5 < z < 3.2 with fesc ∼ 30-100 per cent (de Barros et al. 2016; Shapley et al. 2016; Vanzella et al. 2016; Bian et al. 2017). As shown in Verhamme et al. (2017) the available sample of LyC leakers, both at low and high redshift, show very consistent observational features. Such features include the high Ly α equivalent width EW >40Å, a high ratio of [O iii]λ5007/ [O ii]λ3727, and narrow signatures in the Ly α emission like narrow double peaked profiles (<300–400 km s−1) or Ly α emission close to the systemic velocity (<100 km s−1). Also intense nebular optical lines [O iii]λλ4959, 5007+ H β with EW of 1000–1500 Årest-frame seem often associated with LyC emitters (e.g. Schaerer et al. 2016), as well as their compact star-forming region (Heckman et al. 2011) and the weakness of the low-ionization interstellar absorption lines (e.g. Jones et al. 2013; Chisholm et al. 2017). Overall, these quantities correlate with fesc, as was predicted by radiation transfer models (Verhamme et al. 2015) and photoionization models (e.g. Jaskot & Oey 2013; Zackrisson, Inoue & Jensen 2013; Nakajima & Ouchi 2014). In this work, we report on a serendipitously discovered LyC emitter at redshift 4, dubbed here Ion3, the highest redshift case currently known. We assume a flat cosmology with ΩM = 0.3, ΩΛ= 0.7, and H0 = 70 km s−1 Mpc−1. 2 DATA: FORS AND X-SHOOTER OBSERVATIONS Ion3 was discovered during a FORS2 spectroscopic program executed in visitor mode during the period 2017 September 15–19 (prog. 098.A-0804(B), P.I. Vanzella). Ion3 is a relatively bright object and was inserted as a filler in the MXU FORS mask (I-band magnitude 23.64 ± 0.38) and located at 3΄41΄ from the Frontier Field galaxy cluster AS1063 (Lotz et al. 2017), which lies outside the HST coverage. Given the large separation from the galaxy cluster the resulting magnification is low, a well-constrained μ = 1.15 ± 0.02 (or Δm = 0.15, Caminha et al. 2016). In the following, all the reported quantities are corrected for μ. The data reduction was carried on as described in several previous works (e.g. Vanzella et al. 2014) in which the AB-BA sky subtraction scheme was implemented. The final spectrum consists of 14 h integration with an average seeing of 0.8 arcsec and spectral resolution dv ≃ 600  km s−1 at λ = 7000 Å (R = λ/dλ = 500, grism 300V). Fig. 1 shows the FORS2 spectrum covering the wavelength range 3700–9300Å. During the same program 098.A-0804(B), an additional 1 h integration was obtained with the grism 600B, doubling the spectral resolution (dv ≃ 300  km s−1, R = 1000). We clearly confirmed the double peaked Ly α profile of Ion3 not well resolved with the 300V grating (see Fig. 1, top left). Additional four hours X-Shooter integration on Ion3 was subsequently obtained during 2017 November with an average seeing conditions of 0.8 arcsec (prog. 098.A-0665, P.I. Vanzella), providing a final spectrum spanning the range 3 400–24 000 Å with spectral resolution dv ≃ 35 − 60 km s−1. We refer the reader to Vanzella et al. (2014, 2017) for details about FORS and X-Shooter data reduction. Figure 1. View largeDownload slide The FORS spectrum at resolution dv ≃ 580  km s−1 (grism 300V) of Ion3 is shown along with the most relevant ultraviolet lines. In the top-left insets the Ly α spectra at resolution dv ≃ 300(580)  km s−1 obtained with the grism 600B(300V) are shown. The two-dimensional signal to noise and sky spectra are shown in the middle of the figure, in which the wavelength coverage up to 9300 Å is well sampled, despite the fringing pattern at λ > 7800 Å typical of the E2V blue-optimized CCD. In the bottom the the emerging LyC with S/N>10 is clearly detected. Figure 1. View largeDownload slide The FORS spectrum at resolution dv ≃ 580  km s−1 (grism 300V) of Ion3 is shown along with the most relevant ultraviolet lines. In the top-left insets the Ly α spectra at resolution dv ≃ 300(580)  km s−1 obtained with the grism 600B(300V) are shown. The two-dimensional signal to noise and sky spectra are shown in the middle of the figure, in which the wavelength coverage up to 9300 Å is well sampled, despite the fringing pattern at λ > 7800 Å typical of the E2V blue-optimized CCD. In the bottom the the emerging LyC with S/N>10 is clearly detected. 3 RESULTS 3.1 The Lyman continuum emission The most intriguing feature emerging from Ion3 is the LyC leakage at λ < 912 Å rest-frame detected in the FORS spectrum with S/N=6.4(11.1) if averaged over the interval 880–910(800–910Å), and corresponding to magnitude 27.5 (AB). Fig. 1 shows the two-dimensional spectrum with the LyC signal spatially aligned with the non-ionizing radiation. The possible presence of a foreground object at z < 4 mimicking the LyC signal represents a serious problem in this kind of observations (e.g. Vanzella et al. 2012; Siana et al. 2015), especially when HST imaging is not available. However, we can reasonably exclude such an interloper. First, we note that the probability chance of a superposition is low, less than 1 per cent adopting the above magnitude and the seeing of 0.8 arcsec (see Vanzella et al. 2010). Secondly, there is no trace of any spectral line arising from a foreground object, both in the deep FORS spectrum (that excludes [O ii]λ3727, 3729 at z < 1.45 and Ly α at z > 1.9) and wide X-Shooter spectrum, that easily would have captured several UV and/or optical rest-frame emission lines in the redshift range 0 < z < 4. Moreover, all the properties of Ion3 support a very low column density of neutral gas along the line of sight, making the entire picture consistent with the emerging LyC signal. From the FORS spectrum the observed fluxes at 900 and 1500 Årest-frame are $$F_{\lambda }^{900} = (5.6 \pm 0.08) \times 10^{-20}$$ erg s−1 cm−2 Å−1and $$F_{\lambda }^{1500} = (3.8 \pm 0.09)\times 10^{-19}$$ erg s−1 cm−2 Å−1, respectively, corresponding to a flux density ratio of fν(1500)/fν(900) = 19.1 ± 3.3. Following Vanzella et al. (2012), this ratio translates to a relative escape fraction fesc,rel = 20–100 per cent, assuming an intrinsic ratio of the luminosity densities Lν(1500)/Lν(900) = 1–5 and the median intergalactic medium (IGM) transmission at 900Å, $$<T_{{\rm IGM}}^{900}>=0.26$$ (with a central 68 per cent interval of 0.05–0.40, Inoue et al. (2014); Vanzella et al. (2015). Being a single line of sight the real $$T_{{\rm IGM}}^{900}$$ is unknown, therefore, any combination of Lν(1500)/Lν(900) > 1 and IGM transmission lower than 100 per cent produces fesc,rel in the range 10 per cent–100 per cent, with a fiducial value of 60 per cent adopting median $$<T_{{\rm IGM}}^{900}>=0.26$$ and Lν(1500)/Lν(900) = 3. The inferred ionizing photon production rate from $$F_{\lambda }^{900}$$ is Nphot(900) = 3.5 × 1053s−1, which compared to Starburst99 models for instantaneous bursts (Salpeter IMF; Salpeter 1955, and upper mass limit of 100 M⊙; Leitherer et al. 2014) yields a stellar mass involved in the starburst event of 4 × 106 M⊙ with the age not larger than 20 Myrs, and a number of O-type stars dominating the ionizing radiation of ≃ 1.6 × 104 (with uncertainties mainly dominated by the aforementioned IGM transmission). 3.2 The non-ionizing properties The continuum redward the Ly α line up to ≃ 1850 Å rest-frame is well detected in the FORS spectrum (Figs 1 and 2) with a best-fitting ultraviolet spectral slope of β = −2.50 with 68 per cent central interval [−2.73,−2.28] (Fλ ∼ λβ). Figure 2. View largeDownload slide The FORS spectrum is shown with the main spectral features reported at the systemic redshift (blue crosses). In particular, the P-Cygni N vλ1240, [C ii]*λ1335.71, [S iv]λ1393, 1402, and the He iiλ1640 lines are highlighted with thick lines. The N vλ1240 region (square) is zoomed and shown in the top-right inset (black line), in which three Starburst99 stellar models of 1, 10, and 20 Myr are superimposed to the spectrum (coloured lines). Figure 2. View largeDownload slide The FORS spectrum is shown with the main spectral features reported at the systemic redshift (blue crosses). In particular, the P-Cygni N vλ1240, [C ii]*λ1335.71, [S iv]λ1393, 1402, and the He iiλ1640 lines are highlighted with thick lines. The N vλ1240 region (square) is zoomed and shown in the top-right inset (black line), in which three Starburst99 stellar models of 1, 10, and 20 Myr are superimposed to the spectrum (coloured lines). Such a steep slope is not significantly affected by the atmospheric dispersion (the atmospheric dispersion compensator, LADC, is part of the system at UT1), and is fully consistent with being powered by massive and hot stars (O and early B stars), which are also responsible for the ionizing photons. The strongest spectroscopic signature of these stars is provided by the broad P-Cygni profiles in the resonance transitions of highly ionized species that arise in the stellar winds (e.g. Leitherer et al. 2014). In our FORS spectrum, a N vλ1240 P-Cygni profile is clearly present (see Fig. 2), observed also in local star-forming galaxies (e.g. Heckman et al. 2011, see also fig. 4 of Jaskot et al. 2017). In particular, the height and depth of the N vλ1240 line suggest an age of the burst of a few Myr. We superimposed a Starburst99 (Leitherer et al. 2014) spectral population synthesis model to the FORS spectrum with instantaneous burst with age 1, 10, and 20 Myr old, 0.4 solar metallicity (Salpeter IMF and 100 M⊙ upper mass limit), which reproduces the observations well (especially the 1–10 Myr old templates, Fig. 2). Low ionization interstellar absorption lines like [Si ii]λ1260, [O i]λ1303, and [C ii]λ1334 are very weak or even absent, indicating a very low gas covering fraction consistent with the measured LyC leakage (Heckman et al. 2011; Jones et al. 2013; Chisholm et al. 2017). Interestingly, Ion3 also shows the non-resonant fluorescent emission line [C ii]*λ1335.71 (see Fig. 2). Such a feature has been detected by Jaskot & Oey (2014) on local LyC candidates and interpreted as an evidence of the complex geometry of the neutral gas outside the line of sight, like anisotropic ionizing emission. The systemic redshift zsys has been derived from the [C ii]*λ1335.71 line and more accurately from the X-Shooter detection of the [O ii]λ3727, 3729 doublet, providing zsys = 3.999 ± 0.001. At such redshift, the doublet [N v]λλ1239, 1243 lies in the middle of its P-Cygni profile and the possible He iiλ1640 line is also recognized, though with low significance, 2.7σ. Table 1 summarizes the most relevant spectroscopic properties. Table 1. Spectral and physical properties of Ion3. Line fluxes are reported in units of 10−17  erg s−1 cm−2 (no slit losses are considered). Quantities are corrected for the lensing magnification μ = 1.15. σz is the redshift error on the last digit. Line/λvacuum  Flux(S/N)  z[σz], Resolution(km s−1)  Ly α(0) 1215.7  0.37(5)  3.9902[4], 35 (XSHO)  Ly α(1) 1215.7  1.16(12)  3.9950[3], 35 (XSHO)  Ly α(2) 1215.7  1.37(15)  3.9984[3], 35 (XSHO)  Ly α(3) 1215.7  6.23(32)  4.0033[3], 35 (XSHO)  Ly α(total)  9.13(73)  −, 580 (FORS)  [C ii]*λ1335.71  0.14(4)  4.000[4], 580 (FORS)  [He ii] λ1640.42  0.12(2.7)  4.000[4], 580 (FORS)  [O ii] λ3727.09  0.38(4.5)  3.999[1], 55 (XSHO)  [O ii] λ3729.88  0.25(3)  3.999[1], 55 (XSHO)  SED-fitting output  Value  Uncertainty  M(stellar) (×109 M⊙)  1.5  [1.4–2.2]  Age (Myr)  11  [10–20]  SFR [M⊙yr−1]  140  [110–150]  E(B − V)  ≃ 0.1  [0–0.1]  MUV(1500)  −22.20  ±0.15  H α EW(rest) (Å)  ≃ 1000  [700–1300]  Nphot(900Å)(s−1)  3.0 × 1053  [1053–54]  log10(ξion(Hz erg−1))  25.6  25.4–25.8  fesc, rel  0.60  [0.10–1.00]  Line/λvacuum  Flux(S/N)  z[σz], Resolution(km s−1)  Ly α(0) 1215.7  0.37(5)  3.9902[4], 35 (XSHO)  Ly α(1) 1215.7  1.16(12)  3.9950[3], 35 (XSHO)  Ly α(2) 1215.7  1.37(15)  3.9984[3], 35 (XSHO)  Ly α(3) 1215.7  6.23(32)  4.0033[3], 35 (XSHO)  Ly α(total)  9.13(73)  −, 580 (FORS)  [C ii]*λ1335.71  0.14(4)  4.000[4], 580 (FORS)  [He ii] λ1640.42  0.12(2.7)  4.000[4], 580 (FORS)  [O ii] λ3727.09  0.38(4.5)  3.999[1], 55 (XSHO)  [O ii] λ3729.88  0.25(3)  3.999[1], 55 (XSHO)  SED-fitting output  Value  Uncertainty  M(stellar) (×109 M⊙)  1.5  [1.4–2.2]  Age (Myr)  11  [10–20]  SFR [M⊙yr−1]  140  [110–150]  E(B − V)  ≃ 0.1  [0–0.1]  MUV(1500)  −22.20  ±0.15  H α EW(rest) (Å)  ≃ 1000  [700–1300]  Nphot(900Å)(s−1)  3.0 × 1053  [1053–54]  log10(ξion(Hz erg−1))  25.6  25.4–25.8  fesc, rel  0.60  [0.10–1.00]  View Large 3.3 The multiple peaked Ly α emission In the conventional scenario for Ly α emission, Ly α scatters many times before escaping, which significantly alters and broadens the original line profile. The kinematics, the column density, and geometry of the H i gas are the main ingredients that shape the Ly α emission (not to mention the dust attenuation). A prominent Ly α emission line with rest-frame EW = 40 ± 3 Å is clearly detected in the FORS spectrum (Fig. 1). Fig. 3 shows the same line at the X-Shooter spectral resolution (dv = 35  km s−1), in which we identify four emitting structures marked as 0, 1, 2, and 3. While the presence of blue peaks typically suggest a low column of H i gas (e.g. Henry et al. 2015; Yang et al. 2016), the emission at peak (2) is remarkable and emerges at the resonance frequency (z = 3.9984, i.e. ≲ 40 km s−1 from the systemic), where the opacity to Ly α photons would be the highest (Fig. 3). This is fully consistent with a scenario in which the LyC and (part of) the Ly α photons are escaping along the same optically thin direction (to LyC, $$N_{{\rm H\,\,{\small I}}} <10^{17.2}$$ cm−2) and likely from the same cavity (e.g. Zackrisson, Inoue & Jensen 2013; Behrens, Dijkstra & Niemeyer 2014; Verhamme et al. 2015). The LyC−Ly α escape through a ionized channel discussed in Behrens et al. (2014) consider the possibility that the gas is outflowing perpendicular to a galactic disc (and is reminiscent of a wind breaking out of a galactic disk). In this scenario, the quadruply peaked Ly α emission observed in Ion3 might be associated with a face-on disc. Ion3 represents the highest redshift empirical evidence of such a LyC−Ly α escaping mode. The Ly α escape degenerates with the H i column and the outflow velocity such that fast winds can mimic low columns (Verhamme et al. 2015). In this case, both the LyC emission and a relatively fast wind are detected. The evidence of an outflowing gas is imprinted in the blueshifted interstellar metal lines [S iv]λ1393.76 and [S iv]λ1402.77, with an average dv ≃ −400 ± 150 km s−1. This is also consistent with the well developed red tail of peak (3) possibly suggesting backscattering from the receding gas. Figure 3. View largeDownload slide Top: the one-dimensional X-Shooter spectrum of the Ly α is shown (at dv ≃ 35  km s−1) in the velocity space (with reported the relative velocities among the peaks,  km s−1). The systemic velocity is inferred from the [O ii]λ3727, 3729 doublet (top-right inset) and marked with the red cross. Bottom: the X-Shooter two-dimensional spectrum is shown with the four main structures identified. In the bottom-right inset the same Ly α line is shown at resolution dv ≃ 600  km s−1 as observed with FORS. Figure 3. View largeDownload slide Top: the one-dimensional X-Shooter spectrum of the Ly α is shown (at dv ≃ 35  km s−1) in the velocity space (with reported the relative velocities among the peaks,  km s−1). The systemic velocity is inferred from the [O ii]λ3727, 3729 doublet (top-right inset) and marked with the red cross. Bottom: the X-Shooter two-dimensional spectrum is shown with the four main structures identified. In the bottom-right inset the same Ly α line is shown at resolution dv ≃ 600  km s−1 as observed with FORS. Interestingly, a very similar direct Ly α escape has recently been identified in a lensed z = 2.37 galaxy by Rivera-Thorsen et al. (2017), in which the central peak is also at the systemic redshift and indicative of a possible perforated channel of very low H i optical depth. 3.4 Physical properties from the SED fitting and the nature of the ionizing radiation SED fitting has been performed using BC03 templates (Bruzual & Charlot 2003) including the nebular prescription and assuming exponentially declining star formation histories with e-folding time 0.1 < τ < 15 Gyr, (see Castellano et al. 2016 for details). It has been applied to the ground-based ESO/WFI imaging (B[842], V[843], R[844], and Ic[879], and deep ESO VLT/HAWKI Ks (obtained with 0.39 arcsec seeing, Brammer et al. 2016) and space-based Sptizer/IRAC 3.6 and 4.5μm bands1 (see Fig. 4). While Ion3 is detected at S/N ≲  5 in the ESO/WFI bands, the optical rest-frame continuum is detected with S/N ≳  10 in the Ks and IRAC bands. The most interesting features are the flat continuum at rest-frame wavelengths 4400 and 9000 Å, and the clear excess in the 3.6μm band consistent with an H α emission with rest-frame EW of 1000Å. The best-fitting solution implies Ion3 is a relatively low stellar mass (1.5 × 109M⊙) system undergoing a starburst phase (SFR ≃ 140 M⊙yr−1) consistently with the presence of prominent Ly α, N vλ1240 P-Cygni profile, strong H α and measured LyC. Ion3 appears as a still rapidly growing system with a specific star formation rate of ≃ 90 Gyr−1 (see Table 1 for a summary of the properties of Ion3). The photometric estimate of the H α line luminosity (≃ 2 × 1043 erg s−1) implies a high LyC photon production efficiency, ξion = 25.6 ± 0.2 (Hz erg−1). It resembles the values derived by Bouwens et al. (2016) for the bluest galaxies (β < −2.3, see also Shivaei et al. 2017), and consistent with the values reported from local candidate and confirmed LyC emitters, that also show large rest-frame EW(H α) ∼1000 Å (e.g. Izotov et al. 2017; Jaskot et al. 2017), and eventually similar to those inferred at z > 6 reported by Stark et al. (2017). It is also worth noting that Ion3 belongs to the starburst region of the z ∼ 4.5 SFR−M* bimodal distribution recently identified by Caputi et al. (2017). Based on current data, there is no evidence of, nor any need for, any contribution to the UV emission by an AGN. The relatively large ratios of Ly α/ N v (≃ 17 ± 2) and Ly α/ C iv ≳ 20 tend to exclude the presence of an obscured AGN (e.g. Alexandroff et al. 2013). A relatively shallow Chandra exposure of 130 ksec in the 0.5–2 kev band available in the field 1 yields a limit of 1.6 × 10−15  erg s−1 cm−2 (3σ), corresponding to an upper limit to the X-ray luminosity of 3 × 1043 erg s−1, which rules out a luminous AGN. Assuming a low-luminosity AGN is present, the very young burst detected would imply the ionizing photons are a mixture of stellar and non-stellar radiation escaping along a transparent medium ($$N_{{\rm H\,\,{\small I}}}<10^{17.2}$$ cm−2), that, however, should not be able to attenuate the expected high-ionization emission lines, making Ion3 either a very special case among the AGN category or a pure star-forming dominated object. Another possibility is that Ion3 has been captured just after the AGN has turned off, such that the optically thin channel produced by the previous nuclear activity enables the Ly α and LyC stellar radiation to escape towards the observer. Figure 4. View largeDownload slide The cutouts of Ion3 (top, 6.6 arcsec across) and the best SED fitted with only stellar (red) and stellar and nebular (black) templates are shown. The photometric jump at 3.6μm is evident. Figure 4. View largeDownload slide The cutouts of Ion3 (top, 6.6 arcsec across) and the best SED fitted with only stellar (red) and stellar and nebular (black) templates are shown. The photometric jump at 3.6μm is evident. 4 FINAL REMARKS Ion3 is a bright star-forming galaxy showing copious LyC leakage identified just 1.5 Gyr after the Big-Bang (z = 4). This makes Ion3 the highest redshift confirmed LyC emitter known so far. In particular, The spectral features and the SED-fitting suggest that Ion3 is a young, low-mass system undergoing a starburst phase containing hot and massive stars, with a specific star formation rate of 90 Gyr−1. The FORS and X-Shooter spectra reveal for the first time at such redshift a transparent ionized channel through which the Ly α photons escape at the resonance frequency, plausibly along the same path of the LyC photons. The absence of HST imaging prevents us from deriving any conclusion on the morphology of the galaxy. The image with the best seeing is the VLT/HAWKI Ks-band (0.39 arcsec) from which Ion3 appears marginally resolved with a FWHM ∼2–3 kpc proper. We refer the reader to a future work focused on the detailed analysis of the spatial distribution of the ionizing and non-ionizing radiation, as well as a proper radiative transfer modelling of the Ly α emission. However, it is worth noting that Ion3 shows similar properties of another LyC emitter we identified at z = 3.21 (dubbed Ion2, Vanzella et al. 2016), e.g. a structured Ly α shape, the blue UV slope and weak low-ionization interstellar absorption lines. While in the case of Ion2 we confirmed also very strong [O iii]λ5007 lines (EW>1000 Å rest-frame) and large O32 index ([O iii]λ5007/ [O ii]λ3727>10) making it among of the highest redshift Green Pea galaxy and suggesting a density-bounded condition, Ion3 might also have a perforated medium and will need JWST to probe the rest-frame optical wavelengths (imaging and spectroscopy) and HST to image directly the LyC. Irrespective of the nature of the ionizing radiation, Ion3 represents a unique high-redshift laboratory where ionized channels carved in the interstellar medium by one or more feedback sources can be studied. Ion3 and Ion2 represent ideal reference cases to guide the search for reionizing sources at z > 6.5 with JWST. ACKNOWLEDGEMENTS We thank the referee for constructive comments. EV gratefully acknowledge the excellent support by ESO staff at Paranal during the observations. We thank G. Zamorani, A. Jaskot, S. Oey, D. Schaerer and A. Grazian for useful discussions. FC, AM acknowledge funding from the INAF PRIN-SKA 2017 program 1.05.01.88.04. MM acknowledges support from the Italian Ministry of Foreign Affairs and International Cooperation, Directorate General for Country Promotion. Based on observations collected at the European Southern Observatory for Astronomical research in the Southern hemisphere under ESO programmes P098.A-0804(B), P098.A-0665(B). Footnotes 1 http://www.stsci.edu/hst/campaigns/frontier-fields/FF-Data REFERENCES Alexandroff R. et al.  , 2013, MNRAS , 435, 3306 https://doi.org/10.1093/mnras/stt1500 CrossRef Search ADS   Behrens C., Dijkstra M., Niemeyer J. C., 2014, A&A , 563, 77 CrossRef Search ADS   Bian F., Fan X., McGreer I., Cai Z., Jiang L., 2017, ApJ , 837, L12 https://doi.org/10.3847/2041-8213/aa5ff7 CrossRef Search ADS   Bouwens R., Smit R., Labbé I., Franx M., Caruana J., Oesch P., Stefanon M., Rasappu N., 2016, ApJ , 831, 176 https://doi.org/10.3847/0004-637X/831/2/176 CrossRef Search ADS   Brammer G. B. et al.  , 2016, ApJS , 226, 6 https://doi.org/10.3847/0067-0049/226/1/6 CrossRef Search ADS   Bruzual G., Charlot S., 2003, MNRAS , 344, 1000 https://doi.org/10.1046/j.1365-8711.2003.06897.x CrossRef Search ADS   Caminha G. B. et al.  , 2016, A&A , 587, 80 CrossRef Search ADS   Caputi K. I. et al.  , 2017, ApJ , 849, 45 https://doi.org/10.3847/1538-4357/aa901e CrossRef Search ADS   Castellano M. et al.  , 2016b, A&A , 590, A31 CrossRef Search ADS   Chisholm J., Orlitová I., Schaerer D., Verhamme A., Worseck G., Izotov Y. I., Thuan T. X., Guseva N. G., 2017, A&A , 605, 67 CrossRef Search ADS   de Barros S. et al.  , 2016, A&A , 585, A51 CrossRef Search ADS   Grazian A. et al.  , 2017, A&A , 602, A18 CrossRef Search ADS   Heckman T. M. et al.  , 2011, ApJ , 730, 5 https://doi.org/10.1088/0004-637X/730/1/5 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   Inoue A. K., Shimizu I., Iwata I., Tanaka M., 2014, MNRAS , 442, 1805 https://doi.org/10.1093/mnras/stu936 CrossRef Search ADS   Izotov Y. I., Schaerer D., Thuan T. X., Worseck G., Guseva N. G., Orlitová I., Verhamme A., 2016a, MNRAS , 461, 3683 https://doi.org/10.1093/mnras/stw1205 CrossRef Search ADS   Izotov Y. I., Orlitová I., Schaerer D., Thuan T. X., Verhamme A., Guseva N. G., Worseck G., 2016b, Nature , 529, 178 https://doi.org/10.1038/nature16456 CrossRef Search ADS   Izotov Y. I., Schaerer D., Worseck G., Guseva N. G., Thuan T. X., Verhamme A., Orlitova I., Fricke K. J., 2017, MNRAS , 474, 4514 CrossRef Search ADS   Jaskot A. E., Oey M. S., 2013, ApJ , 766, 91 https://doi.org/10.1088/0004-637X/766/2/91 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   Jaskot A. E., Oey M. S., Scarlata C., Dowd T., 2017, ApJ , 851, L6 CrossRef Search ADS   Jones T. A., Ellis R. S., Schenker M. A., Stark D. P., 2013, ApJ , 779, 52 https://doi.org/10.1088/0004-637X/779/1/52 CrossRef Search ADS   Leitherer C., Ekström S., Meynet G., Schaerer D., Agienko K. B., Levesque E. M., 2014, ApJS , 212, 14 https://doi.org/10.1088/0067-0049/212/1/14 CrossRef Search ADS   Lotz J. M. et al.  , 2017, ApJ , 837, 97 https://doi.org/10.3847/1538-4357/837/1/97 CrossRef Search ADS   Nakajima K., Ouchi M., 2014, MNRAS , 442, 900 https://doi.org/10.1093/mnras/stu902 CrossRef Search ADS   Rivera-Thorsen T. E. et al.  , 2017, A&A , 608, 4 CrossRef Search ADS   Salpeter E. E., 1955, ApJ , 121, 161 https://doi.org/10.1086/145971 CrossRef Search ADS   Schaerer D., Izotov Y. I., Verhamme A., Orlitová I., Thuan T. X., Worseck G., Guseva N. G., 2016, A&A , 591, 8 CrossRef Search ADS   Shapley A., Steidel C. C., Strom A. L., Bogosavljevic M., Reddy N. A., Siana B., Mostardi R. E., Rudie G. C., 2016, ApJ , 826, 24 https://doi.org/10.3847/2041-8205/826/2/L24 CrossRef Search ADS   Shivaei I. et al.  , 2017, preprint (arXiv:1711.00013) Siana B. et al.  , 2015, ApJ , 804, 17 https://doi.org/10.1088/0004-637X/804/1/17 CrossRef Search ADS   Stark D. P. et al.  , 2017, MNRAS , 464, 469 https://doi.org/10.1093/mnras/stw2233 CrossRef Search ADS   Vanzella E., Siana B., Cristiani S., Nonino M., 2010, MNRAS , 404, 1672 Vanzella E. et al.  , 2012, ApJ , 751, 70 https://doi.org/10.1088/0004-637X/751/1/70 CrossRef Search ADS   Vanzella E. et al.  , 2014, ApJ , 783, L12 https://doi.org/10.1088/2041-8205/783/1/L12 CrossRef Search ADS   Vanzella E. et al.  , 2015, A&A , 576, A116 CrossRef Search ADS   Vanzella E. et al.  , 2016, ApJ , 825, 41 https://doi.org/10.3847/0004-637X/825/1/41 CrossRef Search ADS   Vanzella E. et al.  , 2017, ApJ , 842, 47 https://doi.org/10.3847/1538-4357/aa74ae CrossRef Search ADS   Verhamme A., Orlitová I., Schaerer D., Hayes M., 2015, A&A , 578, A7 CrossRef Search ADS   Verhamme A., Orlitová I., Schaerer D., Izotov Y., Worseck G., Thuan T. X., Guseva N., 2017, A&A , 597, 13 CrossRef Search ADS   Worseck G. et al.  , 2014, MNRAS , 445, 1745 https://doi.org/10.1093/mnras/stu1827 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   Zackrisson E., Inoue A. K., Jensen H., 2013, ApJ , 777, 39 https://doi.org/10.1088/0004-637X/777/1/39 CrossRef Search ADS   © 2018 The Author(s) Published by Oxford University Press on behalf of the Royal Astronomical Society http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Monthly Notices of the Royal Astronomical Society: Letters Oxford University Press

# Direct Lyman continuum and Ly α escape observed at redshift 4

, Volume 476 (1) – May 1, 2018
5 pages

/lp/ou_press/direct-lyman-continuum-and-ly-escape-observed-at-redshift-4-eddOZHITpR
Publisher
journal_eissn:11745-3933
© 2018 The Author(s) Published by Oxford University Press on behalf of the Royal Astronomical Society
ISSN
1745-3925
eISSN
1745-3933
D.O.I.
10.1093/mnrasl/sly023
Publisher site
See Article on Publisher Site

### Abstract

Abstract We report on the serendipitous discovery of a z = 4.0, M1500 = −22.20 star-forming galaxy (Ion3) showing copious Lyman continuum (LyC) leakage (∼60 per cent escaping), a remarkable multiple peaked Ly α emission, and significant Ly α radiation directly emerging at the resonance frequency. This is the highest redshift confirmed LyC emitter in which the ionizing and Ly α radiation possibly share a common ionized channel (with NH I < 1017.2 cm−2). Ion3 is spatially resolved, it shows clear stellar winds signatures like the P-Cygni N vλ1240 profile, and has blue ultraviolet continuum (β = −2.5 ± 0.25, Fλ ∼ λβ) with weak low-ionization interstellar metal lines. Deep VLT/HAWKI Ks and Spitzer/IRAC 3.6 and 4.5μm imaging show a clear photometric signature of the H α line with equivalent width of 1000 Å rest-frame emerging over a flat continuum (Ks − 4.5μm ≃ 0). From the SED fitting, we derive a stellar mass of 1.5 × 109 M⊙, SFR of 140 M⊙ yr−1 and age of ∼10 Myr, with a low dust extinction, E(B − V) ≲ 0.1, placing the source in the starburst region of the SFR−M* plane. Ion3 shows similar properties of another LyC emitter previously discovered (z = 3.21, Ion2, Vanzella et al. 2016). Ion3 (and Ion2) represents ideal high-redshift reference cases to guide the search for reionizing sources at z > 6.5 with JWST. gravitational lensing: strong, galaxies: formation, galaxies: starburst, ultraviolet: general 1 INTRODUCTION The definition of a reference sample of Lyman continuum (LyC) emitters at z ≲ 4.5 is crucial to guide the identification of the sources that reionized the Universe at z > 6.5, an epoch when the LyC is not directly observable (Worseck et al. 2014). Several LyC leakers showing escape fraction of $$f_{esc}^{abs} \sim 4{\rm -}15\,\,\rm{per\,\,cent}$$ have been confirmed in the nearby universe, z ∼ 0–0.3 (see Izotov et al. 2016a,b and references therein). Recently, Izotov et al. (2017) identified another LyC emitter with fesc ≃ 46 per cent, the highest value currently measured for the local sample. Most searches at high redshift have proved unsuccessful, due to contamination by foreground sources (Siana et al. 2015), or have yielded fairly stringent limits of fesc < 0.1 (e.g. Vanzella et al. 2012; Grazian et al. 2017). However, three LyC emitters have been confirmed at cosmological distances, 2.5 < z < 3.2 with fesc ∼ 30-100 per cent (de Barros et al. 2016; Shapley et al. 2016; Vanzella et al. 2016; Bian et al. 2017). As shown in Verhamme et al. (2017) the available sample of LyC leakers, both at low and high redshift, show very consistent observational features. Such features include the high Ly α equivalent width EW >40Å, a high ratio of [O iii]λ5007/ [O ii]λ3727, and narrow signatures in the Ly α emission like narrow double peaked profiles (<300–400 km s−1) or Ly α emission close to the systemic velocity (<100 km s−1). Also intense nebular optical lines [O iii]λλ4959, 5007+ H β with EW of 1000–1500 Årest-frame seem often associated with LyC emitters (e.g. Schaerer et al. 2016), as well as their compact star-forming region (Heckman et al. 2011) and the weakness of the low-ionization interstellar absorption lines (e.g. Jones et al. 2013; Chisholm et al. 2017). Overall, these quantities correlate with fesc, as was predicted by radiation transfer models (Verhamme et al. 2015) and photoionization models (e.g. Jaskot & Oey 2013; Zackrisson, Inoue & Jensen 2013; Nakajima & Ouchi 2014). In this work, we report on a serendipitously discovered LyC emitter at redshift 4, dubbed here Ion3, the highest redshift case currently known. We assume a flat cosmology with ΩM = 0.3, ΩΛ= 0.7, and H0 = 70 km s−1 Mpc−1. 2 DATA: FORS AND X-SHOOTER OBSERVATIONS Ion3 was discovered during a FORS2 spectroscopic program executed in visitor mode during the period 2017 September 15–19 (prog. 098.A-0804(B), P.I. Vanzella). Ion3 is a relatively bright object and was inserted as a filler in the MXU FORS mask (I-band magnitude 23.64 ± 0.38) and located at 3΄41΄ from the Frontier Field galaxy cluster AS1063 (Lotz et al. 2017), which lies outside the HST coverage. Given the large separation from the galaxy cluster the resulting magnification is low, a well-constrained μ = 1.15 ± 0.02 (or Δm = 0.15, Caminha et al. 2016). In the following, all the reported quantities are corrected for μ. The data reduction was carried on as described in several previous works (e.g. Vanzella et al. 2014) in which the AB-BA sky subtraction scheme was implemented. The final spectrum consists of 14 h integration with an average seeing of 0.8 arcsec and spectral resolution dv ≃ 600  km s−1 at λ = 7000 Å (R = λ/dλ = 500, grism 300V). Fig. 1 shows the FORS2 spectrum covering the wavelength range 3700–9300Å. During the same program 098.A-0804(B), an additional 1 h integration was obtained with the grism 600B, doubling the spectral resolution (dv ≃ 300  km s−1, R = 1000). We clearly confirmed the double peaked Ly α profile of Ion3 not well resolved with the 300V grating (see Fig. 1, top left). Additional four hours X-Shooter integration on Ion3 was subsequently obtained during 2017 November with an average seeing conditions of 0.8 arcsec (prog. 098.A-0665, P.I. Vanzella), providing a final spectrum spanning the range 3 400–24 000 Å with spectral resolution dv ≃ 35 − 60 km s−1. We refer the reader to Vanzella et al. (2014, 2017) for details about FORS and X-Shooter data reduction. Figure 1. View largeDownload slide The FORS spectrum at resolution dv ≃ 580  km s−1 (grism 300V) of Ion3 is shown along with the most relevant ultraviolet lines. In the top-left insets the Ly α spectra at resolution dv ≃ 300(580)  km s−1 obtained with the grism 600B(300V) are shown. The two-dimensional signal to noise and sky spectra are shown in the middle of the figure, in which the wavelength coverage up to 9300 Å is well sampled, despite the fringing pattern at λ > 7800 Å typical of the E2V blue-optimized CCD. In the bottom the the emerging LyC with S/N>10 is clearly detected. Figure 1. View largeDownload slide The FORS spectrum at resolution dv ≃ 580  km s−1 (grism 300V) of Ion3 is shown along with the most relevant ultraviolet lines. In the top-left insets the Ly α spectra at resolution dv ≃ 300(580)  km s−1 obtained with the grism 600B(300V) are shown. The two-dimensional signal to noise and sky spectra are shown in the middle of the figure, in which the wavelength coverage up to 9300 Å is well sampled, despite the fringing pattern at λ > 7800 Å typical of the E2V blue-optimized CCD. In the bottom the the emerging LyC with S/N>10 is clearly detected. 3 RESULTS 3.1 The Lyman continuum emission The most intriguing feature emerging from Ion3 is the LyC leakage at λ < 912 Å rest-frame detected in the FORS spectrum with S/N=6.4(11.1) if averaged over the interval 880–910(800–910Å), and corresponding to magnitude 27.5 (AB). Fig. 1 shows the two-dimensional spectrum with the LyC signal spatially aligned with the non-ionizing radiation. The possible presence of a foreground object at z < 4 mimicking the LyC signal represents a serious problem in this kind of observations (e.g. Vanzella et al. 2012; Siana et al. 2015), especially when HST imaging is not available. However, we can reasonably exclude such an interloper. First, we note that the probability chance of a superposition is low, less than 1 per cent adopting the above magnitude and the seeing of 0.8 arcsec (see Vanzella et al. 2010). Secondly, there is no trace of any spectral line arising from a foreground object, both in the deep FORS spectrum (that excludes [O ii]λ3727, 3729 at z < 1.45 and Ly α at z > 1.9) and wide X-Shooter spectrum, that easily would have captured several UV and/or optical rest-frame emission lines in the redshift range 0 < z < 4. Moreover, all the properties of Ion3 support a very low column density of neutral gas along the line of sight, making the entire picture consistent with the emerging LyC signal. From the FORS spectrum the observed fluxes at 900 and 1500 Årest-frame are $$F_{\lambda }^{900} = (5.6 \pm 0.08) \times 10^{-20}$$ erg s−1 cm−2 Å−1and $$F_{\lambda }^{1500} = (3.8 \pm 0.09)\times 10^{-19}$$ erg s−1 cm−2 Å−1, respectively, corresponding to a flux density ratio of fν(1500)/fν(900) = 19.1 ± 3.3. Following Vanzella et al. (2012), this ratio translates to a relative escape fraction fesc,rel = 20–100 per cent, assuming an intrinsic ratio of the luminosity densities Lν(1500)/Lν(900) = 1–5 and the median intergalactic medium (IGM) transmission at 900Å, $$<T_{{\rm IGM}}^{900}>=0.26$$ (with a central 68 per cent interval of 0.05–0.40, Inoue et al. (2014); Vanzella et al. (2015). Being a single line of sight the real $$T_{{\rm IGM}}^{900}$$ is unknown, therefore, any combination of Lν(1500)/Lν(900) > 1 and IGM transmission lower than 100 per cent produces fesc,rel in the range 10 per cent–100 per cent, with a fiducial value of 60 per cent adopting median $$<T_{{\rm IGM}}^{900}>=0.26$$ and Lν(1500)/Lν(900) = 3. The inferred ionizing photon production rate from $$F_{\lambda }^{900}$$ is Nphot(900) = 3.5 × 1053s−1, which compared to Starburst99 models for instantaneous bursts (Salpeter IMF; Salpeter 1955, and upper mass limit of 100 M⊙; Leitherer et al. 2014) yields a stellar mass involved in the starburst event of 4 × 106 M⊙ with the age not larger than 20 Myrs, and a number of O-type stars dominating the ionizing radiation of ≃ 1.6 × 104 (with uncertainties mainly dominated by the aforementioned IGM transmission). 3.2 The non-ionizing properties The continuum redward the Ly α line up to ≃ 1850 Å rest-frame is well detected in the FORS spectrum (Figs 1 and 2) with a best-fitting ultraviolet spectral slope of β = −2.50 with 68 per cent central interval [−2.73,−2.28] (Fλ ∼ λβ). Figure 2. View largeDownload slide The FORS spectrum is shown with the main spectral features reported at the systemic redshift (blue crosses). In particular, the P-Cygni N vλ1240, [C ii]*λ1335.71, [S iv]λ1393, 1402, and the He iiλ1640 lines are highlighted with thick lines. The N vλ1240 region (square) is zoomed and shown in the top-right inset (black line), in which three Starburst99 stellar models of 1, 10, and 20 Myr are superimposed to the spectrum (coloured lines). Figure 2. View largeDownload slide The FORS spectrum is shown with the main spectral features reported at the systemic redshift (blue crosses). In particular, the P-Cygni N vλ1240, [C ii]*λ1335.71, [S iv]λ1393, 1402, and the He iiλ1640 lines are highlighted with thick lines. The N vλ1240 region (square) is zoomed and shown in the top-right inset (black line), in which three Starburst99 stellar models of 1, 10, and 20 Myr are superimposed to the spectrum (coloured lines). Such a steep slope is not significantly affected by the atmospheric dispersion (the atmospheric dispersion compensator, LADC, is part of the system at UT1), and is fully consistent with being powered by massive and hot stars (O and early B stars), which are also responsible for the ionizing photons. The strongest spectroscopic signature of these stars is provided by the broad P-Cygni profiles in the resonance transitions of highly ionized species that arise in the stellar winds (e.g. Leitherer et al. 2014). In our FORS spectrum, a N vλ1240 P-Cygni profile is clearly present (see Fig. 2), observed also in local star-forming galaxies (e.g. Heckman et al. 2011, see also fig. 4 of Jaskot et al. 2017). In particular, the height and depth of the N vλ1240 line suggest an age of the burst of a few Myr. We superimposed a Starburst99 (Leitherer et al. 2014) spectral population synthesis model to the FORS spectrum with instantaneous burst with age 1, 10, and 20 Myr old, 0.4 solar metallicity (Salpeter IMF and 100 M⊙ upper mass limit), which reproduces the observations well (especially the 1–10 Myr old templates, Fig. 2). Low ionization interstellar absorption lines like [Si ii]λ1260, [O i]λ1303, and [C ii]λ1334 are very weak or even absent, indicating a very low gas covering fraction consistent with the measured LyC leakage (Heckman et al. 2011; Jones et al. 2013; Chisholm et al. 2017). Interestingly, Ion3 also shows the non-resonant fluorescent emission line [C ii]*λ1335.71 (see Fig. 2). Such a feature has been detected by Jaskot & Oey (2014) on local LyC candidates and interpreted as an evidence of the complex geometry of the neutral gas outside the line of sight, like anisotropic ionizing emission. The systemic redshift zsys has been derived from the [C ii]*λ1335.71 line and more accurately from the X-Shooter detection of the [O ii]λ3727, 3729 doublet, providing zsys = 3.999 ± 0.001. At such redshift, the doublet [N v]λλ1239, 1243 lies in the middle of its P-Cygni profile and the possible He iiλ1640 line is also recognized, though with low significance, 2.7σ. Table 1 summarizes the most relevant spectroscopic properties. Table 1. Spectral and physical properties of Ion3. Line fluxes are reported in units of 10−17  erg s−1 cm−2 (no slit losses are considered). Quantities are corrected for the lensing magnification μ = 1.15. σz is the redshift error on the last digit. Line/λvacuum  Flux(S/N)  z[σz], Resolution(km s−1)  Ly α(0) 1215.7  0.37(5)  3.9902[4], 35 (XSHO)  Ly α(1) 1215.7  1.16(12)  3.9950[3], 35 (XSHO)  Ly α(2) 1215.7  1.37(15)  3.9984[3], 35 (XSHO)  Ly α(3) 1215.7  6.23(32)  4.0033[3], 35 (XSHO)  Ly α(total)  9.13(73)  −, 580 (FORS)  [C ii]*λ1335.71  0.14(4)  4.000[4], 580 (FORS)  [He ii] λ1640.42  0.12(2.7)  4.000[4], 580 (FORS)  [O ii] λ3727.09  0.38(4.5)  3.999[1], 55 (XSHO)  [O ii] λ3729.88  0.25(3)  3.999[1], 55 (XSHO)  SED-fitting output  Value  Uncertainty  M(stellar) (×109 M⊙)  1.5  [1.4–2.2]  Age (Myr)  11  [10–20]  SFR [M⊙yr−1]  140  [110–150]  E(B − V)  ≃ 0.1  [0–0.1]  MUV(1500)  −22.20  ±0.15  H α EW(rest) (Å)  ≃ 1000  [700–1300]  Nphot(900Å)(s−1)  3.0 × 1053  [1053–54]  log10(ξion(Hz erg−1))  25.6  25.4–25.8  fesc, rel  0.60  [0.10–1.00]  Line/λvacuum  Flux(S/N)  z[σz], Resolution(km s−1)  Ly α(0) 1215.7  0.37(5)  3.9902[4], 35 (XSHO)  Ly α(1) 1215.7  1.16(12)  3.9950[3], 35 (XSHO)  Ly α(2) 1215.7  1.37(15)  3.9984[3], 35 (XSHO)  Ly α(3) 1215.7  6.23(32)  4.0033[3], 35 (XSHO)  Ly α(total)  9.13(73)  −, 580 (FORS)  [C ii]*λ1335.71  0.14(4)  4.000[4], 580 (FORS)  [He ii] λ1640.42  0.12(2.7)  4.000[4], 580 (FORS)  [O ii] λ3727.09  0.38(4.5)  3.999[1], 55 (XSHO)  [O ii] λ3729.88  0.25(3)  3.999[1], 55 (XSHO)  SED-fitting output  Value  Uncertainty  M(stellar) (×109 M⊙)  1.5  [1.4–2.2]  Age (Myr)  11  [10–20]  SFR [M⊙yr−1]  140  [110–150]  E(B − V)  ≃ 0.1  [0–0.1]  MUV(1500)  −22.20  ±0.15  H α EW(rest) (Å)  ≃ 1000  [700–1300]  Nphot(900Å)(s−1)  3.0 × 1053  [1053–54]  log10(ξion(Hz erg−1))  25.6  25.4–25.8  fesc, rel  0.60  [0.10–1.00]  View Large 3.3 The multiple peaked Ly α emission In the conventional scenario for Ly α emission, Ly α scatters many times before escaping, which significantly alters and broadens the original line profile. The kinematics, the column density, and geometry of the H i gas are the main ingredients that shape the Ly α emission (not to mention the dust attenuation). A prominent Ly α emission line with rest-frame EW = 40 ± 3 Å is clearly detected in the FORS spectrum (Fig. 1). Fig. 3 shows the same line at the X-Shooter spectral resolution (dv = 35  km s−1), in which we identify four emitting structures marked as 0, 1, 2, and 3. While the presence of blue peaks typically suggest a low column of H i gas (e.g. Henry et al. 2015; Yang et al. 2016), the emission at peak (2) is remarkable and emerges at the resonance frequency (z = 3.9984, i.e. ≲ 40 km s−1 from the systemic), where the opacity to Ly α photons would be the highest (Fig. 3). This is fully consistent with a scenario in which the LyC and (part of) the Ly α photons are escaping along the same optically thin direction (to LyC, $$N_{{\rm H\,\,{\small I}}} <10^{17.2}$$ cm−2) and likely from the same cavity (e.g. Zackrisson, Inoue & Jensen 2013; Behrens, Dijkstra & Niemeyer 2014; Verhamme et al. 2015). The LyC−Ly α escape through a ionized channel discussed in Behrens et al. (2014) consider the possibility that the gas is outflowing perpendicular to a galactic disc (and is reminiscent of a wind breaking out of a galactic disk). In this scenario, the quadruply peaked Ly α emission observed in Ion3 might be associated with a face-on disc. Ion3 represents the highest redshift empirical evidence of such a LyC−Ly α escaping mode. The Ly α escape degenerates with the H i column and the outflow velocity such that fast winds can mimic low columns (Verhamme et al. 2015). In this case, both the LyC emission and a relatively fast wind are detected. The evidence of an outflowing gas is imprinted in the blueshifted interstellar metal lines [S iv]λ1393.76 and [S iv]λ1402.77, with an average dv ≃ −400 ± 150 km s−1. This is also consistent with the well developed red tail of peak (3) possibly suggesting backscattering from the receding gas. Figure 3. View largeDownload slide Top: the one-dimensional X-Shooter spectrum of the Ly α is shown (at dv ≃ 35  km s−1) in the velocity space (with reported the relative velocities among the peaks,  km s−1). The systemic velocity is inferred from the [O ii]λ3727, 3729 doublet (top-right inset) and marked with the red cross. Bottom: the X-Shooter two-dimensional spectrum is shown with the four main structures identified. In the bottom-right inset the same Ly α line is shown at resolution dv ≃ 600  km s−1 as observed with FORS. Figure 3. View largeDownload slide Top: the one-dimensional X-Shooter spectrum of the Ly α is shown (at dv ≃ 35  km s−1) in the velocity space (with reported the relative velocities among the peaks,  km s−1). The systemic velocity is inferred from the [O ii]λ3727, 3729 doublet (top-right inset) and marked with the red cross. Bottom: the X-Shooter two-dimensional spectrum is shown with the four main structures identified. In the bottom-right inset the same Ly α line is shown at resolution dv ≃ 600  km s−1 as observed with FORS. Interestingly, a very similar direct Ly α escape has recently been identified in a lensed z = 2.37 galaxy by Rivera-Thorsen et al. (2017), in which the central peak is also at the systemic redshift and indicative of a possible perforated channel of very low H i optical depth. 3.4 Physical properties from the SED fitting and the nature of the ionizing radiation SED fitting has been performed using BC03 templates (Bruzual & Charlot 2003) including the nebular prescription and assuming exponentially declining star formation histories with e-folding time 0.1 < τ < 15 Gyr, (see Castellano et al. 2016 for details). It has been applied to the ground-based ESO/WFI imaging (B[842], V[843], R[844], and Ic[879], and deep ESO VLT/HAWKI Ks (obtained with 0.39 arcsec seeing, Brammer et al. 2016) and space-based Sptizer/IRAC 3.6 and 4.5μm bands1 (see Fig. 4). While Ion3 is detected at S/N ≲  5 in the ESO/WFI bands, the optical rest-frame continuum is detected with S/N ≳  10 in the Ks and IRAC bands. The most interesting features are the flat continuum at rest-frame wavelengths 4400 and 9000 Å, and the clear excess in the 3.6μm band consistent with an H α emission with rest-frame EW of 1000Å. The best-fitting solution implies Ion3 is a relatively low stellar mass (1.5 × 109M⊙) system undergoing a starburst phase (SFR ≃ 140 M⊙yr−1) consistently with the presence of prominent Ly α, N vλ1240 P-Cygni profile, strong H α and measured LyC. Ion3 appears as a still rapidly growing system with a specific star formation rate of ≃ 90 Gyr−1 (see Table 1 for a summary of the properties of Ion3). The photometric estimate of the H α line luminosity (≃ 2 × 1043 erg s−1) implies a high LyC photon production efficiency, ξion = 25.6 ± 0.2 (Hz erg−1). It resembles the values derived by Bouwens et al. (2016) for the bluest galaxies (β < −2.3, see also Shivaei et al. 2017), and consistent with the values reported from local candidate and confirmed LyC emitters, that also show large rest-frame EW(H α) ∼1000 Å (e.g. Izotov et al. 2017; Jaskot et al. 2017), and eventually similar to those inferred at z > 6 reported by Stark et al. (2017). It is also worth noting that Ion3 belongs to the starburst region of the z ∼ 4.5 SFR−M* bimodal distribution recently identified by Caputi et al. (2017). Based on current data, there is no evidence of, nor any need for, any contribution to the UV emission by an AGN. The relatively large ratios of Ly α/ N v (≃ 17 ± 2) and Ly α/ C iv ≳ 20 tend to exclude the presence of an obscured AGN (e.g. Alexandroff et al. 2013). A relatively shallow Chandra exposure of 130 ksec in the 0.5–2 kev band available in the field 1 yields a limit of 1.6 × 10−15  erg s−1 cm−2 (3σ), corresponding to an upper limit to the X-ray luminosity of 3 × 1043 erg s−1, which rules out a luminous AGN. Assuming a low-luminosity AGN is present, the very young burst detected would imply the ionizing photons are a mixture of stellar and non-stellar radiation escaping along a transparent medium ($$N_{{\rm H\,\,{\small I}}}<10^{17.2}$$ cm−2), that, however, should not be able to attenuate the expected high-ionization emission lines, making Ion3 either a very special case among the AGN category or a pure star-forming dominated object. Another possibility is that Ion3 has been captured just after the AGN has turned off, such that the optically thin channel produced by the previous nuclear activity enables the Ly α and LyC stellar radiation to escape towards the observer. Figure 4. View largeDownload slide The cutouts of Ion3 (top, 6.6 arcsec across) and the best SED fitted with only stellar (red) and stellar and nebular (black) templates are shown. The photometric jump at 3.6μm is evident. Figure 4. View largeDownload slide The cutouts of Ion3 (top, 6.6 arcsec across) and the best SED fitted with only stellar (red) and stellar and nebular (black) templates are shown. The photometric jump at 3.6μm is evident. 4 FINAL REMARKS Ion3 is a bright star-forming galaxy showing copious LyC leakage identified just 1.5 Gyr after the Big-Bang (z = 4). This makes Ion3 the highest redshift confirmed LyC emitter known so far. In particular, The spectral features and the SED-fitting suggest that Ion3 is a young, low-mass system undergoing a starburst phase containing hot and massive stars, with a specific star formation rate of 90 Gyr−1. The FORS and X-Shooter spectra reveal for the first time at such redshift a transparent ionized channel through which the Ly α photons escape at the resonance frequency, plausibly along the same path of the LyC photons. The absence of HST imaging prevents us from deriving any conclusion on the morphology of the galaxy. The image with the best seeing is the VLT/HAWKI Ks-band (0.39 arcsec) from which Ion3 appears marginally resolved with a FWHM ∼2–3 kpc proper. We refer the reader to a future work focused on the detailed analysis of the spatial distribution of the ionizing and non-ionizing radiation, as well as a proper radiative transfer modelling of the Ly α emission. However, it is worth noting that Ion3 shows similar properties of another LyC emitter we identified at z = 3.21 (dubbed Ion2, Vanzella et al. 2016), e.g. a structured Ly α shape, the blue UV slope and weak low-ionization interstellar absorption lines. While in the case of Ion2 we confirmed also very strong [O iii]λ5007 lines (EW>1000 Å rest-frame) and large O32 index ([O iii]λ5007/ [O ii]λ3727>10) making it among of the highest redshift Green Pea galaxy and suggesting a density-bounded condition, Ion3 might also have a perforated medium and will need JWST to probe the rest-frame optical wavelengths (imaging and spectroscopy) and HST to image directly the LyC. Irrespective of the nature of the ionizing radiation, Ion3 represents a unique high-redshift laboratory where ionized channels carved in the interstellar medium by one or more feedback sources can be studied. Ion3 and Ion2 represent ideal reference cases to guide the search for reionizing sources at z > 6.5 with JWST. ACKNOWLEDGEMENTS We thank the referee for constructive comments. EV gratefully acknowledge the excellent support by ESO staff at Paranal during the observations. We thank G. Zamorani, A. Jaskot, S. Oey, D. Schaerer and A. Grazian for useful discussions. FC, AM acknowledge funding from the INAF PRIN-SKA 2017 program 1.05.01.88.04. MM acknowledges support from the Italian Ministry of Foreign Affairs and International Cooperation, Directorate General for Country Promotion. Based on observations collected at the European Southern Observatory for Astronomical research in the Southern hemisphere under ESO programmes P098.A-0804(B), P098.A-0665(B). Footnotes 1 http://www.stsci.edu/hst/campaigns/frontier-fields/FF-Data REFERENCES Alexandroff R. et al.  , 2013, MNRAS , 435, 3306 https://doi.org/10.1093/mnras/stt1500 CrossRef Search ADS   Behrens C., Dijkstra M., Niemeyer J. C., 2014, A&A , 563, 77 CrossRef Search ADS   Bian F., Fan X., McGreer I., Cai Z., Jiang L., 2017, ApJ , 837, L12 https://doi.org/10.3847/2041-8213/aa5ff7 CrossRef Search ADS   Bouwens R., Smit R., Labbé I., Franx M., Caruana J., Oesch P., Stefanon M., Rasappu N., 2016, ApJ , 831, 176 https://doi.org/10.3847/0004-637X/831/2/176 CrossRef Search ADS   Brammer G. B. et al.  , 2016, ApJS , 226, 6 https://doi.org/10.3847/0067-0049/226/1/6 CrossRef Search ADS   Bruzual G., Charlot S., 2003, MNRAS , 344, 1000 https://doi.org/10.1046/j.1365-8711.2003.06897.x CrossRef Search ADS   Caminha G. B. et al.  , 2016, A&A , 587, 80 CrossRef Search ADS   Caputi K. I. et al.  , 2017, ApJ , 849, 45 https://doi.org/10.3847/1538-4357/aa901e CrossRef Search ADS   Castellano M. et al.  , 2016b, A&A , 590, A31 CrossRef Search ADS   Chisholm J., Orlitová I., Schaerer D., Verhamme A., Worseck G., Izotov Y. I., Thuan T. X., Guseva N. G., 2017, A&A , 605, 67 CrossRef Search ADS   de Barros S. et al.  , 2016, A&A , 585, A51 CrossRef Search ADS   Grazian A. et al.  , 2017, A&A , 602, A18 CrossRef Search ADS   Heckman T. M. et al.  , 2011, ApJ , 730, 5 https://doi.org/10.1088/0004-637X/730/1/5 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   Inoue A. K., Shimizu I., Iwata I., Tanaka M., 2014, MNRAS , 442, 1805 https://doi.org/10.1093/mnras/stu936 CrossRef Search ADS   Izotov Y. I., Schaerer D., Thuan T. X., Worseck G., Guseva N. G., Orlitová I., Verhamme A., 2016a, MNRAS , 461, 3683 https://doi.org/10.1093/mnras/stw1205 CrossRef Search ADS   Izotov Y. I., Orlitová I., Schaerer D., Thuan T. X., Verhamme A., Guseva N. G., Worseck G., 2016b, Nature , 529, 178 https://doi.org/10.1038/nature16456 CrossRef Search ADS   Izotov Y. I., Schaerer D., Worseck G., Guseva N. G., Thuan T. X., Verhamme A., Orlitova I., Fricke K. J., 2017, MNRAS , 474, 4514 CrossRef Search ADS   Jaskot A. E., Oey M. S., 2013, ApJ , 766, 91 https://doi.org/10.1088/0004-637X/766/2/91 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   Jaskot A. E., Oey M. S., Scarlata C., Dowd T., 2017, ApJ , 851, L6 CrossRef Search ADS   Jones T. A., Ellis R. S., Schenker M. A., Stark D. P., 2013, ApJ , 779, 52 https://doi.org/10.1088/0004-637X/779/1/52 CrossRef Search ADS   Leitherer C., Ekström S., Meynet G., Schaerer D., Agienko K. B., Levesque E. M., 2014, ApJS , 212, 14 https://doi.org/10.1088/0067-0049/212/1/14 CrossRef Search ADS   Lotz J. M. et al.  , 2017, ApJ , 837, 97 https://doi.org/10.3847/1538-4357/837/1/97 CrossRef Search ADS   Nakajima K., Ouchi M., 2014, MNRAS , 442, 900 https://doi.org/10.1093/mnras/stu902 CrossRef Search ADS   Rivera-Thorsen T. E. et al.  , 2017, A&A , 608, 4 CrossRef Search ADS   Salpeter E. E., 1955, ApJ , 121, 161 https://doi.org/10.1086/145971 CrossRef Search ADS   Schaerer D., Izotov Y. I., Verhamme A., Orlitová I., Thuan T. X., Worseck G., Guseva N. G., 2016, A&A , 591, 8 CrossRef Search ADS   Shapley A., Steidel C. C., Strom A. L., Bogosavljevic M., Reddy N. A., Siana B., Mostardi R. E., Rudie G. C., 2016, ApJ , 826, 24 https://doi.org/10.3847/2041-8205/826/2/L24 CrossRef Search ADS   Shivaei I. et al.  , 2017, preprint (arXiv:1711.00013) Siana B. et al.  , 2015, ApJ , 804, 17 https://doi.org/10.1088/0004-637X/804/1/17 CrossRef Search ADS   Stark D. P. et al.  , 2017, MNRAS , 464, 469 https://doi.org/10.1093/mnras/stw2233 CrossRef Search ADS   Vanzella E., Siana B., Cristiani S., Nonino M., 2010, MNRAS , 404, 1672 Vanzella E. et al.  , 2012, ApJ , 751, 70 https://doi.org/10.1088/0004-637X/751/1/70 CrossRef Search ADS   Vanzella E. et al.  , 2014, ApJ , 783, L12 https://doi.org/10.1088/2041-8205/783/1/L12 CrossRef Search ADS   Vanzella E. et al.  , 2015, A&A , 576, A116 CrossRef Search ADS   Vanzella E. et al.  , 2016, ApJ , 825, 41 https://doi.org/10.3847/0004-637X/825/1/41 CrossRef Search ADS   Vanzella E. et al.  , 2017, ApJ , 842, 47 https://doi.org/10.3847/1538-4357/aa74ae CrossRef Search ADS   Verhamme A., Orlitová I., Schaerer D., Hayes M., 2015, A&A , 578, A7 CrossRef Search ADS   Verhamme A., Orlitová I., Schaerer D., Izotov Y., Worseck G., Thuan T. X., Guseva N., 2017, A&A , 597, 13 CrossRef Search ADS   Worseck G. et al.  , 2014, MNRAS , 445, 1745 https://doi.org/10.1093/mnras/stu1827 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   Zackrisson E., Inoue A. K., Jensen H., 2013, ApJ , 777, 39 https://doi.org/10.1088/0004-637X/777/1/39 CrossRef Search ADS   © 2018 The Author(s) Published by Oxford University Press on behalf of the Royal Astronomical Society

### Journal

Monthly Notices of the Royal Astronomical Society: LettersOxford University Press

Published: May 1, 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
discover and read the research
that matters to you.

over 12 million articles from more than
10,000 peer-reviewed journals.

All for just $49/month ### Explore the DeepDyve Library ### Unlimited reading Read as many articles as you need. Full articles with original layout, charts and figures. Read online, from anywhere. ### Stay up to date Keep up with your field with Personalized Recommendations and Follow Journals to get automatic updates. ### Organize your research It’s easy to organize your research with our built-in tools. ### 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. ### Monthly Plan • Read unlimited articles • Personalized recommendations • No expiration • Print 20 pages per month • 20% off on PDF purchases • Organize your research • Get updates on your journals and topic searches$49/month

14-day Free Trial

Best Deal — 39% off

### Annual Plan

• All the features of the Professional Plan, but for 39% off!
• Billed annually
• No expiration
• For the normal price of 10 articles elsewhere, you get one full year of unlimited access to articles.

$588$360/year

billed annually

14-day Free Trial