The Keck Lyman Continuum Spectroscopic Survey (KLCS): The Emergent Ionizing Spectrum of Galaxies at z ∼ 3^∗

Abstract

We present results of a deep spectroscopic survey quantifying the statistics of the escape of hydrogen-ionizing photons from star-forming galaxies at z ∼ 3. The Keck Lyman Continuum Spectroscopic Survey (KLCS) includes spectra of 124 galaxies with z = 3.05 ± 0.18 and −22.1 ≤ M uv ≤ −19.5, observed in 9 independent fields, covering a common rest-wavelength range 880 ≤ λ 0 /Å < ∼ 1750. We measure the ratio of ionizing to non-ionizing UV flux density f 900 / f 1500 obs , where f 900 is the mean flux density evaluated over the range λ 0 = [880, 910] Å. To quantify f 900 / f 1500 out -the emergent ratio of ionizing to non-ionizing UV flux density -we use detailed Monte Carlo modeling of the opacity of H I in the intergalactic (IGM) and circumgalactic (CGM) medium as a function of source redshift. By analyzing high-S/N composite spectra formed from sub-samples exhibiting common observed properties and numbers sufficient to reduce the uncertainty in the IGM+CGM correction, we obtain precise values of f 900 / f 1500 out , including a full-sample average f 900 / f 1500 out = 0.057 ± 0.006. We further show that f 900 / f 1500 out increases monotonically with Lyα rest equivalent width W λ (Lyα), inducing an inverse correlation with UV luminosity as a by-product. To connect LyC leakage to intrinsic galaxy properties, we fit the composite spectra using stellar population synthesis (SPS) together with simple models of the ISM in which a fraction f c of the stellar continuum is covered by optically-thick gas with column density N HI . We show that the composite spectra simultaneously constrain the intrinsic properties of the ionizing stars (L 900 /L 1500 ) int along with f c , N HI , E(B − V ), and f esc,abs , the escape fraction of ionizing photons. We find a sample-averaged f esc,abs = 0.09 ± 0.01, and that subsamples fall along a linear relation f esc,abs 0.75 W λ (Lyα)/(110 Å) for 0 ≤ W λ (Lyα) < ∼ 60; subsamples with W λ (Lyα) < 0 have f esc,abs consistent with zero. We use the FUV luminosity function, the distribution function n[W λ (Lyα)], and the relationship between W λ (Lyα) and f 900 / f 1500 out to estimate the total ionizing emissivity of z ∼ 3 star-forming galaxies with M UV ≤ −19.5: LyC 6 × 10 24 ergs s −1 Hz −1 Mpc −3 . This value exceeds the contribution of QSOs by a factor of ∼ 1.2 − 3.7, and accounts for > ∼ 50% of the total LyC estimated using indirect methods at z ∼ 3. 11 According to Becker et al. (2011), continuum blanketing from the Lyman α forest increases ∝ (1 + z) 2.8 over the redshift range 2.1 < ∼ zs < ∼ 5.5.

Modelling a bright z = 6 galaxy at the faint end of the AGN luminosity function

“…The three dashes on the right axis indicate the average escape fraction measured after t > 750 Myr, when the SFR remains constant. On average, the three runs present a very similar behaviour, with a fairly low average escape fraction of f esc ∼ 5 − 7%, consistent with the recent results of Steidel et al (2018) on a sample of LBGs at z ∼ 3 and with the detailed simulation of Yoo et al (2020) of an isolated galaxy of similar mass. A 6 As we perform RHD simulations, we could in principle measure the ratio of the total ionizing flux crossing the virial radius divided by the intrinsic ionizing production, as e.g.…”

Outside the Lyman-break box: detecting Lyman continuum emitters at 3.5 &lt; z &lt; 5.1 with CLAUDS

“…In the last few years, the link between the properties of the Lyα line, the escape of the Lyα photons, and the escape of the LyC photons among low and high-redshift galaxies has been explored by a number of independent studies. For example, the correlation between EW(Lyα) and f esc among low redshift galaxies is reported by Verhamme et al (2017); Steidel et al (2018);Fletcher et al (2019) and an indication of a possible trend among EW(Lyα) and R obs is discussed by Marchi et al (2017). In Figure 10, we present rest frame EW(Lyα) as a function of estimated f abs esc shown as black horizontal lines (left panel) for three spectroscopically confirmed LyC leakers and for our two q1 candidates.…”

“…From models we assume (L LyC /L 1500 ) int varies in the range of ∼ 0.1 − 0.9 and depends on several galactic parameters like star formation history, stellar initial mass function, stellar age, and metallicity. The attenuation factor τ LyC I G M is usually determined by analytic models or MC simulations (e.g., Inoue et al 2014;Steidel et al 2018). Finally, we directly measure (F LyC /F 1500 ) obs in this work.…”

“…To account for the fact that any detection of LyC radiation at z > 3.6 is likely to arise from a line of sight that is more transparent than the average transmission, we also adopt the maximum transmission ( 1−D b ) estimated for the redshift range 2.7 z 5, from Steidel et al (2018). The quantity D b is defined as a mean depression in the rest frame continuum interval 920Å -1015Å caused by Lyman line blanketing (Oke & Korycansky 1982).…”

“…Therefore, we use IGM transmission values that describe lines of sight with higher transparency for LyC radiation. For this purpose, we adopt the estimated maximum IGM+CGM transmission values from Steidel et al (2018). A shortcoming of this approach is that the estimated IGM+CGM transmission is only available for the restframe 880Å λ rest 900Å interval and the chance for not having high column density absorber (> 10 16 cm 2 ) at the redshifts of our candidates is less likely.…”