Deep Hubble Space Telescope Imaging on the Extended Lyα Emission of a QSO at z = 2.19 with a Damped Lyman Alpha System as a Natural Coronagraph

Figure 1 shows the spectroscopic data of J1154–0215 from Fathivavsari et al. (2016). The median S/N around the Lyα emission is ∼3 pixel⁻¹ and the resolution is (R ∼ 4000). The reduction and data analysis process for the final spectrum are detailed in Fathivavsari et al. (2016). The MagE spectroscopy allows us to measure the metal lines associated with this PDLA. The metal-line diagnostic will constrain the distance between the absorbers and the QSO (see details in Section 4.1).

Zoom In Zoom Out Reset image size

We use six orbits (about 17,000 s) on the FQ387N narrowband observation of the QSO J1145–0215, which allows us to detect the Lyα flux at the 15σ level in the FQ387N filter.

We distribute our observations into two individual visits, each with three orbits. A standard three-point dither sequence is applied to populate each orbit. We conduct the deep HST/WFC3 narrowband FQ387N imaging of J1154–0215 as explained above. The redshift for the Lyα emission ${z}_{\mathrm{em}}$ = 2.181 and the redshift for the absorber, ${z}_{\mathrm{abs}}$ = 2.1853, which are measured in Fathivavsari et al. (2016). In Figure 1, we plot the spectrum with the filter response curve of the FQ387N filter (${\lambda }_{c}=3874\ \mathring{\rm A}$, FWHM $=\,34\ \mathring{\rm A}$) and the Voigt profile fitting for the DLA. The filter lies within the DLA trough and the Lyα emission is included in the filter.

The data reduction is conducted using WFC3/UVIS (Koekemoer et al. 2002), and the detailed procedures follow Cai et al. (2011, 2015). To optimize the output data quality, we choose a final output pixel scale of 003 instead of an initial pixel scale 004 and final pixfrac parameter 0.7 (shrinking pixel area) after different trials of combinations of parameters. The final output images we obtained from HST for J1145–0215 are shown in the upper left panel of Figure 2.

Zoom In Zoom Out Reset image size

In our HST narrowband image, we detect spatially resolved, diffuse Lyα emission. Using SExtractor (Bertin & Arnouts 1996), we measure the Lyα flux using an elliptical aperture with a Kron factor of 2.5 (0075) (magenta elliptical in upper left panel in Figure 2). The flux density of the source is ${F}_{\mathrm{FQ}387{\rm{N}}}$ = $6.2\pm 0.2\times {10}^{-18}\,\mathrm{erg}\,{{\rm{s}}}^{-1}\,{\mathrm{cm}}^{-2}\,{\mathring{\rm A} }^{-1}$ (total flux ${F}_{\mathrm{total}}$ = $2.11\,\pm 0.07\times {10}^{-16}\,\mathrm{erg}\,{{\rm{s}}}^{-1}\,{\mathrm{cm}}^{-2}$), corresponding to an AB magnitude of 22.68 ± 0.04.

In the upper left panel of Figure 2 we overplot the QSO position (dark point) with its error (green dashed circle). Note that our HST narrowband completely resides in the DLA dark trough. Thus, we do not have QSO continuum flux from our HST observations. The SDSS provides the QSO continuum position. We use two stars in the HST field to correct the offset between the SDSS and the HST coordinate systems (assuming the rotational offset is zero). We find that the offset between HST and SDSS is 009 ± 01. We also examine the morphology of this source. The HST image and the circularly averaged surface brightness (SB) profile shown in the lower left panel of Figure 2 suggest that we detect two resolved components, and we use two Sérsic components (right in Figure 2) to fit J1154–0215. We measure an effective radius of 0.29 ± 0.05 kpc for the major component and a radius of 1.52 ± 0.6 kpc for the minor component. The contour of the SB of the source from the Sérsic components model fitting is shown in the upper right panel of Figure 2. The residual of the fitting for this model is shown in the lower right panel of Figure 2. We also use two Gaussian components to repeat the above fitting and recover similar results of one major component and one minor component. The detailed parameters of the two models are shown in Table 1.

For the MagE spectra from Fathivavsari et al. (2016), we plot the spectra with the filter response curve of the FQ387N filter and the Voigt profile fitting with the error for the DLA in Figure 1. The spectral resolution is 75 km s⁻¹. We attain the H i column density for the DLA as log(${N}_{\mathrm{HI}})$ = 21.76 ± 0.1, consistent with the value fitted by Fathivavsari et al. (2016). We further fit a double-Gaussian profile to the Lyα spectrum inside the DLA dark trough. The integrated Lyα emission flux is ${F}_{\mathrm{Ly}\alpha }$ = $1.56\pm 0.10\times {10}^{-16}\,\mathrm{erg}\,{{\rm{s}}}^{-1}\,{\mathrm{cm}}^{-2}$ from this spectrum. Moreover, we also detect and fit metal lines Si ii 1260, Si* ii 1264 in this spectrum (shown in Figure 3, with a detailed analysis in Section 4.1).

Zoom In Zoom Out Reset image size

From the high-resolution HST FQ387N imaging, we do not detect Lyα emission in the central QSO region, and we now refer to this region as a Lyα "void." This may be explained by the fact that a kiloparsec-scale optically thick cloud is located in the region of the Lyα emission "void." We construct a power-law model without DLA absorption for the SB profile centered at the quasar position (labeled in red in Figure 4). We then add the absorption from the DLA to the model and find the best-fitted model to the data of the SB profile⁹ (model labeled in green and data labeled in blue in Figure 4). The two models (with and without DLA implemented) converge at r ∼ 1.5 kpc. We thus infer that the cloud size optically thick to the Lyα emission is ∼1.5 kpc. Our conclusion also agrees with the physical extent (<10 kpc) of log(${N}_{\mathrm{HI}}$) ≥ 20.3 absorption constrained by lensed quasar pairs in Cooke et al. (2010). The physical picture of the QSO, DLA cloud, and the "void" of Lyα emission are illustrated in the left panel of Figure 5.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

From the last section, we also detect absorption lines, Si ii 1260 and Si* ii 1264 from the MagE spectrum arising from the ground state n and the excited state ${n}^{* }$ of Si⁺. For a single, homogeneous cloud, the column density ratio between the lines from the excited state and the lines from the ground state is equal to the corresponding volume densities ratio of the ions/atoms (Bahcall 1967; Silva & Viegas 2002; Prochaska et al. 2006), namely, $\tfrac{{n}^{* }}{n}$ = $\tfrac{{N}^{* }}{N}$. This ratio may be determined by the ratio of the column density between Si* ii 1264 and Si ii 1260. Since the Si ii 1260 line is saturated, we will put a lower limit on its column density. Since the b value and column density will be degenerated when fitting the line Si ii 1260, we test a range of Doppler parameter (b) value and use the maximum b value (b = 30 km s⁻¹) to obtain a lower limit of the column density of $\mathrm{log}(N/{\mathrm{cm}}^{-2})$ = 15.3 for Si ii 1260. We then fit Si* ii 1264 with a fixed b value = 30 km s⁻¹ and attain the column density of $\mathrm{log}(N/{\mathrm{cm}}^{-2})$ = 13.62 ± 0.13.¹⁰ The results of the Voigt profile fitting of these absorption lines are shown in Figure 3.

If we assume the excited level is populated by indirect pumping of far-UV photons, Prochaska et al. (2006) derived a relation between $\tfrac{{n}^{* }}{n}$ (see Figure 7 in Prochaska et al. (2006)) and the far-UV flux. We can derive that d = $\sqrt{\tfrac{L}{{F}_{{uv}}4\pi }}$, where L is the quasar luminosity in the far-UV range (8 eV < hν < 13.6 eV), and ${F}_{{\text{}}{uv}}$ is the far-UV flux determined from $\tfrac{{n}^{* }}{n}$ based on the relation of population ratio and intensity from Prochaska et al. (2006). The quasar luminosity L is computed by integrating a local power law through the far-UV frequency range ($\int a{\nu }^{-{\alpha }_{Q}}{\text{}}d\nu$), where a is a constant. The local power law in this integral is estimated by the SDSS u band and SDSS g band photometric result. We finally get L = 1.33 × 10⁴⁵ erg ${{\rm{s}}}^{-1}$. We infer ${I}_{{\text{}}{uv}}$ = 32 erg ${{\rm{s}}}^{-1}\,{\mathrm{cm}}^{-2}$ from the ratio of the column density estimated by the pair of Si⁺ lines. Based on the equation d = $\sqrt{\tfrac{L}{{F}_{{\text{}}{uv}}4\pi }}$, we finally get an upper limit of d $\leqslant$ 0.6 kpc, with a lower limit on the column density of Si ii 1260, which is consistent with the fact that this target has the highest Al iii/Si ii ratio reported for a PDLA in Fathivavsari et al. (2016), indicating that it may be close to the active galactic nucleus. The physical picture of the QSO and DLA position is illustrated in the right panel of Figure 5

Combining the photometric and ${Galfit}$ fitting results of the final combined image with the reduced 1D spectrum, there are three main possible origins for the emitting gas for the Lyα emission: (1) emission from star-forming regions in the QSO host galaxy; (2) emission from star-forming regions in the DLA galaxy; (3) fluorescent Lyα emission from the optically thick gas cloud.

4.2.1. QSO Host Galaxies

4.2.2. DLA Galaxies

4.2.3. Fluorescent Lyα Emission

The third possibility for the origin of the Lyα emission is fluorescent Lyα emission from the optically thick gas that is close to the DLA cloud or from a system detached from it. Ionizing photons intercepted by the gas ionize neutral hydrogen atoms and the subsequent recombination has a high probability ($\eta \sim 0.6;$ e.g., Gould & Weinberg 1996) of ending up as Lyα photons.

We can estimate the SB I_α of the fluorescent Lyα emission through $\pi {I}_{\alpha }=h{\nu }_{\alpha }\eta {{fF}}_{i}/{\left(1+z\right)}^{4}$, where $h{\nu }_{\alpha }$ is the Lyα photon energy and the ${\left(1+z\right)}^{4}$ factor accounts for the cosmological dimming. For isotropic quasar emission, the ionizing photon flux F_i at the cloud position is ${F}_{i}=\int {L}_{\nu }/(h\nu )d\nu /(4\pi {d}^{2})$, with L_ν being the quasar luminosity and d being the distance from the quasar to the gas cloud. The integral goes from the Lyman limit ${\nu }_{L}=13.6\,\mathrm{eV}/h$ to infinity. The factor f is introduced to scale the line-of-sight (LOS) flux to that along the quasar-cloud direction, accounting for the anisotropy of the quasar emission of ionizing photons. With the assumption that ${L}_{\nu }={L}_{{\nu }_{L}}{\left(\nu /{\nu }_{L}\right)}^{-\alpha }$, we have

$$\begin{eqnarray}&&{I}_{\alpha }=\displaystyle \frac{1}{{\left(1+z\right)}^{4}}\displaystyle \frac{1}{4{\pi }^{2}{d}^{2}}\displaystyle \frac{\eta f}{\alpha }\displaystyle \frac{{\nu }_{\alpha }}{{\nu }_{L}}{\nu }_{L}{L}_{{\nu }_{L}}.\end{eqnarray} \tag{ 1 }$$

The quasar has ${\nu }_{L}{L}_{{\nu }_{L}}=0.76\times {10}^{45}\,\mathrm{erg}\,{{\rm{s}}}^{-1}$ with $\alpha =1.57$. For $d\sim 0.6\,\mathrm{kpc}$ and $f\sim 0.1$, we obtain the Lyα SB to be at the level of $0.8\times {10}^{-13}\,\mathrm{erg}\,{{\rm{s}}}^{-1}\,{\mathrm{cm}}^{-2}\,{\mathrm{arcsec}}^{-2}$, which is about two orders of magnitude higher than the observed SB shown in Figure 2. To match the observation, we need to put the cloud much farther way from the quasar (e.g., $d\gt 10\,\mathrm{kpc}$ along the LOS) or to have extremely anisotropic emission of the quasar (e.g., $f\lt {10}^{-3}$).

The other difficulty faced by the fluorescent origin of the Lyα emission is the shape of the spectrum. Fluorescent Lyα photons emerging from the thin skin layer of the optically thick cloud would typically have a double-peak profile (e.g., Gould & Weinberg 1996; Zheng & Miralda-Escudé 2002; Adelberger et al. 2006), with peak separation about eight times the velocity dispersion in the cloud. If the velocity dispersion in the cloud is of the order of $50\,\mathrm{km}\,{{\rm{s}}}^{-1}$ (e.g., Adelberger et al. 2006), the two peaks (with separation $\sim 400\,\mathrm{km}\,{{\rm{s}}}^{-1}$) would be well resolved, which does not seem to be the case in the spectrum shown in Figure 1. The tentatively detected blue peak with low flux could indicate gas kinematics, like outflowing, in the cloud.

Overall, based on the Lyα SB and the Lyα line profile, we find that the fluorescent origin of the Lyα emission is not favored.