The First Gamma-Ray Emitting BL Lacertae Object at the Cosmic Dawn

The observations with OSIRIS were performed on 2020 May 26 (seeing ∼13), using the R1000B grism, which provides an intermediate spectral resolution R ∼ 1018, with a 2.06 Å pixel⁻¹ sampling, and a spatial scale of 0508 pixel⁻¹ (with a 2 × 2 binning). The total exposure on-source was 7200 s divided in three individual exposures of 1200 s each. Bias correction, trimming, rotation, and flat-fielding were performed using Python scripts. Cosmic rays were removed using the package cleanest (Cardiel 2020), which allows for an interactive supervision and interpolation of the affected pixels. Wavelength calibration, C- and S-distortion correction, spectrum extraction, and atmospheric extinction correction were carried out using the ${{\rm{R}}}_{{\rm{E}}}{\rm{D}}\mathop{{\rm{m}}}\limits^{\mathrm{uc}}{\rm{E}}$ package.⁹ Because the target was placed in different positions along the slit in each observing block, this allowed a first-order sky subtraction correction by subtracting the average image from each block. In addition, we employed the xnirspec tool (Cardiel et al. 2003) to perform a second-order sky subtraction by computing an averaged sky-spectrum of residuals, which were subtracted by accounting for the specific image distortion at each pixel.

Finally, the absolute flux calibration was obtained using the observation of the spectrophotometric standard star GD153. By fitting the upper boundary of the stellar continuum using the package boundfit (Cardiel 2009), we derived not only the overall response curve, but also a normalized telluric absorption spectrum. A final flux correction factor of 1.32 was derived from the photometric calibration of the acquisition image using available Sloan $r^{\prime}$ data. This factor accounts for light losses due to the seeing being slightly larger than the slit width.

The near-infrared observations with EMIR were carried out on 2020 June 18–19 (seeing ∼11), using the K grism, which provides and intermediate spectral resolution R ∼ 4000, with a 1.73 Å pixel⁻¹ sampling, and a spatial scale of 01945 pixel⁻¹. Two observing blocks consisting in three ABBA sequences/block with an exposure time of 360 s per individual frame, accounted for a total exposure time of 8640 s. The spatial separation between the A and B exposures were 74. The initial data reduction was carried out using PyEmir (Cardiel et al. 2019; Pascual et al. 2019), a dedicated pipeline specifically created for the reduction of EMIR data, including bad-pixel masking, flat-fielding, image rectification, and wavelength calibration. In addition to this, we also performed the absolute flux calibration, following the same procedure employed with the OSIRIS data. In this case the spectrophotometric standard star HD116405 was used. The target spectrum was accordingly corrected for telluric absorption, and a final flux correction factor of 1.11, to account for light losses in the slit, was computed using the acquisition image calibrated with available K_s 2MASS photometric data.

Both OSIRIS and EMIR spectra were corrected for Galactic extinction (E(B − V) = 0.015) following Fitzpatrick (1999) and using R_V = 3.1.

The optical and NIR spectra of J1219 are shown in Figure 2. For a comparison, we also overplot the SDSS spectrum. By comparing the wavelength dependent signal-to-noise ratio (S/N), one can notice the high S/N of the OSIRIS data, which is ∼3 times better than the SDSS one. Because the EMIR spectrum is relatively noisy, a re-binning of 50 pixels was done to achieve a S/N of ∼5 per pixel at the center of the spectrum.

Zoom In Zoom Out Reset image size

Both OSIRIS and EMIR spectra do not exhibit any emission lines suggesting the BL Lac nature of J1219 and, more importantly, the first γ-ray emitting BL Lac source identified beyond z = 3. Considering the beginning of the wavelength drop at ∼5552 Å as the Lyα break, the source redshift can be constrained as z ∼ 3.57. However, the onset is not only set by the redshift but also by proximity effects in the neutral gas distribution and by the spectral resolution. One would expect a large proximity zone of many Mpc of ionized gas around the quasar (due to the quasar), so that the Lyα forest will appear to start at a lower redshift than the quasar itself (see Dall'Aglio et al. 2008). Therefore, the actual source redshift could be somewhere between 3.5 and 3.6. Based on the visual inspection of the SDSS spectrum, Ahumada et al. (2020) reported the redshift of J1219 as z = 3.59 and the same we have adopted in our analysis.

We highlight the expected location of a few prominent broad emission lines, e.g., C iv, in Figure 2 though none of them are detected. One can notice a narrow emission line very close to the expected position of Lyα emission. This feature, however, is the residual of the bright [O i] sky line, which was subtracted during the OSIRIS data reduction. Furthermore, on a comparison with the SDSS spectrum taken on 2011 May 28, the continuum flux level was found to be lower by a factor of ∼3 during GTC observations. This finding suggests the absence of emission lines even during the low jet activity. This, in turn, hints a radiatively inefficient accretion process and hence strengthens the BL Lac nature of J1219.

We used the OSIRIS spectrum to determine the upper limit on the EW of the C iv emission line. The spectrum was brought to the rest frame and fitted with a power law to reproduce the continuum in the wavelength range [1500, 1600] Å. Additionally, the C iv emission line was modeled as a Gaussian function with a variable EW while keeping FWHM fixed to 4000 km s⁻¹, a value typical for blazars (e.g., Shaw et al. 2012; Landoni et al. 2016). Then, we performed a χ² test by varying EW and derived the upper limit to it when χ² > χ² (99.7%), i.e., at 3σ confidence level. This exercise resulted in the EW upper limit of 3.4 Å (Figure 2, bottom panel). We also computed the minimum detectable EW (EW_min) following a model independent approach described in Sbarufatti et al. (2005). In particular, the EW values were derived for wavelength intervals of fixed size (20 Å) along the whole spectrum. The EW_min was considered as the 3σ deviation from the mean (μ ∼ 0 Å) of the obtained distribution and found to be EW_min = 1.9 Å. Altogether, any emission line, if exists, must have the rest-frame EW ≲ 3.5 Å. We repeated the same exercise on the SDSS spectrum and estimated 3σ EW upper limit and EW_min as ∼6 Å and 2.5 Å, respectively. On the other hand, the results acquired from the EMIR data analysis are less constraining, due to low S/N, giving 3σ EW upper limit and EW_min for H_β line as ∼33 Å and 9 Å, respectively.

There are three absorption features in the rest-ultraviolet (UV) continuum longward of the Lyα forest at ∼5770, ∼5870, and ∼6065 Å (Figure 2). The most prominent absorption feature at ∼6065 Å is clearly the C iv λλ1548,1551 doublet of a foreground absorber at z = 2.9135, as indicated by the separation and relative strength of the lines as well as a clear Lyα absorption line at the same redshift. The other two absorption features are also possibly C iv doublet absorption at z = 2.7243 and z = 2.7900. However, the lower depth of the absorption lines, combined with noisy corresponding Lyα absorption in the Lyα forest makes it difficult to confirm. However, these two absorption features at ∼5770 Å ∼ 5870 Å do not correspond to known absorption lines at the redshift of J1219 that are commonly seen in the rest-UV of galaxies or quasars. The feature at ∼5770 Å could be explained by Si iiλ1260, but there are no other apparent absorption features of low ions (O i, C ii, Fe ii, and other less-prominent lines of Si ii). Thus, we deduce that there is no conclusive evidence of any absorption features associated with the blazar itself.

We also note the strong signal at 3700–4176 Å, below the Lyman limit of J1219. This indicates an intergalactic medium (IGM) transparent to hydrogen-ionizing photons for at least 68 Mpc h⁻¹ (proper), nearly 1.5× the measured mean free path of ∼46 Mpc h⁻¹ at this redshift (Prochaska et al. 2009). This is also much larger than the typical "proximity zone" of ionized gas found around luminous quasars at similar redshifts (Dall'Aglio et al. 2008).

The broadband spectral energy distribution (SED) of J1219 using all available observations is shown in the left panel of Figure 3. For a comparison, we also plot the SEDs of nine other γ-ray detected, z > 3 FSRQs studied in Paliya et al. (2020). The SED peak locations for J1219 were determined by fitting a third-order polynomial. For other sources, we used the results of the leptonic radiative modeling done in Paliya et al. (2020). As can be seen, J1219 has both SED humps peaked at considerably higher energies compared to z > 3 FSRQs. At optical-UV frequencies, a prominent thermal bump from the accretion disk is observed from other z > 3 sources. The radiation from J1219, on the other hand, appears to be synchrotron dominated with no traces of the accretion disk emission. Furthermore, the Compton dominance, which is the ratio of the inverse Compton to synchrotron peak luminosities, of z > 3 FSRQs is significantly larger (∼100) than that is observed for J1219 (∼1).

Zoom In Zoom Out Reset image size

Because emission lines are not detected in the optical/NIR spectra of J1219, we have derived the 3σ upper limit on the C iv line luminosity following the same procedure adopted to compute the EW upper limit and found it to be ∼2.5 × 10⁴² erg s⁻¹ (∼1.1 × 10⁴³ erg s⁻¹, when using SDSS data). Considering the line flux ratios from Francis et al. (1991) and assuming a 10% BLR covering factor, the 3σ upper limit on the accretion disk luminosity is ∼2.2 × 10⁴⁴ erg s⁻¹. Assuming the central black hole mass as 10⁹ ${M}_{\odot }$ (Paliya et al. 2020), we get ${\dot{L}}_{\mathrm{acc}}/{L}_{\mathrm{Edd}}\lt 0.002$ or 0.2% of L_Edd. A relatively high upper limit of 2% of L_Edd was obtained when H_β line luminosity derived from the EMIR data (∼10⁴³ erg s⁻¹) was considered. Such a low level of radiatively inefficient accretion is unlikely to photoionize the BLR clouds and can explain the observed lack of emission lines.

We comment on the location of J1219 in the blazar sequence by comparing its SED with the BL Lac PKS 2155−304 (z = 0.12) and other z > 3 blazars (Figure 3, left panel). PKS 2155−304 exhibits a low-luminous and high-frequency peaked SED compared to that of J1219. Other z > 3 FSRQs are more powerful and their SED peaks are located at even lower energies, as expected from the blazar sequence. Furthermore, as the updated Fermi blazar sequence is based on the sources grouped in different γ-ray luminosity (L_γ) bins (Ghisellini et al. 2017), we have calculated synchrotron peak frequency and Compton dominance for all blazars present in the 4LAC catalog and grouped them in different L_γ bins (Figure 3, middle panel). Interestingly, compared to other luminous blazars (L_γ > 10⁴⁷ erg s⁻¹), J1219 has a relatively high-frequency peaked and a low-Compton dominated SED. The physical behavior of J1219, therefore, deviates from the predictions of the Fermi blazar sequence. This stresses the importance of searching high-redshift BL Lacs to critically examine the proposed sequence.

The γ-ray emitting blazars are known to occupy a distinct place in the Wide-field Infrared Survey Explorer (WISE) color–color diagram (e.g., Massaro et al. 2011). In the right panel of Figure 3, we show this plot for 4LAC blazars and show the location of J1219. As can be seen, J1219 is located right in the middle of the WISE blazar strip indicating a close connection between the IR and γ-ray emissions of the source.

The space density of γ-ray emitting BL Lacs has only been constrained to z ≈ 2.5 and measured there to be ≈0.01 Gpc⁻³ (Ajello et al. 2014). Extrapolating the proposed evolution to z = 3.5 finds a space density of an order of magnitude lower, ≈0.001 Gpc⁻³. Considering the comoving volume within that redshift, this implies a prediction that there should be between one and two BL Lacs in the entire universe at that redshift, and this work has discovered one of them. Therefore, identification of more of z > 3 γ-ray emitting BL Lac sources will challenge our current understanding about their cosmic evolution.