Identifying the 3FHL Catalog. V. Results of the CTIO-COSMOS Optical Spectroscopy Campaign 2019

The Large Area Telescope (LAT; Atwood et al. 2009) on board the Fermi Gamma-Ray Space Telescope has enabled great strides in the understanding of blazars, the most powerful class of active galactic nuclei (AGN). Blazars make up the majority of the astrophysical gamma-ray source population (Abdollahi et al. 2020). The broadband spectral energy distribution (SED) of blazars is generally characterized by two distinct bumps, which are attributed to synchrotron radiation at lower energies (infrared to X-rays) and inverse Compton scattering at high energies (X-rays to gamma-rays, Maraschi et al. 1994; Abdo et al. 2011).

The Third Catalog of Hard Fermi-LAT Sources (3FHL; Ajello et al. 2017) reported 1556 sources detected in the 10 GeV–2 TeV energy range during the first seven years of Fermi-LAT's operation. Blazars represent the majority (∼1300/1556) of these 3FHL sources. However, ∼45% (578) of these blazars lack a redshift measurement and ∼21% (232) are of an uncertain type, leaving the 3FHL catalog highly incomplete.

Accurate redshift determinations are essential for studying numerous topics, including blazar emission processes and energetics (e.g., Ghisellini et al. 2017; van den Berg et al. 2019), the role of blazars in cosmic-ray acceleration (e.g., Furniss et al. 2013), the properties and evolution of extragalactic background light (EBL, i.e., the integrated emission from all stars and galaxies in the universe since the reionization epoch; e.g., Abramowski et al. 2012; Ackermann et al. 2012; Domínguez & Ajello 2015; Abdollahi et al. 2018; Desai et al. 2019), the derivation of cosmological parameters using gamma-rays (e.g., Domínguez et al. 2019), and studying and understanding the cosmic evolution of blazars (e.g., Ajello et al. 2014; Paliya et al. 2020).

To this end, spectroscopic campaigns of blazars have been effectively conducted with 4 m, 8 m, and 10 m class telescopes to investigate the nature of these sources as well as to provide accurate redshift measurements (e.g., Sbarufatti et al. 2005; Shaw et al. 2012; Massaro et al. 2014; Paggi et al. 2014; Landoni et al. 2015; Ricci et al. 2015; Álvarez Crespo et al. 2016a, 2016b; Paiano et al. 2017, 2019, 2020; Peña-Herazo et al. 2017, 2020; Marchesi et al. 2018; Desai et al. 2019). The successes of these campaigns prove that the listed classes of telescopes are capable of distinguishing the two blazar subclasses, namely, BL Lacertae objects (BL Lacs) and flat-spectrum radio quasars (FSRQs).

FSRQs show broad emission features in their optical spectra and have higher average luminosities than BL Lacs, thus allowing easy measurement of their redshifts, while BL Lac spectra exhibit weak or very narrow emission lines (equivalent width (EW) ≤ 5 Å; Urry & Padovani 1995; Marcha et al. 1996).

Blazars are further subdivided into three categories based on the position of their synchrotron peak frequency, ${\nu }_{\mathrm{pk}}^{\mathrm{sy}}$. Sources with ${\nu }_{\mathrm{pk}}^{\mathrm{sy}}\lt {10}^{14}$ Hz are classified as low-synchrotron peak blazars (LSP), sources with ${10}^{14}\lt {\nu }_{\mathrm{pk}}^{\mathrm{sy}}\lt {10}^{15}$ Hz are known as intermediate-synchrotron peak blazars, and sources with ${\nu }_{\mathrm{pk}}^{\mathrm{sy}}\gt {10}^{15}$ Hz are categorized as high-synchrotron peak blazars (HSP). In the Fermi catalogs, BCU is a term used for classification for associated counterparts that (a) show blazar-like multifrequency behavior or (b) are classified as blazars of an uncertain type in the Roma-BZCAT (Massaro et al. 2009). The 3FHL contains 232 BCUs, 28 of which were targeted in this work. Spectroscopic observations of such sources will enable us to classify these sources and potentially measure their redshifts.

In this campaign, 51 BCUs have so far been classified into BL Lacs and FSRQ categories using optical spectroscopy. Spectroscopic redshifts were also obtained for 15 of these sources (Marchesi et al. 2018; Desai et al. 2019). In addition, machine-learning algorithms have provided tentative classifications for an additional 51 sources (Kaur et al. 2019; Silver et al. 2020).

In this work, we report the results from spectroscopic observations of 3FHL BCUs performed using the Cerro Tololo Ohio State Multi-Object Spectrograph (COSMOS) mounted on the 4 m Blanco telescope located at the Cerro Tololo Inter-American Observatory (CTIO) facility. The organization of this paper is as follows: In Section 2, we report the criteria used for the sample selection. In Section 3, we describe the observation methodology and the data analysis approach. In Section 4, we summarize the results of this campaign, with additional consideration given to sources that exhibit spectral features, while in Section 5, we present our conclusions.

In this work, we report the observations of 28 sources performed with CTIO in 2019, constituting the third part of this campaign. ⁵ These sources were selected from BCUs in the 3FHL catalog based on the following criteria:

• 1.  
  Lack of a redshift.
• 2.  
  Optical magnitude in the V band of V ≤ 21.5.
• 3.  
  Being bright in 10 GeV–2 TeV band: (f_3FHL > 10⁻¹² erg s⁻¹ cm⁻²). This condition ensures that the completeness of the 3FHL catalog moves to lower fluxes as more optical observations are performed.
• 4.  
  Being observable from Cerro Tololo (−80° < decl. < 20°), above an altitude of 40° and during our scheduled observations from April to June.

Details of all sources analyzed here have been summarized in Table 1. Optical magnitudes have been obtained from the 3FHL catalog (where the authors used USNO ⁶ or SDSS ⁷ archival data).

All the source spectra for this work were obtained using the COSMOS mounted on the 4 m Blanco telescope located at the CTIO facility in Chile. The instrument was configured with the red grism and 09 slit, to cover a wavelength range of λ = [5000−9000] Å, corresponding to a dispersion of ∼4 Å pixel⁻¹ and a spectral resolution of R ∼ 2100. Wavelength calibration was performed with a precision of 0.04 Å. The fluctuations were found to be both positive and negative indicating a nonbiased wavelength calibration process. The observations were performed over three nights in 2019 April and over four nights in 2019 June. However, due to poor weather conditions, data from six nights in total was used in our analysis. The observing dates for all sources along with the exposure times and V-band magnitudes have been reported in Table 1.

Spectra for each source were obtained in triplicate or more, with varying exposure times, and the individual observations were combined to reduce the effects of instrumental noise and cosmic-ray contamination. Data reduction was performed using the standard IRAF pipeline (Tody 1986) including bias subtraction and correction for bad pixels. Flat-field normalization was also performed in order to remove any wavelength-dependent variations present in the flat-field source (but not in the observed spectrum). The reliability of the emission and absorption lines was obtained by comparing the features in each individual exposure. Additionally, all spectra were visually inspected to remove any artificial features.

Wavelength calibration for each source was performed using the mercury–neon (Hg–Ne) lamp. A lamp spectrum was obtained after each observation of a source to account for the possible shifts in the pixel–λ calibration, due to changes in the telescope position throughout the night. The spectra were independently inspected for emission and absorption features typical of those observed in BL Lac and FSRQ spectra (e.g., see Stickel et al. 1991; Marcha et al. 1996; Bade et al. 1998; Shaw et al. 2013). The presence of only absorption features in a spectrum indicates either absorption by the intervening medium (along the line of sight) or by the host galaxy (e.g., Ca ii, G-band, NaD, Mg i, Ca+Fe lines). This enables us to set a spectroscopic lower limit of the redshift. All spectra were flux calibrated using a spectrophotometric standard. The standards were observed twice each night: one at the beginning of the night and one at the end, under the same observing conditions as the rest of the analysis. Finally, each source spectrum was corrected for Galactic reddening, using the E(B − V) value obtained from NASA/IPAC Infrared Science Archive online tool, ⁸ based on Schlafly & Finkbeiner (2011).

Results obtained from the spectral analysis of the sources have been reported in Table 2 (along with source classifications). Here, we report to the highest precision for which there is agreement among different features. The corresponding spectra are shown in Figure 1. To make the features more apparent, normalized spectra of our sources are also shown, which were obtained by dividing the flux-calibrated spectrum by a power-law fit of the continuum. The signal-to-noise ratio (S/N) of the normalized spectra (reported in Table 2) were obtained by measuring the ratio in at least five different regions in the featureless part of each spectrum.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

To identify the various emission and absorption features, all spectra were individually inspected. A feature was considered reliable if its presence was verified in all the individual exposures. Sources were classified as FSRQs if the emission lines in their spectra had EW > 5 Å, and as BL Lacs if emission lines were observed with EW < 5 Å.

In this work, 25 out of 28 sources have been classified as BL Lacs, with 3 sources being classified as FSRQs. Further, redshifts (z) have been provided for 6 sources and lower limits on z for 10 more. The remaining 12 sources exhibited featureless optical spectra.

Comments have been provided for all the sources. Certain source spectra were also observed to be particularly noisy, inhibiting us from identifying any features. Tentative classifications have been provided for such sources.

4.1. Notes on Individual Sources

This identification campaign to classify BCUs in the 3FHL catalog and to measure their redshift began in 2017, when 36 unassociated sources (out of 52 chosen) from 3FHL catalog were classified as BL Lacs with the help of machine-learning algorithms (Kaur et al. 2019). For the second part of this campaign, 28 sources were observed with the 4 m telescope at the Kitt Peak National Observatory (KPNO) in Arizona, in order to perform optical spectroscopy. These results were published in Marchesi et al. (2018), where 27 sources were classified as BL Lacs and one as an FSRQ. Redshift measurements or lower limits were also provided for seven of these sources. The third part of the campaign, for which spectra for 23 sources were obtained using the 4 m telescope at CTIO in Chile, resulted in the categorization of all 23 sources as BL Lacs and the determination of redshift estimates or lower limits for eight sources (Desai et al. 2019). The fourth paper from this campaign resulted in the classification of 15 sources (out of 38) as BL Lacs using machine-learning algorithms (Silver et al. 2020).

There are other works that are complementary to our campaign toward completing the 3FHL catalog. For instance, Paiano et al. (2017, 2019, 2020) have studied 103 Fermi-LAT sources so far and provided redshifts for 65 of them. As part of a large spectroscopic program, Peña-Herazo et al. (2020) have classified 350 Fermi-LAT sources into BL Lac and FSRQ categories. Landoni et al. (2020) have compiled a database of spectroscopic redshifts of BL Lacs that is available online. ⁹

We present here the results of the fifth part of this ongoing optical spectroscopic campaign aimed at the spectral completeness of the 3FHL catalog. We report on 28 sources observed with the COSMOS spectrograph mounted on the 4 m Blanco telescope, located at the CTIO facility in Chile. Based on the optical spectra of our sources, 25 sources were classified as BL Lacs, and the remaining three were classified as FSRQs. ¹⁰ With ∼85% of the sources in the 3FHL catalog being BL Lacs, these classification results are not unexpected. Desai et al. (2019) found all (23) sources observed by them to be BL Lacs, while Marchesi et al. (2018) identified 27 sources as BL Lacs, out of the 28 observed.

We were also able to place redshift constraints on 57% (16/28) of our sources representing a significant improvement in the success rate of such campaigns with 4 m class telescopes (see Landoni et al. 2015; Ricci et al. 2015; Álvarez Crespo et al. 2016a; Peña-Herazo et al. 2017).

The completeness plots for our campaign sources with respect to all blazars in the energy range of 10 GeV–2 TeV are shown in Figure 2. The latest release of the fourth catalog of AGN detected by the Fermi-LAT (4LAC–DR2; Ajello et al. 2020) was used to update the 3FHL catalog in order to include recent redshift measurements and classifications of sources as provided by various campaigns (e.g., Peña-Herazo et al. 2017, 2020; Paiano et al. 2019, 2020; de Menezes et al. 2020). Completeness was obtained by considering the ratio of the number of classified blazars or blazars with z above a given flux divided by the total number of blazars in the same range. As seen from this figure, this campaign has substantially increased the type completeness of the sources in 3FHL catalog, but less so the redshift completeness. Prior to our campaign, 21% of the 3FHL blazars lacked a type and 45% lacked a redshift value. With our campaign, we were able to decrease the fraction of blazars without a type to ∼12% and blazars without redshift to 43%. The type completeness also reaches 95% for F > 3 × 10⁻¹² erg cm⁻² s⁻¹ (see Figure 2, left).

Zoom In Zoom Out Reset image size

Hence, even with 4 m class telescopes like KPNO and CTIO, we have made significant progress in achieving our goal of full completeness in the 3FHL catalog. But since the success rate of redshift determination with 8 m class telescopes is even higher (60%–80%), for the next part of our campaign we will present the results of the observations undertaken using the 8 m Gemini-N and -S telescopes.

Identification of classes and redshifts of high-energy sources is also a crucial step in understanding and constraining the EBL (e.g., Saldana-Lopez et al. 2020). The cosmic gamma-ray horizon (CGRH) provides an estimate of the distance beyond which the universe becomes opaque to gamma-rays (τ_{γ γ} = 1, where τ_{γ γ} is the optical depth of EBL as estimated from the models). This opacity originates when very high-energy (E > 100 GeV) photons interact with the low-energy EBL photons causing electron–positron pair production to occur, thus annihilating the photons in the process (Domínguez et al. 2013). The CGRH as a function of redshift has been plotted in Figure 3 for all the highest-energy photons (HEP) in the 3FHL catalog (Ajello et al. 2017). We highlight in the figure both the objects studied in our spectroscopic campaign and the photometric campaign. The photometric dropout technique is based on the flux attenuation caused by the interaction of source photons by neutral hydrogen and it becomes particularly effective at z > 1 (Rau et al. 2012; Kaur et al. 2017, 2018; Rajagopal et al. 2020). It is clear that both campaigns resulted in several sources for which the HEPs are very near current estimates of the CGRH, as well as some beyond the horizon. ¹¹ Thus, such campaigns are invaluable in further constraining the CGRH as well as probing higher-opacity regions, e.g., τ_{γ γ} ≥ 2, where redshift data are sparse. Estimating a redshift for such high-energy sources will enable a better and more accurate measurement of the EBL.

Zoom In Zoom Out Reset image size

M.R., S.M., and M.A. acknowledge support from NASA contract 80NSSC19K0151. A.D. is thankful for the support of the Ramón y Cajal program from the Spanish MINECO. The authors thank Alberto Alvarez, Claudio Aguilera, and Sean Points for the help provided during the observing nights at CTIO. This work made use of the TOPCAT software for the analysis of data tables.