THE 2009 DECEMBER GAMMA-RAY FLARE OF 3C 454.3: THE MULTIFREQUENCY CAMPAIGN

The flat spectrum radio quasar 3C 454.3 (at a redshift z = 0.859) turns out to be among the most active blazars emitting a broad spectrum ranging from radio to gamma-ray energies. Blazars are a sub-class of active galactic nuclei, with the relativistic jet aligned to the line of sight. Their spectral energy distributions (SEDs) typically show a double humped shape, with the low-energy peak lying between radio and X-rays and the high-energy peak in the GeV–TeV band (Urry & Padovani 1995). A detailed description of blazar leptonic emission models can be found in Maraschi et al. (1992), Marscher & Bloom (1992) and Sikora et al. (1994). The observed spectra can also be modeled in the framework of hadronic models (Mucke and Protheroe 2001; Mucke et al. 2003; Bottcher 2007).

Starting in 2004–2005, 3C 454.3 showed a long period of optical activity, with variability timescale ranging from several months to less than one day. In 2005 May, the source reached a peak magnitude R ≃ 12 showing strong 1 day variability (Villata et al. 2006). A radio peak was detected nine months after the optical peak, with fluxes of 22 Jy at 37 GHz and 20 Jy at 43 GHz (Villata et al. 2007). The source was then quiescent from the beginning of 2006 until mid-2007 (with an R magnitude between 15 and 16; Raiteri et al. 2008). Starting in the second half of 2007 (Vercellone et al. 2008), 3C 454.3 has been detected in a high gamma-ray state by AGILE (Tavani et al. 2009) and, subsequently, also by Fermi Large Area Telescope (Atwood et al. 2009; Abdo et al. 2009). Typically, the level of gamma-ray activity (with a flux of (300–600) ×10⁻⁸ photons cm⁻² s⁻¹ for E > 100 MeV) has been observed to be correlated with the optical emission. Relatively large gamma-ray fluxes were detected during the AGILE observations of 2007 (Vercellone et al. 2008, 2009a; Donnarumma et al. 2009). During 2008, the source that was bright at the beginning of the year, started to fade in the optical band (Villata et al. 2009). AGILE detected the fading in gamma rays too (Vercellone et al. 2010). Fermi reported an averaged spectrum above 200 MeV with a photon index α ≃ 2.3 and a spectral break at E_c ∼ 2.4 GeV, obtained in 2008 August (Abdo et al. 2009).

The intensive monitoring of 3C 454.3 carried out by our group covered the period of extraordinary gamma-ray activity in 2009 November–December (Striani et al. 2009a, 2009b). The campaign involved AGILE for the gamma-ray band, Swift/Burst Alert Telescope (BAT), Rossi XTE/HEXTE, and INTEGRAL/IBIS in the hard X-ray band, Rossi XTE/PCA and Swift/XRT in X-rays, Swift/UVOT in the optical and UV bands, the KANATA observatory and Goddard Robotic Telescope (GRT) in the optical, and Rapid Eye Mount (REM; Zerbi et al. 2001) in the near infrared and optical. AGILE observed the source every day in spinning mode, scanning about 70% of the whole sky every 6 minutes. International Gamma-Ray Astrophysics Laboratory (INTEGRAL) pointed the source in response to a Target of Opportunity (ToO) observations (Vercellone et al. 2009b) and observed it from 2009 December 6 until 2009 December 12. The Rossi XTE satellite observed 3C 454.3 on 2009 December 5 and then daily from 2009 December 8 until 2009 December 17 for typical integrations of ∼3 ks. Swift started to observe 3C 454.3 on 2009 November 27, in response to a ToO, and pointed at the source every day (UVOT performed most of the observations with UV filters). The KANATA 1.5 m telescope performed a long source monitoring in the V band with a time step of 1 day. The fully automated 14'' GRT performed observations in the V and R bands, quasi-simultaneously with Swift, starting on 2009 November 30. REM started the 3C 454.3 monitoring on 2009 December 10 in response to a ToO and observed the source every day in the VRIJHK filters. The GLAST-AGILE Support Program (GASP; Villata et al. 2008, 2009) performed an intensive monitoring campaign of the source in 2009–2010. We used a sub-sample of their data: optical observations reported in this Letter were performed at Lulin, New Mexico Skies, Roque de los Muchachos (KVA), and Saint Petersburg. GASP radio data were taken at Mauna Kea (The Submillimeter Array (SMA), 230 GHz), Noto (43 GHz), Metsähovi (37 GHz), and UMRAO (4.8, 8.0, and 14.5 GHz).

AGILE/GRID data were analyzed using the Build-19 software and the response matrix v10 calibrated in the energy range 100–3000 MeV. Well-reconstructed gamma-ray events were selected using the FM3.119 filter. All the events collected during the passage in the South-Atlantic Anomaly were rejected. We filtered out the Earth-albedo, rejecting photons coming from a circular region of radius 85° and centered on the Earth. We rejected photons with polar direction >35° in the GRID reference system. Gamma-ray data were analyzed with integrations of 1 or 2 days, depending on the source flux. We used the standard AGILE Maximum-Likelihood procedure (ALIKE; see Mattox et al. 1996 for the concept definition) for each data set. The integration over 5 weeks from 2009 November 18 to 2009 December 23 UTC yields a photon index 1.88 ± 0.08 (all the errors reported in this Letter are at 1σ, except where stated).

The INTEGRAL-IBIS (Ubertini et al. 2003) data were processed using the OSA software version 8.0. Light curves (from 20 to 200 keV) and spectra (from 18 to 200 keV) were extracted for each individual science window of revolutions 873 and 874.

The Swift-BAT survey data were obtained applying the "BAT FOV" option. The data have been processed by batsurvey script available through HEASOFT software package with a snapshot (single pointing) interval. To estimate the background, 10 background points around the source in a radius of 50' are selected. The source, 10 background points, and the bright hard X-ray sources (for cleaning purpose) are included in the input catalog of batsurvey. The BAT count rate in the 14–195 keV band has been converted into the energy flux assuming a power-law photon index 1.7 as determined from the INTEGRAL-ISGRI data, see below. In order to match with the HEXTE range, the BAT hard X-ray flux has been rescaled in Figure 1 to the 20–40 keV band.

Zoom In Zoom Out Reset image size

Rossi XTE-PCA (Jahoda et al. 1996) and HEXTE (Rothschild et al. 1998) data were analyzed following the same procedure as described in Vercellone et al. (2010). The data analysis was restricted to the PCU2 in the 3–20 keV energy range for the Proportional Counter Array (PCA) and to the Cluster B in the 18–50 keV range for the HEXTE. The net exposure times were 27.3 ks for PCA and 7.3 ks for HEXTE. The background-subtracted source spectra obtained with both the instruments were simultaneously fit⁴⁰ with an absorbed power-law model, with the photoelectric absorption fixed to 0.134×10²² cm⁻² (Villata et al. 2006). After the introduction of a 2% systematic error, the best-fit value for the photon index is 1.74 ± 0.01.

The Swift-XRT data were processed using the most recent calibration files available. We utilized Swift Software version 3.5, FTOOLS version 6.8, and XSPEC version 12.5.1n. We fitted the data with an absorbed power-law model. We obtained photon indices between 1.51 ± 0.09 and 1.73 ± 0.11 and excess absorption between (0.09 ± 0.06) × 10²² and (0.17 ± 0.03) ×10²² cm⁻² (all uncertainties on XRT spectral fit are at the 90% level).

Swift-UVOT data from each observation sequence were processed by the standard UVOT tool uvotsource using the same version of the Swift software as for the XRT analysis. An extraction region of radius 5 arcsec centered on the source and a suitable background region were used. Magnitudes are based on the UVOT photometric system (Poole et al. 2008).

The optical photometry of the Kanata Observatory data was performed using TRISPEC (Watanabe et al. 2005). The observations were pipeline-reduced, including bias removal and flat-field corrections. We derived the V-band magnitude from differential photometry with a nearby reference star, USNOB 1061-0614254 (see Gonzales-Perez et al. 2001). Photometric stability of this star has been confirmed by our simultaneous photometry for another nearby star, USNOB 1061-0614207.

All REM raw optical and NIR frames, obtained with ROSS (Tosti et al. 2004) and REMIR (Conconi et al. 2004), respectively, as well as images from GRT were corrected for dark, bias, and flat field following standard recipes. Instrumental magnitudes were obtained via aperture photometry, and absolute calibration has been performed by means of secondary standard stars in the field (Raiteri et al. 1998).

Even if it is not shown in Figure 1, the Metsähovi radio data at 37 GHz show a high flux with an increasing trend from 2009 December 1 until 2010 January 14 and a mean flux of ∼20 Jy during the first week of 2009 December. The mean 230 GHz flux is ∼25 Jy. These radio fluxes⁴¹ are comparable to the peak flux measured in 2006 (Villata et al. 2007).

All UV/optical/NIR data presented here were corrected for the Galactic extinction toward 3C 454.3 assuming A_V = 0.349 (Schlegel et al. 1998).

The multifrequency light curves of 3C 454.3 are shown in Figure 1. The exceptional gamma-ray flaring activity is produced during an extended period lasting several weeks. The broadband coverage is excellent, and for the first time allows a detailed multiwavelength study of a gamma-ray blazar peaking above the Vela pulsar flux (for comparison, see the multifrequency coverage for the gamma-ray super-flare of PKS 1622-29; Mattox et al. 1997).

The optical data show variability timescale as short as 1 day or less. Due to a typical interval between optical observations (1 day), we did not show in Figure 1 a detailed representation of the variability. The optical V data indicate a low state of the source (V > 15 mag, flux < 3.9 mJy) until MJD 55160, after which the flux started to increase. The optical flux increased about 50% in less than 1 day from MJD 55166.4 until MJD 55167.4, reaching the value V = 13.7 mag (e.g., 12.8 mJy), followed by a fast flux decrease to V = 14.3 mag (e.g., 7.4 mJy) at MJD 55169.0. Another optical peak was reached at MJD 55172.5 (V = 13.7 mag), and was then followed by a minimum at MJD 55175.0, with V = 14.4 mag (e.g., 6.7 mJy).

In general, the X-ray flux follows⁴² the rising part of the optical emission in the interval MJD 55150-55169. Starting on MJD 55169, the X-ray flux started to fade, with the optical emission remaining in a relatively high state for 4–5 days. Our INTEGRAL hard X-ray data sample the fading phase of the high-activity period.

We performed a discrete correlation function analysis for the optical–gamma and soft X–gamma data set. No significant delay is found within 2 days (compatible with the average bin of gamma-ray light curve).

We focused on the time-dependent spectral analysis of this exceptional activity of 3C 454.3 and used simultaneous broadband data to obtain a detailed account of the source variability. We obtained the SED for four periods. The first period (interval-1) is for a 5 day integration of GRID data, centered at 2009 November 27 09:36 UT (MJD 55162.4), and using the Swift observation at MJD 55162.9, and quasi-simultaneous GASP optical data (from KVA and New Mexico Skies) obtained within 16 hr from the Swift observation (pre-flare SED).

The second period (interval-2) was obtained for the gamma-ray super-flare episode integrating the GRID data for 1 day centered at 2009 December 2 16:48 UT (MJD 55167.7) and using the simultaneous Swift and GRT observations at MJD 55167.0 (first flare SED). The Kanata observatory measured a flux 40% higher in the V band 10 hr after the GRT observation.

The third SED was obtained (interval-3) integrating GRID data for 2 days centered at 2009 December 6 16:48 UT (MJD 55172.7), to match the local maximum of the optical light curve with V = 13.7 mag that is apparently coincident with the secondary gamma-ray maximum near MJD 55174. For this interval, we used the INTEGRAL-ISGRI data collected between MJD 55171.7 and MJD 55174.2, Swift data at MJD 55173.9, Rossi XTE data at 55173.4, and GASP optical data (from Lulin and GRT) obtained within 26 hr from the Swift data.

The last SED (interval-4) was obtained integrating GRID data for 5.5 days centered at 2009 December 15 18:00 UT (MJD 55180.8) and making use of the Swift observation at MJD 55179.1, the Rossi XTE observation at MJD 55179.2, the GASP optical data (from Saint Petersburg and Lulin) simultaneous with Swift within 12 hr, and the near-infrared observations from REM at MJD 55181.0 (post-flare SED).

Radio data for the SEDs were taken by the GASP, using observations from Mauna Kea (SMA, 230 GHz), Noto (43 GHz), Metsähovi (37 GHz), UMRAO (4.8, 8.0, 14.5 GHz), and simultaneous within a few days with the XRT observations. Due to the slow variability of the radio data, we use interpolated radio values in the SEDs. The results are shown in Figure 2.

Zoom In Zoom Out Reset image size

The multifrequency data of the extensive campaign on 3C 454.3 show a remarkable behavior of the source. Starting from MJD 55150, first an overall rise of the gamma-ray emission is detected and then that of the X-ray and optical fluxes. This rise culminates with peak optical/X-ray/gamma-ray emission during a 10 day period centered around MJD 55173. Subsequently, the overall flux decreased and reached again a relative-minimum state around MJD 55200. During the two-month period, the optical and X-ray fluxes vary within a factor of 3, whereas the γ-ray flux grows by a factor of 5–10 compared to the pre-flare value. During the rapid super-flare around MJD 55167.7, the γ-ray flux doubles within 1 day with the optical and average X-ray increase of 50% and 30%, respectively.

We find an overall correlation at all wavelengths for both long and short timescales. However, the unusual gamma-ray flaring and super-flaring activity from 3C 454.3 during the period 2009 November–December is not accompanied by strong emission of similar intensity in the optical or even in the soft X-ray bands. The hard X-ray flux is comparable to the level of spring 2005, e.g., ∼3.5 × 10⁻¹⁰ erg cm² s⁻¹ in 20–200 keV (Pian et al. 2006). This flaring behavior appears to be quite different from other episodes detected in 2007 and 2008 (e.g., Vercellone et al. 2009a; Donnarumma et al. 2009). The synchrotron emission appears to be quite broad and centered around ν ∼ 10¹³ Hz.

Striani et al. (2010) provided a first report of the AGILE-GRID data. A single power-law approximation gives a photon spectral index 1.66 ± 0.32 in the energy band 0.1–1 GeV, integrating the data for 2 days centered at MJD 55167.7. We confirm this result in the analysis presented here integrating gamma-ray data for 1 day, as shown in our Figure 2 (left panel). In addition, we found a similar spectral shape of the gamma-ray emission during the secondary maximum of interval-3. Our Figure 2 (right panel) shows the interval-3 SED as compared with the post-flare SED of interval-4.

The simultaneous observations of 3C 454.3 by AGILE, INTEGRAL, Rossi XTE, and Swift strongly constrain the emission models in the high-energy range. In particular, our interval-3 spectrum (blue solid squares of Figure 2, right panel) represents one of the best constrained multifrequency spectra ever obtained for a flaring blazar from X-ray up to GeV energies.

We present in Figure 2 the results of our spectral modeling based on synchrotron self-Compton (SSC), plus contribution from external seed photons (EC). We used parameters similar to those already implemented to model previous gamma-ray flares of 3C 454.3. (Vercellone et al. 2009a; Donnarumma et al. 2009). We find that the pre- and post-flare spectra (interval-1 and interval-4) are adequately represented by a simple one-zone SSC model plus EC in which the accretion disk and the broad-line region provide the necessary soft radiation field for the inverse Compton components that dominate the X-ray through the GeV energies. The simple one-zone model parameters for the pre- and post-flare intervals are reported in Table 1 (they are also labeled component-1 in the table).

We can fit the super-flare with a simple one-zone model as shown in Figure 2 (dot-dashed curve), with the parameters reported in Table 1. Such a model requires (compared to the pre-flare fit parameters) an increase of the electron energy and density, and a slight reduction of the comoving magnetic field for the whole electron population of the blob (as also reported in Bonnoli et al. 2010). We note, however, that we used the optical data obtained during the rising edge of optical emission for the SED of the super-flare. Using only the V peak emission for the optical portion of the spectra, the one-zone model gives a magnetic field of 0.55 G, closer to the value for the pre-flare.

The time evolution of the source (e.g., the lack of strong optical emission and the gamma-ray spectrum during the super-flare) can also be described adopting a different approach. We assume a long-term rise and fall of the mass accretion rate onto the central black hole. This enhanced accretion causes an overall increase of the synchrotron emission and of the soft photon background scattered off by the primary component of accelerated electrons (component-1). An additional population of accelerated leptons (component-2, co-existing with component-1) can be introduced for the super-flare and secondary flare episodes. This component is a consequence of additional particle acceleration and/or plasmoid ejection near the jet basis. Table 1 reports the parameters that we used to model the Figure 2 spectra (the two-component models are shown as thick solid lines for the super-flare and for the secondary flare). For all intervals, we assume the presence of component-1. This is the only component acting for the pre- and post-flare intervals, with the parameters given in Table 1. The super-flare (secondary-flare) parameters for component-1 are the same as the pre-flare (post-flare) interval. We assumed a bulk Lorentz factor Γ = 25, a jet angle with respect to the line of sight θ = 12, and an accretion disk of bolometric luminosity L_d = 6 × 10⁴⁶ erg s⁻¹ slowly decaying toward L_d = 5 × 10⁴⁶ erg s⁻¹. A broad-line region located 0.5 pc from the black hole reflects 5% of the disk power toward the emitting regions.

The component-2 energy distribution that better reproduces our gamma-ray spectral data is a relativistic Maxwellian of characteristic energy γ_b ≃ 180. Interestingly, this component appears to be strongly energized but not yet modified by additional non-thermal acceleration.

This two-component approach avoids the problem of explaining the time variability of the physical parameters of the whole electron population of the one-zone model, at the price of adding an additional varying component superimposed with the pre-existing one.

To summarize, our multifrequency data for the 2009 December flare of 3C 454.3 provide a wealth of very important information on this puzzling and fascinating blazar. We find that 3C 454.3 is characterized by strong broadband spectral variability and that the modeling of the peak gamma-ray emission episodes suggests more elaborate models than the standard one-zone SSC+EC models of bright blazars.

The AGILE Mission is funded by the Italian Space Agency (ASI) with scientific and programmatic participation by the Italian Institute of Astrophysics (INAF) and the Italian Institute of Nuclear Physics (INFN). This investigation was carried out with partial support from the ASI contract no. I/089/06/2. V.L. acknowledges support from Russian RFBR foundation via grant 09-02-00092. The operation of UMRAO is made possible by funding from the NSF, NASA, and the University of Michigan. The Submillimeter Array is funded by the Smithsonian Institution and the Academia Sinica Institute of Astronomy and Astrophysics. The GASP president acknowledges the ASI support through contract ASI-INAF I/088/06/0.