MAGIC DISCOVERY OF VERY HIGH ENERGY EMISSION FROM THE FSRQ PKS 1222+21

The differential energy spectrum of PKS 1222+21 was reconstructed using the "Tikhonov" unfolding algorithm (Albert et al. 2007), to take into account the finite energy resolution of the instrument and the biases in the energy reconstruction. The energy spectrum, shown in Figure 1, extends up to at least 400 GeV and is well described by a simple power law of the form

$$\begin{equation} \frac{{\it dN}}{{\it dE}}=N_{200}\left(\frac{E}{200\;{\rm GeV}}\right)^{-\Gamma } \end{equation} \tag{ 1 }$$

with a photon index Γ = 3.75 ± 0.27_stat ± 0.2_syst and a normalization constant at 200 GeV of N₂₀₀ = (7.8 ± 1.2_stat ± 3.5_syst) × 10⁻¹⁰ cm⁻² s⁻¹ TeV⁻¹, yielding an integral flux (4.6 ± 0.5) × 10⁻¹⁰ cm⁻² s⁻¹ (≈1 Crab Nebula flux) at E>100 GeV and (9.0 ± 3.6) × 10⁻¹² cm⁻² s⁻¹ (7% of the Crab Nebula flux) at E>300 GeV, at the same level of Whipple upper limit (Section 1). For energies higher than 400 GeV no significant excess was measured. The upper limits corresponding to 95% confidence level (CL) are shown in Figure 1. The systematic uncertainty of the analysis (studied by using different cuts and different unfolding algorithms) is shown by the gray area.

Zoom In Zoom Out Reset image size

We studied the effect of the VHE γ-ray absorption due to pair production with low energy photons of the EBL by using different state-of-the-art EBL models, namely, the models by Dominguez et al. (2011), Kneiske & Dole (2010), Gilmore et al. (2009), Franceschini et al. (2008), and the "max high UV" EBL model described in Albert et al. (2008). For each of the EBL models, the optical depth corresponding to the measured VHE γ-ray energy intervals was computed and the differential fluxes were corrected accordingly to obtain the de-absorbed (or intrinsic) spectrum. The spectrum deabsorbed with the EBL model of Dominguez et al. (2011), shown by the blue squares in Figure 1, is well fitted by a power law with an intrinsic photon index of Γ_intr = 2.72 ± 0.34 between 70 GeV and 400 GeV. Uncertainties caused by the differences between the EBL models are represented in Figure 1 by the blue-striped area. The corresponding spread is smaller than the systematic uncertainties of the MAGIC data analysis.

We investigated the possible presence of a high-energy (HE) cutoff in the VHE range by fitting power laws with different photon indexes and different values for the cutoff. The method adopted is the χ² difference method (see, e.g., Lampton et al. 1976). With the available statistics, at the 95% CL we cannot exclude the presence of a cutoff above 130 GeV for a photon index 2.4 (the lowest possible value compatible with fit uncertainties and with Fermi/LAT data, see Section 3.3) or above 180 GeV for a photon index 2.7. The confidence interval is not bounded on the HE side, i.e., a fit without a cutoff is fully compatible with the data. Further observations with higher statistics are needed to better constrain the location of a possible steepening in the form of a cutoff or spectral break.

Despite the short observation time, the strength of the signal allows us to perform a variability study of the measured integral fluxes above 100 GeV. The light curve binned in 6 minutes long intervals is shown in Figure 2 and reveals clear flux variations. The constancy hypothesis (χ²/NDF = 28.3/4) is rejected with high confidence (probability <1.1 × 10⁻⁵). The fluxes of background events surviving the γ/hadron selection cuts are compatible with being constant, and hence we can exclude a variation of the instrument performance during the observation.

Zoom In Zoom Out Reset image size

To quantify the variability timescale we performed an exponential fit (solid black line in Figure 2). A linear fit is also acceptable but does not allow us to define a timescale unambiguously. For the exponential fit the doubling time of the flare is estimated as 8.6^+1.1_−0.9 minutes. The derived timescale corresponds to the fastest time variation ever observed in an FSRQ in the VHE range and in any other energy range (Foschini et al. 2011), and is among the shortest timescales measured on TeV emitting sources (Abramowski et al. 2010).

In the HE MeV/GeV energy range measured by Fermi/LAT the source showed a significant flare lasting ∼3 days, with a flux peak on 2010 June 18 (MJD 55365) (Tanaka et al. 2011). A dedicated analysis found that the 1/2 hr MAGIC observation fell within a gap in the LAT exposure, thus we analyzed a period of 2.5 hr (MJD 55364.867 to 55364.973), encompassing the MAGIC observation. The LAT analysis for this time bin was performed as in Tanaka et al. (2011), where details can be found. It results in an integral flux (6.5 ± 1.9) × 10⁻⁶ cm⁻² s⁻¹ at energy E>100 MeV. The observation in such a short time does not provide any detection with Fermi/LAT at E>2 GeV. Two Fermi/LAT spectral points up to 2 GeV together with an upper limit at the 95% CL in the range 2–6.3 GeV are combined with the MAGIC data in the spectral energy distribution (SED) shown in Figure 3.

Zoom In Zoom Out Reset image size

The figure also shows bow ties representing uncertainties associated with the spectral fits. The Fermi/LAT spectrum is best described by a single power law with index of 1.95 ± 0.21. In the case of MAGIC data the bow tie refers to the "intrinsic" source spectrum, i.e., to the observed spectrum corrected for EBL absorption, described in Section 3.1. An extrapolation of the intrinsic spectrum in the MAGIC range to lower energies is also shown indicating that: (1) there is a potentially smooth connection between the Fermi/LAT and MAGIC extrapolated data in the 3 to 10 GeV region and (2) the photon index steepens from 1.9 in the Fermi/LAT range to 2.7 in the MAGIC range. These results agree with the analysis of wider temporal intervals during this flare and during the whole active period, in which the source spectrum is well described by a broken power law with an energy break falling between 1 and 3 GeV (Tanaka et al. 2011). Furthermore, it is found that the HE tail (E>2 GeV) of the Fermi/LAT spectrum of PKS 1222+21 extends up to 50 GeV, with a photon index in the range 2.4–2.8.

The interaction of VHE γ-rays with low energy photons of the isotropic EBL is a process with an energy dependent threshold, thus leading to an imprint of the EBL density on the measured VHE γ-ray spectra of extragalactic sources (Mazin & Raue 2007). For PKS 1222+21 (z = 0.432), the measured spectrum spans from 70 GeV to 400 GeV probing EBL photons in the range 0.1–1 μm (i.e., UV to near-infrared).

The EBL constraints using VHE γ-rays are usually derived assuming an intrinsic spectrum of the source (e.g., Aharonian et al. 2006). In FSRQs, the presence of dense radiation fields of soft photons can lead to the internal absorption of VHE γ-rays, mimicking harder-than-intrinsic spectra (e.g., Sitarek & Bednarek 2008). However, for realistic spectral distributions of the internal photon fields it should not change the EBL limits significantly (Tavecchio & Mazin 2009).

In our case the simultaneous data from Fermi/LAT, which is free from internal or external absorptions, have been used to constrain the intrinsic photon index in VHE (e.g., Georganopoulos et al. 2010; Finke & Razzaque 2009). We adopt a method similar to the one utilized by Georganopoulos et al. (2010): the intrinsic spectrum in the VHE regime is assumed to follow the extrapolation of the Fermi/LAT above 3 GeV with a Γ = 2.4. This is a conservative assumption since in reality the spectrum could soften with increasing energy.

The upper limit (95% CL) on the optical depth, τ_max, for VHE γ-rays can be obtained from

$$\begin{equation} \tau _{{\rm max}}(E) = \log \left[ \frac{F_{{\rm intr}} (E)}{F_{{\rm obs}}(E)-1.64\cdot \Delta F(E)}\right], \end{equation} \tag{ 2 }$$

where F_intr(E) is the maximum intrinsic flux at energy E, and F_obs(E) and ΔF(E) are the MAGIC measured flux and its error, respectively. The maximum intrinsic flux has been normalized at 70 GeV assuming the EBL model giving a maximum flux absorption of 30% (Albert et al. 2008). The derived limits on the optical depth are shown in Figure 4 together with a compilation of the predicted optical depths for a source at z = 0.432 computed according to recent EBL models. The limits confirm previous constraints on the EBL models in the UV to near-infrared regimes derived using VHE (Aharonian et al. 2006; Mazin & Raue 2007; Albert et al. 2008) and HE spectra (Abdo et al. 2010a). Given the fact that the EBL models predict for this redshift a stronger absorption with increasing energy, our data do not indicate a softening of the spectrum within the energy range of our observations.

Zoom In Zoom Out Reset image size

In the framework of the currently accepted EBL models, the observed simultaneous VHE and GeV spectra are consistent with a single power law with index ∼2.7 ± 0.3 between 3 GeV and 400 GeV, without a strong intrinsic cutoff. This evidence suggests that the 100 MeV–400 GeV emission belongs to a unique component, peaking at ≈2–3 GeV, produced in a single region of the jet. If the emission process is inverse Compton (IC) scattering on external photons by relativistic electrons in the jet, as commonly assumed, a strong softening of the spectrum is expected above few tens of GeV if the external photons derive from the BLR. This is due to the combination of two effects: the decreased efficiency of the IC scattering occurring in the Klein–Nishina (KN) regime (e.g., Ghisellini & Tavecchio 2009) and the absorption of γ-rays through pair production (Reimer 2007; Tavecchio & Mazin 2009; Liu & Bai 2006).³⁵

The energy above which the KN effects become important can be roughly expressed as E_KN ≃ 22.5 ν⁻¹_o,15 GeV, where ν_o,15 is the frequency of the target photons in units of 10¹⁵ Hz (or E_KN ≃ 75 λ_μm GeV in wavelength μm units). γ-ray absorption becomes effective when E_γγ ≃ 60 ν⁻¹_o,15 GeV (E_γγ ≃ 200 λ_μm GeV). Above this energy a cutoff is then expected. The importance of both effects in the 10–100 GeV band is reduced if the external photon field is associated with the IR torus (ν_o = 10¹³ Hz), as envisioned by the "far dissipation" scenarios (e.g., Sikora et al. 2008). In that case both effects start to be important above ≈1 TeV. The absence of a spectral break or cutoff in the spectrum observed by MAGIC strongly suggests that the γ-ray emission is not produced within the BLR.

The other important result of the MAGIC observation is the evidence of fast variability, t_var ∼ 10 minutes, indicating an extremely compact emission region with transverse dimensions, R ∼ 1.3 × 10¹⁴(δ/10)(t_var/10 minutes) cm. This seems to be difficult to reconcile with the "far dissipation" scenarios if the emission takes place in the entire cross section of the jet (see also Tavecchio et al. 2010). Estimating the size of the BLR, R_BLR from the accretion disk luminosity, L_disk = 5 × 10⁴⁵ erg s⁻¹ (Fan et al. 2006), the distance of the emitting region is expected to be around d>R_BLR = 3 × 10¹⁷ cm. Assuming a conical jet with constant opening angle θ_j (see, however, the suggestion of recollimation; Marscher 1980), its size would be R ∼ θ_jd ∼ 3 × 10¹⁶(θ_j/5°) cm. The absence of absorption features in the VHE spectrum allows also to exclude absorption within the emitting region and, together with the observed variability, to put a lower limit to the Doppler factor of the source. From Dondi & Ghisellini (1995), Equation (3.7), assuming a power-law photon index 1.5 for the spectrum of the optical target photons, we get a lower limit δ>15, in agreement with Doppler factors derived from radio observations (Section 1).

A possibility to reconcile the spectral information (pointing to emission beyond the BLR) and the fast variability is to invoke the presence of very compact emission regions embedded within the large-scale jet, as already proposed by several authors to explain the exceptionally rapid variability in PKS 2155−304, Mkn 501, and AO 0235+164 (Ghisellini & Tavecchio 2008; Giannios et al. 2009; Marscher & Jorstad 2010). An alternative possibility is that the jet experiences a strong recollimation forming a small emitting nozzle (e.g., Nalewajko & Sikora 2009) as already suggested for M87 and PKS 2155−304 (e.g., Bromberg & Levinson 2009; Stawarz et al. 2006). Alternative scenarios involving proton-driven cascades or proton-synchrotron emission in amplified magnetic fields, e.g., generated by filamentation instabilities (Frederiksen et al. 2010), could also play a role.

In conclusion, the MAGIC observations of VHE emission from the FSRQ PKS 1222+21 put severe constraints on emission models of blazar jets. These results were obtained from a short observation of a flaring source thanks to the collaboration between the MAGIC and Fermi projects. Repeated and hopefully longer observations of flaring blazars with MAGIC and Fermi promise substantial progress in the study of extreme blazars.