2
Detector Array (LHAASO-WCDA, the gold star, RA =288.295◦and Dec =19.772◦, [29]) and Very Long Baseline Array
(VLBA, the grey square, RA =288.264◦and Dec =19.773◦, [32]). Pre-GRB 221009A, just two photon events larger than
100 GeV had been found in 14 years of Fermi-LAT observations within 0.5 degree of GRB 221009A, suggesting a rather low
background level at energies above 100 GeV (One was observed with the energy of 268.1 GeV at the location of RA =288.51◦
and Dec =20.08◦. The other 107.1 GeV photon was located at RA =288.47◦and Dec =19.54◦). Giving its spatial and temporal
coincidence with GRB 221009A, we conclude that this 400 GeV photon is indeed physically associated with this monster. We
calculate the probability that this LAT event belongs to GRB 221009A within the (T0+0.3−T0+1) days interval using the
gtsrcprob tool (in the Fermitools package). It turns out to be 0.9999937, corresponding to a significance level of 4.4σ. Note that
this 400 GeV photon is among the ULTRACLEAN class events and the possibility for being a mis-identification of a cosmic ray
is very low. Therefore, we have identified the most energetic GRB photon detected by Fermi-LAT so far. The previous records
are a 95 GeV photon from GRB 130427A [19] and then a 99.3 GeV photon from GRB 221009A at an early time [23].
An intergalactic magnetic field strength of about 4×10−17 G needed in the cascade scenario. The spectral energy
distributions (SEDs) measured in the time intervals after the burst of 0.05 −0.3 days (blue), 0.3−1 days (yellow) and 0.3−250
days (red) are reported in the panel (a) of Fig. 2(see methods subsection The Fermi-LAT energy spectral analysis). We find that
the emissions in these time intervals show different behaviors: for the 0.05 −0.3 days interval, it is dominated by low-energy
radiation; but for the later time intervals of 0.3−1 days and 0.3−250 days, we only detect the single 400 GeV photon without
accompanying low energy emission. In principle, the delayed GeV-TeV emission could be either from the SSC radiation of the
forward shock electrons or the cascade emission of about 10 TeV prompt gamma-rays. The LHAASO collaboration has reported
that the SSC afterglow from a very narrow jet, as proposed in Ref. [33], can explain the TeV emission within first 2000 s [29].
With multi-band afterglow light curves of this event (see Ref. [34] for more details, and an updated paper with the structured jet
[35]), we obtain the expected SSC afterglow emission for the corresponding time intervals plotted as the dashed lines in panel
(a) of Fig. 2. As for the 0.05 −0.3 days interval, we find the SEDs measured by the Fermi-LAT can be well described by this
SSC model. While in the later time intervals of 0.3−1 days and 0.3−250 days, the fluxes measured by Fermi-LAT around
400 GeV are larger by about 3 orders of magnitude than that expected by the SSC model, which implies that it is challenging to
produce such a 400 GeV photon via the SSC process. Hence we concentrate on exploring the cascade scenario.
Within the cascade scenario, the yielding e±pairs have a Lorentz factor of approximately 107, and the delay of the arrival
time of the secondary GeV−TeV photons is governed by the deflection of the e±pairs by the intergalactic magnetic field. To
account for a delay time of 0.4 days, we need an intergalactic magnetic field strength of approximately 4 ×10−17 G (assuming a
coherence scale of 1 Mpc), which is comparable with limits set by Fermi-LAT observations of TeV blazars [36,37] (see methods
subsection Analytical estimate of the intergalactic magnetic field strength).
Then we conduct Monte Carlo simulation of the cascade scenario with the BIGMF =4×10−17 G to obtain the expected cascade
flux (see methods subsection Numerical simulation of the cascade scenario and the estimate of BIGMF.). As shown in panel (a)
of Fig. 2, the expected emission flux at around 400 GeV in the cascade scenario is higher than that of the SSC model, implying
that the former is more likely the origin of the photon.
Discussion
GRB 221009A is the most powerful gamma-ray bursts detected so far. Thanks to its rather low redshift z=0.151, the emission
has been detected up to the energy of about 13 TeV. Because of the high optical depth of the universe to such energetic gamma
rays, the intrinsic spectrum likely extends to an even higher energy range and most of these primary photons have been absorbed
by the far-infrared background before reaching us. The resulting ultra-relativistic e±pairs will up-scatter and then boost the CMB
photons to sub-TeV energy. Motivated by such a prospect, we analyze the long term of the Fermi-LAT gamma-ray observations
in the direction of GRB 221009A and successfully identified a 400 GeV photon, without accompanying any low-energy gamma
rays, at 0.4 days after GRB 221009A.
Motivated by the facts that the SSC afterglow model is hard to account for the data and a simple analytical estimate suggests
that the cascade scenario can account for the data, we conduct the simulation of cascade with the Elmag 3.03 package and adopted
the intrinsic energy spectrum of the GRB as a power-law (PL) with an index of 2.4 reported by the LHAASO collaboration [29].
In the time intervals of 0.3 −1 days and 0.3 −250 days, the cascade spectra with BIGMF =4×10−17 G peak at several hundred
GeV, and the corresponding probabilities that the Fermi-LAT observed one cascade photon with energy larger than 100 GeV are
about 2.0% and 20.5%, respectively (see methods subsection Numerical simulation of the cascade scenario and the estimate of
BIGMF). Hence, we suggest that the detection of the 400 GeV photon is by chance (i.e., it is a small probability event). Anyhow,
the cascade scenario for the 400 GeV photon is preferred over the SSC model with probabilities higher by a factor of 3 −30.
Although the difference is not very significant, the cascade scenario seems to be a better explanation for the delayed 400 GeV
photon. Note that due to the limited observation, it is difficult to draw a strong conclusion. Considering the uncertainty of
intrinsic spectrum above 13 TeV, we also take the power-law with an exponential cutoff(PLEcut) at 20 TeV model into account
(see methods subsection Uncertainty in the intrinsic energy spectrum). The cascade fluxes above 10 GeV for the PLEcut cases
are also significantly higher than that expected by the SSC model, as shown in Fig. 3. For both two intrinsic spectral models (PL
and PLEcut), the possibility of a weaker IGMF BIGMF =1×10−18 G can be ruled out by the Fermi-LAT upper limits at about
10 GeV in the 0.3 −250 days interval. A higher BIGMF, say ≥10−16 G, is also disfavored because of the resulting lower cascade