Content uploaded by Sushmita Agarwal
Author content
All content in this area was uploaded by Sushmita Agarwal on Jun 10, 2024
Content may be subject to copyright.
,
Draft version May 21, 2024
Typeset using L
A
T
E
Xtwocolumn style in AASTeX631
Imprint of “local opacity” effect in gamma-ray spectrum of blazar jet
Sushmita Agarwal ,1Amit Shukla ,1Karl Mannheim ,2Bhargav Vaidya ,1and Biswajit Banerjee 3, 4
1Department of Astronomy, Astrophysics and Space Engineering, Indian Institute of Technology Indore, Khandwa Road, Simrol, Indore,
453552, India
2Julius-Maximilians-Universit¨at W¨urzburg, Fakult¨at f¨ur Physik und Astronomie, Institut f¨ur Theoretische Physik und Astrophysik,
Lehrstuhl f¨ur Astronomie, Emil-Fischer Str. 31, D-97074 W¨urzburg, Germany
3Gran Sasso Science Institute, Viale F. Crispi 7, L’Aquila (AQ), I-67100, Italy
4INFN - Laboratori Nazionali del Gran Sasso, L’Aquila (AQ), I-67100, Italy
ABSTRACT
Relativistic jets from accreting supermassive black holes at cosmological distances can be powerful
emitters of γ-rays. However, the precise mechanisms and locations responsible for the dissipation of
energy within these jets, leading to observable γ-ray radiation, remain elusive. We detect evidence
for an intrinsic absorption feature in the γ-ray spectrum at energies exceeding 10 GeV, presumably
due to the photon−photon pair production of γ-rays with low-ionization lines at the outer edge of
broad-line region (BLR), during the high-flux state of the flat-spectrum radio quasar PKS 1424−418.
The feature can be discriminated from the turnover at higher energies resulting from γ-ray absorption
in the extragalactic background light. It is absent in the low-flux states, supporting the interpretation
that powerful dissipation events within or at the edge of the BLR evolve into fainter γ-ray emitting
zones outside the BLR, possibly associated with the moving very long baseline interferometry radio
knots. The inferred location of the γ-ray emission zone is consistent with the observed variability time
scale of the brightest flare, provided that the flare is attributed to external Compton scattering with
BLR photons.
Keywords: Blazars (164); Gamma-ray astronomy (628); Relativistic jets (1390); Galactic and extra-
galactic astronomy(563); Flat-spectrum radio quasars(2163); High energy astrophysics(739)
1. INTRODUCTION
Accreting supermassive black holes spew out colli-
mated, relativistic jets. Specifically when aligned with
our line of sight, the emission from the jets is Doppler
boosted, thus rendering them visible up to high red-
shifts. In a leptonic scenario, blazars—aligned jetted
objects—emit high-energy radiation through the upscat-
tering of soft seed photons. These soft photons could
originate within the jet through the synchrotron-self-
Compton process or could be contributed by an external
radiation field (Ghisellini & Tavecchio 2009).
Particularly in flat spectrum radio quasars (FSRQs),
the observed high Compton dominance implies an in-
creased influence of external seed photons from the
broad-line region (BLR ; Ghisellini & Tavecchio 2009).
Corresponding author: Sushmita Agarwal
Email: sush.agarwal16@gmail.com
Typically FSRQs are characterized by high black hole
masses, enhanced radiative efficiency from accretion
disks, and accretion rates near the Eddington limit.
As a result, they possess a luminous disk (Maraschi &
Tavecchio 2003). Part of disk radiation is reprocessed,
contributing to the luminosity of the BLR and torus,
thereby influencing the external jet environment close
to the black hole (Ghisellini et al. 2011;Sbarrato et al.
2012). γ-ray photons produced near the black hole on
propagating through this dense photon environment, are
expected to leave an imprint of photon−photon pair
creation at energies from 10 to 200 GeV (Liu & Bai
2006). Thus, an expected observational signature of pro-
duction of high-energy photons within BLR, occurring
within parsec scales from central engine, would mani-
fest as a cutoff feature at E > 10 GeV in high-energy
spectrum. However, the absence of such BLR-induced
cutoff in high-energy spectrum of FSRQs raises con-
cerns about the Compton dominance typically observed
arXiv:2405.09612v2 [astro-ph.HE] 18 May 2024
2
in such sources (Costamante et al. 2018). The absence
of cutoff above 10 GeV indicates that the emission re-
gion of high-energy photons is outside the influence of
the BLR. Multiple detection of very high-energy TeV
photons from FSRQs such as 3C 279 (Aleksi´c et al.
2011a), PKS 1510−089 (H. E. S. S. Collaboration et al.
2013,MAGIC Collaboration et al. 2018), PKS 1222+216
(Aleksi´c et al. 2011b), and PKS 1441+25 (Abeysekara
et al. 2015), further supports the possibility of high-
energy emission region beyond the BLR (Liu & Bai
2006;Donea & Protheroe 2003). In contrast to the pre-
vious results, Poutanen & Stern (2010) presented evi-
dence of a break in the high-energy spectrum. The idea
gains further support from Fermi-LAT observation of
3C 454.3 and 4C +21.35 (Le´on-Tavares et al. 2013;Isler
et al. 2013;Tanaka et al. 2011;Stern & Poutanen 2014).
This constrains the emission site within the region of
influence of the BLR. Additionally, the detection of fast
variability typically observed in FSRQs implies the pro-
duction of high-energy γ-rays towards the edge of the
BLR via a magnetic reconnection scenario (Shukla &
Mannheim 2020;Agarwal et al. 2023).
In this context, we studied a high-redshift (z=
1.522) FSRQ PKS 1424−418 through flux-resolved spec-
troscopy. The source recently exhibited exceptional out-
bursts during 2022, reaching 10 times the average flux
level. The large black hole mass of 4.5×109M⊙in
PKS 1424−418 provides a possibility of strong accre-
tion rate and thus sufficient seed photons from BLR for
the observed Compton dominance (q∼30; Abhir et al.
(2021)). The structure of the paper is as follows: Sec-
tion 2discusses the methods and techniques, Section 3
presents the results, and a discussion is provided in Sec-
tion 4. Our results are summarized in Section 5.
2. METHODS AND TECHNIQUES
2.1. Data Analysis: Fermi-LAT
To examine the varying flux states of PKS 1424−418,
we have chosen ≈15 yr of γ-ray data within the energy
range of 0.1−300 GeV. This data were recorded by
the Large Area Telescope (LAT) on board the Fermi
Gamma-ray Space Telescope, spanning from 2008 Au-
gust 4, to 2023 March 21 (Atwood et al. 2009). Standard
analysis procedure1provided by Fermi Science tools and
the open-source Fermipy package (Wood et al. 2017) are
used to analyze data for the source in the energy range
between 0.1−300 GeV using the latest instrument re-
sponse function P8R3 SOURCE V3. All the photons com-
1https://fermi.gsfc.nasa.gov/ssc/data/analysis/documentation/
Cicerone/Cicerone Data Exploration/Data preparation.html
ing within 20◦of the source location have been analyzed
to account for the broad PSF of the telescope. Addi-
tionally, a zenith angle cut of 90◦,GTMKTIME cut of
DATA QUAL >0 && LAT CONFIG==1 and evtype=3 were
used in the analysis. Only those events are considered
for analysis that is highly probable of being photons
by applying a GTSELECT cut on the event class to ac-
count for the SOURCE class event using evclass=128.
Spectral analysis on the resulting data set was carried
out by including gll iem v07 and the isotropic diffuse
model iso P8R3 SOURCE V2 v1. A region of interest of
10◦centered at the source was used for analysis. To ac-
commodate and model the photons coming from vicin-
ity of the source of interest, the spectral parameters of
sources within 5◦of the region of interest were allowed to
vary. Additionally, sources with a variability index >25
within 15◦were made to vary. However, spectral param-
eters of all the sources outside 5◦of the region of interest
and with variability index <25 were fixed to their 4FGL
catalog values. The flux and spectrum of PKS 1424−418
were determined by fitting a log-parabola model, using
a binned gtlike algorithm based on the NewMinuit op-
timizer. A test statistics (TS) >9 suggests a detection
significance of more than 3σ(√TS ∼3). The TS is
defined as TS = −2 ln(L1/L0) where L0and L1denote
the likelihood value without and with the point source
at the position of interest, respectively.
2.2. Identification of Bright flaring epochs in light
curve (Bayesian Block and HOP Algorithm)
Only periods of significant detection are further con-
sidered for analysis. In this work, a time bin in Fermi-
LAT light curve is considered a detection if it has TS >9
and if flux in one bin is greater than its uncertainty, i.e.
Ft> σt.
We represent the flux points and the associated uncer-
tainties in step-function representation using Bayesian
Block (BB) to detect and characterize variability struc-
tures localized in time (Scargle et al. 2013;Agarwal et al.
2023). Each point of change of the block in the step-
function representation highlights the 3σvariation from
the previous block. To identify the flaring features in the
light curve, the BB is fed into the HOP algorithm, which
is based on a watershed concept used in topological data
analysis (Eisenstein & Hut 1998). This segregates the
light-curve periods into multiple flaring groups (further
identified as HOP flaring groups) and quiescent groups
based on the emergence of BB above the mean flux level.
This has been used in Meyer et al. (2019) to zoom into
periods of flaring epochs and identify signatures of com-
pact emission regions in FSRQs. The flare identification
code by Wagner et al. (2021) is used to identify the dif-
ferent HOP groups. The above classification resulted
3
55000 56000 57000 58000 59000 60000
Time [MJD]
0
1
2
3
4
Flux × 10 6
[Ph cm 2 s 1]
Fermi-LAT
14 day binned
(a)
HOP group
Mean flux level
BB
10 1100101102
Energy [GeV]
10 12
10 11
10 10
E
2
dN
/
dE
[erg cm 2s1]
High state : = 2.0 ± 0.004, = 0.08 ± 0.003
Low state : = 2.15 ± 0.01, = 0.1 ± 0.005
174 231 243 202 149 91
45
6
85 121 128 105 72 45
21
5
(b)
Figure 1. (a) 14 day binned light curve of ≈15 yr of Fermi-LAT observation of PKS 1424−418. The gray region represents
high state and white represents low state (b) The combined spectrum of all “flaring” and “quiescent” epochs marked in panel
(a). Intrinsic spectrum is modeled with a log-parabola model up to 10GeV (solid line with envelope) and then extrapolated up
to ∼300 GeV (dotted lines). The significance of detection (√TS) for each energy bin is mentioned over the respective bins.
in 11 HOP flaring groups and 11 quiescent periods as
indicated in Fig. 1(a).
2.3. Flux distribution
The assessment of flux patterns for various flaring
groups of PKS 1424−418 is conducted based on the tech-
niques presented in the study by Acciari et al. (2021) and
Agarwal et al. (2023). These flux profiles are then sub-
jected to fitting processes using the following functions:
1. Gaussian:
G(x; µG, σG) = NG
σG√2πexp −(x −µG)2
2σ2
G(1)
2. Lognormal :
LN(x; µLN, σLN ) = NLN
xσLN√2πexp −(log(x) −µLN )2
2σ2
LN
(2)
Here, Ni,µiand σiare the normalization constant,
mean, and standard deviation for the fitted profiles, re-
spectively (i= G or LN indicating Gaussian or Lognor-
mal profile). The selection of the more favorable distri-
bution fit is determined by considering the Akaike infor-
mation criterion (AIC) values (Akaike 1974). A lower
AIC for a model means a better description of the data.
3. RESULT
3.1. Activity Periods
A visual inspection of high-energy γ-ray (0.1−
300 GeV) light curve over ∼15 yr displays distinct pe-
riods of high activity and low activity. A significant
highlight of this period was the extended flare in 2022,
which manifested as a pronounced outburst observed in
optical (ATOM; Jankowsky et al. 2022), radio (ATCA;
4
Kadler et al. 2022), and γ-ray (Fermi-LAT and AGILE;
La Mura 2022;Verrecchia et al. 2022) emissions. Flaring
groups, denoted by gray patches in Fig. 1(a), are inter-
spersed with multiple quiescent groups, represented by
white patches in Fig. 1(a). These flaring and quies-
cent groups are identified using the HOP algorithm, as
detailed in section 2.2.
We examine the collective high-energy γ-ray spectrum
(0.1−300 GeV) of all the photons during the “flaring”
and “quiescent” periods. The culmination of all the
flaring groups and quiescent groups are hereby referred
to as the “high state” and “low state”, respectively, in
the paper. The γ-ray spectrum is modeled using a log-
parabola model, parameterized as:
dN
dE =N◦E
Eb−(α+βlog(E/Eb))
(3)
where Ebwas fixed to 4FGL catalog value of
677.45 MeV and Nois the Normalization. We also tested
with the power-law model and found that γ-ray spec-
trum favors a log-parabola model over a power-law, with
a TS value of 231 and 151 during high and low state,
respectively.
For PKS 1424−418, a redshift of z= 1.522 implies a
resulting attenuation due to optical-UV-near-IR extra-
galactic background light (EBL) radiation beyond criti-
cal energy, Ecrit ≈170(1+z)−2.38 GeV = 18.8 GeV (Ack-
ermann et al. 2012). Poutanen & Stern (2010) found a
significant break in bright blazars indicating γ-ray ab-
sorption via photon−photon pair production on helium
II (He II) and hydrogen (H) recombination continuum
photons. Stern & Poutanen (2014) observed a signifi-
cant 20 GeV break in the source frame due to the H
Lyman continuum (LyC) using revised Pass 7 Fermi-
LAT response function and found that any break due
to He II is less significant. Any spectral features in soft
spectrum at energy Esoft interact with γ-ray photons at
energy Ehard and are manifested as an observed attenu-
ation beyond the threshold energy
Eth ≳(mec2)2
Esoft(1 + z)(1 −cos θ)≃10 10 eV
Esoft,LyαGeV
(4)
which is the minimum energy for absorption in a head-
on collision (θ= 180◦). BLR photons may influence the
high-energy spectrum only beyond 10 GeV (equation 4).
Thus, the spectrum below ≈10 GeV is a true estimate
of the unabsorbed intrinsic spectrum of the source.
Combined high-energy γ-ray spectrum of high and low
states up to 10 GeV are fitted with log-parabola model
using equation 3and have consistent values of βparam-
eters as in Fig. 1(b). Since EBL attenuation and BLR
influence is relevant only beyond 10 GeV, consistency in
0.00 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00
MJD (59933+)
0
5
10
15
20
Flux × 10 6
[Ph cm 2 s 1]
Fermi-LAT
Orbit binning (96min) BB
P1 : 0.15 ± 0.06 day [3.8 ]
P2 : 0.18 ± 0.07 day [4.3 ]
P1 P2
0
20
40
60
80
100
Significance
Figure 2. Orbit-binned light curve of a period of fastest
variability in Fermi-LAT of PKS 1424−418. Bayesian Blocks
with a false alarm probability of 5% indicating significant
points of change are plotted on top. The periods of signifi-
cant variability are marked with grey dotted lines.
the βparameter for low- and high-state spectrum hints
at a similar influence of external seed photons up to 10
GeV.
3.2. Fast variability
During the most intense flaring episodes in 2022, the
flux of PKS 1424−418 surged to approximately 10 times
the ≈15 yr flux average. On 2022 December 20, Fermi-
LAT recorded a remarkably fast variability during the
source’s brightest flare. The orbit-binned light curve
(≤96 minutes) in Fig. 2illustrates the variable flux,
with its 3σvariation represented by the correspond-
ing BB. Recently, Fermi-LAT captured a significant
(3.8 σ) intraday variability of 0.15 ±0.06 days during
the brightest flux state of the source (Fig. 2).
The source is visibly variable during the brightest ac-
tivity period between MJD 59760 and 59961, statisti-
cally evident from the pink noise behavior down to 6
hr (power-law index ≈1.16 ±0.25, details in Agarwal
et al 2024, in preparation). With a black hole mass of
MBH = 4.5×109M⊙, light-crossing timescales of ≈0.5
days is a measure of minimum expected variability in
the jet frame (Spada et al. 2001).
Between MJD 59758 and 60024, the flux doubled in
(tvar)obs = 0.15 ±0.06 days shown in Fig. 2. We
quantify the variability timescales associated with these
points of change using tvar = (t2−t1)ln2
ln(F2/F1)(Fos-
chini et al. 2011), where F2and F1are the fluxes at
time t2and t1respectively, and tvar is the flux doubling
and halving timescales. Such short variability timescales
place a tight constraint on the size and location of the
emission region. Here, the minimum size of the emission
region is of the same order as the size of the black hole.
This is the fastest-ever recorded variability in the source
to date. Previously, Abhir et al. (2021) observed a vari-
ability of 3.6 days (4.74σ) during 2012 in the variable
periods of MJD 56015–56020. The observed variability
5
Table 1. Parameters of flux distribution for Low and High state
State Model Mean (µi) Sigma (σi) AIC
(1) (2) (3) (4) (5)
High state Gaussian 3.74 ±0.01 2.00 ±0.01 −55999.5
Lognormal 1.43 ±0.01 0.57 ±0.01 −65411.3
Low state Gaussian 7.36 ±0.01 4.16 ±0.01 −47024.6
Lognormal 2.11 ±0.01 0.51 ±0.01 −35417.7
Note—(1) The state of the light curves —high state or low state.
(2) The models tested on the obtained flux distribution. (3)
Mean of the fitted model. (4) Standard deviation of the fitted
model. (5) AIC values for the fitted model.
constrains the emission blob at a maximum radius of
r′
emm =ctvarδ/(1 + z)=1.5×1015(δ/10) cm. The typ-
ical dissipation distance from supermassive black hole
for a radiation region of size r′
emm, is Rdiss = 2cΓ2
jtvar =
0.025 (δ/10)2pc, assuming the Doppler factor of the blob
δ= Γj(Rieger 2019). Assuming the emission region
covers the entire jet cross section, the 0.025 pc emission
region could be in the vicinity of seed photons from BLR
or accretion disk.
3.3. Flux Distribution
The evaluation of flux patterns across high and low
state is performed using flux distribution as described
in section 2.3. The AIC value suggests high states pre-
fer a log-normal distribution, contrasting with the Gaus-
sian distribution for low states (see Table 1). Lognor-
mal fluctuation hints at a multiplicative process at play,
typically known for accreting galactic sources like X-ray
binaries. This provides hints of the influence of accre-
tion disk on the jet (Uttley et al. 2005), or a minijet
in the jet model due to the Pareto distribution (Biteau
& Giebels 2012) possibly due to magnetic reconnection
scenario at the edge of the BLR as studied in Agarwal
et al. (2023).
3.4. EBL attenuation
The unabsorbed intrinsic spectrum of the source is
estimated from 0.1−10 GeV and extrapolated further
to higher energies up to 100 GeV as in Fig. 1(b). We use
reduced chi-square (χ2) to test if the intrinsic spectrum
is a true representation of the observed spectrum up to
100 GeV as in Costamante et al. (2018). For the high
state, the intrinsic spectrum is rejected with a p-value
<10−5. For low-state, the model is rejected with a p-
value ∼10−4. Notably, the main contribution for high
χ2comes from the high-energy end (E > 10 GeV) of
the spectrum (see Fig. 1(b)).
The observed spectrum shows a contrasting deviation
from the extrapolated intrinsic fit, with a 21.8σdevia-
tion for photons in the high state and a 2.8σdeviation
in the low state within the range of 40−95 GeV. This
deviation is most likely caused by the absorption of γ-
ray photons interacting with EBL photons alone or with
EBL photons in combination with soft photons from the
local jet environment (accretion disk, torus, or BLR)
during their journey to the observer.
To account for EBL absorption, the extrapo-
lated intrinsic spectrum is modified with an ex-
ponential term, e−τ(E,z)
γ,γ , such that Fobs (E) =
Fint(E) exp[−τγ,γ (E , z)] (Kneiske et al. 2004). Here,
τγ,γ (E , z) = b×τmodel
γ,γ (E , z), where τmodel
γ,γ (E , z) is the
predicted optical depth by various EBL models and b
represents the opacity scaling factor. Considering the
level of EBL absorption as described by the model, the
value of bhighlights either of two conditions: (1) b= 0
highlights no EBL attenuation, (2) b= 1 indicates cor-
rect model selection, hence the correct estimate of EBL
absorption.
We examined the observed high- and low-state spec-
trum with 15 EBL-absorbed spectra (Fobs ) with bvalue
ranging from 0.75 - 1.25 to accommodate 25% tolerance
on the optical depths (Ackermann et al. 2012) predicted
by models at different energies ( Kneiske et al. 2004,
Gilmore et al. 2012,Kneiske & Dole 2010,Finke et al.
2010,Franceschini et al. 2008,Dom´ınguez et al. 2011,
Helgason & Kashlinsky 2012,Inoue et al. 2013,Scully
et al. 2014,Stecker et al. 2006). Among the 15 EBL-
absorbed spectra derived from 15 models above, 12 mod-
els resulted in χ2close to 1 in the low state, but follow
a deviation with a significantly high χ2in the high state
as listed in Fig. 3. Since EBL is uniform and isotropic
on large scales, the expected levels of γ-ray absorption
from a source should be flux-independent. Thus, the
high and low states are expected to have similar lev-
els of absorption from the photon−photon absorption in
presence of EBL photons. However, none of the present
EBL models can efficiently explain the absorption levels
beyond 10 GeV for the high state using EBL absorption
alone. A certain level of extra absorption is imprinted on
the high-state spectrum reflecting an additional effect of
intervening interacting photons. Conservatively, the ob-
served high-state spectrum deviates by ∼4.7σbeyond
10 GeV from the EBL-absorbed spectra derived using
the Scully et al. (2014) – high-opacity model. Addition-
ally, for the more widely used EBL model, by Frances-
chini et al. (2008) and Dom´ınguez et al. (2011), the ab-
6
sorption significance beyond E>10 GeV are found to be
>5σ.
4. DISCUSSION
Spectral breaks in γ-ray spectrum are expected at dif-
ferent energies, owing to (a) internal absorption through
photon−photon interactions with external seed photons,
primarily from the BLR, accretion disk, and dusty torus,
and (b) absorption of high-energy photons by the EBL in
the optical, UV, and near-IR bands. Such multiple ab-
sorptions make the detection of high-energy photons in
high-redshift objects extremely challenging due to poor
photon statistics. We report a significant deviation of
the stacked high-state and low-state spectrum from the
fitted log-parabola model at E > 10 GeV as evident in
Fig. 1(b).
At energy beyond 10 GeV, two energy bins have
greater than 3σdeviation in the high state (Fig. 1(b)).
During the low state of the source, the deviation is lim-
ited to 0.6σwithin 17−40 GeV and 2.8σ, within 40 −95
GeV. This deviation is significantly smaller than 4σand
21.8σdeviation within 17 −40 GeV and 40 −95 GeV,
respectively, during the high state of the source (Fig.
1(b)). This deviation carries the imprint of absorbed
high-energy photons during the high states of the source.
The imprint includes absorption from both EBL and ex-
ternal jet photons (i.e. BLR, torus, or accretion disk),
contingent on the emission site’s position during high
states. The absence of the absorption feature in the low
state indicates that external photons, apart from the
EBL, could have a notable impact on the γ-ray spec-
trum during high states. Low-activity periods are typ-
ically associated with the outer parsec-scale regions of
the jet or result from combined emissions along the en-
tire jet length in the absence of a dominant emission
zone. High-activity periods are primarily linked to emis-
sion originating from energetic particles within the in-
ner jet at parsec scales from black hole (Ezhikode et al.
2022). Additionally, high Compton dominance in the
source (q∼30; Abhir et al. 2021) indicates that accel-
erated high-energy electrons in the jet scatter a fraction
of soft photons, emitting γ-rays. This process neces-
sitates proximity of high-energy electrons to dominant
sites of the soft photons, such as the accretion disk, BLR,
or dusty torus. The observed variability of 0.15 ±0.06
days constrains the emission region’s location to be at a
minimum distance of Rdiss >0.025 pc from the central
black hole. In Agarwal et al. (2024, in preparation),
aEHE,max = 65 GeV photon, alongside ∼200 photons
with E > 10 GeV, was detected with over 99% prob-
ability of association with PKS 1424−418 during the
2022 flaring activity. For a flat BLR, γ-ray emission
site is constrained at rmin =rBLR/tan θmin ≃0.45 pc
(Nalewajko et al. 2012) from a central super massive
black hole where rBLR = 0.5 pc (using Ldisk = 2.5×
1047 erg s−1as in Abhir et al. 2021 and Buson et al.
2014 and rBLR = 0.1 pc×(Ldisk/1046)1/2from Nalewa-
jko et al. 2012). The minimum collision angle θmin at
Eth =EHE,max is given from equation 4as,
θmin = arccos 1−2(mec2)2
(1 + z)EHE,maxEsoft,Lyα≃47◦.
(5)
At 0.45 pc, both the BLR and accretion disk could
contribute as sources of external seed photons. However,
if the BLR has a significant ‘tail’ for r > rBLR, the high-
energy emission is produced at least within the parsec
scale.
Hence, the observed attenuation could be attributed
to the “local opacity” effect caused by the BLR and
accretion disk photon. To explore this further, we in-
vestigate potential sources of change in γ-ray opacity in
a local jet scenario.
4.1. Imprints of BLR
Due to varying degrees of ionization, the BLR emits
several strong line features. Poutanen & Stern (2010)
synthesize the most significant features in the BLR
spectrum and its effect on high-energy photon propa-
gation. The interaction of BLR line photons with jet
photons could result in photon−photon absorption at
energy threshold of pair production. The energy thresh-
old for photon−photon pair production depends on col-
lision angle as well energy of soft seed photon, such
that, Eth ∝1/Esoft(1 −cos θ). Thus, the role of angu-
lar distribution of external radiation near the emission
zone becomes important, emphasizing the contribution
of BLR geometry and position of emission site with re-
spect to BLR for changing opacity of γ-ray radiation
(Tavecchio & Ghisellini 2012,Lei & Wang 2014,Abol-
masov & Poutanen 2017). For an emission region within
BLR at r=RBLR, isotropically distributed BLR radi-
ation around the emission region leads to increased op-
tical depth. Toward the outer edge of the BLR, the
optical depth decreases rapidly, due to a shift from
head-on collisions (θ= 180◦) to less favorable angles
of θ < 90◦(Tavecchio & Ghisellini 2012). This decrease
is much more pronounced near the opacity threshold
energy (Abolmasov & Poutanen 2017). Typically, spec-
tral breaks are imprinted on the high-energy spectrum
by two emission lines, hydrogen Lyα(H Lyα) and he-
lium II Lyα(He Lyα), produced near the central en-
gine. Abolmasov & Poutanen (2017) emphasized contri-
butions from various emission lines at larger distances,
dominated mostly by low-ionization lines.
7
10 12
10 11
10 10
Kneiske+ 2004 -- best fit
2
red
,
high
= 4.51
2
red
,
low
= 1.02
Gilmore 2012 -- Fiducial
2
red
,
high
= 4.74
2
red
,
low
= 1.04
Intrinsic spectrum -- Low state
Intrinsic spectrum -- High state
Observed spectrum -- low state
Observed spectrum -- high state
Kneiske and Dole 2010
2
red
,
high
= 7.15
2
red
,
low
= 1.09
10 12
10 11
10 10
Finke+ 2010
2
red
,
high
= 8.64
2
red
,
low
= 1.18
Frankenschini+ 2008
2
red
,
high
= 11.24
2
red
,
low
= 1.25
Dominguez+ 2011
2
red
,
high
= 11.87
2
red
,
low
= 1.3
10 12
10 11
10 10
Gilmore 2012 -- fixed
2
red
,
high
= 10.0
2
red
,
low
= 1.21
Gilmore 2012
2
red
,
high
= 9.4
2
red
,
low
= 1.19
Helgason and Kashlinsky 2012
2
red
,
high
= 10.26
2
red
,
low
= 1.25
10 12
10 11
10 10
Inoue 2013
2
red
,
high
= 4.43
2
red
,
low
= 1.03
Scully et al 2014 -- high Opacity
2
red
,
high
= 3.61
2
red
,
low
= 1.34
Scully et al 2014 -- Low Opacity
2
red
,
high
= 7.14
2
red
,
low
= 1.09
10 1100101102
10 12
10 11
10 10
Kneiske+ 2004 -- high UV
2
red
,
high
= 6.45
2
red
,
low
= 2.33
10 1100101102
Stecker 2006 -- baseline
2
red
,
high
= 13.72
2
red
,
low
= 4.16
10 1100101102
Stecker 2006 -- fast evolution
2
red
,
high
= 15.61
2
red
,
low
= 4.7
Energy [GeV]
E
2
dN
/
dE
[erg cm 2s1]
Figure 3. Observed γ-ray spectrum in 0.1−100 GeV range for the high state and low state. The true intrinsic spectrum in the
energy range of 0.1−10 GeV extrapolated up to 100 GeV for the high and low state is shown in gray dashed and dotted-dashed
lines, respectively. The solid red and blue envelope represents the intrinsic high-state and low-state spectrum, respectively, after
EBL attenuation using opacity scaling factor (b) ranging from 0.75 to 1.25 across different models indicated in each panel. The
mean value of the fit corresponding to b=1 is indicated by a black dashed line. The quoted reduced χ2in each panel is for the
best-fit corresponding to b=1.
8
At rmin = 0.45 pc, toward the edge of the BLR, low-
ionization lines become more influential. This is a re-
sult of large collision angles, in contrast to the small
collision angles associated with high-ionization lines lo-
calized within the inner boundaries of the BLR. Absorp-
tion in the source frame, around ∼10 −30 GeV, may
result from increase in γ-ray opacity due to H Lyαand
LyC (with EBLR,Lyα= 10.2 eV and EBLR,LyC = 13.6
eV). While the resulting break could shift to higher en-
ergies, efficiency of photon−photon absorption may be
reduced with decreased collision angles. The observed
absorption around 100/(1 + z) - 140/(1 + z) GeV cannot
be attributed to the dominant H Lyαand LyC in the
BLR. However, several strong lines at lower energies,
such as Balmer Hαand Hβ(with EBLR,Hα= 1.89eV
and EBLR,Hβ= 2.55 eV), may contribute to the noted
absorption within this energy range (Sol et al. 2013).
Interestingly, the position of emission site at 0.45 pc,
at the outer edge of the BLR, should be dominated by
the low-ionization Balmer lines resulting in absorption
at 40 −95 GeV in observer’s frame. Toward the edge of
the flat BLR, head-on collisions are significantly reduced
as collision angles are limited to θ < 90◦. This should
result in doubling the break energy (Stern & Poutanen
2014). However, if a few BLR photons are positioned
along the jet axis for head-on collisions outside the γ-ray
emitting region, the observed absorption at 40−95 GeV
due to Hα,Hβ could be justified. We implemented the
BLR cutoff on the EBL-absorbed high-state spectrum,
employing the commonly used EBL model by Frances-
chini et al. (2008). The fitting highlighted the impact
of low-ionization lines, as indicated by a significant im-
provement in the χ2from ∼11.2 to ∼2.5, with the
inferred cutoff at 44 GeV. The reduced χ2larger than 1
for the BLR cutoff model is possible due to the choice
of a single exponential cutoff model, which could be too
simplistic for the much more complex contribution from
the BLR lines and continua.
4.2. Imprints of Accretion disk
Absorption beyond 10 GeV indicates soft photon en-
ergies causing attenuation are within Esoft = 1 to 10
eV. While EBL origin would imply consistent absorp-
tion across flux levels, higher absorption during the flar-
ing state suggests a different origin for these optical
photons, possibly from accretion disk at the jet’s base.
PKS 1424−418 displays a bright disk emission, domi-
nant within the optical range (Abhir et al. 2021;Buson
et al. 2014).
A lognormal distribution of flux in high-energy γ-rays
during high state, in contrast with Gaussian distribution
during low state implies distinct emission characteristics
in two states. The preferred lognormal distribution indi-
cates the possible influence of disk photons on jet specif-
ically during high states (Rieger 2019) or a minijet in jet
model due to the magnetic reconnection scenario at the
edge of the BLR (Biteau & Giebels 2012;Agarwal et al.
2023).
At rmin = 0.45 pc, the radiation field from the accre-
tion disks (dominant up to ≈1017 cm ∼0.03 pc) has
a minimal contribution to the opacity of γ-rays. The
alignment of the soft disk photons with the direction of
the γ-rays from the jet reduces their interaction rates
significantly, making the imprint of accretion disk pho-
tons on the high-energy spectrum highly unlikely (Abol-
masov & Poutanen 2017). Additionally, for reasonable
Thomson optical depths of the hot intercloud medium,
the disk photons scattering off free electrons also do not
significantly contribute to the photon−photon pair pro-
duction optical depth. Furthermore, unabsorbed pho-
tons, postaccretion disk influence, undergo additional
absorption from BLR photons scattering at larger an-
gles, resulting in higher opacity. Consequently, the ob-
served γ-ray site is unlikely to be within the inner radius
of the BLR.
5. SUMMARY
In this Letter, we report significant absorption of γ-ray
photons at energies exceeding 10 GeV, possibly due to
photon−photon pair production on low-ionization BLR
photons from the outer edge of the BLR. We detected
this absorption signature in PKS 1424−418 with high
significance during a high state. Conversely, the absence
of this absorption signature in the low state suggests
the presence of an emission region located further away
from the BLR. The absorption feature during the high
state, in contrast with the absence of it during the low
state, supports the interpretation that powerful emis-
sion events originating within or at the edge of the BLR
evolve into fainter emission components outside of the
BLR. This is further supported by the detection of fast
variability during the high state consistent with a size
of the emission site of 0.45 pc which is within the outer
edge of the BLR.
Acknowledgements We thank the anonymous referee
for constructive criticism, which has helped improve the
manuscript. This research work has made use of data,
software and web tools obtained from NASA’s High
Energy Astrophysics Science Archive Research Center
(HEASARC) and Fermi gamma-ray telescope Support
centre, a service of the Goddard Space Flight Center
and the Smithsonian Astrophysical Observatory. BB ac-
knowledges financial support from the Italian Ministry
of University and Research (MUR) for the PRIN grant
METE under contract no. 2020KB33TP.
9
REFERENCES
Abeysekara, A. U., Archambault, S., Archer, A., et al.
2015, ApJL, 815, L22, doi: 10.1088/2041-8205/815/2/L22
Abhir, J., Joseph, J., Pgm, S. R., & Bose, D. 2021,
MNRAS, 501, 2504, doi: 10.1093/mnras/staa3639
Abolmasov, P., & Poutanen, J. 2017, MNRAS, 464, 152,
doi: 10.1093/mnras/stw2326
Acciari, V. A., Ansoldi, S., Antonelli, L. A., et al. 2021,
MNRAS, 504, 1427, doi: 10.1093/mnras/staa3727
Ackermann, M., Ajello, M., Allafort, A., et al. 2012,
Science, 338, 1190, doi: 10.1126/science.1227160
Agarwal, S., Banerjee, B., Shukla, A., et al. 2023,
MNRASL, 521, L53, doi: 10.1093/mnrasl/slad023
Akaike, H. 1974, IEEE Transactions on Automatic Control,
19, 716, doi: 10.1109/TAC.1974.1100705
Aleksi´c, J., Antonelli, L. A., Antoranz, P., et al. 2011a,
A&A, 530, A4, doi: 10.1051/0004-6361/201116497
—. 2011b, ApJL, 730, L8, doi: 10.1088/2041-8205/730/1/L8
Atwood, W. B., Abdo, A. A., Ackermann, M., et al. 2009,
ApJ, 697, 1071, doi: 10.1088/0004-637X/697/2/1071
Biteau, J., & Giebels, B. 2012, A&A, 548, A123,
doi: 10.1051/0004-6361/201220056
Buson, S., Longo, F., Larsson, S., et al. 2014, A&A, 569,
A40, doi: 10.1051/0004-6361/201423367
Costamante, L., Cutini, S., Tosti, G., Antolini, E., &
Tramacere, A. 2018, MNRAS, 477, 4749,
doi: 10.1093/mnras/sty887
Dom´ınguez, A., Primack, J. R., Rosario, D. J., et al. 2011,
MNRAS, 410, 2556,
doi: 10.1111/j.1365-2966.2010.17631.x
Donea, A. C., & Protheroe, R. J. 2003, Progress of
Theoretical Physics Supplement, 151, 186,
doi: 10.1143/PTPS.151.186
Eisenstein, D. J., & Hut, P. 1998, ApJ, 498, 137,
doi: 10.1086/305535
Ezhikode, S. H., Shukla, A., Dewangan, G. C., et al. 2022,
ApJ, 939, 76, doi: 10.3847/1538-4357/ac9627
Finke, J. D., Razzaque, S., & Dermer, C. D. 2010, ApJ,
712, 238, doi: 10.1088/0004-637X/712/1/238
Foschini, L., Ghisellini, G., Tavecchio, F., Bonnoli, G., &
Stamerra, A. 2011, A&A, 530, A77,
doi: 10.1051/0004-6361/201117064
Franceschini, A., Rodighiero, G., & Vaccari, M. 2008, A&A,
487, 837, doi: 10.1051/0004-6361:200809691
Ghisellini, G., & Tavecchio, F. 2009, MNRAS, 397, 985,
doi: 10.1111/j.1365-2966.2009.15007.x
Ghisellini, G., Tagliaferri, G., Foschini, L., et al. 2011,
MNRAS, 411, 901, doi: 10.1111/j.1365-2966.2010.17723.x
Gilmore, R. C., Somerville, R. S., Primack, J. R., &
Dom´ınguez, A. 2012, MNRAS, 422, 3189,
doi: 10.1111/j.1365-2966.2012.20841.x
H. E. S. S. Collaboration, Abramowski, A., Acero, F., et al.
2013, A&A, 554, A107,
doi: 10.1051/0004-6361/201321135
Helgason, K., & Kashlinsky, A. 2012, ApJL, 758, L13,
doi: 10.1088/2041-8205/758/1/L13
Inoue, Y., Inoue, S., Kobayashi, M. A. R., et al. 2013, ApJ,
768, 197, doi: 10.1088/0004-637X/768/2/197
Isler, J. C., Urry, C. M., Coppi, P., et al. 2013, ApJ, 779,
100, doi: 10.1088/0004-637X/779/2/100
Jankowsky, F., Ait-Benkhali, F., Zacharias, M., & Wagner,
S. J. 2022, The Astronomer’s Telegram, 15552, 1
Kadler, M., Stevens, J., Ojha, R., Edwards, P. G., &
Roesch, F. 2022, The Astronomer’s Telegram, 15536, 1
Kneiske, T. M., Bretz, T., Mannheim, K., & Hartmann,
D. H. 2004, A&A, 413, 807,
doi: 10.1051/0004-6361:20031542
Kneiske, T. M., & Dole, H. 2010, A&A, 515, A19,
doi: 10.1051/0004-6361/200912000
La Mura, G. 2022, The Astronomer’s Telegram, 15807, 1
Lei, M., & Wang, J. 2014, PASJ, 66, 7,
doi: 10.1093/pasj/pst008
Le´on-Tavares, J., Chavushyan, V., Pati˜no-´
Alvarez, V., et al.
2013, ApJL, 763, L36, doi: 10.1088/2041-8205/763/2/L36
Liu, H. T., & Bai, J. M. 2006, The Astrophysical Journal,
653, 1089, doi: 10.1086/509097
MAGIC Collaboration, Acciari, V. A., Ansoldi, S., et al.
2018, A&A, 619, A159,
doi: 10.1051/0004-6361/201833618
Maraschi, L., & Tavecchio, F. 2003, ApJ, 593, 667,
doi: 10.1086/342118
Meyer, M., Scargle, J. D., & Blandford, R. D. 2019, ApJ,
877, 39, doi: 10.3847/1538-4357/ab1651
Nalewajko, K., Begelman, M. C., Cerutti, B., Uzdensky,
D. A., & Sikora, M. 2012, MNRAS, 425, 2519,
doi: 10.1111/j.1365-2966.2012.21721.x
Poutanen, J., & Stern, B. 2010, ApJL, 717, L118,
doi: 10.1088/2041-8205/717/2/L118
Rieger, F. 2019, Galaxies, 7, 28,
doi: 10.3390/galaxies7010028
Sbarrato, T., Ghisellini, G., Maraschi, L., & Colpi, M. 2012,
MNRAS, 421, 1764,
doi: 10.1111/j.1365-2966.2012.20442.x
Scargle, J. D., Norris, J. P., Jackson, B., & Chiang, J. 2013,
The Astrophysical Journal, 764, 167,
doi: 10.1088/0004-637X/764/2/167
10
Scully, S. T., Malkan, M. A., & Stecker, F. W. 2014, ApJ,
784, 138, doi: 10.1088/0004-637X/784/2/138
Shukla, A., & Mannheim, K. 2020, Nature
Communications, 11, 4176,
doi: 10.1038/s41467-020-17912-z
Sol, H., Zech, A., Boisson, C., et al. 2013, Astroparticle
Physics, 43, 215, doi: 10.1016/j.astropartphys.2012.12.005
Spada, M., Ghisellini, G., Lazzati, D., & Celotti, A. 2001,
MNRAS, 325, 1559,
doi: 10.1046/j.1365-8711.2001.04557.x
Stecker, F. W., Malkan, M. A., & Scully, S. T. 2006, ApJ,
648, 774, doi: 10.1086/506188
Stern, B. E., & Poutanen, J. 2014, ApJ, 794, 8,
doi: 10.1088/0004-637X/794/1/8
Tanaka, Y. T., Stawarz, L., Thompson, D. J., et al. 2011,
ApJ, 733, 19, doi: 10.1088/0004-637X/733/1/19
Tavecchio, F., & Ghisellini, G. 2012, arXiv e-prints,
arXiv:1209.2291, doi: 10.48550/arXiv.1209.2291
Uttley, P., McHardy, I. M., & Vaughan, S. 2005, MNRAS,
359, 345, doi: 10.1111/j.1365-2966.2005.08886.x
Verrecchia, F., Pittori, C., Piano, G., et al. 2022, The
Astronomer’s Telegram, 15818, 1
Wagner, S. M., Burd, P., Dorner, D., et al. 2021, in
Proceedings of 37th International Cosmic Ray
Conference — PoS(ICRC2021), Vol. 395, 868,
doi: 10.22323/1.395.0868
Wood, M., Caputo, R., Charles, E., et al. 2017, PoS,
ICRC2017, 824, doi: 10.22323/1.301.0824
