Signature of Particle Diffusion on the X-ray Spectra of the blazar Mkn 421

C. Baheeja,1, ∗ S. Sahayanathan,2, 3, † F.M. Rieger,4, 5, ‡ and C.D. Ravikumar1, §

1
Department of Physics, University of Calicut, Malappuram, Kerala, India
2
Astrophysical Sciences Division, Bhabha Atomic Research Centre, Mumbai - 400085, India
3
Homi Bhabha National Institute, Mumbai 400094, India
4
Max Planck Institute for Plasma Physics (IPP), Boltzmannstraße 2, 85748 Garching, Germany
5
Institute for Theoretical Physics, Heidelberg University, Philosophenweg 12, 69120 Heidelberg, Germany
The curvature in blazar spectrum has the potential to understand the particle dynamics in jets.
We performed a detailed analysis of simultaneous Swift-XRT (0.3-10 keV) and NuSTAR (3-79 keV)
arXiv:2404.15171v1 [astro-ph.HE] 23 Apr 2024

observations of Mkn 421. Our analysis of NuSTAR observations alone reveals that, during periods
of low flux, the hard X-ray spectra are best represented by a steep power-law with photon index
reaching ∼ 3. However, the spectrum exhibits significant curvature during its high flux states. To
investigate this, we explore plausible diffusion processes facilitating shock acceleration in the emission
region that can contribute to the observed spectral curvature. Particularly, such processes can cause
gradual fall of the photon spectrum at high energies which can be represented by a sub-exponential
function. The parameter that decides this spectral change can be used to characterise the energy
dependence of the diffusive process. Our results suggest that the X-ray spectra of Mkn 421 are
consistent with a scenario where particle acceleration is mediated through Bohm-type diffusion and
the spectra beyond the synchrotron peak is modulated by the radiative loss process.

I. INTRODUCTION SED modeling of blazars often fails to support such an

interpretation [e.g., 15]. In particular, our earlier study of
Blazars are a class of Active Galactic Nuclei (AGNs) the blazar Mkn 421 reported a strong anti-correlation be-
with a relativistic jet closely aligned to the observer’s tween the spectral indices measured at lower and higher
line of sight [1, 2]. They exhibit rapid flux variations energies around the synchrotron spectral peak, disfavour-
at different time-scales, spanning from minutes to years, ing a simple radiative loss origin [16].
across the entire electromagnetic spectrum [3]. These Narrow-band spectral analysis of blazars frequently re-
highly variable sources are characterized by two promi- veals significant curvature around the peak of the syn-
nent peaks in their spectral energy distribution (SED), chrotron component, that is formally well represented by
extending from radio to γ−ray energies. The low-energy a log-parabola function [17–20]. However, such a func-
SED component (peaking at the IR to X-ray regime) is tion fails to explain the combined optical-UV and X-ray
commonly attributed to synchrotron emission by rela- spectrum [18, 21–23]. Instead, a smooth broken power-
tivistic electrons [2, 4]. On the other hand, the physical law or a power-law particle distribution with an exponen-
process responsible for the emission of the high-energy tial cutoff is often capable of explaining this broad-band
SED component (peaking at the MeV to TeV gamma- spectral component [11, 24]. Recent studies with high-
ray regime) is still under debate [e.g., 5, for a review]. resolution observations have revealed significant spectral
Modeling the broadband synchrotron spectral compo- curvature even beyond the synchrotron peak [22, 25, 26].
nent of blazars often suggests the emitting electron dis- A simple description of first-order Fermi acceleration
tribution to be a broken power-law [6–8]. In principle, at shocks, assuming an energy-independent acceleration
a power-law electron distribution could be produced by and escape time scale, naturally produces a power-law
Fermi-type particle acceleration [e.g., 9, 10], where parti- electron distribution. However, when radiative losses
cles gain energy as they are scattered by magnetic turbu- are taken into account and an energy-dependence is
lence structures embedded in the jet. When electrons are incorporated into the acceleration and/or escape time
scattered by the turbulent structures across a shock front, scales, the resulting particle distribution can deviate
Fermi acceleration will be efficient and is commonly re- from a power-law and exhibit curvature towards high
ferred to as shock acceleration [e.g., 11, 12]. Synchrotron energies [11]. Similarly, a log-parabolic photon spec-
cooling of a power-law electron distribution then results trum suggests the underlying electron distribution to
in a broken power-law distribution with power-law in- be log-parabolic, which could be interpreted in terms
dices differing by one. The corresponding synchrotron of a statistical acceleration scenario with an energy-
photon spectrum would follow a broken power-law with dependent acceleration probability [18]. In fact, un-
indices differing by half [13, 14]. However, broadband der specific conditions, the electron distribution resulting
from a stochastic particle acceleration could mimic a log-
parabola [27]. It is then expected that the synchrotron
∗ baheeja314@gmail.com peak energy anti-correlates with spectral curvature [28].
† sunder@barc.gov.in During the period of 1997−2006, an anti-correlation be-
‡ frank.rieger@ipp.mpg.de tween the synchrotron peak energy and the spectral cur-
§ cdr@uoc.ac.in vature was indeed found through log-parabolic spectral
 2

fitting of the X-ray data from Mkn 421 and other TeV of the source in January 2013, the hard X-ray spectra
BL Lacs, seemingly supportive of a stochastic acceler- were well represented by a steep power-law model with a
ation scenario [21, 27–29]. However, recent studies us- photon index saturating at ∼ 3 [41, 42]. However, the ob-
ing Swift −XRT/NuSTAR observations report no signifi- served X-ray spectrum also shows a significant curvature
cant correlation between these quantities [16, 22, 30–32]. during high-flux states in April 2013 [22]. This curvature
On the other hand, a curved spectrum could also be the persists even in the hard X-rays, which makes it (in spite
outcome of an energy-dependent escape from the accel- of the fact that some blending of components cannot be
eration region. When this energy-dependence is mild, excluded) challenging to attribute this solely to the spec-
the resulting electron distribution closely follows a log- tral transition occurring at the peak of the synchrotron
parabolic shape [23] but deviates significantly otherwise. component [43, 44]. To explore this further, we have
Synchrotron emission by an electron distribution origi- performed a detailed spectral study on the X-ray data of
nating in a model with a strong energy-dependent escape the source. We are particularly interested to understand
time scale has been used to fit the spectra of Mkn 421 whether the observed X-ray spectral characteristics allow
during different flux states [33, 34]. A strong correlation some inferences on the turbulence properties in the jet.
was observed between flux and the energy-dependence of The paper is organized as follows: In Section §II, we dis-
the escape time-scale, and this supports that Bohm type cuss the observation and data reduction procedure, while
diffusion is prominent during high flux states. the X-ray spectral study is described in Section §III. The
Particle acceleration at non- or mildly relativistic shock summary is presented in Section §IV.
fronts has for long been considered as one of the pre-
ferred mechanisms for generating the non-thermal par-
ticle distributions seen in AGN jets [e.g., 11, 12, 35–
38]. The highest energy achieved by the accelerated
particles, as well as the shape of the spectrum around II. OBSERVATION AND DATA ANALYSIS
the maximum energy, are influenced by the balance be-
tween acceleration, escape and the radiative energy loss
rates. In the presence of synchrotron losses, shock ac- Mkn 421 has been observed by NuSTAR and Swift -
celeration (for example) can result in a power-law par- XRT in both flaring as well as quiescent flux states. For
ticle distribution with a modified exponential cutoff, the current study we have selected all the available simul-
∝ exp[−(γ/γc)βe ], where βe is dependent on the under- taneous Swift -XRT and NuSTAR observations till 2018
lying turbulence/diffusion properties [e.g., 39]. Formally, (details are given in Table I). This allows us to analyse
βe = (1 + a) is related to the momentum index a of the the source over a wide range of X-ray energies, from 0.3
spatial diffusion coefficient, κ = (1/3)λ c ∝ γ a , that fa- to 79 keV. The strategies for analysing these observations
cilitates the particle transport. Here, λ is the particle are detailed below.
mean free path and γ is the particle Lorentz factor. In
particular, one may have βe = 1 (a = 0) in the case of
“idealized” hard-sphere scattering (energy-independent
diffusion), βe = 4/3 (a = 1/3) for Kolmogorov-type
turbulence, and βe = 2 (a = 1) for Bohm type diffu- A. NuSTAR
sion (where λ ∼ rg , with rg as the gyro-radius). Since
the corresponding particle acceleration timescale, tacc is NuSTAR [45] is a space-based hard X-ray telescope
proportional to λ, Bohm diffusion typically yields the which operates from 3 to 79 keV energy band with
fastest acceleration rate (i.e., the highest γe,max when an angular resolution of subarcmin. All observations
balanced with synchrotron losses). The resultant syn- are carried out with two co-aligned, independent tele-
chrotron spectra can exhibit some extended curvature scopes called Focal Plane Module A (FPMA) and B
at high energies. In particular, an electron distribution (FPMB). The NuSTAR observations were taken from the
with exponential cutoff index βe will result in a syn- HEASARC interface by NASA and the data were pro-
chrotron spectrum which can  be significantly smoother cessed with NuSTARDAS package (Version 1.4.1) avail-
(sub-exponential) jν ∝ exp −(ν/νc )ζ with ζ ≡ βeβ+2
 e
, able within HEASOFT (Version 6.19). The source spec-
e.g., ζ = 1/2 in the case of Bohm-type diffusion [40]. trum is extracted from a circular region with a radius
The BL Lac object Mkn 421, that we focus on here, of 50 arcsec centered on the source, while the back-
is the nearest (z = 0.031) and one of the well-studied ground is estimated from a circular region with a ra-
TeV blazars. Mkn 421 belongs to the high-frequency BL dius of 70 arcsec that is free of source contamination
Lac (HBL) class, as its synchrotron spectral component but near it. NUPRODUCT (Version 0.2.8) was used
peaks in the X-ray regime. The X-ray spectrum around to obtain source and background spectra after running
the synchrotron peak exhibits significant curvature that NUPIPELINE (Version 0.4.9) on each observation. The
has been interpreted in terms of a log-parabola function. FPMA and FPMB source spectra were then individually
X-ray spectral analysis of Mkn 421 using NuSTAR (3-79 grouped to 30 photons per bin using the tool GRPPHA
keV) observations, reveals that during the low-flux state to ensure improved χ2 statistics.
 3

TABLE I. Details of simultaneous Swift-XRT and NuSTAR observations.

Swift Date & time Exposure NuSTAR Date & time Exposure
Obs.ID (s) Obs.ID (s)
35014034 2013-01-15 02:09:59 3958.859 60002023006 2013-01-15 00:56:07 24181
80050003 2013-02-06 01:20:59 9506.827 60002023010 2013-02-06 00:16:07 19302
80050006 2013-02-17 00:03:59 9201.642 60002023014 2013-02-16 23:36:07 17356
80050007 2013-03-04 23:34:25 984.609 60002023016 2013-03-04 23:06:07 17251
80050011 2013-03-11 23:58:59 8425.937 60002023018 2013-03-11 23:01:07 17472
80050013 2013-03-17 01:22:59 8880.74 60002023020 2013-03-17 00:11:07 16554
80050014 2013-04-02 21:01:59 1644.569 60002023022 2013-04-02 17:16:07 24767
80050016 2013-04-11 00:30:59 1118.631 60002023024 2013-04-10 21:26:07 5757
80050019 2013-04-12 21:53:58 9546.279 60002023027 2013-04-12 20:36:07 7629
32792002 2013-04-14 00:38:59 6327.071 60002023029 2013-04-13 21:36:07 16508
35014062 2013-04-15 23:07:59 534.621 60002023033 2013-04-15 22:01:07 17276
35014065 2013-04-17 00:46:59 8842.132 60002023035 2013-04-16 22:21:07 20278
35014066 2013-04-18 00:49:59 6887.219 60002023037 2013-04-18 00:16:07 17795
35014067 2013-04-19 00:52:59 6132.768 60002023039 2013-04-19 00:31:07 15958
34228110 2017-01-04 00:06:57 6021.027 60202048002 2017-01-03 23:51:09 23691
81926001 2017-01-31 23:27:57 1009.619 60202048004 2017-01-31 23:46:09 21564
34228145 2017-02-28 22:46:56 44.62 60202048006 2017-02-28 22:11:09 23906

B. Swift-XRT power-law (PL), log-parabola (LP), and a power-law with

an exponential cutoff (CPL). These models are defined as
The XRT is a focusing X-ray telescope operating in
F (ǫ) ∝ ǫ−Γ (PL) , (1)
the 0.3-10 keV energy range with an angular resolution
< 20 arcsec [46]. The Swift -XRT observations were where ǫ is the photon energy, and Γ represents the power-
also retrieved from HEASARC interface and the data law index
processed using the XRTDAS software package (Version
3.0.0) available within HEASOFT. We used the observa-  −α−β log(ǫ/ǫ0 )
ǫ
tions performed in Windowed Timing (WT) mode and F (ǫ) ∝ (LP) , (2)
ǫ0
the events with 0-2 grades have been considered in the
analysis. The event files were cleaned and calibrated us- where α is the spectral slope at energy ǫ0 , and β is the
ing standard procedures with the XRTPIPELINE (Ver- spectral curvature, and
sion v0.13.4) task. A circular region of 30 pixel radius
centred at the source was used to extract the source spec- F (ǫ) ∝ ǫ−p exp[−(ǫ/ǫc )] (CPL) , (3)
trum, and a circular region of same size devoid of source
contamination was used to extract the background spec- where p represents the power-law index, and ǫc character-
trum. An annular region with inner and outer radii of 2 izes the position of the cutoff energy. The spectral peak
and 30 pixels, respectively, was used as source and back- (ǫ2 F (ǫ) representation) of the log-parabola function is
ground regions for the observation with pileup (Obs.ID obtained from
80050019). XRTPRODUCTS (Version v0.4.2) was used 2−α

to generate the final spectrum. The XRTMKARF task ǫp = ǫ0 10 2β . (4)

was employed to generate the auxiliary response files
(ARFs), and the response matrice files (RMFs) from the The neutral hydrogen column density, inclusive of
Swift CALDB were used. The source spectra were then both HI and HII was fixed at NH = 2.03 × 1020 cm−2
grouped using the GRPPHA tool to ensure a minimum throughout the analysis. The best-fit parameters of
of 20 counts/bin. these models are presented in Table II. We observe that
most of the low-flux states are well-fitted with a steep
power-law model, yielding a photon index saturating at
III. X–RAY SPECTRAL ANALYSIS AND ∼ 3, as reported earlier [42]. In contrast, the high-flux
RESULTS
states exhibit significant curvature and deviate from a
simple power-law model (see Figure 1). The spectral
A. NuSTAR (3-79 keV) regime fittings using the PL, LP, and CPL models for a sample
of low and high flux states are shown in Figure 1. Figure
To investigate the curvature in the hard X-ray regime, 2 illustrates the change in the reduced chi-square value
we fitted the NuSTAR observations (3-79 keV) of the of the PL fit (left) and the spectral curvature (right) as
source, using three models available in XSPEC namely, the flux increases.
 4

keV2 (Photons cm−2 s−1 keV−1)

keV2 (Photons cm−2 s−1 keV−1)

keV2 (Photons cm−2 s−1 keV−1)

0.1 0.1 0.1

0.01 0.01 0.01

(data−model)/error

(data−model)/error

(data−model)/error
2 2 2

0 0 0

−2 −2 −2
5 10 20 50 5 10 20 50 5 10 20 50
Energy (keV) Energy (keV) Energy (keV)
2 2 2
keV2 (Photons cm−2 s−1 keV−1)

keV2 (Photons cm−2 s−1 keV−1)

keV2 (Photons cm−2 s−1 keV−1)

1 1 1

0.5 0.5 0.5

0.2 0.2 0.2

0.1 0.1 0.1

(data−model)/error

(data−model)/error

(data−model)/error
2 2
0
0 0

−2 −2
−5

5 10 20 50 5 10 20 50 5 10 20 50
Energy (keV) Energy (keV) Energy (keV)

FIG. 1. Spectral fits (NuSTAR alone) using the models PL, CPL and LP (left to right) for the ObsIDs 60002023018 (low-flux
state) and 60002023027 (high-flux state) are shown in the upper and lower panels, respectively. Significant curvature is apparent
in the high flux state.

-8.6 -8.6
-8.8 -8.8
log10Flux (3-79 keV)

-9 -9
log10Flux (3-79 keV)

-9.2 -9.2
-9.4 -9.4
-9.6
-9.6
-9.8
-9.8
-10
-10
-10.2
-10.2
1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6
2
χ red(PL)
0 0.08 0.16 0.24 0.32 0.4 0.48
β

FIG. 2. Scatter plots (NuSTAR alone) showing flux, along with reduced chi-square values for the PL fit (left), and LP curvature
values β (right).

3 3 1.2

2.7 2.7 1.15

2.4 2.4
1.1
red(CPL)

red(CPL)
χ2red(LP)

2.1 2.1
1.05
1.8 1.8
2

2
χ

1
1.5 1.5

1.2 1.2 0.95

0.9 0.9 0.9

0.9 1.2 1.5 1.8 2.1 2.4 2.7 0.9 1.2 1.5 1.8 2.1 2.4 2.7 0.9 0.95 1 1.05 1.1 1.15
χ2red(PL) χ2red(PL) χ2red(LP)

FIG. 3. Scatter plots between the reduced chi-square values of NuSTAR data fitted with the PL, LP, and CPL models, along
with the identity line.
 5

0.5 -8.6
-8.8
0.4
-9

log10Flux3-79 keV
-9.2
0.3
-9.4
β

0.2 -9.6
-9.8
0.1 -10
-10.2
0
2.2 2.4 2.6 2.8 3 3.2 2.2 2.4 2.6 2.8 3 3.2
α α

FIG. 4. Scatter plots (NuSTAR alone) showing LP index α, along with spectral curvature β (left), and flux in the 3-79 keV
range (right).

90 -8.6 -8.6

80 -8.8 -8.8

log10Flux3-79 keV
70 -9 -9
log10Flux3-79 keV

-9.2
εc (keV)

60 -9.2
-9.4
50 -9.4
-9.6
40 -9.6
-9.8
30 -9.8
-10
20 -10 20 30 40 50 60 70 80 90
2 2.2 2.4 2.6 2.8 3 3.2 2 2.2 2.4 2.6 2.8 3 3.2 εc (keV)
p p

FIG. 5. Scatter plots (NuSTAR alone) showing best-fit CPL model parameters: index and cutoff energy (in keV), index and
flux, and cutoff energy and flux (from left to right).

2 2

0.4 1.6 1.6

1.2 1.2
εp (keV)

εp (keV)
β

0.3 0.8 0.8

0.4 0.4

0.2 0 0
2.2 2.4 2.6 2.8 3 2.2 2.4 2.6 2.8 3 0.2 0.3 0.4
α α β
-8.4 -8.4 2

-8.7 -8.7 1.6

log10Flux0.3-79 keV

log10Flux0.3-79 keV

-9 -9
1.2
εp (keV)

-9.3 -9.3
0.8
-9.6 -9.6
0.4
-9.9 -9.9
0
-10.2 -10.2 -10.2 -9.9 -9.6 -9.3 -9 -8.7 -8.4
2.2 2.4 2.6 2.8 3 0.2 0.3 0.4 log10Flux0.3-79 keV
α β

FIG. 6. Scatter plots between the parameters obtained from LP fitting of combined Swift-XRT and NuSTAR data. The upper
panel represents plots between index (α) and curvature (β), α and peak energy (ǫp ), and β and ǫp (from left to right). The
lower panel shows the variation of flux in 0.3 to 79 keV with α, β and ǫp , respectively (from left to right).
 6

TABLE II. Fit parameters of NuSTAR (3-79 keV) spectra as modeled with PL, LP and CPL.
NuSTAR Flux PL LP(ǫ0 =5 keV) CPL
Obs.ID (3-79 keV) Γ χ2red α β χ2red p ǫc (keV) χ2red
60002023006 -9.954±0.004 3.03±0.01 1.05 (563) 2.95±0.02 0.31±0.06 0.92 (562) 2.82±0.05 35.12+9.75
−6.47 0.94 (562)
60002023010 -9.842±0.004 2.95±0.01 1.32 (570) 2.83±0.02 0.41±0.06 1.04 (569) 2.64±0.05 24.99+4.49
−3.41 1.06 (569)
60002023014 -10.187±0.007 3.02±0.02 1.06 (419) 3±0.03 0.11±0.09 1.05 (418) − − −
+16.44
60002023016 -9.779±0.004 3.01±0.01 1.11 (557) 2.95±0.02 0.25±0.06 1.01 (556) 2.85±0.05 44.97−9.8 1.03 (556)
+31.18
60002023018 -9.906±0.005 3.09±0.02 1.04 (509) 3.04±0.02 0.2±0.07 0.99 (508) 2.96±0.05 53.64−14.91 1 (508)
+10.57
60002023020 -9.731±0.005 2.77±0.01 1.18 (595) 2.68±0.02 0.28±0.05 1.04 (594) 2.58±0.04 40.77−7.19 1.06 (594)
60002023022 -9.343±0.002 2.74±0.01 1.51 (898) 2.64±0.01 0.3±0.03 1.03 (897) 2.52±0.02 38.28+3.89
−3.29 1.06 (897)
60002023024 -9.201±0.003 2.9±0.01 1.47 (620) 2.78±0.02 0.39±0.05 1.11 (619) 2.61±0.04 27.88+4.2
−3.33 1.14 (619)
60002023027 -8.624±0.002 2.62±0.01 2.63 (1023) 2.45±0.01 0.44±0.02 0.94 (1022) 2.29±0.02 26.4+1.36
−1.26 1.01 (1022)
60002023029 -9.085±0.002 2.79±0.01 2.06 (917) 2.65±0.01 0.42±0.02 0.98 (916) 2.48±0.02 26.67+1.83
−1.64 1.08 (916)
60002023033 -9.014±0.002 2.59±0.01 1.86 (1019) 2.46±0.01 0.33±0.02 1.01 (1018) 2.34±0.02 35.89+2.5
−2.23 1.02 (1018)
60002023035 -8.934±0.002 2.39±0.01 2.33 (1182) 2.25±0.01 0.35±0.02 1.07 (1181) 2.13±0.01 35.77+1.9
−1.74 1.1 (1181)
60002023037 -9.835±0.004 2.85±0.01 1.31 (568) 2.72±0.02 0.4±0.06 1.05 (567) 2.55±0.05 26.61+4.85
−3.67 1.05 (567)
+23.58
60002023039 -9.868±0.005 2.94±0.02 0.98 (519) 2.88±0.02 0.22±0.06 0.91 (518) 2.79±0.05 50.09−12.58 0.92 (518)
60202048002 -9.269±0.003 2.45±0.01 1.39 (1006) 2.36±0.01 0.22±0.02 1.1 (1005) 2.29±0.02 60.22+7.58
−6.17 1.14 (1005)
60202048004 -9.247±0.003 2.45±0.01 1.72 (1001) 2.31±0.01 0.33±0.02 1.04 (1000) 2.2±0.02 37.22+2.94
−2.59 1.06 (1000)
60202048006 -9.261±0.002 2.49±0.01 1.75 (996) 2.37±0.01 0.31±0.02 1.13 (995) 2.26±0.02 39.53+3.31
−2.89 1.16 (995)

TABLE III. The best-fit parameters of combined Swift-XRT and NuSTAR spectral fitting with a LP and a simple CPL model,
respectively.
Obs.ID Flux LP (ǫ0 =5 keV) CPL
Swift NuSTAR (0.3-79 keV) α β ǫp (keV) χ2red (dof) p ǫc (keV) χ2red (dof)
35014034 60002023006 -9.696±0.003 2.93±0.01 0.34±0.01 0.22±0.02 0.97 (962) 2.38±0.01 11.5±0.42 1.17 (962)
80050003 60002023010 -9.602±0.003 2.81±0.01 0.36±0.01 0.37±0.02 1.27 (1110) 2.23±0.01 10.37±0.3 1.4 (1110)
80050006 60002023014 -9.93±0.004 2.94±0.02 0.25±0.01 0.07±0.01 1.19 (848) 2.51±0.01 14.4±0.87 1.3 (848)
80050007 60002023016 -9.523±0.003 2.92±0.01 0.32±0.02 0.19±0.03 1.1 (822) 2.46±0.02 13.71±0.72 1.34 (822)
80050011 60002023018 -9.639±0.003 2.98±0.01 0.34±0.01 0.17±0.01 1.1 (1001) 2.42±0.01 10.73±0.38 1.35 (1001)
80050013 60002023020 -9.521±0.003 2.67±0.01 0.24±0.01 0.19±0.02 1.16 (1126) 2.29±0.01 16.97±0.69 1.2 (1126)
80050014 60002023022 -9.146±0.002 2.64±0.01 0.31±0.01 0.47±0.03 1.06 (1338) 2.23±0.01 16.45±0.46 1.62 (1338)
80050016 60002023024 -8.972±0.002 2.78±0.01 0.36±0.01 0.41±0.02 1.18 (1039) 2.23±0.01 11.93±0.38 1.41 (1039)
80050019 60002023027 -8.46±0.001 2.45±0.01 0.43±0.01 1.5±0.03 1.07 (1575) 1.91±1.91 12.16±12.16 2.11 (1575)
32792002 60002023029 -8.874±0.001 2.68±0.01 0.34±0.01 0.51±0.02 1.1 (1411) 2.21±0.01 14.15±0.31 1.52 (1411)
35014062 60002023033 -8.847±0.002 2.47±0.01 0.29±0.01 0.79±0.05 1.07 (1391) 2.19±0.01 22.72±0.83 1.33 (1391)
35014065 60002023035 -8.799±0.001 2.27±0.01 0.28±0.01 1.61±0.05 1.17 (1741) 1.96±0.01 21.77±0.52 1.45 (1741)
35014066 60002023037 -9.6±0.003 2.76±0.01 0.22±0.01 0.1±0.01 1.19 (1068) 2.39±0.01 17.67±0.77 1.14 (1068)
35014067 60002023039 -9.625±0.003 2.85±0.01 0.26±0.01 0.11±0.01 1.05 (1001) 2.43±0.01 14.71±0.65 1.21 (1001)
34228110 60202048002 -9.133±0.002 2.34±0.01 0.26±0.01 1.14±0.08 1.09 (1348) 2.12±0.01 28.88±1.44 1.5 (1348)
81926001 60202048004 -9.105±0.002 2.33±0.01 0.29±0.01 1.33±0.08 1.07 (1385) 2.05±0.01 23.45±0.92 1.29 (1385)
34228145 60202048006 -9.116±0.003 2.37±0.01 0.31±0.02 1.29±0.15 1.11 (1044) 2.25±0.02 37.69±2.64 1.18 (1044)

Our analysis provides strong evidence for spectral parameters α and β yields a correlation coefficient, rs =
curvature in the NuSTAR regime, with a LP/CPL -0.41 with a null hypothesis probability, p = 0.104. This
model clearly preferred over a pure PL model in high result is consistent with previous studies which reported
flux states. The plots between reduced chi-square that no significant correlation was observed [22, 33].
values for the spectral fittings with the PL, LP, and However, an anti-correlation is witnessed between α
CPL models (χ2red (PL), χ2red (LP), and χ2red (CPL)) are and the flux (rs = -0.81, p <0.001), indicating that the
shown in Figure 3. Scatter plots showing the best-fit spectra get harder during brighter states of the source.
LP and CPL model parameters, and with the flux, are This harder when brighter behaviour of the source has
depicted in Figures 4 and 5, respectively. To identify already been reported earlier [18, 22, 29, 31, 32]. The
the dependence between these best-fit parameters, we spectral fittings with the simple CPL model allows to
performed a Spearman rank correlation analysis between constrain well the cutoff energy during high-flux states.
various quantities obtained from the spectral fit. The Here no correlation is observed between p and ǫc (rs
Spearman rank correlation study between the LP model = 0.16, p = 0.556), while a significant anti-correlation
 7

between p and flux (rs = -0.79 p <0.001) is seen. does not support a simple cooling break origin of the
peak frequency, rather might be affected by the blending
of different components [e.g., 16]. At hard X-rays, where
the synchrotron spectrum declines, we may expect the
B. Combined NuSTAR and Swift-XRT (0.3-79 emission to be more dominated by a single component,
keV) regime particularly during higher flux states. Additionally, the
high-energy end of the spectrum is likely to be populated
In order to gain further insights, we have also studied by a cooled electron distribution. To explore the spectral
the broad X-ray spectra of Mkn 421 ranging from 0.3 curvature towards high energies in more detail, we thus
to 79 keV using simultaneous Swift -XRT and NuSTAR perform spectral fits to the X-ray data above the syn-
observations, employing the log-parabola (LP) and sim- chrotron SED peak (with ǫp > 0.3 keV) using a modified
ple exponential cutoff (CPL) model, respectively. These CPL (MCPL) model
X-ray spectra exhibits significant curvature, and a LP
model generally provides a better fit when compared to
F (ǫ) ∝ ǫ−p exp[−(ǫ/ǫc )ζ ] (MCPL) , (5)
a simple CPL model. The best-fit parameters are pre-
sented in Table III, and the scatter plots between LP where p represents the power-law index, ǫc character-
parameters and the flux are shown in Figure 6. Again, izes the position of the cutoff energy, and the parameter
we observe no correlation between the α and β (rs = ζ governs the steepness of the cutoff. This function is
0.17, p = 0.521). Additionally, there was no significant added as a local model in XSPEC, and we perform spec-
correlation between β and peak energy, ǫp , (rs = 0.31, tral fitting for the combined simultaneous Swift -XRT and
p = 0.227), whereas α showed a strong negative corre- NuSTAR observations from ǫp to 79 keV.
lation with ǫp (rs = -0.86, p <0.001). Furthermore, we Among the total 17 simultaneous Swift -XRT and
noted a strong negative correlation between α and flux NuSTAR observations we found only for 10 epochs,
(rs = -0.80, p <0.001), and ǫp being significantly corre- the peak falls in between 0.3–79 keV. Most of these
lated with flux (rs = 0.85, p <0.001). These correlations epochs are during high-flux states (except obsID.
suggest that during flares, the spectral index hardens and 80050003+60002023010), and all these epochs are con-
the spectral peak moves towards higher energies. sidered for the MCPL fit. The considered X-ray spectra
The absence of a significant correlation among the LP did not allow us to constrain all parameters of the model.
parameters, even in the broad energy range studied here, Hence, we performed a fitting with p fixed to a value 2,
indicates that the changes in spectral characteristics can- representing a cooled particle distribution. This choice
not simply be ascribed to the energy-dependence of the may be appropriate since we are interested in the spec-
particle acceleration process as proposed in ref. [18]. Ad- trum above ǫp where synchrotron losses dominate. The
ditionally, such a model is unable to account for the sample spectral fits of MCPL model are shown in Figure
broadband SED of blazars [18, 21, 22]. Hence, an al- 7. The modified CPL model represents well the spec-
ternate physically motivated choice could be a CPL type trum above ǫp and the best-fit parameters ǫc and ζ are
model. On the other hand, the foregoing analysis indi- shown in Table IV. The scatter plots between the fitting
cates that a simple (purely exponential) CPL model does parameters, and with the flux are shown in Figure 8. We
not provide a better fit to the broad X-ray spectra of the performed a Spearman correlation analysis and did not
source, particularly in high flux states. To explore this find a significant correlation between the MCPL model
further, we next study the broadband X-ray spectrum parameters ǫc and ζ (rs = 0.45, p = 0.192). Also, no
using a power-law with a modified exponential cutoff, as significant correlations are observed between ǫc and flux
might be expected to occur in shock-type acceleration (rs = 0.42, p = 0.229), and ζ and flux (rs = 0.55, p =
scenarios (see Section §I). 0.102).
Constraining the ζ-parameter in the X-ray spectrum
can provide insights into the parent particle distribution.
C. Probing a power-law with modified exponential In the case of synchrotron emission, the parameter ζ is
cutoff linked to the primary particle distribution through the
relation ζ = βeβ+2 e
[40]. Therefore, the value of ζ is ex-
In the context of shock acceleration scenarios, the elec- pected to be 0.33 in the case of energy-independent diffu-
tron distribution exhibits a simple exponential cutoff sion (βe = 1), while Bohm-type diffusion (βe = 2) results
form only when diffusion is independent of energy. The in ζ=0.5. As can be seen from Table IV, the inferred
shape of the particle distribution deviates from this as ζ-values favour a Bohm-type behaviour.
the diffusion coefficient becomes energy-dependent, lead- For comparison, we also repeated this analysis by fixing
ing to a corresponding change in the synchrotron cutoff ζ at the values corresponding to hard-sphere and Bohm-
(i.e., typically sub-exponential) shape [39, 40]. In HBL type diffusion, which are 0.33 and 0.5, respectively (Table
sources such as Mkn 421, the maximum achievable elec- V). The best-fit parameters revealed that ζ = 0.5 leads
tron energies are limited by synchrotron losses. The spec- to an index p closer to ∼ 2, supporting our previous as-
tral index evolution around the synchrotron (SED) peak sumption of a cooled distribution above ǫp . Therefore,
 8

TABLE IV. Best fit parameters using the MCPL model (cooled p=2) for the energy range ǫp -79 keV.
Obs.ID ǫp ǫc ζ χ2red (dof) Flux
Swift NuSTAR (keV) (keV) (ǫp -79 keV)
80050003 60002023010 0.37±0.02 2.64±0.16 0.54±0.01 1.2 (1106) -9.621±0.003
80050014 60002023022 0.47±0.03 2.95±0.32 0.47±0.02 1.17 (1323) -9.156±0.002
80050016 60002023024 0.41±0.02 2.63±0.27 0.52±0.02 1.18 (1030) -8.988±0.003
80050019 60002023027 1.5±0.03 8.01±0.53 0.6±0.02 1.03 (1458) -8.457±0.002
32792002 60002023029 0.49±0.02 3.54±0.28 0.52±0.02 1.1 (1394) -8.888±0.002
35014062 60002023033 0.79±0.05 6.9±0.66 0.53±0.02 1.09 (1345) -8.853±0.002
35014065 60002023035 1.61±0.05 20.58±0.66 0.71±0.03 1.05 (1613) -8.807±0.002
34228110 60202048002 1.14±0.08 7.9±1.54 0.44±0.04 1.11 (1266) -9.128±0.003
81926001 60202048004 1.33±0.08 14.96±1.01 0.62±0.04 1.06 (1285) -9.109±0.003
34228145 60202048006 1.29±0.15 8.88±1.18 0.5±0.03 1.12 (1015) -9.112±0.003

TABLE V. Best fit parameters using the MCPL model by assuming Bohm(ζ=0.5) and Hard-sphere(ζ=0.33) for ǫp -79 keV fit.
Obs.ID ǫp Bohm: ζ=0.5 Hard-sphere: ζ=0.33
Swift NuSTAR (keV) p ǫc χ2red (dof) p ǫc χ2red (dof)
80050003 60002023010 0.37±0.02 1.95±0.02 1.84±0.1 1.2 (1106) 1.63±0.02 0.11±0.01 1.19 (1106)
80050014 60002023022 0.47±0.03 2±0.02 3.6±0.22 1.18 (1323) 1.72±0.03 0.25±0.02 1.12 (1323)
80050016 60002023024 0.41±0.02 1.97±0.02 2.2±0.14 1.17 (1030) 1.65±0.03 0.13±0.01 1.16 (1030)
80050019 60002023027 1.5±0.03 1.83±0.03 3.33±0.24 1.01 (1457) 1.43±0.04 0.16±0.02 0.98 (1457)
32792002 60002023029 0.49±0.02 1.95±0.02 2.75±0.13 1.1 (1394) 1.64±0.03 0.17±0.01 1.06 (1394)
35014062 60002023033 0.79±0.05 1.93±0.03 4.83±0.41 1.08 (1345) 1.6±0.04 0.3±0.04 1.05 (1345)
35014065 60002023035 1.61±0.05 1.8±0.02 6.27±0.51 1.03 (1613) 1.49±0.04 0.41±0.05 1.03 (1613)
34228110 60202048002 1.14±0.08 2.04±0.03 13.05±1.97 1.11 (1266) 1.82±0.05 1.19±0.25 1.1 (1266)
81926001 60202048004 1.33±0.08 1.86±0.03 6.24±0.7 1.05 (1284) 1.55±0.05 0.4±0.06 1.04 (1284)
34228145 60202048006 1.29±0.15 1.99±0.04 8.32±1.1 1.12 (1015) 1.71±0.06 0.6±0.11 1.12 (1015)

the results appear consistent with a cooled particle dis- to the primary electron distribution. The acceleration of
tribution with a cutoff shaped by Bohm-type diffusion electrons at shocks is a favored mechanism for generating
(βe = 2). non-thermal particle distributions in astrophysical jets.
In the presence of radiative losses like the synchrotron
process, the accelerated electron distribution will be a
broken power-law with a modified exponential cutoff at
IV. SUMMARY
the maximum available electron energy. The resultant
synchrotron spectrum from such a particle distribution
We have conducted a detailed study of the X-ray spec- will always be a power-law with a sub-exponential cut-
tra of Mkn 421 using simultaneous Swift -XRT and NuS- off, and we found that a MCPL function can satisfacto-
TAR observations. Most of our observations considered rily reproduce the data beyond the SED peak. Further,
are during high flux states (unlike ref. [42] for example) the results are consistent with a scenario where the hard
and enables us to investigate the spectral curvature in X-ray spectrum is due to a cooled electron distribution,
the hard X-ray regime more rigorously. Our spectral with the highest energy part shaped by Bohm-type dif-
study of NuSTAR observations using power-law (PL), fusion. For a strong shock that is non-relativistic in the
log-parabolic (LP), and simple exponential cutoff power- jet frame (Γs = Γj Γb (1 − βb βj ) ∼ 1, with Γj ≫ 1 being
law (CPL) models suggests that LP and CPL are clearly the jet Lorentz factor and Γb the ‘blob’ Lorentz factor),
preferred over a simple PL. This provides strong evidence the characteristic acceleration timescale is approximately
of spectral curvature in the 3-79 keV energy regime. We given by t′acc ≃ 10κ′ /u2s where κ′ = (1/3) λ′ c and λ′ ∼
also examined the broad (Swift -XRT and NuSTAR) X- rg′ = γe′ mc c2 /(eB ′ ) in the Bohm limit [e.g., 12]. Balanc-
ray spectra, spanning from 0.3 to 79 keV with LP and ing acceleration with cooling, t′syn = 9m3e c5 /(4e4 γe′ B ′2 ),
CPL models, indicating that LP provides a better fit one can estimate maximum achievable electron energies
compared to CPL. However, the lack of a significant cor- ′
(γe,m ). The corresponding synchrotron photon energy
relation between the LP parameters suggests that the ′
ǫc ∝ γe,m′2
B ′ can be compared to the synchrotron cut-
variations in spectral characteristics cannot be attributed off energies ǫc ∼ 10 keV inferred from observations (Ta-
to the energy-dependence of the particle acceleration pro- ble IV) taking beaming (ǫc ∼ Γj ǫ′c ) into account. The
cess. result then substantiates the initial assumption of non-
The curvature in the X-ray spectrum is closely linked
 9

00080050019+60002023027 00034228110+60202048002

0.5

keV2 (Photons cm−2 s−1 keV−1)

keV2 (Photons cm−2 s−1 keV−1)

2

1
0.2
0.5

0.1
0.2

0.1 0.05
(data−model)/error

(data−model)/error
2 2
0
0
−2
−2
−4
2 5 10 20 50 2 5 10 20 50
Energy (keV) Energy (keV)

FIG. 7. Spectral fit from ǫp -79 keV using the MCPL model for the obsID 00080050019+60002023027 (left) and
00034228110+60202048002 (right).

1.5 -8.4

1.2 -8.7

log10Flux(εp-79 keV)
log10εc (keV)

0.9 -9

0.6 -9.3

0.3 -9.6

0.3 0.4 0.5 0.6 0.7 0.8 0.3 0.4 0.5 0.6 0.7 0.8
ζ ζ
-8.4

-8.7
log10Flux(εp-79 keV)

-9

-9.3

-9.6

0.3 0.6 0.9 1.2 1.5

log10εc (keV)

FIG. 8. Scatter plots between best-fit parameters for the MCPL model (ǫp -79 keV). The upper left panel is for ζ and cutoff
energy (ǫc ), right is for ζ and flux in ǫp -79 keV, and the lower panel is for ǫc and flux in ǫp -79 keV. The dotted vertical lines
represent ζ corresponding to hard-sphere (0.33) and Bohm (0.5).

relativistic shock acceleration. sition of multiple broken power-law components might

also contribute to spectral curvature. Since the highest-
Though the observed spectral curvature in the hard X- energy electrons are expected to probe the shock vicin-
ray supports Bohm-type diffusion during flares in the jet, ity, X-ray spectral analysis along with dedicated polar-
other possibilities are not yet be ruled out. For instance, ization studies [e.g., 37] provides a powerful diagnostics
an energy-dependent escape time scale or the superpo-
 10

of the underlying flow properties. The signature of an ACKNOWLEDGMENTS

electron distribution shaped by Bohm-type diffusion pro-
cess could in principle be further probed by modeling This research has made use of data obtained from
the resultant γ-ray emission by inverse Compton scatter- NASA’s High Energy Astrophysics Science Archive Re-
ing [47], though in practice, this may be challenging to search Center(HEASARC), a service of the Goddard
achieve with current instrumentation [48]. The type of Space Flight Center and the Smithsonian Astrophysi-
diffusion process can also have an impact on the temporal cal Observatory. CB wishes to thank CSIR, New Delhi
behaviour of the source. Hence, studying the light curve (09/043(0198)/2018-EMR-I) for financial support. CB
considering acceleration initiated by different diffusion is thankful to UGC-SAP and FIST 2 (SR/FIST/PS1-
processes will offer additional insights into the underly- 159/2010) (DST, Government of India) for the research
ing characteristics. facilities in the Department of Physics, University of Cali-
cut. FMR acknowledges support by a DFG grant under
RI 1187/8-1.

[1] R. Antonucci, Annual Review of Astronomy and Astrophysics 31,Astronomy

473 (1993),and Astrophysics 413, 489 (2004),
https://doi.org/10.1146/annurev.aa.31.090193.002353. arXiv:astro-ph/0312260 [astro-ph].
[2] C. M. Urry and P. Padovani, [19] A. Tramacere, P. Giommi, E. Massaro, M. Perri,
R.
Publications of the Astronomical Society of the Pacific 107, 803 (1995). Nesci, S. Colafrancesco, G. Tagliaferri,
[3] M.-H. Ulrich, L. Maraschi, and C. M. Urry, G. Chincarini, A. Falcone, D. N. Burrows,
Annual Review of Astronomy and Astrophysics 35, 445 (1997), P. Roming, M. McMath Chester, and N. Gehrels,
https://doi.org/10.1146/annurev.astro.35.1.445. Astronomy and Astrophysics 467, 501 (2007),
[4] G. Ghisellini, I. M. George, and C. Done, arXiv:astro-ph/0611276 [astro-ph].
Monthly Notices of the Royal Astronomical Society 241, 43P[20](1989),
L. Chen, The Astrophysical Journal 788, 179 (2014).
https://academic.oup.com/mnras/article-pdf/241/1/43P/3188071/mnras241-043P.pdf.
[21] A. Tramacere, P. Giommi, M. Perri, F. Verrecchia, and
[5] H. Sol and A. Zech, Galaxies 10, 105 (2022), G. Tosti, Astronomy and Astrophysics 501, 879 (2009),
arXiv:2211.03580 [astro-ph.HE]. arXiv:0901.4124 [astro-ph.HE].
[6] R. M. Sambruna, P. Barr, P. Giommi, [22] A. Sinha, A. Shukla, R. Misra, V. R.
L. Maraschi, G. Tagliaferri, and A. Treves, Chitnis, A. R. Rao, and B. S. Acharya,
Astrophysical Journal, Supplement 95, 371 (1994). Astronomy and Astrophysics 580, A100 (2015),
[7] F. Tavecchio, L. Maraschi, and G. Ghisellini, arXiv:1506.03629 [astro-ph.GA].
The Astrophysical Journal 509, 608 (1998). [23] S. K. Jagan, S. Sahayanathan, R. Misra,
[8] G. Fossati, A. Celotti, M. Chiaberge, Y. H. C. D. Ravikumar, and K. Jeena,
Zhang, L. Chiappetti, G. Ghisellini, L. Maraschi, Mon. Not. Roy. Astr. Soc. 478, L105 (2018),
F. Tavecchio, E. Pian, and A. Treves, arXiv:1805.03684 [astro-ph.HE].
The Astrophysical Journal 541, 166 (2000). [24] A. Sinha, S. Sahayanathan, B. S. Acharya, G. C.
[9] E. Fermi, Physical Review 75, 1169 (1949). Anupama, V. R. Chitnis, and B. B. Singh,
[10] M. Lemoine, Phys. Rev. D 99, 083006 (2019), The Astrophysical Journal 836, 83 (2017).
arXiv:1903.05917 [astro-ph.HE]. [25] J. Bhagwan, A. C. Gupta, I. E.
[11] J. G. Kirk, F. M. Rieger, and A. Mastichiadis, Papadakis, and P. J. Wiita,
Astronomy and Astrophysics 333, 452 (1998), Monthly Notices of the Royal Astronomical Society 444, 3647 (2014)
arXiv:astro-ph/9801265 [astro-ph]. https://academic.oup.com/mnras/article-pdf/444/4/3647/6353693/st
[12] F. M. Rieger, V. Bosch-Ramon, and P. Duffy, [26] H. Gaur, L. Chen, R. Misra, S. Sahayanathan,
Astrophysics and Space Science 309, 119 (2007). M. F. Gu, P. Kushwaha, and G. C. Dewangan,
[13] N. S. Kardashev, Soviet Astronomy 6, 317 (1962). The Astrophysical Journal 850, 209 (2017).
[14] G. B. Rybicki and A. P. Lightman, Radiative Processes [27] A. Tramacere, E. Massaro, and A. M.
in Astrophysics (1986). Taylor, Astrophys. J. 739, 66 (2011),
[15] N. Mankuzhiyil, S. Ansoldi, M. Persic, arXiv:1107.1879 [astro-ph.HE].
E. Rivers, R. Rothschild, and F. Tavecchio, [28] A. Tramacere, F. Massaro, and A. Cava-
The Astrophysical Journal 753, 154 (2012). liere, Astronomy and Astrophysics 466, 521 (2007),
[16] C. Baheeja, S. Sahayanathan, F. M. Rieger, arXiv:astro-ph/0702151 [astro-ph].
S. K. Jagan, and C. D. Ravikumar, [29] F. Massaro, A. Tramacere, A. Cav-
Mon. Not. Roy. Astr. Soc. 514, 3074 (2022), aliere, M. Perri, and P. Giommi,
arXiv:2206.00659 [astro-ph.HE]. Astronomy and Astrophysics 478, 395 (2008),
[17] R. Landau, B. Golisch, T. J. Jones, T. W. Jones, arXiv:0712.2116 [astro-ph].
J. Pedelty, L. Rudnick, M. L. Sitko, J. Kenney, T. Roel- [30] B. Kapanadze, S. Vercellone, P. Romano, P. Hughes,
lig, E. Salonen, S. Urpo, G. Schmidt, G. Neugebauer, M. Aller, H. Aller, O. Kharshiladze, S. Kapanadze, and
K. Matthews, J. H. Elias, C. Impey, P. Clegg, and S. Har- L. Tabagari, Astrophys. J. 854, 66 (2018).
ris, Astrophys. J. 308, 78 (1986). [31] B. Kapanadze, S. Vercellone, P. Romano, P. Hughes,
[18] E. Massaro, M. Perri, P. Giommi, and R. Nesci, M. Aller, H. Aller, O. Kharshiladze, and L. Tabagari,
 11

Astrophys. J. 858, 68 (2018). S. D. Bongiorno, A. Brez, N. Bucciantini, F. Capi-

[32] B. Kapanadze, A. Gurchumelia, D. Dorner, tanio, S. Castellano, S. Ciprini, E. Costa, E. Del Monte,
S. Vercellone, P. Romano, P. Hughes, N. Di Lalla, V. Doroshenko, M. Dovčiak, T. Enoto,
M. Aller, H. Aller, and O. Kharshiladze, Y. Evangelista, S. Fabiani, R. Ferrazzoli, J. A. Garcia,
The Astrophysical Journal Supplement Series 247, 27 (2020). S. Gunji, K. Hayashida, J. Heyl, W. Iwakiri, F. Kislat,
[33] P. Goswami, S. Sahayanathan, T. Kitaguchi, J. J. Kolodziejczak, F. La Monaca, L. La-
A. Sinha, R. Misra, and R. Gogoi, tronico, S. Maldera, A. Manfreda, C. Y. Ng, N. Omodei,
C. Oppedisano, A. Papitto, G. G. Pavlov, M. Pesce-
Monthly Notices of the Royal Astronomical Society 480, 2046 (2018),
Rollins, M. Pilia, A. Possenti, J. Poutanen, J. Rankin,
https://academic.oup.com/mnras/article-pdf/480/2/2046/25440521/sty2003.pdf.
[34] P. Goswami, S. Sahayanathan, A. Sinha, and R. Gogoi, A. Ratheesh, O. J. Roberts, C. Sgrò, P. Slane, P. Sof-
fitta,
Monthly Notices of the Royal Astronomical Society 499, 2094–2103 G. Spandre, D. A. Swartz, T. Tamagawa, R. Tav-
(2020).
[35] A. P. Marscher and W. K. Gear, erna, Y. Tawara, N. E. Thomas, F. Tombesi, A. Trois,
Astrophys. J. 298, 114 (1985). S. S. Tsygankov, R. Turolla, J. Vink, M. C. Weisskopf,
[36] A. Zech and M. Lemoine, F. Xie, and S. Zane, Nature Astronomy 7, 1245 (2023),
Astronomy and Astrophysics 654, A96 (2021), arXiv:2305.13497 [astro-ph.HE].
arXiv:2108.12271 [astro-ph.HE]. [39] V. N. Zirakashvili and F. Aharonian,
[37] L. Di Gesu, I. Donnarumma, F. Tavecchio, I. Agudo, Astronomy and Astrophysics 465, 695 (2007).
T. Barnounin, N. Cibrario, N. Di Lalla, A. Di Marco, [40] K. D. Fritz, Astronomy and Astrophysics 214, 14 (1989).
J. Escudero, M. Errando, S. G. Jorstad, D. E. Kim, P. M. [41] J. Kataoka and L. Stawarz,
Kouch, I. Liodakis, E. Lindfors, G. Madejski, H. L. Mar- The Astrophysical Journal 827, 55 (2016).
shall, A. P. Marscher, R. Middei, F. Muleri, I. Myserlis, [42] M. Baloković , D. Paneque, G. Madejski, A. Furniss,
M. Negro, N. Omodei, L. Pacciani, A. Paggi, M. Perri, J. Chiang, M. Ajello, D. M. Alexander, D. Barret, R. D.
S. Puccetti, L. A. Antonelli, M. Bachetti, L. Baldini, Blandford, S. E. Boggs, F. E. Christensen, W. W. Craig,
W. H. Baumgartner, R. Bellazzini, S. Bianchi, S. D. Bon- K. Forster, P. Giommi, B. Grefenstette, C. Hailey, F. A.
giorno, R. Bonino, A. Brez, N. Bucciantini, F. Capitanio, Harrison, A. Hornstrup, T. Kitaguchi, J. E. Koglin, K. K.
S. Castellano, E. Cavazzuti, S. Ciprini, E. Costa, A. De Madsen, P. H. Mao, H. Miyasaka, K. Mori, M. Perri,
Rosa, E. Del Monte, V. Doroshenko, M. Dovčiak, S. R. M. J. Pivovaroff, S. Puccetti, V. Rana, D. Stern, G. Tagli-
Ehlert, T. Enoto, Y. Evangelista, S. Fabiani, R. Fer- aferri, C. M. Urry, N. J. Westergaard, W. W. Zhang,
razzoli, J. A. Garcia, S. Gunji, K. Hayashida, J. Heyl, A. Zoglauer, S. Archambault, A. Archer, A. Barnacka,
W. Iwakiri, V. Karas, T. Kitaguchi, J. J. Kolodziejczak, W. Benbow, R. Bird, J. H. Buckley, V. Bugaev, M. Cer-
H. Krawczynski, F. La Monaca, L. Latronico, S. Maldera, ruti, X. Chen, L. Ciupik, M. P. Connolly, W. Cui, H. J.
A. Manfreda, F. Marin, A. Marinucci, F. Massaro, Dickinson, J. Dumm, J. D. Eisch, A. Falcone, Q. Feng,
G. Matt, I. Mitsuishi, T. Mizuno, C. Y. Ng, S. L. O’Dell, J. P. Finley, H. Fleischhack, L. Fortson, S. Griffin, S. T.
C. Oppedisano, A. Papitto, G. G. Pavlov, A. L. Peirson, Griffiths, J. Grube, G. Gyuk, M. Huetten, N. Håkansson,
M. Pesce-Rollins, P.-O. Petrucci, M. Pilia, A. Possenti, J. Holder, T. B. Humensky, C. A. Johnson, P. Kaaret,
J. Poutanen, B. D. Ramsey, J. Rankin, A. Ratheesh, M. Kertzman, Y. Khassen, D. Kieda, M. Krause,
R. W. Romani, C. Sgrò, P. Slane, P. Soffitta, G. Span- F. Krennrich, M. J. Lang, G. Maier, S. McArthur,
dre, T. Tamagawa, R. Taverna, Y. Tawara, A. F. Ten- K. Meagher, P. Moriarty, T. Nelson, D. Nieto, R. A.
nant, N. E. Thomas, F. Tombesi, A. Trois, S. Tsygankov, Ong, N. Park, M. Pohl, A. Popkow, E. Pueschel, P. T.
R. Turolla, J. Vink, M. C. Weisskopf, K. Wu, F. Xie, and Reynolds, G. T. Richards, E. Roache, M. Santander,
S. Zane, Astrophysical Journal, Letters 938, L7 (2022), G. H. Sembroski, K. Shahinyan, A. W. Smith, D. Staszak,
arXiv:2209.07184 [astro-ph.HE]. I. Telezhinsky, N. W. Todd, J. V. Tucci, J. Tyler,
[38] L. Di Gesu, H. L. Marshall, S. R. Ehlert, D. E. Kim, S. Vincent, A. Weinstein, A. Wilhelm, D. A. Williams,
I. Donnarumma, F. Tavecchio, I. Liodakis, S. Kiehlmann, B. Zitzer, M. L. Ahnen, S. Ansoldi, L. A. Antonelli,
I. Agudo, S. G. Jorstad, F. Muleri, A. P. Marscher, P. Antoranz, A. Babic, B. Banerjee, P. Bangale, U. B.
S. Puccetti, R. Middei, M. Perri, L. Pacciani, M. Ne- de Almeida, J. A. Barrio, J. B. González, W. Bednarek,
gro, R. W. Romani, A. Di Marco, D. Blinov, I. G. Bour- E. Bernardini, B. Biasuzzi, A. Biland, O. Blanch, S. Bon-
bah, E. Kontopodis, N. Mandarakas, S. Romanopou- nefoy, G. Bonnoli, F. Borracci, T. Bretz, E. Car-
los, R. Skalidis, A. Vervelaki, C. Casadio, J. Escud- mona, A. Carosi, A. Chatterjee, R. Clavero, P. Colin,
ero, I. Myserlis, M. A. Gurwell, R. Rao, G. K. Keat- E. Colombo, J. L. Contreras, J. Cortina, S. Covino,
ing, P. M. Kouch, E. Lindfors, F. J. Aceituno, M. I. P. D. Vela, F. Dazzi, A. D. Angelis, B. D. Lotto,
Bernardos, G. Bonnoli, V. Casanova, M. Garcı́a-Comas, E. de Oña Wilhelmi, C. D. Mendez, F. D. Pierro, D. D.
B. Agı́s-González, C. Husillos, A. Marchini, A. Sota, Prester, D. Dorner, M. Doro, S. Einecke, D. Elsaesser,
R. Imazawa, M. Sasada, Y. Fukazawa, K. S. Kawabata, A. Fernández-Barral, D. Fidalgo, M. V. Fonseca, L. Font,
M. Uemura, T. Mizuno, T. Nakaoka, H. Akitaya, S. S. K. Frantzen, C. Fruck, D. Galindo, R. J. G. López,
Savchenko, A. A. Vasilyev, J. L. Gómez, L. A. An- M. Garczarczyk, D. G. Terrats, M. Gaug, P. Giammaria,
tonelli, T. Barnouin, R. Bonino, E. Cavazzuti, L. Costa- D. G. (Eisenacher), N. Godinović, A. G. Muñoz, D. Gu-
mante, C.-T. Chen, N. Cibrario, A. De Rosa, F. Di berman, A. Hahn, Y. Hanabata, M. Hayashida, J. Her-
Pierro, M. Errando, P. Kaaret, V. Karas, H. Krawczyn- rera, J. Hose, D. Hrupec, G. Hughes, W. Idec, K. Ko-
ski, L. Lisalda, G. Madejski, C. Malacaria, F. Marin, dani, Y. Konno, H. Kubo, J. Kushida, A. L. Barbera,
A. Marinucci, F. Massaro, G. Matt, I. Mitsuishi, S. L. D. Lelas, E. Lindfors, S. Lombardi, F. Longo, M. López,
O’Dell, A. Paggi, A. L. Peirson, P.-O. Petrucci, B. D. R. López-Coto, A. López-Oramas, E. Lorenz, P. Majum-
Ramsey, A. F. Tennant, K. Wu, M. Bachetti, L. Bal- dar, M. Makariev, K. Mallot, G. Maneva, M. Manga-
dini, W. H. Baumgartner, R. Bellazzini, S. Bianchi, naro, K. Mannheim, L. Maraschi, B. Marcote, M. Mari-
 12

otti, M. Martı́nez, D. Mazin, U. Menzel, J. M. Miranda, T. C. Weekes, R. J. White, D. A. Williams, M. D.

R. Mirzoyan, A. Moralejo, E. Moretti, D. Nakajima, Wood, B. Zitzer, H. D. Aller, M. F. Aller, M. Baker,
V. Neustroev, A. Niedzwiecki, M. N. Rosillo, K. Nilsson, D. Barnaby, M. T. Carini, P. Charlot, J. P. Dumm, N. E.
K. Nishijima, K. Noda, R. Orito, A. Overkemping, S. Pa- Fields, T. Hovatta, B. Jordan, Y. A. Kovalev, Y. Y. Ko-
iano, J. Palacio, M. Palatiello, R. Paoletti, J. M. Paredes, valev, H. A. Krimm, O. M. Kurtanidze, A. Lähteenmäki,
X. Paredes-Fortuny, M. Persic, J. Poutanen, P. G. P. J. F. L. Campion, J. Maune, T. Montaruli, A. C. Sadun,
Moroni, E. Prandini, I. Puljak, W. Rhode, M. Ribó, S. Smith, M. Tornikoski, M. Turunen, and R. Walters,
J. Rico, J. R. Garcia, T. Saito, K. Satalecka, V. Scapin, The Astrophysical Journal 695, 596 (2009).
C. Schultz, T. Schweizer, S. N. Shore, A. Sillanpää, [45] F. A. Harrison, W. W. Craig, F. E. Christensen, C. J.
J. Sitarek, I. Snidaric, D. Sobczynska, A. Stamerra, Hailey, W. W. Zhang, S. E. Boggs, D. Stern, W. R.
T. Steinbring, M. Strzys, L. Takalo, H. Takami, F. Tavec- Cook, K. Forster, P. Giommi, B. W. Grefenstette,
chio, P. Temnikov, T. Terzić, D. Tescaro, M. Teshima, Y. Kim, T. Kitaguchi, J. E. Koglin, K. K. Madsen,
J. Thaele, D. F. Torres, T. Toyama, A. Treves, V. Ver- P. H. Mao, H. Miyasaka, K. Mori, M. Perri, M. J.
guilov, I. Vovk, J. E. Ward, M. Will, M. H. Wu, R. Zanin, Pivovaroff, S. Puccetti, V. R. Rana, N. J. Westergaard,
J. Perkins, F. Verrecchia, C. Leto, M. Böttcher, M. Vil- J. Willis, A. Zoglauer, H. An, M. Bachetti, N. M.
lata, C. M. Raiteri, J. A. Acosta-Pulido, R. Bachev, Barrière, E. C. Bellm, V. Bhalerao, N. F. Brejnholt,
A. Berdyugin, D. A. Blinov, M. I. Carnerero, W. P. Chen, F. Fuerst, C. C. Liebe, C. B. Markwardt, M. Nynka,
P. Chinchilla, G. Damljanovic, C. Eswaraiah, T. S. Gr- J. K. Vogel, D. J. Walton, D. R. Wik, D. M. Alexan-
ishina, S. Ibryamov, B. Jordan, S. G. Jorstad, M. Joshi, der, L. R. Cominsky, A. E. Hornschemeier, A. Hornstrup,
E. N. Kopatskaya, O. M. Kurtanidze, S. O. Kurtanidze, V. M. Kaspi, G. M. Madejski, G. Matt, S. Molendi,
E. G. Larionova, L. V. Larionova, V. M. Larionov, D. M. Smith, J. A. Tomsick, M. Ajello, D. R. Bal-
G. Latev, H. C. Lin, A. P. Marscher, A. A. Mokrushina, lantyne, M. Baloković, D. Barret, F. E. Bauer, R. D.
D. A. Morozova, M. G. Nikolashvili, E. Semkov, P. S. Blandford, W. N. Brandt, L. W. Brenneman, J. Chiang,
Smith, A. Strigachev, Y. V. Troitskaya, I. S. Troit- D. Chakrabarty, J. Chenevez, A. Comastri, F. Dufour,
sky, O. Vince, J. Barnes, T. Güver, J. W. Moody, M. Elvis, A. C. Fabian, D. Farrah, C. L. Fryer, E. V.
A. C. Sadun, S. Sun, T. Hovatta, J. L. Richards, Gotthelf, J. E. Grindlay, D. J. Helfand, R. Krivonos,
W. Max-Moerbeck, A. C. R. Readhead, A. Lähteenmäki, D. L. Meier, J. M. Miller, L. Natalucci, P. Ogle, E. O.
M. Tornikoski, J. Tammi, V. Ramakrishnan, R. Reinthal, Ofek, A. Ptak, S. P. Reynolds, J. R. Rigby, G. Tagli-
E. Angelakis, L. Fuhrmann, I. Myserlis, V. Karamanavis, aferri, S. E. Thorsett, E. Treister, and C. M. Urry,
A. Sievers, H. Ungerechts, J. A. Zensus, , and and, The Astrophysical Journal 770, 103 (2013).
The Astrophysical Journal 819, 156 (2016). [46] D. N. Burrows, J. E. Hill, J. A. Nousek, J. A. Kennea,
[43] G. Fossati, J. H. Buckley, I. H. Bond, S. M. Bradbury, A. Wells, J. P. Osborne, A. F. Abbey, A. Beardmore,
D. A. Carter-Lewis, Y. C. K. Chow, W. Cui, A. D. Fal- K. Mukerjee, A. D. T. Short, G. Chincarini, S. Cam-
cone, J. P. Finley, J. A. Gaidos, J. Grube, J. Holder, pana, O. Citterio, A. Moretti, C. Pagani, G. Taglia-
D. Horan, D. Horns, M. M. Jordan, D. B. Kieda, ferri, P. Giommi, M. Capalbi, F. Tamburelli, L. Angelini,
J. Kildea, H. Krawczynski, F. Krennrich, M. J. Lang, G. Cusumano, H. W. Bräuninger, W. Burkert, and
S. LeBohec, K. Lee, P. Moriarty, R. A. Ong, D. Petry, G. D. Hartner, Space Science Reviews 120, 165 (2005),
J. Quinn, G. H. Sembroski, S. P. Wakely, and T. C. arXiv:astro-ph/0508071 [astro-ph].
Weekes, The Astrophysical Journal 677, 906 (2008). [47] E. Lefa, S. R. Kelner, and F. A.
[44] D. Horan, V. A. Acciari, S. M. Bradbury, J. H. Buck- Aharonian, Astrophys. J. 753, 176 (2012),
ley, V. Bugaev, K. L. Byrum, A. Cannon, O. Celik, arXiv:1205.2929 [astro-ph.HE].
A. Cesarini, Y. C. K. Chow, L. Ciupik, P. Cogan, [48] C. Romoli, A. M. Taylor, and F. Aha-
A. D. Falcone, S. J. Fegan, J. P. Finley, P. Fortin, ronian, Astroparticle Physics 88, 38 (2017),
L. F. Fortson, D. Gall, G. H. Gillanders, J. Grube, arXiv:1608.01501 [astro-ph.HE].
G. Gyuk, D. Hanna, E. Hays, M. Kertzman, J. Kildea,
A. Konopelko, H. Krawczynski, F. Krennrich, M. J.
Lang, K. Lee, P. Moriarty, T. Nagai, J. Niemiec, R. A. Appendix A: Sample X-ray spectral fits using a
Ong, J. S. Perkins, M. Pohl, J. Quinn, P. T. Reynolds,
log-parabola model (Figure 9, 10 ).
H. J. Rose, G. H. Sembroski, A. W. Smith, D. Steele,
S. P. Swordy, J. A. Toner, V. V. Vassiliev, S. P. Wakely,
 13

ObsID 60002023027 ObsID 60002023033

keV2 (Photons cm−2 s−1 keV−1)

keV2 (Photons cm−2 s−1 keV−1)

0.5
1

0.5 0.2

0.1
0.2
0.05
0.1
(data−model)/error

(data−model)/error
2 2

0 0

−2 −2

5 10 20 50 5 10 20 50
Energy (keV) Energy (keV)
ObsID 60002023035 ObsID 60202048002
keV2 (Photons cm−2 s−1 keV−1)

keV2 (Photons cm−2 s−1 keV−1)

0.5
0.2

0.2
0.1

0.1
0.05
0.05
(data−model)/error

(data−model)/error
2
2
0
0
−2
−2
−4
5 10 20 50 5 10 20 50
Energy (keV) Energy (keV)
ObsID 60202048004 ObsID 60202048006
keV2 (Photons cm−2 s−1 keV−1)

keV2 (Photons cm−2 s−1 keV−1)

0.2 0.2

0.1 0.1

0.05
0.05
(data−model)/error

(data−model)/error

2
2
0
0
−2
−2
−4
5 10 20 50 5 10 20 50
Energy (keV) Energy (keV)

FIG. 9. NuSTAR X-ray spectra (3-79 keV) along with log-parabola model.
 14

00080050019+60002023027 00035014062+60002023033

keV2 (Photons cm−2 s−1 keV−1)

keV2 (Photons cm−2 s−1 keV−1)

2
0.5
1

0.5 0.2

0.2 0.1

0.1 0.05
(data−model)/error

(data−model)/error
2 2

0 0

−2 −2

1 10 1 10
Energy (keV) Energy (keV)
00035014065+60002023035 00034228110+60202048002

0.5
keV2 (Photons cm−2 s−1 keV−1)

keV2 (Photons cm−2 s−1 keV−1)

1

0.5
0.2

0.2
0.1
0.1

0.05 0.05
(data−model)/error

(data−model)/error
2 2

0 0
−2
−2
−4
1 10 1 10
Energy (keV) Energy (keV)
00081926001+60202048004 00034228145+60002023029

0.5
keV2 (Photons cm−2 s−1 keV−1)

keV2 (Photons cm−2 s−1 keV−1)

0.2
0.2

0.1
0.1

0.05
0.05
(data−model)/error

(data−model)/error

2
2
0
0
−2
−2
−4
1 10 1 10
Energy (keV) Energy (keV)

FIG. 10. Combined Swift–XRT and NuSTAR X-ray spectra (0.3-79 keV) along with log-parabola model.