The Origin of X-ray Emission in the Gamma-ray emitting Narrow-Line Seyfert 1 1H 0323+342
Sergio A. Mundo, Erin Kara, Edward M. Cackett, A.C. Fabian, J. Jiang, R.F. Mushotzky, M.L. Parker, C. Pinto, C.S. Reynolds, A. Zoghbi
MMNRAS , 1–11 (2020) Preprint 1 July 2020 Compiled using MNRAS L A TEX style file v3.0
The origin of X-ray emission in the gamma-ray emittingnarrow-line Seyfert 1 1H 0323+342
Sergio A. Mundo, (cid:63) Erin Kara, , Edward M. Cackett, A.C. Fabian, J. Jiang, , R. F. Mushotzky, M. L. Parker, C. Pinto, C. S. Reynolds, A. Zoghbi Department of Astronomy, University of Maryland, College Park, MD 20742, USA MIT Kavli Institute for Astrophysics and Space Research, Cambridge, MA 02139, USA Department of Physics & Astronomy, Wayne State University, 666 W. Hancock St, Detroit, MI 48201, USA Institute of Astronomy, University of Cambridge, Madingley Road, Cambridge CB3 0HA, UK Tsinghua Center for Astrophysics, Tsinghua University, Beijing 100084, China European Space Agency (ESA), European Space Astronomy Centre (ESAC), E-28691 Villanueva de la Ca˜nada, Madrid, Spain ESTEC/ESA, Keplerlaan 1, 2201AZ Noordwijk, The Netherlands Department of Astronomy, University of Michigan, Ann Arbor, MI 48109-1107, USA
Accepted 2020 June 5. Received 2020 April 24; in original form 2019 August 29
ABSTRACT
We present the results of X-ray spectral and timing analyses of the closest gamma-ray emitting narrow-line Seyfert 1 ( γ -NLS1) galaxy, 1H 0323+342. We use observa-tions from a recent, simultaneous XMM-Newton / NuSTAR campaign. As in radio-quietNLS1s, the spectrum reveals a soft excess at low energies ( (cid:46) keV) and reflection fea-tures such as a broad iron K emission line. We also find evidence of a hard excessat energies above ∼ keV that is likely a consequence of jet emission. Our analysisshows that relativistic reflection is statistically required, and using a combination ofmodels that includes the reflection model relxill for the broadband spectrum, wefind an inclination of i = + − degrees, which is in tension with much lower values in-ferred by superluminal motion in radio observations. We also find a flat ( q = . ± . )emissivity profile, implying that there is more reflected flux than usual being emittedfrom the outer regions of the disk, which in turn suggests a deviation from the thindisk model assumption. We discuss possible reasons for this, such as reflection off of athick accretion disk geometry. Key words: galaxies: active – X-rays: galaxies – galaxies: jets – galaxies: Seyfert –galaxies: individual: 1H 0323+342
Narrow-line Seyfert 1 (NLS1) galaxies are a type of activegalactic nucleus (AGN) characterized by their unique opticalspectral features, such as broad Balmer emission lines withlow widths (FWHM H β < km s − ), weak [OIII] emission([OIII]/H β flux < ∼ − M (cid:12) ,(Grupe & Mathur 2004; Deo et al. 2006)) that accrete nearor above the Eddington limit (e.g., Peterson et al. 2000). Asa result, it is believed that NLS1s form a set of younger AGNthat have yet to transform into more luminous quasars.In addition to their optical properties, NLS1s are brightin the X-ray band and exhibit complex X-ray spectral fea-tures, such as an excess in soft X-ray emission and reflection (cid:63) E-mail: [email protected] features in the hard X-rays. As with other accretion sys-tems around black holes (e.g. black hole X-ray binaries),the primary form of the X-ray emission is a continuum well-modeled by a power law that results from the inverse Comp-ton scattering of seed disk photons by a hot plasma of elec-trons, or a corona, that lies above the disk in the vicinity ofthe black hole. Some of this continuum emission illuminatesthe accretion disk, and the upscattered photons end up ei-ther Compton scattering off of electrons in the disk, or arereprocessed through fluorescence (see Reynolds & Nowak2003 for a review). These reflection mechanisms have alsobeen featured in the spectra of NLS1s in the form of aniron emission line between 6-7 keV and a Compton reflec-tion “hump” that peaks between around 20 and 30 keV, im-plying that the reflection is occurring off of an ionized disk(for reflection features in NLS1s, see e.g. Marinucci et al.2014; Kara et al. 2017). The Fe emission line in these spec-tra is usually broadened and skewed towards lower energies © a r X i v : . [ a s t r o - ph . H E ] J un Mundo et al. due to line-of-sight Doppler boosting and the gravitationalpotential of the black hole, respectively (Fabian et al. 1989;Reynolds & Nowak 2003). Which effect is most dominantdepends on the inclination of the disk relative to our lineof sight; therefore, from reflection processes alone, we canarrive at an estimate for the disk inclination of a NLS1.The origin of the soft excess is still disputed. It canbe modeled as blackbody thermal emission, but this is notphysical because the resulting blackbody temperature is sim-ply too high to be emission from the disk. A way aroundthis is to consider a “warm” Comptonization region that isoptically thicker than the corona, which yields a tempera-ture of around 0.1-0.2 keV that is constant across a widerange of black hole masses and accretion rates (e.g., Gier-li´nski & Done 2004). Other ideas that have been put forthplace atomic processes like reflection or absorption as theculprits. One example of the former is that coronal illumi-nation of the disk could result in the fluorescence of linesat lower energies that end up being blurred due to grav-itational effects (Crummy et al. 2006; Fabian et al. 2009;Walton et al. 2013). The soft excess could also be describedby a disk with a high electron density. With a higher den-sity, bremsstrahlung would have a higher contribution to thespectrum, in the form of an increased temperature at thesurface of the illuminated disk that results from free-freeabsorption (Garc´ıa et al. 2016; Jiang et al. 2018; Mallicket al. 2018; Jiang et al. 2019). This, in turn, may causeblurred reflection at low energies to look more like a black-body spectrum. If a higher density disk is not taken intoaccount, this could result in a perceived excess at lower en-ergies, especially in AGN with smaller supermassive blackholes ( M (cid:28) M (cid:12) ).The spectral features of NLS1s have also been describedby a series of alternative, absorption-based partial coveringmodels. In this family of models, the X-ray variability is notintrinsic to the source, but is rather caused by a varyingpartial covering fraction of clouds that possibly result fromdisk instabilities or radiation-driven outflows. Gallo et al.2004 and Tanaka et al. 2004 described spectral changes inthe NLS1 1H 0707-495 with neutral single and double layerabsorbers, respectively, with the latter assuming that thecovering fraction changed with the clouds’ orbital motion.In addition, Miyakawa et al. 2012 and Mizumoto et al. 2014were able to explain the broad line feature in MCG-6-30-15and 1H 0707-495 with ionized partial covering models, withMizumoto et al. 2014 suggesting that the clouds are pro-duced by funnel-shaped disk winds. However, several stud-ies have shown that most accreting objects have a linearRMS-flux relation, which suggests that the underlying X-ray variability processes are multiplicative in nature and aretherefore intrinsic to the source (e.g., Alston 2019; Uttleyet al. 2005). This is at odds with the inherent features ofpartial covering models, which would show shot noise, ad-ditive variability. Due to these characteristics and the mys-terious nature of some of these observations, NLS1s, alongwith other Seyferts, are not only ideal candidates for study-ing X-ray emission processes and their origins, but also offerinteresting possibilities in the realm of X-ray astronomy per-taining to an AGN’s central region.Roughly 10% of AGN exhibit collimated, relativisticjets that emit in the radio band via synchrotron radiation(e.g., Begelman et al. 1984). Another unsolved problem in the physics of AGN processes is exactly where and how thesejets are launched, and as a result, the connection betweenthe disk, corona, and the jet’s driving mechanism is currentlypoorly understood. In any case, “radio-loud” (RL) AGN withthe X-ray spectral features discussed earlier would providethe most promising environment in which to study the in-teraction between these three and other components. MostNLS1s, however, are “radio-quiet” (RQ), meaning that theirjets are not nearly as powerful as those from blazars orother RL AGN. Therefore, looking for the long-sought disk-corona-jet connection in these AGN proves to be quite diffi-cult.In recent years, a new class of NLS1s has been discov-ered by the Fermi Gamma-Ray Telescope . These are gamma-ray emitting, RL NLS1s ( γ -NLS1), which have features seenin both NLS1s (discussed previously) and blazars, such asflat radio spectra, double-hump spectral energy distribution(SED) (e.g., Abdo et al. 2009), and occasionally superlumi-nal motion (e.g., D’Ammando et al. 2013). The presence ofgamma-ray emission in these galaxies is evidence for a jet,given that in those blazars classified as flat spectrum radioquasars, photons from outside a jet can be inverse Comp-ton scattered to hard X-rays or γ -rays by the relativisticparticles in the jet; this is referred to as external Compton(EC) (e.g., B(cid:32)la˙zejowski et al. 2000; Ghisellini et al. 2009).Studies of the X-ray spectra of these γ -NLS1s have alreadyshown not only a soft excess, but also a “hard excess” abovea few keV that requires a much harder power law componentcompared to the ones usually found in AGN (Bhattacharyyaet al. 2014), possibly representing the jet emission. There-fore, these γ -NLS1s provide us with an unusual laboratoryfor studying properties from both the jet and the thermalemission from the corona simultaneously, and can give usinsight into the dominant mechanism that produces the X-ray emission, while at the same time shedding light on howextragalactic jets are formed.The closest of these exotic AGN is 1H 0323+342 ( α :03 24 41.16, δ : +34 10 45.8), with a redshift of z = . (Zhou et al. 2007). Radio images of this source exhibit su-perluminal motion (1-7 times the speed of light), which inturn implies the presence of a relativistic jet at an angle i = − deg from the line of sight (Fuhrmann et al. 2016).1H 0323+342 also has a double-hump SED characteristic of γ -NLS1s that peaks in the radio band and gamma-rays, im-plying synchrotron emission and synchrotron self-Compton(or EC) mechanisms that would result from a jet.Exactly where the X-ray emission from 1H 0323+342comes from is still unknown. Currently, two main explana-tions have been put forth: it could be the result of inter-actions between the disk and the corona, as in RQ NLS1s(power-law continuum, reflection features, etc.; see Paliyaet al. (2014), Paliya et al. (2019)), or it could simply bea continuation of the double-hump SED that fills the gapbetween the radio and gamma-ray emission from the jet,implying that X-ray emission would be included as a directconsequence of interactions with the jet (for a detailed multi-wavelength spectral analysis for this source, see Kynochet al. 2018).1H 0323+342 has been observed by Swift and
Suzaku ,and the spectra show properties seen in RQ NLS1s, such as asoft excess and a potential broad Fe K α emission line (e.g.,Walton et al. 2013; Paliya et al. 2019). Archival data has MNRAS , 1–11 (2020) -Ray Spectral Analysis of 1H 0323+342 Table 1.
XMM-Newton and
NuSTAR observations used in ouranalysis. Shown are the detector name, observation ID, observa-tion start dates, the duration of the observations, and the effectiveexposure times (for
XMM , after excluding epochs of flaring par-ticle background). All
XMM observations were made in LargeWindow mode.Detector Obs. ID StartDate Duration(s) EffectiveExposure(s)EPIC-pn 0764670101 2015/08/23 80900 620000823780201 2018/08/14 53066 476000823780301 2018/08/18 48212 438000823780401 2018/08/20 47975 454610823780501 2018/08/24 48515 460020823780601 2018/09/05 50818 447000823780701 2018/09/09 49468 46951FPMA/ 60061360002 2014/03/15 108880 101633FPMB 60402003002 2018/08/14 38937 3639060402003004 2018/08/18 31711 2973260402003006 2018/08/20 28137 2640160402003008 2018/08/24 27276 2556560402003010 2018/09/05 32516 3042260402003012 2018/09/09 29921 27795 shown that there is a potentially blue-shifted iron line thatwould require a disk inclination of nearly 90 ◦ due to line-of-sight Doppler boosting (Walton et al. 2013; Yao et al.2015), which contradicts data from radio observations sincethe superluminal motion in the latter indicates that the jet isemitted at an angle close to our line of sight, and thereforealso suggests that the disk is face-on. However, the broadiron line has never been clearly detected due to low signal-to-noise.In this paper, we present an X-ray spectral and timinganalysis with the first such data set from XMM-Newton and
NuSTAR for this source. We aim to find the origin of theX-ray emission in 1H 0323+342, while at the same time ob-taining an estimate of the disk inclination purely from theX-ray spectra. We describe the observations and data reduc-tion in Section 2, present our spectral and timing analysesand their results in Section 3, and discuss these results inSection 4.
To study this source, we use a set of six simultaneousobservations made by both
XMM-Newton and
NuSTAR (twelve total, PI: Kara; see Table 1), as well as an archivalobservation that was made in 2015 (ID 0764670101; PI:D’Ammando).
XMM-Newton
The
XMM-Newton data were reduced with the
XMM-Newton
Science Analysis System (SAS v.16.1.0) and thecurrent calibration files available. We focus on the data fromthe EPIC-pn camera here, which was taken in Large Win-dow mode. All observations were checked for flaring particlebackground, and we set pattern ≤ to choose single anddouble events and flag == to get the best quality data.The data was also checked for pile-up with the SAS epat-plot task, but there was no pile-up present in any of the observations. We extracted the source spectra from circularregions with radii of 40 (cid:48)(cid:48) that were centered on the J2000coordinates of 1H0323+342. The background spectra wereobtained from circular background regions that were offsetfrom the source and were made as large as possible, withradii of 70 (cid:48)(cid:48) .Background-subtracted light curves were also extractedwith the epiclccorr tool and were binned with 10 secondbins (see Fig. 1). Although the flux varies by about a fac-tor of 4, there was no significant change in the spectrumbetween observations, and the hardness ratio between thehard and soft bands remains relatively constant across eachobservation. We therefore decided to combine them to formone spectrum. This co-added spectrum was rebinned withthe grppha tool to ensure that we would have a minimum of25 counts per bin. NuSTAR
We extracted the
NuSTAR data using HEASOFT v.6.22.1and the standard nupipeline task, again from 40 (cid:48)(cid:48) circu-lar regions for both source and background spectra, for eachfocal plane module (FPMA/FPMB). The spectra were alsoco-added with the archival observation and binned with grp-pha to have at least 25 counts per bin. The source spectrumis above the background spectrum up to ∼ keV, so weshow the relevant plots up to this energy. In the past, it has been shown that most accreting sourceshave a linear relationship between absolute RMS and flux.Previous works (e.g., Alston 2019; Uttley et al. 2005; Uttley& McHardy 2001) have shown that this relation implies avariability process that is intrinsic to the source and that islikely a multiplicative process, building up from propagatingmass accretion rate fluctuations occurring in the disk. Sincethe best model to describe this is one where accretion diskfluctuations are the source of the variability, finding sucha relation in 1H 0323+342 would suggest that disk-coronaand reflection models, where the X-rays are assumed to beintrinsic variations originating from the innermost regionsof the disk, would be best to describe the data.We begin a calculation of the rms-flux relationship inthe time domain by binning our
XMM lightcurves in 1000 sbins. At the same time, we compute the excess variance as afunction of time with the same binning, effectively produc-ing a ‘variance light curve’. We take the square root of theexcess variance to obtain the rms amplitude, and sort thesevalues by count rate. We then bin the rms amplitudes by fluxsuch that we have ∼ points in each bin. Errors on the rmsamplitude are calculated as in Vaughan et al. (2003). Weshow in Fig. 2 that we indeed obtain a linear rms-flux rela-tion, with a linear fit of χ = . for 5 degrees of freedom.This therefore supports the use of reflection models for thisdata and implies that the X-ray variations are intrinsic. MNRAS000
XMM lightcurves in 1000 sbins. At the same time, we compute the excess variance as afunction of time with the same binning, effectively produc-ing a ‘variance light curve’. We take the square root of theexcess variance to obtain the rms amplitude, and sort thesevalues by count rate. We then bin the rms amplitudes by fluxsuch that we have ∼ points in each bin. Errors on the rmsamplitude are calculated as in Vaughan et al. (2003). Weshow in Fig. 2 that we indeed obtain a linear rms-flux rela-tion, with a linear fit of χ = . for 5 degrees of freedom.This therefore supports the use of reflection models for thisdata and implies that the X-ray variations are intrinsic. MNRAS000 , 1–11 (2020)
Mundo et al.
Time (s) C oun t R a t e H a r dn e ss R a ti o Figure 1.
Top panel : EPIC-pn broadband lightcurves from 0.3-10 keV in 600 s bins. Light curves were generated with circular extractionregions of 40 arcsec radii. The first observation is archival and is longer than the rest by about 30 ks.
Bottom panel : Hardness ratio (1-4keV to 0.3-1 keV) of the EPIC-pn observations. For the spectral analysis, the observations were co-added as the hardness ratio remainsfairly constant throughout the campaign. r m s ( c t s s - ) r a ti o Flux (cts s -1 ) Figure 2.
Relationship between absolute rms and flux. On aver-age, the absolute rms increases linearly with flux, suggesting thatvariations are intrinsic to the source.
We fit the
XMM-Newton and
NuSTAR spectra simultane-ously using X spec v12.9.1p (Arnaud 1996). We included anoverall multiplicative constant in each of our models ( C A forFPMA and C B for FPMB) to take into account differencesin absolute flux calibration between each detector. We usedthe tbabs model (Wilms et al. 2000) and cross-sections fromVerner et al. (1996) to model the galactic absorption, setting N H , Gal = . × cm − (Kalberla et al. 2005). In all of ourfits, we also check for potentially different spectral shapesbetween XMM and
NuSTAR , but we find that the spectralindex Γ only differs by ∼ and does not affect our results,so we fit with the same photon index. All spectra are plottedin the source rest frame energies. We begin by searching for the broad iron line, which hasonly been marginally detected in previous observations (e.g.,Paliya et al. 2019, 2014; Yao et al. 2015; Walton et al. 2013). To show the highest resolution version of the line and to showwe are making a significant detection, we fit the 3-10 keVspectra with a power law (see Figure 3) and compare thisto a power law plus Gaussian. The addition of the Gaussianimproves the fit by ∆ χ = for 3 additional parameters,and therefore a line is preferred at > . ± . keV in the rest frame of the source thatis broad with respect to the spectral resolution of the instru-ments, with σ = . + . − . keV. This is consistent with Fe K α emission. We also calculate an equivalent width of around ± eV for XMM , which is stronger than in archival,lower signal-to-noise observations (e.g., Kynoch et al. 2018)and again supports our claim of a broad line.Figure 3 also shows a relatively pronounced peak be-tween 6 and 7 keV, at the 6.4 keV of neutral iron fluores-cence. We therefore include a narrower line fixed at 6.4 keV,in addition to the aforementioned Gaussian, and fix its widthto 100 eV. This improves the fit by ∆ χ = for one addi-tional parameter, at a significance of > . %. This suggeststhat, in addition to the broad iron line, there may also be anarrow component from a distant reflector that contributesto the spectrum.A power law fit to the 2-79 keV spectra of XMM-Newton and
NuSTAR shows not only a broad iron linepeaking at 6-7 keV, but also a Compton reflection humpabove 10 keV, indicating reflection off of an ionized disk(see Figure 4). This power law yields a reduced chi-squared χ / d . o . f = χ ν = / = . . Given the nature of thespectra, it is clear that reflection and possibly relativisticbroadening need to be taken into account, so we fit the 2-79keV data with different flavors of the reflection model relx-ill (Dauser et al. 2016). Following up on the narrow com-ponent at 6.4 keV discussed earlier, we start with a neutral,non-relativistic reflection model xillver , with the ionizationparameter log ξ fixed at 0. This gives a fit with χ ν = . , im-proving from the power law fit by ∆ χ = for 4 additionalfree parameters, with a significance of > F -test. The residuals between 6-7 keV, along withthe fact that we detect a fairly broad line, suggest we stillneed to account for relativistic smearing.We use the relativistic model relxill and set the inneraccretion disk radius R in to the innermost stable circular or- MNRAS , 1–11 (2020) -Ray Spectral Analysis of 1H 0323+342 r a ti o Energy (keV)
Figure 3.
The ratio of EPIC-pn to a simple power law; energiesare in the rest frame of the source. The blue line shows the posi-tion of neutral iron at 6.4 keV, the green line that of hydrogen-likeFe XXVI at 6.97 keV. We find an equivalent width of ± eVfor the broad iron line. (a) powerlaw r a ti o r a ti o r a ti o r a ti o Figure 4.
The ratio of EPIC-pn and
NuSTAR spectra to differentmodels. (a) The ratio to a power law. (b) Ratio to neutral reflec-tion model. (c) Ratio to relativistic reflection model. (d) Ratioto relativistic reflection model with an additional neutral, distantreflector. bit. We start by fixing the emissivity index to the Newtonianvalue q = and find a good fit, with χ ν = / = . .We also attempt using a broken power law emissivity, butthis yields a break radius R break that is below the innermoststable circular orbit, which is not physical. We therefore keepan unbroken emissivity for the rest of the paper. We also can-not constrain the spin a . Since this is a jetted source, it mayinvolve a high-spin object, so we use this physical motivationto fix the spin to the maximum value of 0.998. We combine the models in the second and third pan-els of Figure 4 to check for the possibility of an additionalneutral, distant reflector. All relevant parameters in xil-lver are tied to the parameters of relxill , except for thenormalization. This improves the fit by ∆ χ = for one ad-ditional free parameter, at a significance of > i = + − deg, which is in tension withthe low inclination obtained from radio observations. Within90% confidence, this fit yields very similar parameters as theones with just relxill , and we adopt this version for lateranalyses of the broadband spectrum. We report the best-fitparameters in Table 2.Our fits also seem to show consistent negative residu-als at ∼ keV (see Figures 3 and 4). It is important tonote that this is most likely not physical. When we plotthe background spectrum, we see that this is probably abackground feature, namely the EPIC-pn Cu-K α complex ataround those energies. Regardless of our source region sizeor background region placement (keeping the latter withinthe inner 4 CCDs), the feature remains. The residuals aretherefore likely the result of over-subtraction of the EPIC-pnbackground features. We verify this by plotting the spectrumwithout the background subtracted, and find that the fea-ture is not present (see Appendix A). We also do not seethe negative residuals in any of the individual EPIC-MOSspectra. Fitting a power law to the data in the 2-79 keV range andextrapolating the fit to lower energies also reveals a softexcess below 2 keV. We begin our analysis in the broadbandby including a phenomenological disk blackbody componentto help fit the spectrum at soft energies. To account for thepossibility of intrinsic absorption, we use ztbabs . We alsokeep the neutral, distant reflector from the previous section.We first attempt a Newtonian emissivity profile forthe tbabs*ztbabs*(diskbb + xillver + relxill) model.This gives a fit with χ ν = . . Freeing the emissivity indeximproves the fit by ∆ χ = for one additional free parame-ter at a significance > χ / d . o . f . = / = . . Increasing residuals at energies (cid:38) keV (see Figure4) remain, so we include an extra power law to account fora possible contribution of Compton upscattering further outin the jet, similar to Paliya et al. 2019. An additional hardpower law improves the fit by ∆ χ = for two additionalfree parameters with a significance > Γ = . + . − . forthe coronal emission models and Γ hard = . + . − . for the hardpower law, which is consistent with values found in γ -NLS1s(see e.g. Paliya et al. 2019 and the Γ histograms for theirdouble power-law fit, as well as photon index values in Ojhaet al. 2020). It also gives a high inclination of i = + − deg,as before. Moreover, we obtain a relatively flat emissivityindex q = . ± . , suggesting not only that strong generalrelativistic effects are not required to accurately describe theillumination pattern of the disk, but also that more reflectedflux is emitted from the outer regions of the disk. The factthat the observed flux seems to drop slowly at large radii MNRAS000
NuSTAR spectra to differentmodels. (a) The ratio to a power law. (b) Ratio to neutral reflec-tion model. (c) Ratio to relativistic reflection model. (d) Ratioto relativistic reflection model with an additional neutral, distantreflector. bit. We start by fixing the emissivity index to the Newtonianvalue q = and find a good fit, with χ ν = / = . .We also attempt using a broken power law emissivity, butthis yields a break radius R break that is below the innermoststable circular orbit, which is not physical. We therefore keepan unbroken emissivity for the rest of the paper. We also can-not constrain the spin a . Since this is a jetted source, it mayinvolve a high-spin object, so we use this physical motivationto fix the spin to the maximum value of 0.998. We combine the models in the second and third pan-els of Figure 4 to check for the possibility of an additionalneutral, distant reflector. All relevant parameters in xil-lver are tied to the parameters of relxill , except for thenormalization. This improves the fit by ∆ χ = for one ad-ditional free parameter, at a significance of > i = + − deg, which is in tension withthe low inclination obtained from radio observations. Within90% confidence, this fit yields very similar parameters as theones with just relxill , and we adopt this version for lateranalyses of the broadband spectrum. We report the best-fitparameters in Table 2.Our fits also seem to show consistent negative residu-als at ∼ keV (see Figures 3 and 4). It is important tonote that this is most likely not physical. When we plotthe background spectrum, we see that this is probably abackground feature, namely the EPIC-pn Cu-K α complex ataround those energies. Regardless of our source region sizeor background region placement (keeping the latter withinthe inner 4 CCDs), the feature remains. The residuals aretherefore likely the result of over-subtraction of the EPIC-pnbackground features. We verify this by plotting the spectrumwithout the background subtracted, and find that the fea-ture is not present (see Appendix A). We also do not seethe negative residuals in any of the individual EPIC-MOSspectra. Fitting a power law to the data in the 2-79 keV range andextrapolating the fit to lower energies also reveals a softexcess below 2 keV. We begin our analysis in the broadbandby including a phenomenological disk blackbody componentto help fit the spectrum at soft energies. To account for thepossibility of intrinsic absorption, we use ztbabs . We alsokeep the neutral, distant reflector from the previous section.We first attempt a Newtonian emissivity profile forthe tbabs*ztbabs*(diskbb + xillver + relxill) model.This gives a fit with χ ν = . . Freeing the emissivity indeximproves the fit by ∆ χ = for one additional free parame-ter at a significance > χ / d . o . f . = / = . . Increasing residuals at energies (cid:38) keV (see Figure4) remain, so we include an extra power law to account fora possible contribution of Compton upscattering further outin the jet, similar to Paliya et al. 2019. An additional hardpower law improves the fit by ∆ χ = for two additionalfree parameters with a significance > Γ = . + . − . forthe coronal emission models and Γ hard = . + . − . for the hardpower law, which is consistent with values found in γ -NLS1s(see e.g. Paliya et al. 2019 and the Γ histograms for theirdouble power-law fit, as well as photon index values in Ojhaet al. 2020). It also gives a high inclination of i = + − deg,as before. Moreover, we obtain a relatively flat emissivityindex q = . ± . , suggesting not only that strong generalrelativistic effects are not required to accurately describe theillumination pattern of the disk, but also that more reflectedflux is emitted from the outer regions of the disk. The factthat the observed flux seems to drop slowly at large radii MNRAS000 , 1–11 (2020)
Mundo et al.
Parameter xillver + relxill diskbb + xillver + relxill + po N H , Gal ( cm − ) * 0.126 0.126 N H , intrinsic ( cm − ) ... < . K diskbb ... + − K dist . reflection ( − ) . + . − . . ± . K rel . reflection ( − ) . + . − . . + . − . z * 0.06 0.06 T in (keV) ... . ± . Γ . ± .
01 2 . + . − . Γ hard ... . + . − . E cut / kT (keV) >
453 300 A Fe ± + − log ξ (erg cm s − ) < . < R refl . ± . . + . − . i (deg) + − + − R in ( ISCO ) * 1 1 R out ( R g ) * 400 400 R break ( R g ) *
15 15 q . ± . a * .
998 0 . C A . ± .
01 1 . ± . C B . ± .
01 1 . ± . χ /d.o.f. 2354/2379 2693/2679 χ ν Table 2.
90% confidence level parameters for the 2-79 keV and0.5-79 keV ranges, reported with respect to EPIC-pn. K ’s are nor-malizations, C A and C B are the calibration constants for FPMAand FPMB, respectively, and q is the emissivity index. Γ hard is thephoton index of the hard power law, T in the temperature at theinner disk radius, and E cut the cutoff energy in the rest frame ofthe source. R refl and ξ are the reflection fraction and ionizationparameter of the relativistic reflection component, respectively. N H , Gal , R in , z , R break , a and R out were frozen to the correspond-ing values (parameters with asterisk). In addition, the ionizationparameter in the xillver distant reflector model was frozen to 0. could be an indication that the razor-thin disk assumptionis invalid.We also attempt to model the data with a more physi-cally motivated approach. Following the possibility that thesoft excess could arise from reflection off of a high-densitydisk (e.g., Garc´ıa et al. 2016; Jiang et al. 2018; Tomsicket al. 2018), we try to fit the soft excess with the extendedreflection model relxillD . Given that this source is at a lowgalactic latitude, it is likely that there is a fair amount ofuncertainty in the column density, so we allow tbabs to be afree parameter, and remove ztbabs . We also include an addi-tional hard power law to compare directly with our diskbb fit. This tbabs*(xillver + relxillD + po) model yieldsan acceptable fit with χ / d . o . f . = / and an electrondensity log n e = . + . − . , but it requires an iron abundanceof > and a high emissivity index q = + − , which suggests amuch higher degree of relativistic smearing than is expectedgiven the equivalent width of our iron line. As in our pre-vious model, it also requires a very high inclination, with i = + − .For completeness, we also fit the broadband spec-trum with an alternative, ionized partial covering model in-stead of relativistic reflection, replacing relxill with zx-ipcf*powerlaw in our best fit, in order to see if we actu- soft powerlawneutral reflectordiskbbhard powerlawrelativistic reflection k e v ( P ho t on s c m - s - k e v - ) r a ti o Energy (keV)
Figure 5.
Unfolded spectrum and best-fit model. Model compo-nents are also shown. ally require a broad iron line. We also apply zxipcf to thedisk blackbody component and remove the hard power lawcomponent for simplicity. This model, which instead fits thebroad line region with a combination of an absorbed powerlaw and a narrow reflection component, gives an acceptablefit with χ / d . o . f = / = . . It requires a coveringfraction of . + . − . , as well as an outflow velocity of nearly ∼ . c , which is probably needed in order to account for theblueshift of the iron K line. However, the physical processesrepresented by this model would manifest themselves in thevariability as shot noise, or additive, variability processes,but we instead see a linear RMS-flux relationship represent-ing multiplicative variability processes. In our best-fit model, we find an unphysically high disk incli-nation. We confirm this result by generating the statistic sur-face for this parameter, which gives comparable values (seeFigure 6). This is much higher than predicted by superlu-minal motion in radio observations (Fuhrmann et al. 2016),which implies a relativistic jet at an inclination i = − degfrom the line of sight. We therefore attempt to further con-strain the inclination angle by placing hard limits at thesevalues. However, we are unable to constrain the inclinationthis way, as the parameter is pegged at the hard upper limit,so we proceed to fix the inclination to a value of 10 deg. Thisscenario provides a fit with χ ν = . , and we again find thatour best fit with an unrestricted inclination is an improve-ment at a significance of > . ( ∆ χ = for 1 addi-tional free parameter). Therefore, our data suggests that,although we do observe a broad iron line, the broadeningmay be caused almost exclusively by line-of-sight Dopplereffects that result from a high inclination. However, this isat odds with superluminal motion inferred from radio ob-servations. Alternatively, the geometry is not a simple ge-ometrically thin disk, as assumed by the blurred reflectionmodel. MNRAS , 1–11 (2020) -Ray Spectral Analysis of 1H 0323+342 χ Inclination (deg)
10 20 30 40 50 60 70 80
Figure 6. χ statistic vs. disk inclination for our best fit. Thered horizontal line is the 90% confidence level. The blue hashedregion depicts the range set by radio observations (4-13 deg). In addition to the spectral analysis described in the previoussection, we also performed a timing analysis to compare totiming analyses from RQ NLS1s. X-ray reverberation lagshave been used to probe the inner accretion disk region as ameans to gain insight into the origin of the X-ray emissionin NLS1s (e.g., Fabian et al. 2009; Uttley et al. 2014). Theseso called “soft lags” are caused by a reflection-induced softexcess that lags behind coronal power-law photons due to theshort, extra time the latter take to travel from the corona tothe inner disk. The detection of a soft lag therefore providesevidence that the soft excess is due to relativistic reflection;on the other hand, a non-detection of a soft lag could suggestthat the soft excess is possibly a different component that isnot simply reflection.In the next subsections, we look into whether we candetect soft lags in 1H 0323+342, and analyze the relationshipbetween variations in the soft band and those in the hardband by using Fourier techniques outlined in works such asNowak et al. (1999) and Uttley et al. (2014).
We begin the second part of our timing analysis by usingthe available 7 light curves to compute time lags betweenthe power law dominated hard band (1-4 keV) and the softexcess (0.3-1 keV) as a function of temporal frequency, orinverse timescale. If the light curve in, say, the soft band, s , has N time bins of width ∆ t , then its discrete Fouriertransform at each Fourier frequency f n = n /( N ∆ t ) is S = N − (cid:213) k = s k e π ink / N (1)We focus on frequencies higher than × − Hz to avoid rednoise leakage at lower frequencies. We require at least 10frequencies per bin.We can rewrite the Fourier transform in a complex polarform as S = | S | e i φ s . Repeating the process for the hard bandlight curve h , we can write the complex conjugate of its Fourier transform as H ∗ = | H | e − i φ h . The product of S and H ∗ gives us the Fourier cross-spectrum between the two bands: C = | H || S | e i ( φ s − φ h ) (2)This gives the phase difference between the soft and hardbands, and the average lag between the two bands in eachfrequency bin is then obtained by taking the phase of theaveraged cross-spectrum, or φ = arg [(cid:104) C (cid:105)] . This can then bemanipulated to give the time lag at each frequency bin: τ = φ π f (3)Figure 7 shows the resulting lag-frequency spectrum be-tween lightcurves in the 0.3-1 keV and 1-4 keV bands thathad 10 s bins. To decide our upper limit in frequency, wecalculate the frequency range where we expect to see rever-beration, given the black hole mass of this AGN, by usingthe correlation between soft lag frequencies and black holemass from De Marco et al. (2013). Landt et al. (2017) ob-tain a mean mass of . + . − . × M (cid:12) through estimates basedon the ionizing 5100 ˚A continuum luminosity and the widthof the hydrogen broad emission lines. We use this mass inthe correlation in De Marco et al. (2013), and find that thiscorresponds to a frequency range ν lag = ( . ± . ) × − Hz where we can expect to find a soft lag. As a precaution,we extend the frequency range to × − Hz , to account foruncertainties in the mass estimate and the frequency-masscorrelation.While our spectrum shows a low-frequency hard lag thatis consistent with propagating fluctuations in the disk, ourspectrum shows no soft lags at shorter timescales, whichcould be related to a lack of coherence between the soft andhard bands, something we investigate in the next subsection.This non-detection may also in fact support our spectralresults by suggesting that reflection off of the accretion diskis not the dominant process causing the soft excess. In an attempt to further explain the non-detection of softlags, we proceed to calculate the coherence between the softand hard light curves. The coherence provides us with a wayof measuring how correlated two light curves or signals are.Essentially, it tells us to what extent one light curve can bepredicted from the other. The coherence at each frequencyis defined as γ = |(cid:104) C (cid:105)| − n (cid:104) P s (cid:105)(cid:104) P h (cid:105) (4)where the n term is due to Poisson noise contributing tothe square of the cross-spectrum, and the terms in the de-nominator are noise-subtracted power spectra. For unity co-herence, the two light curves would be perfectly coherent,meaning one would be able to predict a light curve from theother through a linear transformation.Figure 7 shows the coherence-frequency spectrum thatresults from our data. In the frequency range where we ex-pect to see reverberation, given this black hole mass, we cansee that for the most part the coherence is below 0.6 and con-sistent with 0. The non-unity coherence at these frequencies MNRAS000
We begin the second part of our timing analysis by usingthe available 7 light curves to compute time lags betweenthe power law dominated hard band (1-4 keV) and the softexcess (0.3-1 keV) as a function of temporal frequency, orinverse timescale. If the light curve in, say, the soft band, s , has N time bins of width ∆ t , then its discrete Fouriertransform at each Fourier frequency f n = n /( N ∆ t ) is S = N − (cid:213) k = s k e π ink / N (1)We focus on frequencies higher than × − Hz to avoid rednoise leakage at lower frequencies. We require at least 10frequencies per bin.We can rewrite the Fourier transform in a complex polarform as S = | S | e i φ s . Repeating the process for the hard bandlight curve h , we can write the complex conjugate of its Fourier transform as H ∗ = | H | e − i φ h . The product of S and H ∗ gives us the Fourier cross-spectrum between the two bands: C = | H || S | e i ( φ s − φ h ) (2)This gives the phase difference between the soft and hardbands, and the average lag between the two bands in eachfrequency bin is then obtained by taking the phase of theaveraged cross-spectrum, or φ = arg [(cid:104) C (cid:105)] . This can then bemanipulated to give the time lag at each frequency bin: τ = φ π f (3)Figure 7 shows the resulting lag-frequency spectrum be-tween lightcurves in the 0.3-1 keV and 1-4 keV bands thathad 10 s bins. To decide our upper limit in frequency, wecalculate the frequency range where we expect to see rever-beration, given the black hole mass of this AGN, by usingthe correlation between soft lag frequencies and black holemass from De Marco et al. (2013). Landt et al. (2017) ob-tain a mean mass of . + . − . × M (cid:12) through estimates basedon the ionizing 5100 ˚A continuum luminosity and the widthof the hydrogen broad emission lines. We use this mass inthe correlation in De Marco et al. (2013), and find that thiscorresponds to a frequency range ν lag = ( . ± . ) × − Hz where we can expect to find a soft lag. As a precaution,we extend the frequency range to × − Hz , to account foruncertainties in the mass estimate and the frequency-masscorrelation.While our spectrum shows a low-frequency hard lag thatis consistent with propagating fluctuations in the disk, ourspectrum shows no soft lags at shorter timescales, whichcould be related to a lack of coherence between the soft andhard bands, something we investigate in the next subsection.This non-detection may also in fact support our spectralresults by suggesting that reflection off of the accretion diskis not the dominant process causing the soft excess. In an attempt to further explain the non-detection of softlags, we proceed to calculate the coherence between the softand hard light curves. The coherence provides us with a wayof measuring how correlated two light curves or signals are.Essentially, it tells us to what extent one light curve can bepredicted from the other. The coherence at each frequencyis defined as γ = |(cid:104) C (cid:105)| − n (cid:104) P s (cid:105)(cid:104) P h (cid:105) (4)where the n term is due to Poisson noise contributing tothe square of the cross-spectrum, and the terms in the de-nominator are noise-subtracted power spectra. For unity co-herence, the two light curves would be perfectly coherent,meaning one would be able to predict a light curve from theother through a linear transformation.Figure 7 shows the coherence-frequency spectrum thatresults from our data. In the frequency range where we ex-pect to see reverberation, given this black hole mass, we cansee that for the most part the coherence is below 0.6 and con-sistent with 0. The non-unity coherence at these frequencies MNRAS000 , 1–11 (2020)
Mundo et al. L a g ( s ) −25007501000 C oh e r e n ce −4 −3 Figure 7.
Time lag and coherence between variations in the soft(0.1-3 keV) and hard (1-4 keV) bands as a function of timescale(temporal frequency). The dashed red line is the frequency abovewhich we would have expected reverberation. Our results yield asoft lag of zero, possibly suggesting the soft excess is not neces-sarily due to reflection, and a relatively low coherence, suggestingthat the soft excess may not be directly correlated with the hardband. shows that there is a non-linearly correlated component inthe soft band. A coherence less than unity may have to dowith an additional soft excess continuum component that isvariable in a way that is not correlated with the power-lawand reflection components.
Previous studies of 1H 0323+342 have had mixed approacheswhen it comes to the modeling of the source’s X-ray spectra.By using a simple comparison in count rates between the soft(0.3-2 keV) and hard (2-10 keV) bands of
Swift -XRT obser-vations, Paliya et al. (2014) showed that 1H 0323+342 mightexhibit a strong soft excess whose variability was not per-fectly correlated with the variability of the hard band, whichmay suggest that at least two spectral components are re-quired to fit the X-ray spectra of this source. They success-fully fit the spectrum using an ionized reflection model, andfound that the fit improved when they took relativistic blur-ring into account. However, they fix the inclination to deg.This resulted in a steep spectrum ( Γ = . ± . ) and a highspin ( a = . ± . ). Ghosh et al. (2018) also assumed thatthe soft excess was due to reflection and arrived at similarresults. However, other studies such as Paliya et al. (2019)and Kynoch et al. (2018) have managed to obtain good fitsby modeling the soft excess with a power law, which adds a layer of ambiguity regarding the best description of theexcess at low energies.In each of these cases (as well as in Walton et al. 2013,Yao et al. 2015), there was some evidence of a potentialbroad Fe K α line, but there were a few aspects that wereunclear, such as a stark difference in the measured blackhole spin between Walton et al. (2013) and Yao et al. (2015)(although the latter froze their inclination to a much lowervalue), and the fact that the emission was relatively weakto begin with. In addition, Paliya et al. (2019) and Kynochet al. (2018) found that a combination of narrow emissionlines was adequate enough to fit the residuals at ∼ keV.Therefore, as for the soft excess, the nature of these residualswas up for debate.We find through our spectral analysis that reflectionmay play a role in both the excess emission in the soft bandand the residuals at higher energies ( ∼ relxill ) and a phenomeno-logical blackbody model, along with a distant reflector, re-sults in a good fit to most of the broadband spectrum, andwe detect an iron line at . ± . keV thanks to the in-creased signal-to-noise ratio that results from combining 6simultaneous XMM-Newton and
NuSTAR observations witharchival data. However, a discussion of the origin of theblackbody-like component is warranted, and whether it isrelated to the jet or is due to Comptonization is still un-clear.A high density reflection model can also describe thedata fairly well, but certain well-known issues arise, suchas the fact that the featureless soft excess requires strongerbroadening than the fit to the iron line alone. An alterna-tive ionized partial covering model also works; however, thevariability imprinted by a partial covering model would notpredict the linear RMS-flux relation we obtain, which sug-gests that the variability of this source is multiplicative innature. This further suggests that the X-ray emission origi-nates in the central region of the AGN.While we detect a hard low-frequency lag that suggestspropagating disk fluctuations, we do not detect a soft lag.RQ AGN with similar variance, exposure, and flux do showsoft lags, which might suggest that this is not a signal-to-noise issue. This might have to do with the presence of anadditional soft excess component that is not simply blurredreflection and that displays variability that is uncorrelatedwith the continuum-dominated hard band.The final piece of the spectral puzzle involves the pos-sible hard excess observed at energies above ∼ keV. Wefind that, in each of our models, the residuals seem to in-crease at higher energies. None of our reflection models wereable to take care of this, which implies that this excess maybe due to non-thermal processes related to the jet. Theotherwise coronal-dominated X-ray spectrum suggests theemission might be contaminated by a combination of syn-chrotron self-Compton mechanisms and Comptonization ofexternal disk photons. This hard excess was also observedin Ghosh et al. (2018). Paliya et al. (2014) and Paliya et al.(2019) also detect a hard excess and manage to fit it witha broken power law and a second power law, respectively,and they attribute this to a combination of thermal coro-nal emission and jet emission with a power-law shape, the MNRAS , 1–11 (2020) -Ray Spectral Analysis of 1H 0323+342 latter of which would result from the AGN undergoing γ -ray flaring. In other words, the same process that producesthe γ -ray emission (such as synchrotron self-Compton or ex-ternal Compton) may be the dominant form of hard X-rayemission above 35 keV. The fact that an additional hardpower law component helps fit the hard excess we observeat a significance of > . seems to suggest a similar be-havior. Fuhrmann et al. (2016) constrain the viewing angle of 1H0323+342 to values between 4 and 13 degrees using obser-vations in the radio band. In X-ray spectra, the inclinationcan be independently determined by the shape of the Fe K α line in the reflection spectrum: if the line is mostly broad andblue-shifted, line of sight Doppler boosting (i.e. larger line-of-sight velocities) has the dominant influence on the lineprofile, and therefore this would correspond to a highly in-clined disk; on the other hand, if the profile is mostly skewedtowards lower energies, it would mean that gravitational andtime dilation redshifts are dominant, corresponding to a diskthat is almost face-on.Our results in both the 2-79 keV and the broadbandranges with relxill provide an inclination that is higherthan the upper limit obtained from radio observations (seeTable 2). Our detection of the iron line also shows that theline is blue-shifted. Walton et al. (2013) had modeled Suzaku observations with relativistic reflection and found that ifthey let the inclination vary, they end up with a highly blue-shifted line that peaks at ∼ keV, and obtain an unphysicalvalue of i = ± deg. Therefore, high inclinations have beensuggested in the past; however, as previously mentioned, theiron line had only been marginally detected. In this work,we finally have a constraint on the iron line profile, and westill measure a high inclination. Given the latter, it seemsthat the broadening of the line can be explained almost ex-clusively through line-of-sight Doppler effects, but this is alot more Doppler boosting than is physically allowed by theinclination measured in the radio observations. This, alongwith the unusually flat emissivity profile from our best fit,seems to suggest that we may have to reconsider our modelassumptions.One way to account for a perceived high inclination isto consider the possibility that the emission is being viewedthrough the acceleration zone of the jet. In this scenario,a highly energetic corona would end up inverse Comptonscattering the reflection spectrum. Studies have shown thatconsidering these processes provides a more complete pictureof the emission from both black hole transients and AGN,with the logic being that if thermal seed photons from thedisk are upscattered by the corona, reprocessed reflectionphotons that emerge in the inner regions of the disk shouldalso be influenced by the same process (e.g., Wilkins & Gallo2015; Steiner et al. 2017). This process could cause the ironline in 1H 0323+342 to be blueshifted, even if the disk isface-on and the jet is closely aligned to the line of sight.Another way to explain our measurements may involveabandoning the thin disk geometry (Shakura & Sunyaev1973) altogether and considering a model where the accre-tion disk is geometrically thick, a scenario which has also been suggested to explain similar inclination mismatches inX-ray binaries (e.g., Connors et al. 2019). In the past fewyears, simulations have been used to understand the un-derlying physical processes behind sources that have veryhigh accretion rates (e.g., McKinney et al. 2015; S¸adowski& Narayan 2015). All of these simulations lead to an accre-tion disk that is both optically and geometrically thick. Thishighly accreting, geometrically thick scenario could lead tooptically thick outflows that restrict the emitted radiationto a funnel-like space. A jet from near the black hole couldthen be able to clear out some of the inner flow and couldpotentially give us a line of sight towards the central X-rayemitting region (McKinney et al. 2015).If 1H 0323+342 were to have a relatively high accretionrate, its accretion disk would not be infinitesimally thin.Since radio observations point to a low-inclination geome-try, the blue-shifted iron line we observe could be caused byradiation-driven outflows in the funnel, and since the pres-ence of the jet would be exposing the central region, grav-itational redshifts could also be used to explain the broad-ening of the line. Kara et al. (2016) develop a simple modelfor iron line reverberation in a funnel geometry where spe-cial relativistic effects and gravitational redshift are takeninto account, and where only an outflow velocity profile forthe funnel walls is considered. Through Monte Carlo simula-tions, they show that an iron line can indeed be blueshiftedand broadened due to reflection off an optically thick out-flow.A very recent study by Thomsen et al. (2019) makesuse of this funnel geometry to make a detailed comparisonbetween the iron line profiles of thin disks and those of a ge-ometrically thick supper-Eddington disk. In particular, theyshow that, in each type of disk, photons emitted from dif-ferent radii have different contributions to the line profile.They show that for thin disks, most of the flux comes fromradii less than R g , since the emissivity profile drops rel-atively quickly at larger radii. However, for a funnel geom-etry, more photons are emitted between − R g becausethe reflected flux does not decrease as quickly at large radii,due to the fact that in this geometry, the emission from thecorona would be able to illuminate more of the disk at largerradii than it would for a thin disk. In addition, they showthat the iron line profiles of the thick disk are significantlymore blue-shifted than those of the thin disk, for reasonsmentioned earlier.We can test if 1H 0323+342 would be an adequatecandidate for a thick-disk model by performing a quickcalculation to estimate its Eddington rate using our data.We use the 2-10 keV luminosity from our best fit modelto estimate the latter. After performing this calculation,we arrive at a value of log( L −
10 keV )=43.8. Adjusting theappropriate value in ergs s − by a bolometric correctionof ∼
20 (Vasudevan & Fabian 2007) to get the bolomet-ric luminosity L = η (cid:219) Mc , we then divide by the Edding-ton luminosity. As in our timing analysis, we use the mass M BH = . + . − . × M (cid:12) obtained by Landt et al. (2017).Our result is then L ∼ . + . − . L Edd . The values we calculatehere are comparable to the luminosity and Eddington ratiovalues obtained in Kynoch et al. (2018) and several otherprevious works (e.g., Paliya et al. 2019, 2014; Pan et al.2018; Landt et al. 2017). Although this ratio does not im-
MNRAS000
MNRAS000 , 1–11 (2020) Mundo et al. ply super-Eddington accretion, it does represent a relativelyhigh accretion rate.It is important to note that the bolometric correctionwe use is really an average obtained for Eddington ratios (cid:46) ∼ . . Our results seem to suggest that these are the physicalphenomena that are relevant for the case of 1H 0323+342. We have presented X-ray spectral and timing analyses of the γ -NLS1 1H 0323+342 with a combination of the first simul-taneous XMM-Newton and
NuSTAR campaign and archivaldata. Our main results are as follows:(i) We definitively measure a broad iron line and a Comp-ton hump in 1H 0323+342. The broad line extends from ∼ . − keV, meaning that there are significantly blueshiftedand redshifted components of the line.(ii) We attempt describing the data with a variety of ab-sorption and reflection based models. We find that the X-rayemission in the 0.5-79 keV range (soft excess and iron line)is well described by a combination of a phenomenologicalblackbody component and relativistic reflection model, sug-gesting that the dominant form of the X-ray emission comesfrom the corona in the vicinity of the black hole.(iii) We find potential evidence of a hard excess at highenergies, suggesting that non-thermal jet emission con-tributes yet another component to the hard X-ray spectrum.(iv) Our measurement of the inclination using a razor-thin disk model is definitively in tension with radio obser-vations at a significance of > ACKNOWLEDGEMENTS
SM and EK acknowledge support from NASA grant80NSSC20K1085.
REFERENCES
Abdo A. A., et al., 2009, ApJ, 707, L142Alston W. N., 2019, MNRAS, 485, 260Arnaud K. A., 1996, in Jacoby G. H., Barnes J., eds, AstronomicalSociety of the Pacific Conference Series Vol. 101, AstronomicalData Analysis Software and Systems V. p. 17Begelman M. C., Blandford R. D., Rees M. J., 1984, Reviews ofModern Physics, 56, 255Bhattacharyya S., Bhatt H., Bhatt N., Singh K. K., 2014, MN-RAS, 440, 106 B(cid:32)la˙zejowski M., Sikora M., Moderski R., Madejski G. M., 2000,ApJ, 545, 107Connors R. M. T., et al., 2019, ApJ, 882, 179Crummy J., Fabian A. C., Gallo L., Ross R. R., 2006, MNRAS,365, 1067D’Ammando F., et al., 2013, MNRAS, 436, 191Dauser T., Garc´ıa J., Walton D. J., Eikmann W., Kallman T.,McClintock J., Wilms J., 2016, A&A, 590, A76De Marco B., Ponti G., Cappi M., Dadina M., Uttley P., CackettE. M., Fabian A. C., Miniutti G., 2013, MNRAS, 431, 2441Deo R. P., Crenshaw D. M., Kraemer S. B., 2006, AJ, 132, 321Fabian A. C., Rees M. J., Stella L., White N. E., 1989, MNRAS,238, 729Fabian A. C., et al., 2009, Nature, 459, 540Fuhrmann L., et al., 2016, Research in Astronomy and Astro-physics, 16, 176Gallo L. C., Tanaka Y., Boller T., Fabian A. C., Vaughan S.,Brandt W. N., 2004, MNRAS, 353, 1064Garc´ıa J. A., Fabian A. C., Kallman T. R., Dauser T., ParkerM. L., McClintock J. E., Steiner J. F., Wilms J., 2016, MN-RAS, 462, 751Ghisellini G., Maraschi L., Tavecchio F., 2009, MNRAS, 396, L105Ghosh R., Dewangan G. C., Mallick L., Raychaudhuri B., 2018,MNRAS, 479, 2464Gierli´nski M., Done C., 2004, MNRAS, 349, L7Grupe D., Mathur S., 2004, ApJ, 606, L41Jiang J., et al., 2018, MNRAS, 477, 3711Jiang J., et al., 2019, MNRAS, 489, 3436Kalberla P. M. W., Burton W. B., Hartmann D., Arnal E. M.,Bajaja E., Morras R., P¨oppel W. G. L., 2005, A&A, 440, 775Kara E., Miller J. M., Reynolds C., Dai L., 2016, Nature, 535,388Kara E., Garc´ıa J. A., Lohfink A., Fabian A. C., Reynolds C. S.,Tombesi F., Wilkins D. R., 2017, MNRAS, 468, 3489Kynoch D., et al., 2018, MNRAS, 475, 404Landt H., et al., 2017, MNRAS, 464, 2565Mallick L., et al., 2018, MNRAS, 479, 615Marinucci A., et al., 2014, MNRAS, 440, 2347McKinney J. C., Dai L., Avara M. J., 2015, MNRAS, 454, L6Miyakawa T., Ebisawa K., Inoue H., 2012, PASJ, 64, 140Mizumoto M., Ebisawa K., Sameshima H., 2014, PASJ, 66, 122Nowak M. A., Vaughan B. A., Wilms J., Dove J. B., BegelmanM. C., 1999, ApJ, 510, 874Ojha V., Chand H., Dewangan G. C., Rakshit S., 2020, ApJ, 896,95Osterbrock D. E., Pogge R. W., 1985, ApJ, 297, 166Paliya V. S., Sahayanathan S., Parker M. L., Fabian A. C., StalinC. S., Anjum A., Pandey S. B., 2014, ApJ, 789, 143Paliya V. S., Parker M. L., Jiang J., Fabian A. C., BrennemanL., Ajello M., Hartmann D., 2019, ApJ, 872, 169Pan H.-W., Yuan W., Yao S., Komossa S., Jin C., 2018, ApJ, 866,69Peterson B. M., et al., 2000, ApJ, 542, 161Reynolds C. S., Nowak M. A., 2003, Phys. Rep., 377, 389S¸adowski A., Narayan R., 2015, MNRAS, 453, 3213Shakura N. I., Sunyaev R. A., 1973, A&A, 24, 337Steiner J. F., Garc´ıa J. A., Eikmann W., McClintock J. E., Bren-neman L. W., Dauser T., Fabian A. C., 2017, ApJ, 836, 119Tanaka Y., Boller T., Gallo L., Keil R., Ueda Y., 2004, PASJ, 56,L9Thomsen L. L., Lixin Dai J., Ramirez-Ruiz E., Kara E., ReynoldsC., 2019, ApJ, 884, L21Tomsick J. A., et al., 2018, ApJ, 855, 3Uttley P., McHardy I. M., 2001, MNRAS, 323, L26Uttley P., McHardy I. M., Vaughan S., 2005, MNRAS, 359, 345Uttley P., Cackett E. M., Fabian A. C., Kara E., Wilkins D. R.,2014, A&ARv, 22, 72Vasudevan R. V., Fabian A. C., 2007, MNRAS, 381, 1235MNRAS , 1–11 (2020) -Ray Spectral Analysis of 1H 0323+342 no r m a li ze d c oun t s s - k e V - −4 −3 r a ti o Figure A1.
APPENDIX A: CHECKING FOR APOTENTIAL ABSORPTION FEATUREBETWEEN 8 AND 10 keV
The background spectrum reveals a peak that correspondsto the energies where we observe negative residuals (FigureA1). This is likely the Cu-K α complex in EPIC-pn, and ournegative residuals are therefore the result of over-subtractionof these features. To be more thorough, we perform an ex-ercise to make sure that the negative residuals are not theresult of a physical absorption process. We fit the EPIC-pnspectrum, without the background subtracted, to a powerlaw; in this case, if there is indeed absorption from an out-flow, for instance, the residuals should remain negative, in-dicating that the feature is not just an artifact.We show this spectrum in Figure A2, and we find thatthe residuals are not present, suggesting that backgroundover-subtraction is at fault. We also perform this exercisefor our individual high and low-flux observations, and stillthe negative residuals do not show up. In addition, none ofthe individual MOS spectra show these residuals. This paper has been typeset from a TEX/L A TEX file prepared bythe author. no r m a li ze d c oun t s s - k e V - r a ti o Energy (keV)
Figure A2.000