A X-Ray Spectroscopy Study of LINERs as Observed by XMM-Newton

(1)

Is Every One LINER Dense at Its Core?

A X-Ray Spectroscopy Study of LINERs as Observed

by XMM-Newton

Adam Andrews (920310)

aandrews@kth.se

Department of Physics

Royal Institute of Technology (KTH)

Supervisor: Serena Falocco

(2)

Typeset in LA_TEX

ISRN KTH/FYS/– – 17:68 – – SE ISSN 0280-316X

(3)

Abstract

Active Galactic Nuclei (AGN) emit high luminosity in nearly all wavelength bands; they have a characteristic X-ray spectrum and can be modelled as accreting Super Massive Black Holes (SMBH). Low-Ionization Nuclear Emission-line Regions (LINERs) are an-other type of galactic nuclei. The aim of this project is to study the X-ray spectra of two LINERs (NGC 1052 and NGC 1961) in order to find any evidence of AGN presence. Data from the XMM-Newton is reduced and processed, utilizing xspec to select and fit models applied to the data, in order to test their statistical significance. The results show that NGC 1052 exhibits clear AGN features in the X-ray band, including a power-law, clear Fe-emission line, reflected emission and variability, as well as starburst presence. The NGC 1961 data displays a power-law and starburst emission; the time span of the observations denies us the possibility of testing for variability. The temperature and the photon index of NGC 1052 are consistent with the values in the literature, as well as our findings of variability. In NGC 1961, we find clear evidence for a hot plasma and pri-mary emission, and also the possibility of another plasma structure, which is supported by other studies on the object. Improvements and further investigations are discussed, mainly focused on expanding the data sets and using this project as a stepping stone for further research.

(4)

Chapter 1 Introduction and Theory

1.1 Aim

The aim of this study is to investigate the X-ray spectra of the two LINERs NGC 1052 and NGC 1961. By studying this range in the electromagnetic spectrum, we wish to apply theoretical models to the data to test for AGN presence: power-law emission, soft excess, Compton reflection, broad and narrow iron lines and variability. In addition, we search for any signs of starburst activity and host galaxy emission.

1.2 History of HE Astrophysics and Background

Although astronomy has existed since the dawn of mankind, high-energy astrophysics was not possible until approximately 50 years ago. This is mainly due to the fact that the atmosphere absorbs most X-rays, and in order to observe non-terrestrial high-energy photons, instruments would be required to be placed above the atmosphere. The first mission to successfully detect extrasolar X-rays was the OSO-3 satellite, launched in 1967 by NASA. OSO-3 was able to detect the strong X-ray source of Scorpius X-1 [63]. In order to circumvent the absorption of high-energy photons by the atmosphere, the project demonstrated that new information could be attained by high-altitude detectors [61].

As a result of this expedition, the 1970s witnessed several groundbreaking research projects aimed at observing the X-ray sky of the universe. More specifically, instead of using rockets to lift scientific equipment to a sufficient altitude, satellites were used to carry these X-ray detectors. Projects included Uhuru, Ariel-5, SAS-3, OSO-8, and HEAO-1, which together greatly advanced X-ray astrophysics and lifted the area into mainstream astronomy. It was hypothesized around this time that most of the X-ray sources were neutron star binaries, with “normal” stars accreting onto a neutron star and thus producing X-ray radiation. These systems were dubbed “X-ray binaries”, and the mass of the heavier object could be studied and calculated. A few percent of galactic centra contained X-ray sources as well, which were called Active Galactic Nuclei. It is

(6)

Telescopic ARray (NuSTAR, 2012). As this thesis will use data from the XMM-Newton Science Archive, it will not consider the other projects as much [61].

Figure 1.1: The XMM-Newton (Source: NASA)

1.3 AGNs and the Unified Model

As a result of the study of astrophysical objects in X-ray spectra, new ways to explore the universe emerged. Since X-rays are linked to very hot or energetic processes, mechanisms close to extremely compact objects were viewed in a new light. Also, X-ray radiation allowed physicists to reach farther out into the universe, making never-before-seen objects observable. Furthermore, X-rays have an intrinsic ability to penetrate many materials, allowing us to probe closer to the sources [19]. Typically, the main action behind X-ray production is accretion, thus compact objects can be studied, especially in their inner regions, with X-ray spectroscopy [77].

In addition, these processes allow us to understand specifically one type of object, namely Active Galactic Nuclei (AGNs). Usually, AGNs are said to consist of a galaxy core containing a SMBH (> 105

M) with the constraint on the Eddington ratio LAGN

LEdd = 0.5,

where LAGN is the bolometric luminosity. Moreover, AGNs usually include a mixture of several structure components. These include an accretion disk, a hot cloud of plasma surrounding the accretion disk (known as the hot corona), high velocity gas clouds known as the Broad-Line Region (BLR), a dusty torus, lower velocity gas clouds known as the Narrow-Line Region (NLR), as well as a central jet. Figure 1.2 shows an AGN and its system consisting of the aforementioned components [26].

(7)

Figure 1.2: The structure of an AGN (Source: NASA)

There is no single definitive observational feature of AGN, but rather a mixture of traits. These include radio continuum emission, optical continuum and line emissions (narrow and broad), X-ray continuum and X-ray line emissions. The continuum emissions are typically emitted from the accretion disk, the hot corona and the jet. We will not further develop the non-X-ray features of AGNs in this report, but rather refer to Krolik and his summary [77]. Due to these general traits of the spectrum, a plethora of categories have emerged to handle all combinations of observational characteristics available. In broad terms, AGNs can be organized as being Type I, Type II, radio-loud or radio-quiet. A Type I AGN has broad and narrow optical lines while a Type II AGN has only narrow optical lines. Radio-loud implies a high ratio of fluxes in radio to optical emission, while radio-quiet has a larger flux in the optical range than radio. Additionally, a Unified Model has been proposed in order to organize all these variations under one parameter. We will see that all AGNs can be explained as variations in a single variable, namely the orientation of the accretion disk [56] [26] [55].

1.3.1 Type I and Type II

Through the observational evidence of (optical) broad line emissions, we characterize AGNs as either Type I or Type II. When discussing the structure, we designated the BLR and the NLR as Keplerian velocity gas clouds and low velocity gas clouds. Basi-cally, Keplerian velocities broaden the optical emission lines, while the relatively narrow emission lines emanate from the NLR. This is because the BLR is closer to the SMBH, thus experiencing a much stronger gravitational force. Important to realize, the reason we can express this difference via the Unified Model as mere orientation differences, is due to the possibility of the BLR being blocked by the dusty torus. If the torus lies in the line of sight between the observer and the BLR, we cannot possibly detect the broadened optical lines directly. However, in some cases, the broadened optical lines are detected

(8)

versus a Type II AGN. Notice how the Type II clearly lacks the broader components which can be seen in the Type I spectra. An example of a Type I source is NGC 1068 [4]. Furthermore, an example of a Type II AGN is NGC 4388 [14]. The Unified Model has the capacity to describe the difference between Type I and Type II via their inclination in relation to the observer.

Figure 1.3: Type I and Type II AGNs (the Y-axis is relative intensity and the X-axis is wavelength in angstroms) (Source: Extragalactic Astronomy)

1.3.2 Radio Loud and Radio Quiet

Turning our attention to the radio band instead, the same principle is applied here; the objects are unified via their inclination angle. As mentioned, AGNs are characterized as either radio-quiet or radio-loud, i.e. categorized via their relation of the radio emission and optical emission. As mentioned previously, radio-loud AGNs have stronger emission in the radio band than compared to the optical band, while radio-quiet has stronger flux in the optical spectrum than the flux in the radio spectrum. For radio-quiets, the optical band is used to separate the subcategories; they either have both broad and narrow optical emission lines, or only narrow optical emission lines (sometimes together with polarized broad optical emission lines), as explained earlier. On the other hand, radio-loud AGNs typically have prominent jet activity. There are generally two main jet profiles. In the first case, the spectrum is said to have a core-dominated spectrum, i.e. a flat spectrum. On the other hand, if the core is covered, most likely by the dusty torus, the emission is said to be lobe-dominated, having more steep-like traits to it, i.e. a decreasing in intensity faster than the flat spectrum. The difference can, again, be explained via the inclination angle: the dusty torus can cover the jet (lobe-dominated), the core can be observed (core-dominated) or non-observable or non-existent jet (radio-quiet). For the radio band, we still apply narrow and broad lines, but also introduce differences based on the relative strength of the optical and radio band [56] [55].

There are plenty of AGNs studied in the radio band. Radio-loud objects traditionally studied are high-luminosity quasars, but a great deal of variations do exist. A more extended list of objects and descriptions of them can be found in the review made by V´eron-Cetty and V´eron [56] or Krolik [77]. Considering such a large range of

(9)

AGN-like objects, it is truly quite helpful to have only one parameter to unify the categories, namely, the inclination of the AGN.

1.4 X-Ray Spectra of AGNs

There are several advantages of and reasons for investigating AGNs via their X-ray spec-tra. First and foremost, the energy of the X-ray radiation constitutes a large portion of the total radiated energy. Furthermore, high-energy X-rays have penetrative capability, which allows us not only to probe the AGN through its dusty barriers, but also has the ability to reveal the closest regions of the SMBH [52]. A typical X-ray spectrum contains primarily four distinct components: The power-law feature, the Compton reflection (or reflection hump), the soft excess and the iron line. A typical AGN X-ray spectrum, with components labeled, can be found in Figure 1.4. Also, a more detailed diagram of how the primary and secondary emission is generated can be found in Figure 1.5. For this thesis, we only investigate the astrophysical objects via their X-ray spectrum. As can be seen, the X-ray spectrum of an AGN is the result of its complex structure; while some of the primary emission from the accretion process can be detected directly, some of it is also reflected against other parts of the system [19] [77].

Figure 1.4: A Typical AGN X-Ray spectrum (Source: Ricci et al. 2011)

The power-law component of the X-ray spectra of AGNs is the most dominant feature, sometimes called the primary emission. It is described as the following: N (E) = cE−Γ, where N (E) is the number of photons per unit area per unit energy per unit time that the observer detects, c is the normalization coefficient and E is the energy (or frequency) of the photons. Γ is called the photon index. Usually, AGNs have a photon index

(10)

Figure 1.5: Diagram of the regions of primary and secondary emission in the AGN core (Source: Fabian et al. 2014 [16])

the spectrum varies: The short time scale and the long time scale. The short time means roughly less than one day, while the long time indicates a span of couple of months. These fast-changing spectra are evidence that the source of the variability in the power-law component is a small-structure feature, most probably emanating from the hot corona (through scattering) or the accretion disk itself. X-rays allow observers to study properties of the SMBH via the X-ray and power-law spectrum [27].

Two other aspects of the X-ray spectrum of AGNs are the Compton reflection and the soft excess [32]. The Compton reflection, or the secondary emission, is a result of the relatively cold material existing close to the SMBH. This can be either the accretion disk or the torus, sometimes even both. In this case, the material has high column density (which is defined as the amount of material per unit area along the observation path, notated as NH) which scatters and reflects the seed photons, i.e. the photons emanating from the central structure of the AGN. The scattering angles of the photons depend strongly on energy, and so the reflected radiation will also depend on the initial energy of the output stream. While the Compton hump appears at higher energies (around 30 keV), the soft excess is a hard-to-explain feature most important below 1 keV, sometimes dominating the total X-ray spectrum of the AGN. The soft excess is a continous form of thermal emission, emanating from a hot, diffuse gas. Surveys performed by Exosat revealed the existence of soft-excess in around 50% of all Type I AGNs [5], and are often shown to be quite variable over time. While this is currently an intensely debated area of research, there are various explanations for the soft excess. Done et al. (2012) explains the source of the soft excess as Compton up-scattering by warm and optically thick material, most probably the accretion disk itself [12] [5]. Another interpretation of the soft excess are soft X-rays created in the NLR and emitted as thermal emission [8]. The soft excess and the Compton reflection are two distinguishable aspects of the X-ray spectrum which further complicate the spectrum of AGNs [77].

While we have covered some of the continuum emission of the X-ray spectra, the most frequently observed line of AGNs is the iron line. Iron lines exist in the energy range of 5.5-7 keV, with narrow iron lines existing in the 6.2-6.6 keV range and broadened iron lines in the 6.0-6.8 keV range. [19]. Detected by the Tenma satellite (launched in

(11)

1983) [37], the line exists at 6.4 keV, with an equivalent width of 100-200 ev. Besides the Fe-line at 6.4 keV (non-ionized), there are additional lines at 6.7 keV (almost fully ionized) and 6.9 keV (fully ionized). This very distinct line is explained as the result of X-ray photons undergoing photoelectric absorption followed by fluorescence in cold material surrounding the SMBH, e.g. the dusty torus or inner disk [15] [19]. However, in some cases the iron lines are broadened; this is believed to be due to relativistic effects of the SMBH. As the material in the accretion disk travels at high speeds, Doppler broadening occurs, which can affect the iron line. Furthermore, the gravitational pull of the SMBH can have relativistic effects on the iron line as well [15]. These relativistically broadened lines emerge from the inner region, i.e. the accretion disk, while the narrow, un-broadened lines are emitted in the dusty torus. The fluorescent iron line is the strongest emission line in the X-ray band; thus, this extremely common emission line supports the model that all AGNs contain the same underlying system of an accreting SMBH [51] [77].

Additionally, while not an intrinsic part of the AGN structure, the surrounding mat-ter affects the X-ray spectrum. More precisely, the column density (NH) between the observer and the AGN has its effects on the observable traits. There are two main pro-cesses occurring: Photoelectric absorption and Compton scattering. Figure 1.6 displays the effects of various NH-levels on a typical X-ray spectrum. For NH < 1021.5cm−2, photoelectric absorption of the interstellar medium alone obscures broad optical lines (Caccianiga et al. 2007) [7] with negligible continuum suppression, i.e. at energies higher than 2 keV. For NH < 1024cm−2, we have both primary emission and reflection features intact in the spectrum, while the direct emission is more dominant. For NH > 1024cm−2, Compton scattering dominates, and the source is said to be Compton thick. The column density has a very clear effect on the X-ray emissions from AGNs, especially in the range in which we collect our data (0.5-12 keV). Together with the power-law, there are other features which can show variability. While it was mentioned that the power-law compo-nent can show short time scale and long time scale variation, it has been shown that the total flux of the AGN can vary over time as well [45]. With variability studies, it has been shown that the soft excess can express change over time [39]. As the dusty torus rotates around the central core of the AGN, the column density can display variability as well. While the variability is not truly periodic, it does increase and decrease over time. Various forms of variability can be studied in AGNs [30] [48].

(12)

Figure 1.6: The effect of NH-levels on the X-ray spectrum from an AGN (Source: Gilli et al.

(2007)) [23]

In summary, we have discussed several features of an AGN spectrum in the X-ray band. The power-law spectrum has high variability and is most probably a direct obser-vation resulting from the accretion disk powered by the hot corona electrons. Another strong component, the soft excess, which may be even more luminous than the total power-law, is relatively poorly understood in light of the power-law function. Several suitable interpretations exist which could explain this sort of pattern. The reflection pro-cesses in the enveloping material yield the fluorescent narrow iron lines and the Compton hump. The interested reader will be referred to sources found in the bibliography ( [77] and [22] in particular) which give a more descriptive explanation of AGNs in the X-ray spectrum.

1.5 The LINER-model, NGC 1052 and NGC 1961

Another type of galactic nuclei are LINERs (Low-Ionization Nuclear Emission Regions), which are defined (as their names reveal) by their low emission lines in the optical spec-trum of their cores. It is estimated that around 30% of nearby galaxies (distance ranging from 20 to 40 Mpc) can be classified as LINERs [53]. The spectra from LINERs in-clude emission lines from weakly ionized or neutral ions which are present (e.g. O+ or N+), while the emission lines from strongly ionized atoms (e.g. He++ or O++) are absent or very weak. Hence, LINER-candidates span a diverse set of galaxies, with large variations in brightness and morphology. One general quantitative selection criterion of LINERs is the detection of broadened Hα-line in the optical spectrum [58]. LINERs have a broad definition and there are many suitable candidates in close space which satisfy the description [33].

The LINER-system’s mechanism to energize and ionize the emissions has not been agreed upon. For this, there are two main explanations. The first suggests that LINERs are fueled in the same way as are AGNs, with a SMBH at the center accreting from a massive disk [28]. The second proposes that the radiation emission is due to star formation regions, also called starburst activity [54]. For both cases, these mechanisms

(13)

power the LINER and ionize the clouds which thus emit the observed optical lines. Alternatively, there is another model: the advection-dominated accretion flow (ADAF). This form of low-luminosity AGN proposes that the cooling processes are inefficient, and thus the energy is not emitted, but rather absorbed by the SMBH. Key features in the X-ray spectrum for an ADAF AGN is the lower luminosity compared to traditional AGNs (around L = 1040− 1042_ergs−1_{), a flatter Γ (around 1.3 − 1.4) and the absence of narrow} iron lines [58]. However, broad iron lines can be detected in ADAF AGNs [15]. In many cases, LINERs are referred to as AGNs [33], even though their energy and ionization mechanisms are yet to be determined [31].

In general, a starburst galaxy is a galaxy where stars are forming at an extremely high rate, typically as a result of a merger. In a paper from 2001, Persic and Repheli have modelled the X-ray feature of starburst galaxies, and compared the theoretical spectra with observational data. The template spectrum of a starburst galaxy can be seen in Figure 1.7, with the various components described. The mechanisms which contribute to the final spectrum are galactic winds, supernova remnants, Compton emission and X-ray binaries. Note that the spectrum peaks at around 2 keV and without a strong, distinct power-law component. These kind of spectra could show themselves in the spectra of LINERs and should be worth mentioning; however, those interested in a further discussion of starburst galaxies are referred to the work by Persic and Repheli [44].

In this thesis, the two LINERs which have been studied are NGC 1052 and NGC 1961 (New General Catalogue). NGC 1052 is a galactic nucleus of a Type IV elliptical galaxy in the constellation of Cetus (the “Whale”). This LINER is located around 18 Mpc from Earth (z = 0.005), and harbors a double-sided radio jet and an extensive neutral hydrogen disk [36]. These two features suggest that a merging occurred with another galaxy ca. 1 billion years ago [46] as well as past starburst activity. There are also aspects which point towards past or present AGN activity (e.g. high variability, iron lines, soft excess, etc.) [53]. The column density of our galaxy in the direction of NGC 1052 is calculated to be 0.0307 ∗ 1022cm−2 [60]. Moreover, NGC 1961 is a spiral galaxy (SBbc) located in the Camelopardalis (the “Giraffe”) constellation, ca. 55 Mpc from us (z = 0.0131). While NGC 1052 does not have any other galaxies in its close proximity, NGC 1961 is the central galaxy in a group containing nine galaxies, separated at a distance of around 1 Mpc. The shape of NGC 1961 is a central bright nucleus with highly irregular arms, pointing towards an earlier merging, stripping or similar interaction. However, no nearby companion suitable to explain this hypothesis can be observed [64]. A bright X-ray corona can be observed to envelop the galaxy, and plenty of supernovae have been discovered [2]. The galactic column density in the direction of NGC 1961 is calculated to be 0.0834 ∗ 1022cm−2 [60]. Additionally, NGC 1961 has been reported earlier to be a possible AGN candidate [13]. No previous studies of NGC 1961 have reported any jet activity. While both are defined as LINERs via their optical spectrum, NGC 1052 and NGC 1961 have structural (existence of jets, within or without the galaxy cluster) and spectral differences (time-dependent variability in NGC 1052) [53] [27].

(14)

Figure 1.7: A normalized X-ray spectrum of starburst galaxies. In increasing flux at 6 keV, the components are: galactic wind, super-nova remnants, faint low-mass X-ray binaries (XRB), Compton emission, high-mass XRB and bright low-mass XRB (Source: Persic and Rephaeli [44])

In light of our previous discussion of AGNs’ optical and radio spectra, we turn our attention towards the lower energy bands in the spectra of our two LINER sources. Fosbury et al. (1978) [21] reports that NGC 1052 displays Type II AGN qualities, with narrow optical emission lines. Furthermore, the radio spectrum displays a clear flat-like shape and variability, possibly extending up to the infrared. Alonso-Herrero et al. (1997) [1] adds to the discussion by considering evidence of broad Balmer lines in polarized light, and states that NGC 1052 displays prototypical LINER properties, concluding that the probable energy source is the same as of an AGN or an ADAF AGN [35] [9]. Similar to NGC 1052, NGC 1961 also displays clear Type II aspects, with no direct broad optical emission lines [24]. For the radio spectrum, NGC 1961 has peculiar spectral index distribution, displaying both steep-like and flat-like features. Lisenfeld et al. (1999) attributes these effects to high variations in former star formation rates and possibly a violent merging process [38]. Some similar qualities in the lower-energy spectra of the objects can be detected; however, NGC 1052 has been much more extensively studied, possibly due to its clear LINER features.

In like manner, the X-ray spectra of NGC 1052 and NGC 1961 have been studied. NGC 1052 demonstrates a flat spectrum with photon index Γ ∼ 0.2 in the 2.0 − 10keV range. Several reports conclude that NGC 1052 is a low-luminosity AGN with pre-dominately advection-flow accretion [57] [25] [36]. Hern´andez-Garc´ıa et al. (2016) finds NH-variability over time for NGC 1052 [30]. The study of Brenneman et al. (2009) sup-ports earlier findings and adds that a broad and narrow iron line is detected, however, it cannot detect any secondary emission from the center [6]. For NGC 1961, Pence et al. (1997) uses data from the ROSAT satellite to find X-rays from starburst activity and spiral arm regions. The X-ray spectrum is characterized by a narrow peak at 0.9 keV [43]. Andersson et al. (2015) studies the X-ray emission from the gaseous halo which envelopes NGC 1961. No AGN-activity is studied, but the temperature profiles

(15)

of the halo are carefully measured [3]. This thesis will build upon the earlier studies by interpreting the findings more carefully: we will test NGC 1961 for AGN activity and test NGC 1052 for more detailed AGN activity, e.g. secondary emission and variability. While NGC 1052, again, demonstrates clear LINER-qualities with evidence for ADAF AGN, the presence of a SMBH in NGC 1961 has not been determined.

(16)

Chapter 2 Method and Instrumentation

2.1 Instrumentation and Software

The XMM-Newton was launched in December 1999 as a mission to study astronomical ob-jects in X-ray emitting regions, together with optical emitting regions. More specifically, its focus was towards star-forming regions, formation and evolution of galaxy clusters and close-range environments of super-massive black holes. Initially, XMM-Newton was planned to be active for only two years; it is still operating today, having being online for almost 18 years. The satellite holds the EPIC-system, RGS-system and a complex arrangement of mirrors, focusing the X-rays onto the instruments. The RGS-system and the mirror-system are not further elaborated upon here, but we rather refer to the techni-cal specification given in the technitechni-cal description [71]. A highly successful space mission, the XMM-Newton is a cornerstone in studying astrophysical X-rays [70]. In Figure 1.1, a schematic diagram of the XMM-Newton can be seen.

The primary instrument of the XMM-Newton for studying X-ray sources is the Euro-pean Photon Imaging Cameras-system (EPIC), which has also produced the data for this project. The EPIC consists of two MOS-CCDs and one pn-CCD, which together cover a total field of view of about 30 arcminutes of the energy range of 0.15 keV to 15 keV. The MOS-CCDs are intended to perform optimally with low-energy X-rays, ranging from 0.2 keV to 10 keV, while the pn-CCD is designed to measure high-energy X-rays, ranging up to 15 keV. An image of one of the MOS-cameras is given in Figure 2.1, and an image of the pn-camera is given in Figure 2.2. Using a triple-camera system, the EPIC instrument aims to cover a wide ranges of signals over energy and angle [71].

(17)

Figure 2.2: The pn camera (Source: ESA)

In order to analyze the data from XMM-Newton, a combination of the two software programs SAS and xspec is used. SAS is useful in extracting the spectra from the observations, while xspec is used for spectral analysis. For the purpose of this thesis, Ubuntu (version 16.04) is run on a Mac (version 10.11.5) using VirtualBox (version 5.1.20). The installation procedure for SAS (release xmmsas-20170112-1337-16.0.0) and xspec (version 12.9.1) followed the instructions from ESA [69]. The default settings for xspec are used (H0 = 70 kms−1M pc−1, q0 = 0, Λ0 = 0.73). For the error values, the 90% confidence level is chosen, i.e. the error in the parameter is set at the value for which the χ2-statistic is 2.706 greater than the best fit. The CCF used to analyze the data are of the version 20170125T122118. The software needed to study data from XMM-Newton is commonly used, and plenty of support has been developed for these platforms.

Via XMM-Newton Science Archive [72], the observations for NGC 1052 and NGC 1961 can be found and downloaded. Within the time frame of this project, there are four public observations available for NGC 1052 (a fifth public data point exists, but contains no scientific data) and eleven for NGC 1961. Some information about the observations has been summarized in Table 2.1. However, those curious about more details concerning the observations are referred to the data base directly, which is open to public access [73].

(18)

Observation ID Object Right Ascension Declination Obs. Date 0093630101 NGC 1052 02h 41m 04.80s -08d 15’ 21.0” 2001-08-15 0306230101 NGC 1052 02h 41m 04.80s -08d 15’ 20.7” 2006-01-12 0553300301 NGC 1052 02h 41m 04.80s -08d 15’ 20.8” 2009-01-15 0553300401 NGC 1052 02h 41m 04.80s -08d 15’ 20.8” 2009-08-12 0673170101 NGC 1961 05h 42m 04.60s +69d 22’ 42.0” 2011-08-31 0673170301 NGC 1961 05h 42m 04.60s +69d 22’ 42.0” 2011-09-14 0723180101 NGC 1961 05h 42m 04.60s +69d 22’ 42.0” 2013-09-11 0723180501 NGC 1961 05h 41m 07.89s +69d 23’ 02.0” 2014-02-17 0723180901 NGC 1961 05h 43m 01.30s +69d 22’ 22.0” 2014-02-19 0723180401 NGC 1961 05h 42m 04.60s +69d 23’ 02.0” 2014-02-21 0723180801 NGC 1961 05h 42m 04.60s +69d 22’ 22.0” 2014-02-23 0723180601 NGC 1961 05h 41m 07.89s +69d 22’ 42.0” 2014-03-07 0723180201 NGC 1961 05h 43m 01.30s +69d 22’ 42.0” 2014-03-09 0723180701 NGC 1961 05h 41m 07.89s +69d 22’ 22.0” 2014-03-11 0723180301 NGC 1961 05h 43m 01.30s +69d 23’ 02.0” 2014-03-15

Table 2.1: Overview of the observations acquired from the XMM-Newton Science Archive

2.2 Data Reduction Procedure

In order to reduce the data into spectra, the handbook published by ESA was followed primarily. Therefore, this section will mainly provide a brief overview of the procedure, but the interested reader is referred to the threads themselves [75], where more details are given. As for the commands themselves, the SAS manual [74] presents and explains them well.

Firstly, the raw data files were processed with the default configurations of the com-mands used. Secondly, the extraction regions of the source and background are deter-mined. The standard commands and settings as exemplified in the data thread [75] are used. Using these, the images can be opened in the program Ds9. Using Ds9, a spec-trum of the source as well as a background specspec-trum can be collected. For the source spectrum, which is the spectrum from the source to be studied, a circle with radius 15 arcsec is placed on the source. This number is chosen to include only galactic nuclei and exclude diffuse emission from the galaxy. When the region is selected, the coordinates in physical parameters are noted.

Next, the background spectrum is extracted. This is done by selecting a suitable region where no other bright sources exist while acting on the same plate as the source. The radius of 30 arcsec was chosen to include enough noise in order to get a clear profile of the background. Then, the background is subtracted from the source spectrum. Given the coordinates and radii of the extraction zones, the data reduction procedure can move on to extract the requested spectra.

Finally, the finalization of the data is made. Together with the coordinates gath-ered from Ds9, the regions are analyzed and the final spectral extraction is performed, producing scientific files for analysis. In addition to a source spectrum and background spectrum, response files are produced. The data is grouped, with the minimum number of counts per each background subtracted energy bin chosen to be 25, in order to

(19)

over-sample the instrumental resolution by at least a factor of 3. The energy band in which to register photons is selected to be 0.5 keV to 12 keV. Additionally, as the MOS-system has two cameras, the results from these instruments are merged into a single spectrum. This is done to produce a higher quality spectrum on which to perform more exact fits. The data is merged using a Perl-code used by Falocco et al. [17] and developed in that collaboration. In essence, using SAS to perform the spectral extraction is a rather straightforward procedure, given that the data threads provided by ESA are followed more or less.

2.3 Quality of the Data

When reducing the data, not all of the observations from NGC 1961 were suitable to be used; the image could be poor in terms of photon hits or the photons could have landed between the CCDs. In these cases, the data was deemed unscientific and therefore discarded. More images of the plates have been added to the Appendix, and an overview of the observations saved and discarded is given in Table 2.2.

Furthermore, another issue with the data output was the two first observations of NGC 1961. According to the ESA Helpdesk, strong X-ray storms occurred during the observations and in order to save the equipment the protective shutter was closed. This is also described in the Quality Report of the observations in the XMM-Newton Science Archive [72]. Hence, the saved data was deemed untrustworthy, since the background radiation was not controlled.

Observation Observation ID Condition Obs. Date

1 0673170101 Discarded 2011-08-31 2 0673170301 Discarded 2011-09-14 3 0723180101 Accepted 2013-09-11 4 0723180201 Accepted 2014-03-09 5 0723180301 Accepted 2014-03-15 6 0723180401 Discarded 2014-02-21 7 0723180501 Discarded 2014-02-17 8 0723180601 Accepted 2014-03-07 9 0723180701 Discarded 2014-03-11 10 0723180801 Discarded 2014-02-23 11 0723180901 Accepted 2014-02-19

Table 2.2: Summary of accepted and discarded observations for NGC 1961

Using xspec, we calculate the exposure time and the total counts of the observations. These values are summarized in Table 2.3.

(20)

Observation ID Object Total Counts 0093630101 NGC 1052 7600 0306230101 NGC 1052 25533 0553300301 NGC 1052 30973 0553300401 NGC 1052 35411 0723180101 NGC 1961 920 0723180501 NGC 1961 824 0723180901 NGC 1961 980 0723180601 NGC 1961 783 0723180301 NGC 1961 1089

Table 2.3: Total counts and exposure times for the observations

2.4 Data Analysis

As has been mentioned, we wish to investigate whether the LINERs reveal any form of AGN processes, and/or starburst activity. Hence, when choosing models to fit to the data, we should select those that are related to AGNs or starburst, and designate initial parameters which seem reasonable for the objects in focus. This is what we use xspec for. Methodwise, we select models and fit the data to them using xspec so that the χ2-value becomes as small as possible. In other words, we minimize the χ2-value to check the goodness of the fit. Graphically, this means we choose our models and xspec selects parameters so that the model overlaps as much as possible the observed spectrum. In addition, we add model components step by step and select initial values for the components from the best fit of previous testing, so as to understand how adding more models improves the fitting. In order to clearly evaluate the underlying processes of the LINERs, the statistical testing procedure must be gradually and rigorously advanced. In addition, we always set the normalization of the newly added model component to zero, so that the current χ2_{-value is initially unaffected by the new model component and} can only improve the fit if the normalization constant increases. In this way, irrelevant model components will have an unchanged normalization after xspec fits the data. If one were to add all the model components at once and then fit, xspec could risk finding a local minimum of the χ2-value, thereby erroneously analyzing the data and displaying false parameters. By following this methodical process, we make sure to only increase the validity of the fit with each step [19].

Specifically in our case, we wish to utilize the models in xspec named wabs [83], zwabs [83], zpcfabs [86], powerlaw, mekal [82], pexriv [81] and gauss. The xspec manual [62] contains more information about the models than has been included here; the sections here function as a short introduction to their application in the project.

The first three models deal with the photon absorption of the hydrogen columns; wabs deals with the absorption in our Galaxy and zwabs/zpcfabs the local absorption in the target galaxies. zpcfabs also includes a Cover − F raction, which describes how large a fraction of the local galaxy is effected by the absorption process [86]. The input value for wabs is the hydrogen column density in our galaxy, for which we calculate with the help of NASA Heasarc’s NH Column Density Database [60]. In order to find the redshift for zwabs (and other models), the NED database was queried.

(21)

The powerlaw model applies a simple power-law component to the fitted spectrum. The powerlaw component is an essential feature of an AGN source and it is crucial to test this. The value of the photon index (or Γ) is usually between 1 and 2.5. Neither the photon index of the power-law nor the normalization can be known and must be fitted by xspec.

The mekal model simulates the energy emission of a hot, dense gas. We apply this representation to test for starburst activity, soft excess or emission from the host galaxy. Generally, starburst activity emits at a mekal-temperature (kT ) around 0.7 keV [42] [34], soft excess at 0.1-0.2 keV [5], and host galaxy at 0.3-0.5 keV [30] [57]. The abundance of metallic elements in relation to the Sun can be varied as well. These elements are C, N, O, Ne, Na, Mg, Al, Si, S, Ar, Ca, Fe and Ni. In our case, mekal operates especially in the lower regions of the X-ray spectrum (from 0.5 keV to 2 keV), and this is important since the power-law component is weaker at lower energies.

Even though we can test non-reflective emission with powerlaw and mekal, a typical AGN spectrum involves plenty of reflection components. To test this, we can apply pexriv, which includes a power-law component with a cut-off feature and a reflection factor. Also, abundances of metallic elements relative to the Sun are included in this model as well. Other factors include the inclination angle i, disk temperature and the ionization parameter ξ, defined as ξ = 4πFion

n , where Fion is the 5 eV - 20 keV irradiating flux and n is the density of the reflector [78]. pexriv does not include an iron line. The pexriv model tests mainly for the Compton reflection. Another important feature of AGNs is the reflection nature of the processes, which can be tested with pexriv.

As discussed, there are narrow lines in the AGN X-ray spectrum due to fluorescence mechanisms. These iron lines can be represented as gaussians, with fixed energy values. Thus, the command gauss is used to represent narrow lines, to further mimic AGN properties in the final spectrum.

Another form of spectral characteristics or emission distortion is the relativistic smear-ing from the strong gravitational fields of the SMBH. This is emulated by the component kdblur2 in xspec. Being a convolution model, it is applied onto other model compo-nents. In this case, we select pexriv as our model component with which to convolve, because the reflection can in principle occur in the inner accretion disk. kdblur2 was developed by Ari Laor [80].

Furthermore, with the help of xspec, we can evaluate other quantities and relation-ships. Using the command lum, we can compute the luminosity of the source. As lum calculates the luminosity of the model components, we can utilize it to separate the various sources, e.g. the primary emission from the secondary emission. Moreover, we calculate the errors of the luminosity from the errors of the model parameters. The formula followed is:

∆N orm N orm =

∆Lum Lum

where N orm is the normalization on which the command lum is applied, ∆ is the error and Lum is the luminosity calculated [85]. Additionally, in order to further constrain

(22)

data most appropriately. Our ambition is to test whether or not an AGN could be at the center of these galactic nuclei, or if there is evidence for starburst activity alone. Further-more, we wish to determine any physical parameters, e.g. photon index, temperature of the plasma, column density, of the source. Likewise, we search for any variability in the fluxes of the these sources. By fitting the observed data to statistical models, we hope to draw conclusions about the source of the LINERs.

2.5 Model Selection

Our aim with using xspec is to find the best theoretical model for the data. The χ2_-value is a measurement of how closely the model matches the spectrum. In order to find the best combination of xspec-models, we apply the model components one at a time, fitting the model to the data after each additional model. In this section, we list the selected models in order to outline and explain the models we use.

2.5.1 NGC 1052

ABS = constant*wabs*zpcfabs

PMGP = powerlaw+mekal+gauss+pexriv

1. Power [constant*wabs*zpcfabs*(powerlaw)]

• The constant-model is added in order to simultaneously fit the MOS-data and pn-data

• The column density for wabs is selected based on the NH-database [60], i.e. for our galaxy in the direction of NGC 1052

• Parameters for zpcfabs are set at their standard starting values, except for the redshift which is set at z = 0.005

• The column density for zpcfabs is free to vary

• The initial value for the Γ is 1.9 and allowed to vary between 1 and 2.5 • The normalization is set to zero and then free to vary

2. Hot Gas [ABS*(powerlaw+mekal)] • The redshift is fixed at z = 0.005

• Initial values are 0.5 keV for the temperature

• The temperature is free to vary, while the abundances, hydrogen density and switch are fixed at standard values

• The normalization starts at zero and is free to vary 3. Fe Emission [ABS*(powerlaw+mekal+gauss)]

• The energy value is 6.37 keV (to compensate for redshift) • Sigma fixed to 0.08 keV

(23)

4. Reflection [ABS*(powerlaw+mekal+gauss+pexriv)]

• The photon index is fixed to the same value as the primary emission’s photon index

• The cut-off energy is fixed to 100 keV

• Relative reflection starts at -1.5, and only varies between -2 and 0 (Negative values for the Relative reflection implies only that the reflection component is included, and not the direct emission as well; see [81] for more information) • The redshift is again fixed at z=0.005

• Ionization is for this model fixed at zero

• Metal abundances, cosine of angle and the disk temperature are fixed at their standard initial values

• The normalization starts at zero and is free to vary 5. Ionized [ABS*(PMGP)]

• Frees the ionization parameter of pexriv • Starts from the values of Reflection 6. Fe Width [ABS*(PMGP)]

• Releases the gaussian sigma, to vary between 0.05 keV and 0.5 keV • Starts from the values of Reflection

7. Metallic [ABS*(PMGP)]

• Links the metallic abundances of mekal and pexriv to assume the same values and allows these to be free

• Starts from the values of Reflection 8. textitIncline [ABS*(PMGP)]

• Varies the angle of the AGN’s accretion disk • Starts from the values of Reflection

9. Second Power [ABS*(PMGP+powerlaw)]

• Adds another power-law inside the absorption components • Inital value for Γ is 1.9, and then free to vary

• Normalization is set to zero, then free to vary • Starts from the values of Reflection

(24)

2.5.2 NGC 1961

1. BH Engine [constant*wabs*(powerlaw)]

• The column density for wabs is selected based on the NH-database [60] for Earth in direction of NGC 1961

• Initial value of the Γ is set to be 1.9, then free to vary between 1 and 2.5 • The normalization of the powerlaw is initially zero, then free to vary 2. AGN [constant*wabs*(powerlaw+mekal)]

• Initial value for the temperature is selected to be 0.5 keV, then free to vary • The redshift is fixed at z = 0.013122

• The abundances, hydrogen density and switch are fixed at standard values • The normalization starts at zero and is free to vary

3. AGN & Hot Gas [constant*wabs*(powerlaw+mekal+mekal)]

• Initial value for the additional temperature is selected to be 0.5 keV, then both are free to vary

• The abundances, hydrogen density and switch are fixed at standard values • The normalization starts at zero and is free to vary

4. Starburst [constant*wabs*(mekal)]

• Initial value for the temperature is selected to be 0.5 keV, then free to vary • The temperature is free to vary, while the abundances, hydrogen density and

switch are fixed at standard values

• The normalization starts at zero and is free to vary 5. Starburst & Hot Gas [constant*wabs*(mekal+mekal)]

• Initial value for the additional temperature is selected to be 0.5 keV, then both are free to vary

• The temperature is free to vary, while the abundances, hydrogen density and switch are fixed at standard values

(25)

Chapter 3 Processed Data and Results

3.1 Data Extraction Procedure

3.2 Spectral Data Plots

From the extraction regions, we perform spectral extractions in order to obtain spectra for analysis. For the sake of an overview, all of the spectra from NGC 1052 and NGC 1961 have been plotted in Figures 3.1a, 3.1b, 3.1c and 3.1d (MOS-spectra and pn-spectra plotted separately).

(26)

1 2 5 10 −3 0.01 0.1 normalized counts s −1 keV −1 Energy (keV) NGC 1052 MOS #1: 2001−08−15 #2: 2006−01−12 #3: 2009−01−15 #4: 2009−08−12

(a) All MOS Spectra of NGC 1052

1 2 5 10 10 −3 0.01 0.1 normalized counts s −1 keV −1 Energy (keV) NGC 1052 pn #1: 2001−08−15 #2: 2006−01−12 #3: 2009−01−15 #4: 2009−08−12 (b) All pn Spectra of NGC 1052 1 2 5 10 −4 10 −3 0.01 normalized counts s −1 keV −1 Energy (keV) NGC 1961 MOS #3: 2013−09−11 #4: 2014−03−09 #5: 2014−03−15 #8: 2014−03−17 #11: 2014−02−19

(c) All MOS Spectra of NGC 1961

1 2 5 10 −4 10 −3 0.01 0.1 normalized counts s −1 keV −1 Energy (keV) NGC 1961 pn #3: 2013−09−11 #4: 2014−03−09 #5: 2014−03−15 #8: 2014−03−17 #11: 2014−02−19 (d) All pn Spectra of NGC 1961 Figure 3.1: Overview of the Raw Spectra

Concerning the NGC 1052 observations, there are a couple of aspects to consider. Firstly, there seems to be a large flux of soft X-rays in the lower energies (ca 0.5-2 keV). This supports the idea that the existence of soft excess should be tested. Secondly, the decreasing intensity at higher energies resembles some sort of power-law function. Thirdly, there is a high variability in the 5-8 keV range, especially between the two first observations and the two last. This sort of spectral variability supports the AGN model, but further testing of variability should be made. Lastly, the sharp peak at around 6 keV points towards an iron line, another component of AGNs. Through the first impressions of the spectra, we have some knowledge of model components to test.

On the other hand, the NGC 1961 spectra reveal other details. Again, we have some form of soft excess, with the peak at around 1 keV. In contrast to NGC 1052, we have lower variability between the observations, but this could be a result of the time span

(27)

between the observations. However, we should still be able to detect some form of short term variability. Towards higher energies, the spectra could be somewhat decreasing, and testing for a primary emission could be necessary. The spectra for NGC 1961 are not as extensive as NGC 1052, and are in many other ways different.

3.3 Spectral Fitting

3.3.1 NGC1052

Table 3.1 provides an overview of all the models which were tested and fitted, as well as their χ2_{-values. As can be seen, Reflection proved to be the best fit of them all, and} thus has been plotted along with all of the observations in Figure 3.2. In this model, we tested for absorption, primary emission from an AGN-engine, soft excess, iron lines and secondary emission, i.e. reflection in the AGN structure. Moreover, some parameters were selected to freely vary; these include the ionization, the sigma of the gaussian, the metallic abundances and the inclination of the LINER.

Table 3.1: Overview of Models and χ2-Values of NGC 1052

Name Modela _χ2

1/ν1 b χ22/ν2 χ23/ν3 χ24/ν4

Power Po 437.16 / 214 433.44 / 225 659.90 / 319 739.83 / 331

Hot Gas Po+Mek 352.53 / 212 352.66 / 223 445.85 / 317 499.57 / 329 Fe Emission Po+Mek+Gau 298.07 / 211 300.55 / 222 364.94 / 316 415.05 / 328 Reflection Po+Mek+Gau+Pex 247.52 / 209 253.25 / 220 338.47 / 314 383.60 / 326 Ionized Po+Mek+Gau+Pex 247.52 / 208 253.25 / 219 338.47 / 313 383.60 / 325 Fe Width Po+Mek+Gau+Pex 247.35 / 208 253.11 / 219 338.42 / 313 382.43 / 325 Metallic Po+Mek+Gau+Pex 247.51 / 208 253.24 / 219 337.00 / 313 381.64 / 325 Incline Po+Mek+Gau+Pex 247.51 / 208 253.24 / 219 337.29 / 313 381.96 / 325 Second Power Po+Mek+Gau+Pex+Po 247.52 / 207 253.25 / 218 338.47 / 312 383.60 / 324

a _{Po = powerlaw; Mek=mekal; Gau=gauss; Pex=pexriv. All models are multiplied by constant*wabs*zpcfabs - the}

photoelectric absorption

b_χ2_{and degrees of freedom of the nth observation}

Further focusing our attention, we examine the parameters of Reflection. These have been listed with their errors in Table 3.2.

(28)

Table 3.2: The Parameter Values of the Best Fit Model (Reflection) for NGC 1052

Model Component Observation #1 Observation #2 Constant Factora 1.01 ± 0.02 1.00 ± 0.02 zpcfabs, NH 1022cm−2a 2.69 ± 0.87 2.77 ± 0.76 zpcfabs, CvrFracta 0.412 ± 0.134 0.443 ± 0.289

powerlaw, Γb 1.87 ± 0.10 1.86 ± 0.10

mekal, kT (keV)c 0.700 ± 0.029 0.693 ± 0.029

Abundancec 1.00 (Fixed) 1.00 (Fixed)

gauss σ d 0.08 (Fixed) 0.08 (Fixed)

pexriv, relref le -2.00 ± 0.614 -2.00 ± 0.597

pexriv, ξe 0.0 (Fixed) 0.0 (Fixed)

pexriv, cos(i)e 0.45 (Fixed) 0.45 (Fixed) powerlaw, Lumf 8.97 ± 1.21 1039 _{9.73 ± 2.7210}39 mekal, Lum f 3.91 ± 1.64 1039 erg/s 3.79 ± 1.29 1039 pexriv, Lum f 2.93 ± 1.33 1041 _{2.90 ± 2.46 10}41 Total Luminosityf _{3.13 ± 1.42 10}41 _{3.04 ± 2.58 10}41 gauss, Equivalent Width 148 ± 30 eV 146 ± 29 eV Model Component Observation #3 Observation #4 Constant Factora 0.975 ± 0.012 0.975 ± 0.011 zpcfabs, NH (1022cm−2)a 8.33 ± 0.25 8.85 ± 0.23 zpcfabs, CvrFracta 0.914 ± 0.047 0.927 ± 0.038

powerlaw, Γb 1.37 ± 0.05 1.40 ± 0.05

mekal, kT (keV)c 0.658 ± 0.020 0.624 ± 0.019

Abundancec 1.00 (Fixed) 1.00 (Fixed)

gauss σd 0.08 (Fixed) 0.08 (Fixed)

pexriv, relref le -2.00 ± 0.48 -2.00 ± 0.70

pexriv, ξe 0.0 (Fixed) 0.0 (Fixed)

pexriv, cos(i)e 0.45 (Fixed) 0.45 (Fixed) powerlaw, Lumf 1.89 ± 2.05 1041 _{3.21 ± 12.4 10}41 mekal, Lum f 2.90 ± 9.01 1039 2.92 ± 6.80 1039 pexriv, Lum f 1.94 ± 1.86 1041 1.89 ± 1.75 1041 Total Luminosityf _{4.36 ± 4.18 10}41 _{4.31 ± 3.99 10}41 gauss, Equivalent Width 134 ± 14 eV 135 ± 15 eV

a _{The factor for the pn-data, the factor for the MOS-data is 1; the column density}

of the target galaxy; the Cover − F raction of the column density

b_{The Photon Index of the powerlaw. The Photon Index of pexriv is set to be}

the same as the powerlaw

c _{The temperature of the soft excess and the metal abundance (in relation to the}

Sun, where 1 is the same composition as the Sun). The abundances of pexriv is set to be the same as mekal’s

d_{The σ of the gauss-model; that represents the iron line emission}

e _{The reflection scaling factor; the ionization parameter; the angle of the target}

galaxy

f

The luminosity of the models. The energy range for powerlaw and pexriv is 2 keV - 12 keV, while the range for mekal is 0.5-2 keV. The values are in erg/s.

(29)

In order to evaluate the iron line, we calculate the equivalent width of the gaussian. These values, along with the error, are included in Table 3.2. The error of the equivalent width is based on the error of the gaussian norm. The calculations are the same as for the error of the luminosity, namely: ∆N orm

N orm = ∆EQW

EQW .

The best fit, Reflection, has been plotted together with the spectra in Figure 3.2.

10−4 10−3 0.01 0.1 normalized counts s −1 keV −1 1 2 5 10 −0.02 0 0.02 normalized counts s −1 keV −1 Energy (keV)

(a) Reflection on the First set of Observations together with the residuals

Γ = 1.87 ± 0.10; NH = 2.69 ± 0.87 1022cm−2 kT = 0.700 ± 0.028 keV 10−4 10−3 0.01 0.1 normalized counts s −1 keV −1 1 2 5 10 −0.02 −0.01 0 0.01 0.02 normalized counts s −1 keV −1 Energy (keV)

(b) Reflection on the Second set of Observations together with the residuals

Γ = 1.86 ± 0.10; NH = 2.77 ± 0.76 1022cm−2 kT = 0.693 ± 0.029 keV 10−4 10−3 0.01 0.1 normalized counts s −1 keV −1 1 2 5 10 −0.01 0 0.01 normalized counts s −1 keV −1 Energy (keV)

(c) Reflection on the Third set of Observations together with the residuals

10−4 10−3 0.01 0.1 normalized counts s −1 keV −1 1 2 5 10 −0.02 −0.01 0 0.01 normalized counts s −1 keV −1 Energy (keV)

(d) Reflection on the Fourth set of Observations together with the residuals

(30)

3.3.2 NGC1961

Similarly, Table 3.3 provides an overview of all the models tested, as well as their χ2 -values. As can be seen, AGN proved to be the best fit of them all, and thus that plot of all the observations is presented in Figure 3.3. For AGN, we tested for absorption, soft excess and for a power-law feature. As the spectra of NGC 1961 are clearly different from NGC 1052, another set of models, to fit the data with, is required.

Table 3.3: Overview of Models for NGC 1961

Name Modela χ32/ν3 χ42/ν4 χ52/ν5 χ82/ν8 χ112/ν11

BH Engine power 150.56 / 47 57.43 / 33 52.67 / 37 55.43 / 23 79.38 / 32

AGN power+mekal 34.27 / 42 51.89 / 31 29.40 / 35 24.60 / 21 26.15 / 30

AGN & Hot Gas power+mekal+mekal 32.94 / 40 29.31 / 29 25.10 / 33 23.41 / 19 23.46 / 28 Starburst mekal 145.38 / 44 189.37 / 33 211.73 / 37 107.37 / 23 130.57 / 32 Starburst & Hot Gas mekal+mekal 135.73 / 42 189.37 / 31 211.73 / 35 24.28 / 21 130.57 / 30

a _{All models are multiplied by constant*wabs}

Correspondingly, we examine the parameters of AGN more closely. Together with their errors, they are listed in Table 3.4.

Table 3.4: The Parameter Values of Model AGN for NGC 1961

Model Component Observation #3 Observation #4 Observation #5 Constant Factora _{0.977 ± 0.075} _{0.972 ± 0.124} _{1.04 ± 0.10} powerlaw Γb 1.00 ± 0.8 1.37 ± 0.17 1.56 ± 0.17 mekal kT (keV)c 0.644 ± 0.023 0.528 ± 0.073 0.477 ± 0.062 powerlaw, Lumd 3.32 ± 1.30 1040 _{5.09 ± 1.06 10}40 _{4.58 ± 0.79 10}40 mekal, Lum d 1.61 ± 3.12 1040 _{8.97 ± 2.49 10}39 _{1.28 ± 1.03 10}40 Total Lumd _{5.66 ± 2.22 10}40 _{7.54 ± 1.57 10}40 _{6.53 ± 1.13 10}40 Model Component Observation #8 Observation #11

Constant Factora 0.864 ± 0.121 1.14 ± 0.11 powerlaw Γb 1.22 ± 0.24 1.38 ± 0.22 mekal kT (keV)c 0.658 ± 0.052 0.535 ± 0.050 powerlaw, Lumd 4.67 ± 1.19 1040 _{4.31 ± 1.20 10}40 mekal, Lum d 1.05 ± 2.70 1040 1.62 ± 1.24 1040 Total Lumd _{7.64 ± 1.95 10}40 _{6.37 ± 1.77 10}40

a _{The factor for the pn-data, the factor for the MOS-data is 1} b_{The Photon Index of the powerlaw}

c _{The temperature of the soft excesses}

d_{The luminosity of the models. The energy range for powerlaw is 2 keV - 12 keV, while}

the range for mekal is 0.5-2 keV. The values are in erg/s

The best fit, AGN, has been plotted together with the observations in Figure 3.3. In Figure 3.3b and Figure 3.3d, the red line is shorter than the black line. This is due to the shorter energy range of the pn-data compared to the MOS. One explanation is that there are not counts in the pn-data after the background subtraction.

(31)

10−5 10−4 10−3 0.01 normalized counts s −1 keV −1 1 2 5 0 5×10−3 0.01 normalized counts s −1 keV −1 Energy (keV)

(a) AGN on the Third set of Observations together with the residuals

Γ = 1.00 ± 0.8; kT = 0.644 ± 0.023 keV 10−5 10−4 10−3 0.01 normalized counts s −1 keV −1 1 2 5 0 5×10−3 normalized counts s −1 keV −1 Energy (keV)

(b) AGN on the Fourth set of Observations together with the residuals

Γ = 1.37 ± 0.17; kT = 0.528 ± 0.073 keV 10−5 10−4 10−3 0.01 normalized counts s −1 keV −1 1 2 5 −0.01 −5×10−3 0 5×10−3 0.01 normalized counts s −1 keV −1 Energy (keV)

(c) AGN on the Fifth set of Observations together with the residuals

Γ = 1.56 ± 0.17; kT = 0.477 ± 0.062 10−5 10−4 10−3 0.01 normalized counts s −1 keV −1 1 2 5 0 0.01 0.02 normalized counts s −1 keV −1 Energy (keV)

(d) AGN on the Eighth set of Observations together with the residuals

Γ = 1.22 ± 0.24; kT = 0.658 ± 0.052 keV 10−5 10−4 10−3 0.01 normalized counts s −1 keV −1 1 2 5 −0.01 0 0.01 normalized counts s −1 keV −1 Energy (keV)

(32)

3.4 Contour Plots

As an illustration, contour plots between parameters were made of the data from NGC 1052. The contour plots between the column density NH in the direction of NGC 10 52 and the photon index is portrayed in Figure 3.4. Furthermore, the contour plots between the Cover − F raction and the photon index is plotted in Figure 3.5. The contour calculations are all done on the model Reflection varying only the parameters mentioned and keeping the other parameters of the model fixed.

There are mainly four reasons for producing contour plots. Firstly, we wish to see the range of the possible values for the parameters. Secondly, the contour plots provide a visualization of the constraint on the parameters. In other words, the contour plots display how well the parameters are determined and the range of their uncertainties. Thirdly, we could detect any inter-parameter relationships that are difficult to quantify through only manual testing of the parameters. Lastly, it allows us to investigate any variability or other differences in the data. By using contour plots, we can more rigorously constrain and analyze our parameters of the models.

There are three significance levels in the contour plots. The innermost region is the 1st σ confidence level (the red line), the second region is the 2nd σ confidence level (the green line) and the third region is the 3rd σ confidence level (the blue line). They represent respectively the 68%, 95.4% and 99.7% probability for those parameters.

(33)

1.831 1.84 1.85 1.86 1.87 1.88 1.89 2 3 4 5 Parameter: nH (10 22 ) Parameter: PhoIndex cross = 2.475e+02; Levels = 2.498e+02 2.521e+02 2.567e+02

+

(a) First set of Observations for NGC 1052

1.83 1.84 1.85 1.86 1.87 1.88 1.89 2 3 4 5 Parameter: nH (10 22 ) Parameter: PhoIndex cross = 2.532e+02; Levels = 2.555e+02 2.579e+02 2.625e+02

+

(b) Second set of Observations for NGC 1052

1.357.8 1.36 1.37 1.38 1.39 8 8.2 8.4 8.6 8.8 Parameter: nH (10 22) Parameter: PhoIndex cross = 3.532e+02; Levels = 3.555e+02 3.578e+02 3.624e+02

+

(c) Third set of Observations for NGC 1052

1.38 1.385 1.39 1.395 1.4 1.405 1.41 1.415 8.4 8.6 8.8 9 9.2 Parameter: nH (10 22) Parameter: PhoIndex cross = 3.991e+02; Levels = 4.014e+02 4.037e+02 4.083e+02

+

(d) Fourth set of Observations for NGC 1052 Figure 3.4: Contour Plots for the Column Density versus the Photon Index

(34)

1.83 1.84 1.85 1.86 1.87 1.88 1.89 0.38 0.4 0.42 0.44 0.46 Parameter: CvrFract Parameter: PhoIndex cross = 2.475e+02; Levels = 2.498e+02 2.521e+02 2.567e+02

+

(a) First set of Observations for NGC 1052

1.830.4 1.84 1.85 1.86 1.87 1.88 1.89 0.42 0.44 0.46 0.48 Parameter: CvrFract Parameter: PhoIndex cross = 2.532e+02; Levels = 2.555e+02 2.579e+02 2.625e+02

+

(b) Second set of Observations for NGC 1052

1.35 1.36 1.37 1.38 1.39 1.4 0.91 0.912 0.914 0.916 0.918 0.92 Parameter: CvrFract Parameter: PhoIndex cross = 3.532e+02; Levels = 3.555e+02 3.578e+02 3.624e+02

+

(c) Third set of Observations for NGC 1052

1.39 1.395 1.4 1.405 1.41 1.415 0.924 0.926 0.928 Parameter: CvrFract Parameter: PhoIndex cross = 3.991e+02; Levels = 4.014e+02 4.037e+02 4.083e+02

+

(d) Fourth set of Observations for NGC 1052 Figure 3.5: Contour Plots for the Cover − F raction versus the Photon Index

(35)

Chapter 4 Discussion and Conclusion

4.1 NGC 1052

Among the results, the best fit model for the observations of NGC 1052 was found to be Reflection. Reflection tests the existence of absorption, a primary emission from an AGN-engine, thermal emission, iron lines and reflected or secondary emission. The xspec-components used as well as the goodness of the statistical fit can be seen in Table 3.1. The parameters of Reflection along with the values of errors are summarized in Table 3.2.

In order to evaluate the statistical significance of the models, we analyze Table 3.1. From model Power to model Reflection, we can see that the χ2-value improves for each of the added model components mekal, gauss and pexriv. Thus, we can say that we detect these model components, i.e. mekal, gauss and pexriv, in the data. In other words, model Reflection detects local absorption, the primary power-law emission from the corona (absorbed in the local rest frame), together with thermal emission, an iron line and reflected emission. These are all typical structural components of an AGN. Thus, we conclude that NGC 1052 is powered by an AGN-like structure.

Again, with attention to Table 3.2, we evaluate the values of the parameters. Our best fit (the fourth observation) for the temperature of the thermal emission, mekal kT , has a value of around 0.624 ± 0.019 keV, which is higher than the values of Weaver et al. (1999), 0.53+0.34_−0.26, but in the same range [57]. Based on the temperature found, Weaver et al. concluded that their mekal-temperature pointed towards a galaxy source in the emission, and thus their “AGN+galaxy”-model provided the best results. Indeed, our value for the temperature is higher, which further suggests the existence of a starburst source. For the photon index, Γ, we have 1.40 ± 0.05 for our best fit (again, the fourth observation). This value is quite reasonable from a purely physical viewpoint, since this is in the expected range for photon indices of AGNs. Furthermore, it is consistent with the values of Hern´andez-Garc´ıa et al. (2013) [29]. Additionally, this photon index value together with the calculated luminosity (< 1042ergs−1) and a narrow iron line allows for the possibility that NGC 1052 is powered by an ADAF AGN. The results of Guainazzi et

(36)

abundances, the cosine of the inclination angle cos(i), the ionization parameter ξ and iron line σ. From Table 3.1, we can see that from Ionization to Incline the χ2_{-value does not} significantly improve. This suggests that we cannot detect these parameter dependencies with substantial evidence. In other words, we can conclude that we do not detect a second powerlaw or any ionized materials. Furthermore, our data is not sensitive to a change from the standard input values of the iron line width, metallicity or inclination of the disk.

The equivalent width of an emission line can be used to evaluate the intensity of the iron line for the spectra. When analyzing NGC 1052, we applied a gaussian in order to fix the iron line at 6.4 keV, in the LINER’s reference frame. The equivalent width of the models can be seen in Table 3.2. Notably, the equivalent widths of the two first observations are larger than those of the two last observations, by a small margin. This is evident with our change in NH and Cover − F raction, both of which increase from the two first observations to the two last observations. One explanation for this phenomenon is the existence of clumpy clouds in the torus. These rotate with Keplerian motion around the central SMBH. Hence, these clouds have re-positioned themselves over the course of our observations, blocking the line-of-sight, resulting in an increase of NH and Cover − F raction [49] [47] [50].

Comparing our results with previous findings, we find that Brenneman et al. (2009) and Guainazzi et al. (1998) have made similar investigations. Brenneman et al. finds an equivalent width of 201 ± 35 eV, which is not consistent with the equivalent widths of our earlier observations. In another keypoint, Brenneman et al. tested other models to fit an iron line (diskline and laor), but found gauss to improve the χ2_{-value the} most [6]. Guainazzi et al. tested a variety of models on observations of ROSAT and ASCA on NGC 1052, some similar to Incline. In one of these models, testing for pure reflection and soft excess, the equivalent width of the iron line was found to be 90+100₋₇₀ eV, which is in the same range as our later observations. Notably, the errors of Guainazzi et al.’s equivalent widths are larger than the ones calculated here, due to a lower quality of data [25]. The equivalent width found in our observations are in the same range as when compared to earlier calculations of the same object, and we thus see some support for our findings.

On the contrary, some model components did not improve the goodness of the fit. Two of these were the convolution of kdblur2 with pexriv or with gauss. Hence, these tests were not included within the best fit model. Furthermore, the iron line of the object was modelled with other xspec models together with gauss. One of these was diskline [79]. However, this model did not not improve the goodness of the fit any more than only gauss could.

Judging from the contour plots of Figures 3.4 and 3.5, the parameter constraints can be discussed. The parameters are all well-constrained, in all observations, at 3 σ confidence level. For NH vs. Photon Index (Figure 3.4), there is an intrinsic anti-correlation between the parameters. Moreover, the parameters do reveal a change over time: the Photon Index decreases from ∼ 1.9 to ∼ 1.4 and the NH increases from ∼ 3 ∗ 1022_cm−2 _{to ∼ 8 ∗ 10}22_cm−2_{. Similarly, Cover − F raction changes over time when} plotted against the Photon Index, although there is no internal correlation between the parameters in these cases. The Cover − F raction increases from initially ∼ 0.4 to ∼ 0.9. As our parameters are well-constrained and show a significant change of values over the course of the observations, we can confidently conclude that our data demonstrate clear

(37)

variability.

4.2 NGC 1961

Comparatively, AGN & Hot Gas fit the spectra of NGC 1961 quite well. In this case, we tested for absorption and two sources of thermal emission, one from stellar evolution and the other from the galactic nuclei. The xspec-components used, as well as the goodness of the statistical fit, can be seen in Table 3.3. The parameters of AGN along with the values of errors are summarized in Table 3.4. Important to note, the pn-spectra of the fourth and eighth spectra are quite poor, with few data points, resulting in poorly constrained results.

In order to discuss the statistical significance of our data from NGC 1961, we consider the significance of any fit improvement based on the χ2 _{decrease and the corresponding} change in the degrees of freedom. From model BH Engine to AGN we detect in our spectrum a significant improvement. The addition of another mekal, model AGN & Hot Gas, does improve the χ2_{-value of the fit, but not to a notable level of significance. Thus,} we cannot confidently conclude that there exists another source of thermal emission close to NGC 1961 without more testing. However, we conclude that we find another thermal feature overlapping the mekal. This can, in principle, be interpreted as AGN emission or as a contribution from the host galaxy [53]. The thermal emission can not be interpreted as AGN soft excess, as will be discussed. Even though the data of NGC 1961 is not as high quality as the data from NGC 1052, i.e. not as many data counts, we can still argue that the source presents traits consistent with AGN features.

From Table 3.4, we see that the values for Γ and the kT are somewhat consistent throughout the observations. Including all observations, the best fit value of Γ is 1.38 ± 0.18 (eleventh observation). This is reasonable compared to the value calculated by Anderson et al. (2011) [2], Γ = 1.4+0.5−0.4. Important to note, however, is that Anderson et al.’s work is based on the hot gaseous halo around NGC 1961, and may thus not be directly comparable: i.e., the extraction regions are not the same for both studies, and in Anderson et al.’s study, it is altered throughout. On the other hand, a similar model of two mekals was tested by Anderson et al. (2015) [3]. For the inner regions of NGC 1961, the temperatures found were similar to our values, 0.535±0.050 keV. Due to this range of values, we conclude that the temperature of the mekal points towards starburst activity. Again, Anderson et al. focuses more attention on the regions around the inner core and also utilizes all of the observations from XMM-Newton of NGC 1961. The photon index and temperatures of our models agree with Anderson’s findings, which focuses on the diffuse emission.

Other models were tested, but were found not to be statistically significant. Two of these model components were absorption of the target galaxy, zwabs or zpcfabs. None of the mentioned models improved the fit or gave reasonable values when multiplied with wabs, and were thus rejected. Also, a gaussian was tested in order to find an iron line

A X-Ray Spectroscopy Study of LINERs as Observed by XMM-Newton

Is Every One LINER Dense at Its Core?