Flux Contribution and Geometry of Charge Exchange Emission in the Starburst Galaxy M82

Recent X-ray studies of starburst galaxies have found that charge exchange (CX) commonly occurs between outflowing hot plasma and cold gas, possibly from swept-up clouds. However, the total CX flux and the regions where CX occurs have been poorly understood. We present an analysis of XMM-Newton observations of M82, a prototype starburst galaxy, aiming to investigate these key properties of CX emission. We have used a blind source separation method in an image analysis of CCD data, which identified a component with the enhanced O–K lines expected from the CX process. Analyzing the XMM-Newton/RGS spectra from the regions identified by the image analysis, we have detected a high forbidden-to-resonance ratio of the O vii Heα triplet as well as several emission lines from K-shell transitions of C, N, and O that are enhanced by the CX process. CX is less responsible for the emission lines of Ne and Mg, and accurate estimation of the CX contribution is confirmed to be crucial in measuring chemical abundances. The temperature of the plasma acting as an electron receiver in the CX process is significantly lower compared to that of the plasma components responsible for most of the X-rays. From the low temperature and an estimation of the CX-emitting volume, we find that the CX primarily occurs in a limited region at interfaces between plasma and gas whose temperatures rapidly decrease due to thermal conduction.


Introduction
Galaxy-scale outflows in starburst galaxies (Veilleux et al. 2005;Rubin et al. 2014) are responsible for removing energetic and chemically enriched materials from the galaxy, injecting them into the circumgalactic medium or even the intergalactic medium (Borthakur et al. 2013;Werk et al. 2016).Such outflows regulate star formation activity within galaxies as well as act as the major driver of cosmic chemical enrichment (Oppenheimer & Davé 2006;Peeples & Shankar 2011).The prevailing picture is that outflows are formed by hot gas shockheated by supernovae (SNe) and stellar winds that entrain dust and cold gas (e.g., Chevalier & Clegg 1985).X-ray emissions from multiphase gas have been widely used as diagnostic tools to investigate their kinematic and chemical properties (e.g., Tsuru et al. 2007;Lopez et al. 2020).
Detection of charge exchange (CX) based on recent observations of nearby starburst galaxies with RGS on board XMM-Newton (Ranalli et al. 2008;Liu et al. 2011Liu et al. , 2012) ) gave us unique insights to the previous studies on outflows.CX is an atomic process predicted to occur between an ionized plasma and neutral matter in an outflow (Lehnert et al. 1999;Lallement 2004), and selectively enhances several emission lines such as the forbidden ( f ) and resonance (r) lines of the O VII Heα triplet.Measurements of the flux of these enhanced lines and their ratios allow us to put strong constraints on key parameters (e.g., bulk and turbulence velocities and density) to understand complicated gas-phase physics (e.g., hydrodynamic effects such as shocks and turbulence, thermal conduction, and nonequilibrium emission processes) that play an important role in the evolution of outflows (e.g., Strickland & Heckman 2009).Although the number of results detecting CX emission in the X-ray band is growing, quantitative flux estimations and investigations of CX geometry have been poorly understood, as distinguishing thermal X-ray plasma and CX emission remains difficult and CX modeling is complex (e.g., Smith et al. 2014).
M82, the prototype of a starburst galaxy, is where CX emission was detected for the first time (Ranalli et al. 2008).Because of its close proximity (3.6 Mpc; Freedman et al. 1994) and large inclination angle (i ∼ 80°; McKeith et al. 1995), M82 is regarded as the ideal target for observational studies on outflows.Biconical outflows from the nuclear region of M82 extend perpendicularly to the galactic disk and their properties have been widely studied across the electromagnetic spectrum, tracing atomic H I and molecular gas (e.g., Walter et al. 2002;Salak et al. 2013), warm ionized gas in Hα (e.g., McKeith et al. 1995;Ohyama et al. 2002), hot plasma emitting X-rays (e.g., Watson et al. 1984;Lopez et al. 2020), and dust in the UV, IR, and submillimeter bands (e.g., Origlia et al. 2004;Hoopes et al. 2005).Past observations of M82 aiming to investigate the nature of X-rays pointed out the discrepancy between the plasma model and the CCD spectra of the O VII Heα triplet in both the galactic disk and outflow regions of M82 (Konami et al. 2011;Lopez et al. 2020).
Here we report on an analysis of XMM-Newton data of M82, combining the application of a blind source separation method together with high-resolution spectroscopy to pin down the origin of the CX emission.Throughout this paper, we adapt 3.6 Mpc as the distance to M82 (Freedman et al. 1994), and the statistical errors are quoted at the 1σ level.
Original content from this work may be used under the terms of the Creative Commons Attribution 4.0 licence.Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.

Observations and Data Reduction
M82 was observed with XMM-Newton many times between 2003 and 2021.In most of these observations, the field of view (FoV) of MOS and the pn CCDs completely covers the entirety of M82 including both outflows.We used MOS and pn data only for the image analysis and identifying spectral characteristics.We reprocessed this data with the Science Analysis System software (SAS) version 19.1.0and the Current Calibration Files, following the cookbook for analysis procedures of extended sources. 5We discarded short-duration Observation IDs whose MOS 1+2 exposure time after the screening is less than 20 ks.The observations that meet the criteria are summarized in Table 1.Furthermore, two data sets (Obs.ID = 0870940101 and 0870940401) observed in the small window mode were not used.We used RGS data only for spectral analysis.These data were reprocessed using the rgsproc task in SAS.We extracted background light curves and filter out observation periods when the background count rate is higher than 0.15 counts s −1 .

Imaging Analysis
Our first purpose is to constrain the spatial distribution of a component with the enhanced O VII Heα triplet expected from the CX process (e.g., Lallement 2004).We used a blind source separation method, the General Morphological Components Analysis (GMCA; Bobin et al. 2015), that was designed specifically for cosmic microwave background reconstruction using Planck data.The method was recently introduced to an analysis of Chandra data by Picquenot et al. (2019), and the authors demonstrated that this algorithm works as a powerful tool to isolate physically meaningful components in data and identify extraction regions of interest.In the case of M82, the observed X-rays arise from multitemperature thermal plasmas (cold: ∼0.2 keV; medium hot: ∼0.5 keV; hot: ∼0.9 keV; and extreme hot: 6 keV) as well as nonthermal emission, in addition to the CX emission we are interested in (e.g., Konami et al. 2011;Lopez et al. 2020).Although the GMCA has no intrinsic knowledge of these different components, if the CX emission has an identifiable spectral and spatial shape in the data, it should be able to extract it.
The primary concept of the GMCA method is to take into account the morphological particularities of distinct components by measuring the sparsity level in the wavelet domain for each energy slice of a 3D data cube of photon position (x, y) and energy (E).Inputs to the algorithm are the data cube and the user-defined number N of components to extract, and the output is a set of N images associated with the spectra.In our analysis, the position (x, y) and energy (E) information in the cube correspond to the sky coordinates of each event in the band images and the energy range we define, respectively.We used the data in the energy range from 0.3 to 2.5 keV where all CX-enhanced lines (e.g., C VI Heγ: 0.459 keV; N VII Lyα: 0.500 keV; and Si XIII Heα: 1.839 keV) discussed in previous work (e.g., Zhang et al. 2014) are covered, and generated 44 band images with binning = 3 in equal 0.05 keV increments, with adapt-merge.Subsequently, we merged them together to build the input data cube.The number N was fixed to 4. For the case of N 5, two of the components show similar images and spectra, which can be interpreted as overfitting of the data.On the other hand, for N 3, the CX component of interest and other components cannot be disentangled well.
Figure 1(a) shows the spectra of the primary extracted components.The spectrum of Component 1 is characterized by soft emission below 1 keV, including the O VII Heα triplet selectively enhanced by the CX process as reported by previous work (e.g., Ranalli et al. 2008).In the spectrum of Component 2, one can see a strong Fe-L complex and Ne VIII Lyα lines.These characteristics are consistent with the interpretation that Component 2 mainly represents thermal emission from moderately hot plasma (kT e ∼ 0.5-1.0keV).The spectrum of Component 3 is described by a featureless spectral continuum dominant above ∼2 keV.Given the count map as shown in Figure 1(b), Component 3 can be considered to be dominated by nonthermal emission from M82 X-1 (Konami et al. 2011).Component 4 consists of a line-like structure at ∼1.5 keV.The structure arises from a neutral Al line (1.49keV) in the instrumental background (Kuntz & Snowden 2008) whose careful handling is often required in the analyses of diffuse sources (e.g., Okon et al. 2020Okon et al. , 2021)).

CCD Spectral Characteristics
The spectral interpretations in Section 3.1 are, however, merely qualitative.To quantify these interpretations we used the MOS 1+2 spectra extracted from the representative regions in Figure 1(d).Regions 1-3 are dominated by GMCA Components 1-3 based on the fraction map in Figure 1(c), defined as f i =N i /∑ i=1−4 N i , where N i is the number of photons in the count map for component i in Figure 1(b).We found that the MOS spectrum from Region 1 exhibit the strong O VII Heα and O VIII Lyα lines expected from the GMCA results (Figure 2).We also confirmed that the enhanced Fe-L complex and Ne IX Lyα lines from a hot thermal plasma are seen in the spectra of Region 2 and that a nonthermal continuum above 2 keV dominates the spectrum of Region 3.

RGS Spectral Analysis
Unfortunately, CCDs cannot spectrally resolve the O VII triplet to detect conclusive evidence of CX via a significantly high f/r ratio.We therefore turned to RGS spectra.Given the deviation of the wavelength of the incident photon against its off-axis angle,6 the difference of the center energy of the O VII f and r corresponds to the source size of 4 ~¢.We focused on a bright peak with an angular diameter 2 ~¢, sufficient to resolve the multiplet lines, in the south outflow as shown in the map of Component 1 (noted as "Peak" in Figure 3(a)).We used all data sets in which the RGS FoVs completely cover the Peak region regardless of the roll angle (Obs.ID = 0112290201, 0206080101, 0560590201, 0560590301, 0870940101, and 0870940401).Figure 3(b) shows the firstand second-order spectra of the Peak region, where the data from all observations are integrated to improve the photon statistics.The f and r lines in the RGS data are clearly resolved and a high f/r ratio is confirmed.
We first measured the line intensity of the O VII Heα f and r lines, as well as O VIII Lyα since their ratios offer helpful guides in complex CX modeling.Given the variation in the roll angle among the data sets, we independently fitted the RGS1 +2 first-order spectra in the energy band of 0.55-0.70keV with a phenomenological model consisting of a bremsstrahlung continuum and five Gaussians accounting for the O VII f (0.574 keV), intermediate (0.569 keV), and r (0.561 keV), O VIII Lyα (0.654 keV), and O VII Lyβ lines (0.666 keV).The central energy of the Gaussian components was fixed to the corresponding line energy.The intensity of the Gaussians and the parameters concerning the continuum were allowed to vary.We convolved RGS response matrices (RMFs) with the spatial profile of source emission in the MOS+pn image in the 0.55-0.70keV band, using the FTOOL ftrgsrmfsmooth.In the spectral fitting, we used XSPEC software version 12.11.1 (Arnaud 1996) and the C-statistic (Cash 1979) on the unbinned spectra.The results are summarized in Table 2. Finally, we calculated the exposure-weighted mean intensity I and the statistical error σ I,stat among the measurements as follows, ( ) , and t i are the intensity, statistical error, and exposure time of observation i, respectively.
Figure 4(a) compares the measured O VII Heα f/r ratio with theoretically expected curves from plasma and CX models computed with PyatomDB7 and the AtomDB CX package8 based on cross sections from the Kronos database (Mullen et al. 2016(Mullen et al. , 2017;;Cumbee et al. 2018), respectively.Here, we presented three cases of the CX process where the collision velocity between the plasma and gas is 500 km s −1 , 1000 km s −1 , or 1500 km s −1 , taking into account the velocity of the outflow plasma estimated with the Chevalier & Clegg (1985) model ∼1000-2500 km s −1 (Strickland & Heckman 2009) and the observed Hα clump velocity ∼ 600 km s −1 (Shopbell & Bland-Hawthorn 1998).We found that the observed f/r ratio requires a large CX contribution (50%) for the O Heα f if the CX is assumed to occur between gas and hot plasma whose temperature kT e is ∼0.2 keV, ∼0.5 keV, or ∼0.9 keV, as reported by previous works (Konami et al. 2011;Lopez et al. 2020).In Figure 4(b), we have plotted the observed flux of O Lyα overlaid with the expected flux level assuming that CX is responsible for 50% of the total flux of O VII Heα f.This comparison requires an upper limit for kT e,CX ∼ 0.35 keV for the plasma component receiving electrons via CX, regardless of the collision velocity.In the following spectral analysis, we thus assume that CX occurs between the gas and the coldest plasma component with kT e ∼ 0.2 keV.
Based on this assumption, we fitted the RGS firstand second-order spectra.We employed bvapec implemented in XSPEC to describe the multiple collisional ionization equilibrium plasma components (hot: kT e,hot ∼ 0.9 keV; medium hot: kT e,med ∼ 0.5 keV; and cold: kT e,cold ∼ 0.2 keV).To model the CX emission, we used ACX29 where the CX temperature kT e,CX is linked to the electron temperature kT e,cold in the cold plasma.ACX2 includes any velocity-dependent effects for the calculation of the cross section not included in the previous ACX model.The collision velocity υ col in ACX2 cannot be constrained and does not significantly affect the final results, so we fix υ col at 1000 km s −1 .We set the (n, l) distribution10 of the exchanged ions to the default value of eight, and assumed the case that one ion repeatedly captures electrons until it becomes neutral, which is available in the current CX model.The electron temperature of the plasma components and the normalization of the plasma and CX components were allowed to vary.We tied the abundances of C, N, O, Fe, and Ni among both components and let them vary.The abundances of metals with line emission not detected were fixed to solar.For the intrinsic absorption for M82 (N H,M82 ) and Galactic absorption (N H,MW ) in the direction toward M82, we used TBabs with solar abundances (Wilms et al. 2000) and TBvarabs with the metal abundances in Origlia et al. (2004), respectively.The hydrogen column density N H,M82 of the former was fixed to 0.4 × 10 21 cm −2 (Dickey & Lockman 1990), whereas that of the latter was left as a free parameter.In addition to these models, we added a power law to account for nonthermal emission from M82 X-1.The photon index Γ was fixed at 0.55 given by Konami et al. (2011) whereas the normalization was left free.To account for variation of the flux profile with energy range, we applied five different RMFs convolved with the five energy band images of 0.45-0.60keV, 0.60-0.70keV, 0.70-0.85keV, and 0.85-1.25 keV.We did not use the spectra of the energy band above 1.25 keV as the significant difference between the profiles of the thermal and nonthermal emission

Note.
a In units of 10 −4 photons cm −2 s −1 .creates a large uncertainty in the RMFs.This modeling gives good fits, as shown in Figure 5.The best-fit parameters are summarized in Table 3.

Charge Exchange Contribution
We have applied the GMCA method to CCD data of XMM observations of M82, and for the first time constrained the location of the component with enhanced CX emission in the starburst galaxy.Analyzing the RGS spectra from the Peak region in the south outflow, we have detected an unusually high O VII Heα f/r ratio as well as clearly resolved emission lines selectively enhanced in the CX process such as C VI Kγ, N VII Lyα, and O VIII Lyα .The high ratio and the presence of line emission using the same parts of the RGS as used here have been previously discussed (e.g., Liu et al. 2011;Zhang et al. 2014).Our results confirm their conclusions and indicate that CX emission accounts for ∼50%, ∼30%, ∼60%, ∼40%, and ∼30% of the total flux of the C VI Kγ, N VII Lyα, O VII f, O VII r, and O VIII Lyα lines, respectively.Some previous work (e.g., Zhang et al. 2014;Lopez et al. 2020) discussed the CX contribution to the Ne VII and Mg IX triplets in addition to the above line emission.For example, Zhang et al. (2014) claimed CX is responsible for more than 30% of the total flux of Ne VII f although the contribution in our spectral fits is less than ∼10%.The discrepancy between their and our results arises from the selection of the recipient plasma in the CX modeling.Figure 6 This change in the modeled CX emission implies that the abundance measurements in the hot plasmas must be updated as well.In Figure 7, we compare the obtained abundance pattern with those given by Zhang et al. (2014) and Lopez et al. (2020).It is worth pointing out that overall our abundances tend to be lower except for carbon, which shows an opposite trend, even if we consider the differences between the 11 To measure the Ne f and r flux, we delete these two transitions in the XSPEC model package and perform the spectral fittings with the best-fit model in Table 3 plus additional Gaussians accounting for the two lines.Detailed procedures are described in Section 5 in Suzuki et al. (2020).
observations.The cause of the abundance variations could be explained by a difference in the source regions due to the different roll angles and/or the position-to-position fluctuations of the column density, although it is hard to conclude it.A collection of core collapse supernova (CCSN) events arising from star formation would naturally lead to supersolar abundance ratios of α elements versus Fe (e.g., Nomoto et al. 2006) et al. (2004) reported high ratios of α elements including O to Fe from near-infrared observations of stars and cold gas in the disk, the puzzling [O/Fe] behavior should be interpreted as the result of significant O depletion rather than Fe enhancement.Some O might suffer severe dust depletion since CCSNe produce large amounts of dust (e.g., Todini & Ferrara 2001).However, in this scenario, other elements are also expected to be depleted (e.g., Savage & Sembach 1996;Boogert et al. 2015), o confirmation is needed to determine if the observed abundance pattern can be explained by a model that includes dust evolution.An alternative possibility is the uncertainties due to the modeling (e.g., are the abundances the same among the three plasma and CX components?Are kT e,CX and kT e,cold really the same?).In either case, follow-up observation with microcalorimeters on board future satellites such as XRISM will offer better understanding of the chemical properties of this starburst galaxy.

Charge Exchange Geometry
The maps of the three components extracted in the GMCA analysis and their comparison with the gas distribution provide an unbiased look at the geometry of the CX component of the galaxy.Figure 8 shows the location of each component overlaid on the Hα image with the Advanced Camera for Surveys (ACS) on board the Hubble Space Telescope (HST; Mutchler et al. 2007).We find Component 1 with enhanced CX emission to be spatially coincident with the distribution of Hα filaments extending to both the northern and southern sides of the galactic disk.Numerical (e.g., Cooper et al. 2008) and hydrodynamical simulations (e.g., Melioli et al. 2013;Schneider et al. 2020) show that these vast Hα structures are created by swept-up gas that has been dragged by plasma outflows.Our results support the idea that the CX process occurs at the interface between the plasma and gas components within the outflows; similar conclusions were reached by Lallement (2004) and Wu et al. (2020) from numerical simulations.
Another clue to the CX geometry comes from the fraction of the CX-emitting volume V CX with respect to the outflow volume V pl .We can estimate V CX to be 3.6 × 10 56 cm 3 (see Appendix), while V pl ∼ 2.1 × 10 65 cm 3 , assuming the morphology is a cylinder with a radius of 1 kpc ( 1 = ¢) and a height of 2 kpc ( 2 = ¢).The fraction ∼10 −9 is significantly smaller than the typical filling factor of clouds within an outflow (0.001-0.1;Sharp & Bland-Hawthorn 2010;Müller-Sánchez et al. 2013), indicating the CX occurs in an extremely limited region near the interface zone.This interpretation is indirectly supported by the fact that the temperature kT e,CX ∼ 0.2 keV of the plasma acting as an electron receiver the CX process is much lower than the other plasma temperatures (kT e,med ∼ 0.5 keV and kT e,hot ∼ 0.9 keV), which account for most of thermal X-rays.Neutral hydrogen, typically the major electron donor in the CX process, is easily ionized so that it cannot deeply penetrate into the hot plasma (e.g., Wise & Sarazin 1989).The low kT e,CX can  c The unit is 10 −4 photons keV -1 cm -2 s -1 at 1 keV.d Degree of freedom.
be explained if the CX process mainly occurs in a layer where the plasma temperature rapidly varies due to thermal conduction by the gas.

Conclusions
We have performed an analysis of X-ray emission from M82, a prototype starburst galaxy, obtained with XMM/ Newton in order to investigate the CX flux contribution and its geometry.We have applied a blind source separation method to analyze the CCD data, and identified the spatial distribution of a component with the enhanced O-K lines expected from the CX process in a starburst galaxy for the first time.Based on the image analysis, we have analyzed the RGS data extracted from a compact peak ( 2 ~¢) in the southern outflow and detected a high forbidden-to-resonance ratio in the O VII Heα triplet as well as several emission lines enhanced in the CX process such as C V Kγ, N VII, and O VIII Lyα.The RGS spectra of all observations are well fitted with a model that consists of three different plasma temperatures (∼0.2 keV, ∼0.5 keV, and ∼0.9 keV) and CX components with an additional nonthermal component.The CX emission accounts for ∼50%, ∼30%, ∼60%, ∼40%, and ∼30% of the total flux of C VI, N VII, O VII f, O VII r, and O VIII Lyα lines, respectively, although the CX contributions to the emission lines of Ne and Mg are less than ∼10%.Including this CX emission component primarily affects the measured abundance measurement of these light elements in the thermal plasma components, tending toward  lower values relative to earlier calculations.The temperature of the plasma acting as an electron receiver in the CX process is significantly lower than the plasma components that emit most of X-rays.From the low temperature and an estimation of the CX-emitting volume, the CX primarily occurs in a thin region near the interface of the plasma and gas whose temperature rapidly decreases due to thermal conduction.
and n r H are the total density of neutral hydrogen and helium of the donor gas and the hydrogen density of the receiver plasma, respectively.The average VEM CX is estimated to be 0.9 × 10 57 cm −3 by applying Equation (1) to the values in Table 2.In order to calculate V CX , we need to obtain ( -, where n r e is the electron density of the receiver plasma.Assuming an average VEM pl = VEM cold + VEM med + VEM hot = 8.1 × 10 61 cm −3 , we can estimate n r H to be ∼0.018cm 3 .Here, we have used the relation n n 1.2 r e r H

Figure 1 .
Figure 1.Results of the GMCA analysis applied to the M82.(a) Output spectra of the four components.Components 1 (purple), 2 (green), 3 (blue), and 4 (orange) predominantly consist of CX, hot plasma, nonthermal emission, and line emission due to the detector background, respectively.The input data before the component extraction (red) is given by the sum of the components.Counts maps of (b) the input and (c-i)-(c-iii) the components.(d-i)-(d-iii) Fraction of each component, defined as f i = N i /∑ i=1−4 N i , where N i is the number of photons in the corresponding count map.The regions enclosed by the magenta lines in panel (d) are the three representative regions whose spectra are plotted in Figure 2. The white circle indicates the Peak region as shown in Figure 3(a).

Figure 3 .
Figure 3. (a) Same as Figures 1(b)-(i) but we show the location of the Peak region (green circle) and RGS spectral extraction regions (cyan) for Obs.ID = 0112290201 and 0206080101, whose roll angles are the maximum and minimum, respectively.The green box displays the area of Figure 7.(b) RGS1+2 first-(black) and second-order (red) spectra where all used data sets are integrated.
(a)  shows the Ne f flux directly estimated with RGS data via the manner 11 and the flux level expected from the CX process as a function of plasma temperature.In the flux calculation, we have assumed that 60% of the observed O VII f line is emitted in the CX process and computed the flux ratio of Ne IX f/O VII f with the same package in Section 3.3 using the abundance ratio Z O /Z Ne = 1.1/0.4from Table2.The Ne f flux curve significantly varies with the plasma temperature since the population of H-like and bare ions that can emit Ne f lines after electron transfer is sensitive to the plasma temperature (Figure6(b)).If we employ kT e,CX ∼ 0.2 keV as demonstrated in Section 3.3, the CX contribution to Ne f is less than ∼10%.On the other hand,Zhang et al. (2014) assume kT e,CX to be ∼0.6 keV, suggesting the CX contribution is overestimated.

Figure 4 .
Figure 4. Comparison between the observational and theoretical (a) O VII f/r ratio and (b) O VIII Lyα flux.The purple area (line) indicates the observed line ratio (flux).The red curve in panel (a) is the ratio predicted from collisional ionization equilibrium plasma as a function of the plasma temperature.The orange, green, and blue curves show the ratios in the CX cases of υ col = 500 km s −1 , 1000 km s −1 , and 1500 km s −1 , respectively.In panel (b), because the three curves completely overlap each other, we only plot the case of υ col = 1000 km s −1 .The gray dot line and arrow represent the constrained upper limit of kT e,CX = 0.35 keV.

Figure 5 .
Figure 5. (a) first-(black) and second-order (red) spectra of Obs.ID = 0112290201 plotted with the best-fit model.The blue, magenta, orange, green, and black curves represent the CX model, the cold, medium hot, hot plasma models, and the sum of the models fit to RGS1+2 first-order data.(b)-(f) Same as panel (a) but for Obs.ID = 0206080101, 0206080101, 0560590201, 0560590301, 0870940101, and 0870940101.

Figure 6 .
Figure 6.(a) Comparison between the observational Ne VII f flux (purple) and that theoretically expected in the CX process with υ col = 500 km s −1 (orange), 1000 km s −1 (blue), and 1500 km s −1 (green), respectively.The gray hatched (∼0.2 keV) and filled (∼0.6 keV) areas correspond to the constrained kT e,CX range in our results and in Zhang et al. (2014), respectively.(b) Ion fraction of Ne in the He-like (green), H-like (blue), and bare state (red) as a function of the plasma temperature.

Figure 7 .
Figure 7. Elemental abundances relative to the solar values of Wilms et al. (2000).The magenta and light green hatched areas correspond to previous measurements in Zhang et al. (2014) and Region S1 by Lopez et al. (2020), that are also normalized by the solar abundances of Wilms et al. (2000) and plotted for comparison.
is ∼14 cm −3 under the assumption of solar abundances of H and He.n r H is given by the EM of the plasma components EM pl , defined as n

Table 2
The Intensities of the O and Ne Emission . Although the [Ne/Fe]( models, [O/Fe] ∼ 0.07 is much smaller.Given that Origlia

Table 3
Best-fit Model Parameters of the Spectra from All Observations aThe volume emission measure integrated over the line of sight, i.e., ∫n e n (H+He) dV.b The volume emission measure integrated over the line of sight, i.e., ∫n e n H dV.