The XMM-Newton Line Emission Analysis Program (X-LEAP). I. Emission-line Survey of O vii, O viii, and Fe L-shell Transitions

The XMM-Newton Line Emission Analysis Program (X-LEAP) is designed to study diffuse X-ray emissions from the Milky Way (MW) hot gas, as well as emissions from the foreground solar wind charge exchange (SWCX). This paper presents an all-sky survey of spectral feature intensities corresponding to the O vii, O viii, and iron L-shell (Fe-L) emissions. These intensities are derived from 5418 selected XMM-Newton observations with long exposure times and minimal contamination from point or extended sources. For 90% of the measured intensities, the values are within ≈2–18 photons cm−2 s−1 sr−1 (line unit (LU)), ≈0–8 LU, and ≈0–9 LU, respectively. We report long-term variations in O vii and O viii intensities over 22 yr, closely correlating with the solar cycle and attributed to SWCX emissions. These variations contribute ∼30% and ∼20% to the observed intensities on average and peak at ≈4 and ≈1 LU during solar maxima. We also find evidence of short-term and spatial variations in SWCX, indicating the need for a more refined SWCX model in future studies. In addition, we present SWCX- and absorption-corrected all-sky maps for a better view of the MW hot gas emission. These maps show a gradual decrease in oxygen intensity moving away from the Galactic center and a concentration of Fe-L intensity in the Galactic bubbles and disk.


INTRODUCTION
The circumgalactic medium (CGM) is a diffuse, multiphase gas surrounding the galaxy.It plays an important role in the formation, evolution, and interactions of galaxies (see the reviews Tumlinson et al. 2017;Donahue & Voit 2022).In particular, the CGM encodes the feedback materials, metals, and energy ejected from the galaxy disk, as well as the accreted gas from the intergalactic medium (IGM).Studying the properties and distribution of the CGM provides valuable constraints on these uncertain processes.
Over the last two decades, the multi-phase CGM in the Milky Way (MW) has been extensively explored using multi-wavelength observations (see the review Putman et al. 2012).X-ray observations reveal that the hot CGM (T ≳ 10 6 K) is a massive component in the multi-phase medium (e.g., Snowden et al. 1995;Henley & Shelton 2010, 2012).Notably, the stars and interstellar medium (ISM) in the MW only account for ≈ 30-50% of the baryons, expected from the cosmic average baryonic fraction (e.g., Sommer-Larsen 2006;Mc-Gaugh et al. 2010).This discrepancy suggests a deficit of ≈ 3 × 10 11 M ⊙ (Anderson & Bregman 2010), positioning the hot CGM as a principal candidate for the "missing baryons".
The hot gaseous phase is primarily probed through the emission or absorption lines of highly ionized metals, such as the O VII Kα and O VIII Kα.These two ions exist at temperatures of ≈ 1-2 × 10 6 K assuming collisional ionization equilibrium (CIE), which is consistent with the expected virial temperature of the MW.Extensive efforts have been made to estimate the total hot gas mass using the oxygen emission lines (e.g., Gupta et al. 2012;Miller & Bregman 2013;Miller & Bregman 2015;Nakashima et al. 2018;Kaaret et al. 2020).However, these estimates yield considerable uncertainty, ranging from ≈ 3-4 × 10 10 M ⊙ to 1-2 × 10 11 M ⊙ (Miller & Bregman 2015;Faerman et al. 2017).
This uncertainty primarily arises from two factors.First, the emission or absorption column is affected by plasma density, temperature, and metallicity together.The emission line intensity scales as I ∝ n 2 e ϵ(T )Z, while the absorption column scales as N ∝ n e f (T )Z, where ϵ and f are the emissivity and ionization fraction, respectively.Accurate assessments of temperature and metallicity are crucial to reliably infer the hot gas density from X-ray observations.
Second, the MW hot gas emission is contaminated by foreground emissions from the solar wind charge exchange (SWCX) and the Local Hot Bubble (LHB).The SWCX occurs when solar wind ions exchange charges with neutral atoms in the Solar System.Its intensity varies temporally and spatially (see Kuntz 2019, for a review), contributing a large amount to the diffuse soft X-ray background (SXRB) at energies below 1 keV (e.g., Cravens 2000;Koutroumpa et al. 2009; Uprety et al. 2016).The LHB is thought to be an irregular "local cavity" extending ∼ 100 pc from the Sun (Lallement et al. 2003).Although the origin of the LHB has many explanations (e.g., Maíz-Apellániz 2001;Linsky & Redfield 2021;Zucker et al. 2022), its gas temperature is measured to be T ≈ 10 6 K (e.g., McCammon & Sanders 1990;Snowden et al. 2000;Liu et al. 2016;Yeung et al. 2023), a temperature high enough to emit X-rays (Sanders et al. 1977).Shadowing studies toward nearby molecular clouds indicate that the LHB contributes negligibly to the observed O VIII emission but accounts for part of the O VII emission (e.g., Smith et al. 2007;Koutroumpa et al. 2011).
The XMM-Newton archive offers unique opportunities to overcome these challenges with observations over two decades.It provides a deep, sensitive, and spatially resolved view of the X-ray sky, and also allows us to consider long-term temporal variations associated with solar activity.The high sensitivity enables us to detect faint emissions due to the hot gas.The spatial resolution enables the removal of contamination due to other X-ray sources, improving the accuracy of the line intensity measurements.
We design the XMM-Newton Line Emission Analysis Program (X-LEAP) aiming to study the diffuse Xray emissions from the foreground SWCX and the MW hot gas.In this first paper, we conduct a line emission survey, which measures three specific spectral features in the SXRB: O VII Kα (≈ 0.57 keV), O VIII Kα (≈ 0.65 keV), and transitions due to the iron L-shell (Fe-L; ≈ 0.8 keV).They are referred to as emission lines for simplicity hereafter, although they are combinations of individual lines.These lines are the strongest in both SWCX and MW hot gas emissions.The oxygen lines serve as primary tracers of the hot gas, while the Fe-L line is a frequently detected feature in SXRB spectra peaking in emissivity at ≈ 7 × 10 6 K.This temperature is above MW's virial temperature, suggesting that the Fe-L feature may trace an even hotter phase of the MW gas.Recent CGM studies show evidence of such hotter component with a super-virial temperature ≈ 10 7 K, associated with the Fe-L line (e.g., Das et al. 2019;Bluem et al. 2022;Ponti et al. 2023;Bhattacharyya et al. 2023).The origin of this component is debated, with no consensus among these studies.Including this feature in our survey is crucial to further explore and understand its properties and origin.
Because of the complexities of measuring these lines, we adopt a line extraction method introduced by Henley & Shelton (2012, hereafter HS12).This method does not estimate temperature or emission measure but extracts line intensities directly from the spectra.This method is insensitive to assumptions of the underlying physical model of the hot gas, SWCX, and other background emissions, reducing measurement uncertainties.This is particularly important because the SXRB includes contributions from multiple emission sources that are difficult to separate.
Here, we present the line measures of the O VII, O VIII, and Fe-L lines from the 22-year XMM-Newton archive data, along with initial findings, including intensity fluctuations in foreground SWCX and all-sky intensity distributions of the MW.
The paper is structured as follows: Section 1 outlines the data selection and reduction process.The emission components in the SXRB and our approach to modeling and fitting are introduced in Section 3. In Section 4, we present a catalog of observed O VII, O VIII, and Fe-L intensities.We also discuss the distribution and correlations of these intensities.Additionally, we compare our intensity measurements with those from HS12.Our approach to characterizing long-term SWCX variation is described in Section 5. Section 6 introduces methods  1.
for SWCX-and absorption-corrections, and presents the corrected all-sky maps.Section 7 discusses further implications of the sample, including the use of O VII line centroids as a diagnostic of solar activity, the evidence of magnetospheric SWCX, and the spatial dependence of SWCX on ecliptic latitude.We also discuss the forthcoming works for this program.Finally, Section 8 summarizes the key findings.

DATA SELECTION AND REDUCTION
We present a new survey of O VII, O VIII, and Fe-L emission features in the SXRB using the XMM-Newton archive data before 2022 January 1st.From 15,035 total observations, our data selection and reduction processes yield 6155 data points across 6090 observations, with some observations containing multiple exposures.The dataset is further reduced to 5418 observations, based on the quality check of spectral fitting in Section 3.10.These observations are optimized to study the soft X-ray emissions of the MW hot gas and the SWCX.In particular, they have relatively long exposure times and low contamination due to soft proton backgrounds, external galaxies, and galaxy clusters.The selection criteria and data processing are summarized in Figure 1 and detailed in this section.

Initial Observation Selection
In the X-LEAP program, we only consider data obtained from the EPIC-MOS detectors (MOS1 and MOS2) because of their large field of view (FOV, ≈ 30 ′ ) and high sensitivity at energies of the interested lines.We consider only observations with EPIC-MOS exposure times longer than 7 ks to ensure a reasonable signalto-noise ratio, enabling reliable detection of faint emission.In addition, we only consider observations in "Full Frame" or "Partial Window" modes.These modes observe less bright and/or time-varying X-ray sources compared to other modes and read out all seven or most CCDs (CCD#2-7).This ensures full or near-full FOV coverage, making them more suitable for diffuse emission studies.Consequently, 4486 observations are ignored from the above criteria.

Data Processing
The data processing starts with the observation data files (ODFs) obtained from the XMM-Newton Science Archive1 .For each ODF, the calibrated photon event files for both EPIC-MOS detectors are generated using the XMM-SAS emchain script.To improve data quality, public software tools are used, including HEAsoft version 6.30.12 , SAS version 20.0.03 (Gabriel et al. 2004), XMM-Newton Extended Source Analysis Software4 (XMM-ESAS; Kuntz & Snowden 2008;Snowden & Kuntz 2011) and FTOOLS5 (Blackburn 1995).The following subsections outline the detailed steps adopted to clean and optimize these calibrated photon event files.

Soft Proton Background
The soft proton background (SPB) arises from interactions between the EPIC-MOS detectors and soft protons (SPs), observable as flares in the photon count rate.These flares can last for seconds to hours and significantly contaminate observed data (Snowden et al. 2004).To minimize this contamination, we utilize the XMM-SAS espfilt script.This script identifies the flares as time intervals with count rates 2.5σ above the median, assuming a Gaussian distribution of the observed photon count rates.The resulting Good Time Interval (GTI) file is used to generate the SP-filtered event file.We only include observations with at least one exposure having a GTI > 5 ks in the following analysis.This flare-filtering process cannot remove all SP contamination (Henley & Shelton 2010).The remaining SPB contamination is modeled as a power law component in the subsequent spectral analysis (Section 3.7; see also Kuntz & Snowden 2008).The flare-filtering process removes 942 observations from the final sample.

The Quiescent Particle Background
The quiescent particle background (QPB) is induced when high-energy particles interact with the detectors and the surrounding instruments.It includes both continuum and instrumental line emissions.The continuum and weak lines are calculated for each observation using the XMM-ESAS mos back script (Kuntz & Snowden 2008), and then subtracted to obtain the QPB-cleaned SXRB spectrum.
Besides the modeled QPB spectrum, the QPB contains two strong instrumental lines: Al Kα and Si Kα, located at 1.49 keV and 1.74 keV, respectively.These lines cannot be well-modeled by the QPB spectrum.They are represented as two Gaussian lines in the spectral model, detailed in Section 3.8.

Unusable CCD Removal
Among all seven CCD chips in each EPIC-MOS detector, some chips may occasionally be exceptionally bright or completely nonfunctional (Kuntz & Snowden 2008).In particular, MOS1 CCD#3 is damaged and no longer in use due to a micrometeorite strike on 2012 December 11th.Such anomalous CCD chips are identified using the XMM-SAS emanom script.It computes the count rate ratio between the 2.5-5.0 keV and 0.4-0.8keV energy bands for each functional chip, and compares to a predefined threshold.Chips with ratios above this threshold are classified as "good".Conversely, nonfunctional chips, unable to generate photon counts, are classified as "off" (Kuntz & Snowden 2008).Our analysis only considers exposures with at least three "good" CCD chips, ensuring coverage of > 40% of the FOV.In total, 21 observations are excluded due to insufficient CCD chips.

Point Source Removal
Point sources in the FOV may contaminate the diffuse emission.They are identified and excluded using the XMM-SAS cheese script in this study.We customize the cheese parameters to optimize point source detection.First, the point spread function (PSF) threshold scale is fixed to 0.1, allowing the removal of a larger fraction of the point-source flux.Second, the minimum separation for point sources is set to 2.5 ′′ , enabling the detection of point sources in crowded fields.Finally, the point-source flux threshold rate is set to 3 × 10 −15 erg cm −2 s −1 , improving the identification and exclusion of fainter sources.Following these adjustments, the cheese script is performed to search point sources in three energy bands: 400-1250 eV, 1250-7200 eV, and 400-7200 eV.This approach ensures the detection of soft X-ray sources, hard X-ray sources, and those with broad energy emissions.
The script identifies the locations and sizes of detected point sources, recording them in a region file.For each observation, we combine the cheese detected regions from all EPIC-MOS exposures to create a master mask, leaving only the unmasked areas for subsequent analysis.If the master mask covers > 50% of the FOV, we exclude the observation to prevent potential contamination from extended sources in the FOV, which are not part of the SXRB we aim to measure.As a result, 705 observations are removed because of this exclusion criterion.
2.2.5.Extended X-ray Sources: Galaxy Clusters and Nearby Galaxies Galaxy clusters and nearby galaxies are extended sources emitting in the soft X-ray band.They serve as background contamination in studies of the MW hot gas or SWCX emission.They can neither be excluded using cheese because they lack clear boundaries, nor be characterized by a spectral model due to their complex emission spectra.To minimize this contamination, we estimate their angular size and mask them from the FOV.We assume all these sources are spherical for simplicity.
For galaxy clusters, we adopt the Meta-Catalogue of X-ray Detected Clusters of Galaxies (MCXC; Piffaretti et al. 2011), a catalog of nearby X-ray bright galaxy clusters, which provides the redshift, physical radius (r 5006 ), and position (i.e., Galactic longitude l and latitude b).
The angular size of each cluster is calculated using the r 500 , beyond which the X-ray is negligible in observation.
For nearby galaxies, we adopt the Kourkchi & Tully (2017) galaxy catalog, which provides distance, position, and K s -band luminosity over 15,000 galaxies within 3500 km s −1 .Since only a few massive galaxies have their hot gas halos detected (e.g., Anderson & Bregman 2010;Bogdán et al. 2013), we consider only those within 5 Mpc with stellar masses above 10 10 M ⊙ .Galaxies beyond 5 Mpc have detectable X-ray emissions primarily from their centers, which can be easily masked using the cheese script.Stellar masses are derived using the K s -band luminosity and the infrared mass-to-light ratio from Kourkchi & Tully (2017).Besides the galactic center, the disk may also emit detectable X-rays (e.g., Li & Wang 2013).Thus, we adopt an 8 kpc detection radius, a radius that covers most emissions from the nearest galaxies, such as the Triangulum Galaxy (M33), the Large Magellanic Cloud (LMC), and the Small Magellanic Cloud (SMC).
For each XMM-Newton observation, we calculate the overlap area between the FOV and the extended sources using the circle-circle intersection formula: (1) where r is the FOV radius of EPIC-MOS fixed at 14 ′ , R is the angular radius of the source, and d is the angular separation between the source and the observation direction.We exclude observations where the overlap area is > 50% of the FOV in the following analysis.For observations with < 50% overlap, we add their coordinates and estimated radii into the pre-generated master mask.This approach automatically excludes them along with those detected point sources.Figure 2 illustrates an example of the exclusion results for galaxy clusters and nearby galaxies.
This method may not always yield completely accurate results in all scenarios due to the assumptions involved in size estimation.To address this, we introduce a "flag" parameter.A flag value of "1" or "2" is assigned to observations in which galaxy clusters and nearby galaxies cover between 10% and 50% of the FOV (See Table 1 for all observation flags).The "flag" parameter allows a manual exclusion of observations if needed.
In addition, an X-ray bright diffuse enhancement is discovered around M31, extending up to ≈ 10 • -20 • (Qu et al. 2021), which is much larger than the estimated radius.Thus, we manually flag all observations within 4.4 degrees from M31's center based on a visual inspection of the all-sky intensity maps as shown in Section 4.
In total, 670 and 872 observations are excluded because of the large covering fractions (> 50%) of galaxy clusters and nearby galaxies, respectively.

SXRB Spectra
The SXRB spectra are extracted from the SP-filtered photon event lists using the entire FOV, excluding any anomalous CCDs and masked regions.This extraction is performed using the XMM-ESAS mos-spectra script, which also generates the corresponding redistribution matrix files (RMFs) and ancillary response files (ARFs).Then, we use the grppha script in FTOOLS to group the SXRB spectra with associated QPB spectra, RMFs, and ARF files and bin them with a minimum of 100 counts per channel.The SXRB spectra are extracted in the 0.3-3.2keV band, which includes all three emission lines of interest (i.e., O VII, O VIII, and Fe-L) and is broad enough to model the residual SPB.

COMPONENTS AND SPECTRAL MODEL
In the X-LEAP program, three spectral features (i.e., O VII, O VIII, and Fe-L) are of particular interest to investigate the hot gas in the MW and the SWCX variations.An accurate measure of their intensities requires precise spectral modeling to separate them from other components.In this section, we describe the components considered in this work and their implementations in spectral fitting.

Overview of the Spectral Components
The SXRB spectrum is composed of multiple spectral components, including the internal "quiescent" background (Section 2.2.2), the external "flaring" background (SPB; Section 2.2.1), the SWCX emission, the LHB emission, the MW CGM emission, and the cosmic X-ray background (CXB) emission.Both the X-ray emissions from the MW CGM and the CXB are subject to absorption by gas and dust in the MW (see Section 3.5 for details).
The EPIC-MOS detectors are configured to observe the same sky regions simultaneously.To amplify signals from diffuse hot gas emissions, we adopt a joint-fitting approach, treating both closely-timed EPIC-MOS spectra as a paired spectrum.Specifically, we link all parameters corresponding to consistent components across both detectors in the fitting process.Parameters associated with the instrumental backgrounds are not linked because the instrument background observed in one MOS detector slightly differs in the other (Snowden & Kuntz 2014).The spectral fitting is done using XSPEC version 12.12.1 7 (Arnaud 1996) with metal abundances from Lodders (2003).This table is chosen for its widespread 7 https://heasarc.gsfc.nasa.gov/xanadu/xspec/usage and because its oxygen and iron abundances align more closely with recent measurements, compared to the default table from Anders & Grevesse (1989).The following subsections detail how each component is modeled in XSPEC.The spectral components and their parameter settings are summarized in Table 2.In Figure 3, we show an example of a paired SXRB spectrum along with its best-fit model and all model components.

The O VII, O VIII and Fe-L Emission Lines
The O VII Kα, O VIII Kα, and Fe-L lines represent three spectral features in the SXRB.These lines do not refer to single ionic transitions.Instead, each line is a set of transitions with similar transition energies originating from specific ions.
The O VII and O VIII lines originate from both hot ionized gases and the SWCX process.The O VII line is a triplet consisting of the forbidden line at 561 eV, the intercombination line at 569 eV, and the resonance line at 574 eV.The O VIII line is a doublet consisting of two close lines at 653.5 eV and 653.7 eV (Foster et al. 2012).The EPIC-MOS spectral resolution is ≈ 40 eV at 500 eV (Lumb et al. 2012), much larger than the splitting of both the O VII triplet (≈ 13 eV) and the O VIII doublet (< 1 eV), so we model these two lines as Gaussian functions.The O VII line centroid is allowed to vary in the 535-595 eV energy range, whereas the O VIII line centroid is fixed at 653.6 eV.Given the spectral resolution exceeds the intrinsic line widths and the energy splitting, the observed line widths are mainly influenced by instrumental resolution.Consequently, we fix the line widths at 0, while leaving the normalizations of the Gaussian functions as free parameters.
The Fe-L line is composed of a set of iron emission lines in the ≈ 0.7 − 1.0 keV range from Fe XVI to Fe XXI.
The most prominent lines include Fe XVII at ≈ 0.83 keV, Fe XVIII at ≈ 0.87 keV, and Fe XIX at ≈ 0.92 keV (Foster et al. 2012).These lines contribute the most to the total emissivity in the ≈ 0.7 − 1.0 keV range at temperatures between 0.2 − 1.0 × 10 7 K.Previous studies suggest that this feature may be indicative of a hotter phase of the MW CGM with kT ≈ 0.41-0.72 keV (e.g., Das et al. 2019;Ponti et al. 2023).However, it may be contaminated by the hot corona of unresolved stars (e.g., Wulf et al. 2019).We investigate this feature to gain a deeper insight into the nature and origin of this feature.It is modeled as a Gaussian function with the line centroid varying within the 0.7-1.0keV energy range.The line width is fixed at 50 eV, consistent with the standard deviation of the transition energies of these iron lines.The normalization is varied as a free parameter.
We employ the Astrophysical Plasma Emission Code (APEC) model for MW and LHB emissions (Smith et al. 2001), which will be introduced in the following sections.To avoid double-counting the line emissions, we disable them in the APEC model.Adopting the approach of Lei et al. (2009), we set emissivities of the O VII and O VIII Kα lines to zero in the APEC line emissivity file (apec v3.0.9 line.fits,Foster et al. 2012).Note that we do not disable the O VII Kβ line at 666 eV near the O VIII Kα line at 653 eV.This may slightly lower the observed O VIII intensity.For the Fe-L emission, we examine the emissivity data in the same file within kT = 0.5-0.8keV.At each temperature, we identify the 50 strongest emission lines and disable only iron lines by setting their emissivities to zero.
Figure 4 shows the O VII, O VIII, and Fe-L emissivities at different APEC temperatures using abundances from Lodders (2003).It demonstrates that in a single-temperature plasma under CIE, the emissivities for O VII, O VIII, and Fe-L emissivities peak at ≈ 2 × 10 6 K, ≈ 3 × 10 6 K, and ≈ 7 × 10 6 K, respectively.Since line intensity scales as I ∝ n 2 e ϵ(T )Z, a high intensity implies a plasma temperature close to these peak values.Moreover, the intensity ratios, such as I OVII /I OVIII ∝ ϵ OVII (T )/ϵ OVIII (T ), provide a direct measure of plasma temperature.These temperature diagnostics have been extensively explored in Qu et al. (2023, hereafter Paper II).

the MW CGM
The diffuse gas surrounding the MW, known as the MW CGM, contains a hot phase with temperatures typically ranging from 1-3 × 10 6 K, leading to X-ray emissions.Such hot gas is thought to be an optically thin plasma in CIE and is commonly modeled using the APEC model (represented as apec in XSPEC; Smith et al. 2001).In our MW APEC model, we fix the metal abundance at 0.5 Z ⊙ but allow temperature and normalization to vary.The hot gas is expected to have temperatures kT ≈ 0.1-0.3keV, which is most sensitive to the O VII and O VIII lines we disabled (Miller & Bregman 2016).Thus, we adopt a wider allowed temperature range of 0.1-1.0keV for this model.
It is worth noting that a cooler phase with kT < 0.1 keV might exist (e.g., Kuntz & Snowden 2000), potentially enhancing the lower end of the SXRB spectrum.However, this component is barely detectable by XMM-Newton or Chandra and would likely overlap with the LHB component described in Section 3.6.Furthermore, our line extraction method directly measures line intensity from the spectrum, making it insensitive to assumptions regarding hot gas models, as discussed in Introduction 1.

The CXB
The CXB is defined as the total emission from all the extragalactic sources in the X-ray band, which dominates the SXRB spectrum above 1 keV.The primary contributors to the CXB emission are active galactic nuclei (AGNs), with some contributions from galaxy clusters and starburst galaxies (Gilli et al. 2007;Moretti et al. 2009).The spectral characteristics of the CXB, largely influenced by the AGNs, typically exhibit power law distributions.This is a consequence of synchrotron radiation, a common emission mechanism in AGNs (Piconcelli et al. 2005).We model the CXB using a powerlaw (powerlaw in XSPEC) with a photon index fixed to Γ = 1.46, which represents the average spectral shape of the distant AGN population (Chen et al. 1997).Although the CXB is generally isotropic, it still exhibits some level of variation across the sky.Therefore, the normalization of CXB remains a free parameter for each observation.

MW Absorption
Emissions from both MW CGM and CXB travel through cold dust and clouds, leading to absorption in soft X-rays.This absorption is modeled using the phabs absorption model (Balucinska-Church & McCammon 1992) in XSPEC, with an updated He cross-sections from Yan et al. (1998).The absorption efficiency is determined by the hydrogen column density (N H ), a key input for the phabs model.
We estimate N H values based on the 2018 Planck dust emission survey, which measures the dust opacity at 353 GHz (τ 353 ; Planck Collaboration et al. 2014).This opacity traces both the molecular and atomic portions of the absorbing medium, a good tracer for N H particularly in regions containing both H 2 and H I. Specifically, the dust-inferred N H , N H,dust , is calculated based on the visible extinction map 8 derived from the dust survey, using the relation N H (cm −2 ) = 2.21×10 21 A V (Güver & Özel 2009).For comparison, we also calculate the H I-inferred N H using the HI4PI survey (HI4PI Collaboration et al.

2016).
For each observation, we determine the N H using the averaged N H within a 0.3 • radius, slightly larger than the FOV of EPIC-MOS.
The results show that the N H,dust values align with those inferred from H I in over 50% of the observations, specifically at |b| > 30 • , without exceeding by more than a factor of one.However, at |b| < 30 • -a region rich in molecular hydrogen -the N H,dust median is approximately 2.5 times higher than the H I inferred N H median.This significant difference suggests that N H,dust more accurately accounts for the total absorbing medium, making it a better tracer, specifically in H 2 dense regions.
The derived N H,dust values are also used in Section 6 for estimating the deabsorbed MW emission.

The LHB
The LHB is a cavity filled with hot and diffuse gas surrounding the Solar System.It extends ≈ 100 pc in all directions away from the Sun (Liu et al. 2016), with an estimated temperature of kT ≈ 0.084 keV (e.g., Bluem et al. 2022;Yeung et al. 2023) and a relatively constant electron density of 4 × 10 −3 cm −3 (Yeung et al. 2023).We model the LHB's emission using another APEC model with the temperature, abundance, and redshift parameters fixed at 0.084 keV, 1.0, and 0, respectively.The normalization remains a free parameter.We note that the O VII, O VIII, and Fe-L lines are disabled in both the MW CGM and LHB APEC models to prevent 8 https://pla.esac.esa.int/#mapsdouble-counting these intensities when measuring them directly using Gaussian functions.

Residual Soft Proton Contamination
The SPB is part of the instrumental background, which primarily comes from soft protons interacting directly with the detector.It can not be fully removed by the flare-filtering process discussed in Section 2.2.1.Here, we model the residual SP contamination using a power-law not folded through the instrumental effective areas, since these protons do not interact with the parts of the instrument that shape the effective area.The power-law index parameter is allowed to vary between 0.2-1.3following the XMM-ESAS manual (Snowden & Kuntz 2011).The normalization is a free parameter, allowed to vary across observations.

The Instrumental Lines
As previously discussed in Section 2.2.2, the Al and Si instrumental lines at 1.49 keV and 1.74 keV can not be adequately characterized by the mos back script.They are modeled as two Gaussian functions with line centroids confined to the ranges of 1.46-1.52keV and 1.72-1.78keV, respectively.The line widths are fixed at 0, and the normalizations remain as free parameters.

Model Fitting
In XSPEC, our spectral model is structured as constant(phabs(apec MW + powerlaw CXB ) + gaussian OVII + gaussian OVIII + gaussian FeL + apec LHB ) + gaussian Al + gaussian Si , complemented by a residual SPB (powerlaw SPB ) unfolded with the response matrix.From left to right, the constant accounts for the total usable FOV for the detector, which is calculated by the scale proton script in XSPEC.This factor allows us to directly derive FOV averaged parameters.The phabs(apec MW + powerlaw CXB ) shows how the absorption of MW hot gas emission and the CXB is corrected in XSPEC.The following three Gaussian functions represent the three lines of interest.The apec LHB after that represents the unabsorbed LHB emission.The two Gaussian functions at the end represent the Al and Si instrumental lines.
The model fitting is performed in two steps.We first use the χ 2 -statistic in XSPEC to fit each paired spectrum, providing initial estimates for the best-fit parameters.Although χ 2 fitting provides a quick and reasonable parameter estimation, it can sometimes find local minima and yield less accurate uncertainty estimations, particularly with multiple free parameters.Therefore, we adopt the Markov Chain Monte Carlo (MCMC) method secondly after the χ 2 fitting.This approach allows for a more robust and efficient exploration of the parameter space.
The MCMC is performed with the Goodman-Weare algorithm (Goodman & Weare 2010) via the chain type command in XSPEC.It starts with the best-fit parameters from the χ 2 fitting as initial values for the walkers, using the chain gaussian deltas 100 command.Considering 20 free parameters in the model, 40 MCMC walkers are used with each run for a minimum of 10,000 steps.Convergence is continuously monitored using the Gelman-Rubin convergence diagnostic (R, Gelman & Rubin 1992), a widely used method to assess MCMC chain convergence.A value of R < 1.1 generally indicates good convergence.If any spectral line parameters fail to meet this threshold, the fitting process is extended with additional steps until convergence is achieved.Our data analysis uses only the last 2000 points from these MCMC chains.
In some cases, the signal from the hot gas and SWCX may be weak or indistinguishable from noise.To distinguish real signals from noise, we apply the following criteria: if the median of an MCMC chain exceeds the width of its 68% credible interval, we classify it as a measurement and report the median and the 68% credible interval.If not, we classify it as an upper limit and only report the 95th percentile, indicating that we are 95% certain the intensity is below this value.

Visual Inspection of Fitting Results
Although the spectral model is specifically designed for SXRB spectra, it may not fit with cases where bright X-ray sources are not fully masked or excessive SP contamination remains during high SPB activity.These could bias line intensity measurements, necessitating manual exclusion of such observations through visual inspection.After inspecting each spectrum (e.g., Figure 3) and its best-fit model, we identify dozens of poorly-fitted spectra caused by the reasons above and resulting in significant residuals at certain energies.Additionally, a few spectra show a feature approximately 10 eV lower than the median energy of the O VII line centroids.Such unexpected diffuse emissions are likely shifted O VII emissions from unmasked diffuse sources beyond the MW.Therefore, we exclude observations with χ 2 /DoF > 5 or E OVII less than 555 eV, considering the spectral resolution and the instrumental gain (see Section 7.1 for more detail).
The remaining problematic observations are concentrated in low galactic latitude regions.These regions, dense with gas and dust, exhibit strong X-ray absorption and complex emissions from the Galactic disk, complicating line measurements.Therefore, we exclude observations within |b| < 2 • , the region where most problematic observations are found.
Based on visual inspection and low latitude exclusion, we further exclude 1921 observations from our sample.This results in a final sample of 5418 observations or 5470 paired exposures.

Catalog
The line measure results are presented in Table 3, along with corresponding observational information.Columns in this table include ( 1

Classification of Bright Observations
The SXRB measurements can be enhanced by various astronomical objects, including SWCX around comets or solar planets within the Solar System, Galactic bubbles, diffuse superbubbles (SBs) or supernova remnants (SNRs) within the MW, and galaxy clusters or galaxies beyond the MW.They may not exhibit clear structures within the FOV, yet can enhance the intensity measurements.Their contamination may not be fully excluded using the procedures in Section 1, leading to the need for additional data filtering.
We add a classification parameter, "Flag", in our catalog to classify observations, which are likely contaminated by these objects.The classification is based on the presence of X-ray bright objects in the FOV or the observation direction.Observations contaminated by known galaxy clusters and nearby galaxies are flagged with "1" and "2" in previous Section 2.2.5.We further expand the initial classification by visually inspecting the allsky maps in Figure 5.We focus on isolated observations with measured line intensities at least three times higher than their neighboring observations.These observations are then classified based on their target names and/or observation directions.In addition, we notice five extended regions with local intensity enhancements: eROSITA bubbles (eRBs; Predehl et al. 2020), Cygnus SB (Cash et al. 1980), Eridanus SB (Ochsendorf et al. 2015), Vela SNR (Helfand et al. 2001), and Monogem SNR (Thorsett et al. 2003).They are remnants of past astronomical events with angular radii ∼ 10 • .Observations within these regions are directly flagged based on their angular sizes in soft X-rays.The size of eRBs is approximated by projecting ellipsoids at the Galactic center with dimensions of 14 kpc × 9 kpc × 9 kpc and tilted by 10 • to align with observations.The sizes of the other four are estimated based on visual inspection of the line intensity maps in Figure 5.All classification flags are summarized in Table 1.
We note that the clean sample (Flag="0") is most suitable for diffuse gas studies due to the absence of contamination from bright X-ray sources.Observations flagged "1" and "2" are generally usable, as we mitigate contamination from galaxy clusters and nearby galaxies (see Section 2.2.5), though a few may still have notable contamination from extended sources.Flags of "3" to "7" indicate observations within five large X-ray structures.They are high-intensity regions resulting from past events and not fully mixed with the hot CGM.Observations with Flags of "8" to "11" should only be used as references for other studies as they do not represent the MW's hot gas emission.

Line Intensity Properties
The observed O VII, O VIII, and Fe-L line intensities have medians of ≈ 6 photons cm −2 s −1 sr −1 (line unit; L.U.), ≈ 1 L.U., and ≈ 1 L.U., respectively.Furthermore, 90% of these intensities fall within the ranges of ≈ 2-18 L.U., ≈ 0-8 L.U., and ≈ 0-9 L.U.Here, we explore the intensity distributions and correlations between these lines and discuss the implications of these relationships.Figure 6 presents two-dimensional histograms, showing the correlations between the intensities of these emission lines.Each observation is represented by all points from the MCMC chain.Blue dashed lines represent the collisional ionization equilibrium (CIE) model predictions at different temperatures.
All panels show strong positive correlations, indicating a roughly consistent intensity ratio.This consistency, as predicted by the CIE model, points to a relatively narrow temperature range in MW hot gas.A more detailed study on MW's temperature distribution is presented in Paper II.In addition, the red contours enclosing the eRBs' measurements indicate that the eRBs contribute most of the high-intensity measurements.The double peak in the contours suggests that the eRBs may consist of two components, which may be evidence of multi-phase structures within the eRBs.Furthermore, the presence of high-intensity observations outside the eRBs suggests the existence of other high-intensity emission regions in the hot gas.These regions may represent other small X-ray structures or X-ray bright phenomena within the hot gas.

Data Comparison
In this section, we compare our measurements of O VII and O VIII to those reported by HS12.There are several differences in our methodology from that of HS12, which could potentially result in inconsistent O VII and O VIII measurements: 1. We use version 20.0.0 of the SAS software, whereas HS12 use version 11.0.1.
2. We use cheese to mask out point sources and two catalogs to mask out known galaxy clusters and nearby galaxies.In contrast, HS12 use the XMM-Newton Serendipitous Source Catalogue to mask out point sources and some other sources by hand.
3. The metal abundances in our study are from Lodders (2003), while the metal abundances use in HS12 are from Anders & Grevesse (1989).
4. We treat the CXB normalization parameter as a free parameter in our model, while HS12 fixes it at 7.9 cm −2 s −1 sr −1 keV −1 .
5. The column density accounting for the MW and CXB absorption is based on the dust emission map from the Planck survey, rather than the H I map.
6.We include the Fe-L line as another Gaussian function in our model.Additionally, we disable strong iron emission lines in the 0.7-1.0keV band in the APEC model.
7. We determine our measurements and uncertainties using the MCMC method, as opposed to the χ 2statistic method used by HS12.To visualize the impact of these differences, we conduct a comparative analysis using 1312 shared observations between our study and HS12.In Figure 7, we show all shared measurements with ours on the X-axis and HS12's on the Y-axis.The figure shows that our measurements largely align with those of HS12, with O VII and O VIII medians 0.19 and 0.15 L.U. higher than HS12's.
The discrepancy in O VIII is most likely caused by the metal abundance table adopted in the spectral fitting.In our model, the MW CGM is modeled using an APEC model with O VIII lines (≈ 654 eV) disabled, but the O VII Kβ line (≈ 666 eV) remains.Shifting from the Anders & Grevesse (1989) to the Lodders (2003) abundance table nearly halves the oxygen abundance (i.e, from 8.51 × 10 −4 to 4.90 × 10 −4 ).This change results in a weaker O VII Kβ intensity in the MW APEC model, which can lead to a slight increase in the measured intensity of the O VIII line.This is tested by randomly fitting 500 SXRB spectra using both abundance tables while keeping other settings unchanged.The results show that ≈ 80% of the O VIII values are higher when the Lodders (2003) table is used, supporting this explanation.
In conclusion, our O VII and O VIII measurements generally align with those of HS12 with minor differences due to methodological differences.

TEMPORAL VARIATION IN SWCX
Studying the temporal variation of SWCX helps us understand its characteristics and accurately derive the intrinsic MW emission by subtracting this variation.
Hot gas emissions from the LHB and MW are expected to remain constant over decades.Therefore, any temporal variations in a given direction are attributed to SWCX (e.g., Kuntz 2019).
The SWCX emission in the SXRB has two primary contributors, the magnetospheric SWCX and the heliospheric SWCX.The former arises from interactions between highly ionized species in the solar wind with the neutral gas in Earth's atmosphere and typically exhibits variations on timescales of hours to days (see Section 7.2 for a detailed discussion).The heliospheric SWCX occurs when the solar wind interacts with the neutral ISM entering the heliosphere and generally varies over the solar cycles (Cravens et al. 2001).Notably, in certain directions such as those parallel to the Parker spiral, the variation timescales can be shorter than a week (Kuntz 2019).Qu et al. (2022) reported a temporal variation of the observed O VII and O VIII lines over a solar cycle using XMM-Newton data from HS12.This variation is highly correlated with the solar cycle and is likely associated with the heliospheric SWCX.The 22-year data presented in this study allows a more precise investigation of this long-term variation over an entire solar magnetic cycle, which will be further investigated in future work.

Characterizing the Long-Term SWCX
The XMM-Newton's uneven pointing can introduce observational biases.These biases are mitigated in halfyear or longer intervals, as XMM-Newton observes the brighter areas of the sky twice a year.This scheduling enables us to investigate long-term variations by plotting the median intensities over half-or 1-year intervals using the clean sample, as illustrated by the dots in Figure 8.Although the results reveal clear trends correlating with solar activity, a more robust approach independent of pointing biases is necessary to confirm the authenticity of the observed variations.
Here, we apply a "close-pair" method that utilizes observation pairs projected within ≤ 2 • , since the MW emission shows a strong correlation within 2 • (e.g., Kaaret et al. 2020;Paper II).By only considering the clean sample, we identify over 26000 close pairs and calculate their intensity differences (i.e., I 1,OVII − I 2,OVII ).They are then grouped into half-year intervals based on the observation dates.The median values for each interval are shown in the left panel of Figure 9.For better visualization, each pair is counted twice by inverting the order.
We reconstruct the temporal variation of the SWCX using the difference calculated based on close pairs.In this fitting, we only use pairs with the observation date T 1 later than T 2 , meaning only pairs above the diagonal in Figure 9.We then adopt a non-parametric model to fit these re-binned differences in the Bayesian framework using the emcee implementation (Foreman-Mackey et al. 2013).It has a likelihood function of where y and σ m are the observed difference and uncertainty.For O VII and O VIII, the model y m represents the difference calculated based on a step function with 44 free parameters, characterizing the SWCX variations in each half-year bin from 2000 to 2022.In contrast, the long-term variation in Fe-L intensity is modeled with 22 parameters corresponding to 1-year intervals, due to its lower signal-to-noise ratio.The parameter σ p is an empirical parameter to account for additional scatter in data, which is different for O VII, O VIII, and Fe-L.Specifically, the fitted values are 0.67 ± 0.03, 0.22 ± 0.01, and 0.03 ± 0.01, respectively.The MCMC sampling assigns 150 walkers, each taking 30,000 steps.Only the last 10% of the chain is used to derive the results for best-fit parameters.Figure 9 shows the best-fit model and residuals of the O VII difference, for example.

The Empirical SWCX Model
The results derived from the "close-pair" method, along with the regrouped median intensities are shown in Figure 8.The fitted model is represented by reddish curves with medians, 1σ, and 2σ intervals outlined.The gray curves represent the re-scaled and smoothed sunspot number (SSN), as a direct tracer for solar activity.The figure shows that O VII and O VIII emissions rise with an increase in SSN and decrease when solar activity diminishes.In contrast, this trend is less prominent for Fe-L emissions.
To evaluate the consistency between the "close-pair" and "rebin" approaches, we calculate the normalized differences, (I cp − I rebin )/σ.The result shows that ∼ 80% of the O VII differences and ∼ 70% of both O VIII and Fe-L differences fall within ±1, indicating a high consistency between the two approaches.
We examine the relations between the SSN and each line emission by applying the Pearson correlation coefficient test on the median intensities of the three lines and the median SSNs.The calculated correlation coefficients are 0.85, 0.83, and 0.44, respectively, with corresponding p-values of 2.43 × 10 −13 , 2.17 × 10 −12 , 4.00 × 10 −2 .These p-values represent the probabilities that the observed relationships occurred by chance.The results suggest strong positive correlations between SSN and the O VII or O VIII intensities, while the correlation with Fe-L emission is moderate.
In addition, Figure 8 shows long-term variations in O VII, O VIII, and Fe-L intensities with medians of 1.8, 0.3, and 0.1 L.U., against observed medians of 6.2, 1.4, and 0.8 L.U., respectively.This implies SWCX contributes roughly 30%, 20%, and 10% on average to the observed intensities, significantly impacting O VII and O VIII emissions, with a more modest effect on Fe-L emission.Moreover, the O VII and O VIII variations can reach up to approximately 4.5 L.U. and 1.3 L.U., consistent with Henley & Shelton (2010).

SOFT X-RAY EMISSION OF THE MW
The MW's intrinsic emission is contaminated by foreground SWCX and LHB emissions, and partially absorbed by the medium along the line of sight.In the following, we quantify and correct these factors to derive the true MW emission.
The LHB is known to contribute more to the 1/4 keV band than the 3/4 keV band (Kuntz & Snowden 2000), implying it produces more O VII emission than O VIII emission.Shadowing observations towards the foreground molecular cloud, MBM 12, reveal an observed O VII emission of 1.8 +0.5 −0.6 L.U., with about 0.28 L.U. expected from the LHB (Koutroumpa et al. 2011).Further modeling of the observed O VII emission, accounting for multiple components including the LHB, suggests that the LHB's contribution is < 1 L.U. (e.g., Miller & Bregman 2015;Wulf et al. 2019).In addition, we estimate the LHB emission based on the emission measure, which is typically less than 7 × 10 −3 cm −6 pc (Liu et al. 2016), and ∼ 2 × 10 −3 cm −6 pc at medium Galactic latitudes (|b| ∼ 20 • ; Yeung et al. 2023).Assuming a temperature of 0.084 keV, the expected O VII and O VIII emissions from the LHB are under 0.6 and 8.4 × 10 −4 L.U., respectively.These findings indicate that the O VII emission from the LHB is generally less than 1 L.U., contributing a maximum of 17% to the median observed O VII intensity (≈ 6 L.U.).Since the LHB's contribution is relatively small and varies with direction, we have deferred such detailed analysis to future research.
To remove the SWCX contribution, we subtract the long-term variation of each line shown in Figure 8 from the observed intensities.Although Fe-L shows only a moderate correlation with solar activity, we include it in the correction for consistency.
After subtracting the foreground from the observed emission, the unabsorbed MW emission can be calculated as where e τ eff serves as the absorption correction factor with τ eff being the effective optical depth.τ eff is determined by σN H,eff , where σ is the absorption crosssection (Balucinska-Church & McCammon 1992;Yan et al. 1998) and N H,eff is the effective column density for a given line of sight.
In the spectral modeling, N H,eff is simply the total N H along each observation's line of sight, derived from the dust emission map (N H,dust , Section 3.5).This approach assumes that both the CGM and CXB emissions are located behind all absorbing layers.This assumption is commonly used (e.g., Henley & Shelton 2012;Miller & Bregman 2015) because most MW hot CGM emissions are situated beyond the MW disk, a region where the majority of absorbing material lies.However, this approach overestimates the absorption correction in high N H,dust regions, particularly in the MW disk.Because the X-ray-emitting gas is mixed with the absorbing medium rather than being entirely behind it.
Determining the effective absorption factor for these regions requires understanding the spatial distributions of both the high-column-density cool gas and X-rayemitting hot gas.However, a more sophisticated model that considers both absorption and emission simultaneously is beyond the scope of this paper.
Here, we empirically derive the maximum-allowed absorption factor mainly for high N H,dust observations, based on data quality.First, we assume that the absorbing medium and the X-ray-emitting gas are cospatial and the source function (S) remains constant across all absorption optical depths.In this scenario, the amount that escapes from the absorbing region is proportional to the integral τ 0 e −τ dτ , or equivalently, 1 − e −τ .We then derive a maximum optical depth (τ max ) by assuming e −τmax = σ I /I, where I and σ I are the measured line intensity and corresponding uncertainty for each observation.The calculated τ max represents the maximum optical depth allowed by the data quality as emissions beyond this threshold are obscured by the measurement uncertainty.Therefore, absorption correction is only ap- e ϵ(T )Z, meaning a higher intensity suggests an increased plasma density, metallicity, or emissivity at temperatures optimal for emission (i.e., ≈ 2 × 10 6 K, ≈ 3 × 10 6 K, and ≈ 7 × 10 6 K for O VII, O VIII, and Fe-L).The LHB emission is not corrected and may contribute to these line emissions, particularly for O VII (see Section 6).
plied to nearby gas with optical depths within τ max .Considering the cospatial absorbing and emitting gas, the unabsorbed and observed intensities can be calculated as Sτ max and S(1 − e τmax ).Thus, the maximumallowed absorption factor is derived as τ max /(1−e −τmax ).
For each observation, we always use the lower value between the dust-inferred and the maximum-allowed absorption factor.Consequently, ≈ 10% of the observations are corrected using the later factor, primarily concentrated in the Galactic disk region.
We note that using the data-limited factor primarily accounts for nearby gas absorption, which might lead to an underestimation of total absorption.Additionally, selecting the lower of two distinct correction factors may introduce inconsistencies, especially where these methods significantly diverge in different galactic regions.However, considering our data limitations, this is likely the most suitable approach we can adopt.
Upon applying SWCX and absorption corrections, we applied Gaussian smoothing to the data.Recent studies hint at the presence of small-scale temperature structures in the MW's hot gas, revealing significant variations at approximately 4 • − 6 • angular scales (Kaaret et al. 2020;Paper II).As such, we choose to apply the smoothing with a standard deviation of 4 • to enhance the visibility of these variations in all-sky maps.The smoothed all-sky maps are displayed in Figure 10, with large-scale structures (dash lines) highlighted.Hydrogen-dense areas with N H,dust > 5 × 10 21 cm −2 are outlined in black.Emissions from these areas are typically corrected using the maximum-allowed absorption factor.This plot shows that the O VII and O VIII intensities decrease away from the Galactic center, tracing the hot CGM at ≈ 2 × 10 6 K and ≈ 3 × 10 6 K.
In contrast, the Fe-L emission matches closely with the eROSITA bubbles (eRBs) and Galactic disk, implying two different origins of this hotter plasma (≈ 7 × 10 6 K).Previous studies have noticed such plasma in eRBs (e.g., Das et al. 2019;Ponti et al. 2023) and can be explained by significant energy injections from the Galactic center (Kataoka et al. 2018).However, the origin of the Fe-L disk remains controversial.A possible contributor to this emission is M dwarfs, the most abundant stars (∼ 75%) in the solar neighborhood (Reylé et al. 2021).Despite their low mass (≈ 0.1 − 0.6 M ⊙ ) and luminosity, M dwarfs generally exhibit higher soft X-ray luminosity compared to the Sun due to more flare activities (e.g., Osten et al. 2005;Wargelin et al. 2008).Observations targeting l, b = 165 • , 5 • indicate excess emissions in the Fe-L band, which can be explained by including an extra spectral emission model of M dwarfs (Wulf et al. 2019).A more recent study of the eFEDS field (l ∼ 220 • − 230 • and b ∼ 20 • − 40 • ) also observes such excess emissions, which can be well-fitted using an additional APEC model with kT ∼ 0.7 keV (Ponti et al. 2023).Including the same stellar model from Wulf et al. (2019), they find that the stellar contribution accounts for a significant portion, but not all.They propose that the rest of the excess emissions are from the Galactic corona, a plasma phenomenon likely arising from hot outflows or Galactic fountains originating in the Galactic disk.
As the contributions from different sources remain under debate, our Fe-L emission data could be crucial for future explorations in this area.

DISCUSSION
Using the X-LEAP sample, we characterize the longterm variation in O VII and O VIII emissions, and construct SWCX-and absorption-corrected all-sky maps for all three emission lines of interest.In this section, we explore further implications of the X-LEAP sample, including using the E OVII as a solar activity tracer, evaluating the impact of magnetospheric SWCX on shortterm variations, investigating the potential spatial dependence of SWCX on ecliptic latitude, and discussing future work.

O VII Line Centroid as an SWCX Tracer
The O VII line is a triplet consisting of a resonance line, an inter-combination line, and a forbidden line, as detailed in Section 3.2.The resonance line dominates the MW hot gas emission due to efficient collisional excitation, while the forbidden line dominates in SWCX emission because of the downward electron cascade favoring triplet states (see Porquet et al. 2010 for a detailed review).Thus, the E OVII can theoretically be used to trace the strength of SWCX.
The EPIC-MOS detectors have a spectral resolution of ≈ 40 eV at 500 eV, unable to resolve the resonance and forbidden lines separated by ≈ 13 eV.However, shifts in the E OVII may still be detectable.Ponti et al. (2023) detects an E OVII shift towards the forbidden line when the SWCX component dominates using SRG/eROSITA, which has a spectral resolution comparable to XMM-Newton 's (Predehl et al. 2021).
To examine E OVII shifts in MOS1 and MOS2, we resample the observed O VII line centroids, together with the instrumental line centroids in 1-year bins using the clean sample.Interestingly, both instrumental centroids shift similarly from their rest energies over time (top panel, Figure 11), likely due to temperature variations in the MOS detectors (Abbey et al. 2003).According to A smaller total χ 2 in the offset case suggests a closer relation between the shift and offset.Bottom: the observed and corrected EOVII over time.The correction is done using the Al-inferred offset results to minimize the instrument-induced variations.The resulting trend suggests a potential anticorrelation with solar activity, implying EOVII as a viable tracer of solar activity, even with limited spectral resolution.
the XMM-Newton technical notes9 , line centroid shifts are corrected by assuming a linear relationship between the rest and observed energies: However, after correction, line energy can still vary with an uncertainty of 5 eV over the full energy range for EPIC-MOS 9 , consistent with the observed shifts.Due to limited spectral resolution, remaining gain and/or offset changes can not be accurately determined.To evaluate the more dominant factor, we first assume that these shifts are solely due to either gain or offset changes.We then use the Al-inferred gain and offset (i.e., gain = E Al /E obs,Al and offset = E Al − E obs,Al ) to calibrate the observed Si line centroids (middle panel, Figure 11).The results suggest a smaller total χ 2 and lower dispersion (i.e., σ = 1.04 eV compared to 1.09 eV) in the offset case, implying a tighter association between the shift and offset.Thus, we correct the E obs,OVII using the Al-inferred offset results to minimize the instrumentinduced variations.The bottom panel of Figure 11 presents both the observed and corrected E OVII over time.The results show a potential anti-correlation with solar activity, especially during solar maxima and minima.
The findings suggest that the E OVII effectively traces solar activity, even with limited spectral resolution.Future X-ray missions like HUBS and LEM (Bregman et al. 2023;Khabibullin et al. 2023) will have higher spectral resolutions capable of resolving the O VII triplet lines.This improvement will allow the use of line ratios between the forbidden and resonant lines to assess solar activity levels, or even distinguish the SWCX and MW emissions directly.

Evidence of Magnetospheric SWCX
The magnetospheric SWCX is sensitive to changes in solar wind conditions and Earth's magnetic field (Ishikawa et al. 2013), which may lead to observable intensity variations over short periods.To investigate this, we identify same-direction observations in our dataset and find two representative observation sets of Mars (Dennerl et al. 2006) and AA Tauri, a young variable star at 150 pc (Loomis et al. 2017).Both objects are effectively masked out by cheese as point sources, and their FOVs remain almost unchanged.Therefore, any observed intensity changes are independent of both the target emissions and changes in observation direction.
Figure 12 shows the observed variations in O VII intensities of the Mars and AA Tauri observations, along with time-averaged solar proton flux data (grey line) retrieved from the Advanced Composition Explorer 10 (ACE).In the top two panels, the O VII intensities exhibit significant intensity enhancements of ≈ 5 L.U. within a month and ≈ 23 L.U. within a day.These magnitudes of increase are within those reported by others (e.g., Snowden et al. 2004;Fujimoto et al. 2007;Koutroumpa et al. 2007).Additionally, intensity peaks correlate with periods of maximum proton flux, suggesting that they are induced by rapid changes in solar activity.In contrast, the 2009 Mars observations (bottom row) display stable intensities, unsurprising given the proton flux is lower than in 2003.
These findings indicate that short-term variations in O VII intensity are primarily driven by changes in solar activity.This highlights the need to account for magnetospheric SWCX in future X-ray studies.

Spatial Dependence of SWCX on Ecliptic Latitude
Apart from the temporal variations, SWCX also exhibits spatial variations (e.g., Fujimoto et al. 2007;Ringuette et al. 2023).Here, we investigate its correlation in ecliptic latitude by dividing the clean sample into low (β low ≡ |β| < 30 • ) and high latitude (β high ≡ |β| ≥ 30 • ) sub-samples.In the top panel of Figure 13, we show the 1-year re-binned O VII intensities for these sub-samples, along with their best-fit models derived from the "close-pair" method outlined in Section 5.1.The lower panel presents the differences between the two sub-samples and the solar cycle, traced by the SSN.The difference suggests that O VII intensities are higher at high latitudes, especially during solar maxima.Conversely, during periods of low solar activity, the intensity difference diminishes, approximating zero.Since the major differences occur during solar maxima, this 10 https://izw1.caltech.edu/ACE/ASC/level2/index.html ) over 1-year bins.Bottom: the intensity differences and 1σ uncertainty from the "close-pair" result, along with smoothed SSN.The O VII intensity is generally higher at higher ecliptic latitudes, especially during solar maxima.This variation likely results from the varying solar wind properties at different latitudes.
latitude-dependent variation is likely attributed to the difference in solar wind properties across ecliptic latitudes.
Understanding the spatial distribution of SWCX is challenging due to multiple factors, including anisotropic solar wind and neutral gas distributions.Future research should aim to develop a more comprehensive model that incorporates these variables.

Future Work
Considering the limitations in the SWCX and LHB modeling, our future work will first focus on further refining the foreground correction, accounting for shortterm and spatial variations in SWCX, among other potential factors.An accurate foreground model is the basis for further study of the MW hot gas, including the temperature structures (also see Paper II) and the density distribution.The density distribution is essential for estimating the total hot gas mass in the MW (e.g., Miller & Bregman 2015), which can provide further insights into fundamental questions, such as the "missing baryon" problem and feedback processes.These lines represent spectral features at energies of ≈ 0.56 keV, ≈ 0.65 keV, and ≈ 0.80 keV.The dataset is specifically optimized to study diffuse emissions from both the MW hot gas and the SWCX, minimizing the contamination from point and extended sources, as well as from irrelevant backgrounds.
The key findings are summarized as follows: 1.All line measures are summarized in a machinereadable table, which is available online.
2. In the dataset, the observed O VII, O VIII, and Fe-L line intensities have medians of ≈ 6 L.U., ≈ 1 L.U., and ≈ 1 L.U., respectively, with 90% falling within the ranges of ≈ 2-18 L.U., ≈ 0-8 L.U., and ≈ 0-9 L.U. (Figure 6).The strong positive correlation between these intensities suggests consistent intensity ratios, in line with the CIE model's prediction of a narrow temperature range in MW hot gas.In addition, observations inside eRBs contribute most high-intensity measurements, while high-intensity measurements outside the eRBs point to the existence of small-scale structures in the MW hot gas.
4. We present all-sky maps of SWCX-and absorption-corrected O VII, O VIII, and Fe-L intensities (Figure 10).These maps represent the soft X-ray emissions of the MW hot gas at different bands.The oxygen emissions extend beyond the eRBs and decrease from the Galactic center.Conversely, the Fe-L emission closely traces the eRBs and the Galactic disk, implying two different origins.
5. The Al and Si instrumental line centroids shift similarly away from their rest energies, likely due to temperature variations in the EPIC-MOS.These temperature changes alter line energies across the entire MOS spectrum, thereby distorting the E OVII measurements.After addressing this effect, the corrected E OVII shows an anticorrelation with solar activity (Figure 11).This suggests that E OVII can serve as an independent tracer of SWCX contributions, even when the spectral resolution is lower than the splitting of the O VII triplet.6.The O VII intensities vary significantly over hours or days across observations of AA Tauri and Mars (Figure 12).These variations correlate closely with the solar proton flux, suggesting that they are associated with the magnetospheric SWCX.
7. The SWCX exhibits spatial dependence, with O VII intensity being higher at high ecliptic latitudes, especially during solar maxima (Figure 13).This variation is likely due to changing solar wind properties at different latitudes.Particularly, it lowers the observed intensity at the anti-center region, as its ecliptic latitude is low (≈ 5 • ).

Figure 1 .
Figure 1.Flowchart illustrating the data reduction process and observation classification.Explanations of the flags are in Table1.

Figure 2 .
Figure 2. Examples of point source and extended source removal on EPIC-MOS2 images.The sequential panels depict images from ObsIDs: 0675010901 (the left two panels) and 0304850901 (the right two panels).Each set presents the image before and after the removal of point sources (green ellipses) and either a galaxy cluster (Abell 3718 in blue) or a nearby galaxy (NGC 253 in orange).See Section 2.2.5 for size determination of the extended sources.

Figure 3 .
Figure 3.An example of a joint-fitted SXRB spectrum and its best-fit model.Left: the paired spectrum extracted from ObsID: 0724770201.The O VII (≈ 0.56 keV), O VIII (≈ 0.65 keV), and Fe-L (≈ 0.80 keV) lines are modeled as three Gaussian functions and highlighted.Instrumental lines are covered by a grey shade.Right: MOS1 spectrum only.All spectral components are present illustrating individual contributions to the modeled spectrum.The soft proton background (SPB) is not folded through the instrumental response.

Figure 4 .
Figure 4. Left: the O VII, O VIII, and Fe-L line emissivities at different APEC temperatures(Foster et al. 2012).They are disabled in the APEC model to prevent double-counting line intensities when extracting them using Gaussian functions.Right: the emissivity ratios.The observed line intensity ratios are consistent with the emissivity ratios if the emission originates from a single-temperature plasma.
) the XMM-Newton observation ID; (2) the name of each EPIC-MOS pair; (3) the observation start date in YYYY-MM-DD; (4) the mean GTI of each EPIC-MOS pair; (5) and (6) the observation direction in galactic coordinates (l, b); (7) the O VII line centroid (E OVII ) median with 68% credible interval; (8), (9), and (10) the observed O VII, O VIII, and Fe-L intensities.Measurements are reported with medians and 68% credible intervals, while upper limits are given at the 95th percentile.The criteria for distinguishing these two are in Section 3.9; and (11) the classification flag (detailed in Section 4.2).The observed intensities of the three lines are illustrated in all-sky maps shown in Figure 5.

Figure 6 .
Figure 6.Top: intensity distributions of O VII, O VIII, and Fe-L lines.Measurements and upper limits are shown in bar-type and dashed step-type histograms, respectively.Bottom: correlations of observed O VII, O VIII, and Fe-L intensities.Each panel displays the two-dimensional histograms of intensities (grayscale) using points from the MCMC chain.The blue dashed lines represent the CIE model predictions at different temperatures, while the contours (red) represent intensities measured inside eROSITA bubbles (eRBs).The eRBs contribute most of the high-intensity emissions, with some exceptions due to small X-ray structures or X-ray bright phenomena outside the eRBs.Positive relations between emission lines imply a consistent temperature of the MW's gas, with a relatively narrow temperature dispersion.A detailed study on MW temperature structures is in Paper II.

Figure 7 .
Figure 7. Top row: comparison of shared O VII and O VIII intensity measurements between this work and HS12.Bottom row: weighted differences derived from intensity difference (∆I) divided by total uncertainty (σ).Interpretation of the differences is in Section 4.4.

Figure 8 .
Figure 8.Long-term variations in the observed O VII, O VIII, and Fe-L line intensities from 2000 to 2022.Sunspot numbers (SSN) are smoothed and rescaled individually for better visualization.The circles represent intensities in halfor 1-year bins using the clean sample, adjusted by subtracting minimum values of 3.20, 0.54, and 0.46 L.U., respectively.The reddish curves represent the medians, 1σ, and 2σ uncertainties derived from the "close-pair" method.The variations in O VII and O VIII intensities correlate with the solar cycle, suggesting a strong association with SWCX.Moderate Fe-L correlation indicates that Fe-L is not primarily produced by SWCX (see text for details).

Figure 9 .
Figure 9. Example of MCMC analysis to derive long-term variations in O VII intensity from close pairs.Left: the medians of O VII intensity differences in half-year intervals.Middle: the intensity differences reconstructed after MCMC fitting.Right: residuals.The MCMC fitting process and results are detailed in Section 5.1 and Section 5.2.

Figure 10 .
Figure 10.All-sky maps depicting the MW soft X-ray emissions.From top to bottom: SWCX-and absorptioncorrected O VII, O VIII, and Fe-L maps.These maps are centered on the Galactic center and smoothed with a Gaussian function of σ = 4 • .Dashed lines indicate large-scale structures introduced in Figure 5. Black outlines hydrogendense areas where N H,dust > 5 × 10 21 cm −2 .The oxygen line emissions extend beyond the eROSITA bubbles (eRBs) and decrease from the Galactic center.The Fe-L line emission closely matches the eRBs and Galactic disk.The emission intensity is proportion to n 2e ϵ(T )Z, meaning a higher intensity suggests an increased plasma density, metallicity, or emissivity at temperatures optimal for emission (i.e., ≈ 2 × 10 6 K, ≈ 3 × 10 6 K, and ≈ 7 × 10 6 K for O VII, O VIII, and Fe-L).The LHB emission is not corrected and may contribute to these line emissions, particularly for O VII (see Section 6).

Figure 11 .
Figure 11.Top: yearly shifts in observed Al and Si instrumental line centroids in both MOS1 and MOS2.These shifts are with an uncertainty of 5 eV, likely due to temperature variations in the instrument.Middle: calibrated Si line centroids derived from Al-inferred gain or offset adjustments.A smaller total χ 2 in the offset case suggests a closer relation between the shift and offset.Bottom: the observed and corrected EOVII over time.The correction is done using the Al-inferred offset results to minimize the instrument-induced variations.The resulting trend suggests a potential anticorrelation with solar activity, implying EOVII as a viable tracer of solar activity, even with limited spectral resolution.

Figure 12 .
Figure 12.Left column: O VII intensity measurements/upper limits obtained from observations of the same targets within one month, one day, and two days, respectively.The gray lines represent the 12-hour and 1-hour averaged proton fluxes retrieved from ACE.Right column: The corresponding FOVs for these observation sets.The AA Tauri observations maintain the same direction, while the Mars observations shift by ≈ 0.5 • .Significant variations in the top two rows are likely associated with the magnetospheric SWCX.

Figure 13 .
Figure13.O VII intensity variations in ecliptic latitude.Top: median O VII intensities derived from "rebin" and "close-pair" methods in low and high ecliptic latitudes (|β| < 30 • and |β| ≥ 30 • ) over 1-year bins.Bottom: the intensity differences and 1σ uncertainty from the "close-pair" result, along with smoothed SSN.The O VII intensity is generally higher at higher ecliptic latitudes, especially during solar maxima.This variation likely results from the varying solar wind properties at different latitudes.
this study, we introduce the X-LEAP program and report a new set of O VII, O VIII, and Fe-L line measures.They are extracted from SXRB spectra based on 5418 EPIC-MOS observations conducted before 2022.

Table 1 .
Classification of Bright X-Ray Observations

Table 2 .
Spectral Model Settings in XSPEC Component a Model b Parameter c Initial Value d Range e b Model components associated with each emission source.c Parameters required for each model component.d Initial values assigned to these parameters.e Allowed ranges for each parameter.f Absorption model applied to account for dust and gas absorption attenuating MW CGM and CXB emissions.The column density for each observation is fixed to the total NH derived from the 2018 Planck dust emission survey (see Section 3.5).

Table 3 .
Properties of Observed O VII, O VIII, and Fe-L Lines a Ten typical samples from the catalog.The complete dataset is available in a machine-readable table online.b Paired EPIC-MOS exposures observe the same sky region simultaneously.If an observation has more than one pair, they are listed separately, as illustrated by the last two rows in the table.c The median and 68% credible interval for the O VII line centroid, derived from MCMC chains.d Intensities for O VII and O VIII Kα, and Fe-L spectral lines extracted from MCMC chains.Medians and 68% credible intervals are reported for measurements, while the 95% percentile is reported for upper limits.e Classification of bright X-ray observations for manual data filtering. a