Paradigm shift in understanding the Galactic eROSITA Bubbles

The magniﬁcent bubbles at the Galactic center provide a great channel to un- derstand the eﬀects of feedback on galaxy evolution. The newly discovered eROSITA bubbles show enhanced X-ray emission from the shells around bub- bles. Previous works assumed that the X-ray emitting gas in the shells has a single 17 temperature component and that they trace the shock-heated lower-temperature 18 Galactic halo gas. Here we show that the thermal structure of the eROSITA 19 bubble shells is more complex. Using Suzaku observations we ﬁnd with high 20 conﬁdence that the X-ray emission from the shells is best described by a two- 21 temperature thermal model, one near Galaxy’s virial temperature at kT ≈ 0 . 2 keV 22 and the other at super-virial temperatures ranging between kT = 0 . 4 − 1 . 1 keV . 23 Furthermore, we show that temperatures of the virial and super-virial compo- 24 nents are similar in the shells and in the ambient medium, although the emission 25 measures are signiﬁcantly higher in the shells. We argue that the X-ray bright 26 eROSITA bubble shells are the signature of compressed isothermal radiative 27 shocks. The age of the bubbles is constrained to 70 – 130 Myr . This expansion 28 timescale, as well as the observed non-solar Ne/O and Mg/O ratios, favor the 29 stellar feedback models for the formation of the Galactic bubbles, settling a 30 long-standing

The Galactic bubbles are expanding into the MW halo; we therefore examine the spatial 48 distribution of the X-ray emission from the bubble shells and from the halo around them 49 to constrain their thermal structure. We conducted a survey of Suzaku observations with 50 this goal. We selected 230 archival Suzaku observations of the soft diffuse X-ray background 51 (SDXB) to characterize the X-ray emission from the Galactic bubbles (Galactic longitudes 52 300 • < l < 60 • ) and from the surrounding extended halo (60 • < l < 300 • ). 53 In order to extract the Galactic bubbles/halo emission from the SDXB, it is crucial to 54 accurately model the other components of the SDXB, such as the Local Bubble (LB), solar 55 wind charge exchange (SWCX), the cosmic X-ray background (CXB), and the instrumen-56 tal background. We included emission from these components in the spectral fitting (see 57 methods). Spectral fits to the Suzaku spectra show that the X-ray emission of the bubble 58 shells as well as the outer halo is best described by two-thermal components, a warm-hot 59 phase near the Galaxy's virial temperature kT ≈ 0.2 keV (2.3 × 10 6 K) and a hot phase 60 at super-virial temperatures ranging between kT = 0.4 − 1.1 keV (0.5 − 1.3 × 10 7 K). The 61 best fit models also require overabundance of nitrogen (N/O) in the warm-hot phase, both 62 from and around the bubbles. Toward a few Galactic bubble sightlines, the best fit model 63 also requires super-solar abundance ratios of neon and/or magnesium to oxygen (Ne/O and 64 Mg/O). Fig. 1 shows the X-ray emission maps of the warm-hot and the hot components of 65 the Galactic bubbles and the surrounding halo emission. 66 The presence of the warm-hot, virial-temperature gas in the Galactic halo has been known 67 for years [4,5,6,7,8], however the super-virial temperature gas was recently discovered. 68 The first robust detection was in the sightline to 1ES1553 + 113 passing close to the North 69 Polar Spur/Loop-I region of the Galactic bubbles [9, 10]. Later, the similar temperature hot 70 gas was detected toward three other sightlines passing close to and away from the Galactic 71 bubbles [11]. These studies showed the presence of the hot gas in the Galactic halo, but it 72 1 https://www.mpe.mpg.de/7461950/erass1-presskit 2 The CGM of the Milky Way is usually referred as the Galactic "halo". CGM is a more prevalent term for external galaxies. Both the terms mean essentially the same, and we will use these terms interchangeably.
was not known how ubiquitous it is.
In this work we have detected the hot gas toward a large number of sightlines distributed 74 all over the sky. We confirmed with high confidence that the super-virial temperature plasma 75 is widespread in the Galaxy and it is not necessarily associated with the Galactic bubbles 76 only (Extended Data Fig. 2). This has significant implications for our understanding of the 77 bubbles.

78
The Galactic bubbles are believed to have formed by the GC feedback (e.g., [12,13,14]), 79 that has generated shocks in the northern and the southern hemispheres, and these shocks   However, the temperatures of the warm-hot and hot components are similar in/outside the 89 shells. X-ray surface brightness of a gaseous medium depends on its temperature as well as 90 the EM. Our results show that the Galactic bubble shells have higher EMs but not higher 91 temperature in comparison to the surrounding halo, contrary to the current proposed models 92 of the bubbles [1]. Since the EM is proportional to the density square, we argue that the 93 higher X-ray surface brightness of the Galactic bubble shells as seen in the eROSITA all-sky 94 map is a result of the compressed denser gas, but it is not hotter than the surrounding 95 medium.

96
Previous studies used a single temperature model with fixed relative abundances to define 97 the X-ray emission and inferred that the Galactic bubble shells have temperature of kT ≈ 98 0.3 keV [15,16,17,18,19,20] or kT ≈ 0.4 keV [21]. This is higher than the temperature of the 99 MW CGM of ∼ 0.2 keV, which led them to conclude that the bubble shells represent shock 100 heated gas. Further, using the ratio of the pre-and post-shock temperatures, these works 101 estimated the shock speed, age and energy of the bubbles (see methods). We show that the 102 use of a single temperature model to represent the shell emission was too simplistic, leading 103 to incorrect physical model of the bubbles. In this work using the new and better spectral about 14 kpc [1] above and below the Galactic plane. Whether the Galactic bubbles are the 118 result of the AGN activity or the star-formation driven outflows is still a topic of extensive 119 debate. The age of the bubbles provide an important clue to differentiate the two feedback 120 mechanisms. The AGN wind driven models require small age ∼ 3 − 12 Myr [22,23,24], 121 while the stellar feedback models based on the star formation rate at the GC estimate the  We performed Suzaku data reduction with HEAsoft version 6.29. We only used the data 139 from the back-illuminated (BI) XIS1 detector, as this has better sensitivity at low energies 140 than the front-illuminated (FI) XIS0 and XIS3 detectors. We combined the data taken in band due to solar X-rays scattered off the Earth's atmosphere [29].

148
The activity of our own Sun can affect the space weather and contaminate data taken  The goal of this work is to analyze the diffuse emission, hence it is important to remove the 166 point sources. We generated the 0.5 − 2.0 keV images and identified the bright point sources. 167 We selected the point source exclusion regions of radii of 1 ′ − 3 ′ (c.f Suzaku XRT's half-power diameter of 1.8' to 2.3'). Then we extracted the diffuse emission spectrum from the entire 169 field-of-view after excluding the point source regions. We produced the redistribution matrix 170 files (RMFs) using the xisrmfgen ftool, in which the degradation of energy resolution and its 171 position dependence are included. We also prepared ancillary response files (ARFs) using 172 xissimarfgen ftool with the revised recipe 4 . For the ARF calculations we assumed a uniform 173 source of radius 20 ′′ and used a detector mask which removed the bad pixel regions. We We performed all the spectral fitting with Xspec version 12.11.1 5 . We modeled all the 179 thermal plasma components in collisional ionization equilibrium (CIE) with the APEC (ver-180 sion 3.0.9) model and used solar relative metal abundances [31]. For absorption by the 181 Galactic disk, we used the phabs model in XSPEC.

182
Suzaku provides an opportunity to resolve the different components of the SDXB due to  We started with fitting the Suzaku SDXB spectra with a three-component model. The

193
temperature of the foreground component was frozen at kT = 0.1 keV (e.g., [32,33,34,194 35]), but we allowed the normalization to vary. We modeled the Galactic bubbles (or the 195 extended CGM) emission as single temperature collisionally ionized plasma characterized by 196 temperature (kT) and emission measure (EM), and with fixed metallicity 6 . We fixed the 197 total metallicity to 1 (in solar units) for both the thermal components as the total metallicity 198 and normalizations (or EM) are degenerate in the APEC model. We allowed the power-law 199 photon index and the normalization to vary in the spectral fits.  Since N vii and Ne ix have strong transitions at 0.5 keV and 0.9 keV, respectively, we 205 allowed the nitrogen and neon relative abundances to vary in our above model. That provided 206 a slightly better fit but still left significant excess emission at the higher energy side (0.8 − 207 4 https://heasarc.gsfc.nasa.gov/docs/suzaku/analysis/xisnxbnew.html 5 https://heasarc.gsfc.nasa.gov/xanadu/xspec/ 6 The X-ray emission data does not contain any line or edge of hydrogen. Thus we cannot obtain absolute metal abundances from X-ray emission data alone. Instead, the X-ray observations provide constraints on relative metal abundances, for example N/O, C/O, Ne/O.  The eROSITA all-sky soft X-ray map shows giant, bi-polar, X-ray emitting shell-like struc- Post-shock conditions of compression and heating send the shocked gas out of thermal 295 equilibrium and consequently it radiates profusely. Eventually, the radiative cooling lowers 296 the temperature until the gas attains a radiative balance, and very often the final equilibrium 297 temperature is the same as the initial pre-shock temperature of the medium. A shock where 298 the post-shock gas cools to the initial temperature is commonly referred to as an isothermal 299 shock (see [36] for the detailed theory of shocks). As we noted above, the temperatures of Multiple studies have attempted to characterize the X-ray emission from the Galactic bubbles 346 [1,15,16,17,18,19,20]. These authors assumed a single temperature for the X-ray emitting 347 shells, and measured it to be ∼ 0.3 keV. They interpreted that this emission arises in the 348 weakly shock-heated Galactic halo gas at ∼ 0.2 keV, and they estimated a Mach number of 349 the shock of M ≈ 1.5 using the R-H conditions for the temperature [1,15,18]. 350 We argue that the X-ray spectral model is more complex. First, we did not assume a 351 single-temperature model to describe the X-ray emitting shell gas. We found that a two 352 temperature model provides a better fit to both the shell gas, and the ambient halo gas.

353
This enabled us to determine that the shock is radiative, instead of adiabatic. Secondly, 354 previous studies used the fixed abundances ratio (0.2-0.3 solar) for the thermal model of the 355 bubbles, which fails to detect any non-solar relative abundances (due to metal enrichment 356 or metallicity inhomogeneity), as is commonly seen in the star-formation related feedback.

357
Additionally, we find that the previous claims of an adiabatic shock are inconsistent with non-radiative shock is bounded by a value of (γ + 1)/(γ − 1) which equals 4 for γ = 5/3.

368
Thus we see that the shocks are not adiabatic, but instead are radiative, as we argue here.  We are grateful to Prof. Barabara Ryden for her notes of the "Radiate Gas Dynamics" 373 graduate course at Ohio State. We gratefully acknowledge support through the NASA 374 ADAP grants 80NSSC18K0419 to AG and NNX16AF49G to SM.