The X-ray bright bubbles at the Galactic Centre provide an opportunity to
understand the effects of feedback on galaxy evolution. The shells of the
eROSITA bubbles show enhanced X-ray emission over the sky background.
Previously, these shells were assumed to have a single temperature
component and to trace the shock-heated lower-temperature halo gas.
Using Suzaku observations, we show that the X-ray emission of the shells
is more complex and best described by a two-temperature thermal model:
one component close to the Galaxy’s virial temperature and the other at
supervirial temperatures. Furthermore, we demonstrate that temperatures
of the virial and supervirial components are similar in the shells and in the
ambient medium, although the emission measures are significantly higher in
the shells. This leads us to conclude that the eROSITA bubble shells are X-ray
bright because they trace denser gas, not because they are hotter. Given
that the pre- and postshock temperatures are similar and the compression
ratio of the shock is high, we rule out that the bubble shells trace adiabatic
shocks, in contrast to what was assumed in previous studies. We also observe
non-solar Ne/O and Mg/O ratios in the shells, favouring stellar feedback
models for the formation of the bubbles and settling a long-standing debate
on their origin.
Forest laws, Indian forest laws, why they are important
Thermal and chemical properties of the eROSITA bubbles from Suzaku observations
1. Nature Astronomy
natureastronomy
https://doi.org/10.1038/s41550-023-01963-5
Article
Thermalandchemicalpropertiesofthe
eROSITAbubblesfromSuzakuobservations
Anjali Gupta 1,2
, Smita Mathur2,3
, Joshua Kingsbury 1,2
, Sanskriti Das 2
& Yair Krongold4
TheX-raybrightbubblesattheGalacticCentreprovideanopportunityto
understandtheeffectsoffeedbackongalaxyevolution.Theshellsofthe
eROSITAbubblesshowenhancedX-rayemissionovertheskybackground.
Previously,theseshellswereassumedtohaveasingletemperature
componentandtotracetheshock-heatedlower-temperaturehalogas.
UsingSuzakuobservations,weshowthattheX-rayemissionoftheshells
ismorecomplexandbestdescribedbyatwo-temperaturethermalmodel:
onecomponentclosetotheGalaxy’svirialtemperatureandtheotherat
supervirialtemperatures.Furthermore,wedemonstratethattemperatures
ofthevirialandsupervirialcomponentsaresimilarintheshellsandinthe
ambientmedium,althoughtheemissionmeasuresaresignificantlyhigherin
theshells.ThisleadsustoconcludethattheeROSITAbubbleshellsareX-ray
brightbecausetheytracedensergas,notbecausetheyarehotter.Given
thatthepre-andpostshocktemperaturesaresimilarandthecompression
ratiooftheshockishigh,weruleoutthatthebubbleshellstraceadiabatic
shocks,incontrasttowhatwasassumedinpreviousstudies.Wealsoobserve
non-solarNe/OandMg/Oratiosintheshells,favouringstellarfeedback
modelsfortheformationofthebubblesandsettlingalong-standingdebate
ontheirorigin.
The all-sky survey performed by the eROSITA X-ray telescope has
shownalargehourglass-shapedstructureinthecentreoftheMilkyWay
(MW)1
, called the ‘eROSITA bubbles’. The X-ray bright quasi-circular
feature in the northern sky, which includes structures such as the
North Polar Spur and the Loop I, has been known since its discovery
by ROSAT2
. The eROSITA map shows X-ray emission from a similarly
huge quasi-circular annular structure in the southern sky; together
they seem to form giant galactic X-ray bubbles emerging from the
Galactic Centre (GC).
Thelarge-scaleX-rayemissionobservedbyeROSITAinitsmedium
energyband(0.6–1.0 keV)showsthattheintrinsicsizeofthebubbles
is several kiloparsecs across1
. The eROSITA bubbles show striking
morphologicalsimilaritiestothewell-knownFermibubblesdetected
inγ-raybytheFermitelescope3
,buttheyarelargerandmoreenergetic.
TheFermiandeROSITAbubbles(collectivelywecallthemthe‘Galactic
bubbles’) provide an exciting laboratory for studying the feedback
because of their size and the location in the Galaxy. These bubbles
aremagnificentstructuresinjectingenergy/momentumintotheMW
circumgalacticmedium(CGM)orhalo.(TheCGMoftheMWisusually
referredastheGalactic‘halo’.CGMisamoreprevalenttermforexternal
galaxies. Both the terms have essentially the same meaning, and we
will use these terms interchangeably.) To understand the feedback
process,itisimportanttodeterminethethermal,kineticanddynamic
structureofthesebubbles.
TheGalacticbubblesareexpandingintotheMWhalo;wetherefore
examine the spatial distribution of the X-ray emission from the bub-
ble shells and from the halo around them to determine their thermal
structure. We conducted a survey of Suzaku observations with this
Received: 22 January 2022
Accepted: 31 March 2023
Published online: xx xx xxxx
Check for updates
1
Columbus State Community College, Columbus, OH, USA. 2
Department of Astronomy, The Ohio State University, Columbus, OH, USA. 3
Center for
Cosmology and Astro-Particle Physics, The Ohio State University, Columbus, OH, USA. 4
Instituto de Astronomia, Universidad Nacional Autonoma de
Mexico, Mexico City, Mexico. e-mail: agupta1@cscc.edu
2. Nature Astronomy
Article https://doi.org/10.1038/s41550-023-01963-5
shows the X-ray emission maps of the warm-hot and the hot compo-
nents of the Galactic bubbles and the surrounding halo emission.
Out of our 150 sightlines that probe the Galactic bubbles region,
thehotthermalcomponentisrequiredathighconfidence(F-testprob-
ability)of>99.99%in55sightlines,at>90.0%in80sightlines,andat1σ
significancein127sightlines.Fortheregionsoutsidethebubbles,the
hotcomponentisrequiredattheconfidenceof>99.0%in26sightlines,
at>90.0%in51sightlinesandat1σsignificancein64sightlines,outof
our80sightlines.Figure2showsthehot-componentF-testprobability
mapforallthesightlinesinvestigatedinthiswork.
The best-fit models also require overabundance of nitrogen at
theconfidenceof>90.0%(averageN = 4.2 ± 0.2 solar)inthewarm-hot
phase, both from and around the bubbles. Towards the 10 Galactic
bubbles sightlines (but not outside the bubbles), the best-fit model
also requires supersolar abundance ratios of neon to oxygen at >1σ
significance (average Ne = 2.1 ± 0.2 solar). Supersolar magnesium to
oxygenratio(Mg = 3.6 ± 1.4 solar)isalsorequiredalongonesightline.
The presence of the warm-hot, virial-temperature gas in
the Galactic halo has been known for years4–9
; however, the
supervirial-temperaturegaswasrecentlydiscovered.Thefirstrobust
detection was in the sightline to 1ES1553 + 113 passing close to the
North Polar Spur/Loop I region of the Galactic bubbles10,11
. Later, the
similar-temperature hot gas was detected towards three other sight-
lines passing close to and away from the Galactic bubbles12
. These
studies showed the presence of the hot gas in the Galactic halo, but it
wasnotknownhowubiquitousitis.
Inthisworkwehavedetectedthehotgastowardsalargenumber
ofsightlinesdistributedalloverthesky.Weconfirmedwithhighcon-
fidence that the supervirial-temperature plasma is widespread in the
Galaxy and it is not necessarily associated with the Galactic bubbles
only (Fig. 2). This has important implications for our understanding
ofthebubbles.
The Galactic bubbles are believed to have formed by the GC
feedback (for example, refs. 13–15); this has generated shocks in the
northern and the southern hemispheres, and these shocks have been
expandingintotheGalactichalo.Theshapeandspeedofshockstravel-
ling through the MW CGM depend on the CGM density, pressure and
temperature. Thus to characterize the properties of the shocks, we
examined the variation in thermal parameters of the warm-hot and
the hot phases of the shocked (bubble shells) and unshocked (outer
halo)plasmaoftheGalactichalo.
Figure 3 shows the distribution of emission measures (EMs) and
temperaturesofboththethermalcomponentsasafunctionofGalactic
longitude.WeseethattheEMsarequitehigherforsightlinespiercing
goal.Weselected230archivalSuzakuobservationsofthesoftdiffuse
X-raybackground(SDXB)tocharacterizetheX-rayemissionfromthe
Galactic bubbles (Galactic longitudes 300° < l < 60°) and from the
surroundingextendedhalo(60° < l < 300°).
ToextracttheGalacticbubbles/haloemissionfromtheSDXB,itis
crucial to accurately model the other components of the SDXB, such
as the Local Bubble, solar wind charge-exchange, the cosmic X-ray
backgroundandtheinstrumentalbackground.Weincludedemission
fromthesecomponentsinthespectralfitting(Methods).
Results
Atwo-temperaturespectralmodel
Typically the Galactic bubbles/halo emission is described by a
single-temperature thermal component. However, our spectral fits
totheSuzakuspectrashowthattheX-rayemissionofthebubbleshells,
and of the outer halo, is best described by two-thermal components
(Methods), a warm-hot phase near the Galaxy’s virial temperature
kT ≈ 0.2 keV (2.3 × 106
K) and a hot phase at supervirial temperatures
rangingbetweenkT = 0.4 keVandkT = 1.1 keV(0.5–1.3 × 107
K).Figure1
60°
30°
0°
–30°
–60°
–30°
–60°
150° 120° 90° 60° 30° 330° 300° 270° 240° 210°
0.040
0.020
0.010
0.005
0.001
0.2 0.4 0.6 0.8 1.0 1.2 1.4
kT (keV)
0.2 0.4 0.6 0.8 1.0 1.2 1.4
kT (keV)
Emission measure (cm–6
pc)
0.010
0.005
0.001
Emission measure (cm–6
pc)
0°
60°
30°
0°
150° 120° 90° 60° 30° 330° 300° 270° 240° 210°
0°
Fig.1|X-rayemissionmapsfromourSuzakusurveyoftheGalacticbubbles
andthesurroundinghaloregions.Figuresontopandbottomshowthe
distributionofthewarm-hotandthehotphases,respectively.Thecolourofeach
circleindicatestemperature,andtheradiusisproportionaltotheEM.Thesolid
redlinemarksX-rayeROSITAbubblesandthereddashedlinesrepresenttheedge
oftheγ-rayFermibubbles.
100
95
90
85
80
75
70
F-test statistic (%)
0°
150°
–30°
–60°
30°
60°
120° 90° 60° 30° 330° 300° 270° 240° 210°
0°
Fig.2|F-testprobabilitymapforthehot-componentsignificancerequired
overthestandardthree-componentSDXBmodelfortheSuzakuobservations
investigatedinthiswork.Emptycircleswithredcrossesmarkthesightlinesfor
whichaddingahotthermalcomponentdidnotimprovethefit.
3. Nature Astronomy
Article https://doi.org/10.1038/s41550-023-01963-5
thebubblesthanfortheouterhalosightlines.However,thetempera-
tures of the warm-hot and hot components are similar in/outside the
shells. X-ray surface brightness of a gaseous medium depends on its
temperature and the EM. Our results show that the Galactic bubble
shellshavehigherEMs,butnothighertemperature,thanthesurround-
ing halo, contrary to the current proposed models of the bubbles1
. As
theEMisproportionaltothedensitysquare,wearguethatthehigher
X-ray surface brightness of the Galactic bubble shells as seen in the
eROSITA all-sky map is a result of the compressed denser gas, but it is
nothotterthanthesurroundingmedium.
Theshockproperties
TheeROSITAbubblesareprobablyproducedbyshocksthathavebeen
driven into the northern and the southern Galactic halo. The speed
andshapeofshocksdependonthetotalenergyinputandthethermal
parametersoftheambientplasma.Multiplestudieshaveattemptedto
characterizetheX-rayemissionfromtheGalacticbubbles1,16–21
.These
authors assumed a single temperature for the X-ray-emitting shells,
and measured it to be ~0.3 keV. They interpreted that this emission
arises in the weakly shock-heated Galactic halo gas at T ≈ 0.2 keV, and
they estimated a Mach number of the shock of M ≈ 1.5 using the Rank-
ine–Hugoniot (R–H) conditions for the assumed temperature1,16,19
.
WehavefoundthattheX-rayspectralmodelismorecomplexthan
previouslyassumed.Theshellsarebestdescribedbyatwo-temperature
modelandthetemperaturesinandaroundtheshellsaresimilar.This
showsthattheshellsarenotshockheated;theshellsarebrightbecause
theytracedensergas,nothottergas.Wecomparedthethermalparam-
etersofthebubbleshellsandthepreshockhalogastoinfertheshock
propertiesfurther(followingDraine22
).ThegasdensityoftheeROSITA
bubbles’shellsareestimatedfromthemeasuredvaluesoftheEMs.The
EMisgivenbyn2
L,wherenisthedensity(assumingauniformmedium)
and L is the line-of-sight path length. The average line-of-sight path
lengthisaboutL ≈ 5 kpc,forashellofouterradiusof~ 7 kpcandthick-
nessof~4 kpc(fromPredehletal.1
).Thisresultsinanaveragedensityof
nshell ≈ 1.6 × 10−3
cm−3
within the shells. The Galactic halo studies (both
observational and theoretical) have estimated the halo density to be
about 2–5 × 10−4
cm−3
at a distance of 10 kpc (approximate location of
theshells)fromtheGC8,23–25
.Adoptingtheunperturbedhalodensityof
no = 4 × 10−4
cm−3
(thesameasusedbyPredehletal.1
),wecalculatedthe
compressionratioofshocktobe~4.0.Foraweakadiabaticshock,the
postshockdensitycanonlymarginallyincreaseaccordingtotheR–H
conditionfordensity.Thelargecompressionratiowemeasureisincon-
sistentwiththeassumptionofaweakadiabaticshockinPredehletal.1
.
Furthermore,theestimated0.3-keVplasmadensityof0.002 cm−3
in Predehl et al.1
is a factor of about 5 times larger than their adopted
valueofthepreshockedhalodensityof4 × 10−4
cm−3
.However,accord-
ingtotheR–Hconditionfordensityforanon-radiativeshockofM = 1.5,
thedensityratioshouldbe~1.7instead.Eveninthelimitofaverystrong
shockM → ∞,thedensityjumpforanon-radiativeshockisboundedby
avalueof(γ + 1)/(γ − 1)bwhichequals4forγ = 5/3,andcannotbeashigh
as5.Thusweseethattheshockscannotbeadiabatic.
Detailed theoretical calculations of the shock properties of the
eROSITA bubbles are beyond the scope of this paper. Any successful
modeloftheseenigmaticbubblesmustexplaintheobservedthermal
andchemicalpropertiespresentedinthispaper.
Discussion
Comparisonwithpreviousstudies
Previous studies used a single-temperature model with fixed rela-
tive abundances to define the X-ray emission and inferred that the
Galactic bubble shells have temperatures of kT ≈ 0.3 keV (refs. 16–21)
orkT ≈ 0.4 keV(ref.26).ThisishigherthanthetemperatureoftheMW
CGM of ~0.2 keV, which led them to conclude that the bubble shells
representshock-heatedgas.Furthermore,usingtheratioofthepre-and
0.05
0.04
0.03
0.02
0.01
0
0.35
0.30
0.25
0.20
0.15
0.10
EM
(cm
–6
pc)
kT
(keV)
1.4
1.0
1.2
0.8
0.6
0.4
0.2
kT
(keV)
EM
(cm
–6
pc)
180 140 100 60 20 340
/ (°)
300 260 220 180
0 180 140 100 60 20 340
/ (°)
300 260 220 180
0
180
0
0.0025
0.0050
0.0075
0.0100
0.0125
0.0150
0.0175
140 100 60 20 340
/ (°)
300 260 220 180
0 180 140 100 60 20 340
/ (°)
300 260 220 180
0
a
b
Fig.3|Distributionoftheemissionmeasuresandthetemperaturesofthe
warm-hotandthehotcomponentsoftheX-rayemission. a,b,Thewarm-hot
component(a)andthehotcomponent(b).Themapcoversb > 15°andb < −15°.
Thereportederrorsareof1σsignificance.TheGalacticbubblesregionisshown
bythegrey-shadedband.Theredverticalbarsincludeerrorsaswellasthe
dispersionofthedataover10°bins.
4. Nature Astronomy
Article https://doi.org/10.1038/s41550-023-01963-5
postshocktemperatures,theseworksestimatedtheshockspeed,age
andenergyofthebubbles.Weshowthattheuseofasingle-temperature
model to represent the shell emission was too simplistic, leading to
incorrect physical modelling of the bubbles. In this work, using the
better spectral models, we accurately measured the temperatures,
EMsandrelativemetalabundancesoftheplasmainthebubbleshells.
Weestimatedthattheaveragedensityofthewarm-hotcomponent
of the bubbles is about four times larger than the preshock halo gas.
Foranadiabatic(non-radiative)shock,themaximumdensityjumppos-
sibleisequalto4,inthelimitofaverystrongshockwithMachnumber
M → ∞.Butsuchastrongshockshouldalsocauseasubstantialincrease
in temperature. Given that the pre- and postshock temperatures are
similar, and the compression ratio of the shock is high, we rule out
that the bubble shells trace adiabatic shocks, in contrast to what was
assumedinPredehletal.1
(Methods).
Activegalacticnucleusorstellarfeedback?
ThephysicaloriginoftheGalacticbubblesisstillunderdebate.Since
the discovery of the Fermi bubbles, there have been a lot of efforts to
understand the formation mechanism of the bubbles, with several
theoretical models proposed in the literature. On the basis of their
feedback mechanisms, these models can be broadly divided into two
categories:oneisthenuclearstar-formingactivitysimilartostarburst
galaxiesandtheotheristhepastactivegalacticnucleusactivityofthe
GC supermassive black hole.
Metal abundance measurements provide a useful insight on the
originofthebubbles.Inthestar-formationactivityscenarios,thebub-
blesareenrichedbymetalsproducedbySNeandstellarwinds,whose
abundancesaredifferentfromthatintheinterstellarmedium.Onthe
otherhand,intheactivegalacticnucleuswindscenario,theabundance
ofthewindwouldbethesameastheambientinterstellarmediumthat
accretes onto the GC supermassive black hole. In this work, we have
measuredsupersolarabundancesofneonandmagnesium,compared
withoxygen,towardsafewsightlinespassingthroughthebubbles;this
supportsthestarformation-relatedfeedbackscenariofortheforma-
tionoftheGalacticbubbles.
Methods
Data selection and reduction
In this work, we analysed the Suzaku archival observations probing
the eROSITA bubbles’ regions towards the centre of the Galaxy, as
well as the surrounding fields. For the Galactic bubbles’ regions we
selected observations with exposure time of ≥20 ks. As can be seen in
the eROSITA all-sky map, the surrounding fields are much fainter in
X-rays; therefore, we selected the observations with higher exposure
times of ≥50 ks. Furthermore, to avoid the contamination from the
Galactic disk, we chose targets at least 15° above/below the Galactic
plane. This yielded multiple observations of 150 and 80 fields, prob-
ingtheGalacticbubblesandthesurroundingregions,respectively.
WeperformedSuzakudatareductionwithHEAsoftv.6.29.Weused
thedatafromtheback-illuminatedX-rayimagingspectrometer1(XIS1)
detector only, as this has better sensitivity at low energies than the
front-illuminatedXIS0andXIS3detectors.Wecombinedthedatataken
inthe3 × 3and5 × 5observationmodes.Weappliedextrascreeningto
thedatainadditiontothestandardscreeningdescribedintheSuzaku
Data Reduction Guide. To minimize the detector background, we
excludedtimeswhenthecut-offrigidity(COR)oftheEarth’smagnetic
fieldwaslessthan8 GV(thedefaultvalueisCOR = 2 GV).Furthermore,
weincreasedthefiltervalueoftheanglebetweenSuzaku’slineofsight
andthelimboftheEarth(ELV)fromthedefaultvalueof5°to10°.This
minimizestheexcesseventsinthe0.5–0.6 keVbandduetosolarX-rays
scatteredofftheEarth’satmosphere27
.
TheactivityofourownSuncanaffectthespaceweatherandcon-
taminate data taken by space observatories. The Sun was at its mini-
mum in the 11-year solar activity cycle when Suzaku was launched on
10July2005,approachingitsmaximumfromearly2011to2014.Solar
X-rays interact with the neutral oxygen in the Earth’s atmosphere and
generate a fluorescent emission line at 0.525 keV (ref. 28). This line in
thesoftX-raybandcanbedetectedbyinstrumentsonboardsatellitesin
thelow-Earthorbits,suchasSuzaku.Guptaetal.12
reportedthat,infour
Suzaku spectra taken in 2014, the O i intensity was about 25% to 130%
oftheOviiintensity(attemperaturesofafewmillionkelvin,theOvii
andOviiiemissionlinesarethedominantfeaturescharacterizingthe
MWCGMorthebubbles).TheOicontaminationcanbeminimizedby
removingeventstakenduringtimeintervalswhentheelevationangle
fromthebrightEarthlimb(theDYE_ELVparameter)islarger28
,aswedid.
Forobservationstakenin2011–2015,wecarefullyquantifiedtheO
ifluorescencelinecontaminationinouranalysis(fordetailsseeref. 12).
WeexaminedtheOiemissionwithrespecttodifferentDYE_ELVvalues
(>20°, >40° and >60°) and selected the best value for the DYE_ELV
parameterthatprovidedagoodbalancebetweenoptimizingtheeffec-
tive exposure time and mitigating the O i contamination. We then
modelledtheresidualOiemissionwithaGaussianlineinthespectral
analysis. For observations taken before 2011, we applied standard
screeningofDYE_ELV > 20°.
Thegoalofthisworkwastoanalysethediffuseemission;henceit
wasimportanttoremovepointsources.Wegeneratedthe0.5–2.0 keV
imagesandidentifiedthebrightpointsources.Weselectedthepoint
sourceexclusionregionsofradiiof 1′
− 3′
(cf.SuzakuXRT’shalf-power
diameter of 1.8′ to 2.3′). Then we extracted the diffuse emission spec-
trum from the entire field of view after excluding the point source
regions. We produced the redistribution matrix files using the xisrm-
fgenftool,inwhichthedegradationofenergyresolutionanditsposi-
tion dependence are included. We also prepared ancillary response
files using xissimarfgen ftool. For the ancillary response file calcula-
tions,weassumedauniformsourceofradius20"andusedadetector
maskthatremovedthebadpixelregions.Weestimatedthetotalinstru-
mentalbackgroundfromthedatabaseofthenightEarthdatawiththe
xisnxbgenftool.
Spectralanalysis
We performed all the spectral fitting with Xspec v.12.11.1 (ref. 29). We
modelledallthethermalplasmacomponentsincollisionalionization
equilibriumwiththeAPEC(v.3.0.9)modelandusedsolarrelativemetal
abundances30
.ForabsorptionbytheGalacticdisk,weusedthephabs
modelinXspec.
Suzaku provides an opportunity to resolve the different compo-
nentsoftheSDXBasaresultofitslowandstabledetectorbackground
even at low energies (0.3–1.0 keV). The SDXB spectrum is usually
describedbyathree-componentmodelconsistingof(1)aforeground
component of the Local Bubble and solar wind charge-exchange,
modelled as an unabsorbed thermal plasma emission in collisional
ionization equilibrium, (2) a background component of cosmic X-ray
background (made of unresolved point sources) modelled with an
absorbed power law and (3) the MW halo emission, modelled as an
equilibrium thermal plasma absorbed by the cold gas in the Galactic
disk(thehaloemissiontowardstheGCisdominatedbythebubbles1
).
Recentlywefoundthatinafewobservationsanextraabsorbedthermal
componentand/orenhancedNeabundanceisrequiredtoexplainthe
excess emission near 0.7–0.9 keV in the Suzaku12
and XMM-Newton11
SDXBspectra.
We started with fitting the Suzaku SDXB spectra with a
three-component model. The temperature of the foreground com-
ponent was frozen at kT = 0.1 keV (for example, refs. 31–34), but we
allowed the normalization to vary. We modelled the Galactic bubbles
(or the extended CGM) emission as single-temperature collisionally
ionized plasma characterized by temperature (kT) and EM, and with
fixed metallicity. The X-ray emission data do not contain any line or
edgeofhydrogen.Thuswecannotobtainabsolutemetalabundances
fromX-rayemissiondataalone.Instead,theX-rayobservationsprovide
5. Nature Astronomy
Article https://doi.org/10.1038/s41550-023-01963-5
constraints on relative metal abundances, for example N/O, C/O and
Ne/O. We fixed the total metallicity to 1 (in solar units) for both the
thermalcomponentsasthetotalmetallicityandnormalizations(orEM)
are degenerate in the APEC model. We allowed the power-law photon
indexandthenormalizationtovaryinthespectralfits.
This three-component model provided a poor fit to most of the
datasets, showing strong excess emission at low (~0.4–0.5 keV) and
high(0.8–1.0 keV)energybands.AnexampleoftheSuzakuspectrum
for one observation showing these excess emissions is shown in Sup-
plementaryFig.1(toppanel).
SinceNviiandNeixhavestrongtransitionsat0.5 keVand0.9 keV,
respectively, we allowed the nitrogen and neon relative abundances
to vary in our above model. That provided a slightly better fit but still
left notable excess emission at the higher energy side (0.8–1.0 keV).
To fit the higher-energy excess emission we added an extra thermal
componenttoourmodel.Thissubstantiallyimprovedthefitformost
of our datasets. An example of the best-fit two-temperature model is
showninSupplementaryFig.1(bottompanel).Thetemperatureofthe
secondthermalcomponentismuchhigher(kT = 0.4–1.1 keV)thanthat
of the first (kT ≈ 0.2 keV, known as the warm-hot component); we call
thisthehotcomponent.
A recent study35
has shown that the CGM spectra can be fitted
with a non-equilibrium ionization model. This, in principle, could be
an alternative to our two-temperature model. To test this possibility,
we fitted our data with the non-equilibrium ionization model, but
found that the fits were worse. Therefore, we use our results of the
two-temperaturemodelinallthefurtherdiscussion.
We have used abundances from Anders and Grevesse30
in the
aboveanalysis.Weobtainedsimilarresultsusingtheabundancesfrom
Loddersetal.36
.
Distributionofthermalparameters
Galactic bubbles’ region. The temperature of the warm-hot com-
ponent from the bubble shells is consistent within errors, with an
average value of kT = 0.205 ± 0.003 ± 0.002 keV (statistical and sys-
tematic errors). The EMs of the warm-hot component of the bubbles
regionsvariesgreatlyintherange2.2–46.9 × 10−3
cm−6
pcwithamean
of 13.9 × 10−3
cm−6
pc (and median of 12.7 × 10−3
cm−6
pc). Overabun-
dance of nitrogen by 1.3–10.3 solar in the warm-hot phase is required
for most of the observations that are not contaminated by the local
O i emission. In observations contaminated by O i, we were not able
toconstrainthenitrogenabundance;therefore,wefixedthattosolar.
A few sightlines also require supersolar abundances of neon and
magnesium,comparedwithoxygen.
The measured temperatures and EMs of the hot gas in the bub-
ble regions are in the range of 0.4–1.1 keV and 0.4–13.9 × 10−3
cm−6
pc,
respectively, with mean values of 0.741 ± 0.018 keV and
2.3 × 10−3
cm−6
pc. The emission from the hot component is substan-
tiallyfainterthanthatfromthewarm-hotcomponent.
Extended halo region.Thewarm-hotcomponenthasauniformtem-
perature of kT = 0.201 ± 0.004 ± 0.003 keV, similar to those in the
bubbles’ regions. The hot component has a temperature in the range
of kT = 0.4–1.2 keV, with a mean value of 0.837 ± 0.028 keV. The aver-
agetemperatureofthehotcomponentisslightlylowerinthebubbles’
regionincomparisontotheouterhalo,althoughthetwoareconsistent
with each other within 3σ. However, for both components, the EMs
in the extended halo regions are much lower than the EMs from the
bubbles regions. The EMs of the warm-hot phase are in the range of
0.8–14.2 × 10−3
cm−6
pcwithameanof4.4 × 10−3
cm−6
pc.Thehotphase
EMs are much lower, with a range of 0.2–1.5 × 10−3
cm−6
pc and a mean
of 6.1 × 10−4
cm−6
pc. We also found that nitrogen is overabundant by
1.0–11.4solarinthewarm-hotphase.However,supersolarabundances
of neon and magnesium, compared with oxygen, are not required
towardsanyofthesightlines.
Thermal parameters of the Galactic bubbles and the extended
haloregionsaregiveninSupplementaryTable1.
Northernversussouthernbubbles.Wecomparedthethermalproper-
tiesofthenorthern(b> 15°)andthesouthern(b< −15°)bubbles.Wehave
plottedthetemperaturesandEMsoftheGalacticbubblessightlinesver-
sustheGalacticlatitudeinSupplementaryFig.2.Thesightlinesprobing
thenorthernbubblehavecomparativelyhigherEMsthanthesouthern
bubble, but their temperatures are similar. For the northern bubble,
the warm-hot and the hot components have average temperatures of
0.203 ± 0.003 ± 0.002 keV and 0.734 ± 0.018 ± 0.010 keV and average
EMs of 14.8 ± 0.9 ± 0.2 × 10−3
cm−6
pc and 2.5 ± 0.2 ± 0.1 × 10−3
cm−6
pc,
respectively. The warm-hot and the hot components of the south-
ern bubble have similar temperatures of 0.210 ± 0.005 ± 0.003 keV
and 0.759 ± 0.024 ± 0.020 keV, but have lower EMs of
9.4±1.1±0.3×10−3
cm−6
pcand1.6±0.4±0.1×10−3
cm−6
pc,respectively.
TheEMofthewarm-hotcomponentdecreaseswithGalacticlati-
tudeouttoaboutb ± 45,thenbecomescomparativelyuniform.Thehot
component EM variation shows a similar trend but is less prominent.
ThedecreaseintheEMwithGalacticlatitudeisinagreementwiththe
eROSITAX-rayemissionall-skymap,whichshowsverybrightemission
at the base of the bubbles, with the surface brightness falling mono-
tonically away from the base. We do not find any such relation in the
distribution of temperatures with the Galactic latitude. This further
confirms that regions with brighter emission in the eROSITA all-sky
maphavehigherEMsbutarenothotterthanthesurroundingmedium.
For the northern and the southern bubbles, the total X-ray
surface brightness (0.5–2.0 keV) of the warm-hot component is
3.1 ± 0.6 × 10−15
ergs cm−2
s−1
arcmin
−2
.Assumingaprojectedareaofthe
eROSITAbubblesof35° × 35° × πforeachbubble(fromPredehletal.1
),
we calculated a total flux of 6.5 ± 0.9 × 10−8
ergs cm−2
s−1
and
4.1 ± 0.5 × 10−8
ergs cm−2
s−1
for the northern and southern bubbles,
respectively. Further assuming a distance of 10.6 kpc (from Predehl
et al.1
), we estimated the luminosities of the northern and southern
bubbles to be 8.7 ± 1.3 × 1038
ergs s−1
and 5.6 ± 0.8 × 1038
ergs s−1
,
respectively.
Dataavailability
ThedatapresentedinthispaperarepubliclyavailableattheHighEnergy
AstrophysicsScienceArchiveResearchCenter(HEASARC)archive.
References
1. Predehl, P. et al. Detection of large-scale X-ray bubbles in the
Milky Way halo. Nature 588, 227–231 (2020).
2. Snowden, S. L. et al. First maps of the soft X-ray diffuse
background from the ROSAT XRT/PSPC All-Sky Survey. Astrophys.
J. 454, 643–653 (1995).
3. Su, M., Slatyer, T. R. & Finkbeiner, D. P. Giant gamma-ray bubbles
from Fermi-LAT: active galactic nucleus activity or bipolar galactic
wind? Astrophys. J. 724, 1044–1082 (2010).
4. Nicastro, F. et al. Chandra discovery of a tree in the X-ray forest
toward PKS 2155-304: the local filament? Astrophys. J. 573,
157–167 (2002).
5. Williams, R. J. et al. Probing the local group medium toward
Markarian 421 with Chandra and the Far Ultraviolet Spectroscopic
Explorer. Astrophys. J. 631, 856–867 (2005).
6. Williams, R. J., Mathur, S. & Nicastro, F. Chandra detection of
local warm-hot gas toward Markarian 279. Astrophys. J. 645,
179–185 (2006).
7. Williams, R. J., Mathur, S., Nicastro, F. & Elvis, M. Chandra and Far
Ultraviolet Spectroscopic Explorer observations of z~0 warm-hot
gas toward PKS 2155-304. Astrophys. J. 665, 247–256 (2007).
8. Gupta, A., Mathur, S., Krongold, Y., Nicastro, F. & Galeazzi, M. A
huge reservoir of ionized gas around the Milky Way: accounting
for the missing mass? Astrophys. J. Lett. 756, L8–L14 (2012).