Far-UV Observations of a Thermal Interface in the Orion-Eridanus Superbubble

0.18 MB·English

by  J. Kregenow

Accepted for Publication inApJL. Contact: [email protected]
PreprinttypesetusingLATEXstyleemulateapjv.6/22/04
FAR-UV OBSERVATIONS OF A THERMAL INTERFACE IN THE ORION-ERIDANUS SUPERBUBBLE
J. Kregenow1, J. Edelstein, E. Korpela, B. Welsh, C. Heiles2, K. Ryu3, K. Min, Y. Lim, I. Yuk4, H. Jin, K. Seon
Accepted for Publication inApJL. Contact: [email protected]
ABSTRACT
Diffusefar-UVemissionarisingfromtheedgeoftheOrion-Eridanussuperbubblewasobservedwith
theSPEAR imagingspectrometer,revealingnumerousemissionlinesarisingfrombothatomicspecies
and H . Spatial variationsin line intensities of Civ, Siii, andOvi, in comparisonwith soft X-ray,Hα
2
and dust data, indicate that these ions are associatedwith processes at the interface between hot gas
inside the bubble and the cooler ambient medium. Thus our observations probe physical conditions
6
of an evolved thermal interface in the ISM.
0
0 Subject headings: ultraviolet: ISM — line: identification — ISM: bubbles — ISM: lines and bands
2

n 1. INTRODUCTION hot and cooler gas are likely to be interacting. SPEAR
a (aka FIMS) spectrally images diffuse FUV background
The Orion-Eridanus Superbubble (OES) is a large
J radiation with λ/∆λ∼550, using two large field of view
cavity in the interstellar medium (ISM) created by
6 (FOV) imaging spectrographs optimized for measuring
some combination of stellar winds and supernovae
2 from the enclosed Orion OB1 stellar association, (e.g. diffuseemission: theShort’S’(900-1150˚A)andLong’L’

1 Reynolds & Ogden1979). Itscomparableproximityand (1350-1750 ˚A) channels. The instrument and on-orbit
v size (∼300 pc) give it a large apparent angular extent of performance are described in Edelstein et al. (2005a).
4 ∼30◦ (Burrows & Guo1996),allowingfordetailedstudy
8 of its structure. The complicated morphology and dy-
2. OBSERVATIONSANDDATAREDUCTION
5 namics,describedbyGuo et al.(1995),indicatethatthe
1 OES is still evolving and expanding into the ISM and SPEAR observed a fixed location in the OES where
0 thusitlikelyincludesavarietyofinterfacesandasymme- hot X-ray emitting gas abuts cooler Hα and infrared
6 triesresultingfrominteractingphasesofinterstellargas. (IR) emitting gas. The imaged field (see Fig.1)
0 The OES has clearly-delineated thermal boundaries be- spans the bright outermost Hα front called ’Arc B’ by
h/ tweenthe ambientISMandthesuberbubblecavity. The Boumis et al. (2001), and is large enough to simultane-
p cavity is bright in soft X-ray (SXR) emission, indicating ously measure both the hot X-ray region inside and the
- that it is filled with hot T∼106K gas (see Heiles et al. colder IR region outside the bubble. The target was ob-
o 1999). For example, the X-ray edge of the OES toward servedover 60 orbits in Dec 2003 for a cumulative expo-
tr Galactic South (lower right of Fig. 1) coincides with a sure time of 31 ks.
s ridge of Hα emission produced by cooler T∼104 K gas, The data were processed as described in
a
beyond which still-cooler neutral H and dust are seen in Edelstein et al. (2005b). We removed data recorded
v: 21-cm (Heiles et al. 1999) and infrared (Burrows et al. at times with high count rate (>10 s-1 for S and
Xi 1993). >20 s-1 for L channel) to mitigate terrestrial airglow
We took a detailed far ultraviolet (FUV) emission contamination. Spectra were binned by 1 ˚A, smoothed
ar spectrum with the SPEAR instrument to study the by a 3 ˚A running boxcar, and fluxed to Line Units (LU,
physical conditions across one edge of the OES, where photons s−1cm−2sr−1). All of the L channel photons
were mapped to the sky and the image inspected for
′
the presence of stars. Two <20-wide image features
are consistent with stellar profiles, with a flux of
45o
∼3-5×10−14 ergs−1cm−2˚A−1. Ifthe featuresare indeed
-15o caused by faint UV stars, then they contribute no more
◦
than 7% of the background continuum flux over 1 – an
unimportantsourceoferrorinourspectralline analysis.
Potential stellar contributions to the S channel are even
less significant.
-30o
Fig. 1.— Composite 1/4 keV, Hα, and dust map image 3. SPECTRALANALYSIS
(McGlynnetal. 1998) of an OESinterface ingreyscale [color im-
Therawspectrashowastrongunderlyingbackground
age online],overlaidwith SPEAR slitposition. TheSPEAR field
spans the edge of the OES wall, sampling both hot inside and component in addition to the many emission lines of in-
cooler outside material. The gridlines show ecliptic latitude and terest. We used a spectral model (Korpela et al. 2005)
longitude. tosimultaneouslyfitthebackgroundandoverlyingemis-
sion lines. The background model has an equivalent
1 SpaceSciences Lab,UnivofCA,Berkeley,CA94720 total intensity of 3500 continuum units (CU, photons
23 AKsotrreoaDAedpvta,nUcendivInosftCoAf,SBcie&rkeTleeyc,hC,3A059-4770210,Daejeon s−1cm−2sr−1˚A−1) in the S channel and 1000 CU in
4 KoreaAstro&SpaceSciInst,305-348,Daejeon, Korea L. The background signal primarily consists of con-
2 Kregenow et al.
tributions from instrumental dark noise (80% of the
background in S channel) and starlight scattered by TABLE 1
interstellar dust (∼60% in L), with instrumentally- Modeled Emission Linesin theOES
scattered geocoronal Ly-α emission also contributing to
Wavelength Species Intensity S/N
the S background. The starlight is fit by a canoni- λ(˚A) ID (103LU)
cal power-law (index=0.26) stellar luminosity function
whichis absorbedandscatteredby interveningdustand 977 Ciii <3.6 1.7
gas with N(Hi)≃4×1020cm−2, a value similar to the 990a,b Oi 0.2 0.4
N(Hi)≃5×1020cm−2 measured in this location by the 990b Niii 0.9 1.6
1013 Siii? 0.7 1.3
Leiden/Dwingeloo Survey (Hartmann & Burton 1997). 1032 Ovi 2.0 3.4
We subtract the fitted background model from the raw 1038 Ovi 1.0 1.8
1067 Ari? 0.8 1.5
spectrum to obtain a net emission spectrum. Many
1074a Hei 2.4 4.2
emission lines are apparent in this net spectrum. The
1085 Nii 1.8 3.0
strongest are terrestrial in origin - arising from airglow 1134a Ni 5.3 7.2
in Earth’s upper atmosphere - including the bright Ly- 1358a Oi 3.6 17.1
man series from 912-1026 ˚A, Oi at 990/1041 ˚A, some 1394 Siiv 0.9 5.5
1403b Siiv 0.4 2.8
fraction of Ni at 1133 ˚A, and Oi prominently at 1356 ˚A 1404b Oiv] 0.7 4.2
and possibly blending with Heii at 1641 ˚A. We identify 1417 Siv? 0.2 1.6
andattributemostoftheremaininglinestoastrophysical 1533 Siii 2.1 13.0
1549 Civ 2.2 12.7
sources, namely originating in the OES or ISM.
1640b Heii 0.5 2.0
Several significant emission features could not be ac-
1641a,b Oi 0.9 3.4
counted for by neutral or ionized atomic species. A the- 1657 Ci 0.6 2.3
oretical spectrum of H fluorescence lines (Draine 2004, 1671 Alii 1.9 6.3
2
private communication), however,provides a good fit to
most of these lines and blends. The total H2 flux is aAttributedtoAirglow
19 kLU in S channel, and 32 kLU in L. This is ∼10% bBlendedLine
of the integrated flux from the scattered stellar contin-
uum fit, whose band-averaged value is ∼400 and ∼570
CU in the S and L channel, respectively. Thus we have
(1533 ˚A). Energies of 114, 48, and 8 eV are required to
discovered a significant H component toward the OES.
2
produce Ovi, Civ, and Siii, respectively.
We defer discussion of this component to future work;
FortheCivandSiiianalysis,wedividedtheL-channel
see Ryu et al. (2005) for an analysis of the physical pa-
◦
FOV into 1.0 bins and made separate spectra for each
rametersand proposedlocation andoriginof the H gas
2
location. The Civ and Siii intensity were clearly dimin-
detectedbySPEAR overtheentireEridanusregion. For
ished in some locations, so we defined a baseline spec-
thepresentanalysis,wesubtractedthemodeledH com-
2
trum with relative intensity zero in the location where
ponenttoobtainanetatomicemissionspectrum,shown
both lines were weakest. We then subtracted this base-
in Fig. 4.
line Civ/Siii spectrum fromthe other spectra after scal-
We find lines fromspecies with a wide rangeofioniza-
ing to fit the continuum in a quiet region (around 1500
tionpotentials(6-114eV),indicativeofbothcoolandhot
interstellar gas along the sight-line. The spectral model ˚A). Assuming that the continuum did not change be-
(Korpela et al. 2005) was used to estimate the emission tween FOV locations, scaling and subtracting the base-
line strengths. Some ofthe mostsignificant(S/N>3)as- linespectrumleavesonlytheexcesslineemissionateach
trophysical line model detections are Ovi and Nii in S location. WethenfitGaussianlineprofilestotheresidu-
channel, and Siiv, Oiv], Siii, Civ, and Alii in L chan- alstomeasuretheCivandSiiiexcessintensity. Thereis
nel. A list of identified atomic lines, observed line cen- little if any Siii emission in the baseline spectrum, so we
ters to the nearest 1 ˚A, total line flux, and signal to assumethattheexcessintensityissimilartotheabsolute
noiseratioisshowninTable1. Thespectralmodelfinds intensity for Siii. We note that the absolute intensity of
traces of other lines including Feii, Ari and Siii, along Civ, however, may be significantly underestimated be-
with other undetected lines. Before these identifications cause there is a strong, broad Civ absorptionline in the
can be made secure, however, more careful analysis is scatteredstellarbackgroundnottakenintoaccounthere.
required to account for potential confusion from H flu- We measured Ovi relative intensity variations using
2
orescence and airglow lines. Such analysis was done for a similar approach. The Ovi (1032 ˚A) line is resolved,
Ciii (977˚A), which is close to the Lyγ (973˚A) airglow but close to the intense Lyβ airglow line (1026 ˚A),
line. We only establish an upper limit (90% confidence) which dominates all features in the S channel. So
to the Ciii flux, which is included in the Table 1. instead of scaling to the continuum, we performed a
near-unity scaling to fit the short wavelength shoulder
3.1. Spatial Variation of FUV Emission Intensity
of the Lyβ line profile. Fig 2 shows the spectra for the
Inordertounderstandthe originoftheobservedFUV four locations after this scaling but before subtraction.
emission-line species, we have examined the spatialvari- Note how the Ovi intensity is clearly enhanced in two
ationofemissionalongthelongdimensionoftheSPEAR adjacent locations. (At the distance of the OES, this
FOV for three bright emission lines that trace species of 1∼2◦ zone corresponds to 5∼10 pc.) Assuming that the
very different ionization states, and compared the spa- airglow line profile did not change between locations,
tial variations to other bands. The emission lines used scaling to Lyβ and subtracting the baseline leaves only
inthis analysisareOvi(1032˚A), Civ(1549˚A),andSiii the excess Ovi emission at each location. Again there
FUV Obs of O-E Superbubble 3
Fig. 2.—SPEAR spectra,zoomedinontheOvidoubletinthe Fig. 3.— Spatial Variation of emission along SPEAR’s FOV.
wingsofLyβ,for4locationsspanningtheOESedgewithatypical EmissionlineintensityinLineUnitsvs. skypositionisplottedfor
errorbarat1031˚A.ThesmoothsolidlineisaGaussianfittothe Ovi(1032 ˚A), Civ, and Siii emissionin 1◦ bins. For comparison,
spectrumatRA=3.47h,whereOviisclearlyenhanced. the lower panel shows relative (scaled) Dust, Hα, and 1/4 keV x-
rayemission. ThesharppeakinHαatRA=3.4histheexpanding
bubbleedge.
is little if any Ovi emission in the minimum location,
so we assume that the excess intensity is approximately
equal to its absolute intensity. The spatial variation analysis results are shown in
the upper panel of Fig. 3, which plots line strength
of the three analyzed species per one-degree bin along
4. DISCUSSION the FOV: four bins in S channel, and seven in L. The
The global structure and coherence of the OES is a high-ionization lines Ovi and Civ are expected to trace
matter of some debate. In particular, it has been sug- high-temperature gas, while the low-ionization Siii line
gestedthatoneofthetwoprominentHαfilamentsinthe should trace lower-temperature gas and dust. Since the
region(’ArcA’,seeBoumis et al.(2001))maybepartof SPEARFOVspansathermalboundaryintheOES,sep-
amoredistantshellstructureunrelatedtotheOES.But arate spectra sample locations both inside and outside
Arc A only grazes the edge of SPEAR’s FOV where we thesuperbubble. Thelowerpanelshowsspatialvariation
measure no enhancement, while Arc B’s FUV enhance- in emission from three other wavebands for comparison:
ment is quite prominent in the center of the FOV. Since 1.4keVSXR(Snowden et al.1995),dust(Schlegel et al.
the features are unconfused and the contribution from 1998), and Hα (Finkbeiner 2003). (Data obtained from
Arc A is minimal, we presume that our measurements NASA’s SkyView facility, http://skyview.gsfc.nasa.gov).
are dominated by Arc B. We note that both Civ and Ovi intensity are highest at
The present SPEAR observations reveal interstellar RA= 3.45h, where Hα emission peaks. This behavior
gas with a wide range of ionization toward the OES strongly supports the inference that the majority of Civ
whose origin and location we wish to clarify. The emis- and Ovi emission originates from the OES thermal in-
sionlinespeciesdetectedrangefromlowionizationstate terface. Theambientmediaoutsidethebubblearelikely
ions (e.g. Siii, Alii) which can be created entirely by far too cold to produce Ovi, while the hot, thin X-ray
photo-ionizationof the cold neutral ISM by the ambient emitting gas inside has too low of an Ovi or Civ ioniza-
interstellarradiationfield,tospecies(e.g. Siiv,Civ)that tion fraction and emission measure to produce measur-
can be created either by photo-ionization or by thermal able emission. The Siii emission, in contrast, increases
ionization,tospecies(e.g. Ovi,Heii,Oiii])likelycreated outside the bubble where the enhanced dust reddening
only by thermal excitation because of their high ioniza- map indicates cool ambient gas resides.
tion potentials or low optical depth. Whether the interface contains a fast interstellar
The OES has been previously observed twice in the shock is an important question about its nature.
UV. Paresce et al. (1983) observed regions of the OES Hartigan et al. (1987) provide a means for probing this
nearourtargetwithanarrow-bandphotometerandcon- using the Ovi/Ciii (977˚A) flux ratio, a sensitive indica-
cludedthattheirmeasuredincreaseintheUVcountrate tor of shock velocity. In an observation a few degrees
over the bubble edge was most likely due to some com- awayfromoursatasimilarlocationontheHαinterface,
bination of H fluorescence, thermal, or Hi two-photon Murthy et al. (1993) detect Ciii at 20 kLU but do not
2
emissionfromionizedgas. Twosmall,<1◦,fields(within robustly detect Ovi, at 10±10 kLU. Our firm 3 kLU de-
◦
5 of our targetbut not spanning the bubble edge) were tection of Ovi is consistent with their result, being well
observed with the Voyager low-resolution spectrometer belowtheirdetectionthreshhold,butourCiiiupperlimit
(Murthy et al. 1993). They attributed a strong FUV is much smaller at 3.6 kLU. The Ovi/Ciii ratios from
continuumtostarlightscatteredbyinterstellardustwith Voyager and SPEAR data, <1 and >0.8 respectively,
possiblesmalladditionalcontributionsfromunidentified could arise from a shock with velocity 140-180 km/s ac-
emission lines and two-photon continuum emission from cordingtotheHartiganmodels. However,Naranan et al.
Hiradiativedecay. Thesepreviousdatawereinsufficient (1976) compared the OES to Vela and Cygnus – two
to detect spectral line variation across the bubble inter- middle-aged supernova remnants known to contain fast
face, leaving questions unanswered about how the FUV shocks – and found Eridanus to be older, larger, and
emission is linked to the physical properties of the bub- fainter in X-ray emission, concluding that if the OES
ble’s interface with the ambient interstellar media. hasasupernovaoriginthenitismoreevolvedwithmuch
4 Kregenow et al.
Fig. 4.—FluxedSPEAR spectra(thickline)aftersubtractionofmodeledcontinuaa andH2 areoverplotted withamodeledcomposite
spectrum (thin line) of both astrophysical and atmospheric airglow emission lines. Geocoronal airglow lines are indicated by the Earth
symbols,including4thorderHei537˚A.
lower velocities. Additionally, the SPEAR observations not at the bubble edge. We measured intensity varia-
of Vela (Nishikida et al. 2005) and Cygnus (Seon et al. tions of Ovi, Civ, and Siii emission across the interface
2005) show two ordersof magnitude more OVI and CIII between the hot bubble interior and the ambient cooler
emission than Eridanus, and have OVI/CIII ratios be- media. The Ovi and Civ emission are greatly enhanced
tween one and two. Moreover,there are no direct obser- at the interface, and clearly show an ionized stratifica-
vations of high-velocity gas in the Eridanus region that tion. While these ions could be formed in a fast shock,
couldproducehighly-ionizedspeciessuchasOvibyther- we suggest that a quiescent thermal interface model is
mal heating from fast shocks. Both Heiles et al. (1999) alsoconsistentwiththeFUVobservationsandwithother
and Reynolds & Ogden (1979) showed the OES Hi/Hα previous observations finding only low-velocity gas. We
shell to be expanding at <20 km/s, and interstellar ab- have also observed a significant H component, despite
2
sorptionobservations(Welsh & Lallement2005)showno the moderate total hydrogen column in this direction.
high-velocityabsorptioncomponents. Finally,iftheVoy- The spatial variation of emission line intensity across
ager and SPEAR measurements are sampling a similar theinterfaceholdsmuchpromiseforprobingthe physics
interface, then it is puzzling that the Voyager measure- of the interface, especially when applied to species of
ment of Ciii so far exceeds the SPEAR upper limit. We widelyvaryingionizationpotential. Tofurtherprobethe
believethedifficultyinassesmentofbackgroundsubtrac- physicsofthe OESregion,weintendtoextendouranal-
tionandseparatingLyγ fromCiiiinVoyager datacould ysis to additional spectral lines and to two more deep
havecompromisedtheresultandexplainthediscrepancy. observations SPEAR made of the OES toward nearby
We conclude that a fast shock is one possibility at fields. We will also test interface models predicting the
the interface, but there is not a plurality of evidence to spectral components. These new data can contribute to
supportit. We suggestthataslowshockora’quiescent’ theunderstandingofinterstellarthermalheatingsources
(non-shock) thermal interface between the hot X-ray and interfaces by comparing the FUV emission line ob-
producing interior cavity of the OES and the cool servations with predictions made with interface models
ambient medium could be responsible for producing the invoking shocks, conduction, and turbulent mixing.
highly-ionized species we observe.
5. CONCLUSIONS SPEAR/FIMS is a joint project of KASSI, KAIST,
We have discovered numerous diffuse atomic and and U.C. Berkeley, funded by the Korea MOST and
molecular FUV emission lines emanating from both hot NASA Grant NAG5-5355. J. Kregenow is supported by
and cool gas toward the OES. Of these lines, only Ovi the National Physical Sciences Consortium Fellowship,
has been previously detected in the OES at all, though and C. Heiles in part by NSF grant AST 04-06987.
REFERENCES
Boumis,P.etal.2001,MNRAS,320,61 McGlynn, T., Scollick, K., & White, N. 1998, in IAU Symp. 179:
Burrows, D. & Guo, Z. 1996, in Roentgenstrahlung from the NewHorizonsfromMulti-WavelengthSkySurveys,465–+
Universe,221–224 Murthy,J.,Im,M.,Henry,R.,&Holberg,J.1993,ApJ,419,739
Burrows,Singh,Nousek,Garmire,&Good1993, ApJ,406,97 Naranan,Shulman,Friedman,&Fritz1976, ApJ,208,718
Edelstein,J.etal.2005a, ApJ,inpress(thisvolume) Nishikida,K.etal.2005, ApJ,inpress(thisvolume)
—.2005b,ApJ,inpress(thisvolume) Paresce,F.,Bowyer,S.,&Jakobsen, P.1983,A&A,124,300
Finkbeiner,D.P.2003, ApJS,146,407 Reynolds,R.J.&Ogden,P.M.1979,ApJ,229,942
Guo,Burrows,Sanders,Snowden,&Penprase1995,ApJ,453,256 Ryu,K.etal.2005, ApJ,inpress(thisvolume)
Hartigan,P.,Raymond,J.,&Hartmann,L.1987,ApJ,316,323 Schlegel,D.J.,Finkbeiner,D.P.,&Davis,M.1998,ApJ,500,525
Hartmann, D. & Burton, W. B. 1997, Atlas of galactic neutral Seon,K.I.etal.2005,ApJ,inpress(thisvolume)
hydrogen(Cambridge;NewYork:CambridgeUniversityPress) Snowden,S.L.,etal.1995,ApJ,454,643
Heiles, C., Haffner, L. M., & Reynolds, R. J. 1999, in ASP Conf. Welsh,B.Y.&Lallement,R.2005,A&A,436,615
Ser.168:NewPerspectives ontheInterstellarMedium,211–+
Korpela,E.J.etal.2005,ApJ,inpress(thisvolume)