The Milky Way halo as a QSO absorption-line system. New results from an HST/STIS absorption-line catalogue of Galactic high-velocity clouds

1.8 MB·English

by  Peter Herenz

Astronomy&Astrophysicsmanuscriptno.hvc˙paper (cid:13)c ESO2013
December11,2013
The Milky Way halo as a QSO absorption-line system
New results from an HST/STIS absorption-line catalogue
of Galactic high-velocity clouds
P.Herenz1,P.Richter1,2,J.C.Charlton3,andJ.R.Masiero4
1 Institutfu¨rPhysikundAstronomie,Universita¨tPotsdam,Karl-Liebknecht-Strasse24/25,14476Potsdam-Golm,Germanye-mail:
[email protected]
2 Leibniz-Institutfu¨rAstrophysikPotsdam(AIP),AnderSternwarte16,14482Potsdam,Germany
3 3 DepartmentofAstronomyandAstrophysics,PennsylvaniaStateUniversity,UniversityPark,PA16802,USA
1 4 JetPropulsionLaboratory,4800OakGroveDrive,Pasadena,CA91109,USA
0
2 AcceptedbyA&Aon17Dec2012

n
Abstract
a
J
Weusearchival UVabsorption-linedatafromHST/STIStostatisticallyanalysetheabsorptioncharacteristicsofthehigh-velocity
7
clouds (HVCs) in the Galactic halo towards more than 40 extragalactic background sources. We determine absorption covering
O] farnaacltyiosinsswofelcoown-ceanntdraitneteornmeSdiiiaiteabisoonrspt(iOoni,cComii,pSoiniei,nMtsginii,HFVeCii,s,Sifoiiri,wChiivc,hanwdeSiinivve)stiingathteetrhaengdeisftcrib=ut0io.2n0o−f 0co.7l0u.mFnordednestaitiileesd,
b-values, and radial velocities. Combining information for Siii and Mgii, and using ageometrical HVC model weinvestigatethe
C contributionofHVCstotheabsorptioncrosssectionofstrongMgiiabsorbersinthelocalUniverse.WeestimatethattheGalactic
. HVCswouldcontributeonaverage∼ 52percenttothetotalstrongMgiicrosssectionoftheMilkyWay,ifourGalaxyweretobe
h observedfromanexteriorvantagepoint.WefurtherestimatethatthemeanprojectedcoveringfractionofstrongMgiiabsorptionin
p
theMilky Wayhalo anddisc froman exterior vantage point ishf i = 0.31 for ahalo radiusof R = 61kpc. Thesenumbers,
o- together withtheobserved number density of strongMgiiabsorbec,rsMsgaIIt low redshift,indicate thatthecontribution of infallinggas
r clouds(i.e.,HVCanalogues)inthehalosofMilkyWay-typegalaxiestothecrosssectionofstrongMgiiabsorbersis<34percent.
t Thesefindingsareinlinewiththeideathatoutflowinggas(e.g.,producedbygalacticwinds)inthehalosofmoreactivelystar-forming
s
a galaxiesdominatetheabsorption-crosssectionofstrongMgiiabsorbersinthelocalUniverse.
[
Keywords.Galaxy:halo-galaxies:halos-ISM:structure

1
v
5 1. Introduction Collinsetal.(2003);Richteretal.(2005);Richteretal.(2009);
4
Shulletal. (2011)), indicating that HVCs and IVCs have vari-
3
The so-called high-velocity clouds (HVCs) in the halo of the ousorigins.SomeHVCs,inparticularthe“MagellanicStream”
1
. MilkyWayarebelievedtorepresentanimportantphenomenon (MS),mostlikelyoriginatefromgasofsmallersatellitegalaxies
1 related to the ongoing formation and evolution of galaxies that are being accreted by the Milky Way. Other HVCs possi-
0 at z = 0. HVCs (in their original definition) represent neu- blyrepresentmetal-deficientgasthatisinfallingfromtheinter-
3
tral gas clouds seen in Hi 21cm emission that are circulat- galacticmedium.ThemostlikelyoriginfortheIVCs(whichpre-
1
ing with high radial velocities through the inner and outer re- dominantlyhavenearlysolarmetallicities)isthe“galacticfoun-
:
v gions of the Galactic halo. HVCs in the Milky Way and other tain” (Shapiro&Field (1976); Houck&Bregman (1990)). In
Xi galaxies are assumed to connect the central regions of galax- thegalacticfountainmodelhotgasisejectedoutoftheGalactic
ieswiththesurroundingintergalacticmedium(IGM).Following disc by supernova explosions. The gas then cools and (partly)
r
a the usual classification scheme, HVCs in the Milky Way have fallsbacktowardsthediscintheformofcondensed,neutralgas
radial velocities of |v | > 90 kms−1 compared to the lo- clouds.
LSR
cal standard of rest (LSR). Such high velocities are not in Recent distance estimates of several IVCs and HVCs indi-
agreement with a standard Galactic disc rotation model. In cate that most of the IVCs appear to be located within 2 kpc
the velocity range 40 kms−1 < |v | < 90 kms−1 there from the Galactic disc, in accordance with the scenario that
LSR
are the so-called intermediate velocity clouds (IVCs), which IVCs represent gas structures related to the galactic fountain
complete the classification of Galactic halo clouds. Since their (Wakkeretal.(2008);Smokeretal.(2011)).MostoftheHVCs
discovery more than 40 years ago (Mulleretal. (1966)), a appeartobelocatedatdistances<20kpc(Wakkeretal.(2007);
lot of progress has been made in understanding the distribu- Thometal.(2006),Thometal.(2008)),withtheprominentex-
tion, origin, and physical properties of HVCs and IVCs (see ception of the MS, which most likely is located as far as 50
the reviews by Richter (2006) and Wakkeretal. (1998)). QSO kpc(Gardiner&Noguchi(1996)).Thesedistancesindicatethat
absorption-line measurements have shown that HVCs span a HVCsdonotrepresentLocalGroup(LG)objectsthatarerelated
relatively large range in metallicities from ∼ 0.1 to 1.0 solar to the missing dark-matter (DM) halos in the LG (Blitzetal.
(e.g., Wakkeretal. (1999); Richteretal. (1999); Richteretal. (1999)),butratherindicategascirculationprocessesin the im-
(2001);Wakker(2001);Gibsonetal.(2001);Trippetal.(2003); mediateenvironment(d <100kpc)oftheMilkyWay.Yet,with
1
P.Herenz,P.Richter,J.C.Charlton,andJ.R.Masiero:TheMWhaloasQSOabsorber
a total Hi mass of ∼ 108 M , HVCs contribute ∼ 0.7M yr−1 Table1.ProminentionsintheHST/STISspectra
⊙ ⊙
to the Milky Way’s gas-accretion rate (Richter (2012);Wakker
(2004)). Clearly, detailed studies of IVCs and HVCs are of Grating R=λ/∆λ Ion
fundamental importance for our understanding of the past and E140M 45,800 Siii,Siiii,Siiv,Oi,Cii,
presentevolutionofourGalaxy. Civ,Alii
Hi 21cm observations of nearby spirals (e.g., M31, E230M 30,000 Feii,Mgii
NGC891) indicate that the IVC/HVC phenomenon is not re-
strictedtotheMilkyWay,butreflectsgas-circulationprocesses
ments(e.g.,from FUSE) and from21cmstudies of neutralhy-
at large scales that are characteristic of low-redshift galax-
drogen,butthestrategiesofhowtousethesesupplementarydata
ies in general (Thilkeretal. (2004); Oosterlooetal. (2007);
toderivecolumndensitiesandotherparametersarequitediffer-
Fraternalietal. (2007)). However, because of the limited sen-
ent among the above listed studies. As a result, there currently
sitivity and beam size of 21cm observations of more distant
does not exist an homogeneous and coherent sample of HVC
galaxies it is currently impossible to spatially resolve individ-
absorption-lineparameters(columndensities,Dopplerparame-
ualgascloudsinthehalosofgalaxiesbeyondtheLocalGroup.
ters)forimportantionsobtainedbymulti-componentVoigtpro-
As an alternative method, QSO absorption spectroscopy has
file fitting based on the same fitting criteria for all HVC sight-
turned out to be a powerfultechnique to trace neutraland ion-
lines.Acoherentabsorption-linesampleisdesired,however,to
ized gas in the extended halos of galaxies at low and at high
provide a meaningful comparison between Galactic HVC ab-
redshift. For this, absorption lines from low and high ioniza-
sorbers and intervening QSO absorbers. This motivated us to
tion metals, such as Mgii and Civ in intervening absorbers,
reanalyse HST/STIS data of Galactic HVCs using the profile-
have been analysed extensively (Bergeron&Boisse´ (1991);
fittingtechniqueforallHVCabsorptionfeaturesdetected.
Charlton&Churchill(1998); Steideletal. (2002); Nestoretal.
(2005);Kacprzaketal.(2010)).However,sincemostoftheion
transitions of interest are located in the ultraviolet (UV), QSO
2. Dataaquisitionandanalysismethod
absorption spectroscopy of gaseous structures in and around
galaxiesandintheintergalacticmedium(IGM)atz=0requires Our HST/STIS data set contains 47 sightlines through the
spectroscopic data from space-based UV observatories, such Galactic halo towards QSOs and other AGNs. All spectra are
as the Hubble Space Telescope (HST) and the Far Ultraviolet publiclyavailableinthe MikulskiArchiveforSpace Telescope
SpectroscopicExplorer(FUSE).Asaconsequence,theamount (MAST). Fig.1 shows the sky distribution of the 47 sightlines
andthequality(intermsofsignal-to-noise,S/N)ofabsorption- plotted on a Hi 21cm map of the Galactic HVCs from the
line data of intervening metal absorbers rising in the halos of Leiden/Argentine/Bonn(LAB)all-skysurvey(Kalberla&Haud
low-redshiftgalaxiesisrelativelylimited. (2006)).Fromaninspectionoftheoriginalobservingproposals
Generally,QSO absorption-linemeasurementsindicatethat weconcludethatmorethan70percentofthe47selectedAGN
the circumgalactic environment of galaxies is characterized wereselectedbecauseoftheirultraviolet(UV)brightnessand/or
by a complex spatial distribution of multiphase gas that re- becauseoftheknownpresenceofinterveningmetal-absorption
flects both, the gas accretion processes of galaxies from the systems, as indicated by previous UV data from earlier HST
intergalactic medium and from merger events and the outflow spectrographswith lower spectral resolution. We therefore can
of gaseous material from galactic winds (e.g., Fanganoetal. assume that the 47 sightlines are not biased towards particular
(2007);Bouche´etal.(2012)).Yet,theexactmorphologicalrela- HVCs or HVC regions. Fig.1 indicates, however, that there is
tionbetweeninterveningmetalabsorbersandtheGalacticHVC a clear overabundanceofQSO sightlinesin the northernskyat
populationhasnotreallybeenestablished. b > 30 deg with both detections and non-detections of high-
In this paper, we reanalyse HVC absorption lines from velocityhalogasintheSTISdata.Thisnon-uniformskydistri-
archival UV spectral data obtained with the Space Telescope butionoftheQSOsinoursampleisfurtherdiscussedinSect.3.
ImagingSpectrograph(STIS)onHST andcomparetheabsorp- All STIS spectra considered in this paper were recorded
tion characteristics (absorption cross section, column-density using the E140M (λ = 1150 − 1700 Å) and the E230M
distribution function) of several ions with that of intervening (λ = 1600 − 3100 Å) high-resolution Echelle gratings of
metal absorption systems at low redshift. We focus on the ab- STIS. These instruments provide a spectral resolution of R ∼
sorption properties of Siii and Mgii in HVCs and QSO ab- 45,800 (E140M) and R ∼ 30,000 (E230M), corresponding
sorbers,asthesetwoionshaveverysimilarionizationpotentials to a velocity resolution of ∼ 7 kms−1 and ∼ 10 kms−1
andthusareparticularlywellsuitedforsuchacomparison.From FWHM, respectively. The STIS data were reduced using the
this comparisonwe deriveinformationon the relation between standardSTISreductionpipeline(Brown(2002)).Separate ex-
GalacticHVCsandQSOabsorbersandprovideanestimatefor posures were combined following the proceduresdescribed by
thecrosssectionofneutralandionizedgasstructuresintheha- Narayananetal. (2008). Table 1 lists prominent ions in the
losofgalaxies(seealsoRichter(2012);Richteretal.(2011)). two wavelength ranges that can be used to study the absorp-
Many of the HVC sightlines studied in this paper have tion properties of Galactic HVCs. The ion transitions consid-
been analysed in great detail by various different groups ered in this study include Cii λ1334.5, Civ λλ1548.2,1550.8,
(e.g.,Wakkeretal.(1999);Sembachetal.(2001);Richteretal. Oiλ1302.2,Siiiλλ1190.4,1193.3,1260.4,1304.4,1526.7,Siiii
(2001); Gibsonetal. (2001); Trippetal. (2003); Collinsetal. λ1206.5, Siiv λλ1393.8,1402.8, Mgii λ2796.4,2803.5, Feii
(2003); Foxetal. (2004); Richteretal. (2009); Collinsetal. λλ2382.8,2600.2, and Alii λ1670.8. Laboratory wavelengths
(2009);Shulletal.(2009))usingHST/STISdata.Thesegroups and oscillator strengths have been taken from the compilation
usedvariousdifferentspectralanalysistechniques(apparentop- ofMorton(2003).
ticaldepthmethod,AOD;curve-of-growth;profilefitting)tode- For our study we consider only HVCs, i.e., absorptionfea-
rive column densities and Doppler-parameters from their data. tures that have radial velocities|v | ≥ 90 kms−1, but not the
LSR
In addition, some of these studies (but not all) have incorpo- IVCs.Allinall,28outofthe47sightlinesobservedwithSTIS
ratedsupplementaryspectraldata fromlower-resolutioninstru- exhibitsignificantHVCabsorptionfeaturesinatleastoneofthe
2
P.Herenz,P.Richter,J.C.Charlton,andJ.R.Masiero:TheMWhaloasQSOabsorber
Figure1.Hi21cmskymapofHVCcomplexes,basedontheLABsurvey(Kalberla(2003),Kalberlaetal.(2005)).Thesymbols
mark the STIS sightlines inspected in this paper. Filled circles indicate sightlines with HVC detections, whereas empty squares
indicatesightlineswherenoHVCabsorptionwasfound.TheLABall-skymapwaskindlyprovidedbyT.Westmeier.
ionslisted above.Table 2 providesa summaryof the HVC de- 3. Results
tectionsinourSTISQSOsample.Thisoveralldetectionrateis
affectedbythestronglyvaryingS/Nratiosinthespectraandthe 3.1.Coveringfractionsofindividualions
resulting differing detection limits for the individual ion lines
The physical conditions in the gas, in particular the ionization
(seealsoSect.3).
conditions,areknowntovarysubstantiallyamongtheGalactic
The HVC spectral features have been fitted by multi- HVC population. HVCs span a large range in temperatures
component Voigt profiles, from which we obtain column den- andgasdensities,theyare subjecttothermalinstabilities, ram-
sitiesandDoppler-parameters(b-values)fortheindividualHVC pressure stripping, photoionization from the Galactic disc and
absorptioncomponents.Forthefittingprocesswehaveusedthe from the extragalactic UV background, collisional ionization
fitlyman routine implemented in the ESO-MIDAS software from hot material ejected by supernova explosions in the disc,
package(Fontana&Ballester(1995)).Afteraninitialinspection gasmixingprocesses,andotherrelatedphenomena.Asaresult,
of the velocity structure in each HVC we have simultaneously thecharacteristicabsorptionpatternsofHVCs(i.e,theobserved
fitted all low ions (neutral species and singly-ionized species; absorptionfrequencyof metalionsand their relativestrengths)
e.g.,Oi,Siii)ineachvelocitycomponentwithasingleb-value. canbeusedtoconstrainthephysicalconditionsinHVCs.
Thisapproachisjustified,asthelowionsareexpectedtoreside Allinall,wefit67individualhigh-velocityabsorptioncom-
inthesamegasphaseinHVCswherebisdominatedbyturbu- ponents(Voigtcomponents;seeabove)inourdataset.Ofthese,
lence (i.e., the thermalcontributionto b is expected to be neg- 47componentshavevelocityseparations∆v > 30kms−1 from
ligible). The absorption componentsof intermediate ions (e.g., neighbouringHVC absorptioncomponentsand thuscan be re-
Siiii)andhighions(e.g.,Siiv,Civ)werefittedindependentlyof garded as individualentities, hereafter refered to as absorption
the low ions(leadingto other b-valuesfor these ions),as these systems.
ionsmost likely trace a gasphase differentfromthat tracedby
Animportantparameterthatcharacterizesthedistributionof
thelowions. neutralandionizedHVCgasintheMilkyWayhaloisthecov-
BecausemanyspectrahaverelativelylowS/N,notallHVC ering fraction, f (X), for each ion X that is considered. As in-
c
velocitycomponentscanberesolvedwiththecurrentHST data dicatedin Table 3, we definethe coveringfractionas the num-
set.Moreover,high-resolution,high-S/NopticalspectraofHVC ber of sightlines that exhibitsignificantHVC absorptionin the
sightlines indicate that there often are a large number of ve- ion X divided by the total number of sightlines along which
locitysub-componentsinHVCswhoseidentificationwouldre- HVC absorption above the limiting column density threshold
quireaspectralresolutionmuchhigherthancurrentlyprovided (log N ) couldbe detected.Thecolumndensitythresholdfor
min
byspace-basedUVspectrographs(e.g.Weltyetal.(1999)).This each ionwas calculatedfromthe relevantion transitionsin the
systematic uncertainty is, however, not restricted to HVCs but STISwavelengthrange(seeSect.2)togetherwiththelocalS/N
is relevant also for the analysis of intervening metal absorbers ratio.
atlowzusingUVdatawithlimitedS/Nandspectralresolution Table 3 shows the values of f for the ions Oi, Cii, Siii,
c
(e.g.,Richteretal.(2004);Ribaudoetal.(2011)).FortheVoigt Mgii, Feii, Siiii, Siiv, and Civ, togetherwith the limiting col-
profilefittingpresentedinthispaperourstrategywastofindthe umndensities,logN ,asdeterminedfromourline-fittinganal-
min
minimum number of velocity components (Voigt components) ysis.Inthiswayweobtaincoveringfractionsfortheabove-listed
thatarerequiredto obtaina satisfyingfit tothe STISHVCab- ions between 0.20 (Siiv) and 0.70 (Cii, Siiii). Assuming that
soptionprofiles.Thisallowsustocompareourresultstostudies the sightlines and the HVCs are randomly distributed over the
ofinterveningabsorbers,forwhichsimilarfittingstrategieswere skywith f ≤ 1,andconsideringPoisson-likestatistics, theto-
c
chosen(e.g.Churchilletal.(2003)). talskycoveringfractionofHVCabsorptioninourdatais f =
c
All HVC fitting results are listed in Tables B.1−B.5 of the 0.70±0.15. This HVC coveringfraction is in excellent agree-
Appendix. mentwiththevalueof f =0.68±0.04derivedbyLehneretal.
c
3
P.Herenz,P.Richter,J.C.Charlton,andJ.R.Masiero:TheMWhaloasQSOabsorber
Table2.SummaryofQSOSightlinesandHVCDetections
QSOName zem l b HVCstatus HVCvelocityrange HVCname RelevantSTISgrating Detectedions
(deg) (deg) (yes/no) (kms−1) (ifknown)
PKS2155−304 0.117 18 −52 yes −138...−190 ... E140M Siii,Cii,Siiii,Siiv,Civ
NGC5548 0.020 32 +71 no - - E140M,E230M -
B2121−1757 0.110 33 −42 no - - E140M -
Mrk509 0.034 36 −30 yes −295...−310 GCN E140M,E230M Cii,Siiii,Siiv,Civ
CSO873 1.010 38 +84 no - - E230M -
PHL1811 0.192 47 −45 yes −130...−270 GCN E140M Siii,Cii,Oi,Alii,Siiii,Siiv,Civ
PG1630+377 1.480 60 +43 yes −160...−60 - E230M Mgii,Feii
PG1444+407 0.270 70 +63 yes −80...−90 - E140M Siii,Cii,Oi,Civ
PG1718+481 1.083 74 +35 yes −100...−215 CExtension E230M Feii
NGC7469 0.016 83 −45 yes −190...−400 MS E140M Siii,Cii,Oi,Siiii,Siiv,Civ
3c351 0.372 90 +36 yes −130...−230 ComplexC E140M Siii,Cii,Oi,Alii,Siiii,Siiv,Civ
Mrk290 0.030 91 +48 yes −128 ComplexC E230M Feii
Akn564 0.030 92 −25 no - - E140M -
H1821+643 0.297 94 +27 yes −130...−170 OuterArm E140M Siii,Cii,Oi,Alii,Siiii,Siiv,Civ
HS1700+6416 2.740 94 +36 no - - E140M -
PG1634+706 1.337 103 +37 yes −100...−215 ComplexC E230M Feii
Mrk279 0.031 115 +47 yes −150...−200 ComplexCsouth E140M Siii,Cii,Oi,Alii,Siiii,Siiv
PG1259+593 0.472 121 +58 yes −120...−145 ComplexCIII E140M Siii,Cii,Oi,Alii,Siiii,Civ
PG1248+401 1.030 123 +77 no - - E230M -
Mrk205 0.071 125 +42 yes −110...−230 ComplexCsouth E140M Siii,Cii,Oi,Alii
3c249.1 0.310 130 +39 yes −135 - E140M Siii,Cii,Siiii
PG0117+21 1.500 132 −41 yes −134 - E230M Mgii,Feii
NGC3516 0.009 133 +42 yes −160...−170 - E140M,E230M Siii,Cii,Feii,Mgii,Siiii
PG1206+459 1.160 145 +70 no - - E230M -
HS0624+6907 0.370 146 +23 no - - E140M -
NGC4051 0.002 149 +70 no - - E140M -
NGC4151 0.003 155 +75 yes +120...+145 ... E140M Siii,Cii,Siiii,Civ,Feii,Mgii
Mrk132 1.760 159 +49 yes −140...+80 - E230M Mgii,Feii
NGC4395 0.001 162 +82 no - - E140M,E230M -
PKS0232-04 1.440 174 −56 no - - E230M -
HS0747+4259 1.900 177 +29 no - - E230M -
PG0953+415 0.239 180 +52 yes −150 ComplexM E140M Siii,Cii,Siiii,Alii
HS0810+2554 1.510 197 +29 no - - E230M -
Ton28 0.330 200 +53 no - - E140M -
PKS0405−123 0.570 205 −42 no - - E140M -
PG1116+215 0.177 223 +68 yes +180...+190 - E140M,E230M Siii,Cii,Oi,Feii,Mgii,Siiii,Siiv,Civ
TonS210 0.117 225 −83 yes −150...−235 CHVC224.0−83.4−197 E140M,E230M Siii,Cii,Oi,Siiii,Siiv,Civ
HE0515−4414 1.713 250 −35 yes +120...+230 - E230M Mgii,Feii
PG1211+143 0.081 268 +74 yes +169...+184 ... E140M Siii,Cii,Oi,Siiii,Civ
PKS1127−145 1.187 275 +44 no - - E230M -
PG1216+069 0.330 281 +68 yes +210...+270 ... E140M Siii,Siiii,Civ
NGC3783 0.010 287 +23 yes +180...+250 LeadingArm(MS) E140M,E230M Siii,Cii,Oi,Alii,Feii,Mgii,Siiii
3c273 0.160 290 +64 no - EPn E140M -
RXJ1230.8+0115 0.117 291 +63 yes −216...−310 ... E140M Siii,Cii,Oi,Siiii,Siiv
PKS0312−770 0.223 293 −38 yes +160...+240 MB E140M,E230M Siii,Cii,Oi,Feii,Mgii,Siiii
PG1241+176 1.280 293 +80 no - - E230M -
PKS1302−102 0.290 309 +52 no - ... E140M -
(2012), based on a much larger combined COS/STIS data set. Table3.CoveringfractionsofindividualionsinHVCs
ThegoodagreementwiththeLehneretal.resultsindicatesthat
the non-uniformsky distributionofthe QSOs(asmentionedin
Sect.2)hasnosignificantinfluenceonthedeterminationofthe
Ion N/N a f b logN c
HVCcoveringfractionfromourSTISQSOsample. Cii 21/3to0t 0.c70 13.2m0in
Note thatwe do notconsiderabsorptionbyAlii in oursta- Civ 12/30 0.40 13.00
tistical analysis, because for the Alii λ1670.8 line (the only Oi 14/29 0.48 13.65
detectable Alii line in our data) there is a gap between +80 Siii 20/30 0.67 12.25
kms−1 ≤ v ≤ +200 kms−1 at the red end of the Echelle Siiii 21/30 0.70 12.15
LSR
grating.Tocomparethecoveringfractionsoftheindividualions Siiv 6/30 0.20 12.90
witheachother,andtorelatethemtoHisky-coveringfractions Mgii 10/19 0.53 12.70
Feii 10/21 0.48 12.90
determined from 21cm all-sky surveys, one needs to consider
the relative abundancesof the elements(C, O, Si, Mg, and Fe)
aNumber of HVC detections above column-density threshold/total
in HVCs. The ionizationconditionsand dust-depletionproper-
numberofsightlines;
tiesoftheabsorbinggascanalsoaffecttheinterpretationofthe
bcoveringfraction;
coveringfractions.Thesefactorswillbe consideredin thesub- cminimumcolumndensitythresholdconsidered.
sequentsections.
3.2.Siiiabsorption for Siii λ1260.4; Morton (2003)). Our simultaneous fitting of
these lines therefore provides particularly reliable values for
Inourstatisticalanalysis,wefocusonSiiiabsorptioninHVCs. N(Siii) and b(Siii) in both strong and weak HVC absorption
TheSTISE140MdatacontainfiveSiiitransitions(atλ1190.4, components.The ionizationpotentialof Siii (E = 16.4 eV)
SiII
λ1193.3,λ1260.4,λ1304.4,andλ1526.7)thatspanalargerange is very similar to that of Mgii (E = 15.0 eV), suggest-
MgII
inoscillatorstrengths(f =0.133forSiiiλ1526.7and f =1.176 ing that both ions trace the same gas phase in HVCs. In ad-
4
P.Herenz,P.Richter,J.C.Charlton,andJ.R.Masiero:TheMWhaloasQSOabsorber
Figure2.DistributionofLSRvelocitiesofalldetectedSiiiHVC Figure3.DistributionofSiii columndensitiesinHVCs, based
absorptionsystems. onVoigt-profilefittingof38HVCabsorptioncomponents.
dition, the cosmic abundancesof Si and Mg are almost identi- itydistributionsuggestsanetinfallofhigh-velocityneutraland
cal (log(Si/H) = −4.44 and log(Mg/H) = −4.42, assuming weaklyionizedgastowardstheMilkyWaydisc.Highionssuch
⊙ ⊙
solar relative abundancesfrom Asplundetal. (2005)). Because as Ovi, in contrast, show an equal distribution of positive and
Mgii is the most commonly used ion to study circumgalactic negativevelocitiesintheGalactichalo(Sembachetal.(2003)),
gas at low and intermediate redshift in optical quasar spectra supportingtheideathattheytracebothinfallingandoutflowing
(e.g.,Kacprzaketal. (2008);Bouche´etal. (2012)), the absorp- gasthatishighlyionized.
tion propertiesof Siii and Mgii in HVCs can be directly com-
paredtothestatisiticalpropertiesofinterveningMgiiabsorbers
3.2.2. Columndensities
atlowredshift(seeSect.4.4).
ThecoveringfractionforHVCSiiiabsorptioninthehalois Fromourline-fittinganalysiswefindthatthe19HVCsthatare
fc(Siii) = 0.67 for log N(Siii) ≥ 12.25 (see previous section; detected and accurately measured in Siii absorption are com-
Table 3). For comparison,the filling factor of Hi in HVCs de- posed of 38 individualabsorptioncomponents.For these com-
rivedfrom21cmsurveysis fc(Hi) ≈ 0.15forlogN(Hi) ≥ 18.3 ponents we obtain typical Siii column densities in the range
and fc(Hi) ≈ 0.30 for log N(Hi) ≥ 17.8 (Wakker (2004). The log N(Siii) = 12.5− 15.0 (see Tables B.1−B.5). A histogram
higherdetectionrateofSiiiabsorptioncomparedtoHiemission showing the distribution of the Siii column densities is pre-
suggests that more than half (0.37/0.67 = 0.55) of the high- sented in Fig.3. The medianlogarithmiccolumndensity is log
velocitySiiiabsorberstraceneutralandionizedgasinthehalo N(Siii)=13.48.
belowthetypicaldetectionlimitofcurrent21cmobservationsat Fig.3indicatesawidespread,inhomogeneousdistributionof
logN(Hi)<17.8. theSiiicolumndensitiesinHVCabsorptioncomponentswitha
HVCs that are detected in metal absorption without hav- prominentpeaknearlog N(Siii) ∼ 12.8andanothermaximum
inganHi21cmcounterpartcommonlyarereferredtoas“low- nearlogN(Siii)∼13.8.Most(21/31or68percent)oftheHVC
columndensityHVCs”(LCDHVCs;Richteretal.(2009)). absorptioncomponentshavelogN(Siii)≥13.2.
Tables B.1−B.5 show that only some of the Siii absorp-
tion components with log N(Siii) < 13.0 represent satellite
3.2.1. Radialvelocities
components of stronger HVC absorbers. There exist a distinct
In Fig.2 we show the distribution of LSR velocities of the 19 population of isolated, weak HVC absorbers with relatively
HVCs for which Siii absorption was detected and accurately low column densities of Siii and other low ions (e.g., towards
measured(onlyhigh-velocitySiiiabsorptiontowardsNGC3516 PG1211+143,PKS2155−304,NGC4151).Theseabsorbersbe-
isnotconsideredherebecauseofthelowdataquality).Absolute long to the class of highly-ionized high-velocity clouds (e.g.,
valuesforv rangebetween|v | = 90and370kms−1.The Sembachetal. (1999), Sembachetal. (2003)) and to the low-
LSR LSR
highestvelocityabsorberisfoundtowardsNGC7469andisre- column-densityHVCs(Richteretal.(2009)).Thissuggeststhat
latedtotheMagellanicStream(Table2).Notethattheregionbe- theinhomogeneousdistributioninFig.3reflectstheactualphys-
tweenv = −90to+90kms−1 (i.e.,theIVCvelocityregime) ical properties of neutral and weakly-ionized gas structures in
LSR
isnotconsideredinthisstudy. theGalactichalo,andisnotanartifactfromouranalysis.
Outofthese19sightlines,12(63percent)showabsorption BasedontheSiiicolumndensitiesshowninFig.3wehave
atnegativevelocities.OnemayarguethatlargeHVCcomplexes constructed a column-density distribution function (CDDF) of
at negative velocities (such as Complex C) together with the HVC Siii absorption components (Fig.4). The CDDF can be
limitedsamplesize leadsto anobservationalbiastowardsneg- definedas f(N) = m/∆N, wherem is the numberof absorbers
ative velocities. However, optical observations of Caii absorp- in the column-density bin ∆N. The CDDF is usually approx-
tion in HVCs, based on a ten-times larger data sample of ran- imated by a power law in the form f(N) = CN−β, where
domlydistributedQSOsightlines,alsoindicatethatthemajority β ≈ 1.5 for Hi in HVCs, as derived from 21cm observations
of the neutral HVC absorbersexhibit negativeradial velocities (e.g.,Lockmanetal.(2002)).AscanbeseeninFig.4,theCDDF
(BenBekhtietal.(2008),BenBekhtietal.(2012)).Thisveloc- of Siii HVC absorption components deviates from a simple
5
P.Herenz,P.Richter,J.C.Charlton,andJ.R.Masiero:TheMWhaloasQSOabsorber
66
55
44
mbermber 33
NuNu
22
11
00
00 55 1100 1155 2200 2255 3300 3355
bb [[kkmm ss--11]]
Figure4. Column-density distribution function of Siii absorp- Figure5. Distribution of Siii Doppler parameters in HVCs,
tioncomponentsinHVCs.Thedataindicatethatasimplepower basedonVoigt-profilefittingof40HVCabsorptioncomponents.
law (red and blue solid lines) representsa rather poorapproxi- Theblacksolidlineindicatesafittothedistributionwithalog-
mationtothedistributionofSiiicolumndensities. normalfunction(seeSect.3.2.3).Thebluesolidlinemarksthe
medianvalueat9.2kms−1
power law, showing instead a plateau at log N(Siii) ∼ 13.8.
TheshapeoftheCDDFthusreflectstheinhomogeneousdistri- kms−1.Thenon-thermalcomponentmayincludeturbulentmo-
butionofSiiicolumndensitiesinHVCabsorptioncomponents tionsinthegasandunresolvedvelocitystructureinthelines.
showninFig.3.Ifwe forcea power-lawfitwith a singleslope Since Si is a relatively heavy element (A = 28), it is
Si
to the Siii CDDF for column densities log N(Siii) ≥ 12.5, we expected that for neutral and partly ionized HVCs gas with
obtain β = 1.29 ± 0.09 and logC = 4.57 ± 1.20 (Fig.4, red T ≤ 2×104 K(e.g.,BenBekhtietal.(2012);theirFig.15)the
solidline).Thisslopeissomewhatshallowerthanthecanonical thermalcontributiontob(Siii)is≤3.5kms−1.Thisimpliesthat
valueofβ≈1.5derivedfromHi21cmobservationsofGalactic the observed line widths of most of the Siii HVC absorption
HVCs(e.g.,Lockmanetal.(2002)).Ifweinsteadrestrictourfit featuresare subjectto broadeningmechanismsotherthan ther-
to the range log N(Siii) ≥ 13.7, we obtain much steeper slope malbroadening,suchasmacroscopicturbulenceandgasflows.
of β = 1.59±0.23 and logC = 8.94±3.31 (Fig.4, blue solid Moreover,itisverylikelythatmanyoftheSiiiabsorptioncom-
line).ThisslopefitsbettertotheslopederivedforHi,butissub- ponents seen in the STIS data are composed of smaller (unre-
stantiallysmallerthantheslopederivedforopticalCaiiabsorp- solved)substructures.In fact, opticalabsorption-linestudiesof
tioninIVCsandHVCs(β=2.2±0.3;BenBekhtietal.(2008), IVCs and HVCs at very high spectral resolution and high S/N
BenBekhtietal.(2012)).ThesteeperslopeofCaiicomparedto clearlyindicatethatthereexistssubstantialvelocity-structurein
SiiimostlikelyisaresultofthestrongdepletionofCaintodust neutralhalocloudsatalevelofafewkms−1 (see,e.g.,theIVC
grains,becausehigh-columndensitycloudstendtohavehigher andHVCinthedirectionoftheMagellanicClouds;Weltyetal.
depletion valuesthan low-columndensity clouds(see also dis- (1999)).
cussion in BenBekhtietal. (2012)). Note that if some of the
HVCabsorptioncomponentswouldbecomposedofseveral,un-
3.2.4. Sub-componentstructure
resolvedvelocitycomponents,theslopeoftheCDDFwouldbe
steeper,too.
While the smallest substructures in the HVC absorbers obvi-
ously are not resolved in the STIS data, most of the detected
Siiiabsorptionfeaturesdoshowseveralindividualvelocitysub-
3.2.3. Dopplerparameters
componentsthatareseparatedfromeachotherby> 10kms−1,
InFig.5weshowthedistributionofSiiiDopplerparameters(b- typically,andthatwehavefittedasindividualabsorptioncompo-
values)inHVCs,basedontheVoigt-profilefittingofthe38Siii nents.Since forinterveningQSO absorbersthevelocityspread
absorptioncomponents.Themeasuredb-valuesrangefrom1to of the detected absorption feature often is used as an observa-
33 kms−1 with a median value of ∼ 9.2 kms−1. The distribu- tional parameter to constrain the characteristic environmentof
tioncanbefittedbyalog-normalfunction(solidlineinFig.5), the absorber host (e.g., Charlton&Churchill (1998)), it is in-
which peaks at b = 7 kms−1. The median b value is b = 9 teresting to study the velocity structure of Galactic HVCs and
kms−1. Note that Siii b-valuesthatare smaller than the instru- compareit to the absorptionpropertiesof interveningsystems,
mentalresolutionintheE140Mgrating(∼7kms−1)canbereli- inparticularweakandstrongMgiiabsorbers.
ablydeterminedsincewearefittingsimultaneouslyseveralSiii In Fig.6 we show the number distribution of Siii absorp-
lines with different oscillator strengths (i.e., the corresponding tion componentsper HVC for all spectra in which HVC gasis
curve-of-growthiswell-defined). detectedin Siii. About80 percentof themeasuredHVCs have
It is commonly assumed that the Doppler parameter of an one or two velocity components that can be resolved with the
absorberis composedof a thermalcomponent(b ) and a non- STISdata(onecomponent:32percent;twocomponents:47per-
th
thermal component(b ), so that b2 = b2 +b2 . The thermal cent). For comparison,BenBekhtietal. (2012) find for optical
nth th nth
component depends on the temperature of the gas, T, and the CaiiabsorptioninIVCsandHVCsthatmorethan70percentof
atomicweight(A)oftheabsorbingion:b ≈0.129(T[K]/A)1/2 theCaiiabsorbersareseenassingle-componentsystems,while
th
6
P.Herenz,P.Richter,J.C.Charlton,andJ.R.Masiero:TheMWhaloasQSOabsorber
ThereareseveralstudiesonHVCsthathavecombinedSTIS
E140M data with FUV data from FUSE (e.g., Richteretal.
(2001); Sembachetal. (2004b)) to make use of several other
(weaker) Oi transitions at λ < 1040 Å. However, only for a
few lines of sight in our sample there are FUSE data of suffi-
cientqualitytodetermineN(Oi)atanaccuracysimilartothatof
N(Siii);inthisstudy,wethereforedonotconsideranyavailable
FUSEdata.
ThecoveringfractionofHVCOiabsorption(f (Oi)=0.48)
c
is smaller than that of Siii, but the column-densitylimit above
which f (Oi) is considered is 1.4 dex higher than that of Siii
c
(Table 3). The relative solar abundanceof O comparedto Si is
log(O/Si)⊙ =+1.15(Asplundetal.(2005)),sothatforanHVC
withsolarrelativeabundancesofOandSi(andwith100percent
of these elements in the gas phase) our STIS data are slightly
(0.25dex) more sensitive for Siii absorptionin HVCs than for
Figure6.DistributionofSiiivelocity-componentsinHVCs. absorptionby Oi. On the one hand, the Siii columndensity in
HVCs may be reduced by the depletion of Si into dust grains
(e.g., Richteretal. (2001); Richter&deBoer (2004)); on the
other hand, Siii traces neutral and weakly-ionized gas, so that
thefractionoftwo-componentabsorbersislessthan20percent. Siii/Oimaybehigherthan(O/Si)⊙.Consequently,theseeffects
This difference is not surprising, however, since Caii in IVCs (dustdepletionand ionization)are oppositeandmayevencan-
and HVCs is expected to trace relatively confined neutral gas celeachotherout.Toestimatewhetherornotthisisthecasein
regionsinthehaloclouds,whereasSiii tracesbothneutraland an HVC one wouldneed to knowthe localdustpropertiesand
weakly ionized gas regions (i.e., multi-phase gas regions) that ionizationconditionsinthegas,whicharedifficulttodetermine.
arespatiallymoreextended.
The five sightlines, along which high-velocity Siii is de-
On a first look, the distribution of absorption components tected without associated Oi absorption (see Table 2), exhibit
in HVCs appears to be very similar to the distribution found relativelyweakHVCabsorptioninSiii,suggestingthatforsome
for strong intervening Mgii absorbers (Prochteretal. (2006); ofthesecloudsOiλ1302.2maybejustbelowthedetectionlimit
onecomponent:∼ 50percent;twocomponents:∼ 20percent).
(seealsoRichteretal.(2009)).
However,becauseinterveningMgiiabsorberstracegasindiscs
and halos of galaxies (and thus often are fully saturated over
a large velocity range) and because the spectral resolution of 3.3.2. Cii
the Mgii SDSS data used by Prochteretal. (2006) is very low
(R≈2000),thissimilaritydoesnotprovideanycluestothecon- With an ionization potential of E = 24.4 eV singly-ionized
CII
nection between strong Mgii systems and HVCs. Using high- carbontracesneutralandmildlyionizedgasinHVCs.Theonly
resolution optical spectra, Churchilletal. (2003) indeed find a availableCiitransitionintheSTISE140Mwavelengthrangeis
muchlargernumberof∼ 8absorptioncomponentsperabsorp- locatedat1334.5Å.ToderiveN(Cii)fromthe(mostlyfullysat-
tionsystemsforstrongMgiiabsorberswithHicolumndensities urated)λ1334.5lineoneneedstoassumethatb(Cii)issimilarto
belowthatexpectedforneutralgasdiscs,butsimilartothosein theDopplerparameterderivedforSiii(orotherloworinterme-
GalacticHVCs. diate ions).However,in view of the higherionizationpotential
Even if one considers the longer absorption path length of Cii compared to Siii, Cii absorption may arise in a some-
throughagalaxyhalofromanexteriorvantagepoint(Churchill what different (possibly more extended)gas phase, so that this
etal.study)comparedtothepathlengththroughtheMilkyWay assumption may be invalidfor most of the HVC absorbers.As
halo from the position of the Sun (our study), the four-times a consequence, the Cii column densities derived for our HVC
highernumberofabsorptioncomponentspersystemclearlyin- samplearepossiblyafflictedwithlargesystematicuncertainties
dicates that the majority of the strong Mgii absorberswith log thatwecannotaccountfor.
N(Hi) ≤ 20.2 studied by Churchilletal. (2003) trace gaseous From our data we derive a covering fraction of f (Cii) =
c
structuresinhalosthearekinematicallymorecomplexthanthe 0.70;Table 3),which is insignificantlyhigherthan thatof Siii.
Galactic HVC population.As we will see later,this scenariois
The column density threshold is log N = 13.20, which is
min
supported by the very large absorption cross section of strong ∼ 1 dex higher than that of Siii (see Table 3). This differ-
Mgii absorbersthatare bothmorecommonandspatially more
ence compensates the expected abundance difference between
extended(Sect.4).
thesetwoelements,ifsolarrelativeabundancesareassumed(log
(C/Si)⊙ = +0.88).Therefore,theCiiandSiiitransitionsinthe
STIS data provide roughly the same sensitivity to neutral and
3.3.Remarksonotherions
weaklyionizedgasinHVCs.
3.3.1. Oi
OiisanexcellenttracerofHi,becausebothatomshavethesame 3.3.3. Mgii
ionization potential and they are coupled by a strong charge-
exchange reaction. There is only one (strong) transition of Oi TheMgiidoubletnear2800Å isobservedonlywith theSTIS
available in the STIS E140M wavelength range (at 1302.2 Å). E230Mgrating,sothatthereareonly19sightlinesalongwhich
Thus,N(Oi)canbedeterminedfromtheSTISdataalone,only high-velocityMgiicanbestudiedinourdatasampleatinterme-
undertheassumptionthatb(Oi)=b(Siii). diatespectralresolution(FWHM∼10kms−1).
7
P.Herenz,P.Richter,J.C.Charlton,andJ.R.Masiero:TheMWhaloasQSOabsorber
As mentionedabove,Mgii and Siii are expected to trace a
similar gas phase in HVCs and both ions are expected to have
verysimilar columndensities.TheMgiifilling factorin HVCs
is f (Mgii)=0.53(Table3),whichissomewhatlowerthanthat
c
of Siii. This is notsurprising,however,since the Mgii column
density threshold (log N = 12.70) is ∼ 0.5 dex higher than
min
thatofSiii.InSect.4wewillcombinetheE230MdataforMgii
andtheE140MdataforSiiitocomparetheabsorptionstatistics
ofHVCswiththatofinterveningMgiiabsorbers.
3.3.4. Feii
TheionizationpotentialofFeii(E =16.2eV)isverysimilar
FeII
tothatofSiiiandMgii,sothesethreeionsareexpectedtoarise
in the same gas phase (neutraland weakly ionized gas). In the
E230MwavelengthbandthereareseveralFeiitransitionsavail-
ableincludingthetworelativelystrongtransitionsat2382.8and
Figure7. Example for a CLOUDY photoionization model of
2600.2Å.AlsotheE140Mbandcontainsanumberof(weaker)
GalacticHVCgasatadistanceofD=50kpcfromtheGalactic
Feiitransitions. disc,withanHicolumndensityoflogN(Hi)=19andametal-
ThecoveringfractionofHVCFeiiabsorptionis f (Feii) =
c licity of 1.0 solar. Shown are the expected column densities of
0.48(forlogNmin =12.90),thusverysimilartothatofMgii(see Siii andMgii asa functionof thegasdensity,n , andthe ion-
H
above).This is expected,since the threshold columndensity is
izationparameter,U.Becauseofthesimilarionizationpotentials
0.2dexhigherthanforMgii,whilethesolarabundanceofFeis
andcosmicabundancesofbothelements/ionsthecolumndensi-
0.08dexlower(log(Fe/H)⊙ = −4.55;Asplundetal.(2005)).In tiesofSiiiandMgiiinHVCsareexpectedtobeverysimilar.
summary,bothionsareequallysensitivetotracepredominantly
neutral and weakly-ionized gas in HVCs. There are two sight-
lines that show Feii absorption at high velocities, while these absorption sometimes is associated with common 21cm HVCs
twosightlinesarenotcoveredinMgii(Table3). (e.g., Foxetal. (2009)), or with low-column density HVCs
(Richteretal. (2009)), where it is thought to arise in the inter-
3.3.5. Siiii faceregionsbetweenneutralHVCgasandthehotcoronalgas.In
addition,thereexistsapopulationofhighly-ionizedHVCs(e.g.,
Siiiihasoneverystrongtransitionat1206.500Å;withanion- Sembachetal.(1999);Sembachetal.(2003))thatprobablyrep-
izationpotentialof E = 33.5eV thision tracesdiffuse ion- resentlow-density,gasstructuresinthehaloandthatmostlikely
SiIII
ized gas in HVCs and their (more or less) extended gaseous arephotoionized.Thesestructuresmayarise indiffusegaseous
envelopes. As for Cii and Oi the determination of a reliable material that originates in the IGM and that is being accreted
column density for Siiii is basically impossible, as the Siiii by the Milky Way (“warm accretion”),or that results from the
λ1206.500lineisoftenheavilysaturatedandtheassumptionthat break-upofmoremassiveHVCsastheyinteractwiththecoro-
b(Siiii)=b(Siii)maybeinvalidformostcases. nalgasinthehalo(oneprominentexampleistheHVCComplex
We find f (Siiii) = 0.70,whichisidenticalto thevaluede- GCN,whichisdetectedinCivandSiivtowardsPKS2155−304
c
rivedforCii(i.e.,allHVCsthatarereliablydetectedinSiiiiin andMrk509;seeWinkeletal.(2011)).
oursamplearealsodetectedinCii).Thecolumndensitythresh- As covering fractions we derive fc(Civ) = 0.40 for log
oldconsideredforthisestimateislogNmin(Siiii)=12.15(com- Nmin(Civ) = 13.00 and fc(Siiv) = 0.20 for log Nmin(Siiv) =
pared to log Nmin(Cii) = 13.20; see above). Since (C/Si)⊙ = 12.90.Thesecoveringfractionsaresmallerthantheonederived
+0.88(Asplundetal. (2005)) it followsthat Siiii andCii trace for Ovi in the Milky Way halo (fc(Ovi)≥ 0.59;Sembachetal.
the same physical regions in HVCs at roughlythe same sensi- (2003)), implying that Ovi is more sensitive for detecting
tivity.Thecoveringfractionof f (Siiii)=0.70islowerthanthat highly-ionized halo gas and/or the Ovi absorbing gas phase is
c
derivedby Collins et al.(2009; f = 0.84 for log N (Siiii) = spatiallymoreextendedthantheCivandSiivabsorbingphase.
c min
12.50)basedonthesameSTISdata.Thereasonforthisdiscrep-
ancyisthattheseauthorsalsoincludeveryweak(andalsospu-
4. HVCsasinterveningmetal-lineabsorbers
rious)absorptionfeaturesin their statistics thatwe donotcon-
sider as secure HVC Siiii detections. In addition, for our anal- Deep Hi 21cm observations of M31 and other nearby spi-
ysiswetake intoaccountonlythoseHVCcomponentsthatare
ral galaxies (e.g., NGC891) clearly show that the HVC phe-
(in velocity space) well separated from lower-velocitymaterial
nomenon is not restricted to the Milky Way, but represents an
(i.e., IVCs). Our value of f (Siiii)= 0.70 is, however,in excel-
c ubiquitous component of spiral galaxies in the local Universe.
lent agreement with the covering fraction of f = 0.68±0.04
c It indicates the various gas-circulation processes in the inner
of UV-selected HVCs based on a much larger STIS/COS data
and outer halos of star-forming galaxies (Thilkeretal. (2004);
samplerecentlypresentedbyLehneretal.(2012).
Oosterlooetal.(2007);Fraternalietal.(2007);Richter(2012)).
Our results on the coveringfractionof the differentions in the
3.3.6. CivandSiiv MilkyWayHVCs,togetherwithstatisticsofinterveningmetal-
lineabsorbersinQSOspectra,nowcanbeusedtoinvestigatethe
The high ions Civ and Siiv are known to trace a gas phase absorption-crosssectionofHVCanaloguesinthelocalUniverse
in HVCs that is different from that traced by low ions such and to provide an estimate of the contribution of HVCs to the
as Oi, Cii, Siii, Mgii, and Feii. High-velocity Civ and Siiv numberdensityofinterveningmetalabsorbersatlowredshift.
8
P.Herenz,P.Richter,J.C.Charlton,andJ.R.Masiero:TheMWhaloasQSOabsorber
Inthefollowing,wewillfirstbrieflydiscussthegeneralrela- thathaveW >0.3Å wouldbeclassifiedasstrongMgiisys-
2796
tionbetweenthenumberdensityofinterveningmetalabsorbers tems if seen as QSO absorbers from far away. However, since
andtheabsorption-crosssectionofgalaxiesandtheirhalos.As all HVCs in the Milky Way halo have log N(Hi) ≤ 20.2 one
asecondstep,we willthencombinetheobservedskycovering would expect that HVC analogues in the halos of Milky Way-
fractionsofSiiiandMgiiintheMilkyWayHVCswithstatistics type galaxies contribute to the population of strong Mgii ab-
on interveningMgii absorbersto study the the spatial distribu- sorberspredominantlyintherange0.3Å≤W ≤1.0Å while
2796
tionofneutralandweaklyionizedgasinthehalosofMilkyWay- thestrongestoftheinterveningMgiiabsorberswithW >1.0
2796
typegalaxies(seealsoRichteretal.(2011);Richter(2012)). Å are related to discs, disc-halo interfaces, and galactic winds
(e.g., Bouche´etal. (2012)). In the usual QSO absorber classi-
fication scheme, HVCs would appear as sub-dampedLyman α
4.1.Absorption-crosssectionofgalaxiesandtheirhalos
systems(sub-DLAs)andLyman-limitsystems(LLS;seeRichter
LetusdefinedN/dz(X)astheabsorbernumberdensityofQSO (2012)).
metal-line aborbers per unit redshift measured for a given ion In Table 4, sixth row, we show the measured Mgii equiva-
X, ng as the space density of galaxies at z = 0, Rh(X) as the lent widths for the eleven (out of 19 possible) HVCs in which
(mean)galaxyhaloradius(withinwhichmetalabsorptiontakes Mgiiabsorptionhasbeendetected.StrongMgiiabsorptionwith
place), and f (X) ≤ 1 as the mean coveringfraction of the ion
c W > 0.3 Å is measured for six HVCs, suggesting that
2796
X inthedisc/halogasforradiir ≤ Rh(X).Numberdensityand the covering fraction of HVCs with strong Mgii absorption is
geometric cross section of the gas in and around galaxies are f = 6/19 ≈ 0.32. Because of the limited statistical rele-
c,sMg
directly related to each other (e.g., Kacprzaketal. (2008)), so vance of this result, we also considerSiii as a proxyfor Mgii.
thatforz=0 Asmentionedearlier,SiiiandMgiihaveverysimilarionization
potentialsandsolarabundances,so thatitisexpectedthatboth
dN = nghfcicπR2h. (1) ionstracethesamegasphaseinHVCsandtheyhaveverysim-
dz H0 ilar columndensites.To demonstratethe expectedsimilarityof
theSiiiandMgii columndensitiesinHVCs we showin Fig.7
The space density of galaxies can also be expressed as
n =Φ⋆Γ(x,y),whereΦ⋆isthenumberdensityofL⋆ galaxies. anexampleofaCLOUDYphotoionizationmodelofaGalactic
g HVC with an Hi column density of 1019 cm−2, a distance to
Γ(x,y) is the incompleteGamma functionin which x = α+1,
theGalacticdiscof50kpc,andwithsolarabundancesofSiand
whereαistheslopeatthefaintendoftheSchechtergalaxylu-
minosity function.The parameter y is defined as y = L /L⋆, Mg(modelfromRichteretal.(2009)).Inthisfigure,N(Siii)and
min N(Mgii)areplottedagainsttheionizationparameter(U)andthe
where L is the minimum luminosity of galaxies contribut-
min
gasdensity(n ).Overa largerangeofdensitiesandionization
ingtothepopulationofabsorbinggashalos.Therefore,ifn is H
g parameters the expected column densities of Siii and Mgii lie
knownforagivenluminosityrangeanddN/dz(X)and f (X)≤
c
within0.2dexofeachother,demonstratingthatthese twoions
1aremeasuredforagivenion,equation(1)allowsustoestimate
indeedtracethesamegasphaseinHVCswithalmostidentical
the characteristic size (R (X)) of the absorbing region around
h columndensities.
galaxies.Foramoredetaileddiscussionoftheseparameterssee
Richter(2012). TotransformthemeasuredSiiiequivalentwidthsalongthe
19suitedE140MsightlinesintoMgiiequivalentwidths,wecon-
siderSiiiλ1260,whichisthestrongestSiiitransitioncoveredby
4.2.CoveringfractionofstrongMgiiabsorptionintheMilky ourSTISdata (Table 4,fifth row).From atomicdata itfollows
Wayhalo that(fλ) ≈(fλ) (Morton(2003)).
SiII1260 MgII2796
For the study of the physical properties and absorption cross For a typicalHVC Dopplerparameterrange of b = 5−12
section of gas in and around galaxies at low and high red- kms−1 with one or two absorption components and under the
shift via QSO absorption-linespectroscopy the Mgii ion plays abovediscussedassumptionthat N(Siii) = N(Mgii)anequiva-
a crucialrole. The so-called strong Mgii systems are interven- lentwidthof300mÅ intheMgiiλ2796linecorrespondstoan
ing metal-absorptionsystems thathave Mgii equivalentwidths equivalentwidth of 140−200 mÅ in the Siii λ1260 line (see
W >0.3Å;theyusuallyareassociatedwithluminousgalax- also Narayananetal. (2008)). We here use the lower threshold
2796
ies (L > 0.05L⋆) at impact parameters < 35h−1 kpc (e.g., ofW =140mÅ toseparatestrongandweakMgiiabsorbers
1260
Bergeron&Boisse´ (1991);Steideletal.(2002);Kacprzaketal. in HVCs in an indirect manner. The observed HVC Siii/Mgii
(2010)).These absorbersare expectedto trace neutraland ion- absorptionstrengthstowardsPG1116+215andTonS210arein
izedgasinthediscsofgalaxiesandtheirgaeoushalos(includ- excellentagreementwiththisconversionscheme(seeTable4).
ingHVCs).Theso-calledweakMgiisystemshaveW2796 ≤ 0.3 Based on this method, we find that 12 of the 20 Siii HVC
Å;theyappeartobelesstightlyassociatedwithgalaxiesandare absorbers listed in Table 4 represent strong Mgii systems.
typicallyfoundatlargerdistancesfromluminousgalaxies,inthe Combining this result with our direct Mgii measurements out-
range 35− 100h−1 kpc (Milutinovic´etal. (2006); Rigbyetal. lined above we have 6 relevant Mgii plus 8 relevant Siii de-
(2002)). tections along 41 independent sightlines, so that the total cov-
FromtheLCDHVCsurveybyRichteretal.(2009)andfrom ering fraction of strong Mgii absorption in Galactic HVCs is
thisstudyitisexpectedthatonlythemostmassiveHVCs(HVCs estimated as f = 14/41 = 0.34 ± 0.09. This covering
c,sMg
with neutral column densities log N(Hi) ≥ 17.2) display the fractionfor strong Mgii absorptionis identicalto the observed
absorptioncharcterisiticsofstrongMgii absorbers,whilethere covering fraction of Hi in HVCs with log N(Hi) ≥ 17.8,
exists a population of HVCs with Hi column densities log based on 21cm HVC surveys (see Wakker (2004)), and refer-
N(Hi)<17.2thatwouldappearasweakMgiiabsorbersifseen encestherein).Itisalsoverysimilartothecoveringfractionof
asQSOabsorption-linesystem.Thehigh-velocityabsorbersto- Caii absorption in HVCs with log N(Caii) ≥ 11.2, as derived
wardsTonS210andNGC4151representexamplesforthisclass from a large sample of optical QSO spectra (BenBekhtietal.
oflow-columndensityHVCs.NotethatbydefinitionallHVCs (2008), BenBekhtietal. (2012)). Caii traces predominantly
9
P.Herenz,P.Richter,J.C.Charlton,andJ.R.Masiero:TheMWhaloasQSOabsorber
neutralgasinHVCswithHicolumndensitieslogN(Hi)≥17.4 galaxiesfromanexteriorvantagepoint.Thecodealsocalculates
(Richteretal. (2011); BenBekhtietal. (2012)). Therefore, the the sky covering fraction of each gas phase if a vantage point
similar absorption cross sections imply that Caii and strong inside the sphere is chosen; it thus allows us to link the distri-
Mgiitracethesametypeofhaloclouds,namelymassiveHVCs bution of Milky Way HVCs with the frequency of intervening
thatareopticallythickinHi;onlythesecloudswouldbeseenas QSOabsorbers.Formoredetailsonthecodeanditsapplication
strongMgiiabsorbersiftheMilkyWayhalowouldbeobserved togaseousgalaxyhalosseeRichter(2012).Oneimportantresult
asaQSOabsorption-linesystemfromanexteriorvantagepoint. from the study by Richter (2012) is that the observedsky cov-
eringfractionofHVCsintheMilkyWayhalo(interiorview)is
fullyconsistentwiththeprojectedcoveringfractionofHiclouds
4.3.OnthecoveringfractionofstrongMgiiinthehalosof
inthehaloofM31(exteriorview).Thissuggeststhatthedistri-
MilkyWay-typegalaxies
butionofneutralgasinthehalosofbothgalaxiesissimilar(in
Being located within the Milky Way disc, we see the distribu- astatisticalsense)andthatbasicallyallHVCsliewithin50kpc
tionandcoveringfractionofHVCsintheGalaxyhalofromthe fromthediscs.
inside-outperspective(interiorview).Forasphericalhalowith Wenowusethecodehalopathtoestimatethetotalabsorp-
radius R the absorption path-length through the halo is always tioncrosssectionandmeancoveringfractionofstrongMgiiin
∼ Randtheobservedsky-coveringfraction, f ,ofHVCgasre- HVCanaloguesaroundMilkyWay/M31typegalaxiesfroman
c
flectsthespatialdistributionofgasintegratedfromtheinnerto external vantage point. As model input we adopt the observed
theouterregionsoftheMilkyWayhalo.Ifagalaxyanditshalo skycoveringfractionof fc,sMgII,halo,i =0.34ofstrongMgiiinthe
is seen from an exterior view point, the absorption path length MilkyWay fromthe interiorvantagepoint.We furtherassume
through the halo (and disc, eventually) depends on the impact that the observedsky coveringfraction of strong Mgii absorp-
parameterofthesightline,whiletheobservedcoveringfraction tionintheMilkyWayHVCs(interiorview)isrepresentativefor
dependsonthepathlengthandtheradialgasdistributioninthe non-star forming disc galaxies of similar mass and further as-
halo.Therefore,ifwewanttoputintorelationthecoveringfrac- sumethattheprojectedcoveringfractionofhaloMgii(exterior
tionofMgii/SiiiintheMilkyWayhalowiththeobservednum- view)declinesexpontially(i.e., fMgII(r)declinesinsamewayas
ber density of intervening Mgii/Siii absorbers at low redshift fHI(r);seeequation2).
(equation1),thedifferentvantagepointsneedtobeconsidered. Under these assumptions, we find (using the halopath
Inaddition,oneneedstoconsidertheabsorptioncrosssectionof code) that the projected covering fraction of strong Mgii from
gaseousdiscs,asinterveningabsorberspassesbothdiscandhalo HVCsinthehalosofMilkyWay-typegalaxiesasseenfroman
componentsofgalaxies.Thisimportantaspectwillbediscussed external vantage point is hfc,sMgIIi ≈ 0.2 for a halo radius of
inSect.4.4. r = 61kpc1.TheprojectedhaloareacoveredbystrongMgii
3
To investigate the radial distribution of gas in the halos of in HVCs aroundMilky Way-type galaxies then turns out to be
galaxies and the resulting covering fractions, Hi 21cm stud- A =2340kpc2.
sMgII,halo
ies of nearby galaxies are of crucial importance. However,
onlyfor a few nearbyspiralgalaxies(e.g.,M31and NGC891;
4.4.OnthecontributionofHVCstotheabsorberdensityof
Thilkeretal. (2004); Oosterlooetal. (2007)) are the 21cm ob-
strongMgiiabsorbers
servations deep enough to provide meaningful constraints on
the distribution of neutral gas in their halos. In a recent study, OurSTISmeasurementsimplythatHVCsandtheirdistantana-
Richter (2012) has demonstrated that the projected covering logues have a non-negligable absorption cross section in low-
fractionof21cmHVCsaroundM31stronglydecreaseswithra- andintermediateionssuchasSiiiandMgii,andthustheseob-
diusandcanbefittedbyanexponentialintheform jectsareexpectedtocontributetotheobservednumberdensity
of strong Mgii absorbers. However, for a quantitative estimate
fHVC =2.1exp(−r/h), (2) ofthecontributionofHVCstodN/dz(Mgii)onealsoneedsto
consider the the absorption cross section of gaseous discs, as
where r is the projected radius in [kpc] and h = 12 kpc is interveningabsorberspasses bothdisc andhalocomponentsof
thescaleheightforHiinHVCs.Anexponentialdeclineofthe galaxies.Whileadetaileddiscussionoftheabsorptioncrosssec-
covering fraction of neutral/ionized gas in the halos of Milky tion of gaseousgalaxy discs is beyondthe scope of this paper,
Way-type galaxies is further supported by high-resolution hy- we providesomesimpleestimatesthathelpto evaluatetherel-
drodynamicalsimulationsofgalaxies(Ferna´ndezetal.(2012)). evanceofHVCsfortheabsorptioncrosssectionofstrongMgii
Richter(2012)developedthenumericalcodehalopaththat absorbers.
canbeusedtocalculatethecoveringfractionofneutralandion- For the Mgii-absorbing disc component in our Milky
ized halo gas from any given vantage point inside and outside Way/M31 model galaxy we assume a radius of r = 30
disc
the halo sphere. The halopath code assumes that the neutral kpc (for log N(Hi) > 17.5), based on the M31 21cm data of
andionizedgasingalaxyhalosisdistributedsphericallyaround Braunetal. (2009), and a covering fraction of f = 1.
c,sMgII,disc
the neutral gas discs of these galaxies. Instead of modelingin- The mean absorption cross section for strong Mgii of a sam-
dividual halo clouds or halo-gas structures (which would re- ple of randomly inclined gas discs with these properties then
quireknowledgeaboutthe size distributionof suchstructures), is A ≈ 1810 kpc2, which is ∼ 77 percent of the cross sec-
disc
the code uses the volume filling factor of a given gas phase tionofthesurroundingHVCpopulation(seeprevioussection).
(e.g.,neutralorionizedgas)asafunctionofgalactocentricdis- Since the areas covered by discs and halo clouds are overlap-
tance, fv(R), as main input parameter. The function fv(R) can ping from an exterior vantage point, projection effects need to
be parametrizedfor anindividualgalaxyorfora populationof be taken into account. Using the halopath code we calculate
galaxies. A correspondingmodel for the gas discs can also be that the absorption cross section of strong Mgii of gas discs
included.Thecodethendeliverstheabsorptioncrosssectionfor
eachgasphaseasafunctionofgalaxyimpactparameterandthe 1 Wedefiner asthehaloradiusbeyondwhichtheprojectedcovering
3
totalareacoveredbygasin the discsandhalosofthemodeled fractionofstrongMgiifallsbelowthe3percentlevel.
10