The host-galaxy response to the afterglow of GRB 100901A

3.7 MB·English

by  Olga E. Hartoog

Mon.Not.R.Astron.Soc.000,1–??(2013) Printed18January2013 (MNLATEXstylefilev2.2)
The host-galaxy response to the afterglow of GRB100901A⋆
Olga E. Hartoog1,2 , Klaas Wiersema3, Paul M. Vreeswijk4, Lex Kaper1,
†
Nial R. Tanvir3, Sandra Savaglio5, Edo Berger6, Ryan Chornock5, Stefano Covino7,
Valerio D’Elia8,9, Hector Flores10, Johan P. U. Fynbo11, Paolo Goldoni12, An-
dreja Gomboc13,14, Andrea Melandri7,15, Alexei Pozanenko16, Joop Schaye2, Anto-
3
nio de Ugarte Postigo17, Ralph A. M. J. Wijers1
1
0
2
(Affiliationscanbefoundafterthereferences)
n
a
J
18January2013
6
1

] ABSTRACT
O
C ForGamma-RayBurst100901A,wehaveobtainedGemini-NorthandVeryLargeTele-
scope opticalafterglowspectra at fourepochs:one hour,one day,three days and one week
.
h afterthe burst, thanksto the afterglowremainingunusuallybrightat late times. Apartfrom
p
a wealth of metal resonance lines, we also detect lines arising from fine-structure levels of
o- the ground state of Feii, and from metastable levels of Feii and Niii at the host redshift
r (z=1.4084).Theselinesarefoundtovarysignificantlyintime.Thecombinationofthedata
st andmodellingresultsshowsthatwedetectthefalloftheNiii4F9/2 metastablelevelpopula-
a tion,whichtodatehasnotbeenobserved.Assumingthatthepopulationoftheexcitedstates
[ isduetotheUV-radiationoftheafterglow,weestimateanabsorberdistanceofafewhundred

1 pc.Thisappearstobeatypicalvaluewhencomparedtosimilarstudies.Wedetecttwointer-
v veningabsorbers(z = 1.3147,1.3179).Despite the wide temporalrangeof the data, we do
2 notseesignificantvariationintheabsorptionlinesofthesetwointerveningsystems.
1
Key words: gamma-ray burst: individual: GRB100901A; gamma-rays: bursts - galaxies:
9
3 abundances-galaxies:ISM-galaxies:distancesandredshifts
.
1
0
3
1 1 INTRODUCTION Savaglio,Glazebrook&LeBorgne 2009; Hjorthetal. 2012). For
v: areview of longGRBsandtheirhost galaxies, seee.g.,Savaglio
Shortly after the first detection of an optical afterglow asso-
i (2006,2012).
X ciated with a long gamma-ray burst (GRB, vanParadijsetal.
1997), it became clear that GRBs may be useful probes of
r Optical spectra (i.e., rest frame UV at redshifts around z
a distant galaxies: afterglowscan be very bright and have asimple ∼
1 2) of GRBafterglowsgenerally show strong absorption lines
power law continuum at ultraviolet (UV) to optical wavelengths, −
of metals, both at low and high-ionisation stages, as well as
against which otherwise undetectable absorbing systems, such
high neutral hydrogen column densities (see Fynboetal. 2009
as the gas in the host galaxy, leave an observable signature
foralargeandrelativelyunbiased sample).Absorptionlinespro-
(see e.g., Vreeswijketal. 2001; Savaglio,Fall&Fiore 2003;
vide the redshift and allow estimation of column densities of the
Vreeswijketal. 2004; Bergeretal. 2006; Fynboetal. 2006;
host galaxy of the GRB and possibly other absorbers along the
Prochaskaetal. 2007). Long GRBs are formed as end products
line of sight. This provides valuable information on the metal-
oftheevolutionofmassivestars(Woosley1993;Paczyn´ski1998)
licity and dust content of galaxies at arbitrary redshift, regard-
and should therefore probe star-forming regions. Their optical
less of the brightness of these systems, which is very difficult
afterglows fade away on short timescales ( days), making it
∼ to obtain otherwise. The highest redshift lines in the afterglow
possibletostudytheirhost galaxiesinemissionaswell(seee.g.,
spectrum formally only provide a lower limit on the host red-
shift. However, the presence of fine-structure lines can confirm
thegalaxyasthehost,becausethesearisefromexcitedstatesthat
⋆ BasedonobservationsobtainedattheGeminiObservatoryatMaunaKea
are expected to have been populated by the UV-afterglow radia-
under programme GN-2010B-Q-7, and observations obtained with ESO
TelescopesattheParanalObservatoryunderprogramme085.A-0009(B). tionoftheGRB(Prochaska,Chen&Bloom2006;Vreeswijketal.
[email protected] 2007). These lines are not seen in foreground line-of-sight ab-
†
c 2013RAS
(cid:13)
2 OlgaE. Hartoogetal.
sorbers in quasar (QSO) spectra, except for some relatively low- otherspectraatrespectively1and7daysaftertheburstwereob-
energy excited states from carbon. Due to the transient nature tained.A fourthspectrumwasobtainedapproximately 3days af-
of the afterglow, the absorption lines from these excited states tertheburstwiththeX-shooterspectrographmountedontheESO
are expected to vary in strength, which may provide informa- VeryLargeTelescope(VLT),underprogramme085.A-0009(B)(PI
tion on the internal distribution of gas and kinematics within the Fynbo).SeeTable1foradetailedlogoftheobservationspresented
host galaxy; often the only way to obtain such information for inthispaper.
these distant and mostly faint galaxies. Fine-structure line vari-
ation has been measured and modelled for a handful of bursts
(e.g., Vreeswijketal. 2007; Dessauges-Zavadskyetal. 2006; 2.1 Gemini-N/GMOSspectroscopy
D’Eliaetal. 2009a; Ledouxetal. 2009; deUgartePostigoetal.
Gemini-N/GMOS is a low-resolution long slit spectrograph,
2011; DeCiaetal. 2012; Vreeswijketal. 2013), which allowed
equippedwiththreedetectors.Fortheobservationspresentedhere,
constraintstobeplacedonthedistancebetweentheburstandthe
a slit width of 0.75 and the B600 grism with the G5307 order
absorbing material (seeVreeswijketal. 2013 foradescription of ′′
suppressionfilterhavebeenused,whichresultedinthewavelength
themethodsused).
coverage and resolving power reported in Table 1. We used four
Theprimarydifficultyinprobingcolumndensityvariationsin
exposures per epoch, using dithers in both dispersion and spatial
GRBsightlinestofittothemodelsdescribedabove,istherequire-
coordinates tosample over the chip gaps and regions affected by
mentonafterglowbrightness:forfainterafterglowsthenecessary
amplifierlocation.TheGMOSspectrahavebeenreducedwiththe
highsignal-to-noiseandhighspectralresolutionaredifficulttoob-
IRAFpackagesforGeminiGMOS(version1.9),usingarclineex-
tain with current instrumentation. This also limitsthis method to
posurestaken directlybeforeandafterthescience data.Thefour
early-timedataandalimitedrangeofinstruments,makingitdiffi-
exposuresarecombinedafterextraction.Theresultingspectrawere
culttobuildupasample.Inthecasesdescribedabove,thevariation
normalisedandnofluxcalibrationwasperformed.
ismeasuredwithhighresolutionspectrographs,andallspectraare
obtainedwithinafewhourspostburst(observer’sframe).
The optical afterglow of GRB100901Aremained unusually
2.2 VLT/X-shooterspectroscopy
bright out to very late times, which allowed us to collect spectra
over a time span of a week with the Gemini-North Multi-Object VLT/X-shooterisacross-dispersede´chellespectrographwherethe
SpectrographandX-shooterontheESOVeryLargeTelescope. In incominglightissplitintothreewavelengtharmsusingdichroics,
thispaperweinvestigatethebehaviouroftheexcitedlevel popu- coveringthefullopticaltonearinfraredwavelengthrangesimulta-
lationsover thisunusually long timespan, and theapplication of neously(D’Odoricoetal.2006;Vernetetal.2011).TheX-shooter
the excitation models to low and intermediate resolution spectra. spectrahavebeentakeninnoddingmodewith1 2binning(i.e.,
×
The observations and data reduction are described in Section 2. binninginthedispersiondirection)intheUV-Blue(UVB)andVi-
Westudy the metal resonance linesat the host-galaxy redshift in sual(VIS)arms,usinga100kHz/high-gainreadoutand4 600s
×
Section 3.1. In Section 3.1.1 we examine the dust content of the exposuretimes.Theselectedslitwidthswere1.0,0.9and0.9 for
′′
host galaxy. The two intervening absorbers are analysed in Sec- theUVB,VISandnear-infrared(NIR)arms,respectively.A5 nod
′′
tion 3.1.3. In Section 3.2 the fine-structure line variability is de- throwalongtheslitwasusedtoimproveskysubtraction.Thespec-
scribedandmodelled.InSection3.3webrieflydiscusstheemis- tra were reduced using the ESO pipeline software version 1.2.2,
sionlinesfromthehostgalaxy.WediscussourresultsinSection4 using the so-called physical model mode (Goldonietal. 2006;
andsummariseourconclusionsinSection5. Modiglianietal. 2010). We used calibration data (bias, dark, arc
ThroughoutthepaperweadoptastandardΛCDMcosmology lamp,flat-fieldandflexurecontrolframes)takenthenextdayoras
withH = 71kms 1Mpc 1,Ω = 0.27,Ω = 0.73.Weindicate closeintimeaspossibletothescienceobservations.Thebinsizes
0 − − m Λ
linetransitionsbytheirvacuumwavelengths. oftheresultingspectrawere0.2,0.2,and0.5ÅforUVB,VISand
NIR spectra, respectively. Extracted spectra were flux calibrated
using spectrophotometric standard star exposures, also taken the
same night, resulting in roughly flux-calibrated spectra spanning
2 OBSERVATIONS
thenear-ultraviolettothenear-infrared.Wecautionthattheweather
GRB100901Awas detected by Swift on 2010 September 1 at conditionsduringthisnightwerepoor,andlikelynon-photometric.
13:34:10UT.WithatotaldurationofT =439 33s(Immleretal. Weusedatelluricstandardstar(HD4670,aB9Vstar)tocorrect
2010; Sakamotoetal. 2010) the burs9t0is clea±rly classified as a thetelluricabsorptionfeaturesintheNIRafterglowspectrum,us-
long burst. The optical afterglow candidate at RA (J2000) = ingtheIDLspextoolpackage(Vacca,Cushing&Rayner2003,see
01h49m03.42s, Dec =+22 4530.8 (90% confidence error radius alsoWiersema2011fortheuseonX-shooterdata).
◦ ′ ′′
ofabout0.81 )wasidentifiedinanUltraviolet/OpticalTelescope
′′
(UVOT) exposure which started 147s after the Burst Alert Tele-
scope(BAT)trigger(Immleretal.2010).Automaticobservations
3 ANALYSIS
withtheFaulkesTelescopeNorthidentifiedanuncataloguedobject
atthepositionconsistentwiththeUVOTcandidate(Guidorzietal. TheGMOSepoch1spectrum(seeFigure1)isthatofabrightafter-
2010).Severalfollow-upphotometryeffortsfollowed,resultingin glowincludingmanymetalabsorptionlines.Ataredshiftof1.4084
thelightcurvepresentedandanalysedinGombocetal.(inprep) we detect strong resonance lines of Feii, Mnii, Crii, Alii, Aliii,
and Gorbovskoyetal. (2012) (see also Figure 8). The first opti- Znii and Siii. Additionally, at the same redshift, line transitions
cal spectrum of the afterglow was taken with the Gemini Multi- arising from excited states of Feii and Niii are clearly detected
ObjectSpectrograph(GMOS)onGemini-North(programmeGN- (seeSection3.2).Thereforeweidentifythisasthehost-galaxyred-
2010B-Q-7,PITanvir),approximately1hr15mins(0.0526days) shiftz (Chornocketal.2010).WedonotdetectlinesfromFeiii,
h
afterbursttrigger(Chornocketal.2010).Withthisinstrument,two as observed in the case of GRB080310 (DeCiaetal. 2012). No
c 2013RAS,MNRAS000,1–??
(cid:13)
Theafterglowspectrum ofGRB100901A 3
Figure1.NormalisedGMOSspectrum(epoch1)oftheafterglowofGRB100901A.Theredlabelsindicatetheabsorptionlinesofthehost(zh=1.4084);the
greenandbluelabels(withverticaloffset)indicatetheabsorptionlinesformedintwointerveningsystemsatz1=1.3147andz2=1.3179.Themagentalabels
showunshiftedNailines(z=0),resultingfromtheabsorptionintheMilkyWay.Thenumbersatthelabelsindicatetheconfigurationofthelowerleveland
thevacuumrestwavelengthofthetransition.ThehostshowsresonancelinesfromtheionsMnii,Feii,Crii,Znii,Siii,AliiandAliii,aswellaslinesfrom
excitedstatesofFeiiandNiii,whicharefoundtovarysignificantly(seeSection3.2).IntheinterveningsystemswedetectMgi,Mgii,Feii,AliiandAliii.
c 2013RAS,MNRAS000,1–??
(cid:13)
4 OlgaE. Hartoogetal.
Table1.Overview ofthespectroscopic observations oftheafterglow ofGRB100901A,arrangedbyinstrumentandepoch.Columns(1)and(2)indicate
telescope,instrument,setup,andslitdimensions.Column(3)givesthewavelengthcoverageforthissetupandtheresolvingpowerR= λ/∆λ,with∆λthe
FWHMofanunresolvedline.Column(4)givesthemidtimeoftheepochsofobservationindaysaftertheburstBATtrigger.Column(5)liststheinterpolated
observedR-magnitudeoftheafterglowatthisepoch(Gombocetal.,inprep).Column(6)liststhetotalexposuretimeoftheobservationinthisepoch.Columns
(7)and(8)givetheseeingandtheairmass.Theseeingismeasuredbyfittingagaussianalongthespatialdirectionoftheslit,aroundthecentralwavelengthof
eachspectrum.Column(9)givesthesignal-to-noiseratio(S/N)inthecontinuumat4150and6100ÅperGMOSpixel(δλ=0.91Å).Column(10)givesthe
namewewillusetorefertotheepoch.
Telescope Instrument SpectralRange TimeafterBurst mR Exp.Time Seeing Airmass S/N/(δλ=0.91Å) Abbrev.
andSetup andResolution (days) (s) ( ) at4150,6100Å
′′
Gemini-North GMOS, 3810 6710Å 0.0526 17.8 4 400 0.62 1.06 38,95 epoch1
− ×
B600+G5307 R=1000 1900 1.0095 18.6 4 500 0.52 1.01 16,70 epoch2
− ×
0.75′′×330′′ 7.0270 22.3 4×1200 0.68 1.05 5,13 epoch3
VeryLarge X-Shooter 3000 25000Å 2.7492 20.2 4 600 1.30 1.48 9,7 xsh
− ×
Telescope UVB1.0 11 R=5400
′′× ′′
VIS0.9 11 R=7400
′′× ′′
NIR0.9 11 R=5800
′′× ′′
Lyman-αisdetected,becauseitistoofarintheblueforthecov- 3.1 Resonancelines
ered wavelength range. Two intervening absorbers are visible at
TomeasuretheobservedequivalentwidthW oftheabsorption
z = 1.3147andz =1.3179(Chornocketal.2010).Inthesesys- λ,obs
1 2 linesinthespectrum,welocallyfitgaussianfunctionstothenor-
temswedetectFeii,Alii,Aliii,MgiandMgii(restframeequiv-
malised spectrum with ngaussfit, a task in the stsdas package in
alentwidthW < 1Å).Wenotethattheselinesareweakeratthe
λ IRAF.Inprinciple,W canbeobtaineddirectlybyintegration,
interveningsystemredshiftsthanatthehostredshift.Thereareno λ,obs
but as can be seen in Figure 1, many lines are blended. In some
fine-structurelinesdetectedatanyoftheinterveningabsorberred-
casesonelineneeds tobefixedinordertomeasurethepotential
shifts.Naiλλ5892,5898aredetectedatz=0andarethusdueto
variationoftheother.ngaussfitcande-blenduptothreelines,with
foregroundabsorptionintheGalaxy.
the possibility to keep part of the parameters fixed. This method
The GMOS epoch2 spectrum (not fully shown in a figure)
allowsthatinspectrawithlowS/N,anabsentorveryweaklineis
looks similar, though the signal-to-noise ratio (S/N) is lower due
sometimesformallybestfittedwithalinewithnegativeequivalent
tothe decreasing brightness of theafterglow. Thelinetransitions
width(i.e.,anemissionline,seealsoFigure6and7);fortheselines
arisingfromthemetastableNiii4F level(seeTableA4)appear
9/2 weobtainupperlimitsforW .
morepronouncedinepoch2thaninepoch1;theFeiifine-structure λ,obs
Theformalerrorontheequivalentwidthisobtainedfromthe
lineshavebecomeslightlyweaker.
errorspectrum:
SinceGMOS epoch3 isobserved afterthebreak inthelight
curveat 2days(Gombocetal.,inprep),whichcausesthebright- line
∼
ness to drop more quickly, the S/N is lower than for the other σWλ,obs =δλvt (σi/Fc)2, (1)
epochs,despitethelongerintegrationtimeandcomparableweather Xi
conditions.Still,strongmetalresonancelinesareclearlydetected. withδλthespectralbinwidth,σ thevalueoftheerrorspectrum
i
Fine-structure lines that were present in epoch1 and lines from inbiniandF thecontinuumflux.Thesummationrunsovertwo
c
metastablelevelsthatwerepresentinepoch2aremuchweakeror timesthefull-width-at-half-maximum(FWHM).Furthermore,we
notdetectedanymore(seeSection3.2). add3%oftheequivalent widthtotheformalerrortoaccount for
Thequality of theX-shooterspectrum (xsh, after2.75 days, the uncertainty in placing the continuum. In the low-resolution
seeTable 1) islow due topoor weather conditions and largeair- GMOS spectra, all absorption lines are unresolved and can be
mass, but because of the extraordinarily large wavelength cover- fit with a gaussian with a fixed FWHM of 3.6Å (the resolution
age of this instrument, this spectrum does provide additional in- element), which is determined from the width of the lines in the
formation.Moreover,thehigherresolutionallowustoexcludethe arc-lamp frames. In xsh we fix FWHM = 1.0Å and 0.7Å in
existence of different velocity components down to 70kms−1 the UVB and VIS-arms, respectively, based on the widths of arc
withrespectto 200kms−1 forGMOS.Thestronga∼ndsaturated lines and telluric emission lines. Some strongly saturated lines
Mgiiλλ2796,2∼803, Mgiλ2853, and Caiiλλ3934,3969 absorp- (forexampletheMgiidoublet)inthesespectracannotbefitwith
tion lines from the host galaxy are outside the range of GMOS gaussians and are therefore integrated numerically. The resulting
due to the redshift, but they are clearly detected in xsh-VIS (not restframeequivalentwidthsW =W /(1+z )oftheresonance
λ λ,obs h
shown in a figure). Naiλ5892,5897 falls in a region with atmo- linesinthefirsttwoepochsarelistedinTable2.
spheric absorption and can therefore not be detected. We do not
detect the red wing of the Ly-α absorption line due to low S/N. Toinfercolumndensitiesofthedifferentionsfromtheabsorp-
Furthermore,inxsh-NIR,wemarginallydetectforbiddenoxygen tionlineequivalentwidths,wemakeuseofamulti-ionsinglecom-
emissionline[Oiii]λ5007atthehost-galaxyredshift(seeFigure9 ponentcurve-of-growth(MISC-COG)analysis(Spitzer1978).The
andSection3.3). Dopplerlinewidthisinprinciplerelatedtothetemperatureandthe
We do not detect prominent interstellar (dust) extinction levelofturbulenceintheabsorbinggas.Limitedbyspectralreso-
featuressuchasthe2175Åfeature(seee.g.,El´ıasdo´ttiretal.2009) lution,itisdifficulttodistinguishbetweenthisandbroadeningdue
ordiffuseinterstellarbandsinanyofthespectraatthehost-galaxy tothecomplexphysicalstructureofabsorbingcloudsinthe sight
redshift. line.Thereforewespeakofan’effective’Dopplerparameter,which
doesnotcarryinformationaboutthetemperatureofthegas,butis
c 2013RAS,MNRAS000,1–??
(cid:13)
Theafterglowspectrum ofGRB100901A 5
merelyasimplificationinthemodel(Jenkins1986).Theassump-
tionweneedtomakeinthisanalysisisthattheeffectiveDoppler
parameterbisthesameforallionspecies,andthatitissingle(one
velocitycomponent). Inotherwords,weassumethatthevelocity
structureisthesameinalllines,andthatitisdominatedbysmall-
scale effect such as thermal velocities and turbulence and not by
differencesinbulkvelocitiesofdifferentgasclouds.
3.1.1 Hostgalaxycolumndensities
We use three unblended Feii resonance lines (λ2249, λ2260 and
λ2586) todetermine bfortheline-of-sightgasinthehost galaxy
(see Table 2). Because these lines cover both the linear and the
flatpartoftheCOG,bothbandthecolumndensityforFeii,N ,
FeII
canbeconstrained.Theyaredependentvariables,thereforefitthem
simultaneously with a grid method, yielding b = 22.1+1.7kms 1
1.6 −
andlogN /cm 2=15.23+0.07. −
FeII − 0.08 Figure 2. Curve-of-growth (COG) for the resonance lines in the host
The COGwiththe bes−t fitvalue for bisshown inFigure2.
galaxyofGRB100901A.Thesolidlinegives thebestfitCOGwithb =
Theionspeciesotherthanironhavelinetransitionseither onthe 22.1kms−1,thedottedlinesshowtheeffectoftheerroronb.Thedashed
linearorontheflatpartoftheCOG;therefore,bhastobeassumed line,whichcoincideswiththelinearpartoftheCOG,isaCOGforb= ,
tobethesameforallionspeciesinordertodeterminethecolumn thatcanbeusedasanapproximationforweaklines. ∞
densities. The resulting column densities of the ions are listed in
Table2.Formostlineswehavetakenanaverageofthefirsttwo
epochs, except Mgii and Caii which can only be measured in
xsh. The equivalent widths of the Alii lines are slightly largerin
epoch1 while those of Aliii are largerin epoch2. While not very
significant,itmaybeasignofionisationduetotheGRBafterglow
(see e.g., Vreeswijketal. 2013). Due to the low significance we
decidednottotakethisintoaccountintheexcitationmodellingin
Section3.2.
Spectralresolutionlimitstheabilitytodistinguishdifferentve-
locitycomponents inabsorption lines:aninstrumental FWHM =
3.6Å(GMOS spectra) corresponds to 200kms−1. In xsh-VIS,
avelocitycomponentdifferenceof70k∼ms−1wouldbedetectable,
butwedonotseeindicationsformorethanonevelocitycomponent
inanyofthelines.ThiscouldbeduetothelowS/Nofxsh,because
mostintermediatetohigh-resolutionspectraofGRBsshowabsorp-
tionlineswithatleasttwocomponents(butseee.g.,Ledouxetal.
2009). Figure 3. [Fe/Zn] as a function of the Znii column density in QSO-
Prochaska (2006) warns that column densities derived with DLAs(reddiamonds),GRB-DLAs(blackcircles)andGRB100901A(blue
the MISC-COG method are systematically underestimated, espe- square). QSO-DLA values are from Luetal. (1996); Prochaska&Wolfe
ciallywhentheunderlyinglinestructureisthatofseparatedclumps (1996, 1997); Pettinietal. (1999, 2000); Prochaska&Wolfe (2000);
ofgasatdifferentvelocityshifts.Inthiscaseonederivesahighb Prochaskaetal.(2001).GRB-DLAvaluesarefromSavaglio,Fall&Fiore
(2003); Fynboetal. (2002); Fioreetal. (2005); Watsonetal. (2006);
withMISC-COG analysis, and therefore a lowercolumn density.
Penpraseetal.(2006);Vreeswijketal.(2007);D’Eliaetal.(2009b,2011).
Resultsforasinglecomponent COGanalysiswithb & 20kms−1
are highly suspect, according to Prochaska (2006). That COG
analysisofalow-resolutionspectrumcanbeaccurateisshownby
D’Eliaetal.(2011),whocomparelowandhigh-resolutionspectra
3.1.2 Dustdepletion
for the afterglow of GRB081008. The COG analysis gives the
sameresultaslinefitstothehigh-resolutionspectrumwithin3σ. Thestrengthoftheabsorptionlinesinthespectrumareameasure
They conclude that this is linked to the low level of saturation oftheabundanceofanelementinthegasphaseonly.Apartfrom
of the lines. The equivalent widths of the low-ionisation lines in theoverallgas-to-dustratio,thefractionofaspecificelementthat
GRB100901Aare found to be in the lowest 10% of the sample islockedontodustdependsontheelementspeciesandonthena-
of69low-resolutionspectraconductedbydeUgartePostigoetal. tureofthedust.Therefore,thepatternofrelativeabundances, the
(2012), which strengthens the case that a COG analysis on this depletionpattern,mightgiveinsightintotheamountandnatureof
spectrumcanprovideaccurateresults.However,columndensities thedust.InTable2welisttheabundanceratioswithrespecttothe
resulting from COG analysis always need to be treated with refractory(depleted)elementFeandthenon-refractoryelementZn,
caution. comparedtotheratioinasolarabundanceenvironment.Weusethe
following,commonlyusednotation
c 2013RAS,MNRAS000,1–??
(cid:13)
6 OlgaE. Hartoogetal.
Table2.Themeasuredrest-frameequivalent width Wλ fortheresonance lines atthehostredshift,andthecorresponding ioncolumndensities, obtained
withcurve-of-growth(COG)analysis.Thefirstcolumnliststheionandthevacuumrestwavelength(Å);thesecondcolumngivestheoscillatorstrength fλ.
Columns(3)and(4)givetherest-frameWλinÅ measuredinepoch1andepoch2(GMOS)respectively.Forlinesforwhichwehaveagoodmeasurementboth
inepoch1andepoch2,theweightedaverageisused;otherwiseweusethevaluemeasuredinepoch1.ThevaluesthatarefinallyusedfortheCOGanalysis
arelistedinColumn(5).Column(6)givesthebestfittingcolumndensityfortheion,fittedtoaCOGwitheffectiveDopplerparameterb=22.1+11..67kms−1.
Columns(7)and(8)givetherelativeabundanceswithrespecttoironandzincusingsolarvaluesfromAsplundetal.(2009)(fornotationseeEqua−tion2).The
referencesforthewavelengthandoscillatorstrengthareincolumn(9).
Line fλ Wλ/Å Wλ/Å <Wλ/Å> log(Nion/cm−2) [X/Fe] [X/Zn] ref
epoch1 epoch2
Alii 1670.7874 1.740 0.418 0.093 0.338 0.098 0.380 0.067 13.57+0.41 0.61 0.42 1.84 0.42 1
± ± ± 0.32 − ± − ±
−
Aliii 1854.7164 0.539 0.113 0.054 0.162 0.060 0.135 0.040 13.02+0.13 a a 1
Aliii 1862.7895 0.268 0.044±0.051 0.113±0.056 0.075±0.038 ...−0.16 1
± ± ±
Siii 1808.013 2.080 10 3 0.292 0.062 0.267 0.068 0.281 0.046 15.96+0.18 +0.72 0.19 0.51 0.19 2
× − ± ± ± 0.16 ± − ±
−
ZZnniiii 22006226..616346b 52..041600×1100−11 00..139612±00..004416 00..220860±00..004583 00..139277±00..003315 13.5.2..−+00..0077 +1.23±0.10 33
× − ± ± ±
Crii 2056.2539 1.030 10 1 0.150 0.040 0.151 0.046 0.151 0.030 13.73+0.09 +0.36 0.12 0.87 0.12 3
Crii 2066.161 5.120×10−2 0.096±0.038 0.111±0.045 0.102±0.029 ...−0.10 ± − ± 3
× − ± ± ±
Feii 2249.8754c 2.190 10 3 0.129 0.034 0.151 0.040 0.139 0.026 15.23+0.07 1.23 0.10 4
Feii 2260.7793 2.620×10−3 0.152±0.034 0.168±0.040 0.159±0.026 ...−0.08 − ± 4
Feii 2586.6495 7.094×10−2 0.705±0.044 0.727±0.048 0.715±0.032 ... 4
× − ± ± ±
Mnii2576.877 3.610 10 1 0.265 0.032 0.257 0.037 0.262 0.024 13.23+0.04 +0.07 0.09 1.16 0.04 5
MMnniiii22650964..446929 12..988000×××1100−−−11 00..126046±±±00..006340 0.207d±±0.034 00..210674±±±00..006243 ......−0.05 ± − ± 55
Mgii2796.352 6.123 10 1 0.920 0.147e 15.06+0.83 0.27 0.83 1.50 0.83 6
Mgii2803.531 3.054×10−1 0.918±0.126e ...−0.60 − ± − ± 6
× − ±
Caii 3934.777 6.500 10 1 0.907 0.220e 13.74+0.47 0.33 0.50 1.56 0.50 7
Caii 3969.591 3.220×10−1 0.836±0.185e ...−0.50 − ± − ± 7
× − ±
aThequantities[Aliii/Fe]and[Aliii/Zn]donothaveanyphysicalmeaning,sincetheAliiiisprobablymainlyfromadifferentregionthanAlii,Feiiand
Znii(seealsoSavaglio&Fall2004).
bThecontaminationbyCriiλ2026isdeterminedfromtheotherCriilinesandisnegligible.ThecontributionofMgiλ2026cannotbedetermined,butwe
assumeitissmallbasedontheabsenceofMgiλ1827.
cFeii2344,2374,3282and2600arealsoclearlydetected,butthoseareblendedwithfine-structurelinesandthereforenotusedintheabundancestudy.
Estimatesoftheirequivalentwidthscanbefoundintheappendix.
dTheequivalentwidthofthelineisfixedinthesecondepochinordertomeasurethevariedstrengthoftheblendedfine-structureline(s).
eAbsorptionlinesareoutsidetheGMOSrange,measurementisfromxshspectrum.
Referencesforatomicdata:(1)NISTAtomicSpectraDatabase,(2)Bergeson&Lawler(1993b),(3)Bergeson&Lawler(1993a)(4)Verneretal.(1999),
(5)Morton(2003),(6)Verneretal.(1996),(7)Morton(1991).
N(X) n(X) of other GRB-DLAs and QSO-DLAs1. The relative abundance
[X/Y] log log , (2)
≡ N(Y)!− n(Y)! valuesindicateastrongerdepletion(i.e.,moredust)inGRB-DLAs
⊙ thaninQSO-DLAs,butamongtheGRB-DLAs,GRB100901Ais
notaparticularlyspecial case.BothGRB-DLAsandQSO-DLAs
appear to follow the overall trend that systems with stronger Zn
column densities show stronger depletion. The fact that the two
with N(X) thecolumn density of element X,and n(X) the num- classespopulatedifferentregionsinthediagramismainlyasight
ber density of element X in a solar environment (Asplundetal. lineeffect.QSOsprobegalaxiesinrandomorientationswithonly
2009). Differences from zero indicate dust content. We thus as- a small chance that the densest region of a galaxy is intersected,
sumethatrelativeabundances(bothingasanddust)aresolar.Fe, whilesightlinestoGRBafterglowsprobethegastowardsthepro-
ZnandCrareironpeakelements,whichmakesthisareasonable genitorregion,sotheywillingeneralshowhighcolumndensities
assumption, although the main production pathway of Zn is not (seealsoSavaglio2006;Prochaskaetal.2007;Fynboetal.2009).
fullyidentified(Umeda&Nomoto2002).Discrepanciesinrelative
abundancesmightalsobeduetoadifferentstar-formationhistory
thantheMilkyWay.
1 DLA:DampedLyman-αsystem:asightlineabsorberwith NHI > 2
InFigure3weshow[Fe/Zn]asafunctionoftheZncolumn 1020cm 2(Wolfe,Gawiser&Prochaska2005).ADLAsysteminaGRB×
−
density for the host of GRB100901A, compared with a sample afterglowspectrumisusuallyduetothehostgalaxy.
c 2013RAS,MNRAS000,1–??
(cid:13)
Theafterglowspectrum ofGRB100901A 7
Furthermore,forGRB-DLAswithrelativelylowcolumndensities
and weak absorption lines, [Fe/Zn] is not well constrained (i.e.,
not reported in literature) because the fading afterglow prevents
buildingupS/Noverseveralnightsofobservations.Thisisnotthe
caseforQSO-DLAs,whichiswhytheseobjectsalsopopulatethe
lowN partoftheparameterspace.
ZnII
A more detailed analysis of the dust content can be per-
formedbycomparingtheheavy-elementdustdepletionpatternof
GRB100901Awithobservationsintheinterstellarmedium(ISM)
oftheMilkyWay(Savage&Sembach1996).Weusethemethod
describedinSavaglio&Fall(2004).Weconsidertheobservedde-
pletionpatternsintheMWsightlinestroughawarmhalo(WH),
warmdiskandhalo(WDH),warmdisk(WD)andcooldisk(CD)
asmodels.Forthefit,weusetwofreeparameters.Oneisthedust-
Figure4.DepletionpatternforthehostofGRB100901A,comparedwith
to-metalratiorelativetotheGalacticvaluesκ =κ /κ (Jisone
GRB J differentcomponentsintheMilkyWay(Savage&Sembach1996),which
ofthe4depletionpatterns).Thesecondisproportionaltothemetal-
areobserved patterns that canbeadjusted bychanging thedust-to-metal
licity(notknownbecauseNHIisnotmeasuredinGRB100901A). ratioκ.ThecolumndensityofNiiiisnotdirectlymeasuredbutfollowsfrom
The two parameters are scaled until they reproduce the observed themodellingofthefine-structureandmetastablelevels,seeSection3.2.1.
heavy-elementdepletionpatternbest(minimumχ2). Weincludea3σupperlimitforTiiibasedontheabsenceofthedoubletat
Resultsusingmeasurementsof7elementsareshowninFig- arestwavelengthof1910Å.
ure4.TheWHdepletionpatterngivesthesmallestχ2 amongthe
4patterns,buttheWDpatternisalsoareasonablefit.Thedust-to-
metalsratioisbetweentheMWvalueand10%higher.NiandMg
arethemostuncertainmeasurements.TheMgiicolumndensityis
likelyalowerlimitbecauseitsuffersfromverystrongsaturation.
TheNiiicolumndensityisnotmeasuredfromtheresonancelines
directly, but follows from the modelling of the time evolution of
theNiii4F metastablelevel(seeSection3.2.1andTable5).Ex-
9/2
cludingthesetwoelementsfromthefitdoesnotchangeourresults
much.
For the observed column of metals, and assuming a rate of
visual extinction per column density of metals like in the MW
(Bohlin,Savage&Drake1978),theexpectedopticalextinctionin
GRB100901AwouldbeA 0.5.Thisishigherthantheonede-
V ∼
rivedfromthespectralenergydistribution(SED)ofGRB100901A
(A =0.21,Gombocetal.,inprep).Thisinconsistencyistypically
V
observedinGRBsightlines(Schadyetal.2011)andindicatesthat
therateofextinctionintheISMofGRBhostspercolumnofmetals
isdifferentfromwhatisobservedintheMW.
3.1.3 Interveningabsorbers
Two intervening absorbers are detected: int1 at z = 1.3147 and
1
int2 at z = 1.3179. If these two systems are not physically as-
2
sociatedwitheachother, andtheredshift differenceisdominated
bycosmologicalexpansion,thevelocitydifferencecorrespondsto
aco-movingseparationof3.9Mpc.Iftheabsorbersbelongtoone
system,thevelocitydifferenceis414kms 1,andcouldbedueto
−
motionofgalaxieswithinacluster.
Because of the long time span that our observations cover
(until long after the jet break, Gomboc et al., in prep), the data
set is suited to look for variations in the strength of the lines
fromtheinterveningabsorbers.Thiscanbeinterestinginthecon-
Figure5.Curve-of-growthforthetwointerveningsystemsinthesightline
text of the previously supposed discrepancy between the number
toGRB100901A.Thelinesusedandthebest-fitcolumndensitiesarelisted
of strong Mgii absorbers (Wλ(2796) > 1Å) in QSO and GRB inTable3.
sight lines; in the latter the redshift number density was found
tobeabout twotimesashigh(Prochteretal.2006; Verganietal.
2009;Cucchiaraetal.2009.However,Cucchiaraetal.(2012)used
alargersampleanddidnotreproducethisdiscrepancy.
We do not detect variability above the 2σ level in any of
the two intervening absorbers of GRB100901A, and we see no
c 2013RAS,MNRAS000,1–??
(cid:13)
8 OlgaE. Hartoogetal.
systematictrendintheensembleoflinesperabsorber,norperion excited stateinan epoch, the equivalent widths of the transitions
species.Inprinciple,thenon-variationoftheequivalentwidthsof arising from thisexcited level are placed on the COG by assign-
theinterveningabsorberresonancelinesgivesalowerlimitonthe inganN tothislevel.Thebest-fit N isthevalueforwhich
level level
sizeoftheabsorbingclouds(i.e.,thescaleonwhichtheabsorbing theensembleof transitionsfittheCOGbest(minimisationofthe
gasishomogeneous).Iftheprojectedapparentsizeoftheafterglow χ2).Figures6and7showtheCOGoftheNiii4F andtheFeii
9/2
becamelargerthantheprojectedsizeoftheinterveningabsorber, 6D levelperepoch(columns),togetherwiththeobservedtransi-
7/2
theabsorptionlinesfromthissystemwouldbecomeweakersince tionsfromthislevelandgaussianfitstothelineprofiles.Thebest-
partofthelightwouldreachtheobserverunabsorbed.Weuse the fit N and its errors are found with a Monte Carlo simulation,
level
descriptionoftheapparentsizeoftheBlandford-McKeespherical which weapply asfollows. In every iteration, for each transition
expansiondescribedbyGranot,Piran&Sari(1999),whereweuse anequivalentwidthisrandomlypickedfromanormaldistribution
E = 6.3 1052 erg (Gorbovskoyetal. 2012). We assume the with as mean the measured value and as sigma the measured er-
iso ×
closecircum-burstmediumtohaveaconstantdensityn=1cm 3. ror.Forthissimulatedensembleof equivalent widths, thebest-fit
−
Atthetimeofthelastepoch(7.027days),wefindthesourcetobe N isdeterminedbyminimisingtheχ2.After10000iterations,
level
9.6 1016cm.Thisgivesalowerlimitonthesizeof eachof the thedistributionofbest-fitN ’sisfitwithanasymmetricgaussian
× level
absorbingcloudsof0.03pc.Althoughthisisnotaverymeaningful (i.e.,anormaldistributionwithadifferentσonbothsides),from
lower size-limit for a gas cloud, we point out that in general it whichwederivetheoverallbest N anditslowerandupper1σ
level
is difficult to put limits on absorber sizes, but see D’Eliaetal. error.Becausethefine-structurelinesliemostlyonthelinearpart
(2010a)formethodsappliedtoGRBabsorbersandPetitjeanetal. oftheCOG,theeffectofbissmall,andtheuncertaintyofthisvalue
(2000); Ellisonetal. (2004); Balashevetal. (2011) for examples isnottakenintoaccount.Iftwoormoretransitionsofalevelare
ofsizeestimatesofQSOline-of-sightobjects. detectedinanepoch(i.e.,if W > σW ),weobtainavaluefor
| λ| | λ|
N ,otherwiseweobtainanupperlimit(seeTable4).
level
Because the lines of the intervening systems do not vary
in strength, we can average the equivalent widths measured in
3.2.1 UV-pumpingmodels
epoch1 and epoch2. The MISC-COG analysis can be applied,
because wedetect many clearlines fromthese systems. Figure5 Thecolumndensitiesofthefine-structureandmetastablestatesare
shows the best fitting COGs for both intervening absorbers. We found to vary in time (see Table 4 and Figure 8). The presence,
carryoutthesameapproachasforthehost,byfirstconstrainingb andvariation,offine-structurelinesinGRBafterglowspectracan
withtheFeiitransitions.Wefindb = 12.6+11..42kms−1 forint1and generallywellbeexplainedbyexcitationduetotheUVfluxofthe
b=20.6+45..67kms−1forint2.Theequivalent−widthsandthecolumn GRBafterglow.Prochaska,Chen&Bloom(2006)suggestthatthis
densities−ofeachionspeciesarelistedinTable3. isthedominantexcitationmechanism,especiallywhenvariability
ismeasured.Theearlyhigh-resolutionspectroscopicdataobtained
forGRB060418(Vreeswijketal.2007)allowedsystematictestsof
otherexcitationmechanismssuchasabackgroundIRfieldandcol-
3.2 Detectionandvariabilityoftransitionsarisingfrom
excitedlevelsofFeiiandNiii. lisionalexcitation,butalsostronglyfavourstheUV-pumpingsce-
nario.ForGRB100901AweassumethatUV-pumpingistheonly
Attheredshiftofthehostgalaxywedetectlinesfromfine-structure relevantphoto-excitationmechanism.
levelsofFeii(6D ,6D ,6D and6D )andmetastablelevels We apply the photo-excitation model introduced by
7/2 5/2 3/2 1/2
ofFeii(4F and4D )andNiii(4F ).Theequivalentwidthof Vreeswijketal. (2007, 2011, 2013). For technical details we
9/2 7/2 9/2
manyoftheselinesvarieswithtime.Thisprovidestheopportunity refer to the 2013 paper. The general idea is that the UV-flux of
to derive the distance between the burst location and the absorb- the afterglow temporarily excites ions in a cloud with thickness
ingmaterial,assumingthatthepopulationoftheseexcitedstatesis l at a distance d from the location of the burst. As the afterglow
duetotheUV-radiationoftheafterglow(seeSection3.2.1).With brightnessfadesintime,theexcitedlevelsdepopulateafteratime
high-resolutionspectraitwouldbepossibletodirectlymeasurethe whichisdeterminedbytheEinsteincoefficientA or,equivalently,
ul
columndensitiesoftheionsthatareintheexcitedstates,viaVoigt theoscillatorstrengthofthetransitionsconsidered.Foragivenset
profile fits to the fine-structure lines (see e.g., D’Eliaetal. 2007; ofinputparameters,thecolumndensityofeachexcitedlevelasa
Vreeswijketal. 2007). In order to obtain column densities from functionoftime,N (t),ispredictedbythemodel.Bycomparing
level
the equivalent widths, we need to take an intermediate step with this temporal behaviour to the measured N (t) for all levels
level
the COG, where we fix b to the value found from the resonance simultaneously,wecanoptimisetheparameters.
Feiilines(seeSection3.1.1);i.e.,weassumethat theionsinthe Theafterglowfluxisincludedasaninterpolatedseriesof R-
excited statesareinthesame absorbing clouds astheions in the band magnitudes2 m (t) (seeupper panel Figure8),which corre-
R
groundstate;fortheimplicationsofthisassumptionseetheendof spondstotherest-frameUV-flux(at 2700Å)responsibleforpop-
∼
Section3.2.1.Notethatthevalueofbwillonlyhaveaverysmall ulatingthelevels.Themonochromaticfluxinthehost-galaxyrest
influenceonthecolumndensitiesoftheexcitedions,becausethe frameattheGRB-facingsideoftheabsorbingcloudiscomputed
linesareallweakandliemostlyonthelinearpartoftheCOG. asfollows:
of allTatrbalnessitAio1n,sAf2ro,mA3thaendfinAe4-sitnrutchteurAepapnedndmixetgaisvteabalneolevveervlsieowf Frest(t)= F0·10[mR(t)−AR,gal]/−2.5 λrest(1+zh) βν DL 2 (3)
FeiiandNiiithatwouldinprinciplebeobservableintheGMOS ν 1+zh " 6415Å # (cid:20) d (cid:21)
spectral range at thisredshift. Dueto thelow spectral resolution,
manyoftheselinesareblendedsuchthattheindividual variation 2 The observed R-band magnitudes come from GCNs
ofthecomponentscannotbemeasured.Thelinesthatareusedin Andreev,Sergeev&Pozanenko (2010a,b,c,d); Andreevetal. (2010);
themodellingareshowninboldfaceinTablesA1toA4. Kurodaetal.(2010a,b);Volnovaetal.(2010);assembledbyGombocetal.
InordertoobtainthecolumndensityNlevelofionsinaspecific (inprep.).
c 2013RAS,MNRAS000,1–??
(cid:13)
Theafterglowspectrum ofGRB100901A 9
Table3.Therest-frameequivalentwidthsWλforthelinesfromtwointerveningsystemsint1andint2atrespectivelyz1=1.3147andz2=1.3179.<Wλ/Å>
istheweightedaverageoftherestframeequivalentwidth(inÅngstrom)measuredinGMOSepoch1andepoch2.TheCOGsforthetwosystemsareshown
inFigure5.ThecolumndensitiesperionspecieshavebeencalculatedwiththeCOGwiththebestfittingb;theerrorsonthecolumndensitiesaretheresultof
takingintoaccounttherangeinb.EspeciallyforlinesontheflatpartoftheCOG(e.g.,Mgii),theeffectonNduetobislarge.
int1(z1=1.3147) int2(z2=1.3179)
b=12.6+11..42kms−1 b=20.6+45..67kms−1
Line fλ <Wλ/Å> −log(Nion/cm−2) <Wλ/Å> − log(Nion/cm−2)
Aliii 1854.7164 0.539 0.053 0.046 12.74+0.29 0.229 0.049 13.42+0.27
Aliii 1862.7895 0.268 0.073±0.047 ...−0.43 0.171±0.049 ...−0.17
± ±
FFFeeeiiiiii 222332474449...248167205944 132...221599270××111000−−123 000...310183722±±000...000323086 14.2..8....−+00..0195 00..107364±a 00..002274 13.5.7..−+00..0089
Feii 2382.7641 3.432×10−1 0.365±0.030 ... 0.325±0.029 ...
Feii 2586.6495 7.094×10−2 0.296±0.027 ... 0.136±0.023 ...
Feii 2600.1722 2.422×10−1 0.405±0.028 ... 0.344±0.027 ...
× − ± ±
Mgii2796.352 6.123 10 1 0.532 0.030 14.86+0.26 0.845 0.035 14.83+1.39
Mgii2803.531 3.054×10−1 0.531±0.030 ...−0.47 0.792±0.034 ...−0.68
× − ± ±
Mgi 2852.9642 1.83 0.143 0.023 12.17+0.09 0.118 0.022 12.01+0.09
± −0.11 ± −0.11
athislineisblendedwithNiii4F9/2λ2166.23atthehostredshift.
Table 4. Lowerlevel column densities log Nlevel(t)/cm−2 for Feii and Niiiexcited states derived from fine-structure line equivalent widths with COG
analysis.Allerrorsandlimitsare1σ. (cid:16) (cid:17)
Feii Niii
time(d) 6D9/2 6D7/2 6D5/2 6D3/2 6D1/2 4D7/2 4F9/2
epoch1 0.0526 15.20+0.0813.77+0.0513.58+0.0613.30+0.0712.92+0.1112.54+0.1113.25+0.10
0.09 0.04 0.10 0.09 0.14 0.18 0.15
epoch2 1.0095 15.29+−0.0713.68+−0.0513.46+−0.0913.14+−0.09<12.8−3 12.54+−0.1113.73+−0.07
0.10 0.05 0.13 0.15 0.24 0.12
xsh 2.7492 15.16+−0.12<13.4−6 <14.0−2 <13.4−7 <13.40 <13.0−3 13.54+−0.09
0.18 0.13
epoch3 7.0270 15.34+−0.21<13.37 <13.80 <13.39 <13.47 <12.97 <13.7−5
0.48
−
in which F = 3.02 10 20ergs 1cm 2Hz 1 is the flux tion A = 0.2, the data favoursaconfiguration withavery large
0 × − − − − V
of Vega at the effective wavelength (6410Å) of the R-band (> 1kpc) absorbing cloudat asmallerdistance. Fixingthecloud
(Fukugita,Shimasaku&Ichikawa 1995), A = 0.264 is the size l leads to absorber distances of d 165 275pc. Figure 8
R,gal ∼ −
Galacticextinction(Schlegel,Finkbeiner&Davis1998),λ isthe showsthefittothedataforthemodelwithl6500pc(solidlines)
rest
wavelengtharrayoftherelevanttransitionsatwhichthefluxisre- andthemodelwithl=1pc(dottedlines).Bothmodelsaredecent
quiredtocalculatetheamountofexcitation,β isthespectralslope fits:ourdataarenotconstrainingenoughtodiscriminatebetween
ν
andD = 9.968 109pcistheluminositydistance.Theeffective the close-large and the distant-small cloud case. But in all cases,
L ×
Dopplerparameterbisfixedto22.1kms 1,consistentwithhowwe the influence of theGRB reaches distances of a few hundreds of
−
convertedequivalentwidthstocolumndensities(seeSection3.2). parsecs.InSection4.1wewilldiscusstheimplicationsofthises-
Parametersthatcanbeconstrainedfromthelightcurveand/orSED timatedabsorberdistanceandcompareittovaluesfoundinother
fitting(Gombocetal.,inprep)arekeptfixed:opticalspectralslope GRBs.
β = 0.82, and optical extinction in the host galaxy A = 0.21
ν V
assumingaSmallMagellanicCloudextinctionprofile(Gombocet
al.,inprep).WeperformedmodelswithA =0aswelltostudythe Though not detectable with the spectral resolution used, it
effect.A isnotincludedinEquation3bVecausethemodelallows couldbepossiblethatground-stateFeiiispresentinmorevelocity
tospecifyVwheretheopticalextinctiontakesplace.Weplacetheex- componentsthanexcitedFeii,i.e.,thatthereisFeiithatisnotasso-
ciatedwiththeexcitedgas.IfthereexistsmoreFeiifurtheraway,
tinctionintheabsorbingcloud,andapplythenecessarycorrections
thenthiswoulddecreasetheamountofFeiithatisassociatedwith
totheflux.
the excited Feii, and hence this would move the absorber closer
Table 5 shows the fit results of the model to the measured totheburst,i.e.,wewouldinferasmallerGRB-absorberdistance.
N (t) as listed in Table 4, with different requirements for the Veryroughly, if thisfraction of Feii associated toexcited Feii is
level
parameters. The model without extinction fits the data best with only50%(ratherthanthenowassumed100%),thenitwouldde-
a small cloud at 250pc, but when we include optical extinc- creasethedistancebyafactorof √2.Soevenwhenassumingsuch
∼
c 2013RAS,MNRAS000,1–??
(cid:13)
10 OlgaE. Hartooget al.
Figure6.Curves-of-growth(COG)andspectrumexcerptsforthethreedetectedlinesfromtheNiii4F9/2metastablelevel.Everycolumnrepresentsanepoch,
withtheobservedtimesincetheburstindicatedontop.TheupperplotshowstheCOGforb=22.1kms−1towhichwefittheequivalentwidthsforthethree
transitions(reddiamonds).Thebest-fitcolumndensityNlevelisindicated,togetherwiththeχ2νoftheCOG-fit.BelowtheCOG,thethreeindividuallinesare
displayedonavelocityscale,withthecorrespondinggaussianlineprofilefit.Therestframeequivalentwidthandoscillatorstrengthareindicated.Weusethe
samelinelabelcolouringasinFigure1.
Table5.Photo-excitation modellingresultsfortheenvironmentofGRB100901Awithdifferentsettings.Column(1)extinction AV inthehostgalaxy,(2)
requirementforthedistance dfromthebursttotheabsorber,(3)requirementforthethicknessloftheabsorbingcloud,(4)fitresultd,(5)fitresultl,(6)
pre-burstcolumndensitiesofFeiiand(7)Niii(groundstate)and(8)thereducedchi-squareofthefit.SeealsoFigure8.
input output
AV d l d/pc l/pc log NFeII/cm−2 log NNiII/cm−2 χ2ν
(cid:16) (cid:17) (cid:16) (cid:17)
0.0 free free 250 75 0+240 15.25+0.05 13.91+0.07 1.54
0.2 minimal100pc free 100+±61 1−127 569 15.26−+00..0055 14.14−+00..0180 0.73
0.2 free fixedto1pc 275− 12 1 ± 15.24−+00..0065 13.94−+00..1036 1.35
± 0.06 0.07
0.2 free fixedto100pc 249 13 100 15.24−+0.05 13.95−+0.06 1.33
± 0.06 0.07
0.2 free maximal500pc 165 73 500+ 15.22−+0.05 14.02−+0.10 1.15
± −469 −0.06 −0.13
c 2013RAS,MNRAS000,1–??
(cid:13)