Constraining the Role of SN Ia and SN II in Galaxy Groups by Spatially Resolved Analysis of ROSAT and ASCA Observations

14 Pages·0.44 MB·English

by  A. Finoguenov

Mon.Not.R.Astron.Soc.000,1–14(1999) Printed1February2008 (MNLaTEXstylefilev1.4)
Constraining the Role of SN Ia and SN II in Galaxy
Groups by Spatially Resolved Analysis of ROSAT and
ASCA Observations
⋆
A. Finoguenov1 and T.J. Ponman2
1Space Research Institute, Profsoyuznaya 84/32, 117810 Moscow, Russia.
2School of Physics& Astronomy, Universityof Birmingham, Edgbaston, Birmingham B15 2TT, UK
9
9
Accepted forMNRAS
9
1

n
ABSTRACT
a
We present the results of modelling the distribution of gas properties in the galaxy
J
groups HCG51, HCG62 and NGC5044, and in the poor cluster AWM7, using both
9
ASCA SIS and ROSAT data. The spectral quality of the ASCA data allow the ra-

dial distribution in the abundances of several elements to be resolved. In all systems
1
apart from HCG51, we see both central cooling flows, and a general decline in metal
v
0 abundances with radius. The ratio of iron to alpha-element abundances varies signif-
0 icantly, and in comparison with theoretical supernova yields, indicates a significant
1 contribution to the metal abundance of the intergalactic medium (IGM) from type
1 Ia supernovae. This is seen both within the groups, and also throughout much of the
0 cluster AWM7. The total energy input into the IGM from supernovae can be calcu-
9 lated from our results, and is typically 20-40 per cent of the thermal energy of the
9 gas,mostly fromSNe II. Our resultssupport the idea that theSN II ejecta havebeen
/
h more widely distributed in the IGM, probably due to the action of galaxy winds, and
p the lower iron mass to light ratio in groups suggests that some of this enriched gas
- has been lost altogether from the shallower potential wells of the smaller systems.
o
r Key words: galaxies:general– galaxies:evolution– X-rays:galaxies
t
s
a
:
v
i
X 1 INTRODUCTION ofthemetalsinSNII.MorerecentlyFukazawa etal.(1998)
have studied a much larger sample of 40 clusters, and have
r In this paper we present a study of the X-ray emission
a also discriminated out the central regions, which may be
of galaxy groups using the good spatial resolution of the
dominated by strong cooling flows. Their results indicate a
ROSATPSPCinconjunctionwiththesuperiorspectralca-
significantcontributionfromSNIa,inadditiontoSNII,es-
pability of the ASCA SIS. For relatively cool systems like
peciallyinlowermasssystems.Inprinciple,ASCAcanpro-
groups (Tgas ∼ 1 keV) the X-ray spectrum contains fairly videmoredetailedinformationonthespatialdistributionof
strong signatures from a number of elements, and this al-
elements,ratherthanjustonintegratedabundances,andin
lows us to investigate the distribution of heavy elements in
thepresentstudyweexploit thiscapability.Ourworkcom-
theIGM.MakinguseofthedifferentelementyieldsofSNIa
plements that of Fukazawa et al. (1998), in that we derive
and SN II,we can use this information to estimate the role
detailed distributions for a small sample, where they study
of different typesof SNein enriching theIGM.
theintegrated properties of a large collection of systems.
According to standard theories of the chemical evolu-
tionofearly-typegalaxies,theirstarformationwilltriggera Thedistributionofmetalsinlowmassclustersisofpar-
galactic wind, causing a loss of the enriched gas, which can ticularinterest,sinceitisforsuchsystemsthattheenergetic
beretainedinthepotentialsofgroupsandclustersofgalax- effect of galaxy wind injection should have thegreatest im-
ies(e.g.Renzini etal.1993).Inthefirstsubstantialstudyof pact on the IGM. It appears that such effects have already
cluster abundances using ASCA, Mushotzky et al. (1996) been seen in the low beta values for groups and in a steep-
studied the integrated abundances of O, Ne, Si, S and Fe ening of the L : T relation at low temperatures (Ponman
in four galaxy clusters, and concluded that the balance be- et al. 1996, Cavaliere, Menci & Tozzi 1997). These winds
tween α-elements and iron favours the origin of most or all bring with them metals, so that a study of metal distribu-
tions places constraints on the history of wind injection. If,
forexample,windshaveblownmuchoftheSNIImetalsout
⋆
E-mail:[email protected] of groups altogether, we would expect them to have lower
(cid:13)c 1999RAS
2 A. Finoguenov and T. J. Ponman
iron mass to light ratios, and abundance ratios tilted more
towards SN Ia than is the case in rich clusters. A study of Table 1.Opticalcharacteristics.
the role of SN Ia and SN II in groups may also shed light
on the mystery of iron production by ellipticals. The low Name D LB,cD LaB RLB Rv Rc
abundanceobserved byASCAin hot elliptical galaxy halos Mpc 1011L⊙ 1011L⊙ Mpc Mpc kpc
causes substantial difficulties for theoretical models of their HCG62 82 0.95 1.09 1.09 27
chemical history (Arimoto et al. 1996). HCG51 155 2.8 1.19 1.19 59
NGC5044 54 0.68 1.6b 0.5 1.19 180
For our study, we require good photon statistics, and
AWM7 106 1.9 13.2b 1.84 1.89 220
have therefore picked three of the brightest X-ray groups.
Two of these (HCG51 and HCG62) are compact groups,
whilst the third (NGC5044) is a loose group. In practice
a blueluminosityofearly-typemembergalaxies
(Mulchaey et al. 1996) the properties of compact groups
b Assuming30%contributionofspiralstoatotalblueluminosity
appeartobesimilartothoseof(X-raybright)loosegroups,
andthedistinction between thetwomay notbefundamen-
tal. For comparison, we also studya richer system,AWM7, ofallthesourceswereusedasaninputforASCAdatamod-
(which should lie above the L:T break), but one which is elling.ForimaginganalysisoftheROSATdataweusedthe
still cool enough to allow ASCA to determine abundances software described in Snowden et al. (1994) and references
for a range of elements. therein. For the ROSAT/PSPC observation of HCG62 we
In our study we use a 3-dimensional approach to mod- alsoperformedanindependentthree-dimensionalmodelling
elling ofASCAdatafortheseobjects, and wealso compare analysisusingtheBirminghamclusterfittingpackage(Eyles
our results for HCG62 with an independent 3-dimensional et al. 1991) for comparison with our ASCA results.
analysis of ROSATPSPCdata usingtheBirmingham clus- In our approach to the analysis of ASCA data, where
ter fitting package. Ho = 50 km s−1 Mpc−1 is assumed all thederived spectra are fitted simultaneously, and where
throughout the paper. Unless clearly stated otherwise, all correlation between different regions is rather high, a sim-
errors quotedin thispapercorrespond to90 percent confi- pleminimization of χ2 provestobeanill-posed task(Press
dencelevel. et al. 1992, p.795). The best fitting model displays large
fluctuationsintemperatureandmetallicity profilesasitat-
temptstofitnoiseinthedata.Thesolutiontothisproblem
is toaccept amuch smoother model which gives an accept-
2 ANALYSIS
able fit (rather than the very best fit) to the data. This is
IntheanalysisofASCAdata,presentedbelow,weuseddata achieved through regularization, which applies a prejudice
fromtheSIS0andSIS1detectors,whichprovideenergyres- infavourofsomemeasureofsmoothness(Press etal.1992,
olution of ∼75 eV at 1.5 keV. A detailed description of the p.801). We have made a default assumption that a linear
ASCAobservatoryaswellastheSISdetectorscanbefound function(oflogradius)isagood representationofourtem-
in Tanaka, Inoue & Holt (1994) and Burke et al. (1991). peratureandabundancedistributions.Departuresfrom lin-
Standard data screening was carried out using FTOOLS earity are quantified by calculating the second derivative
version 3.6. The effect of the broad ASCA PSF was sim- of the distribution, which is squared and added to the χ2
ulated, as described in Finoguenov et al. (1998), and the statistic.
analysis allows for theprojection onto the sky of the three- Inusinganyregularization technique,special attention
dimensional distribution of emitting gas. In this approach, shouldbepaidtothebalancebetweenmaximizingthelikeli-
weuseROSATdatatoderivethesource’ssurfacebrightness hoodandoptimizingthesmoothnessofthesolution.Thisis
profile,whichisrepresentedbyaoneortwocomponent“β- traditionally doneby introducing a weighting constant into
model” (Jones&Forman 1984). Thisisused toconstructa the regularization term. We set this constant so as to ob-
three-dimensionalmodelwhichisprojectedontotheregions tain a solution which lies within the 68 per cent confidence
oftheSISdetectorschosenforextractingspectra.Thethick- regionsurroundingtheminimumχ2 (i.e.unsmoothed)solu-
nessoftheshellsinour“onion peeling”techniqueischosen tion in then-parameterspace. The allowed offset in χ2 was
to avoid any drastic variation of the temperature (T) and evaluated using theprescription of Lampton et al. (1976).
metallicity (Z) within one bin, which would invalidate our Since regularization introduces a dependence of the
single component spectral modelling. measurement in a given spatial bin on that from the ad-
GIS data were not used in our analysis due to their jacent bins (due to the operation of the prejudice towards
lower energy resolution and uncertainties in calibration be- smoothness), presenting formal error bars at every point is
low 2 keV. Our techniquefor analyzing the ASCA data re- misleading. Instead, more valid representation is an area of
stricts us to usage of ASCA pointings with less than 25′ possible parameter variation with radius. We choose a con-
offset from thesource centre,dueto thesimilar restrictions fidence level of 90 per cent for these representations. Thus,
in calibration data for the ASCA ray-traced PSF and the points will represent the best-fit solution, while a shaded
importance of accounting for the stray light at larger offset zone will mark the area of solutions allowable within the
angles (cf Ezawa et al. 1997). In practice this restriction is chosen confidence level.
of importance only for theAWM7 analysis. Theanalysis presentedin thispaperdiffersfrom previ-
All fits are based on the χ2 criterion. No energy rebin- ousworkontheseobjects(e.g. Fukazawa etal.1996),which
ning is done, but a special error calculation was introduced have generally ignored the effects of projection and PSF
following Churazov et al. (1996), to avoid ill effects from blurring.SincethePSFisenergy-dependent,simplifiedanal-
small numbers of counts. ROSAT (Truemper 1983) images ysescangivemisleadingresults(Takahashi etal.1995).For
(cid:13)c 1999RAS,MNRAS000,1–14
Constraining the Role of SN Ia and SN II in Galaxy Groups 3
relativelycoolsystems,likethegroupsstudiedhere,thereis
an additional effect connected with thepossible presence of (a)
temperaturegradients.Determination ofiron abundanceat
temperatures ∼ 1 keV is based on L-shell lines. Individual
lines in the L-shell complex are unresolved by ASCA, and
merge into a broad peak with a position which is sensitive
totemperature.Ifthetemperaturestructureisnotresolved
)
and a mean temperature is used, then iron abundance will
beunderestimatedbyasignificantfactor,becausetherewill −1
besome mismatch between peak positions. V
To illustrate this effect, we simulate an SIS spectrum
e 0.7 0.8 0.9 1 2 3
using the characteristics of the HCG62 data (ARF file, re- Energy, keV
k
sponsefile,durationoftheobservation).Themodelinvolves
a mix of 2 temperatures: 0.7 keV and 1.2 keV, with both (b)
components having equal emission measures (a “norm” pa-
1
rameter of 2.×10−4 in XSPEC, typical for the groups an- −
s
alyzed here) and all element abundances equal to 0.5 so-
lar. MEKAL plasma code is used throughout this illustra-
tion. When the simulated data are fitted with a single- s
temperature model we find kT = 0.823 ± 0.026, Mg= t
0.18±0.14, Si=0.11±0.08 and Fe=0.19±0.03 with emis-
n
sionmeasurenormalizationbeing7.3±1.6×10−4.Hencethe
u
abundances are underestimated, and the emission measure
overestimated. Such effects may compromise any study in o 0.7 0.8 0.9 1 2 3
Energy, keV
which temperature structure is not resolved. This is likely c
to apply even to thecurrent study,in central cooling flows, (
(c)
wheresteeptemperaturegradientsandmultiphasegas may
bepresent.
In all our spectral modelling, we use the MEKAL x
model(Mewe etal.1985, MeweandKaastra 1995, Liedahl u
et al. 1995). Abundances of He and C were set to so-
l
lar values as in Anders & Grevesse (1989), where ele- f
ment number abundances relative to hydrogen are (85.1,
12.3, 3.8, 3.55, 1.62, 0.36, 0.229, 4.68, 0.179)×10−5 for
d
O,Ne,Mg,Si,S,Ar,Ca,Fe,Ni, respectively. In the case of the
threegroups,onlyMg,SiandFefeaturesareclearlyseenin e 0.7 0.8 0.9 1 2 3
thedata.WecuttheASCAspectraabove2.2keV,soSand l Energy, keV
Ar features are outside the range of measurements, and ar- a
range elements into four groups, based on their production c (d)
history: O and Ne; Mg; Si, S and Ar; and Ca, Fe and Ni.
S
TheanalysisyieldsmeaningfulvaluesforabundancesofMg,
SiandFe.OandNeresultsarenot derivedforgroups, and
we do not group O with other elements, like Si, in order to
avoid possible effects from ASCA calibration uncertainties
atlowenergies. Forthesamereason,wedonotpresentany
results for O abundancein theanalysis below.
Our choice of the MEKAL plasma code leads to sys-
tematicallylowertemperatures(atthe10–20percentlevel) 0.8 1 2 4 6
in the temperature range below 1 keV, compared to the Energy, keV
Raymond-Smith code (Raymond et al. 1977). There have
been made several attempts to quantify the uncertainty in
the Fe abundances determined using L-shell line emission.
Figure1.SpatiallyresolvedASCASISspectraof(a)NGC5044,
It was shown in a study by Matsushita (1998), that if Fe
atradiiof0′–2′–8′–16′ frombottomtotopplots,(b)HCG62,at
abundanceisdecoupled from theotherheavy elements, the
radii0′–1.5′–8′–16′ frombottomtotopplots,(c)HCG51,atradii
range in Fe abundances derived from different codes is 20 0′–2′–6′frombottomtotopplots,and(d)AWM7,atradii0′–3′–
per cent. We therefore add a systematic error of ±10 per 10′–25′ from bottom to top plots). Y-axis values are arbitrarily
cent at 1σ confidencelevel to thederived Fe abundances. scaledforclarity.
Using the 0.7–7.0 keV band in the case of AWM7, we
fit all major elementsseparately, except for thegrouping of
Ca and Ni with Fe, and obtain useful results for the abun-
dances of Ne, Si, S and Fe. A special comment should be
made on Mg. It has been noted (Mushotzky et al. 1996)
(cid:13)c 1999RAS,MNRAS000,1–14
4 A. Finoguenov and T. J. Ponman
Figure 2. Temperature profile of NGC5044. ROSAT PSPC Figure 3. Element abundances in NGC5044. ROSAT PSPC
points are represented by black crosses, solidline represents the points are represented by black crosses, solid line represents the
best-fitcurvedescribingASCAresultswithfilledcirclesindicat- best-fitcurvedescribingASCAresultswithfilledcirclesindicat-
ing the spatial binning used. Shaded zone around the best fit ing the spatial binning used. Shaded zone around the best fit
curvedenotes the90percentconfidence area. curvedenotesthe90percentconfidencearea.Lightshadedzone
ontheFeabundance plotindicatesthecontributiontothiszone
ofa10percentsystematicerror,asdiscussedinthetext.
that inthecase ofcluster emission, thismay beaffected by
the proximity of the poorly understood 4–2 transition lines
ofiron,whichcannotberesolvedwithASCASIS.Thisprob- analysis discussed in the remainder of this section are col-
lem affects our AWM7 data, however at the lower temper- lected together in Table 2. The radius within which these
atures of groups, the relevant Fe L-shell lines are weak and masses are derived is listed in the second column of the ta-
hence uncertainties in modelling them has little impact on ble. All error intervals are quoted at 90 per cent confidence
theabundance of Mg determined. Spatially resolved ASCA level.Limits on masses areobtained byintegrating thecor-
SIS spectra obtained for the sources in our study are illus- responding lower and upper 90 per cent boundary on the
trated in Fig.1. abundance estimation and adding an error on gas mass es-
timation in squared.
3 RESULTS
In this section we present the details of the analysis and
3.1 NGC5044
results for each individual system analyzed. Optical data,
whichwillbeusedforcomparisonwiththeX-rayresults,are The galaxy group NGC5044 is well-suited for spatially re-
provided in Table 1. Columns in this table are: (1) system solved spectroscopy, it is nearby (z = 0.0087), and shows
name, (2) adopted distance in Mpc, (3) B-band luminos- brightandfairly symmetricaldiffuseX-rayemission (David
ityofthecentralgalaxy forthetwosystemswith dominant etal. 1994).Thegrouphasadominantcentralellipticalsur-
central galaxies, (4) total luminosity of early-type galaxies, roundedbymanysmallergalaxies.FromanalysisofROSAT
and (5) radius within which this optical luminosity is mea- PSPC observations (David et al. 1994) the diffuse X-ray
sured. Optical data were taken from Hickson et al. (1992) emission of this object extends to ∼ 20′ radius (1′ corre-
for HCG51 and HCG62, David et al. (1994) for NGC5044 sponds to 15.7 kpc). A strong temperature gradient is ob-
andBeers etal.(1984)forAWM7.Forthelatterobjectwe served in the centre of the diffuse emission, pointing to the
adopt a spiral fraction of 0.3, average for clusters, since no presenceofacoolingflow.Deprojectiongivesacentraltem-
measurementsareavailabletodate,howeverourresultsare perature ∼0.8 keV. The mass accretion rate is significant
insensitive to this choice. The remaining columns give: (6) onlywithinthecentral40kpcandamountsto∼10M⊙ per
virialradiuscalculatedusingaformuladerivedfromsimula- year. The gas cooling time at r≈40 kpcis 109 yr.
tionsbyNavarro etal.(1994),Rvirial=1.04Tk0e.5V(Ho/50)−1 ASCA observations of NGC5044 were carried out dur-
Mpc,whereTisluminosityweightedX-raytemperature(ex- ing June 20–21 1993. We used a representation of the sur-
cluding the cooling flow), and (7) optical core radius taken face brightness profile from ROSAT PSPC data given by a
fromHickson etal.(1992),Ferguson&Sandage(1990)and β-modelwithparameters β=0.51, ra =10.6 kpc,obtained
Dell’Antonio et al. (1995). from our analysis of ROSAT PSPC data in the 0–40′ re-
Masses of gas and various metals derived in the X-ray gion, excluding point sources and a “cooling wake”, found
(cid:13)c 1999RAS,MNRAS000,1–14
Constraining the Role of SN Ia and SN II in Galaxy Groups 5
Table 2.Cumulatedgasandmetalmasses.
Name Rout Mgas MFe MSi MMg MS MNe
kpc 1012 M⊙ 108 M⊙ 108 M⊙ 108 M⊙ 108 M⊙ 108 M⊙
HCG62 410 1.656(1.56–1.76) 3.82(1.8–5.8) 2.24(0.5–3.4) 0.23(0.1–3.8) — —
HCG51 270 0.609(0.56–0.65) 4.05(3.0–5.1) 1.99(1.6–2.3) 1.59(0.6–1.9) — —
NGC5044 270 0.965(0.92–1.00) 4.09(2.7–4.9) 1.75(1.1–2.2) 0.49(0.2–1.5) — —
AWM7 790 21.34(21.1–21.5) 95.2(74–115) 75.7(61–89) — 9.4(1.3–18) 113(55–149)
Figure 4. Temperature profile of HCG62. Results of a three- Figure 5. Mg, Si and Fe abundance profiles for HCG62 from
dimensional analysis of ROSAT PSPC data, fitting a variety of ASCAdata.DatarepresentationissimilartoFig.3.
modelstothedata,areshownindashedlines.Filledcirclesindi-
catingthespatialbinningandtheshadedareashowingthe90per
The ASCA abundances drop from 0.3, 0.5 and 0.5 solar at
cent confidence level areaaround thebestfit(solidline)present
r=40 kpctonear0.0, 0.2 and0.1 solar byr=200 kpcfor
theASCASISresults.
Mg, Si and Fe, respectively. Shown errors on Fe abundance
are dominated by the assumed systematic uncertainty in
byDavid etal. (1994).TocomparewithourASCAresults,
modelling.
we carried out an identical 3-dimensional spectral analysis
with ROSATPSPC data.
In Fig.2 we present atemperature profile of NGC5044.
3.2 HCG62
ROSATPSPCpointsarerepresentedbyblackcrosses,solid
line represents the best-fit curve describing ASCA results HCG62, is a compact group, taken from the catalogue of
withfilledcirclesindicatingthespatialbinningused.Shaded Hickson (1982), which was compiled by locating compact
zonearoundthebestfitcurvedenotes90percentconfidence configurations of galaxies on the Palomar All Sky Survey
area.TheASCAdataconfirmapronouncedcooling flowat in theE-band.HCG62 is themost luminous of theHickson
thecentrewith atemperaturedrop from 100 kpcto10 kpc groupsintheX-ray(Ponman etal.,1996)andananalysisof
byafactoroftwo.ASCAandROSATresultsagreeatradii theROSATPSPCdataforHCG62ispresentedinPonman
exceeding30kpc,withthedisagreementatsmallradiiindi- &Bertram(1993).TheASCASISobservationanalyzedhere
catingthepresenceofcomplextemperaturestructureinthe was carried out on January 14 1994.
centralcoolingflow.Ourresultsagreewellwiththeprevious For our analysis we adopt a distance of 82 Mpc, which
findingofDavid etal. (1994)thatbetween60and250kpc corresponds to a scale 1′= 23.9 kpc. The X-ray surface
thegas is nearly isothermal. brightness profile is characterized by a double β-model fit-
Element abundance profiles are shown in Fig.3. (We ted to the ROSAT PSPC data in the r =0−20 ′ interval.
excludethecoolingregion,whereabundanceresultsmaybe The derived parameters of (β, ra) are (0.66, 10.3 kpc) and
unreliable.) A significant decrease with radius is observed, (0.30, 52 kpc). Results of our analysis of ASCA SIS data
andweseeforthefirsttimethatthisappliestoMgandSiin were then compared with the results of three-dimensional
asimilarwaytoFe.Theabundanceofirondeterminedfrom modellingoftheROSATPSPCdatausingtheBirmingham
ROSATdata(greypoints)alsoshowsadecreasewithradius. cluster fitting package (Eyles et al. 1991), which also al-
(cid:13)c 1999RAS,MNRAS000,1–14
6 A. Finoguenov and T. J. Ponman
Figure 6. Temperature profile of HCG51, derived from ASCA Figure 7. Mg,Siand Feabundance profiles inHCG51 derived
SISdata.Thebest-fitcurvedoesnotindicatethepresenceofthe fromASCASISdata(datarepresentationissimilartoFig.3).No
coolingflowinthe data, althoughthe uncertainty intheparam- significantgradients areseen.
eterestimationislargefortheinnerpoints.
link ASTERIX analysis system, to derive a surface bright-
lowsfortheeffectsofprojection andenergydependentPSF ness profile for use in the ASCA datamodelling procedure.
blurring, but adopts simple analytical models to represent Diffuse emission is centered on the brightest elliptical
the continuous gas density and temperature distributions, galaxy in the group, and is detected in the HRI data out
rather than the regularised discrete distributions employed to6′ from thegroup centre.Anadequatedescription of the
in ourASCA analysis. profile in terms of β-models requires the introduction of a
In Fig.4 we compare the temperature distributions de- secondcomponent givingparameters(β,ra)equalto(11.1,
rivedfromtheASCAandROSATdatausingthesetwodif- 33.2 kpc) and (0.30, 81 kpc).
ferent analysis systems. In the case of the ROSAT data, a ASCAobservationsofHCG51werecarriedoutonJune
variety of different models for the temperature distribution 3–51994usingtwo-CCD modefortheSIS.Weadoptadis-
have been fitted to the data, to give an indication of the tanceof154.7MpctoHCG51,whichgivesascale1′=45kpc.
model dependence of the results. In general the two sets of The derived temperature and abundance profiles are
resultsshowanencouraginglevelofagreement.Thetemper- presented in Fig.6 and 7. In contrast to the other systems
ature profile is characterized bycooling at thecentre and a studied here, HCG51 does not show pronounced signs of a
gradual decrease toward theedge of thegroup. cooling flow. The temperature is ∼ 1.35 keV, with some
TheelementdistributionsderivedfromtheASCAanal- indicationofagentledeclinewithradius.Nevertheless,con-
ysis,presentedinFig.5,showstronggradientsofMg,Siand sidering the confidence area for temperature estimation, it
Feabundance,inthesamesenseasNGC5044.Atr=20kpc is possible to fit in the central cooling zone. The derived
Mg and Si abundance are 0.4 solar, Fe is 0.3 solar. All de- abundances of Mg, Si and Fe do not show any significant
clineto∼0.1solarbyaradiusof400kpc.TheFeabundance variation with radius. The lack of pronounced central cool-
gradientimpliedbyROSATdatamatchestheASCAdatain ingandtheflatabundanceprofiles(contrastingwithstrong
the outer parts, but rises to ∼0.6 solar at 40 kpc, although gradientsin theothersystems) stronglysuggest thatrecent
this difference near the centre is still within the combined gas mixing has taken place in HCG51.
90% errors from both analyses.
3.4 AWM7
3.3 HCG51
ASCASISobservationsofAWM7wereobtainedduringAu-
From results of the compact group survey of Ponman gust 7–8 1993 and February 10–12 1994. We adopt a dis-
et al. (1996), HCG51 is one of the most X-ray luminous tancetoAWM7of105.6 Mpc,sothat1′=31kpc.Asurface
of the compact groups. Apart from a short observation in brightness model was taken from the work of Neumann &
the ROSAT All Sky Survey, the group was never observed Boehringer(1995),givingvaluesoftheparameters(β,ra)in
by the ROSAT PSPC. However a 25 ksec observation was the two β-model approximation of (0.25, 5 kpc) and (0.53,
obtainedinMay1995,andhasbeenreducedusingtheStar- 102kpc).Therelativenormalizationofthecomponentswas
(cid:13)c 1999RAS,MNRAS000,1–14
Constraining the Role of SN Ia and SN II in Galaxy Groups 7
Figure8. TemperatureprofileofAWM7.Filledcirclesrepresent Figure 9. Abundances ofNe, Si,Sand FeinAWM7, obtained
best fit to our ASCA SIS data with shaded area showing the 90 fromASCASISdata(datarepresentationissimilartoFig.3).An
percentconfidencelevel.Crosses,denotedwithadottedlinerep- initialabundancedecreaseisfollowedbyflatteningataradiusof
resentROSAT/PSPCdatafromNeumann&Boehringer(1995), ∼200kpc.
solidblackcrossesdenote theresultsofEzawa etal.(1997). All
confidence intervals on the plot are given at 90 per cent confi-
cent, and Ne is poorly constrained. Hence allowing for the
dence.
effectsofcentralcoolingisonlylikelytoincreaseabundance
gradients,andamplifythecontributionofα-elementsinthe
chosen to give a central density ratio of 2, as obtained by
centreof thecluster, which we discuss below.
Neumann & Boehringer (1995).
ToinvestigatethesystematiceffectsofASCAPSFcal-
In Fig.8 we compare the temperature profile derived
ibrationuncertainties,weexploredtheeffectsofvaryingthe
here from ASCA data with the previous findings of Neu-
spatialbinningofthedata.Theconclusion wederivedfrom
mann & Boehringer (1995) and also from an annular spec-
theseexercisesisthatthedetectedabundancegradientsare
tral analysis (neglecting projection effects) of ASCA data
robust,exceptionforthecentralexcessofsulphurinAWM7,
by Ezawa et al. (1997). The latter authors have extended
which was found tobe lower with coarser binning.
their analysis to larger radii than the present work, due to
amore detailed treatmentof stray light effects, butdid not
apply any regularization, resulting in rather coarse spatial
resolution.Thethreeanalysesareinbroadagreementapart 4 DISCUSSION
fromnearthecentre,whereROSATdatashowthestrongest
Using the above results for the abundance distributions of
cooling flow,while theanalysis of Ezawa et al. (1997) does
iron and α-elements in these systems, we now proceed to
notrevealanytemperaturedecreaseatall.Ourtemperature
examine the contributions from SNe II and SNe Ia to the
profile is intermediate, but disagrees significantly with the
metalenrichment.Wealsoderivetheironmasstolightratio
ROSAT results only at r < 50 kpc, indicating the proba-
(IMLR)for theintracluster gas ineach system, which gives
ble presence of a multi-phase cooling flow, as discussed by
clues about overall metal production and loss, and given
Markevitch & Vikhlinin (1997).
our derived supernova rates we evaluate the importance of
In Fig.9 we present abundance profiles derived for Ne,
energy injection by SNe. Finally, we consider our results in
Si, S and Fe. All abundances show a decrease with radius
thecontext of hierarchical clustering models.
which, with the possible exception of Fe, flattens off out-
side 200 kpc. Central values are 0.4, 1.1, 0.6 and 0.7 solar,
respectively, dropping at r =800 kpc to 0.3, 0.5, 0.1 and
4.1 Roles of different types of SNe in chemical
0.2 solar. The strong cooling flow may be expected to af-
enrichment of the IGM
fect derived abundances in the inner regions, though the
effects will be less severe than in cooler systems. To gauge The heavy elements liberated in type Ia and type II super-
theseeffects we havefittedspectra from thecentral 3′ with novaediffermarkedlyintheratioofirontotheelements(O,
a model including a cooling flow component. The inclusion Ne, Mg, Si and S) generated by the alpha-process. For ex-
of such a component has a negligible effect on the derived ample, the Si/Fe abundance ratio is at least 4 times higher
Fe abundance (which is largely based on the Fe K line for in the ejecta of SN II, compared to SN Ia. Unfortunately,
thissystem),whilstabundancesofSiandSriseby∼35per the complexity of the theory of SN II explosions leads to
(cid:13)c 1999RAS,MNRAS000,1–14
8 A. Finoguenov and T. J. Ponman
11 11
00..88 00..88
00..66 00..66
00..44 00..44
00..22 00..22
00 00
00 00..22 00..44 00..66 00..88 11 00 00..22 00..44 00..66 00..88 11
Figure 11. Input of SN Ia into the iron mass, calculated from
Si/FeandNe/Feratios,assumingT95(leftpanel)andW95(right
panel) yields for SN II. Points correspond to measurements at
different radii in AWM7. Error bars are 1σ, and the points are
notindependent duetoregularization.
shaded area at the boundary points has an elliptical form.
TheelementratiospredictedforSNIaandthechosenmod-
elsof SNII areshown bystraight lines. Inaddition weplot
a line corresponding to a 60 per cent contribution of SN Ia
to the iron mass, calculated using the T95 model for SN II
yields and theTNH93 model for SN Ia yields. Also marked
Figure 10. Ne, Si and S mass vs Fe mass in AWM7. Crosses
are the results of Mu96, which agree with our values well
represent ASCA SIS data from Mu96. Filledcircles indicate our
at comparable radii. Note that Si/Fe (and Ne/Fe) is lower
best-fitsolutionwithshadedregionsdenotingthe90percentcon-
than the Mu96 value at smaller radii, but continues to rise
fidenceintervals.Thelong-dashedlineshowsthetheoreticalmass
ratioforSNIayield(TNH93model),whilstSNIIyieldsfromthe towards SN II-likeratios at larger radii.
compilationofGibson etal.(1997) areshownbysolidanddot- An anomaly immediately apparent in Fig.10, is that
ted lines denoting a choice of T95 and W95 yields, respectively. whilstNeandSimassesshowsimilarbehaviourintheSNIa
(These coincide in the case of Si.) The dashed line represents a –SNIIreferenceframe,theS/Feratioisapproximatelycon-
60 per cent contribution of SN Ia to the iron enrichment, using stant through the cluster. The anomalous behaviour of the
T95yieldsforSNII. S/Fe ratio was also noted in integrated spectra by Mu96.
Wenotethatnotonlytheabsolutevalue,butalsothetrend
a substantial uncertainty in SN II yields between different in the S abundance is anomalous. The observed S/Fe ra-
models.First attemptstoconfronttheabundancemeasure- tio is consistent with 100 percent Feproduction by SNeIa
mentsinclusterswith theoreticalpredictionsweremadeby throughout the cluster, while from other elements it varies
Loewenstein&Mushotzky(1996).Theirconclusionwasthat from70percentdownto30percent,frominnertoouterra-
for some models one doesn’t need any SN Ia to reproduce dialpoints.SuchbehaviourappearstoimplythattheSNII
the observed element ratios. They also found that none of model yields for sulphur are incorrect or that the observed
the models they considered could self-consistently explain abundanceis modified bysome additional physical process.
all theelement ratios observed. Leavingsulphurasideasunreliable,wecanderiveinde-
TurningtoourobservationofAWM7–thisisoneofthe pendent estimates of the fraction of the iron which is con-
objectsstudiedbyLoewenstein&Mushotzky(1996,LM96), tributed by SN Ia from the Si/Fe and Ne/Fe ratios, using
based on the measurements of Mushotzky et al. (1996, a given pair of models for the SN Ia and SN II yields. This
Mu96), but we are able to build a more detailed picture, can be used as a test of the consistency of the models. In
sincewehaveresolvedtheradialelementdistributions.SNII Fig.11wecomparetheresultsforAWM7usingtheT95and
yieldsaretakenfromtherecentworkofGibson etal.(1997), W95modelsforSNII(theTNH93modelyieldsareadopted
where an attempt was made to put different theoretical for SN Ia in each case). Both models appear acceptable in
predictions onto the same mass grid for more direct com- this case. For comparison, the SN II model of Thielemann
parison. We use models T95 and W95(A,Z⊙), from Gibson et al. (1996; yields for Fe and Si are 0.11 and 0.08 in so-
etal.(1997),whicharebothintegratedoveraSalpeterIMF. lar masses), which LM96 note can give a reasonable match
SN Ia yields are subject to much less uncertainty, and are to most of their integrated abundances without any SN Ia
takenfromThielemann etal.(1993,TNH93).Averagestel- contribution, fails to explain the high Si/Fe ratio which we
lar yields for Ne, Mg, Si, S, Fe in solar masses are 0.232, observeat the outerradii in our AWM7 analysis.
0.118, 0.133, 0.040, 0.121 for the T95 model; 0.181, 0.065, Our data on the abundance ratios in the groups are
0.124, 0.058, 0.113 for the W95 model; 0.005, 0.009, 0.158, presented in Fig.12. Again, these ratios indicate significant
0.086, 0.744 for theTNH93 model. contributionsfrom both SNIaand SNII,but thevariation
In Fig.10 we consider the behaviour of the cumulative oftheelement ratios with radiusisless pronounced thanin
massesofα-elementsplottedagainstthecumulativemassof AWM7.TheroleofSNeIainironproductionisthestrongest
iron.Theshadedzonerepresentsthe90percentconfidence inNGC5044,wheretheyprovide75(54−98)percentofthe
areafordeviationfrom eitherparameter,sotheform ofthe ironwithintheregionanalyzed,whileinHCG62andHCG51
(cid:13)c 1999RAS,MNRAS000,1–14
Constraining the Role of SN Ia and SN II in Galaxy Groups 9
Figure 12. Mg and Si mass vs Fe mass in NGC5044 (left), HCG51 (center) and HCG62 (right). Line coding for the theoretical SNe
yieldsanddatarepresentationissimilartoFig.10.
from SN Ia and SN II. The situation is complicated by the
11 11
likelihood thattherewillalso besomecontamination ofthe
00..88 00..88 IGMbymetalsreleasedwithingalaxiesbystellarmassloss.
00..66 00..66 Assumingthatmost of themetals in theIGMareliberated
from early-type galaxies (Arnaud et al. 1992) the abun-
00..44 00..44
danceratiosofsuchwindswouldbesimilartothatofSNII,
00..22 00..22
sothiscontributionwill effectively beincludedintheSNII
00 00 componentinourdecomposition.Aswewill seeinthenext
--00..22 --00..22 section, the stellar wind contribution, if it can escape from
galaxies, may be significant, but should not be a dominant
--00..44 --00..44
component in theobserved mass of metals.
00 00..22 00..44 00..66 00..88 11 00 00..22 00..44 00..66 00..88 11
For our decomposition of the iron profile into the two
Figure 13. Input of SN Ia into the iron mass, calculated from major SN types, we use the Si and Fe abundance measure-
Si/FeandMg/Feratios,assumingT95(leftpanel)andW95(right ments and the yields of T95. This approach has theadvan-
panel)yieldsforSNII.SolidlinedenotesNGC5044data,dashed tages that Si and Fe are measured for all systems studied
–HCG51anddotted–HCG62.Errorbarsare1σ,andthepoints here, the Si abundance is the best constrained of the α-
arenotindependent duetoregularization. elements, and the spread in the Si/Fe yields between the
T95 model and the various models of W95 is small. This
the corresponding values are somewhat lower and amount provides more confidence in the derived results. In Fig.14
to 48 (10−100) and 68 (40−95) per cent, respectively. wepresenttheironprofileattributedtoSNII(upperpanel)
ConsistencybetweentheestimatesoftheSNIacontribution and SN Ia (lower panel) for all each of our systems apart
fromSi/FeandMg/FeissomewhatbetterfortheT95model from HCG51, which has essentially uniform abundances of
than for W95. But both models are acceptable within the all elements, as was shown earlier.
errors, as is illustrated in Fig.13. The main feature seen in Fig.14 is the strong central
TheneedforasignificantSNIacontributiontoexplain concentration of the SN Ia iron profile, in comparison to a
the abundance ratios in the systems we study has interest- flatterdistributionintheSNIIironcontribution.Notethat
ingimplications forthedebateon thechemicalevolution of thefact that theSi/Feratio varies significantly with radius
ellipticalgalaxies.Inparticular,oneoftheoptionsdiscussed in three of these systems implies the need for at least two
byArimoto et al. (1996) toaccount for theapparentlylow differentsourcesofenrichment,irrespectiveofthedetailsof
abundances seen in the hot gas within ellipticals, was that specific models for supernovayields.
the SN Ia rate has always been low. This option is ruled ToanalyzethecentralexcessintheSNeIacontribution,
out by our results, and also by the findings of Fukazawa wefittedaKingprofile.Thevaluesofthefittedcoreradiiare
et al. (1998). 60(27–121)kpcforHCG62,117(65–213) kpcforNGC5044
and 171 (104–262) kpc for AWM7, which compares with
theircorrespondingopticalcoreradiiof27,180and230kpc.
4.2 Iron profiles for SN Ia and SN II
We return to the broad similarity between these core radii
The iron abundance gradients seen in most of our systems below.
show that the IGM is not completely mixed. In addition IfitisassumedthatSNIIejectaescapeprimarilyfrom
toFe,abundancegradientsaredetectedforα-elements:Mg early-type galaxies, where they are generated by an initial
andSiinHCG62andNGC5044,andNe,SiandSinAWM7. massive burst of star formation accompanying the birth of
As discussed above in relation to AWM7, the balance be- thesegalaxies(e.g.Matthews1989),orbysubsequentbursts
tween the contribution of different SNe types could be a triggered by galaxy merging ( Schweizer & Seitzer 1992),
variable function of radius. To explore this further, we use then their injection will be closely associated with active
theratioofα-elementstoironasafunctionofradiustode- star formation, and take place via highly energetic winds.
compose the iron profile into radially varying contributions The fact that the SN II productsare so widely distributed,
(cid:13)c 1999RAS,MNRAS000,1–14
10 A. Finoguenov and T. J. Ponman
Figure14. Ironprofilesdecomposedintothetwomajortypesof Figure 15. Spatially resolved IMLR. The solid line repre-
SNe. Solidlines represent AWM7 data, dashed line– NGC5044, sents AWM7 data, long dashed line – NGC5044, short-dashed
dotted –HCG62. Results for HCG51 arenot shown since abun- –HCG51,dotted–HCG62.Errorsdonotexceed20percentand
dances shownosignificantvariations. arenegligiblecomparedtotrendsseenintheFigure.
4.3 Iron mass to light ratios
in contrast to the SN Ia ejecta, which are expected to be TheIMLRisacomparisonoftheamountofheavyelements
released over a much longer timescale after star formation, with optical luminosity, as an indicator of the ability of a
suggests that the bulk of the SN II activity occurred early system to synthesize and retain elements. Having resolved
in thelife of thesystem. This is consistent with thepicture the spatial behaviour of the elements, it is straightforward
of Ponman, Cannon & Navarro (1999), who find that most to compare it with the distribution of optical light, rather
ofthesupernovaenergyinjection mustactuallyprecedethe thananintegralvalue.Inthefollowing,wewillassumethat
collapse of groups and clusters in order to account for the thedistribution ofgalaxies inoursystemsisadequatelyde-
magnitude of therise seen in the entropyof thegas. scribedbyaKingprofile,characterized bysomecoreradius
The totallevel of FeattributedtoSNII productsindi- (Rc).Inadditionweseparatethecentralgalaxyfromthedis-
catestheabilityofthesystemtoretainSNII-drivenwinds. tributioninthecasesofAWM7andNGC5044,whereitcon-
Groups are characterized by a small spread around a value tributesasubstantialfraction ofthetotallight.Thismeans
of0.1solar,withaveragevaluesof0.07±0.02forNGC5044 effectively an addition of a central ”point” of light. Param-
and0.11±0.03 forHCG62, howevertheSNII contribution eterscharacterizingtheopticalluminosity,LB,cD,LB,RLB
in both systems drops below 0.05 solar at large radii. The and Rc, are presented in Table 1.
SNIIcontributiontothewellmixedIGMinHCG51isequal We normalize the distribution to the observed values
to 0.10±0.02, within theregion analyzed. (LB) within some radius (RLB) for AWM7 and NGC5044.
In AWM7, theSN IIcontribution totheFeabundance ForHicksongroupswerequiretheobservedblueluminosity
outside the core is reasonably flat, at a level of 0.13±0.01 (Hickson et al. 1992) to be reached at the virial radius of
solar. Inthecoreofthecluster,theSNIIcontributionrises eachsystem.AccordingtothestudyofArnaud etal.(1992)
byafactoroftwo.Thiscentralexcess,whichcanbecharac- the metal content in clusters is correlated with the total
terized by aKingprofile with a 55 (30–90) kpccore radius, blue luminosity of early-type member galaxies rather than
has some similarity with ASCA results for A1060 (Tamura with thelate-types.The valuesofLB,listed in Table1 and
et al. 1996), where a concentration of α-elements towards used in calculating the IMLR, therefore includes only the
the cluster centre was detected. Both clusters are nearby contributionfromearly-types.Inthecaseofcompactgroups
and it is quite probable that we are resolving the contri- there is good evidence that the core galaxies cataloged by
bution to the metals arising from stellar wind losses in the Hicksonareaccompaniedbyabroaderdistributionofmostly
centralcDgalaxy.ThefactthattheIMLRinthecentralre- fainter galaxies. This may increase the optical luminosity
gionofAWM7isbelowthelevelexpectedfrom stellarmass somewhat. InthecaseofHCG62, thestudyofDeCarvalho
loss(seenextsection)supportsthissuggestion.However,in et al. (1997) indicates that the optical luminosity we use
NGC5044, which also contains a dominant central galaxy, mightbeincreasedbyuptoonethird,ifalltheextragalaxies
nosuchexcessexistsbeyondtheinnerradiusof40kpcem- (which havenot been typed) were early-types.
ployed in ouranalysis. AcomparisonoftheintegratedIMLRintheIGM,plot-
(cid:13)c 1999RAS,MNRAS000,1–14