The X-ray Emission in Post-Merger Ellipticals

8 Pages·0.2 MB·English

by  Ewan O'Sullivan

Mon.Not.R.Astron.Soc.000,1–8(0000) Printed1February2008 (MNLaTEXstylefilev1.4)
The X–ray Emission in Post–Merger Ellipticals
Ewan O’Sullivan1, Duncan A. Forbes1,2 and Trevor J. Ponman1
1School of Physicsand Astronomy, University of Birmingham, Edgbaston, Birmingham B15 2TT
(E-mail: [email protected])
2Astrophysics & Supercomputing, Swinburne University,Hawthorn VIC 3122, Australia
1
0
0
2
1February2008

n
a
J ABSTRACT

6 The evolution in X–ray properties of early–type galaxies is largely unconstrained. In
1 particular, little is known about how, and if, remnants of mergers generate hot gas
halos. Here we examine the relationship between X–ray luminosity and galaxy age

1 for a sample of early–type galaxies.Comparing normalizedX–ray luminosity to three
v differentageindicatorswefindthatLX/LB increaseswithage,suggestinganincrease
1
in X–ray halo mass with time after a galaxy’s last major star-formationepisode. The
7
long-term nature of this trend, which appears to continue across the full age range of
2
our sample, poses a challenge for many models of hot halo formation. We conclude
1
that models involving a declining rate of type Ia supernovae, and a transition from
0
1 outflow to inflow of the gas originallylost by galactic stars,offers the most promising
0 explanation for the observed evolution in X-ray luminosity.
/
h Key words: galaxies: interactions – galaxies: elliptical and lenticular – galaxies:
p evolution – X-rays: galaxies
-
o
r
t
s
a
1 INTRODUCTION X–rayluminosityactuallydeclines.Forexample,NGC7252
:
v (aprimeexamplefortheremnantofamergerthatoccurred
i A potential problem with forming ellipticals from merging 0.7Gyrago)showssomeevidenceofahothalo,butonethat
X
spirals is how to account for the different gas properties in is much smaller than thehalos seen in typical ellipticals.
ar the two types of galaxies. In particular, spirals contain rel- Two post–merger ellipticals were studied in the X–ray
atively high masses of cold (T ∼ 100 K) gas, whereas el- by Fabbiano & Schweizer (1995). They found that neither
lipticals have very little. The opposite situation is true for galaxy had an extensivehothalo. Giventhecorrelation be-
the hot (T ∼ 106 K) gas masses – normal spirals contain tween X–ray luminosity and isophotal boxiness (thought to
rather little hot gas, while ellipticals may possess exten- be a signature of a past merger) demonstrated by Bender
sive hot halos. Since the total gas masses per unit stellar etal.(1989),thiswasasomewhatsurprisingresult.Ifpost–
mass are broadly comparable in early and late-type galax- merger ellipticals are to resemble ‘normal’ ellipticals they
ies,thisraisesthepossibilitythatinefficientmerger–induced needtoacquireahotgashalo.Possiblemechanismsinclude:
starformation mightheatthecoolinterstellargasinspirals 1) The late infall (i.e. after a few Gyr) of HI gas associated
to form the hot halo in post–merger ellipticals. Recently, withthetidaltails(Hibbardetal.1994).Thiscoldgasmay
Georgakakisetal.(2000)haveshownthatthecoldgasmass be shocked to X–ray emitting temperatures as it falls back
indeeddecreasesinan‘evolutionarysequence’frommerging intothe merger remnant.
spirals to post–merger ellipticals. 2) A reservoir of hot gas, possibly expelled at the nuclear
However, it is not clear that gas heated in an intense merger stage, might infall from large radii, where its low
starburstcanberetainedwithinthegalacticpotential.Read density makes it undetectable to current X–ray satellites.
& Ponman (1998) examined the X–ray properties of eight 3) After the initial violent starburst, the continued mass
on–going mergers placed in a chronological sequence. Their loss from stars and heating from stellar winds recreates a
studyrevealed that material is ejected from merging galax- hot ISM.
ies soon after thefirst encounter. Massive extensions of hot
gas are seen (involving up to 1010 M⊙) at the ultralumi- The first study to attempt an ‘evolutionary sequence’
nous peak of the interaction, as the two nuclei coalesce. for post–merger ellipticals was that of Mackie & Fabbiano
There is evidencefor thisin in the Antennae, Arp 220, and (1997).For32galaxiestheyshowedaweaktrendforLX/LB
NGC 2623. However, after this phase of peak activity, the to increase with a decreasing Σ parameter. The Σ parame-
(cid:13)c 0000RAS
2 Ewan O’Sullivan et al.
ter is defined by Schweizer & Seitzer (1992), and is a mea- 2.1 LX/LB versus Spectroscopic Age
sure of a galaxy’s fine structure, i.e. optical disturbance.
Fine structure only persists in dynamically young galaxies,
TheyshowedthatitcorrelateswithbluecoloursandBalmer
soFig.1providesinformation onlyontheearlygalaxy evo-
line strength, and is thus a rough indicator of dynamical
lution of these objects. A more general measure of age is
youth. The Mackie & Fabbiano trend therefore suggested
needed to show how X–ray properties evolve over a longer
that post–merger ellipticals became more X–ray luminous,
timescale. Galaxy spectroscopic ages are now available for
foragivenopticalluminosity,asthegalaxyaged.Theyalso
alarge numberofearly–typegalaxies from thecatalogue of
plotted LX/LB against Hβ absorption line EW, which re-
Terlevich & Forbes (2000). These ages are generally based
vealedasimilartrend.AmorerecentstudybySansometal.
on Hβ absorption line measurements and the stellar popu-
(2000) confirms the trend in X–ray overluminosity with Σ
lation modelsofWortheyetal.(1994). Thelineindexmea-
for 38 galaxies, and suggests that it might be explained by
surements come from the galaxies’ central regions, and are
the build up of hot gas in post–merger galaxies. However,
luminosity weighted. Thus they are dominated by the last
bothΣandHβ EWhavetheirdrawbacks–Σisonlysemi–
majorburstofstarformation,whichispresumablytriggered
quantitative at best and the strength of Hβ is affected by
by a major merger event. Although not a reliable absolute
both stellar age and metallicity.
measure of the age of each galaxy, these spectroscopic ages
Here we reexamine the trend seen by Mackie & Fab-
doprovideuswithamuchmoreusefulestimateoftheirages
biano (1997) for a larger sample, and investigate two new
measures of galaxy age: residual from the Fundamental relative to one another. In Fig. 2 we show LX/LB plotted
against spectroscopic age (from Terlevich & Forbes 2000).
Planeandspectroscopicage.Usingthesethreemeasureswe
explore the X–ray luminosity evolution of post–merger el- The plot shows a large degree of scatter, most notably
lipticals. in the age range between 4 and 10 Gyrs. This is likely to
Throughout the paper we assume H0 = 75 km s−1 be in part caused by the uncertainties in calculating ages,
Mpc−1 and normalise LB using the solar luminosity in the which we estimate lead a typical error of ∼20%. We also
B band,LB⊙ = 5.2×1032 erg s−1. expectamean1-σ errorinX–rayluminosityof∼15%.Typ-
ical error bounds for points at 1, 7 and 15 Gys are shown
inFig. 2bythediamondsat thetop oftheplot.Thegraph
also contains a relatively large number of upper limits, so
wehaveused thesurvivalanalysis packages available under
2 RESULTS IRAFtoassessthestrengthofanycorrelationandfitregres-
sionlines.TheCoxproportionalhazard,Spearman’srhoand
We first reexamine the trend of normalized X–ray lumi-
generalized Kendall’stau testsare usedtodeterminecorre-
nosity with Fine Structureparameter Σ, presented initially
lationstrength.Linearregression fittingiscarriedoutusing
by Mackie & Fabbiano (1997) for 32 galaxies. Here we
theexpectation and maximization (EM) algorithm and the
use a sample of 47 early–type galaxies with Σ taken from
Buckley–James (BJ) algorithm. In all cases we found that
Schweizer&Seitzer(1992), andnormalized LX valuesfrom
thetwo fittingalgorithms agreed closely.
our recent catalogue (O’Sullivan et al. 2000) of X–ray lu-
Using thefull sample of 77 early-typegalaxies, we find
minosities (mostly based on ROSAT data). The LX val-
uesareapproximately bolometric. TheLB valuesarebased aancdorargeela.tLioinneofiftasttloeatshte99sa.6m%plseigpnriofidcuanceceslboeptewseoefn0L.0X6/6L±B
wherever possible on BT magnitudes taken from Prugniel 0.022(EM)and0.063±0.025(BJ).Theformerisshownas
&Simien (1996). Iftheseareunavailable,valuesfrom NED
thedashed line in Fig. 2.
are used. Fig. 1 shows the sample of 47 early–type galaxies
Recent work on the X–ray properties of galaxies in
plotted in order of decreasing Σ or increasing age.
groups (Helsdon et al. 2000) lead us to suspect that some
AsnotedbyMackie&Fabbiano,thescatteratlowval-
ofthescatterseen inFig.2mightbecausedbyinclusion of
ues of Σ is large, covering around 2 orders of magnitude in
group or cluster dominant galaxies (marked as crossed cir-
LX/LB,whileathigherΣthescatterappearstobesmaller
cles)inoursample.Itisalsoimportanttonotethatthereis
and LX/LB limited to low values. A similar trend was seen
evidencetosuggestthatthelargestellipticalsmayhavenon–
by Sansom et al. (2000) for 38 galaxies. This suggests that
solarabundanceratios(e.g. Carollo etal.1993).Thiscould
dynamicallyyounggalaxieshavelowX–rayluminositiesand
lead totheirages beingunderestimated byasmall amount,
thatagingproducesarangeofluminosities,presumablyde-
again addingtothescatter.Wethereforeremovedfrom our
pendenton other factors.
sampleanumberofgalaxieslistedbyGarcia(1993)asgroup
In order to test the strength of this relation we apply
dominant or known cDs and retested the sample. This im-
Kendall’s K test to the X–ray detections from our sample.
proves the correlation slightly (to >99.75%) and makes the
The test does not assume any distribution in the data, and
line fit slightly shallower; 0.063 ± 0.019 (EM) and 0.061 ±
the K statistic is unit normal distributed when at least 10
0.022 (BJ). The EM fit is plotted as a solid line in Fig. 2.
data points are present. For our data we find K = -2.31054
(using 27 data points), indicating an anti correlation be- ToconfirmthattheincreaseinLX/LBoccursacrossthe
tween Log LX/LB and Σ of ∼2-σ significance. For com- range of ages, rather than just the first few Gyrs, we also
parison, we also apply the test to the sample of Sansom et binnedthesampleintosubsetsbyage.Theagerangeofeach
al. , and find a correlation of very similar significance, K = subset was chosen so as to have roughly equal numbers of
-2.28749 for 30 detections. detections in each bin. We then calculated a mean LX/LB
foreachbin,usingtheKaplan–Meierestimator.Theresults
are shown in Table 1, and as large crosses on Fig. 2. These
show the same trend for increasing LX/LB with age, and
(cid:13)c 0000RAS,MNRAS000,1–8
X–rays from Post–Mergers 3
32
31
30
29
10 8 6 4 2 0
Figure 1. NormalizedX–rayluminosityversusfinestructureparameter Σforearly–typegalaxies.Filledcirclesaredetections, arrows
denoteupperlimits.Dynamicallyyounggalaxies(highΣ)havelowLX/LB.
Age No.ofdetections No.ofupperlimits LogLX/LB error
(Gyrs) (ergs−1 L−B1⊙)
A<4 9 12 29.56 ±0.11
4≤A<6 6 11 29.83 ±0.15
6≤A<8 9 4 29.93 ±0.22
8≤A<10 9 6 30.15 ±0.14
A≥10 8 4 30.32 ±0.09
Table 1.MeanLX/LB valuesfor5agebins,calculatedusingtheKaplan–Meierestimator.
indicate that the trend is continuous across the age range distributethecorrespondingprobability).Whenthisoccurs,
covered by thedata. the Kaplan–Meier estimator treats the limit as a detection
TheKaplan–Meierestimator(Feigelson&Nelson1985) and hencemay bebiased.
includesupperlimitsbyconstructingaprobabilitydistribu- Of the five bins chosen, two (the youngest and second
tion function for the data in which the probability asso- oldest)haveupperlimitsastheirlowestvalues.Forthesec-
ciated with each upper limit is redistributed equally over ondoldestbin,theproblemdatavalueisonlyslightlylower
detected values which lie below the upper limit. So long as than a detected point, so the bias will be small. However,
there is no systematic difference between the detected and the youngest bin has two upper limits as its lowest points,
undetected systems, this should be a reasonable procedure. oneofwhichisconsiderablybelowthenearestdetection.To
A problem arises when the lowest point in a given bin isan check for bias, we recalculated a mean log LX/LB of 29.66
upperlimit(sincetherearenolowerpointsoverwhichtore- for this bin excluding these points, compared to 29.56 pre-
(cid:13)c 0000RAS,MNRAS000,1–8
4 Ewan O’Sullivan et al.
Figure2. Log(LX/LB)plottedagainstspectroscopicage.Crossedcirclesrepresentgrouporclusterdominantgalaxies,filledcirclesall
otherdetections. ArrowsdenoteupperlimitsandthefilledtrianglerepresentsNGC7252,ayoungpost–merger.Thethreediamondsat
thetopofthegraphshowtypicalageandLX/LB errorsforgalaxiesofage1,7and15Gys.Thetwolinesarefitstothedata,including
(dashedline)orexcluding(solidline)thegroupandclusterdominantgalaxies.ThelargecrossesrepresentmeanLX/LB forthefiveage
binsdescribedinthetext.ThedottedlineshowstheexpectedincreaseinLX/LB causedbythedecreaseofLB withage.Normalisation
ofthislineisarbitrary,andthepositionshowncanbeconsideredtorepresenta“worstcase”scenario.
viously. This indicates that although the youngest bin may tion,andthatthenewstarscreatedinthemergeralsohave
beslightly biased, it still supportsthe trend observed. solar metallicity. The relative mass ratios are 90% progeni-
tor stars and 10% new stars. The total B band luminosity
Asafurthercheckoftherobustnessofourconclusions,
fading of this composite stellar population is shown in Fig.
weexaminedtheLX/LB:Agerelation usingdetectedpoints
only,andfindacorrelationsignificantat∼2.1-σ.Theupper 2 (we assumed that the fading of the progenitor stars after
limits also show a correlation with age (at ∼2.3-σ). These reaching18Gyrsoldisinsignificant).Itcanbeseenthatthe
starburstfadesrapidlyinthefirstfewGyrs,roughlymatch-
trends cannot be selection effects in the data, since neither
sourcedistance,norX-rayexposuretimearecorrelatedwith ing the trend seen in LX/LB, but at later times there is
age. They therefore point to a genuine trend in the data. verylittlechange,whereastheobservedmeanLX/LB value
Galaxieswithlargerspectroscopicagestypicallyhavehigher continuestorise.SoalthoughtheLX/LB evolutionatearly
times could be simply driven by the fading starburst, this
LX/LB than youngerones.
cannot account for theevolution at late times.
Asthemerger–inducedstarburstfades,wewouldexpect
theblueluminosity todeclinewith time. Thiswill causean A further possibility to be checked is that our sample
increase in LX/LB, even for constant LX, which will com- might contain a trend in mean optical luminosity with age.
bine with any trends in LX which are present. With this Since LX/LB has been widely reported (see however Hels-
is mind, we show in Fig. 2 the expected change due to a don et al. (2000)) to rise with LB, this could lead to a con-
fading starburst from stellar population models (Worthey sequent correlation between age and LX/LB. To test this,
1994). Here we have crudely assumed that the progenitor we show in Fig. 3 a plot of log LB with measured age. An
galaxies consisted of a solar metallicity, 15 Gyr old popula- anti–correlationbetweenLB withagecouldproducetheob-
(cid:13)c 0000RAS,MNRAS000,1–8
X–rays from Post–Mergers 5
servedtrendinLX/LB.Fig. 3doesnotappeartoshowany a merger by the shock heating or photoionization of cool
such trend, and in order to test for a correlation, we apply gas from the progenitor galaxies. In some ongoing mergers
Kendall’sKtesttothedata.WefindaKstatistic(whichis the tidal tails are thought to contain up to half the Hi gas
unit normal distributed) of ∼0.44 (for 77 galaxies), and ex- originally present in the progenitor galaxies (Hibbard et al.
cluding group dominant galaxies reduces this to ∼0.33 (for 1994).Whenthisgasfalls backintothebodyofthegalaxy,
65 galaxies). This result indicates that there is no trend in shockheatingshouldbecapableofheatingittoX–raytem-
the data, even at the 1σ level. We therefore conclude that peratures.Analternativeisthatthetemperatureofthegas
our sample does not show an anti–correlation of LB with iscausedbyheatinginthestarburstphaseofthemerger.In
age,andthattheLX/LB trendwithagereflectsanincreas- thiscase thehot gas would beblown out tolarge radii, but
ing X–ray luminosity. might be contained by the dark matter halo of the galaxy
(Mathews & Brighenti 1998). At these large radii it would
betoodiffusetobedetected,butwouldeventuallyfallback
2.2 LX/LB versus Fundamental Plane Residual intothe galaxy, forming theobserved X–ray halo.
As further evidence of X–ray evolution with age we have In terms of available masses of gas, both these mod-
compared LX/LB to residual from the Fundamental Plane els appear to be viable formation processes. Bregman et al.
(FP). Work by Forbes et al. (1998) has shown that galaxy (1992)derivedX–rayandcoldgasmassesforalargesample
ageappearstobea“fourthparameter”affectingtheposition of early–type galaxies, finding MX of between ∼ 106 and
of a galaxy relative to the FP plane. Galaxies below the 1011 M⊙. The mean X–ray gas mass was a few 109 M⊙.
plane (negative residual) are generally young, while older On the other hand, very few detections of Hi were made
objects lie on or above the plane (positive residual). Fig. 4 in early–type galaxies, although Sa galaxies were found to
shows212galaxies withFundamentalPlaneresidualstaken containbetween107and∼5×1010 M⊙ ofHi .StudiesofHi
from Prugniel & Simien (1996). masses in later–type galaxies (Huchtmeier & Richter 1989;
Aswith Figs. 1and 2,thisshowsatrendofincreasing Huchtmeier & Richter 1988) give an average of a few 109
LX/LB with age. For the complete data set, the correla- M⊙,dependingon therangeofmorphologies chosen.Given
tion strength is >99.98%. However, only a small numberof thatlargeellipticalgalaxiesarelikelytohavebeenformedby
pointslieoutsidetherange–0.5<FPresidual<0.5.Theseare themergerofmanysmallerspiralgalaxies,theseresultssug-
unlikely to be representative of the general population, but gestthatfairlymodestconversionefficienciescouldproduce
will havea strong influenceon thestatistical tests. Exclud- theexpectedX–rayhalogasmasses eitherfrominfalling Hi
ing them lowers the correlation strength to ∼99.95%, with or via starburst heating and later infall.
a slope of 1.39 ± 0.40 (EM). This line is shown in Fig. 4. However, the infalling Hi model fails to explain the
Again the trend shown in Fig. 4 is one of increasing X–ray timescale over which we observe the LX/LB trend. In the
luminosity, relative to theoptical, as a galaxy evolves. well known merging galaxy NGC 7252, 50% of thecold gas
currently in the tidal tails is expected to fall back into the
main body of thegalaxy within thenext2–3 Gyr(Hibbard
& van Gorkom 1996). This suggests that we should expect
3 DISCUSSION
to see a rapid build up of X–ray emission in the first few
Intheprevioussectionwehaveusedthreeageestimatorsto gigayears after merger. This increase should be made even
show a strong correlation between normalized X–ray lumi- more noticeable by the fading of the stellar population, as
nosityandgalaxy age. TheX–rayemission from early–type thisistheperiodduringwhichLB isdroppingmostquickly.
galaxies is known to be produced by sources of two main Rapidgeneration oftheX–ray haloisinconsistent with our
types; discrete sources such as X–ray binaries, and hot gas. results.
The contribution of discrete sources has been shown to be Amodelinvolvinginfallofhotgasislikelytohavesim-
important in low luminosity early-type galaxies (Fabbiano ilar problems. It is difficult to see how gas falling back into
et al. 1994; Irwin & Sarazin 1998; Sarazin et al. 2000) and thebodyofthegalaxycandosoatthesteadyrateneededto
inlate-typegalaxies,wheretheyaregenerallythedominant produceasmoothincreaseinLX/LB.Infallmaybedelayed,
source of X–ray emission. On the other hand, most early– as thehot gas reaches larger radii than the cold, but it will
typegalaxieshaveastronghotgascomponent,andmassive still occur on a galaxy dynamical timescale. Models involv-
ellipticalsarecertainlydominatedbyhotgasemission(Mat- ing cooling and infall of gas onto central dominant galaxies
sushita et al. 2000; Matsushita 2000). As the contribution in groups or clusters (Brighenti & Mathews 1999) may be
of hot gas is known to vary a great deal, while the contri- abletoproduceaninflowoveralongerperiod,buttheseare
bution from discrete sources is generally believed to scale unlikely to be generally applicable. Group dominant galax-
with LB (Matsushita et al. 2000; Fabbiano et al. 1992) , it ies have exceptionally large dark halos, making them more
seems likely that the trend we observe in LX/LB is caused abletoretain (oraccrete) hot gas. Theyalso lieat thebot-
bychangesinthehotgascontentoftheseobjects withage. tom of a group potential well, and may be at the centre of
Wenowgoontodiscusspossiblehothaloformationmecha- agroupX–rayhalo, providingthem withan extrareservoir
nisms which may beresponsible for this trend of increasing ofhotgas.Theseconditionsdonotapplytothemajorityof
X–ray gas content with age. early-typegalaxies.Thereforewesuggestthatinfallof(cold
or hot) gas is unlikely to be the dominant process in the
generation of X–ray halos in typicalellipticals.
3.1 Gas infall
It has been suggested (Hibbard & van Gorkom 1996) that
the hot gas in elliptical galaxies may be produced during
(cid:13)c 0000RAS,MNRAS000,1–8
6 Ewan O’Sullivan et al.
11
10.5
10
9.5
9
0 5 10 15
Figure 3. Optical luminosity versus spectroscopic age. Crossed circles represent cluster or group dominant galaxies and filled circles
all other detections. The triangle is the recent merger remnant NGC 7252. The cE galaxy, M32, has been excluded from the plot and
associatedstatisticaltests.
3.2 Ongoing stellar mass–loss and galaxy winds ulations(e.g.Cappellaroetal.1999)haverevisedtheclassic
Tammann(1982)valueoftherateinoldstellarpopulations
Sinceinfallofeitherhotorcoldgasseemsunabletoprovide
down by a factor 3-4, and the evolution of this rate with
thesortoflongtermtrendinLX/LB seeninFig.2,wenow population age is very model-dependent. Given our contin-
looktomodelsinvolvinggasgeneratedbymasslossfromthe
uing ignorance about the precise nature of SNIa, it must
galactic stars. This source of gas is fairly well understood,
therefore be regarded as largely unknown.
and can provide the sort of gas masses required to account
for the halos of many early-type galaxies (though proba- The specific energy of the injected gas determines
bly not the very brightest – Brighenti & Mathews 1998), whethergasisretainedinahydrostatichothalo,orescapes
provided that the gas is retained within thegalactic poten- the galaxy as a wind. The way in which the specific energy
tial.Twomaintimescalesareinvolved.Oneisthemassloss evolves, depends upon the evolution of the SNIa rate rela-
ratefromstars,primarilygiantstars.Thesecondistherate tive to the stellar wind loss rate. There are therefore two
of type Ia supernovae (SNIa) which provide the main heat fundamentally different classes of models: those in which
source (after the brief SNII phase). These two factors, and the SNIa rate is constant, or declines more slowly than the
theinterplaybetweenthem,canpotentiallyleadtochanges stellar mass loss rate (e.g. Loewenstein & Mathews 1987;
in the hot gas content of elliptical galaxies on timescales David et al. 1991) and those in which the supernova rate
muchlongerthanadynamicaltime,andsohavethepoten- dropsmorequicklythanthegasinjection rate(Ciotti etal.
tial to explain the slow trend we observe. The stellar mass 1991; Pelegrini & Ciotti 1998). Broadly speaking, when the
lossrateisfairlywellunderstood(e.g.Ciottietal.1991),and specific energy of the injected gas is smaller than its grav-
forasingle-agedstellarpopulationitdeclinesapproximately itational binding energy it will be retained in a hot halo.
as t−1.3. However, the SNIa heating rate, which dominates It may subsequently cool in a luminous cooling flow. How-
theeffectivespecificenergyoftheinjectedgas,ishighlyun- ever, if the specific energy of the gas substantially exceeds
certain.RecentestimatesoftheSNIarateinoldstellarpop- its binding energy it will escape from the galaxy in a fast
(cid:13)c 0000RAS,MNRAS000,1–8
X–rays from Post–Mergers 7
32
31
30
29
-0.8 -0.6 -0.4 -0.2 0 0.2 0.4
Figure 4. X–ray luminosity versus Fundamental Plane residual for early–type galaxies. Filledcircles represent detections and arrows
upperlimits.Thesolidlineisafittoallpointswith–0.5<FPresidual<0.5
wind, which results in a much lower X–ray luminosity. The contrast, thespectroscopic age of ourgalaxies probably de-
first class of models, in which the specific energy rises with notes the time since a merger-induced starburst involving
time, therefore produces an evolution from an early inflow onlyafewpercentofthestellarmass.Inthiscase,theearly
phase to a later wind phase, predicting a declining X-ray stellar mass loss rate will be much lower than in the Ciotti
luminosity with time. Such models are clearly inconsistent etal models, andtheX-rayluminosity correspondingly less
with our results. (scaling approximately as thesquare of themass loss rate).
The main change in LX will be a rise associated with the
ModelsinwhichtheSNIaratedropsmorequicklythan
t−1.3 evolve, in broad terms, from a low luminosity wind slowingwindandtransformation intoahydrostatichalo.In
addition, as shown in Fig. 2, the decrease in optical lumi-
phase, towards a more luminous halo/cooling flow phase.
nosity as the starburst population fades will produce a rise
The timescale for this evolution depends upon a variety of
in LX/LB overthe first 1-2 Gyr.
factors, suchasthedepthandshapeofthepotential,butit
is typically ∼ 10 Gyr (Ciotti et al. 1991). In fact, none of
the models studied to date shows the rather simple mono-
tonic rise in LX with age which we observe. The models of
4 CONCLUDING REMARKS
Ciotti et al. (1991) predict an initial luminous phase, when
thegaslossratefromstarsisveryhigh.TheX–rayluminos- We have examined the evolution of the X–ray properties
ity thendeclines as thewind loss rate drops, and thenrises ofearly–typegalaxies. Forthreegalaxyageestimators(fine
again during the transition to a bound hydrostatic halo, at structureparameter,FundamentalPlaneresidualsandspec-
which point the X-rayluminosity is essentially equal to the troscopic ages) we findthat thenormalized X–ray luminos-
SNIa luminosity. This whole development describes a fairly ity evolves with time. In particular, the X–ray luminosity,
symmetrical dip and rise, lasting ∼ 5−15 Gyr, which is which reflects the mass of hot halo gas, appears to increase
again not what we seein thedata.However,thismodel de- at a steady rate over ∼ 10 Gyrs.
scribes a galaxy in which all stars are formed at t = 0. In Comparing the long term trend in LX/LB which we
(cid:13)c 0000RAS,MNRAS000,1–8
8 Ewan O’Sullivan et al.
observewithexpectationsfrompossiblemechanismsforhot Hibbard J. E., Guhathakurta P., van Gorkom J. H.,
haloformation,weconcludethattheonlyviablemechanism Schweizer F., 1994, AJ, 107, 67
appears to be the slow evolution from an outflowing wind
to hydrostatic halo phase driven by a declining SNIa rate. HuchtmeierW.K., Richter O.-G., 1988, A&A,203, 237
Infalling gas seems unlikely to be the main cause of such
HuchtmeierW.K., Richter O.-G., 1989, A&A,210, 1
a long term trend. Our results suggest that some of the
scatter seen in the global LX versus LB relation is due to Irwin J., Sarazin C., 1998, ApJ,494, L33
the evolutionary state, and past merger history, of early–
typegalaxies. Loewenstein M., Mathews W. G., 1987, ApJ,319, 614
Mackie G., Fabbiano G., 1997, in Arnaboldi M.,
Acknowledgements
Da Costa G. S., Saha P., eds, ASP Conf. Ser. 116:
We would like to thank R. Brown for early work on this
TheNatureofEllipticalGalaxies;2ndStromloSym-
project and S. Helsdon, W. Mathews and A. Renzini for
posium. p.401
useful discussions. We also thank an anonymous referee for
several suggestions which have improved the paper. This
MathewsW.G.,BrighentiF.,1998,inASPConf.Ser. 136:
research has made use of the NASA/IPAC Extragalactic
Galactic Halos. p. 277
Database (NED) which is operated by the Jet Propulsion
Laboratory, California Institute of Technology, under con- Matsushita K., Ohashi T., Makishima K., 2000, PASJ, 52,
tract with the National Aeronautics and Space Adminis- 685
tration. E. O’S. acknowledges the receipt of a PPARC stu-
dentship. Matsushita K., 2000, preprint,astro-ph/0009276
O’Sullivan E., Forbes D. A., Ponman T. J., 2000, in prepa-
ration
REFERENCES
Pelegrini S., Ciotti L., 1998, A&A,333, 433
Bender R., Surma P., Dobereiner S., Mollenhoff C., Made- Prugniel P., Simien F., 1996, A&A,309, 749
jsky R.,1989, A&A,217, 35
Read A. M., Ponman T. J., 1998, MNRAS,297, 143
Bregman J. N.,HoggD.E.,RobertsM. S.,1992, ApJ,387,
SansomA.E.,HibbardJ.,SchweizerF.,2000,AJ,120,1946
484
Sarazin C. L., Irwin J. A., Bregman J. N., 2000, preprint,
Brighenti F., Mathews W. G., 1998, ApJ, 495, 239
astro-ph/0009448
Brighenti F., Mathews W. G., 1999, ApJ, 512, 65
Schweizer F., Seitzer P., 1992, AJ, 104, 1039
Cappellaro E., Evans R.,Turatto M., 1999, A&A,351, 459
TammannG.A. inReesM.,StonehamR.,eds,Supernovae:
CarolloC.M.,DanzigerI.J.,BusonL.,1993,MNRAS,265, A Survey of Current Research, Dordrecht: Reidel,
553 1982
Ciotti L., D’Ercole A., Pelegrini S., Renzini A., 1991, ApJ, Terlevich A.I., Forbes D. A., 2000, MNRAS,submitted
376, 380
WortheyG.,FaberS.M.,GonzalezJ.J.,BursteinD.,1994,
David L. P., Forman W., Jones C., 1991, ApJ,369, 121 ApJS,94, 687
Fabbiano G., Schweizer F., 1995, ApJ, 447, 572
FabbianoG.,KimD.W.,TrinchieriG.,1992,ApJS,80,531
Fabbiano G., Kim D.-W., Trinchieri G., 1994, ApJ, 429, 94
Feigelson E. D., Nelson P. I., 1985, ApJ, 293, 192
Forbes D., Ponman T., Brown R., 1998, ApJ, 508, L43
Garcia A. M., 1993, A&AS,100, 47
Georgakakis A., ForbesD. A., Norris R.P., 2000, MNRAS,
318, 124
Helsdon S. F., Ponman T. J., O’Sullivan E., Forbes D. A.,
2000, MNRAS,submitted
Hibbard J. E., van Gorkom J. H., 1996, AJ, 111, 655
(cid:13)c 0000RAS,MNRAS000,1–8