The Morphology of the Sub-Giant Branch and Red Clump Reveal No Sign of Age Spreads in Intermediate Age Clusters

2.8 MB·English

by  Nate Bastian

Mon.Not.R.Astron.Soc.000,1–11(2014) Printed19January2015 (MNLATEXstylefilev2.2)
The Morphology of the Sub-Giant Branch and Red Clump
Reveal No Sign of Age Spreads in Intermediate Age
Clusters
N. Bastian1 & F. Niederhofer2,3
5
1 1 Astrophysics Research Institute, Liverpool John Moores University,146 Brownlow Hill, Liverpool L35RF, UK
0 2 European Southern Observatory, Karl-Schwarzschild-Straße 2, D-85748 Garching bei Mu¨nchen, Germany
2 3 Universita¨ts-SternwarteMu¨nchen, Scheinerstraße 1, D-81679 Mu¨nchen, Germany

n
a
J

5 Accepted. Received; inoriginalform
1

] ABSTRACT
A
G Arecentsurpriseinstellarclusterresearch,madepossiblethroughtheprecisionof
. HubbleSpaceTelescopephotometry,wasthatsomeintermediateage(1−2Gyr)clus-
h
ters in the Large and Small Magellanic Clouds have main sequence turn-off (MSTO)
p
widths thatare significantlybroaderthan wouldbe expectedfor a simple stellar pop-
-
o ulation (SSP). One interpretation of these extended MSTOs (eMSTOs) is that age
r spreads of the order of ∼ 500 Myr exist within the clusters, radically redefining our
t
s view of stellar clusters, which are traditionally thought of as single age, single metal-
a licity stellarpopulations.Herewe testthis interpretationbystudying otherregionsof
[
the CMD that should also be affected by such large age spreads,namely the width of

1 the sub-giant branch (SGB) and the red clump (RC). We study two massive clusters
v intheLMCthatdisplaytheeMSTOphenomenon(NGC1806&NGC1846)andshow
5 that both have SGB and RC morphologies that are in conflict with expectations if
3 largeagespreadsexistwithin the clusters.We concludethat the SGB andRC widths
8 are inconsistent with extended star-formation histories within these clusters, hence
3
age spreads are not likely to be the cause of the eMSTO phenomenon. Our results
0
are in agreement with recent studies that also have cast doubt on whether large age
.
1 spreads can exist in massive clusters; namely the failure to find age spreads in young
0 massiveclusters,alackofgas/dustdetectedwithinmassiveclusters,andhomogeneous
5 abundances within clusters that exhibit the eMSTO phenomenon.
1
: Key words: galaxies - star clusters, Galaxy - globular clusters
v
i
X
r
a
1 INTRODUCTION magnitudediagram (CMD) for NGC 1846, an intermediate
age(1−2Gyr)massive(∼105M⊙)clusterintheLMC,and
A major open topic in stellar cluster research is whether
foundthatthemainsequenceturn-offregionwasnotwellde-
clusterscanhostmultiplestar-formationeventsorextended
scribedbyasimplestellarpopulation(SSP),butratherwas
star-formationepochs.Youngmassiveclusters(YMCs)have
extended. The authors interpreted this spread as a spread
stellar populations that are well reproduced as single burst
in age of ∼ 300 Myr and postulated that NGC 1846 was
events, with upper limits on potential age spreads of 1−
the result of a rare cluster merger, where the two clusters
3 Myr (cf. Longmore et al. 2014). Globular clusters, on the
had a ∼300 Myr age difference. However, subsequent work
otherhand,hostchemicalabundanceanomalieswithintheir
hasshownthatthemajorityofintermediateage(1−2Gyr)
stellarpopulations(e.g.,Grattonetal.2012),andwhileage
clusters in the LMC and SMC display this extended main-
spreadscannot bediscerned attheseold ages, somescenar-
sequenceturn-off(eMSTO-e.g.,Mackeyetal.2008;Milone
ios to explain the anomalies invoke multiple star-forming
etal.2009; Goudfrooij etal.2011a,b; Kelleretal.2012; Gi-
epochs, spanning tens to hundreds of Myr (e.g., D’Ercole
rardietal.2013),rulingoutrareeventslikeunequalageclus-
etal.2008;Conroy &Spergel2011), althoughotherscenar-
ter mergers. If interpreted as being caused by age spreads,
ios have little or no age spreads (e.g., Krause et al. 2013;
these results suggest, rather, that star-formation can pro-
Bastian et al. 2013b).
ceed for 300−600 Myr within massive clusters, and that
Mackey & Broby Nielsen (2007) constructed a colour-
(cid:13)c 2014RAS
2 Bastian & Niederhofer
age spreads of this magnitude should be a common feature explanation(e.g.,Correntietal.2014)1.Onewaytotestthe
in massive clusters. abovescenariosistolookforotherfeaturesintheCMDthat
mayalsoreflectagespreads.Girardietal.(2009)introduced
Such age spreads have been searched for in younger
the red clump as such a test, showing that the red clump
massive clusters, where such spreads should be readily ap-
in NGC 419 is more elongated than would be expected in
parent.Larsenetal.(2011)studiedtheresolvedCMDsofsix
a simple stellar population. Interestingly, Trumpler 20 also
massive (105−106M⊙) clusters in nearby galaxies, and did displays an extended or “dual red clump” although it does
not findevidencefor age spreadsof more than20−30 Myr
not show any evidence for an age spread within it (Platais
(and being consistent with an SSP in most cases). Bastian
et al. 2012). Li et al. (2014a) found, on the other hand,
& Silva-Villa (2013) looked at the CMDs of NGC 1866 and
thattheyoungereMSTOclusterNGC1831, hasacompact
1856,twoclusterswithmassesof∼105M⊙,andagesof180 red-clump which is inconsistent with the presence of a sig-
and280Myr,respectively.Noagespreadswerefound,with
nificantagespreadwithinthecluster.Itiscurrentlyunclear
upper limits of any potential spreads of ∼ 30 Myr. Nieder-
whattheroleofrotationwill beontheredclumpmorphol-
hoferetal.(2015)extendedtheBastian&Silva-Villasurvey
ogy.Extra-mixingfromrotationwouldlikelygivelargercore
to an additional eight LMC young massive clusters, includ-
masses,whichmayresultinanextendedredclump,however
ingNGC1850,a∼2×105,100Myrcluster,andalsodidnot
this effect may be compensated by a longer main sequence
findanyevidenceforagespreads.Itisimportanttonotethat
lifetime. Detailed modelling is necessary to address this.
the YMCs studied so far have current masses (and escape
Another feature of the CMD that is sensitive to age
velocities)wellinexcessoftheintermediateageclustersthat
spreadsisthewidthofthesub-giantbranch(e.g.,Marinoet
showeMSTOs.Hence,unlessextremeclustermasslossisin-
al. 2012). Liet al. (2014b) haveshown that theSGB width
voked, the YMCs have the same properties (or even higher
inNGC1651,anintermediateageLMCclusterthatdisplays
massandescape velocities) astheintermediateage clusters
theeMSTOphenomenon,isconsistentwithasimplestellar
that display the eMSTO features did when they were at a
population (i.e., with no age spread). The authors place an
similar age.Finally,Bastian etal.(2013a) examinedthein-
upperlimit of 80Myronanyagespread withinthecluster,
tegrated spectra (or resolved CMDs) of ∼ 130 YMCs with
significantly less than the age spread of 450 Myr inferred
ages between 10−1000 Myr and masses of 104 −108M⊙, from the eMSTO. While the SGB width does suffer from
looking for evidence of ongoing star-formation, and none
lowernumbersofstarsthantheMSTOorredclumpregions
were found. These studies have called into question the in-
of the CMDs, a number of clusters contain enough stars to
terpretationoftheeMSTOfeaturebeingduetoagespreads.
place strict limits on any age spreads present within the
SinceagespreadsarenotseeninYMCs,alternativeex- cluster.
planations for the eMSTO have been put forward. Bastian Here we use the same technique as Li et al. (2014b),
&deMink(2009) suggestedthatstellarrotation maycause studying the width of the SGB in two intermediate age
the eMSTO feature, as stellar rotation effects the struc- LMCclustersthatdisplaytheeMSTOphenomenon,namely,
ture of the star and the inclination angle of the star rela- NGC1806&NGC1846.In§2wepresenttheobservations,
tive to the observer will change the effective temperature, whilein§3weanalysetheCMDsofthetwoclusters,testing
henceobservedcolour.However,Girardietal.(2011)calcu- thehypothesisofsignificantagespreadswithintheclusters.
latedisochronesforamodestlyrotatingstarandfoundthat Finally, in § 4 we discuss our results and present our con-
theextended lifetime of thestar, dueto rotation, may can- clusions.
cel the stellar structure effects, resulting in a non-extended
MSTO. Platais et al. (2012) found that in theintermediate
age (∼ 1.3 Gyr) Galactic cluster Trumpler 20, that mod-
2 OBSERVATIONS AND MODELS
estlyrotatingstarswereblueshiftedfromthenominalmain
sequence, in agreement with the Girardi et al. (2011) pre- ForthepresentworkweuseopticalHubbleSpaceTelescope
dictions, suggesting that stellar rotation may not be the (HST) imaging taken with the Advanced Camera for Sur-
cause of the eMSTO phenomenon. However, no rapid ro- veys (ACS) Wide Field Camera (WFC) from programmes
tating stars were present in the cluster, and further studies GO-9891(PIGilmore) andGO-10595 (PIGoudfrooij). The
are needed to confirm thisas a general result. observations, reduction, and photometry are all described
in detail in Milone et al. (2009). We use the field-star sub-
Yang et al. (2013) used the Dartmouth stellar evolu-
tracted (based on a suitable nearby reference field) cata-
tionary code to investigate these contradictory results, and
logues for our analysis, and refer the interested reader to
foundthat themixingefficiency caused byrotation isakey
Miloneetal.(2009)fordetailsonthisprocess.Wehavecor-
parameter, and that in fact stellar rotation may cause eM-
rectedtheobservedphotometry forextinction,usingvalues
STOs, depending on the specific choice of the rotational
ofAV =0.05and0.08anddistancemoduliof18.5and18.42
mixingparameter.Miloneetal.(2013)foundthatwhilethe
young (∼ 150 Myr) cluster NGC 1844 in the LMC did not
havean age spread, its CMD does display features that are 1 Althoughnote that these scenarios use“correction factors” to
not well described by an SSP model, namely a broadened translatetheobservedpresentdaymassesandradiioftheclusters
main sequence at similar magnitudes as the eMSTO in the to the initial values. These factors invoke extreme cluster mass
intermediate age clusters. Hence, as CMDs are studied in loss, causing the intermediate age clusters to have been signifi-
more detail, new features are still being discovered. cantlymoremassiveinthepast.Themodelsthatareusedforthis
(D’Ercoleetal.2008)weremadeforinnerMilkyWaytidalfields,
Hence, the cause of the eMSTO is still under debate, i.e. they were not meant for the LMC/SMC and hence would
with some studies still advocating age spreads as the best predictsignificantlysmallermasslossinthesesmallergalaxies.
(cid:13)c 2014RAS,MNRAS000,1–11
eMSTOs Are Not Caused By Age Spreads 3
for NGC 1806 and 1846 (Goudfrooij et al. 2011b), respec- age distribution is shown in Fig. 2. We then fit a gaussian
tively.ThevaluesweretakenfromGoudfrooijetal.(2011b), tothedistribution(shownasabluedash-dottedline)tothe
although independent analysis using the present datasets data, and the best fit parameters are given in the panel.
found good agreement. We will use this age distribution in constructing synthetic
For our analysis we adopt the stellar isochrones of CMDs to test if other regions of the CMD are consistent
Marigo et al. (2008), with Z=0.008 and Y=0.25. However, with theestimated SFH.
we note that the conclusions of this paper are unchanged if We then generated two synthetic cluster CMDs. The
wewould haveadopted theBaSTI (Pietrinferniet al.2004) first is the expected distribution if the eMSTO was due to
or theDartmouth (Dotter et al. 2008) isochrones with sim- an age spread (i.e. adopting the age spread in Fig. 2). The
ilar parameters. While the absolute ages of the clusters do secondwasundertheassumptionofasingleagepopulation
depend on the model adopted, the relations between the (i.e.,anSSP)withanageof1.44Gyr.Syntheticstellarpop-
MSTO,SGBand RCpositions remain unchanged,i.e. ifan ulations were then made adopting a Salpeter (1955) stellar
agespreadispresentitwillshowupinthesamewayineach IMF and theMarigo et al. (2008) isochrones, applying rep-
of the different portions of theCMD. resentative errors taken from the observations. Stars along
the same colour/magnitude region used to define the SGB
and the RC in the observations were selected and analysed
in the same way as theobservations.
3 RESULTS
In Appendix A we test all the used techniques against
3.1 NGC 1806 synthetic clusters with known (input) age spreads. Addi-
tionally,wetestourrelativelysimplisticagespreadestimate
Fig.1showstheCMDofNGC1806,centredontheMSTO,
based on the cut across the eMSTO (note that this is the
SGB, and red clump portions of the distribution. (Blue)
sametechniqueadoptedbyGoudfrooij etal.(2011b;2014))
lines show isochrones of age 1.41 (top), 1.58 (middle) and
against a full SFH based on the STARFISH package.
1.86 Gyr (bottom), i.e. the range required to explain the
observed spread in the MSTO region. Some of the spread
observed in the MSTO region is due to binaries and pho-
tometric errors, hence the isochrones used to bracket the
eMSTO are upper limits to any actual spread within the
clusters. It should be noted that different authors have de- 3.1.3 The morphology of the Sub-Giant Branch (SGB)
rived somewhat conflicting age spreads, with Goudfrooij
For the SGB, we can quantify the offset between the ob-
et al. (2011b) suggesting age spreads of ∼ 500 Myr for
served distribution and that expected from a population
NGC1806andMiloneetal.(2009)suggestingalowervalue
with an extended SFH or a single burst. This is done by
of ∼200 Myr. We identify candidate SGB branch stars be-
measuring the magnitude offset of each observed star (or a
tween 1.06F435W −F814W 61.6 as (red) diamonds.
starfromoursyntheticCMD)identifiedasbeingpartofthe
As can already be seen from the figure, the SGB stars
SGB from a nominal isochrone.
clusteraroundtheyoungestisochroneneededtoexplainthe
We adopt an isochrone at 1.44 Gyr, as this best de-
eMSTO feature, and do not span the full range expected if
scribes the observed SGB in this cluster, and looked at the
large age spreads were present in thecluster.
differencebetweeneachobservedSGBstarandthenominal
isochrone. The results are shown in Fig. 3. We also show
3.1.1 Photometric Errors theexpectedmagnitudedifferencebetweenotherisochrones
andthenominal isochroneasvertical dashedlines(labelled
The average photometric errors were derived from the ob-
in the panel).
served main sequence distribution below the MSTO. To
As can be seen in Fig. 3, theextended SFH simulation
do this we “verticalised” the main sequence between 21 6
(the average of 1,000 monte carlo realisations) fails to re-
mF435W 6 23.5. We then made histograms of the stars in produce both the peak of the observed distribution as well
(F435W −F814W)vertical in 0.25 magnitude bins, and fit as the width. The expected and observed peaks in the dis-
gaussians to thedistribution. While there was clearly a tail tribution are offset by ∼0.2 mag. The SSP simulation well
of stars to the red (due to binaries) the core of each distri-
describes the width of the core of the observed histogram,
bution was well described by a gaussian with dispersion of
but the observations display extended tails on both sides,
0.028 mag. We took this to be the typical error in colour
suggestingthatphotometricerrorsalonecannotexplainthe
and correspondingly an error of 0.02 mag in brightness.
fulldistribution.However,wenotethattheobservedspread
inthedistribution(correctedforphotometricuncertainties)
is an upper limit to the actual spread, as we have not cor-
3.1.2 Inferred Age Spreads from the MSTO and Synthetic
rected for the effects of differential extinction, nor binarity,
Cluster CMDs
in our sample. Additionally, if stellar rotation is affecting
We begin by estimating the age spread present as inferred theMSTOthenwemayexpecttoseesomeinfluenceonthe
bytheextendedMSTO.Todothis,wetakeaslice through SGB, which causes some spread in the observed SGB stars
thedata(shownasthesolidredboxinFig.1),andfindthe (e.g., Girardi et al. 2011).
best fitting isochrone for each observed star. This is sim- Finally, we note that the base of the red-giant branch
ilar to what was done in Goudfrooij et al. (2011b; 2014), also appears narrower than would be expected if large age
althoughwenotethatwehavenotattemptedtoincludethe spreads were present. Again, the stars cluster along the
effects of binaries in this simple calculation. The resulting youngest isochrone that fits theMSTO region.
(cid:13)c 2014RAS,MNRAS000,1–11
4 Bastian & Niederhofer
3.1.4 Red Clump Position
In Fig. 4 we show a zoom in of the CMD of NGC 1806
centred on the base of the red giant branch and the red
clump (RC). Additionally, we show four isochrones with
ages between 1.31 and 1.77 Gyr. As was the case for the
SGB,wenotethattheposition oftheobservedstarsdonot
conform to the expectations if there was a significant age
spreadpresentwithinthecluster.Whileasignificantspread
in colour is expected, thereis little spread in observations.
We quantify this in a similar way as was done for the
SGB.Wesimply assign anagetoeachoftheobserved(and
synthetic) stars based on the closest isochrone. The results
are shown in Fig. 5. Note that the RC width is consistent
with the expectations of an SSP, while an extended SFH
results in a distribution that is clearly at oddswith theob-
servations. The reason why this “pseudo-age distribution”
is not symmetric (the input age distribution is) is that the
Figure1.TheCMDofNGC1806alongwithisochronesofages
isochrones bunch up at older ages at the RC, so small pho-
1.41, 1.58, and 1.86 Gyr, which span the width of the eMSTO.
tometric errors may lead to large age differences.
Stars identified as part of the SGB are highlighted in red. Note
that the SGB stars do not span the same width in age as the
eMSTO. The red box denotes the cut across the MSTO where
3.1.5 The Effects of Binarity
theextended star-formationhistoryisinferredfrom.
We have explored the effect of binaries on the SGB mor-
phology, by creating CMDs of synthetic clusters with and
without binaries. For the binary populations we assume a
150
flat mass ratio and differing binary fractions, ranging from
NGC 1806
zero (only single stars) to 1 (all stars have binary compan-
MSTO region
ions).OnlyinthecasesofhighbinaryfractionwastheSGB
Observed Distribution
significantly affected, shifting the stars to brighter magni-
tudes. For realistic binary fractious we found that theSGB 100 EPxetaekn daegde S= 1F.H61 Gyr
Dispersion = 141 Myr
is only marginally broadened by binaries, consistent with
the more detailed simulations of Milone et al. (2009 - their Nr
Figs. 31−36).Thereason forthisistherelativelylow frac-
tion of near-equal mass binaries in the clusters (10−15% - 50
Milone et al. 2009). Due to the intrinsic brightness of SGB
stars, low mass companions do not significantly contribute
to thecombined fluxof a binary pair.
Due to the relatively short duration of the RC phase 0
in stellar evolution, and the fact that RC stars are much 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4
Age (Gyr)
brighter than their MS counterparts, binaries are not ex-
pectedtohavealargeeffectontheRCpositionormorphol-
ogy.OnlyintherarecircumstancethataRCstarhasanRC Figure2.Theextendedstar-formationhistoryinferredfromthe
starbinarycompanionwouldthemagnitudebesignificantly cutacrosstheMSTO.The(blue)dash-dottedlineshowsthebest
fitgaussiantothedistribution,withtheparameterslistedinthe
affected.
panel.
3.2 NGC 1846
STO region) is clearly offset from theobserved distribution
We have carried out a similar analysis of the colour- (Fig.8).TheSGBmorphologysuggestsasignificantlylower
magnitude diagram of NGC 1846, again focussing on the age (∼1.4 Gyr) than that inferred from theeMSTO region
SGB and RC regions. The CMD of NGC 1846 is shown in (∼ 1.6 Gyr), and also a narrower age spread. The inferred
Fig.6.AswasthecaseforNGC1806,differentauthorshave age from the SGB is that of the youngest isochrone needed
suggested different agespreadsfor thiscluster,with Milone tomatchtheeMSTO,whichis1.4Gyr,similartothatfound
et al. (2009) suggesting a spread of ∼ 250 Myr, Rubele et for NGC 1806.
al. (2013) finding 330 Myr and Goudfrooij et al. (2011b) From this we conclude that the SGB morphology is
suggesting ∼ 600 Myr. We adopt a nominal isochrone of inconsistent with the presence of a significant age spread
1.4 Gyrfor ouranalysis. withinthecluster.Additionally,fromFigs.9&10itisclear
We carried out the same analysis as was done for that the RC position and morphology is also inconsistent
NGC 1806, and the results are shown in Figs. 7, 8, 9 & with the inferred age spread from the MSTO, being best
10. described bya single isochrone (an SSP).
As was found for NGC 1806, the expected SGB mor- As was the case for NGC 1806, we again note that we
phology for an extended SFH (as inferred from the eM- have not included the effects of binarity nor differential ex-
(cid:13)c 2014RAS,MNRAS000,1–11
eMSTOs Are Not Caused By Age Spreads 5
15 Age [Gyr] 2.014.914.816.717.619.612.514.417.411.314.218.213.117.112.017.02 60 NGC 1806
NGC 1806 RC region
SGB region
50
Observed Distribution
10 40 SSP + Errors
1.44 Gyr
Observed Distribution
Nr Nr
SSP + Errors 30
1.44 Gyr Extended SFH (MSTO)
Peak age = 1.61 Gyr
5 EPDxeistapeken rdasegioden S= =1F .H164 1(1 MG MSyTryrO) 20 Dispersion = 141 Myr
+ Errors
10
0 0
-1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4
Magnitude difference Age [Gyr]
Figure 3. Analysis of the SGB of NGC 1806. The histogram Figure5.Theinferredstar-formationhistoryofNGC1806based
showstheobservedmagnitudedifferencebetweeneachstariden- onthemorphologyoftheredclump(RC).
tifiedasbeingpartoftheSGB andanominalisochrone(chosen
tobe1.44Gyr).Vertical dashed linesshowthe expected magni-
tudedifference ofdifferentisochrones,labelledinthepanel. The combinations (e.g., including F555W) based on new pho-
dash-dotted(red)lineshowstheexpected distributionofstarsif tometry,Milone etal.(in prep),andfindconsistent results.
theunderlyingdistributionwasanSSPconvolvedwiththeobser-
Ourresultsarenotinagreementwiththosereportedby
vational errors.Thedash-dot-dot (blue) lineshows theexpected
Rubeleet al.(2013) whofittheSFHofNGC1846 with the
distribution adopting the parameters that best fit the eMSTO
STARFISH code (Harris & Zaritsky 2001) and found that
region (assuming that an age spread is the cause). All distribu-
the CMD was best reproduced by an extended SFH. Their
tions shownareforthe samenumber of stars inthe SGB region
(N =41inthiscase). fitincludedtheSGBandRCportionsoftheCMD,aswellas
theMSTO,soitcouldhavefoundaninconsistencybetween
the different CMD regions. The code, however, is weighted
to regions with the most stars, hence was heavily weighted
19.5 to the MSTO region of the CMD. Figure 7 of Rubele et
al.(2013)clearlyshowshighχ2 valuesforthisclusteratthe
NGC 1806
position of the RC, inferring that the best fitting extended
1.31 Gyr
20.0 1.41 Gyr SFHdidnotreproducethisregion,consistentwithwhatwe
1.58 Gyr
1.77 Gyr havereported here.
Mackey et al. (2013) have found evidence that
5W 20.5 NGC 1846 displays systematic rotation, similar in magni-
43 tude to the velocity dispersion of the cluster. One expla-
F
nation for this phenomenon is that a dynamically colder
“2nd generation” formed with a net rotation within an ex-
21.0
isting“1stgeneration”(e.g.,Bekki2010).However,sincethe
cluster does not display any abundance variations (Mackey
et al. in prep.) it is not possible to differentiate between
21.5
the“1st”and“2ndgenerations”(onlythe“2ndgeneration”
1.5 1.6 1.7 1.8 1.9 2.0 stars would be expected to show rotation in this model).
F435W - F814W
Hence,itmaysimplybethattheclusterasawholerotates,
with no relation to the eMSTO phenomenon or any poten-
Figure 4. The CMD of NGC 1806 highlighting the base of the
tial age spreads within the cluster. Indeed, rotation may
redbranchandredclump(RC).Fourisochronesareoverplotted.
be a common property of massive clusters, and has been
Notethatagesyoungerthan1.32Gyrorolderthan1.58Gyrdo
notfittheRCmorphology.Thenarrownessoftheobservedbase observedinyoungmassiveclusters,e.g.,R136andGlimpse-
oftheRGBandRClocationareinconsistentwithsignificantage C01, H´enault-Brunetet al. (2012) and Davieset al. (2011),
spreadswithinthecluster.Theboxshowstheregionfromwhich respectively.
thestar-formationhistoryofRCstarswasinferred.
4 DISCUSSION AND CONCLUSIONS
tinction (and also for thepotential spread caused bystellar
rotation), hence any observed spread is an upper limit to The two clusters analysed in the present work (NGC 1806
the intrinsic spread in the distribution. Additionally, there & NGC 1846) display sub-giant branches (SGBs) and red
is likely some amount of stellar field star contamination in clumps(RCs)thataresignificantlynarrowerandoffsetfrom
theused CMD. what would be expected if age spreads of 200−600 Myr
We have also tested our results using different filter would be present within the clusters. The observed spreads
(cid:13)c 2014RAS,MNRAS000,1–11
6 Bastian & Niederhofer
19.5
NGC 1846
1.31 Gyr
20.0 1.41 Gyr
1.58 Gyr
1.77 Gyr
W
5 20.5
3
4
F
21.0
21.5
1.5 1.6 1.7 1.8 1.9 2.0
F435W - F814W
Figure 6.ThesameaFig.1butnowforNGC1846. Figure 9. The CMD of NGC 1846 highlighting the base of the
redbranchandredclump(RC),similartoFig.4
200
NGC 1846 NGC 1846
80 RC region
150 MSTO region Observed Distribution
Observed Distribution 60 S1.S4P1 G+ yErrrors
Nr 100 Extended SFH EPxetaekn daegde S= 1F.H5 9( MGSyTrO)
PDeisapke rasgioen = =1. 5193 6G Myryr Nr Dispersion = 136 Myr
40
50
20
0
1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4
0
Age (Gyr)
1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4
Age [Gyr]
Figure 7.ThesameasFig.2butnowforNGC1846.
Figure 10. The inferred star-formation history of NGC 1846
basedonthemorphologyoftheredclump(RC).
14
Age [Gyr] 2.014.914.816.717.619.612.514.417.411.314.218.213.117.112.017.02
12 NGC 1846 areinfactupperlimitstotheactualspreads,aswehavenot
accounted for the affects of differential extinction (Platais
SGB region
10 et al. 2012) or binarity in our sample (or for the potential
effects of stellar rotation). Li et al. (2014b) have found a
Observed Distribution
8 similarbehaviourintheCMDofNGC1651,confirmingthat
Nr S1.S4P1 G+ yErrrors the lack of spreads in the SGB is common in clusters that
6
Extended SFH (MSTO) display an eMSTO. Hence, our results are not consistent
Peak age = 1.59 Gyr
Dispersion = 136 Myr withtheinterpretationofagespreadsbeingthecauseofthe
4 + Errors
eMSTOphenomenon,incontradictionwithpreviousclaims
(e.g., Goudfrooij et al. 2011b; Rubele et al. 2013; Correnti
2
et al. 2014).
0 Rubele et al. (2013) fit the star-formation history of
-1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4 two intermediate age clusters, NGC 1846 and NGC 1783,
Magnitude difference
under the assumption that the eMSTO was due to an age
spread. The authors used the STARFISH star-formation
Figure 8.AnalysisoftheSGBofNGC1846(similartoFig.3). history code (Harris & Zaritsky 2001) which compares syn-
theticobservationsbasedonlinearcombinationsoftheoret-
ical isochrones (plus errors) to the observations, perform-
ing a χ2 minimisation. The authors report age spreads of
250−300 Myr for the two clusters. Much of the weight of
(cid:13)c 2014RAS,MNRAS000,1–11
eMSTOs Are Not Caused By Age Spreads 7
theanalysis comes from theMSTO region, duetothelarge We can also use the cluster population properties of
numberofstarsinthatphase,andarelativelylowweightis the LMC to test the Goudfrooij et al. (2011b; 2014) sce-
appliedtoportionsoftheCMDwithfewerstars(e.g.,SGB, nario. While the current masses of the eMSTO clusters are
RGB,RC).However,wenotethatforbothclusters,thebest similar to the most massive YMCs (e.g., NGC 1850, 1866
fit extended SFH fails to reproduce the morphology of the and 1856 - Niederhofer et al. 2015), Goudrooij et al. (2011;
redclump(inparticularintheF435W−F814W filters),in 2014) posit that the intermediate age clusters were much
the sense that the models predict too many red RC stars. more massive (by a factor of ∼4) at birth (and also signif-
We also note that the position of the RC was not matched icantly denser). Based on size-of-sample effects (e.g., Gieles
in the analyses of NGC 1806, 1846, 411, 1718 and 2203 by &Bastian2008) wewouldexpectthattheSFRoftheLMC
Goudfrooij et al. (2011a; 2014), again calling into question was ∼4 times higher while the eMSTO clusters were form-
theinterpretationoflargespreadswithintheseclusters.This ing (c.f., Maschberger & Kroupa 2011). However, no such
is similar to the results found in the present work. We will increase in the SFH, based on resolved stars, is seen from
analyse the CMD of NGC 1783 in detail in a future work 1−2 Gyr ago relative to today in the LMC (e.g., Harris
(Niederhofer et al. in preparation). & Zaritsky 2009; Weisz et al. 2013). The lack of increase
Based on the SGB and RC locations (and also on the in the SFH from 1−2 Gyr is consistent with the fact that
base of the red-giant branch), we infer that NGC 1806 and the clusters in this age range would have been born with
1846 haveyoungeragesthanhasbeenreported inthepast, similar masses (at the upper end of the mass function) as
with the best fitting age from the SGB agreeing with the thatobservedintheYMCsintheLMCtoday.Asdiscussed
youngest age inferred from the MSTO region (see § A3 for above,theclustersareunlikelytobetidallylimitedsotheir
a discussion of the degeneracies between age and metallic- currentmassesshouldbegoodapproximationsoftheinitial
ity in the CMD fitting of these clusters). Hence, it appears masses. This can be quantified by using the LMC cluster
thatsomeeffectiscausingstars tomovered-wardin colour sample of Baumgardt et al. (2013). There are nine clusters
spaceattheMSTO.NGC1651displaysthesamebehaviour, more massive than 5× 104 M⊙ (a high value was taken
withan inferredageof∼2Gyr,suggesting thatitisagen- to avoid any completeness limit issues) with ages less than
eral property of the eMSTO intermediate age clusters, and 500Myr,or0.018clusters/Myr.Takingthesamemasslimit
not an age effect (Li et al. 2014b). This is the behaviour but looking at the 1−2 Gyr age range, we find 15 such
predicted if stellar rotation was the cause of the eMSTO clusters, or 0.015 clusters/Myr, very similar to that found
(Bastian & de Mink 2009; Yang et al. 2013). At present, it forthepast500Myr.Hence,theclusterpopulationsuggests
is unclear how rotation will affect the SGB and RC, which that the cluster formation rate, like the SFR derived from
we will leave to a futurework. CMD analyses, was roughly the same for the past 500 Myr
Goudfrooij et al. (2014) found a correlation between as it was from 1−2 Gyr in the LMC. Neither the SFH,
the fraction of blue/bright stars on the MSTO (i.e., young based on resolved stars, nor the cluster population suggest
stars if interpreted as due to age effects) and the fraction an increase of a factor of ∼ 4 in the LMC from 1−2 Gyr
of “secondary RC” stars to the main body of RC stars, i.e. relative to now, that would be expected in the Goudfrooij
stars that are bluer and fainter than the main red clump. etal.(2011b;2014)scenario.Weconcludethatitisunlikely
Theauthorsusethistoargueforthepresenceofagespreads thattheeMSTOclusterhadmasses,henceescapevelocities,
within their cluster sample, although, as shown here, the significantly higher in the past.
mainbodyoftheRCdoesnotshowevidenceforagespreads. Girardi et al. (2009) studied at SMC intermediate age
Goudfrooij et al. (2011, 2014) have suggested that a clusters, NGC 419, focussing on the RC morphology. They
correlation existsbetweenthewidthoftheeMSTOandthe find that the MSTO is broader than would be expected
escape velocity of the cluster. However, this result depends from a nominal isochrone, and suggest an age spread of
heavily on large “correction factors” to change the current ∼ 500 Myr within the cluster. We note that the SGB ap-
observed escape velocities to the “initial ones”. Without pearstoshowthesamemorphology,i.e.clusterstowardsthe
these correction factors many of the clusters that display youngest isochrones, as NGC 1806 and NGC 1846 studied
theeMSTO featurewouldhavelowerescapevelocitiesthan here. However, the RC position is consistent with that ex-
youngclustersthathavebeenshownnottohaveagespreads pected for an older age, unlike that found for the clusters
within them (e.g., Bastian & Silva-Villa 2013; Niederhofer studied here.
etal.2015).Theapplied“correctionfactors”arebasedona Ourresults corroborate a numberof other resultsfrom
numberofextremeassumptions.First,themodelsofcluster the literature, that have shown that age spreads of hun-
dissolution thatare adoptedare notappropriate toclusters dredsofMyrarenotpresentwithinclusters.Asdiscussedin
in the LMC (e.g., the tidal field experienced by the model theintroduction,anumberofstudieshavesearched forevi-
clustersisaMilkyWaylikeGalaxyat4kpcfromthegalac- denceofongoingstar-formation,orlargeagespreads,within
tic centre, unlike the much lower tidal field of the LMC - older (> 10 Myr) clusters, with none found (e.g., Perina
e.g., Lamers et al. 2005). Secondly, it is assumed that the et al. 2009; Larsen et al. 2011; Bastian & Silva-Villa 2013,
clustersbegintheirlivestidallylimitedinordertomaximise Bastian et al. 2013a; Niederhofer et al. 2015). Additionally,
cluster mass-loss, whereas most clusters in the LMC/SMC Cabrera-Ziri et al. (2014) used an integrated optical spec-
that have been studied to dateare not tidally limited (e.g., trum of Cluster 1 in NGC 34, a ∼ 107M⊙, 100 Myr clus-
Glatt et al. 2011). This means that cluster mass loss is in ter, to estimate its star-formation history. The cluster was
fact minimal, regardless oftheinitial stateof masssegrega- found to be well described by an SSP (i.e. no age spread
tionpresentinthecluster.Hence,realisticcorrectionfactors was found), with upper limits of 10−20% of the cluster
wouldbemuchsmallerthanthoseadoptedinGoudfrooijet massbeingmadeupofyoungerstars. Theauthorshaveex-
al. (2011b; 2014) and Correnti et al. (2014). tended this study by looking at a number of other massive
(cid:13)c 2014RAS,MNRAS000,1–11
8 Bastian & Niederhofer
106−108M⊙clusters,includingUVandopticalphotometry Bastian,N.,&Silva-Villa,E.2013,MNRAS,431,L122
in the analysis, and have not found evidence for extended Bastian, N., Cabrera-Ziri, I., Davies, B., & Larsen, S. S. 2013a,
SFHs(Cabrera-Ziri et al. in preparation). MNRAS,436,2852
Additionally,Bastian&Strader(2014)searchedforgas Bastian,N.,Lamers,H.J.G.L.M.,deMink,S.E.,etal.2013b,
and dust in 12 LMC/SMC YMCs, the presence of which is MNRAS,436,2398
Bastian,N.,&Strader,J.2014, MNRAS,443,3594
requiredifnewstarsaretoform,andfoundnoneinanyclus-
Baumgardt,H.,Parmentier,G.,Anders,P.,&Grebel,E.K.2013,
ter,downtolimitsof<1%ofthestellarmass.Cabrera-Ziri
MNRAS,430,676
et al. (2015) extended this analysis to study seven massive
clusters (> 106M⊙) in the Antennae merging galaxies, us- BCeakbkrei,raK-Z.i2r0i,10I,.,ABpaJsLti,a7n2,4N,L.,9D9avies, B., et al. 2014, MNRAS,
ingtheAtacamaLargeMillimetreArray(ALMA)tosearch
441,2754
for gas within the clusters (with ages of 50−250 Myr). No
Cabrera-Ziri,I.,etal.2015,MNRAS,submitted
gaswas detectedwithintheclusters, withan upperlimit of Conroy,C.,&Spergel,D.N.2011,ApJ,726,36
∼8% of the stellar mass present in gas. Correnti,M.,Goudfrooij,P.,Kalirai,J.S.,etal.2014,ApJ,793,
Mucciarelli et al. (2008) searched for chemical spreads 121
(expected if large age spreads exist, in particular if self- Davies, B., Bastian, N., Gieles, M., et al. 2011, MNRAS, 411,
enrichment scenarios are invoked) in three LMC clusters 1386
that display the eMSTO phenomenon, and no abundance D’Ercole,A.,Vesperini,E.,D’Antona,F.,McMillan,S.L.W.,&
spreads werefound. Mucciarelli et al. (2014) haveextended Recchi,S.2008,MNRAS,391,825
Dotter,A.,Chaboyer,B.,Jevremovi´c,D.,etal.2008,ApJS,178,
this analysis to NGC 1806 (also studied in the present
89
work)andagaintheauthorshavenotfoundanyabundance
Gieles,M.,&Bastian,N.2008, A&A,482,165
spreads. Finally, Mackey et al. (in preparation) have also
Girardi,L.,Rubele,S.,&Kerber,L.2009,MNRAS,394,L74
searched for abundance spreads, targeting NGC 1846, and
Girardi, L., Eggenberger, P., & Miglio, A. 2011, MNRAS, 412,
likefor theotheryoungandintermediate age clustersstud-
L103
ied to date, no chemical spreads were found. While a lack Girardi, L., Goudfrooij, P., Kalirai, J. S., et al. 2013, MNRAS,
of abundance spreads does not necessarily mean that age 431,3501
spreadscannotexistwithintheclusters,itdoeshoweverre- Glatt,K.,Grebel,E.K.,Jordi,K.,etal.2011,AJ,142,36
quire either 1) cluster/cluster mergers with different ages Goudfrooij,P.,Puzia, T.H.,Kozhurina-Platais,V.,&Chandar,
(although this would not lead to an extended MSTO, but R.2011a,ApJ,737,3
rather a dual MSTO) or 2) that the cluster was able to re- Goudfrooij,P.,Puzia, T.H.,Chandar, R.,& Kozhurina-Platais,
move stellar ejecta (e.g., the ejecta of AGB stars) at the V.2011b,ApJ,737,4
same time as it was accreting primordial gas from its sur- Goudfrooij, P., Girardi, L., Kozhurina-Platais, V., et al. 2014,
ApJ,797,35
roundings,andnothavethetwogasflowsinteractandmix.
Gratton,R.G.,Carretta,E.,&Bragaglia,A.2012,A&ARv,20,
Hence, significant fine-tuningwould be required.
50
Taken together, we conclude that age spreads are not
Harris,J.,&Zaritsky,D.2001, ApJS,136,25
thecauseoftheeMSTOfeaturefoundinmanyintermediate
Harris,J.,&Zaritsky,D.2009, AJ,138,1243
age (1−2Gyr)clusters. Hence,alternative models, suchas
H´enault-Brunet, V., Gieles, M., Evans, C. J., et al. 2012, A&A,
stellarrotationorinteractingbinariesshouldbeinvestigated 545,L1
further.Newalternativescenariosshouldalsobeencouraged Keller,S.C.,Mackey, A.D.,& DaCosta, G. S.2011, ApJ, 731,
and developed. 22
Keller,S.C.,Mackey,A.D.,&DaCosta,G.S.2012,ApJL,761,
L5
Krause,M.,Charbonnel,C.,Decressin,T.,Meynet,G.,&Prant-
ACKNOWLEDGMENTS zos,N.2013,A&A,552,A121
Lamers,H.J.G.L.M.,Gieles,M.,&PortegiesZwart,S.F.2005,
We are grateful to Antonino Milone for providing his pho-
A&A,429,173
tometry and catalogues. We would like to thank Dou-
Larsen, S. S., de Mink, S. E., Eldridge, J. J., et al. 2011, A&A,
gal Mackey, Gary Da Costa Selma de Mink, and Richard
532,A147
de Grijs for helpful discussions and comments on the
Li,C.,deGrijs,R.,&Deng,L.2014a, ApJ,784,157
manuscript. The anonymous referee is thanked for sugges- Li,C.,deGrijs,R.,&Deng,L.2014b, Nature,516,367
tionsthathelpedimprovethepaper.NBispartiallyfunded Longmore, S. N., Kruijssen, J. M. D., Bastian, N., et al. 2014,
by a Royal Society University Research Fellowship. The PPVI,inpress(arXiv:1401.4175)
results presented here are partially based on observations Mackey,A.D.,&BrobyNielsen,P.2007,MNRAS,379,151
made with the NASA/ESA Hubble Space Telescope, and Mackey,A.D.,BrobyNielsen,P.,Ferguson,A.M.N.,&Richard-
obtained from the Hubble Legacy Archive, which is a col- son,J.C.2008,ApJL,681,L17
laboration between the Space Telescope Science Institute Mackey,A.D.,DaCosta,G.S.,Ferguson,A.M.N.,&Yong,D.
2013,ApJ,762,65
(STScI/NASA), the Space Telescope European Coordinat-
Marigo,P.,Girardi,L.,Bressan,A.,etal.2008, A&A,482,883
ing Facility (ST-ECF/ESA) and the Canadian Astronomy
Marino,A.F.,Milone,A.P.,Sneden, C.,etal.2012,A&A,541,
Data Centre (CADC/NRC/CSA).
A15
Maschberger,T.,&Kroupa,P.2011,MNRAS,411,1495
Milone, A. P., Bedin, L. R., Piotto, G., & Anderson, J. 2009,
A&A,497,755
REFERENCES
Milone, A. P., Bedin, L. R., Cassisi, S., et al. 2013, A&A, 555,
Bastian,N.,&deMink,S.E.2009,MNRAS,398,L11 A143
(cid:13)c 2014RAS,MNRAS000,1–11
eMSTOs Are Not Caused By Age Spreads 9
Mucciarelli, A., Carretta, E., Origlia,L., & Ferraro, F. R. 2008,
AJ,136,375
Mucciarelli, A., Dalessandro, E., Ferraro, F. R., Origlia, L., &
Lanzoni,B.2014, ApJL,793,L6
Niederhofer, F., Hilker, M., Bastian, N., Silva-Villa, S., 2015,
A&A,inpress(arXiv:1501.02275)
Perina,S.,Barmby,P.,Beasley,M.A.,etal.2009,A&A,494,933
Pietrinferni,A.,Cassisi,S.,Salaris,M.,&Castelli,F.2004,ApJ,
612,168
Platais,I.,Melo,C.,Quinn,S.N.,etal.2012, ApJL,751,L8
Rubele, S., Girardi, L., Kozhurina-Platais, V., et al. 2013, MN-
RAS,430,2774
Salpeter,E.E.1955,ApJ,121,161
Weisz, D. R., Dolphin, A. E., Skillman, E. D., et al. 2013, MN-
RAS,431,364
Yang,W.,Bi,S.,Meng,X.,&Liu,Z.2013, ApJ,776,112
FigureA1.ThesameasFig.1butnowforthesyntheticpopu-
lation.ThesyntheticclusterhasagaussianSFH,withapeakat
1.59Gyrandadispersionof136Myr.
APPENDIX A: TESTING THE ADOPTED
METHODS
In order to test whether the age spreads derived from the
eMSTOaccuratelyreproducetheobservations(i.e.whether
150 Synthetic cluster
thecutperpendiculartotheMSTOisnotaffectedbydiffer-
MSTO region
inglifetimesofstarsinthatphaseinthenarrowmassrange
considered) we have performed a series of tests, which are Synthetic Distribution
discussedhere.Wenote,again,thatthistechniquehasbeen Extended SFH
100 Peak age = 1.56 Gyr
usedpreviously,namelyinGoudfrooijetal.(2011a,b;2014). Dispersion = 115. Myr
Nr
50
A1 Synthetic Distributions
Firstwecreatedasyntheticpopulationofstarswithanage
spread similar to the ones inferred from the perpendicular 0
cut across the eMSTO for NGC 1806 and NGC 1846. For 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4
Age (Gyr)
this we adopted the same isochrones used to analyse the
observations (same metallicity), assumed a Salpeter (1955)
FigureA2.ThesameasFig.2butnowforthesyntheticpopu-
stellar initial mass function, and created a population with
lation.
a Gaussian age spread with a peak at 1.59 Gyr and a dis-
persion of 136 Myr. No extinction was added to the data.
Additionally, we adopted the same distance modulus used
for NGC 1806. 12
We then ran this synthetic distribution through the Age [Gyr] 2.014.914.816.717.619.612.514.417.411.314.218.213.117.112.017.02
same procedures used in the present work. The results are 10 Synthetic cluster
shown in Figs. A1−A5. We find that the peak of the age SGB region
distributioniswellrecovered,andthatthedispersion found 8
Synthetic Distribution
intheanalysistechniquesresultsin aslightly (∼10−20%)
smaller dispersion than is input. Hence, we are slightly un- Nr 6 SSP + Errors
1.44 Gyr
derestimatingtheactualagespreadwithintheclusterusing
the analyses technique of cutting across the eMSTO. How-
4 Extended SFH (MSTO)
ever, as can be seen in Figs. A1−A5, such an age spread Peak age = 1.56 Gyr
Dispersion = 115 Myr
wouldeasybedetectedinthemorphologiesandpositionsof + Errors
2
theSGB and RCs.
We have run the above tests for a number of ages be-
0
tween 1 and 2 Gyr and age dispersion of 100−150 Myr
-1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4
andfoundconsistentresults.However,foragesof∼1.8Gyr Magnitude difference
and above it is difficult to infer short age dispersions using
thetechniquesusedhere.Agespreadscanstillberecovered, FigureA3.ThesameasFig.3butnowforthesyntheticpopu-
howevertheir actual extentbecomes highly uncertain. This lation.
is discussed in detail in Keller, Mackey & Da Costa (2011).
(cid:13)c 2014RAS,MNRAS000,1–11
10 Bastian & Niederhofer
19.5 1.0
NGC 1806
Synthetic cluster
STARFISH Result
1.31 Gyr 0.8 Adopted SFH
20.0 1.41 Gyr n
11..5787 GGyyrr butio
ntri 0.6
F435W 20.5 Mass Co
e 0.4
v
21.0 elati
R
0.2
21.5
0.0
1.5 1.6 1.7 1.8 1.9 2.0 1.0 1.2 1.4 1.6 1.8 2.0
F435W - F814W Age [Gyr]
FigureA4.ThesameasFig.4butnowforthesyntheticpopu- 1.0
lation. NGC 1806
STARFISH Result
0.8 Adopted SFH
n
o
uti
b
60 Synthetic cluster ntri 0.6
RC region Co
50 ss
Synthetic Distribution Ma
e 0.4
v
40 SSP + Errors ati
1.44 Gyr el
R
Nr 30 Extended SFH (MSTO) 0.2
Peak age = 1.56 Gyr
Dispersion = 115 Myr
20 0.0
1.0 1.2 1.4 1.6 1.8 2.0
Age [Gyr]
10
FigureA6.Acomparisonbetweentheestimatedstar-formation
0
historybased ontheobservedeMSTO inNGC 1806(top panel)
1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4
Age [Gyr] andNGC1846(bottompanel).Twomethodshavebeenusedand
areshownineachofthepanels.Thedashed(blue)linesrepresent
theadoptedSFHbasedonacutperpendiculartotheMSTO(see
FigureA5.ThesameasFig.1butnowforthesyntheticpopu-
§ 3.1.2). The histogram shows the results of fitting the MSTO
lation.
regionofeachclusterwiththestar-formationhistoryfittingcode
STARFISH. Note that both methods result in very similar age
distributions.
A2 STARFISH SFH Fitting
In addition to the synthetic cluster experiments discussed
above,wehavealsofittheobservedCMDsofNGC1806and
NGC 1846 with the STARFISH (Harris & Zaritsky 2001)
star-formation history code. This is the same technique
which was used by Rubele et al. (2013). We adopted the tocomparedirectlywiththeadoptedSFHbasedonthecut
same isochrones as used in the other analyses performed in along theMSTO, werestricted theSTARFISHfittingto a
the current work; adopting the same metallicity, distance regionaroundtheMSTO.Wechoseboxeswiththefollowing
modulus, and extinction as before. limits: 0.56F435W −F814W 61.5and19.5 6F435W 6
This STARFISH code finds the best-fitting model by 22.00.
creatinglinearlycombinedsyntheticCMDsofsimplestellar The results for NGC 1806 and NGC 1846 are shown
populations (i.e. single isochrones) and comparing them to in Figs. A6. The histogram shows the derived SFH from
the data. We used isochrones between 8.70 6 log(t/yr) 6 STARFISHwhilethedashed(blue)linesshowtheadopted
9.48, logarithmically spaced in steps of 0.03. As with the distribution based on the cut across the eMSTO. We find
previous experiments, we adopt a Salpeter (1955) stellar goodagreementbetweenthetwomethods,hencetheanaly-
IMF.We created an analytical model for the observational sisperformedinthecurrentworkisabletoprovideareason-
errors, assuming anerror forthebrightest starsin theclus- able estimate of the potential age spread along the MSTO.
tersofabout0.01whichincreasesexponentiallyto0.04ata It is this potential age spread that has been tested against
magnitudeofabout24intheF435Wband.Furthermore,we the morphology and position of the SGB and RC. In both
assumed a binary fraction of 0.3 for both clusters. In order regions we donot findevidence for such age spreads.
(cid:13)c 2014RAS,MNRAS000,1–11