An XMM-Newton observation of the young open cluster NGC 2547 coronal activity at 30 Myr
a r X i v :a s t r o -p h /0512441v 1 16 D e c 2005
Mon.Not.R.Astron.Soc.000,1–23(2004)Printed 5February 2008
(MN L
a T E X style ?le v2.2)An XMM-Newton observation of the young open cluster
NGC 2547:coronal activity at 30Myr
R.D.Je?ries 1,P.A.Evans 1,J.P.Pye 2and K.R.Briggs 3
1
Astrophysics Group,School of Chemistry and Physics,Keele University,Keele,Sta?ordshire ST55BG,United Kingdom 2Department of Physics and Astronomy,University of Leicester,Leicester LE17RH 3
Paul Scherrer Institut,5232Villigen PSI,Switzerland
Submitted October 2005
ABSTRACT
We report on XMM-Newton observations of the young open cluster NGC 2547which allow us to characterise coronal activity in solar-type stars,and stars of lower mass,at an age of 30Myr.X-ray emission is seen from stars at all spectral types,peaking among G-stars at luminosities (0.3–3keV)of L x ?1030.5erg s ?1and declining to L x 1029.0eg s ?1among M-stars with masses 0.2M ⊙.Coronal spectra show evidence for multi-temperature di?erential emission measures and low coronal metal abundances of Z ?0.3.The G-and K-type stars of NGC 2547follow the same relationship between X-ray activity and Rossby number established in older clusters and ?eld stars,although most of the solar-type stars in NGC 2547exhibit saturated or even super-saturated X-ray activity levels.The median levels of L x and L x /L bol in the solar-type stars of NGC 2547are very similar to those in T-Tauri stars of the Orion Nebula cluster (ONC),but an order of magnitude higher than in the older Pleiades.The spread in X-ray activity levels among solar-type stars in NGC 2547is much smaller than in older or younger clusters.
Coronal temperatures increase with L x ,L x /L bol and surface X-ray ?ux.The most active solar-type stars in NGC 2547have coronal temperatures intermediate between those in the ONC and the most active older ZAMS stars.We show that simple scaling arguments predict higher coronal temperature in coronally saturated stars with lower gravities.A number of candidate ?ares were identi?ed among the low-mass members and a ?aring rate (for total ?are energies [0.3–3keV]>1034erg)of 1every 350+350?120ks was found for solar-type stars,which is similar to rates found in the ONC and https://www.wendangku.net/doc/7217798863.html,parison with ROSAT HRI data taken 7years previously reveals that only 10–15percent of solar-type stars or stars with L x >3×1029erg s ?1exhibit X-ray variability by more than a factor of two.This is comparable with clusters of similar age but less than in both older and younger clusters.The similar median levels of X-ray activity and rate of occurrence for large ?ares in NGC 2547and the ONC demonstrate that the X-ray radiation environment around young solar-type stars remains relatively constant over their ?rst 30Myr.
Key words:stars:activity –stars:late-type –stars:coronae –stars:rotation –open clusters and associations:individual:NGC 2547–X-rays:stars
1INTRODUCTION
X-ray emission from the hot coronae of cool stars is now a well established phenomenon (e.g.see the review by G¨u del 2004).The emission arises from magnetically con?ned and heated structures with temperatures in excess of 106K.In stars that have reached the zero age main sequence (ZAMS)or older,the driving mechanism for this magnetic activity is thought to be a stellar dynamo:stars with convective envelopes and rapid rotation are relatively luminous X-ray
sources compared with slower rotating stars of similar spec-tral type.There is now a well-founded age-rotation-activity
paradigm (ARAP –see Je?ries 1999;Randich 2000),es-tablished via observations of many open clusters with ages from 50Myr to several Gyr (e.g.Stau?er et al.1994;Stern,Schmitt &Kahabka 1995;Je?ries,Thurston &Pye 1997),whereby younger stars tend to be more rapidly rotating and hence exhibit strong X-ray emission up to a saturated level,where the ratio of X-ray to bolometric ?ux,L x /L bol ?10?3.
c
2004RAS
2R.D.Je?ries et al.
As stars get older,they lose angular momentum and eventu-ally spin down to rotation rates where L x/L bol<10?3and decreases further thereafter.
For very young stars in star forming regions with ages <10Myr,a direct connection between rotation and X-ray activity is much less clear and the presence of an(accre-tion)disc may play a role in either stimulating or inhibiting the observed levels of X-ray activity(see Feigelson et al. 2003;Flaccomio et al.2003a;Flaccomio,Micela&Sciortino 2003b;Stassun et al.2004;Preibisch et al.2005).Feigelson et al.(2003)?nd no correlation between rotation and X-ray activity and a“saturation level”of only L x/L bol?10?3.8 for pre-main-sequence(PMS)stars,both with and without discs,in the Orion Nebula Cluster(ONC).They suggest that a less e?cient,turbulent“distributed”dynamo may act throughout the convective zones of these stars.On the other hand Flaccomio et al.(2003a,b)and Stassun et al. (2004)suggest that ONC stars with accretion discs bias the average L x/L bol downwards,perhaps as a result of intrin-sic absorption or changes in the magnetic?eld geometry. Preibisch et al.(2005)show that active accretion,rather than the mere presence of a disc is possibly responsible for the wider spread and lower median level of X-ray activity among the very young ONC stars.
NGC2547is an interesting open cluster in the context of studying the transition between the early behaviour of stellar coronae in star forming regions and the development of the well tested ARAP at older ages.It has a precisely determined age of either30±5Myr determined from?t-ting isochrones to its0.3–1.2M⊙stars as they descend their PMS tracks(Naylor et al.2002),or35±3Myr determined from the re-appearance of lithium in the atmospheres of even lower mass stars(Je?ries&Oliveira2005).It is old enough that inner circumstellar discs have dispersed–no accretion-related Hαemission or L-band near infrared excesses are seen from its solar-type members(e.g.Je?ries,Totten& James2000;Young et al.2004).However,cluster members with M<1.4M⊙are still in the PMS phase,stars with M<0.4M⊙are fully convective(Siess,Dufour&Forestini 2000;D’Antona&Mazzitelli1997)and it is signi?cantly younger than other well-studied open clusters like IC2391 (50±5Myr)and the Alpha Per cluster(90±10Myr).
NGC2547was observed at X-ray wavelengths by the ROSAT High Resolution Imager(HRI).Je?ries&Tolley (1998)found a rich population of low mass,X-ray active cluster candidates with1029 In this paper we present the results of an XMM-Newton observation of NGC2547using the European Photon Imag-ing Camera(EPIC),which seeks to characterise the coronal emission of solar-type(and lower mass)stars at?30Myr. The sensitivity of these X-ray images is better than the ROSAT HRI data,enabling us to identify cluster members with lower activity and measure X-ray emission from clus-ter members with lower mass.There is also some spectral resolution available with the EPIC data that allows us to test whether the X-ray spectra of the solar-type members of NGC2547are signi?cantly di?erent to active stars in other open clusters.Finally,we are able to look for possible vari-ability in the level of X-ray emission of these young stars on a timescale of7years,which is comparable with the solar magnetic activity cycle. Section2describes the observations,data analysis and identi?cation of X-ray sources with members of NGC2547. Section3deals with spectral analysis of the X-ray data, whilst section4uses the spectral information to calculate intrinsic luminosities and search for evidence of dynamo-related activity.Section5looks at the X-ray variability of NGC2547members,both within the observation(?ares, rotational modulation)and on the longer7year timescale. Section6places NGC2547in context with younger and older clusters and discusses the evolution of X-ray activity, coronal temperatures and coronal variability.Our conclu-sions appear in section7. 2OBSER V ATIONS AND DATA ANALYSIS NGC2547was observed by XMM-Newton between UT 23:16:33on2April2002and UT13:17:53on3April2002 using the EPIC instrument,for a nominal exposure time of49.4ks.The two EPIC-MOS cameras and the EPIC-pn camera were operated in full frame mode(Turner et al.2001; Str¨u der et al.2001),using the medium?lter to reject optical light.The nominal pointing position of the observation was RA=08h10m12.0s,Dec=?49d13m0.0s(J2000.0).As we shall show in subsequent sections(2.1,2.2and4.1),these data yield an X-ray luminosity threshold(0.3–3.0keV)for the weakest detected sources of 8×1028erg s?1for NGC 2547cluster members near the centre of the EPIC?eld of view. 2.1Source Detection Version6.0of the XMM-Newton Science Analysis System was used for the initial data reduction and source detec-tion.Unfortunately,the data were a?ected by several peri-ods of high background.Data from the three cameras were individually screened for high background periods and these time intervals were excluded from all subsequent analysis. Observation intervals were excluded where the total count rate(for single events of energy above10keV)in the in-struments exceed0.35s?1and1.0s?1for the MOS and pn detectors respectively.The remaining useful exposure times were29.0ks and29.4ks for the MOS1and MOS2cameras, but only13.7ks for the pn camera,which is more sensitive to high background intervals. Images were created using the evselect task and a spa-tial sampling of2arcseconds per pixel.The event lists were ?ltered to exclude anomalous pixel patterns and edge ef-fects by including only those events with“pattern” 12. The contrast between background and source events was c 2004RAS,MNRAS000,1–23 XMM-Newton observations of NGC25473 also increased by retaining only those events with energies between0.3and3keV.The edetect XMM-Newton observations of NGC 2547 5 Table 3.X-ray sources that were not identi?ed with candidate members of NGC 2547.This table contains 55rows and is only available in the electronic edition.The format is identical to that of Table 1. 6 8 10 12 14 16 18 20 V B-V Figure 1.V,B-V diagram for X-ray selected members of NGC 2547(open squares).The isochrone is derived from the models of D’Antona &Mazzitelli (1997)using an age of 30Myr,a distance modulus of 8.1,reddening and extinction appropriate for E (B ?V )=0.06and a colour-T e?relation calibrated using the Pleiades (see Naylor et al.2002).Dots represent all photometric cluster candidates within 15arcminutes of the X-ray pointing. 3SPECTRAL ANALYSIS 3.1 X-ray spectra Ten cluster candidates were chosen for a detailed spec-troscopic examination.These ten sources were those with the largest number of detected X-ray photons,with 300-670counts in the pn detector and 230-530counts in the MOS1/MOS2detectors respectively.Source spectra of these stars were extracted from circular or elliptical regions with radii ?12arc sec.This relatively small extraction region was used to minimise the signi?cant subtracted background.A larger (45arcsec)extraction radius for the brightest source was used to check that the smaller extraction radius did not change the derived spectral parameters.The same ?ltering expression used to generate the images was used for source extraction.For 9of the stars spectra were obtained from all three EPIC instruments.Star 11lay in a region of the pn detector which was excluded by the selection expression de-scribed earlier,and thus only MOS data were used for this star.Annuli around each source were used to estimate the background. Redistribution and ancillary response matrices were generated using the rmfgen and arfgen tasks.One ma-trix was generated per instrument (based on star 3)and then used for each of the stars.Energy ranges above 0.3keV were considered in the analysis,but the spectra were binned such that there were at least ten source counts per bin and then modelled using xspec . A single optically thin thermal plasma (mekal –Mewe, 10 12 14 16 18 20 0.5 1 1.5 2 2.5 3 3.5 4 V V-I Figure 2.V,V-I diagram for X-ray selected members of NGC 2547.Symbols as for Fig.1.The isochrone is derived as for Fig.1,but this time an age of 25Myr gives a better ?t (see Je?ries &Oliveira 2005). I R-I Figure 3.I,R-I diagram for X-ray selected members of NGC 2547.Symbols as for Fig.1.The isochrone is derived as for Fig.1,but using an age of 25Myr. Kaastra &Liedahl 1995)component modi?ed by photo-electric absorption (a 1-T model)was used as an initial model.The column density of the absorption was ?xed at 3×1020cm ?2.This corresponds to the reddening estimated for bright cluster members and is unlikely to be uncertain by more than a factor of two (see Je?ries &Tolley 1998)2. 2 Experiments which allowed the column density to be a free pa-rameter showed that the X-ray spectra could only constrain the column density to be 1021cm ?2because of the lack of sensi-tivity to this parameter at energies >0.3keV and the possibility to compensate for changes in column density with changes in the emission measure of a cool coronal component c 2004RAS,MNRAS 000,1–23 6R.D.Je?ries et al. Figure4.Examples of one and two component thermal models (as described in the text)?tted to the pn data of star number3. The metal abundance was a free parameter(in the form of a multiple of the solar metal abundances of Anders&Grevesse 1989)while the normalisation of the mekal component was allowed to optimise independently for the three EPIC instru-ments to counter the e?ects of any cross-calibration uncer-tainties.In general,good agreement was found between the three normalisations.The best-?t parameters andχ2values of these model?ts are given in Table4.An example spectral ?t is shown in the top panel of Fig.4. The1-T model?ts are on the whole statistically ac-ceptable.In two cases(stars3and10),the1-T model is rejected at99per cent con?dence.A second mekal com-ponent was added to the models(a2-T model),which of course improved the?t in all stars(Table5and see the lower panel of Fig.4).For four objects the upper bound to the temperature of the second component could not be con-strained;these stars are indicated by an asterisk in Table5. All of the2-T models are statistically acceptable,and there are signi?cant reductions chi-squared values in8cases(sta-tistically justi?ed at>95per cent according to a likelihood ratio test). Little weight should be attributed to this.It has com-monly been found that multi-temperature?ts are required to?t coronal X-ray spectra once a su?ciently precise spec-trum is obtained.Even then,a2-T model is probably a crude approximation to the true di?erential emission measure (DEM).The pattern here appears to be that the DEM could be approximated with a2-T model with a lower tempera-ture T1?0.6keV and an upper temperature T2?1.5keV. In spectra with insu?cient counts or where one component is signi?cantly larger(in terms of the number of X-ray pho-tons produced)than the other,then a1-T?t is adequate with T1 Perhaps the biggest di?erence between the1-T and2-T models is that adding the extra mekal component re-laxes somewhat the requirement for a very low metallicity in the1-T?ts.Even so,signi?cantly subsolar metallicities are implied by all the2-T?ts,with an average Z?0.3. This value is likely dominated by the e?ects of a group of strong(unresolved)iron lines around1keV.The deduction of low coronal metallicity is a common feature of spectral ?ts to X-ray data from low-mass stars with high levels of magnetic activity(e.g.Briggs&Pye2003;G¨u del2004). Telleschi et al.(2005)studied solar analogues at a range of ages,?nding that coronal iron abundances decrease from solar values for an average coronal temperature of4MK to half-solar at temperatures of10MK.The NGC2547stars have average(emission measure weighted)coronal temper-atures>10MK,so a coronal metallicity of Z?0.3is not surprising. 3.2Hardness ratios Due to low numbers of counts,spectral?tting barely pro-vides constraints on the temperature distribution of the coronal plasma even in the brightest NGC2547sources. Nevertheless,there is su?cient evidence here that these coronae follow the often observed pattern that,given suf-?cient statistics,a multi-component thermal model?ts the data better than a single temperature.To extend our analysis to fainter sources,the hardness ratio,de?ned as (H?S)/(H+S),where S is the count rate in the0.3–1.0keV band and H is the count rate in the1.0–3.0keV band,was modelled in terms of a2-T corona.The purpose is to provide a physical interpretation for any trends in the spectral distribution with type of star or overall X-ray ac-tivity level and also to estimate how any spectral changes might in?uence the conversion from X-ray count rates into ?uxes(see Section4). Figure5shows plots of the hardness ratios in the pn detector versus colour and also versus count rate.The plots for the MOS detectors are similar,but noisier,and are omit-ted for brevity.These hardness ratios were simulated us-ing a2-T model,with?xed temperatures of T1=0.6keV, T2=1.5keV,a?xed metallicity of Z=0.3and N H= 3.0×1020cm?2.This simple,though not unique,model is justi?ed by the spectral?tting results in section3.1.The emission measure ratio of the two components is altered to generate a given hardness ratio.The required emission measure ratio(hot/cold)is indicated against the y-axes of Figure5. Recent work on high quality X-ray spectra of nearby stars with varying activity levels has identi?ed a trend of increasing emission measure weighted mean coronal tem-perature with X-ray activity(e.g.Telleschi et al.2005). The same trend,albeit with poorer ROSAT spectra,has also been identi?ed among stars of the Pleiades cluster (Gagn′e,Caillault&Stau?er1995a).There is evidence for c 2004RAS,MNRAS000,1–23 XMM-Newton observations of NGC25477 Table4.Parameters andχ2(and reducedχ2)values for1-T?ts to sources3–12in Table1.The tabulated emission measures assume a cluster distance of417pc and are the average of the emission measures derived from the pn and MOS instruments.Uncertainties quoted are90per cent con?dence limits for one parameter.The?nal column gives the probability of attaining aχ2at least as high as observed if the1-T model were valid. 30.791 1.05(0.981.11)53.39(53.3353.45)0.16(0.140.21)130.7(1.49)0.01 40.5360.68(0.650.74)53.55(53.4953.62)0.13(0.100.18)57.5(0.80)0.89 5 1.485 1.38(1.261.66)53.43(53.3453.48)0.12(0.070.21)105.7(1.22)0.08 60.9940.80(0.740.87)53.52(53.4653.58)0.05(0.030.07)71.7(0.96)0.59 7-0.060.69(0.650.73)53.16(53.0553.27)0.23(0.160.35)59.7(1.07)0.34 8-0.080.70(0.660.77)53.13(53.0053.24)0.25(0.180.37)59.2(0.99)0.50 90.6220.87(0.801.03)53.15(53.0653.24)0.12(0.080.17)40.5(0.96)0.54 100.6930.82(0.760.88)53.23(53.1453.30)0.12(0.090.18)82.1(1.55)0.01 110.755 1.02(0.911.11)53.47(53.4053.54)0.10(0.060.15)54.5(1.19)0.18 120.704 1.05(0.951.35)53.06(52.9653.15)0.13(0.080.20)50.4(1.33)0.09 Star kT1kT2EM1EM2Zχ2(χ2ν) (keV)(keV)(cm?3)(cm?3) this in Fig.5in terms of increasing hardness ratio and therefore hotter coronae as the pn count rate(and hence X-ray luminosity)rises.A?tted relationship of the form HR(pn)=0.23(±0.04)×log10(pnrate)?0.07(±0.06)(shown as a dashed line)appears a reasonable description.However, there is also evidence for an intrinsic scatter in this relation-ship,especially at pn count rates of0.005–0.01s?1.The re-duced chi-squared is2.53,indicating that further parameters may be important.The plots of hardness ratio versus colour reveal why this may be so.There are two groups of stars with the highest hardness ratios:at0.7 Three other features of the plots are worthy of comment. First,there seem to be a lack of K stars(1 c 2004RAS,MNRAS000,1–23 8R.D.Je?ries et al. -1.2 -1-0.8-0.6-0.4-0.2 0 0.2 0.4 0.001 0.01 0.1 H R (p n ) pn count rate (/s) 0.0 1.04.0EM ratio -1.2 -1-0.8-0.6-0.4-0.2 0 0.2 0.4 0 0.5 1 1.5 H R (p n ) B-V 0.0 1.04.0EM ratio -1.2 -1 -0.8-0.6-0.4-0.2 0 0.2 0.4 0 0.5 1 1.5 2 2.5 3 3.5 H R (p n ) V-I 0.0 1.04.0EM ratio Figure 5.The hardness ratio from the pn detector (see sec-tion 3.2)for NGC 2547members versus overall count rate in the pn detector,versus B ?V and versus V ?I .Indicated against the y-axis of each plot are the emission measure ratios of the hot to cold components that would provide that observed hard-ness ratio in a simple 2-T coronal model with T 1=0.6keV and T 2=1.5keV.The dashed line in the upper plot is a simple straight line ?t to the data. 4X-RAY ACTIVITY LEVELS IN NGC 25474.1 X-ray luminosity To look at the overall level of activity in the cluster as a function of colour/spectral type/T e?requires a means of cal-culating X-ray luminosity and the ratio of X-ray to bolomet-ric luminosity.For simplicity,and also because the available spectral information is too sparse to do otherwise for most sources,a uniform conversion factor between X-ray count-rate and unabsorbed X-ray ?ux in the 0.3–3.0keV range was assumed for each of the three cameras.To do this,the mean hardness ratio in each of the cameras was compared with the two-component spectral model predictions discussed in sec-tion 3.2.From this it seems appropriate to adopt an average X-ray spectrum consisting of a 2-T model with an emission 1e+29 1e+30 1e+31 0.5 1 1.5 L x (0.3-3 k e V e r g s -1)B-V 1e+29 1e+30 1e+31 0 0.5 1 1.5 2 2.5 3 3.5 L x (0.3-3 k e V e r g s -1) V-I Figure 6.X-ray luminosity (0.3–3.0keV,assuming a distance of 417pc)as a function of B ?V and V ?I for NGC 2547candidate members.The dashed lines indicate the approximate sensitivity limit of the X-ray observations at the centre of the X-ray ?eld. 1e-07 1e-06 1e-05 1e-04 0.001 0.01 0 0.5 1 1.5 L x /L b o l B-V 1e-06 1e-05 1e-04 0.001 0.01 0 0.5 1 1.5 2 2.5 3 3.5 L x /L b o l V-I Figure 7.The ratio of X-ray to bolometric luminosities for stars in NGC 2547.The dashed lines representing the approximate sen-sitivity limit of the X-ray observations for a star on the cluster isochrone at the centre of the X-ray ?eld. c 2004RAS,MNRAS 000,1–23 XMM-Newton observations of NGC25479 measure ratio of hot/cool=0.7and Z=0.3.This yields conversion factors(from counts in the0.3–3.0keV range to a?ux in the0.3–3.0keV range)of1.86×10?12,6.72×10?12 and6.59×10?12erg cm?2per count for the pn,MOS1and MOS2detectors respectively3. There is little uncertainty injected by assuming a uni- form conversion factor regardless of hardness ratio.The vari- ation in the conversion factor is less than±6per cent for the most extreme hardness ratios found in our data-with harder coronae leading to larger conversion factors.We have also tested variations in the assumed metal abundances be- tween Z=0.1and Z=1.0.Changing the metal abun- dance while keeping other parameters?xed leads to conver- sion factor variations of±5per cent,with more metal-poor coronae yielding larger conversion factors.Altering the ab- sorbing column from our assumed value of3×1020cm?2to 1020cm?2or1021cm?2leads to conversion factors that are only6per cent smaller or25per cent larger respectively. Figure6shows the X-ray luminosity(0.3–3.0keV)of cluster members versus B?V and V?I.A weighted mean L x was used where measurements from more than one EPIC detector were available.A distance of417pc was assumed. Uncertainties in L x have been estimated using the count rate errors,but we added a further10per cent systematic error in quadrature for each EPIC count rate to cover un- certainties in the detector PSF modelling in the SAS ede- tect 10 R.D.Je?ries et al. -1.2-1-0.8-0.6-0.4-0.2 0 0.2 0.4 1e-06 1e-05 1e-04 0.001 0.01 H R (p n ) L x /L bol 0.0 1.04.0EM ratio -1.2 -1-0.8-0.6-0.4-0.2 0 0.2 0.4 1e+07 1e+08 1e+09H R (p n ) F x (erg s -1 cm -2) 0.0 1.04.0EM ratio Figure 8.Hardness ratio,measured by the pn detector,as a function of X-ray activity.The dashed lines are minimised chi-squared ?ts –see text. in this plot hints at a dependence on another parameter.Figure 8shows how the pn hardness ratio varies with the activity indicators L x /L bol and the surface X-ray ?ux,F x (radii as a function of colour were obtained from the evolu-tionary models used to obtain the isochrone in Fig.2).Only those stars with a V ?I measurement were used,to avoid any scatter introduced by the inclusion of early-type stars in which the dominant optical source is probably not the source of the X-rays. Curves of the form HR(pn)=0.17(±0.03)×log 10(L x /L bol )+0.13(±0.10)and HR(pn)=0.32(±0.04)×log 10F x ?3.00(±0.33)were ?tted,which do indicate signi?cant correlations between hardness ratio and X-ray activity.According to the 2-T modelling of the hardness ratio (see section 3.2)this corresponds to an emission measure ratio of hot to cool plasma which changes from about zero at L x /L bol ?10?5,F x ?3×107erg s ?1cm ?2to unity at L x /L bol ?10?3,F x ?3×108erg s ?1cm ?2.However,the reduced chi-squared of these ?ts are 2.47and 1.95respectively (with 82degrees of freedom).Both the signi?cance of the correlations and the scatter are almost the same as found for the hardness ratio versus pn count-rate relationship in section 3.2.Thus L x ,L x /L bol or F x could be used to predict how hot a coronae will be,but are not deterministic in the sense that an rms hardness ratio scatter of about 0.15exists at a given L x ,L x /L bol or F x .This cannot be explained by the measurement errors and corresponds roughly to a variation of a factor 2in the emission measure ratio of the hot to cool plasma in the 2-T model. 1e-06 1e-05 1e-04 0.001 0.01 L x /L b o l P/t conv Figure 9.X-ray activity (in the range 0.1-2.4keV –see text)as a function of Rossby number for NGC 2547(?lled squares),a range of solar-type stars from the ?eld,Pleiades and other young clusters (from Pizzolato et al.2003)and PMS stars with 0.5 4.4The rotation-activity connection Je?ries et al.(2000)published projected equatorial veloci-ties (v sin i )for 23probable members of NGC 2547which have 0.62 This relationship was re-examined using the XMM-Newton data,but adopting a more physical approach that incorporates both rotation and the properties of the subpho-tospheric convection zone.An approximate Rossby number,the ratio of rotation period to convective turnover time at the base of the convection zone,was calculated for each of the 23NGC 2547members in the Je?ries et al.(2000)sam-ple.The relationship between convective turnover time and B ?V from Noyes et al.(1984)was used.This is more appro-priate for main sequence stars,but the NGC 2547stars con-sidered here are almost at the ZAMS so this should not lead to serious errors (see Gilliland 1986and section 6.1).The period was estimated from the v sin i and a radius obtained from the D’Antona &Mazzitelli (1997)isochrone shown in Fig.1.We divided each v sin i by π/4to correct for an av-erage projection e?ect. Figure 9shows X-ray activity,expressed as L x /L bol ver-sus the Rossby number.Similar data are plotted for a large number of G/K ?eld and cluster stars.These are from the compilation of Pizzolato et al.(2003),who also used the Noyes et al.(1984)estimation of convective turnover time,and for a group of solar-type (0.5–1.2M ⊙)ONC stars with rotation periods from Getman et al.(2005),where a mean convective turnover time of 250years has been assumed (see Preibisch et al.2005).This latter assumption may intro-c 2004RAS,MNRAS 000,1–23 XMM-Newton observations of NGC254711 Table6.Estimated?are parameters. 31.8×10308.6×103083×1034 51.9×10291.1×1031>36>2×1035 102.0×10304.8×1030122×1034 129.6×10298.5×1030>4>2×1034 131.3×10302.7×1030168×1033 158.9×10291.6×1030121×1034 832.6×10291.3×1030362×1034 12R.D.Je?ries et al. 0 0.02 0.04 0.06 0.08 0.1 0.12 0 10000 20000 30000 40000 50000 M O S 1+M O S 2 c -r a t e (s -1 ) Time (s) #3 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0 10000 20000 30000 40000 50000 M O S 1+M O S 2 c -r a t e (s -1 ) Time (s) #5 0 0.005 0.01 0.015 0.02 0.025 0.03 0 10000 20000 30000 40000 50000 M O S 1+M O S 2 c -r a t e (s -1 ) Time (s) #9 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0 10000 20000 30000 40000 50000M O S 1+M O S 2 c -r a t e (s -1 ) Time (s) #11 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 10000 20000 30000 40000 50000M O S 1+M O S 2 c -r a t e (s -1 ) Time (s) #13 0.005 0.01 0.015 0.02 0 10000 20000 30000 40000 50000M O S 1+M O S 2 c -r a t e (s -1 ) Time (s) #20 0 0.002 0.004 0.006 0.008 0.01 0.012 0.014 10000 20000 30000 40000 50000 M O S 1+M O S 2 c -r a t e (s -1 ) Time (s) #32 GTIs 0 0.01 0.02 0.03 0.04 0.05 0 10000 20000 30000 40000 50000 M O S 1+M O S 2 c -r a t e (s -1 ) Time (s) #4 0 0.005 0.01 0.015 0.02 0.025 0.03 0 10000 20000 30000 40000 50000 M O S 1+M O S 2 c -r a t e (s -1 ) Time (s) #6 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0 10000 20000 30000 40000 50000 M O S 1+M O S 2 c -r a t e (s -1 ) Time (s) #10 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0 10000 20000 30000 40000 50000 M O S 1+M O S 2 c -r a t e (s -1 ) Time (s) #12 0 0.005 0.01 0.015 0.02 0.025 0.03 0 10000 20000 30000 40000 50000 M O S 1+M O S 2 c -r a t e (s -1 ) Time (s) #15 0.005 0.01 0.015 0.02 0 10000 20000 30000 40000 50000 M O S 1+M O S 2 c -r a t e (s -1 ) Time (s) #26 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0 10000 20000 30000 40000 50000 M O S 1+M O S 2 c -r a t e (s -1 ) Time (s) #83 Figure 10.X-ray light curves (0.3–3.0keV)for selected members of NGC 2547that exhibit short-term variability (plus the ultra-fast rotator star 4which appears not to show variability).Possible ?aring periods are indicated with a horizontal line.All the curves begin at MJD 2452367.4786,feature data from the MOS1plus MOS2detectors taken during good time intervals and are background subtracted.The reader should note that the light curves are based on data with frequent small coverage gaps.The data points in the plots represent the best estimate of the count rates over either 2ks or 4ks bins.The “good”time intervals are indicated by a broken line in the plot for star 32.There is only one extended time period of a few ks that has no good data. c 2004RAS,MNRAS 000,1–23 XMM-Newton observations of NGC254713 a Rossby number of0.024–approaching the super-saturated part of Fig.9.Star11actually has L x/L bol?7×10?4,which could indicate the onset of super-saturation,but is not con- clusive.Stars9and32are at least moderately fast rotators with v sin i of27and12km s?1and hence have rotation periods that are shorter than1.9and3.6days respectively. Star6is a slow rotator with v sin i=6km s?1and star 26is a close spectroscopic binary with two slowly rotating components(Je?ries et al.2000).It is worth noting that the fastest known rotators in the cluster(stars4and21in this paper,labelled RX30a and RX35in Je?ries et al.2000), with v sin i values of86and160km s?1respectively,show no signs of variability at the 20per cent level.The X-ray light curve for star4is shown in Fig.10. 5.2Long term variability The previous observation of NGC2547with the ROSAT HRI(Je?ries&Tolley1998),taken in December1995,allows an investigation of variations in X-ray activity on a?7year timescale.The XMM-Newton list of cluster members was cross-correlated against the HRI source positions allowing a correlation radius of up to16arcseconds,re?ecting the larger positional uncertainties in the HRI data.There are72 correlations which are identi?ed in Table7by their“RX”numbers from Je?ries&Tolley.RX86was correlated with, and lay mid-way between,XMM-Newton sources111and 113.In the absence of any additional information the X-ray ?ux in the HRI was split equally between the two. In Je?ries&Tolley(1998)it was claimed that the most active stars in NGC2547were under-active by nearly a fac-tor of two(in terms of L x/L bol)compared with other young clusters,especially among the G and K stars.This seems not to be the case in the XMM-Newton observations(see sections4.1and4.4)and the discrepancy needs explain-ing.One possibility that can now be ruled out is that the NGC2547stars have peculiarly hot or cold coronae result-ing in under-estimated?uxes from the HRI count-rates.Like XMM-Newton,the HRI count-rate to?ux conversion factor is quite insensitive to variations in temperature or column density and the coronal parameters we have deduced here are close to the1keV coronal temperature assumed by Jef-fries&Tolley.After further careful investigation we have found two other e?ects that resulted in under-estimated HRI ?uxes. First,the spectral response of the HRI was varying with time throughout the ROSAT https://www.wendangku.net/doc/7217798863.html,ing an updated gain and response matrix shift appropriate for the time of the NGC2547observation(see David et al.1999for details) the count-rate to?ux conversion factor was re-calculated for(a)the0.1–2.4keV range considered by Je?ries&Tolley (1998)and(b)the0.3–3.0keV range for comparison with the XMM-Newton data.In both cases we have assumed the mean coronal model discussed in section4.1.The case(a) conversion factor is4.12×10?11erg cm?2per count,which is 1.31times larger than the conversion factor used by Je?ries &Tolley.The case(b)conversion factor which is used to estimate the L x and L x/L bol(using the same distance and bolometric corrections as in Table2)values in Table7was 3.40×10?11erg cm?2per count. Second,modelling of the HRI PSF was fairly crude at the time of the analysis performed by Je?ries&Tol- 1e+29 1e+30 1e+29 1e+30 L x ( H R I ) e r g s - 1 Lx (XMM) erg s-1 #5 #26 #89 B-F stars G-K stars M stars 1e-07 1e-06 1e-05 0.0001 0.001 0.01 1e-07 1e-06 1e-05 0.0001 0.001 0.01 L x / L b o l ( H R I ) Lx/Lbol (XMM) #5 #26 #89 B-F stars G-K stars M stars Figure11.A comparsion of X-ray activity(in the0.3–3.0keV band)observed with XMM-Newton and the ROSAT HRI,sepa-rated by7years.Di?erent symbols(and colours in the electronic version)distinguish stars of F-type or earlier,G/K-type or M-type.Upper limits in the HRI are denoted by downward pointing triangles.Two upper limits in the XMM-Newton data are shown with leftward pointing arrows.The straight lines in the plots rep-resent equality between the measurements and variations by fac-tors of0.5and2.0respectively.Several signi?cantly variable stars are identi?ed and commented upon in the text. ley(1998).Their point source parameterisation assumed a Gaussian PSF which got broader with o?-axis distance. This is a considerable simpli?cation compared with the more complex PSF modelling discussed by Campana et al. (1999)in the context of constructing the Brera Multiscale Wavelet(BMW)ROSAT HRI source catalogue(Panzera et al.2003).A signi?cant extended“halo”to the PSF means that the count-rates provided by Je?ries&Tolley(1998) may have been signi?cantly underestimated.To test this the Je?ries&Tolley catalogue was correlated against the BMW catalogue,?nding78matches.The Je?ries&Tol-ley count-rates were systematically lower by a factor of 1.152(±0.042)+0.012(±0.004)θ,whereθis the o?-axis ange c 2004RAS,MNRAS000,1–23 14R.D.Je?ries et al. Table 7.Correlations between XMM-Newton and ROSAT HRI detections (from Je?ries &Tolley 1998)of photometric cluster candidates.There are 108entries in the Table,corresponding to Tables 1and 2.The full table is only available electronically.The columns list (1)the running XMM-Newton source identi?cation number (2)the RX identi?er used by Je?ries &Tolley,(3)the separation between the XMM-Newton and HRI positions,(4)–(5)the HRI countrate (modi?ed from the Je?ries &Tolley value as explained in section 5.2)and its uncertainty,(6)–(8)three values of L x /L bol (for the 0.3–3keV energy range,using B ?V ,V ?I and R ?I indices)and (9)an estimate of L x (0.3–3keV)using the HRI count rates in column (4).If there is no correlation with an HRI source (indicated by a zero in column (2))then the last 4columns are upper limits. 353 2.5 1.28E-03±1.70E-04 4.27E-04 4.08E-04 3.47E-04 1.05E+30430 2.7 4.21E-03±2.90E-04 4.83E-04 4.63E-04 3.92E-04 3.56E+30 RX ID (N02)V V ?I HRI count rate (s ?1) HRI L x /L bol HRI L x (0.3–3keV)XMM L x /L bol XMM L x in arcminutes.The scatter around this relationship was 0.15(rms).We have chosen to correct the Je?ries &Tolley count-rates by this factor rather than use the BMW count-rates.The rationale for this is that the BMW catalogue does not include 24of the 102sources found by Je?ries &Tolley,15of which have XMM-Newton detections of cluster counter-parts and 4of the remaining 9are also closely correlated with photometric cluster candidates.It seems that Je?ries &Tolley were better at ?nding X-ray sources even if their count-rates were systematically too low. For the remaining 36cluster candidates detected by XMM-Newton which have no HRI counterparts an estimated upper limit to the ?ux observed by the HRI was made by looking at the minimum detected HRI count-rates (after the correction described above)as a function of o?-axis angle.Where no RX number is given in Table 7this indicates that the HRI count-rate given is an upper limit. Two signi?cant HRI sources (RX 20and RX 23)were found which correlate with photometric cluster candidates and lie within the EPIC ?eld of view,but which were not detected by XMM-Newton .Both of these X-ray sources are closely correlated with M-dwarfs from the membership lists of Naylor et al.(2002)and 2-sigma upper limits for the count-rates and X-ray ?uxes were calculated for these.The details are given in Table 8. A comparison of the X-ray activity as judged by XMM-Newton and the ROSAT HRI (as listed in Table 7)in the form of both L x and L x /L bol is shown in Fig.11.Di?erent symbols (and colours in the electronic version of the paper)are used to represent stars of approximately F-type or earlier (V ?I <0.67),spectral types G or K (0.67 B ?V otherwise. Figure 11shows that there is excellent agreement be-tween the intrinsic X-ray ?uxes of stars observed with both XMM-Newton and the ROSAT HRI.The majority of stars have varied by less than a factor of two between the ob-servations.There appears to be some evidence that stars with lower X-ray luminosities were brighter at the time of the HRI observations.Of course this is counterbalanced to some extent by the upper limits on HRI ?uxes in the same region of the diagram and by the larger error bars for these sources.Another point to consider is that for X-ray sources with low signal-to-noise,there is an inevitable upward bias if (as is the case here and in Je?ries &Tolley 1998)the posi-tion of the X-ray source is a free parameter in the count-rate determination algorithm.This is because the (weak)source tends,on average,to be located at the position of a posi-tive noise peak.This upward bias will predominantly a?ect the weaker sources in the less sensitive HRI observations.On that basis we don’t believe there is strong evidence that lower luminosity sources are more variable,or for any sys-tematic discrepancy between the HRI and EPIC ?uxes as a function of X-ray activity or spectral type. Various ways of parameterising long-term variability can be found in the literature.To make comparisons we have estimated the fraction of stars that have varied by more than a factor of two and the mean (and median)absolute devia-tion from equal luminosity.These statistics were calculated for two limited samples:(A)L x >3×1029erg s ?1;(B)the G/K star sample as de?ned above.These subsets were cho-sen to minimise the number of upper limits and to avoid the weak X-ray source bias discussed above.For samples A and B there are 8/60and 5/40objects that varied by a factor of two or more,treating the upper limits as detections.The fractions are more likely to be 9/60and 4/40given a more thoughtful consideration of where these upper limits lie.The high L x M-stars could be more variable than average;5/15of the M-dwarfs in sample A are variables.However,there is an obvious selection e?ect favouring a high fraction in this very incomplete sample.The mean (and median)abso-lute deviations from equal luminosity of samples A and B c 2004RAS,MNRAS 000,1–23 XMM-Newton observations of NGC254715 are almost identical at0.098(0.095)dex and0.107(0.094)dex respectively(i.e.a factor of?1.25).The upper limits are treated as detections in this estimate,but their inclusion does not a?ect the result signi?cantly. Assuming that the error bars in Fig.11represent a nor-mal distribution,we have made simulations under the ad-ditional assumption that the two measured luminosities are equal.We?nd that we would expect mean absolute devia-tions of0.087and0.088for samples A and B respectively in any case.This strongly suggests that the majority of sources have not varied at all and that the slightly larger observed mean absolute deviations are attributable to a handful of strongly varying objects,which are discussed below. There are only three clear examples where we are con?-dent that variations of more than a factor of2have occurred (i.e.with deviations well in excess of the estimated errors). These are sources5,26and89,which are labelled in Fig11. Source5is an M-dwarf that has clearly undergone an intense?are during the XMM-Newton observation(see sec-tion5.1and Fig.10).The“quiescent”L x estimated in Ta-ble6is consistent with the upper limit derived from the HRI observation. Source26is a moderately active G-type star,identi-?ed as a short period spectroscopic binary by Je?ries et al. (2000).This was identi?ed as variable in the EPIC data,but the light curve only shows a gradual decrease in the X-ray ?ux during the observation.It is however feasible that we have seen the decay phase of a long duration(>50ks)?are that is responsible for increasing the average L x as viewed by XMM-Newton by a factor of3compared with ROSAT. Source89corresponds to the optically brightest star in the cluster,HD68478with a B3iv spectral type.It is just possible that this star is massive enough to generate X-rays in a stellar wind with L x/L bol?10?8–10?7(e.g. Cassinelli et al.1994).However,the strong variability of almost a factor of4between the EPIC and HRI observations is not expected from early-type X-ray sources(Bergh¨o fer et al.1997),so we hypothesise that it is more likely that the X-ray emission arises from an as-yet-undiscovered late-type companion that?ared during the HRI observation. 6CORONAL ACTIVITY AT30MYR The main focus of this paper is to gather information on the X-ray coronae of PMS stars at?30Myr and put them in context with what we know about X-ray activity in younger star forming regions,like the ONC and older,well-studied clusters such as the Pleiades(age?120Myr)and Hyades (age?600Myr).We can then ask whether the X-ray ac-tivity of such stars conforms to our expectations based on the age-rotation-activity paradigm(ARAP).Or,are other factors besides the changing stellar structure and decreasing rotation rate important,such as the apparent suppression of X-ray activity by accretion in the ONC? Thanks to the relative insensitivity of estimated coronal ?uxes to assumptions about the temperature structure of the coronae,we can split our discussion into a consideration of the evolution of the overall coronal energy losses followed by the evolution of coronal temperatures and variability.6.1The Evolution of X-ray activity Figure12shows the cumulative X-ray luminosity func-tions(XLFs)for NGC2547and the equivalent functions for L x/L https://www.wendangku.net/doc/7217798863.html,parison plots are shown for the ONC, Pleiades and Hyades.These latter samples were obtained from the works of Stelzer&Neuh¨a user(2001)and Preibisch &Feigelson(2005).To provide a fair comparison,all the X-ray?uxes have been adjusted to correspond to the0.5–8.0keV band considered by Preibisch&Feigelson(2005). This has been achieved by subtracting0.14dex from the0.1–2.4keV Pleiades and Hyades?uxes(see Preibisch&Feigel-son)and by subtracting0.04dex from the NGC25470.3–3.0keV?uxes(see section4.1). The Pleiades and Hyades XLFs account for X-ray?ux upper limits in non-detected cluster members using“sur-vival analysis”techniques.No account of upper limits is taken for the ONC and NGC2547data.In the ONC es-sentially all cluster members were detected at X-ray wave-lengths(see Getman et al.2005).This is also true for NGC2547members in certain mass or colour ranges,but not in others(see section4.2).It is not easy to take account of upper limits to the X-ray?uxes of undetected NGC2547 members because,outside of photometric candidacy,there is no list of con?rmed cluster members.Hence the inclusion of X-ray upper limits for photometric candidates which turn out to be non-members could bias the results in a way that is very di?cult to assess.Instead we deal with the XLF of detected photometric candidates,accepting that where the X-ray census is incomplete then the XLF will be overesti-mated. X-ray emission is known to be mass-dependent,or at least to depend on the structural properties of a star–which are both mass and age dependent.Fig.12provides XLFs for three di?erent mass subsets.These are not the same as spectral type subsets because the ONC stars are su?cently young that most stars are on their Hayashi tracks and sig-ni?cantly cooler than they will appear when they reach the ZAMS.The mass estimates for the ONC sample are dis-cussed by Getman et al.(2005).For NGC2547we use a rela-tionship between V?I and mass from the isochrone adopted in Fig.2to to make similar mass-range selections;that0.9 M<1.2M⊙corresponds to1.035 V?I>0.670;that 0.5 M<0.9M⊙corresponds to2.47 V?I>1.035;and that0.1 M<0.5M⊙corresponds to3.94 V?I>2.47. In section4.2arguments were presented to suggest that the former two samples are likely to be complete for NGC2547 and su?er little contamination.The lower-mass sample is likely to be very incomplete for NGC2547.There are no X-ray detections with V?I>3.2and a signi?cant number of X-ray undetected candidates with2.8 6.1.1G stars at the ZAMS The XLFs of stars that are or will become G-stars at the ZAMS show large changes with time.These changes could be caused by decreases in the convective turnover time(τconv) and decreases in the rotation period,but may also be con- c 2004RAS,MNRAS000,1–23 16R.D.Je?ries et al. 0 0.2 0.4 0.6 0.8 1 27 28 29 30 31 32 f (>L x ) log (L x ) (erg/s) 0.9-1.2 M sun G stars at ZAMS N2547ONC Pleiades Hyades 0 0.2 0.4 0.6 0.8 1 27 28 29 30 31 32 f (>L x ) log (L x ) (erg/s) 0.5-0.9 M sun K stars at ZAMS N2547ONC Pleiades Hyades 0 0.2 0.4 0.6 0.8 1 27 28 29 30 31 32 f (>L x ) log (L x ) (erg/s) 0.1-0.5 M sun M stars at ZAMS N2547ONC Pleiades Hyades 0.2 0.4 0.6 0.8 1 -6-5.5-5-4.5 -4-3.5-3-2.5-2-1.5 f (>L x /L b o l ) log (L x /L bol ) 0.9-1.2 M sun G stars at ZAMS N2547ONC Pleiades Hyades 0.2 0.4 0.6 0.8 1 -6-5.5-5-4.5 -4-3.5-3-2.5-2-1.5 f (>L x /L b o l ) log (L x /L bol ) 0.5-0.9 M sun K stars at ZAMS N2547ONC Pleiades Hyades 0.2 0.4 0.6 0.8 1 -6-5.5-5-4.5 -4-3.5-3-2.5-2-1.5 f (>L x /L b o l ) log (L x /L bol ) 0.1-0.5 M sun M stars at ZAMS N2547ONC Pleiades Hyades Figure 12.Evolution of X-ray luminosity and ratio of X-ray to bolometric luminosity for stars in three mass ranges from the very early PMS in the Orion nebula cluster through NGC 2547to ZAMS stars in the Pleiades and Hyades clusters.X-ray luminosities are presented in the 0.5–8.0keV band.Data for other clusters comes from Stelzer &Neuh¨a user (2001)and Preibisch &Feigelson (2005).Note that for the lowest mass range in NGC 2547,the X-ray data are incomplete and therefore the X-ray luminosity functions are biased upwards (see text). nected with the disappearance of circumstellar material.For a 1M ⊙star,Gilliland (1986)shows that τconv decreases by a factor of 10between 1and 30Myr,but hardly changes thereafter.The same evolutionary models also show the to-tal moment of inertia of the star decreasing by a factor of 10between 1and 30Myr and staying constant thereafter.Without angular momentum loss (AML)the rotation pe-riods would decrease with the moment of inertia and little change in the Rossby number would be expected.The other major structural change is of course the development of a radiative core which takes place after a few Myr.Looking at the upper panels of Fig.12we see that whilst the upper lev-els of X-ray luminosity and L x /L bol are higher in the ONC than NGC 2547,the median and minimum levels are very c 2004RAS,MNRAS 000,1–23 XMM-Newton observations of NGC254717 close.This is an important result–indicating that the high level of X-ray irradiation seen in the vicinity of the young ONC stars is maintained for at least?30Myr prior to a quite signi?cant decrease by the age of the Pleiades. In terms of the ARAP we might conclude that there was very little AML between1and30Myr so that stars in NGC2547and the ONC had similar Rossby numbers. However,we know that this is not true as rotational veloci-ties have been measured for reasonable samples of solar-type stars in both clusters.The Rossby numbers in the ONC are signi?cantly smaller(see Fig.9).It is still an unsolved puz-zle as to how this angular momentum is lost but probably involves interactions between circumstellar material and the coronal magnetic?elds as well as losses through a magne-tised stellar wind.X-ray activity levels are maintained at high levels in NGC2547because many(but not all)stars have not yet slowed su?ciently to place them on the declin-ing portion of the relationship between activity and Rossby number in Fig.9.On the other hand,the range of activ-ity levels in the ONC is too large to be explained by the ARAP.This is discussed at length by Stassun et al.(2004) and Preibisch et al.(2005)who conclude that stars without active accretion have saturated or super-saturated X-ray ac-tivity that is equivalent to fast-rotating ZAMS stars,but that active accretors often have signi?cantly suppressed X-ray activity.Active accretion is not found in the NGC2547 stars and the spread in X-ray activity is much smaller and quite consistent with the spread in rotation rates and a small amount(less than a factor of two–see section5.2)of vari-ability. Between NGC2547and the Pleiades there is an order of magnitude decline in the median X-ray activity,a similar and perhaps even larger decline in the minimum activity lev-els,but comparatively little decline in the peak levels of X-ray activity.In terms of the ARAP,this can be understood if the fastest rotating G-stars in the Pleiades have yet to spin down su?ciently that their X-ray activity falls below the saturation level.Queloz et al.(1998)?nd that about14per cent of0.9–1.1M⊙Pleiads have equatorial velocities exceed-ing20km s?1and hence Rossby numbers 0.2which lead to saturated X-ray activity.Conversely,to explain their much lower X-ray activity,the slowest rotators in the Pleiades must rotate at rates that are factors of?3lower than the slowest rotators in NGC2547.This tallies with the available observations.From the projected equatorial velocity mea-surements in Je?ries et al.(2000)we see that only4out of20G-type NGC2547members have v sin i<10km s?1, whereas that fraction is about50per cent in the Pleiades, with the slowest10per cent having rotational velocities sim-ilar to those in the even older Hyades(Queloz et al.1998). Hence the solar-type stars undergo a period of rapid spin-down during the?rst100Myr on the ZAMS,followed by a much longer“plateau”phase.This behaviour can plausibly be explained if the convective envelope and radiative core are rotationally coupled on a timescale somewhere between the age of the Pleiades and Hyades(Sills,Pinsonneault& Terndrup2000). 6.1.2K stars at the ZAMS The situation is somewhat di?erent in the lower mass stars. Here,the structural evolution means that whilstτconv de-creases by a factor of2–6for0.5–0.9M⊙stars between1and 30Myr,there is a corresponding decrease in the moment of inertia by factors of6–8for the whole star.In the absence of AML we might expect the Rossby number to decrease by factors of1.3–3.Beyond30Myr there is little evolution of eitherτconv or moment of inertia,so the Rossby number would remain nearly constant in the absence of AML(see Gilliland1986).The K-stars in the NGC2547sample will have developed a radiative core during the last?20Myr. The most notable property of the NGC2547XLF for K-stars is the very narrow spread in L x and L x/L bol.In terms of the ARAP this is not explained by a narrow spread in rotational rates,but instead by the majority of objects rotating fast enough to lie on the saturated,or even super-saturated portions of the relationship between X-ray activ-ity and Rossby number.It is therefore startling to see that a signi?cant fraction(?30per cent)of ONC stars in this mass range have lower L x/L bol values than even the least ac-tive NGC2547members,although they do still have higher values of L x.There is no explanation of this in terms of the ARAP unless the K-stars of NGC2547have an angu-lar momentum distribution that is skewed to higher values compared with the ONC.But even if that were true,there are clear examples of ONC objects with very small Rossby numbers that have L x/L bol?10?4or even lower.Again, it is accreting objects in the ONC that tend to have these lower levels of L x/L bol.If anything,the bolometric lumi-nosities derived by Getman et al.(2005)for these stars may be underestimates,making the result more signi?cant(see discussion in Hillenbrand1997). Between NGC2547and the Pleiades there is an order of magnitude decrease in the minimum and median levels of X-ray activity,but comparable levels of maximum X-ray activity.Again,this can be interpreted in terms of the ro-tation distributions in the two clusters.A fraction(?20 per cent)of K-type Pleiades have equatorial velocities ex-ceeding15km s?1that would result in saturated levels of coronal emission.Almost half have spun down to less than 7.5km s?1(Queloz et al.1998).Unfortunately only three v sin i measurements exist for NGC2547stars in the same mass range(13.4,9.3and19.2km s?1),so we cannot test this hypothesis in detail.On the basis of the NGC2547XLF we predict that NGC2547members with masses of0.5–0.9M⊙all have Rossby numbers less than0.3and equatorial veloc-ities in excess of10km s?1. 6.1.3M stars at the ZAMS The lowest mass stars we have considered have a much slower evolutionary timescale on the PMS.Hence changes inτconv and moment of inertia are expected both before and after the age of NGC2547.In the absence of AML there should be a decrease in the Rossby number by factors of a few between1and30Myr,but only small changes af-ter that(Gilliland1986).A notable di?erence between this subsample and the higher mass stars considered previously is that many of the NGC2547stars may still be fully con-vective.In the D’Antona&Mazzitelli(1997)models,this transition takes place at0.4M⊙at30Myr,corresponding to V?I=2.58.There is no great change apparent in Figs6 and7at this colour,except perhaps for the development of a small tail of high activity objects. c 2004RAS,MNRAS000,1–23 18R.D.Je?ries et al. Unlike the higher mass subsamples,the X-ray census of M stars in NGC2547is de?nitely not complete and so the XLFs will be biased upwards.There are no detections of any NGC2547stars with M<0.2M⊙,although con?rmed cluster members(see Je?ries&Oliveira2005)are certainly within the XMM-Newton?eld of view.All we can say is that the majority of NGC2547M stars are expected to exhibit saturated or even super-saturated levels of X-ray emission and this is consistent with the observed XLFs.The peak levels of X-ray luminosity and activity are close to those in the ONC and several times higher than in the Pleiades and Hyades.It is hard to be sure that this is a signi?cant di?erence though,because approaching the?ux sensitivity limit of the survey one is bound to upwardly bias the?uxes of the detected objects simply through?uctuations in the background and the intrinsic X-ray variability of the stars.It would be fascinating to do a deeper survey in order to estab-lish whether NGC2547(like the ONC0.1–1.2M⊙stars)ex-hibits a signi?cant fraction of objects with L x/L bol>10?2.5 and objects with L x/L bol<10?4.The con?rmed existence of either of these two populations might lend support to the suggestion that young,fully convective stars might not conform to the ARAP possibly through the operation of a dynamo that is distributed throughout the convective inte-rior rather than at the interface between the radiative core and convective envelope(Feigelson et al.2003). 6.2The evolution of coronal temperatures In section 4.3we showed that average coronal tempera-tures,indicated by crude hardness ratios,increased with L x, L x/L bol and F x,although there was considerable scatter in these relationships.As these activity indicators decline with time(also with considerable scatter)an interesting question is whether coronal temperatures,and by implication coronal structures and heating mechanisms,change solely with coro-nal activity,or whether there is some other time-dependent variable involved.It has been known for some time that this is the case for older stars,with average coronal temperatures decreasing among solar type stars between ages of?70Myr and9Gyr.This evolution appears to be largely governed by the gradual disappearance of the hotter( 10MK)coro-nal plasma(e.g.G¨u del,Guinan&Skinner1997;Telleschi et al.2005).Telleschi et al.(2005)propose a relationship of the form L x∝T4.26between coronal luminosity and mean coronal temperature, This dependence of coronal temperature on activity and age apparently continues to younger ages.Whilst the most active?eld stars studied by G¨u del et al.(1997)have a coronal temperature of about10MK,detailed X-ray spec-troscopy of T-Tauri stars(both accreting and non-accreting)?nds that their coronae are dominated by a very hot com-ponent(>20MK)that is seen only in the largest?ares in the solar corona(Skinner&Walter1998;Grosso et al.2004). What is not clear is whether these hotter coronae are merely an extension of the trend de?ned by older stars,which is pre-sumably driven by changes in rotation rate,or whether the signi?cant increase in coronal temperatures is attributable to di?erences in the stellar structure or the presence of discs in PMS stars. Figure13illustrates where the stars of NGC2547,at an age of?30Myr,?t into this progression.The emission 10 20 30 40 50 60 1e+29 1e+30 1e+31 1e+32 A v e r a g e T ( M K ) L x (erg/s) 1e-05 0.0001 0.001 0.01 A v e r a g e T ( M K ) L x/L bol Figure13.Emission measure weighted mean coronal temper-ature as function of L x and L x/L bol for low-mass stars in NGC2547and several other young clusters(see text).The dashed line is a relationship derived for?eld G-stars by Telleschi et al. (2005)that has been extrapolated beyond2×1030erg s?1.Ob-jects in NGC2547which are“contaminated”with?aring emis-sion are encircled. measure weighted mean coronal temperature is shown as a function of L x and L x/L bol for the8stars in Table5 with B?V 0.536and which are unambiguously low-mass stars.The dashed line indicates the relationship de-rived for0.07–9Gyr G-type stars in the?eld,that have 2×1028 L x 2×1030erg s?1(Telleschi et al.2005).Com-parable data were gathered from the literature for low-mass stars in IC2391(age50Myr,Marino et al.2005),Blanco1 (age?100Myr,Pillitteri et al.2004)and the Pleiades(age 120Myr,Briggs&Pye2003).These comparison samples were chosen on the basis that the spectra were also obtained with the XMM-Newton EPIC instrument and modelled with two thermal components with abundance as a free parame-ter.Finally,data were added for a sample young PMS stars in the ONC(Getman et al.2005).These spectra were ob-tained with the Chandra ACIS instrument,but modelled in a similar manner to the NGC2547stars.Objects were se-lected that had0.3 It appears that the PMS stars of the ONC have coronae c 2004RAS,MNRAS000,1–23 XMM-Newton observations of NGC254719 that are much hotter than a simple extension of the trend de?ned by older stars with lower L x.This becomes even more apparent when considering mean coronal temperature versus L x/L bol,where the higher coronal temperatures of the ONC appear to be a consequence of youth rather than a higher overall level of magnetic activity.The ONC stars are photospherically cooler than the young cluster samples because of their younger evolutionary stage and therefore using X-ray surface?ux(F x∝T4e?L x/L bol)as an activity indicator would yield an even worse correlation. The stars of NGC2547form an intermediate popula-tion.They lie signi?cantly above the mean relationship de-?ned by the?eld G-stars.Of course there is a bias towards the most active stars in NGC2547,and the objects with the highest mean temperatures(stars5and12)were seen to?are during the observation,but this is also the case in the comparison stars in the other clusters.When compared at a given activity level(de?ned by L x/L bol)the NGC2547stars lie between the older clusters and the ONC.We emphasize that the Blanco1,Pleiades and IC2391samples were anal-ysed with the same instrument and in an entirely consistent manner with our approach.We also checked whether our as-sumed hydrogen column density of3×1020cm?2could in-?uence this result.Doubling the column density(see section 3.1)reduces the mean coronal temperature in our NGC2547 sample by only?10per cent–insu?cient to account for the observed di?erences. “Activity”,either judged by L x,L x/L bol or F x,is not capable of predicting what the average coronal temperature will be,so it is di?cult to attribute a causal e?ect to these quantities.Instead there must be an additional age-related factor which should be taken into account.It is unlikely that this factor has anything to do with active accretion or the presence of circumstellar material.We can con?dently rule these out in the case of NGC2547and of19objects included in the ONC sample only4are classical T-Tauri stars and about half show evidence of a near-IR excess,but these do not appear to have systematically higher temperatures.This leaves changes in the interior structure of the PMS stars, the absolute rotation rate or perhaps the surface gravity as the most likely causative e?ects of the higher coronal temperatures. One possibility is that the higher coronal temperatures in the younger stars are a manifestation of a more turbu-lent dynamo distributed throughout the entirely convective interior of the young ONC stars(e.g.see the discussion in Feigelson et al.2003and references therein).This could lead to di?ering coronal structures and higher temperatures,per-haps as a result of a less ordered magnetic?eld and more frequent?aring interactions on a variety of scales.Interme-diate coronal temperatures in the older NGC2547stars may be due to the beginning of strong magnetic braking,di?er-ential rotation between the developing radiative core and envelope and hence the presence of a tachocline region,and initiation of anα?dynamo(e.g.Parker1993).The in?u-ence of a distributed convective dynamo that depends on a deep convection zone would diminish. There are(at least)two problems with these ideas: First,there is no clear evidence that large?ares are more common in ONC stars than the most active stars in NGC2547or indeed older ZAMS cluster(see section6.3.1). Second,if being young and fully convective is the recipe for hotter coronal temperatures then>20MK coronae should be present in all stars with M<0.4M⊙(V?I>2.65) in NGC2547.Although we have no detailed spectra with which to test this,the hardness ratios are reasonably in-dicative.Simulating a2keV thermal plasma with a similar column density and abundance to the other NGC2547stars yields a pn hardness ratio of0.00.There is a hint that coro-nae get hotter for V?I>2.5,but hardness ratios are comfortably below zero for stars of any spectral type(see Fig.5).Of course this analysis neglects any cool coronal component and it is true that there are a few fully convec-tive stars with hardness ratios as large as the hottest stars with detailed spectral?ts.More detailed spectral modelling of better data is required. An alternative explanation for hotter coronae in PMS stars arises from simple scaling laws(see also Jordan&Mon-tesinos1991).Suppose that the coronal plasma and mag-netic?eld approach pressure equilibrium,such that p∝B2, where p is the coronal pressure and B the coronal magnetic ?eld.We could further assume that X-ray emission is ob-served from magnetic loops that extend to some fraction of a pressure scale height,so that L∝T g?1,where L is the loop semi-length and g the surface gravity.The?nal ingre-dient is the scaling law between loop length,pressure and temperature derived for hydrostatic loops by Rosner,Tucker &Vaiana(1978),who show that T∝p1/3L1/https://www.wendangku.net/doc/7217798863.html,bining these we?nd T∝B g?1/2.(1) For large(unsaturated)Rossby numbers,mean?eld dy-namo theory suggests that B should increase with decreasing Rossby number(Durney&Robinson1982),but for small (saturated)Rossby numbers it is quite possible that feed-back e?ects limit any further growth in B or magnetic activ-ity.Hence at large Rossby numbers and for stars that have already reached the main sequence with similar gravities, coronal temperature will chie?y depend on Rossby number and hence rotation(or L x/L bol or L x for a given spectral type)–as observed in older G dwarfs(Telleschi et al.2005). For small Rossby numbers and saturated coronae the de-pendence should be dominated by gravity variations.The models of D’Antona&Mazzitelli(1997)suggest that the surface gravity of a0.8M⊙star changes from log g=3.74 to log g=4.48(in cgs units)between1and30Myr,with only a small further decrease thereafter.Accordingly,equa-tion1predicts that stars in the saturated regime will exhibit a corresponding drop in coronal temperature of a factor2.3 between1and30Myr–approximately what is seen.A prob-able complicating factor is that the scale heights in the ONC stars will be large enough for centrifugal forces to play a role.Indeed,Jardine&Unruh(1999)have shown that coro-nal loops extending beyond the Keplerian co-rotation ra-dius would be unstable and the simple pressure scale height equals the co-rotation radius at T?30MK for a star with log g=3.5and rotation period of5days. If coronal temperature just scales with gravity as de-scribed above then this is readily testable by determining the relative coronal heights ONC and ZAMS stars.The lower gravity ONC coronae should be approximately10 times more extended(T is2.5times hotter and g is4times c 2004RAS,MNRAS000,1–23 20R.D.Je?ries et al. smaller)than in active ZAMS stars4.In practice,there is little strong evidence for this.Whilst the coronae of ZAMS stars have usually been found to be consistent with an ex-tent of a pressure scale height(or less)from eclipse mapping or combinations of emission measure and density estimates (see G¨u del2004for a review),the same diagnostics are not yet available for PMS stars.Perhaps the only technique ap-plied to both classes of objects is the modelling of strong X-ray?ares with loop models.Favata et al.(2005)have used the analysis of strong?ares to determine loop half lengths of109–1010m in several ONC stars.Equivalent analyses of strong?ares on very active ZAMS stars do imply signi?-cantly smaller loop lengths(?3×108m–G¨u del et al.2001; Scelsi et al.2005)with an extent of about a pressure scale height.It must be emphasised though that these approaches to interpreting spatially unresolved?ares are highly model dependent. 6.3The evolution of coronal variability 6.3.1Flaring In principle we could attempt to compare the“?are rate”in NGC2547stars with objects of similar mass in older and younger clusters.In practice this turns out to be di?cult be-cause there is little uniformity in the literature about how a ?are should be de?ned and numerous selection e?ects con-cerning the detection of?aring events that are di?cult to account for.In particular,care must be taken in comparing data with di?ering sensitivities(for instance from clusters at di?erent distances)or from instruments with di?erent energy band-passes. In section5.1and Table6a number of possible?aring events were identi?ed.Although we have identi?ed one ma-jor?are on an M-star in NGC2547,the majority of such sources are so weak in our data that a proper assessment of the?aring frequency in M-stars is impossible.The situation is better for solar-type stars,where we would claim to have been able to detect all?ares with integrated energies exceed-ing1034erg,provided that they had total durations 30ks. We have detected4such?ares and2others that fall just outside these thresholds from a sample of28solar-type stars with masses0.8 ?120 ks. Equivalent statistics are presented or can be calcu-lated from a few sources in the literature for older and younger stars.Gagn′e et al.(1995a)used ROSAT obser-vations of the Pleiades to?nd12?ares with peak L x of 1030–2.5×1031erg s?1.We estimate that they observed3?ares from G-stars with energies in excess of1034erg(mak-ing a small approximate correction for the di?erences in band-passes)from an e?ective monitoring time of60ks on 33G-stars.Hence the comparable?aring statistic to our NGC2547measurement is1?are every660ks.Wolk et al. (2005)have presented a detailed search for?ares in ONC stars with0.9 4Note this is in absolute units,not relative to the stellar radius.dataset.They conclude that the equivalent?are rate statis-tic is1?are every650ks.In terms of duration and peak lu-minosity,the?ares in NGC2547are quite comparable with the large?ares seen in the ONC and https://www.wendangku.net/doc/7217798863.html,rger?ares (>1035erg)are seen occasionally in the ONC,but they are reasonably rare and it is not necessarily signi?cant that none were observed in the solar-type stars of NGC2547. Hence there appears to be little or no change in the rate of occurrence of large?ares in solar-type stars during their ?rst～100Myr.If anything the?are rate seems larger in NGC2547than the ONC,but the reader should note that 2of the4?ares in NGC2547,those in stars10and15, may not have been classi?ed as?ares in the ONC study of Wolk et al.(2005),because their peak luminosities were less than three times their quiescent luminosity.On the other hand,some of the ONC?ares were so long(>30ks)that such variability may not have been classed as a?are in this paper.A longer observation of NGC2547is required for a more reliable comparison. A similar?are rate in NGC2547,the ONC and the Pleiades seems at odds with explanations of the enhanced coronal temperatures in very young PMS stars which rely on enhanced?aring rates.On the other hand,only the rate of occurrence of the very brightest?ares have been measured here.It is still possible that the?are energy spectrum has a shallower slope in the ONC than NGC2547or the Pleiades, resulting in coronal heating by more numerous smaller?ares. 6.3.2Long term variations Comparisons of long term variability as a function of age needs to be restricted to those studies which(i)de?ne vari-abilty in terms of an amplitude,rather than stating that some fraction of stars are“variable”which will just be an increasing function of the sensitivity of the observations;(ii) consider upper limits where necessary and do not restrict themselves to stars detected in all observations and(iii)ad-equately account for systematic di?erences in the estimated luminosities from di?erent instruments.Several papers have been found that ful?l these criteria. In section5.2it was established that only10–15per cent of G/K stars or stars with L x>3×1029erg s?1in NGC2547show variations of a factor 2on timescales of 7years.Similarly,Simon&Patten(1998)?nd only2/28 members of IC2391(age?50Myr)have X-ray luminosities that vary by more than a factor of two on a2year timescale. In older clusters Gagn′e et al.(1995a)and Micela et al. (1996)found that?25per cent of low-mass Pleiads(age ?120Myr)showed variations of a factor two or more on timescales of1-2years and perhaps40per cent on timescales of11years.Pillitteri et al.(2005)found that at least8/22 G-and K-type stars in Blanco1(age?100Myr)varied by just over a factor of two on a6year timescale.Marino et al. (2003)showed that older(but still relatively X-ray active) solar-type(F7-K2)?eld stars were even more variable than their counterparts in the Pleiades.Micela&Marino(2003) have shown that the Sun would exhibit variations by factors of?10if observed in a similar manner over the course of its magnetic activity cycle.Favata et al.(2004)have presented the?rst compelling evidence for solar-like X-ray variations in an old?eld G2star,where L x varied by a factor of10in 2.5years. c 2004RAS,MNRAS000,1–23