Modeling the Ne IX Triplet Spectral Region of Capella with the Chandra and XMM-Newton Gratings

arX

iva

stro

-ph

0308

317v

1 1

9 A

ug 2

003

Draft version February 2 2008Preprint typeset using LATEX style emulateapj v 51403

MODELING THE NE IX TRIPLET SPECTRAL REGION OF CAPELLA WITH THE CHANDRA ANDXMM-NEWTON GRATINGS

Jan-Uwe NessHamburger Sternwarte Universitat Hamburg Gojenbergsweg 112 D-21029 Hamburg Germany

Nancy S Brickhouse and Jeremy J DrakeHarvard-Smithsonian Center for Astrophysics 60 Garden Street Cambridge MA 02138

David P HuenemoerderMIT Center for Space Research 70 Vassar Street Cambridge MA 02139

Draft version February 2 2008

ABSTRACT

High resolution X-ray spectroscopy with the diffraction gratings of Chandra and XMM-Newton offersnew chances to study a large variety of stellar coronal phenomena A popular X-ray calibrationtarget is Capella which has been observed with all gratings with significant exposure times Wegathered together all available data of the HETGS (155 ks) LETGS (219ks) and RGS (53 ks) forcomparative analysis focusing on the Ne IX triplet at around 135 A a region that is severely blendedby strong iron lines We identify 18 emission lines in this region of the HEG spectrum including manyfrom Fe XIX and find good agreement with predictions from a theoretical model constructed usingthe Astrophysical Plasma Emission Code (APEC) The model uses an emission measure distributionderived from Fe XV to Fe XXIV lines The success of the model is due in part to the inclusion ofaccurate wavelengths from laboratory measurements While these 18 emission lines cannot be isolatedin the LETGS or RGS spectra their wavelengths and fluxes as measured with HEG are consistentwith the lower resolution spectraIn the Capella model for HEG the weak intercombination line of Ne IX is significantly blended by ironlines which contribute about half the flux After accounting for blending in the He-like diagnosticlines we find the density to be consistent with the low density limit (ne lt 2 times 1010 cmminus3) howeverthe electron temperature indicated by the Ne IX G-ratio is surprisingly low (sim 2 MK) compared withthe peak of the emission measure distribution (sim 6 MK) Models show that the Ne IX triplet is lessblended in cooler plasmas and in plasmas with an enhanced neon-to-iron abundance ratio

Subject headings atomic data ndash line identification ndash stars coronae ndash stars individual (Capella) ndashstars late-type ndash X-rays stars

1 INTRODUCTION

Investigation of stellar coronae in the X-ray wavelengthband has until very recently been restricted to instru-ments with intrinsically low spectral resolution such asthe proportional counters and CCDs on satellites such asEinstein ROSAT ASCA and BeppoSAX1 The limita-tions of low resolution spectra have restricted X-ray stud-ies of stellar coronae to measurements of luminositiesplasma temperatures and estimates of elemental abun-dances The spatial resolution of X-ray emitting plasmathat is routinely available for studies of the solar coronawith satellites such as SOHO and TRACE is still todaya dream of X-ray astronomers

Despite limited spectral information substantialprogress has been made in understanding the gross char-acteristics of stellar coronae throughout the HR diagram

Electronic address jnesshsuni-hamburgdeElectronic address [bhousejdrake]head-cfaharvardeduElectronic address dphspacemitedu

1 Exceptions are the Einstein spectra of Capella with much bet-ter resolution obtained by Mewe et al (1982) with the objectivegrating spectrometer (OGS) covering the range 5 to 30 A with aresolution lt1 A and of Vedder amp Canizares (1983) with the focal-plane crystal spectrometer

For example Schrijver Mewe amp Walter (1984) showedfrom Einstein IPC observations of a sample of 34 late-type stars that coronal temperature was directly corre-lated with X-ray luminosity and stellar rotation rateSimilar results were obtained for a larger sample bySchmitt et al (1990) and later based on the so-calledhardness ratio obtained from ROSAT PSPC observationsby Schmitt (1997)

These results raised a question as to the nature ofthe high temperature plasma in active stellar coronaeVaiana amp Rosner (1978) had pointed out that the Suncompletely covered with active regions would have an X-ray luminosity of sim 2 times 1029 erg sminus1 However the mostactive solar-like stars can have X-ray luminosities up totwo orders of magnitude higher than this and so cannotsimply be scaled-up versions of the solar corona Thehot plasma on the most active stars must be structureddifferently from that in typical solar active regions Theradiative loss of a hot optically-thin collision-dominatedplasma is essentially proportional to the volume emis-sion measure defined as the product of electron densitysquared and the emitting volume n2

eV Increasing ei-ther the emitting volumes or plasma density results inan increase in X-ray luminosity

2 Ness Brickhouse Drake amp Huenemoerder

A key plasma parameter for inferring the size of X-rayemitting regions is therefore the plasma density Directspectroscopic information on plasma densities at coro-nal temperatures on stars other than the Sun first be-came possible with the advent of ldquohighrdquo resolution spec-tra (λ∆λ sim 200) obtained by EUVE that were capableof separating individual spectral lines Even with thisresolution the available diagnostics have often tended tobe less than definitive owing to the poor signal-to-noiseratio of observed spectra or to blended lines This hasbeen especially the case for the active stars Studies ofdensity-sensitive lines of Fe XIX to Fe XXII in EUVEspectra of RS CVn stars revealed tempting evidence forhigh densities of ne sim 1012 to 1013 cmminus3 at coronal tem-peratures near 107 K (Drake 1996 Dupree et al 1993)Such high densities suggest that emitting structures arecompact static loop models such as those described byRosner Tucker amp Vaiana (1978) would have heights ofsim 1000km with confining field strengths of up to 1 kGand surface filling factors of 1 to 10 In the case ofthe evolved active binary Capella Brickhouse (1996) ob-tained ne sim 1012 cmminus3 from Fe XIX to Fe XXII butne sim 109 cmminus3 based on Fe XII to Fe XIV suggestingthat the cool and hot plasma are from distinctly differ-ent structures In contrast to the high densities reportedby Dupree et al (1993) only upper limits were found byMewe et al (2001) (ne lt 2 minus 5 times 1012 cmminus3) using thesame line ratios of Fe XX to Fe XXII but with the higherspectral resolution of Chandra LETGS spectra Theypoint out that the Fe XIX to Fe XXII line ratios are onlysensitive above 1011 cmminus3 such that no tracer for lowdensities for the hotter plasma component is available

The high spectral resolution of Chandra and XMM-Newton has significantly changed the situation regard-ing plasma diagnostics for stellar coronae High spec-tral resolution coupled with large effective area allowsapplication of line-based diagnostic techniques at X-raywavelengths Line ratios of the strong H-like Lyman se-ries which are sensitive to the Boltzmann factor and theHe-like triplets which exploit the competition betweencollisional excitation and recombination-driven cascadescan be useful temperature diagnostics Simply examin-ing the presence or absence of principle lines of differ-ent ionic states of different elements gives an indicationof the emitting plasma temperatures The He-like sys-tems also provide density diagnostics based on the low-lying metastable level 1s2s 3S1 above the 1s2 1S0 groundstate The He-like oxygen diagnostic has been used tosupport the previous EUVE result for Capella show-ing that the lower temperature plasma (sim2 MK) alsohas lower density (Audard et al 2001 Brickhouse 2002Brinkman et al 2000 Canizares et al 2000 Ness et al2001)

We compare the capabilities of this current generationof high spectral resolution X-ray instruments with a de-tailed study of the Ne IX triplet as a density and tem-perature diagnostic We focus our study on the Capellabinary system (HD 34029 α Aurigae G8 III + G1 III)Capella has been extensively studied from X-ray to radiowavelengths and is the brightest steady coronal sourcein the X-ray sky As such it has been a key calibrationtarget for the Chandra high-energy and low-energy trans-mission grating spectrometers (HETGS and LETGS re-spectively) as well as the XMM-Newton reflection grat-

ing spectrometers (RGS) and has been observed on sev-eral occasions by both satellites (Audard et al 2001Brinkman et al 2000 Canizares et al 2000) The emis-sion measure distribution of Capella shows a steep butnarrow enhancement at 6 MK (Brickhouse et al 2000Dupree et al 1993) making it ideal for studying theblending of Ne IX with high temperature lines

The purpose of this study is threefold (i) assess theaccuracy and reliability of state-of-the-art plasma radia-tive loss models for describing the emission spectrum ofCapella in the region of the He-like complex of Ne (ii)determine how well these models might describe the spec-tra of stars both more and less active than Capella andwith different coronal temperatures and (iii) gain furtherinsight into the plasma density in the Capella coronae inthe key temperature range 4 to 6MK

2 HE-LIKE IONS IN HIGH RESOLUTION STELLARSPECTRA

21 The Use of the He-like Triplet Diagnostics

He-like ions are produced at temperatures ranging be-tween sim 2MK for C V and N VI and 10MK for Si XIIIThe densities over which the diagnostics are sensitive in-crease with increasing element number Z such that thelower Z ions C V N VI and O VII with temperaturesof peak emissivity at 10 16 and 20MK respectivelyprovide diagnostics for densities up to sim 1012 cmminus3 Thehigher temperature (gt 6 MK) ions Mg XI and Si XIIIare sensitive at densities above sim 1012 cmminus3 Ne IXtypically formed at sim 4MK is sensitive to densitiesbetween sim 1011 and 1013 cmminus3 In many cases mea-surements of Ne IX are the only chance to fill the gapbetween the cooler plasma and the hotter plasma Inaddition some stars show a neon overabundance (egBrinkman et al 2001 Drake et al 2001 Gudel et al2001 Huenemoerder Canizares amp Schulz 2001) makingNe IX lines more easily detectable in those cases How-ever blending with Fe lines in the Ne IX triplet spectralregion can compromise the diagnostic utility when ana-lyzing plasmas hot enough to produce them (especiallyFe XIX at sim 6 MK)

The theory of the He-like triplets was originally devel-oped by Gabriel amp Jordan (1969) The density diagnos-tic is often referred to as the R-ratio where R = fidenoting the forbidden line flux with f and the inter-combination line flux with i The relation can be param-eterized as

R =R0

1 + neNc

(1)

with low-density limit R0 and electron densityne The critical density Nc is the density atwhich R = 12R0 Pradhan amp Shull (1981) andBlumenthal Drake amp Tucker (1972) calculated theo-retical values of R0 and Nc and determined the Z-dependence of the density-sensitivity as shown in Fig-ure 1 Larger R-ratios mean relatively weaker intercom-bination lines since the i is intrinsically weak at lowdensities it may be difficult to measure accurately

Strong ultraviolet radiation fields can also de-populatethe metastable levels of the He-like triplets and changethe R-ratios For the case of Capella and Ne IX suchfields can be neglected (Ness et al 2001)

For an optically thin plasma a temperature diagnostic

Modeling the Ne IX Triplet with HETGS LETGS and RGS 3

Fig 1mdash The low-density limits R0 and the critical densitiesNc for the different He-like ions show systematic anti-correlatedtrends with wavelength due ultimately to the nuclear charge (fromBlumenthal Drake amp Tucker 1972 Pradhan amp Shull 1981) LargeR0 (C V) implies that the i flux may be difficult to measure LargeNc may be outside the interesting coronal range (Si XIII) Ne IXappears to have ideal values for coronal physics

is

G =i + f

r (2)

where r is the resonance line flux In collisional ionizationequilibrium (CIE) the G-ratio decreases with tempera-ture primarily because the excitation of the triplet lev-els (by dielectronic recombination-driven cascades) de-creases faster than the collisional excitation to the 1P1

(see Smith et al 2001) Assuming a priori that theplasma is in CIE the G-ratio is a direct diagnostic ofthe temperature of Ne IX emission and can be comparedto predictions based on an emission measure distribution

The He-like triplets have been used for measuring tem-peratures and densities in the solar corona Both theO VII and Ne IX triplets have been studied for quiescentemission and flares (eg Acton et al 1972 Keenan et al1987) In solar flares the Ne IX lines appeared blendedMcKenzie (1985) and Doyle amp Keenan (1986) suggestedseveral candidate lines for these blends

Generally stars with lower temperature coronae do notshow significant blending around the Ne IX triplet Pro-cyon (Ness et al 2002b) is a good example In hottercoronae the blending is more problematic To date theR-ratios measured from MEG spectra of Capella havenot been used to determine densities (Ayres et al 2001Phillips et al 2001) Ness et al (2002a) have attemptedto disentangle the blending for the LETGSHRC-S spec-trum of Algol by assuming a priori a G-ratio of 08

22 Comparison among the New Instruments

The new generation of X-ray telescopes of XMM-Newton and Chandra has opened new dimensions in sen-sitivity and resolution Spectroscopic measurements canbe carried out with both instruments using the intrinsicresolution of the CCDs (ACIS EPIC) and the disper-sive instruments for higher resolution Three gratingsare providing data the RGS on board XMM-Newtonand the HETGS and LETGS on Chandra The HETGS

consists of two sets of gratings with different periods theHigh Energy Grating (HEG) and the Medium EnergyGrating (MEG) which intercept X-rays from the innerand outer mirror shells respectively and thus are usedconcurrently The RGS and HETGS use the EPIC andACIS-S CCD detectors respectively while the LETGScan use either ACIS-S or the micro-channel plate detectorHRC-S Figure 2 shows the effective areas and the wave-length ranges for the grating instruments While RGSHEG and MEG operate in the spectral range 40 ALETGS covers a much larger wavelength range from 2 to175 A All the gratings have uniform resolving power fortheir entire wavelength range

Fig 2mdash Effective areas for the Chandra and XMM-Newtongratings with ranges of other observatories shown as dotted linesAt 136 A the RGS has the largest area while HEG is the leastsensitive The bottom panel shows detail in the Ne IX region

As can be seen from Figure 2 the large wavelengthrange covered by LETGS allows the extraction of totalfluxes luminosities and hardness ratios corresponding toROSAT (5 ndash 124 A) or Einstein (3 ndash 84 A) With LETGSall He-like triplets ranging from C V and N VI up toSi XIII can be measured simultaneously LETGS also ob-tains ∆n = 0 Fe L-shell lines including density-sensitivelines of Fe XIX to Fe XXII previously obtained at fivetimes lower spectral resolution with EUVE

Since LETGS cannot sort orders via detector intrinsicenergy resolution integrated fluxes and luminosities arestill source-model-dependent For a log T = 68 plasma


the apparent energy flux obtained by dividing counts bythe first order effective area is about 20 larger than thetrue first order flux when 11 orders are accounted forIn this ideal isothermal model only 80 of the countsare from first order

An advantage of the HETGS is its very high spec-tral resolution near 1 keV In addition order sorting forHEG MEG and the RGS is possible using the energyresolution of the CCD detectors which is not possiblefor the LETGSHRC-S configuration This allows accu-rate determination of source-model-independent broad-band fluxes and luminosities

The RGS covers a wavelength range similar to Einsteinwith especially large effective areas above 10 A (Fig 2)The spectral resolution of the RGS is comparable to thatof LETGS in their region of overlap

The mirror point spread function primarily defines thedispersed line profile of the HEG MEG and LEG Inthe case of the RGS significant scattering wings alsoarise from the reflection gratings themselves Gratingperiod variance detector pixelization and aspect recon-struction are additional factors For Ne IX the RGSoffers the highest sensitivity even in second dispersionorder MEG and LETGS have larger effective areas thanHEG The spectral resolution of the HEG however isthe most important factor for a realistic assessment ofthe effects of blending Here we explore the diagnosticutility for observations of Ne IX with each instrument

23 Spectral Models

Atomic data are fundamental to interpretation of theobserved spectra We compare observations with modelsproduced using the Astrophysical Plasma Emission Code(Smith et al 2001) Version 122 APEC incorporates col-lisional and radiative rate data appropriate for modelingoptically thin plasmas under the conditions of collisionalionization equilibrium The code solves the level-to-levelrate matrix to obtain level populations and produces lineemissivities as functions of electron temperature and den-sity APEC also calculates spectral continuum emissionfrom bremsstrahlung radiative recombination and two-photon emission

We rely on APEC inclusion of the H- and He-likeatomic data discussed by Smith et al (2001) the HUL-LAC calculations of D Liedahl for the Fe L-shell ionswith additional R-matrix calculations available fromCHIANTI V20 (Landi et al 1999) and isosequence scal-ing for Ni L-shell ions Our models assume the ionizationbalance of Mazzotta et al (1998) and the solar abun-dance model of Anders amp Grevesse (1989) Referencewavelengths are from quantum electrodynamic calcula-tions for H- and He-like ions (Drake 1988 Ericsson 1977)and from laboratory measurements for emission lines ofFe L-shell (Brown et al 1998 2002) and Ni L-shell ions(Shirai et al 2000) Additional L-shell wavelengths arederived from HULLAC energy levels

Despite significant improvements to spectral model-ing over the past several years the theoretical atomicdata in APEC and other plasma models remain largelyuntested over the broad range of applicable densitiesand temperatures The deep observations of three late-type stars (Capella Procyon and HR 1099) used for

2 Available at httpcxcharvardeduatomdb

in-flight calibration measurements of the dispersion rela-tion and line spread functions of the gratings are alsouseful for determining the extent of agreement betweenmodels and observations and for highlighting issues thatmight require additional atomic physics work The ac-curacy of each rate as well as the completeness of linelists is important to the correct interpretation of spec-tral diagnostics especially at lower spectral resolution(eg Brickhouse et al 2000) This work is part of a com-prehensive effort known as the ldquoEmission Line Projectrdquoto benchmark the atomic data in plasma spectral mod-els (Brickhouse amp Drake 2000) Work to date on theCapella HETGS spectra has focused on the identificationof strong lines for which HEG and MEG give good agree-ment (eg Ayres et al 2001 Behar Cottam amp Kahn2001 Canizares et al 2000) while we aim to identifyweak lines and test a comprehensive model

Atomic data for the He-like diagnostic lines themselvesare exceptionally good mdash many rates have been bench-marked in controlled laboratory experiments Toka-mak data confirm the calculation of the Ne IX R-ratio (Coffey et al 1994) The transition probabilityfor the forbidden line which determines the density-sensitivity of the Ne IX R-ratio has been measuredto 1 accuracy on an electron beam ion trap (EBIT)(Wargelin Beiersdorfer amp Kahn 1993) it has also beenconfirmed by Bragg crystal spectrometer measurementson the EBIT (Wargelin 1993) Smith et al (2001)showed that the largest systematic uncertainties in thetheory for O VII come from the limited number of energylevels used in calculating the cascades following dielec-tronic recombination

APEC models for Ne IX lines are in good agreementwith the laboratory measurements Simplifying assump-tions often found in the literature such as the lack oftemperature-sensitivity in the R-ratio are no longer nec-essary since the APEC models are calculated over densetemperature and density grids

For weak diagnostic lines such as the Ne IX in-tercombination line two challenging line measurementissuesmdashdetermining the continuum level and assessingline blendingmdashare greatly aided by APEC The APECline list includes approximately 40 lines with referencewavelengths between 1335 and 1385 A Furthermorethe APEC emissivity table contains sim 1000 lines in thatregion down to 4 orders of magnitude weaker than thestrong resonance line of Ne IX This list includes Ne IXdielectronic recombination satellite lines and Fe and NiL-shell lines with principal quantum number n le 5APEC is sufficiently complete over the HETGS bandpassthat it can be used to select line-free spectral regions forcontinuum fitting

3 OBSERVATIONS AND DATA REDUCTION

Our approach is to use the six observations of HEG cal-ibration data of Capella to obtain a very long total expo-sure time (154685ks) and benchmark the APEC mod-els in the wavelength region around 136 A The Capellaobservations discussed in this paper are summarized inTable 1 For each of the three grating instruments thedata from several observations have been combined toproduce high signal-to-noise spectra Standard pipelineprocessing for HETGS LETGS and RGS are describedin Canizares et al (2000) Brinkman et al (2001) and


Audard et al (2001) respectivelyThe HETGSACIS-S data were obtained from the

Chandra archive and reprocessed with CIAO softwareVersion 223 and calibration database CALDB Version210 The standard technique for separating the differ-ent grating orders employs a variable-width CCD pulse-height filter4 that follows the pulse-height versus wave-length relation for each spectral order Since some obser-vations were made during times of uncertain or changingdetector response (primarily CCD temperature changesthat affected the gain) the standard filter regions didnot trace the actual distributions of events in the pulse-height-wavelength plane very accurately To mitigatethis we used the constant fractional energy width op-tion available in the CIAO event-resolution program(tg resolve events) with a fractional width of 03 Theconstant width region accepts more inter-order back-ground particularly at short wavelengths but this isnegligible for our purposes because of the very low back-ground level We have also bypassed the step of po-sition randomization of events detected within a givenpixel which is part of the standard image processing ofACIS-S Otherwise we use the standard CIAO pipelineextraction We use only the first dispersion ordersThe effective areas for positive and negative first ordershave been computed for each dataset using observation-specific data to account for aspect and bad pixels TheHEG effective areas for the individual observations in thewavelength range around 13 A are shown in Figure 3along with the exposure-weighted average effective areato be used for analysis of the combined spectra

The LETGS observations and data reduction are de-scribed in Ness et al (2001) They consist of ninedatasets added with a total exposure time of 21854ksEffective areas are from 2001 February (see Pease et al2000 for a preliminary description) The RGS obser-vations were taken on 2000 March 25 with an expo-sure time of 5292 ks Data processing is described inAudard et al (2003) Effective areas have been calcu-lated for the observations with SAS53

We derive X-ray luminosities given in Table 1 of LX =13 times 1030 28 times 1030 and 17 times 1030 ergs sminus1 using thedistance of 1294 plusmn 015pc (Perryman et al 1997) Dif-ferences are consistent with the different passbands andresponses of the instruments Table 1 also shows that theluminosities for HETGS and LETGS are fairly constantwith time (for HETGS over 15 years) supporting ourcomparison of spectra taken with different instrumentsat different times

4 DATA ANALYSIS

41 Line Flux Measurements

Table 2 lists 18 strong emission line features measuredin the plus and minus first order HEG spectra between1335 and 1385 A including several features not previ-ously reported for Capella Line counts are measuredwith the CORA line fitting tool developed by Ness et al(2001) and described and refined by Ness amp Wichmann(2002) The raw counts obtained with CORA representthe fitted number of expected counts for a given con-

3 Available at httpcxcharvardeduciao4 Encoded in the calibration database as the Order Sorting and

Integrated Probability or OSIP file

Fig 3mdash Effective areas for the different observations (labeledwith observation identification) with the HEG plus first order (top)and minus first order (bottom) The average is indicated by a thickline and is obtained by weighting with the exposure times Thelarge dip on the short wavelength end of the minus first order isthe chip gap while smaller dips are caused by the removal of badpixels The widths and shapes of these dips are determined by theaspect dither and the chip geometry

tinuum (plus background) requiring Poissonian statis-tics to be conserved Measurement errors are given as1 σ errors and include statistical errors and correlatederrors in cases of line blends but include no system-atic errors from eg the continuum placement Linewidths are fixed rather than fit since line broadening isexpected to be predominantly instrumental and line pro-files are represented by analytical profile functions (Gaus-sians for HEG and MEG Lorentzians for RGS and forthe LETGS modified Lorentzians F (λ) = a(1 + λminusλ0

Γ )β

with β = 25 Kashyap amp Drake (2002)) The continuumis chosen as an initial guess to be a constant with leveldetermined by the HEG region at 136 A which containsno apparent lines in the data as well as no significantemission lines in the APEC database (The placement ofthe continuum is discussed further in sect43)


Table 1 List of Observations

Obs ID Exp Time Countsa Observation Start - End log LXb

(ks) (YYYY-MM-DDThhmmss) [erg sminus1]

Chandra HETGS HEG MEG HEG MEG00057 28827 11714 47549 2000-03-03T162853 - 2000-03-04T011803 3010 301901010 29541 10773 42907 2001-02-11T122249 - 2001-02-11T212017 3006 301201099 14565 5934 23634 1999-08-28T075317 - 1999-08-28T121637 3012 302001103 40479 18109 73303 1999-09-24T060921 - 1999-09-24T182256 3016 302401235 14571 5916 24177 1999-08-28T121637 - 1999-08-28T163957 3012 302101318 26701 11133 46338 1999-09-25T132639 - 1999-09-25T215117 3014 3022

HETGS Sum 154685 63579 257908 middot middot middot 3012 3020Chandra LETGS LEG LEGc

01167 1536 41215 middot middot middot 1999-09-09T131006 - 1999-09-09T172608 3045 middot middot middot01244 1237 32981 middot middot middot 1999-09-09T174227 - 1999-09-09T210836 3045 middot middot middot01246 1500 41701 middot middot middot 1999-09-10T030606 - 1999-09-10T071608 3047 middot middot middot01248 8536 240047 middot middot middot 1999-11-09T134224 - 1999-11-10T132505 3043 middot middot middot01420 3030 83533 middot middot middot 1999-10-29T224929 - 1999-10-30T071427 3044 middot middot middot62410 1133 31491 middot middot middot 1999-09-09T234357 - 1999-09-10T025248 3048 middot middot middot62422 1168 31202 middot middot middot 1999-09-12T182642 - 1999-09-12T214120 3046 middot middot middot62423 1480 39328 middot middot middot 1999-09-12T233744 - 1999-09-13T034428 3045 middot middot middot62435 2234 58201 middot middot middot 1999-09-06T003540 - 1999-09-06T064801 3042 middot middot middot

LETGS Sum 21854 599699 middot middot middot middot middot middot 3041 middot middot middotXMM-Newton RGS1 RGS2 RGS1 RGS20121920101d 5292 169423 186267 2000-03-25T113659 - 2000-03-26T025349 3022 30300121920101e 5292 45115 44524 2000-03-25T113659 - 2000-03-26T025349 3029 3015

aCounts are background subtracted values The two dispersion directions are co-addedbLX was determined within the wavelength range 3 to 20 AcHigher order photons are included without correctiondFirst ordereSecond order

Table 2 Best-fit Line Fluxes for Separate Dispersion Orders of HEG

λ Amplitudea Fluxb Aeff λ Amplitudea Fluxb Aeff

(A) (cts) (ph cmminus2 ksminus1) (cm2) (A) (cts) (ph cmminus2 ksminus1) (cm2)

+1st order -1st order13449 plusmn 0002 3159 plusmn 184 0511 plusmn 0030 400 13449 plusmn 0003 3006 plusmn 179 0474 plusmn 0028 41013553 plusmn 0008 1017 plusmn 108 0169 plusmn 0018 388 13555 plusmn 0006 899 plusmn 101 0148 plusmn 0017 39213700 plusmn 0004 1907 plusmn 144 0331 plusmn 0025 372 13700 plusmn 0005 1828 plusmn 140 0312 plusmn 0024 37813358 plusmn 0021 311 plusmn 64 0050 plusmn 0010 407 13356 plusmn 0020 311 plusmn 64 0048 plusmn 0010 42113381 plusmn 0007 406 plusmn 72 0065 plusmn 0012 404 13378 plusmn 0007 395 plusmn 70 0061 plusmn 0011 41713404 plusmn 0007 520 plusmn 80 0084 plusmn 0013 399 13404 plusmn 0007 602 plusmn 84 0097 plusmn 0013 40213427 plusmn 0009 640 plusmn 87 0103 plusmn 0014 401 13426 plusmn 0007 748 plusmn 93 0117 plusmn 0015 41313469 plusmn 0004 1348 plusmn 125 0219 plusmn 0020 398 13468 plusmn 0004 1453 plusmn 129 0236 plusmn 0021 39813507 plusmn 0002 1936 plusmn 150 0318 plusmn 0025 393 13507 plusmn 0002 2040 plusmn 157 0330 plusmn 0025 40013524 plusmn 0002 3160 plusmn 187 0522 plusmn 0031 391 13523 plusmn 0001 2617 plusmn 174 0425 plusmn 0028 39813650 plusmn 0008 712 plusmn 91 0122 plusmn 0016 378 13646 plusmn 0007 708 plusmn 92 0119 plusmn 0015 38513675 plusmn 0009 558 plusmn 83 0096 plusmn 0014 375 13673 plusmn 0009 421 plusmn 74 0073 plusmn 0013 37513722 plusmn 0006 547 plusmn 82 0096 plusmn 0014 370 13722 plusmn 0008 477 plusmn 77 0082 plusmn 0013 37613740 plusmn 0008 422 plusmn 73 0074 plusmn 0013 368 13742 plusmn 0008 502 plusmn 78 0089 plusmn 0014 36613779 plusmn 0004 984 plusmn 109 0175 plusmn 0019 364 13778 plusmn 0007 943 plusmn 105 0166 plusmn 0019 36713797 plusmn 0003 1466 plusmn 129 0262 plusmn 0023 362 13797 plusmn 0006 1171 plusmn 114 0205 plusmn 0020 36913828 plusmn 0004 1858 plusmn 141 0335 plusmn 0025 359 13826 plusmn 0003 1720 plusmn 138 0313 plusmn 0025 35513847 plusmn 0007 504 plusmn 80 0091 plusmn 0014 357 13843 plusmn 0006 554 plusmn 85 0102 plusmn 0016 350

aMeasured line counts with 1σ errors The line widths are all 0005 A and the assumed continuum level (including background)for each dispersion order is 350 cts Aminus1bThe total exposure time was 1547 ks

The wavelengths and line counts are listed in Table 2along with fluxes determined using the total exposuretime and effective areas given in the last column for eachline We find the spectra from both dispersion directionsto agree reasonably well and use the plus and minus firstorder-summed spectrum for further analysis (Table 3)The summed spectrum is shown in Figure 4 along withthe empirical model ie continuum plus 18 lines with

best-fit centroids and fluxesTo predict count spectra from the other instruments

we convolved the HEG empirical model with each instru-mental response We allowed global wavelength shifts(dλ) to account for different absolute dispersion calibra-tions and a normalization correction (Scal) for differ-ences in effective area calibration A χ2 minimizationadjusts these two parameters to transform the model to


Table 3 Line Identification for HEG Measurementsa

λbobs λc

ref λcerr Ion Transition Fluxobs Fluxd

model

(A) (A) (A) ndash (ph cmminus2 ksminus1) (ph cmminus2 ksminus1)

13354 13355 0009 Fexviii 2p5 2P32 ndash 2s2p5(3P )3p 2P32 0048plusmn0007 0027613377 13385 0004 Fexx 2s2p4 4P52 ndash 2s2p3(5S)3d 4D72 0062plusmn0008 00577

13390 middot middot middot Fexx 2p3 2P32 ndash 2p122p323d52 middot middot middot 0004613401 13395 middot middot middot Fexviii 2p5 2P32 ndash 2s2p5(3P )3p 2D52 0091plusmn0009 00459

13407 0004 Fexviii 2p5 2P32 ndash 2s2p2

122p3

323p32 middot middot middot 00238

13409 middot middot middot Fexx 2s2p4 4P52 ndash 2s2p2

122p323d52 middot middot middot 00169

13418 middot middot middot Fexx 2s2p4 4P32 ndash 2s2p122p2

323s middot middot middot 00108

13424 13423 0004 Fexix 2p4 3P2 ndash 2p3(2D)3d 1F3 0110plusmn0010 0050413431 middot middot middot Fexxi 2s2p3 3S1 ndash 2s2p122p323s middot middot middot 0008013432 middot middot middot Fexxi 2p3 2P12 ndash 2p2(3P )3d 2P32 middot middot middot 00042

13446 13447 0004 Ne ix 1s2 1S0 ndash 1s2p 1P1 0492plusmn0021 0397813465 13462 0003 Fexix 2p4 3P2 ndash 2p3(2D)3d 3S1 0228plusmn0015 0114513504 13497 0005 Fexix 2p4 3P2 ndash 2p122p2

323d32 0325plusmn0018 02008

13507 0005 Fexxi 2s2p3 3D1 ndash 2s2p2


13521 13518 0002 Fexix 2p4 3P2 ndash 2p3(2D)3d 3D3 0471plusmn0021 0442513551 13550 0005 Ne ix 1s2 1S0 ndash 1s2p 3P2 0158plusmn0012 00021

13551 0005 Fexix 2p4 3P2 ndash 2p122p2

323d52 middot middot middot 00290

13553 0005 Ne ix 1s2 1S0 ndash 1s2p 3P1 middot middot middot 0053113554 middot middot middot Fexix 2p4 3P2 ndash 2p2




13645 13645 0004 Fexix 2p4 3P2 ndash 2p3(2D)3d 3F3 0118plusmn0011 0071213648 middot middot middot Fexix 2p4 3P2 ndash 2p3(4S)3d 3D3 middot middot middot 00175

13671 13674e middot middot middot Fexix 2p4 3P1 ndash 2p122p2

323d52 0085plusmn0010 00133

13675 middot middot middot Fexix 2p4 3P1 ndash 2p122p2




13697 13699 0005 Ne ix 1s2 1S0 ndash 1s2s 3S1 0322plusmn0017 0184513719 13732f middot middot middot Fexix 2p4 3P1 ndash 2p122p2

323d52 0089plusmn0010 00573

13738 13746 middot middot middot Fexix 2p4 1D2 ndash 2p3(2D)3d 1F3 0081plusmn0010 0069013775 13767 0005 Fexx 2p3 4S32 ndash 2p2(3P )3s 4P52 0170plusmn0013 00494

13779 0005 Nixix 2p6 1S0 ndash 2p5(2P )3s 1P1 middot middot middot 0067613794 13795 0005 Fexix 2p4 3P2 ndash 2p122p2

323d52 0234plusmn0015 01778

13795 0005 Fexix 2p4 1D2 ndash 2p3(2D)3d 3P2 middot middot middot 0020713824 13825 0002 Fe xvii 2p6 1S0 ndash 2s2p63p 1P1 0325plusmn0018 03058

13839 0005 Fexix 2p4 3P2 ndash 2p2


13839 0005 Fexix 2p4 1D2 ndash 2p2


13843 13843 0006 Fexx 2p3 4S32 ndash 2p2(3P )3s 4P32 0094plusmn0011 00233

aOnly the strongest lines modeled are identified In cases of line blends weaker lines up to about 10 of the strong line flux are alsolistedbMeasured HEG wavelengths have been corrected according to the scalingcSources for reference wavelengths with errors are given in the textdModel fluxes are based on the emission measure distribution derived from the lines in Table 4 (shown in Fig 7)eBrown et al (1999) list a group of 5 lines (ldquoO21rdquo) with a central wavelength 13676 plusmn 0004 although APEC has not assigned this

wavelength to any of the model linesfAPEC includes two relatively strong Ne ix satellite lines which may contribute to a centroid shift

each instrument Figure 5 shows the MEG and LETGSmodel and data while Figure 6 shows RGS1 and RGS2The instrumental line spread function width (σ) was nota free parameter since it is well determined by calibra-tion The continuum (cont) was also not a free parameterbut was set at the flux level determined from the HEGspectrum We find excellent agreement between the ob-served counts and the transformed HEG model withinthe systematic errors expected from calibration

42 Emission Measure Distribution from the HEGSpectrum

In order to compare our measurements with models weconstruct a rough emission measure distribution using se-

lected iron lines from ionization stages Fe XV to Fe XXIV(see Table 4) We use the strongest line from each ion(except for Fe XVII) and use other lines to check themodel for consistency The EUVE line fluxes are givenby Ayres et al (2001) and are corrected for an interstel-lar column density NH = 18 times 1018 cmminus2 (Linsky et al1993) Assuming collisional ionization equilibrium weconstruct an emission measure distribution at low density(ne = 10 cmminus3) using the APEC line emissivities ǫ(T )Each emission line specifies an emission measure curve4πd2 Fluxobsǫ(T ) which represents the emission mea-sure as if each temperature contributed the total emis-sion in the line It is an upper limit to the final self-consistent emission measure distribution determined by


Fig 4mdash The HEG summed spectrum of Capella using plus andminus first order In order to construct an empirical model eigh-teen emission lines are used with Gaussian line widths σ = 0005 A(equivalent to FWHM= 0012 A) The continuum level (includingweak lines and background for combined plus and minus first or-ders) is chosen constant with 700 cts Aminus1 The binsize is 25mAand the exposure time is 1547 ks

using the lower envelope of the ensemble of curves as theinitial estimate in an iterative optimization to a solutionwhich predicts all measured line fluxes Figure 7 showsthe individual emission measure curves with the best-fitmodel The figure also compares an earlier emission mea-sure distribution obtained from simultaneous EUVE andASCA observations (Brickhouse et al 2000)

43 Modeling the Continuum

Up to this stage of the analysis the continuum hasbeen taken as a constant value empirically determinedfrom the spectrum Since the value of the continuumunder the weak lines is crucial to the proper derivationof the R-ratio (see Brickhouse 2002) we discuss our con-tinuum modeling method in some detail In this sectionwe derive a formal result which turns out to be close toour initial estimate in general iteration to subtract thenew continuum flux from lines might be necessary

In principle background weak lines and continuumcan be treated separately in the analysis The back-ground which includes non-source events from the detec-tor cosmic rays as well as source photons redistributedby mirror scatter detector aliasing (CCD readout streak-ing) and grating scatter is more than a factor of 20lower than the source-model continuum we derive TheHETGS spectra are not background-subtracted and sothe background is implicitly included in the continuummodel However the HETGS background rejection isvery high background near 13 A is estimated5 to beless than 0017 cts ksminus1 Aminus1 arcsecminus1 and the extrac-tion width is 5 arcsec Hence we only expect about 13cts Aminus1 per order in the summed HEG spectrum Weaklines not directly measured (ie not listed in Table 3)are treated in sect52 as a source of systematic uncertaintyto the diagnostic line ratios

5 See the ldquoProposersrsquo Observatory Guiderdquo Section 83httpcxcharvardeduproposerPOG

Fig 5mdash Best-fit model obtained from the HEG spectrumscaled and overlaid on the measured spectra from MEG (top) andLETGS (bottom) The scaling parameters dλ Scal σ and cont(described in the text) are listed in the upper right box The good-ness of fit χ2N is also given For MEG the binsize is 50mA andthe exposure time is 1547 ks For LETGS the binsize is 100mAand the exposure time is 2185 ks

The emission measure distribution constructed in sect42was used to predict both continuum and line fluxesunder the assumption of solar abundances The tabu-lated APEC linelist includes only lines with emissivitiesgreater than 10 times 10minus20 ph cm3 sminus1 weaker lines areincluded as a ldquopseudo-continuumrdquo and are added to thecontinuum emission spectrum from bremsstrahlung ra-diative recombination and two-photon emission Theresulting model continuum is flat between 13 and 14 Aand is about 30 lower than the continuum level usedin sect41 (an initial guess chosen to match at 136 A) adifference of only about 3 counts per emission line

This model continuum spectrum was then fit to aset of line-free spectral regions defined as spectralbins for which the model line flux is less than 20of the model continuum flux Both HEG and MEGwere fit jointly using the Sherpa package in CIAOwith the Cash statistic appropriate for low-count bins(Freeman Doe amp Siemiginowska 2001) With the con-tinuum shape fixed by the emission measure distributiononly a single parameter for the continuum level or ldquonor-malizationrdquo was allowed to vary The resulting fit con-tinuum level was 6 higher than the model continuumlevel calculated assuming solar abundances Models with


Table 4 HEG Iron Line Measurements Useful for Emission MeasureDistribution

λobs λref Ion Transition Counts Fluxobs Aeff

(A) (A) ndash (ph cmminus2 ksminus1) (cm2)

28415a 28416 Fexv 3s2 1S0 ndash 3s3p 1P2 middot middot middot 105b middot middot middot

33541a 33541 Fexvi 3s 2S12 ndash 3p 2P32 middot middot middot 339b middot middot middot

15262a 15261 Fexvii 2p6 1S0 ndash 2p5(2P )3d 3D1 841008 plusmn 2968 1128 plusmn 0040 481915015 15014 Fexvii 2p6 1S0 ndash 2p5(2P )3d 1P1 271934 plusmn 5271 3354 plusmn 0065 524117098 17096 Fexvii 2p6 1S0 ndash 2p5(2P )3s 3P2 523964 plusmn 2295 2459 plusmn 0108 137714207a 14208c Fexviii 2p5 2P32 ndash 2p4(1D)3d 2D52 134792 plusmn 3792 1331 plusmn 0037 6547middot middot middot 14208c Fexviii 2p5 2P32 ndash 2p122p3

323d52

16075 16071 Fexviii 2p5 2P32 ndash 2p4(3P )3s 4P52 286749 plusmn 1728 0868 plusmn 0052 213514670a 14664 Fexix 2p4 3P2 ndash 2p3(2D)3s 3D3 164356 plusmn 1401 0190 plusmn 0016 558616110 16110 Fexix 2s2p5 3P2 ndash 2p122p2

323p12 42796 plusmn 722 0136 plusmn 0023 2037

12847a 12846 Fexx 2p3 4S32 ndash 2p122p323d32 372794 plusmn 2084 0222 plusmn 0012 1087512828 12824 Fexx 2p3 4S32 ndash 2p122p323d32 248643 plusmn 1842 0148 plusmn 0011 10882middot middot middot 12827 Fexx 2p3 4S32 ndash 2p122p323d52

middot middot middot 12822 Fexxi 2s2p3 3D1 ndash 2s2p122p323d52

12286a 12284 Fexxi 2p2 3P0 ndash 2p3d 3D1 482550 plusmn 2327 0233 plusmn 0011 1340211771a 11770 Fexxii 2s22p 2P12 ndash 2s23d 2D32 177567 plusmn 1480 0077 plusmn 0006 1490112750 12754 Fexxii 2s2p2 2D32 ndash 2s2p123s 90500 plusmn 1040 0050 plusmn 0006 1161011741a 11736 Fexxiii 2s2p 1P1 ndash 2s3d 1D2 107723 plusmn 1210 0047 plusmn 0005 1466611176a 11176 Fexxiv 2p 2P32 ndash 3d 2D52 90073 plusmn 1132 0032 plusmn 0004 18211

aThis line is used to construct the emission measure distributionbThese measurements are from EUVE (Ayres et al 2001)cSum of both lines in the blend are used in the model

stricter criteria for determining line-free regions give sim-ilar results The continuum level we adopt between 13and 14 A thus seems robust

Since the emission measure distribution was deter-mined only from iron lines its absolute value should belowered by 6 to give agreement with the fit continuumlevel requiring a 6 higher iron-to-hydrogen abundanceratio One can draw the preliminary conclusion fromthis analysis that the iron-to-hydrogen abundance ratiois close to solar determining accurate abundance ratiosrequires additional checks from other spectral featuresand systematic error analysis which is beyond the scopeof this paper

5 RESULTS AND INTERPRETATION

51 Identification of the Emission Lines

Using our derived emission measure distribution wepredict the spectrum of Capella under the assumptionsof solar abundances and low density We convolve thismodel spectrum with the instrumental response allow-ing only for the small systematic global shift in wave-lengths and compare it to the observed HEG spectrumThe model fluxes are given in Table 3 along with themeasured fluxes and the measured spectrum with thismodel is shown in Figure 8 While the model is cal-culated at ne = 10 cmminus3 we have used APEC V13density-dependent calculations to check the sensitivity ofour model iron spectrum For ne lt 108 cmminus3 no effectslarger than a few percent are found A few weak linesin the model show significant density-sensitivity above108 cmminus3 but no observable lines show more than 20change up to ne = 1013 cmminus3

Except for one nickel line only iron and neon lines arepresent in our model of this spectral region The most

striking discrepancies are seen for the Fe XIX lines at13462 A (flux underpredicted by sim a factor of two) at13497 A (flux of blend underestimated by sim one thirdwavelength discrepancy consistent with errors on labora-tory wavelengths) and around 1372 to 1375 A (wave-length discrepancies well within theoretical errors nolaboratory wavelengths available) Minor disagreementis seen around 134 A The nickel line at 13779 A is un-derpredicted by a factor of two probably an abundanceeffect Since the goal is to assess blending of the neondiagnostic lines the overall level of agreement is a goodindicator of our knowledge of the blending spectrum (es-pecially for Fe XIX) We note that (1) all strong pre-dicted lines are accounted for in the observed spectrum(2) the worst flux disagreement is a factor of two (3)laboratory iron wavelengths agree with the observationswithin the errors while theoretical iron wavelengths ap-pear to agree to within one resolution element of theHEG spectrometer

Since we have derived the emission measure distribu-tion from iron lines we can allow the three neon diag-nostic line fluxes to vary with independent scaling fac-tors A([r i f ]) in order to obtain best-fit neon abun-dances and densities These parameters are iteratedwith a χ2 minimization After applying the scaling fac-tor A(r) = 103 for the Ne IX resonance line we findFluxmodel = 041 ph cmminus2 ksminus1

The intercombination line (with an adjusted flux 0075ph cmminus2 ksminus1) is significantly weaker in the model thanin the observed spectrum This is of course explainedby the contaminating lines of Fe XIX and Fe XX thatcontribute to the observed feature Table 3 and Figure 9show that the iron lines can contribute almost half of the


Fig 6mdash Same as Figure 5 for the RGS1 (top) and RGS2 (bot-tom) For RGS1 the binsize is 5442mA and for RGS2 the binsizeis 6048mA The exposure time is 5292 ks

Fig 7mdash Best-fit emission measure distribution derived fromiron lines in ionization stages Fe XV to Fe XXIV (solid) Theresult from Brickhouse et al (2000) is also shown (thick dotted)Individual line emission measure curves (thin dotted) are shownfor the lines in Table 4

total flux measured at 1355 A such that care must betaken with deblending this line even for HEG spectraThe forbidden line has an adjusted flux 028 ph cmminus2

ksminus1

52 Treatment of Weak Line Contamination

Lines which are too weak to be identified and mea-sured in the spectrum (ie lines not found in Table 3) aretreated as sources of systematic error to the diagnosticline fluxes The total weak line flux in the region between1335 and 1385 A is predicted to be 19 times the modelcontinuum flux however unlike the flat continuum thisline emission is not randomly distributed Furthermoremost of the stronger of the unidentified lines have refer-ence wavelengths with small estimated errors Thus itis reasonable to compute the fluxes of weak lines withinthe diagnostic line profile to estimate the degree of con-tamination The models indicate that the resonance linecontamination is sim 5 while the forbidden and inter-combination lines each have sim 3 contamination (Thisis in addition to the contamination from the 50 blend-ing of the intercombination line already discussed)

53 Density and Temperature Diagnostics with Ne IX

for Capella

The Ne IX line fluxes constrain electron densities andtemperatures When blends are neglected the Ne-Feblended measurement (Table 3) yields an R-ratio of21plusmn03 which indicates a density log(ne) = 116 plusmn 01Instead accounting for the blends in the intercombi-nation line based on the emission measure model (seeFig 9) we derive the low-density limit R = 395 plusmn 07resulting in log(ne) lt 102 Figure 10 shows that thelow-density limit is found at all relevant temperatures

The G-ratio is less affected by blending The raw mea-surements result in G = 097plusmn007 while the deblendedfluxes give G = 093 plusmn 009 which leads to T= 21 plusmn09MK (Fig 11)

The temperature derived from the G-ratio is a factor oftwo lower than the temperature of maximum emissivity(4MK) but probably consistent given the uncertaintiesHowever it is inconsistent with expectations based onthe emission measure distribution Figure 12 shows thatsim 94 of the line emission in our model comes from tem-peratures above 4 MK This discrepancy is perhaps mosteasily interpreted as an abundance effect in an inhomo-geneous plasma though could also be caused by othereffects Standard emission measure analysis is not validunless the emitting plasma has uniform abundances thiscould easily be incorrect for active binary systems

If we allow for this possibility in the case of Capella wecould conclude that the dominant 6 MK plasma predictedby the peak in the emission measure distribution musthave a lower than solar neon-to-iron ratio However ifthis were the case in order to preserve the approximatelysolar neon-to-iron abundance ratio indicated by our line-to-continuum analysis (the scaling factor Ar = 103 forthe Ne IX resonance line) the 2MK plasma must havea much higher neon-to-iron ratio

Young et al (2001) have used FUSE to show thatthe Fe XVIII emission formed at 6 MK is associatedpredominantly with the G8 giant while Johnson et al(2002) using HSTSTIS have found Fe XXI emissionat sim 10MK to be predominantly on the rapidly rotatingG1 giant These latter observations were concurrent withsome of those included in this analysis Earlier measure-ments with HSTGHRS had found each star contributingroughly half the Fe XXI emission (Linsky et al 1998)


Fig 8mdash Measured spectrum of Capella obtained from HEG (histogram with error-bars) APEC individual predicted emission lines(dotted lines) and their sum (heavy solid line) Predicted line fluxes were estimated using the emission measure distribution described insect42 Binsize is 25mA and exposure time is 1547 ks

Fig 9mdash Same as Figure 8 in more detail showing the Ne IXintercombination line with contamination

Our results suggest that a high neon-to-iron abundance

ratio is associated with the G1 star Such a result inthe case of the more rapidly rotating star would notbe unprecedented As noted in sect21 several analysesof Chandra and XMM-Newton spectra of active stars in-dicate abundance ratios for neon-to-iron considerably inexcess of the accepted solar value Drake et al (2001)also noted similar results arose in earlier analyses oflow resolution ASCA studies If the G1 Hertzsprunggap giant has 3times solar neon-to-iron ratio in its coronaand contributes only 25 at the emission measure dis-tribution peak (based on the FUSE Fe XVIII measure-ment) then the clump giant would need to have less thanabout 40 of the solar neon-to-iron abundance ratio Aself-consistent analysis is difficult without more stringentconstraints on the emission measure distribution espe-cially below 6 MK However though the solar FIP effectpresents an existing observational framework for a plausi-ble neon-to-iron ratio significantly below the solar valueit seems coincidental that the abundance ratio of the G8clump giant would conspire to produce a global averageneon-to-iron abundance ratio which agreed so accuratelywith that of the Sun Furthermore preliminary analysesof the clump giants γ Tau and β Cet do not indicate low


Fig 10mdash Determination of the plasma density from the R-ratio using f and i from the direct measurement (fi = 21 plusmn 02)and from the model accounting for line blends (fi = 39 plusmn 07)Shaded areas represent 1σ errors on the measured ratios Theupper limit for the density is log ne lt 102 APEC models for thedensity-sensitive curves are shown for T = 20 40 63 80 and10 MK with the curve denoting the 63 MK model correspondingto the peak of the emission measure distribution The R-ratioincreases with temperature

Fig 11mdash Determination of the plasma temperature from theG-ratio for the fluxes adjusted for line blending (dashed line) Er-rors on the measured ratio are represented by dotted lines APECmodels are given for 1010 1011 1012 and 1013 cmminus3 with thesolid curve denoting the 1010 cmminus3 (ie low-density) model

neon-to-iron ratios (Brickhouse amp Dupree 2003 in prepa-ration Drake et al 2003 in preparation respectively)

While we cannot totally rule out line blending the re-quired contamination would be far larger than expectedIf the G-ratio were high because of a contaminated fline the contaminating line would be the fifth strongestiron line in this region and therefore hard for us to havemissed The discrepancy might also arise from errors inthe ionization balance The shape of the emission mea-sure distribution derived primarily from the ionizationbalance of iron is subject to significant atomic data un-

64 66 68 70 72Log T (K)

000

010

020

030

040

050

060

Con

trib

utio

n F

ract

ion

Fig 12mdash The fraction of Ne IX resonance line emission aris-ing from each temperature in the plasma (solid curve) derived bymultiplying the emission measure distribution of Figure 7 with thefractional emissivity curve from APEC (dash-dotted) The inter-combination and forbidden lines have virtually the same depen-dences and are not shown

certainties (Brickhouse Raymond amp Smith 1995) Onefurther possible effect concerns the breakdown of the fun-damental assumptions implicit in the coronal approxima-tion that the plasma is optically thin and is in ioniza-tion equilibrium A consistent treatment of the inhomo-geneities in the system (two stars with different coronalstructures contributing to the emission) is needed in or-der to test these assumptions We defer further study ofthis complicated issue to future work

For Capella we have used iron lines observed in HETGand EUVE to determine the emission measure distribu-tion For observations of other stellar coronae whichmay have less sampling of the iron ionization stages orlower signal-to-noise than we have for Capella the con-struction of an emission measure distribution becomesmore complicated A different approach is suggested(eg Schmitt amp Ness 2003) where the emission mea-sure distribution is constructed by use of the H-like Lyαand He-like resonance lines including a model distribu-tion of magnetic loops

54 Blending as a Function of Plasma Temperature

As was noted in sect21 spectra of stellar coronae whichare dominated by plasma at significantly lower tempera-tures than that of Capella show considerably less blend-ing around the Ne IX features because the populationsof the blending ions (primarily Fe XIX) are small It isof interest to examine this more quantitatively in orderto determine the temperature regimes for which Ne IXcan be easily used

Figure 13 shows the cooling functions for the neonand blending iron lines It is apparent that in plasmaswith temperatures log T below sim 68 the measurementof Ne IX will not be severely complicated by blendingIn plasmas with higher temperatures the blending willbe very strong unless neon is significantly overabundantcompared to iron Large neon-to-iron as well as neon-to-hydrogen ratios seem common among active stars suchas in HR 1099 (Brinkman et al 2001 Drake et al 2001)II Peg (Huenemoerder Canizares amp Schulz 2001) and


AR Lac (Huenemoerder et al 2003) Generally in allkinds of low-pressure plasmas with high densities butlow temperatures the blending with Fe XIX is less se-vere and thus disentangling the Ne IX lines should befairly straightforward

6 CONCLUSIONS

The Ne IX triplet is an important tool for estimat-ing plasma densities not only for coronal plasmas asin Capella but for plasmas in general in the densityrange between 1010 and 1013 cmminus3 However we haveshown that the intercombination line of Ne IX is severelyblended in plasma with temperatures log T amp 68 Thisis the case in active coronae as well as in solar flaressuch that deblending of high temperature lines is alsoimportant for solar flare diagnostics

60 62 64 66 68 70 72 74Log T(K)

0

20

40

60

80

100

Em

issi

vity

(x

10minus

17 p

h cm

3 sminus

1 )

Ne IXminus

Fe XVIIminus

Fe XVIII

Fe XIXminus

Fe XXminusFe XXI

Fig 13mdash Total emissivities as functions of temperature for eachion emitting strong lines in the wavelength range between 1335 and1385 A Calculations are APEC models for solar abundances andlow density Table 3 gives the wavelengths for the lines identifiedin this work

Through our detailed study of Capella spectra we havefound that APEC models are sufficiently accurate andcomplete that all the significant observed lines in theNe IX region can be reasonably identified in the HEGspectra The laboratory wavelengths of Fe XIX fromBrown et al (2002) provide a significant improvementto the accuracy over the wavelengths derived from theHULLAC energy levels The model fluxes for these linesare also in good agreement with the observations In theAPEC models for Capella the Ne IX intercombinationline is significantly blended Given the predicted flux ofthe blending lines the fi ratio is consistent with thelow density limit (ne lt 2 times 1010 cmminus3)

Interestingly we find that the temperature-sensitiveG-ratio is inconsistent with the emission measure dis-

tribution derived from iron in the sense that it indi-cates significantly cooler electron temperatures Sincethe strong peak in the emission measure distribution at6 MK is likely produced predominantly by the G8 star(Young et al 2001) abundance differences between thetwo coronae at first seem the most likely explanation forthis apparent inconsistency This would require the neon-to-iron abundance ratio to be lower in the G8 corona thanin the G1 corona In contrast the neon-to-iron ratio inthe G1 corona would have to be higher than solar valuessuch that the average ratio in observations of the Capellasystem as a whole appeared solar

If correct this interpretation could have significant im-plications for coronal analyses in general since manycoronal sources are binary systems comprised of two coro-nae Furthermore if coronae on individual stars are com-positionally inhomogeneous (as is the case on the Sun)models will have to account for this In future work wewill investigate more detailed models to determine whatabundance and temperature differences are required toexplain the Capella G-ratio and emission measure dis-tribution inconsistency We will also continue the de-tailed assessment of blending and atomic data for theseand other spectra which may shed further light on thisproblem

In the case of Capella spectra obtained by LETGS andRGS the spectral resolution is inadequate to derive thesame results independently from this isolated spectral re-gion We have not yet explored the possibility of usingother stronger isolated lines from Fe XIX to Fe XXI tospecify the iron contribution to the blended Ne IX spec-tral region and then perform a more constrained fit Thepotential problem with this approach is that typical un-certainties in the APEC model line flux ratios (sim 20 to30) may be too large to sufficiently constrain the blend-ing spectrum With a single test case such as Capellawe cannot explore all the parameter space in the mod-els and thus any conclusions concerning the treatmentof blending cannot yet be generalized Nevertheless thegood agreement between the APEC models and the HEGobservations is encouraging

We thank Manuel Gudel and Marc Audard for sup-port in reducing the RGS data and Priya Desai for helpwith the Chandra data reduction We also thank GregBrown for making the pre-publication data from theLLNL EBIT available to us We acknowledge supportfor J-U Ness from NASA LTSA NAG5-3559 J J Dand N S B were supported by NASA contract NAS8-39083 to SAO for the CXC and D P H was supportedby NASA through Smithsonian Astrophysical Observa-tory (SAO) contract SVI-61010 to MIT for the CXC

REFERENCES

Acton L W Catura R C Meyerott A J amp Wolfson C J1972 Sol Phys 26 183

Anders E amp Grevesse N 1989 Geochim Cosmochim Acta 53197

Audard M Behar E Gudel M Raassen A J J Porquet DMewe R Foley C R amp Bromage G E 2001 AampA 365 L329

Audard M Gudel M Sres A Raassen J J amp Mewe R 2003AampA 398 1137

Ayres T R Brown A Osten R A Huenemoerder D P DrakeJ J Brickhouse N S amp Linsky J L 2001 ApJ 549 554

Behar E Cottam J amp Kahn S M 2001 ApJ 548 966Blumenthal G R Drake G W amp Tucker W H 1972 ApJ 172

205Brickhouse N S 1996 IAU Colloq 152 Astrophysics in the

Extreme Ultraviolet eds S Bowyer and RF Malina DordrechtKluwer 105


Brickhouse N S amp Drake J J 2000 Rev Mexicana AstronAstrofis Ser Conf 9 24

Brickhouse N S Dupree A K Edgar R J Liedahl D ADrake S A White N E amp Singh K P 2000 ApJ 530 387

Brickhouse N S 2002 in Stellar Coronae in the XMM-Newtonand Chandra Era eds F Favata and J J Drake ASP ConfSer 277 13

Brickhouse N S Raymond J C amp Smith B W 1995 ApJS97 551

Brinkman A C et al 2000 ApJ 530 L111Brinkman A C et al 2001 AampA 365 L324Brown G V Beiersdorfer P Liedahl D A Widmann K amp

Kahn S M 1998 ApJ 502 1015Brown G V Beiersdorfer P Liedahl D A Widmann K Kahn

S M amp Clothiaux E J 2002 ApJS 140 589Canizares C R et al 2000 ApJ 539 L41Coffey I H Barnsley R Keenan F P amp Peacock N J 1994

J Phys B 27 1011Doyle J G amp Keenan F P 1986 AampA 157 116Drake G W 1988 Canadian J Phys 66 586Drake J J 1996 ASP Conf Ser 109 Cool Stars Stellar Systems

and the Sun 9 203Drake J J Brickhouse N S Kashyap V Laming J M

Huenemoerder D P Smith R amp Wargelin B J 2001 ApJ548 L81

Dupree A K Brickhouse N S Doschek G A Green J C ampRaymond J C 1993 ApJ 418 L41

Ericsson G W 1977 J Phys Chem Ref Data 6 3Freeman P Doe S amp Siemiginowska A 2001 Proc SPIE 4477

76Gabriel A H amp Jordan C 1969 MNRAS 145 241Gudel M et al 2001 AampA 365 L336Huenemoerder D P Canizares C R amp Schulz N S 2001 ApJ

559 1135Huenemoerder D P Canizares C R Drake J J amp Sanz-

Forcada J 2003 ApJ submittedKashyap V amp Drake J J 2002 in Stellar Coronae in the XMM-

Newton and Chandra Era eds F Favata and J J Drake ASPConf Ser 277 509

Johnson O Drake J J Kashyap V Brickhouse N S DupreeA K Freeman P Young P R amp Kriss G A 2002 ApJ 565L97

Keenan F P McKenzie D L McCann S M amp Kingston A E1987 ApJ 318 926

Landi E del Zanna G Breeveld E R Landini M BromageB J I Pike C D 1999 AampAS 135 171

Linsky J L et al 1993 ApJ 402 694Linsky J L Wood B E Brown A amp Osten R A 1998 ApJ

492 767Mazzotta P Mazzitelli G Colafrancesco S amp Vittorio N 1998

AampAS 133 403McKenzie D L 1985 ApJ 296 294Mewe R et al 1982 ApJ 260 233Mewe R Raassen A J J Drake J J Kaastra J S van der

Meer R L J amp Porquet D 2001 AampA 368 888Ness J-U et al 2001 AampA 367 282Ness J-U Schmitt J H M M Burwitz V Mewe R amp

Predehl P 2002a AampA 387 1032Ness J-U Schmitt J H M M Burwitz V Mewe R Raassen

A J J van der Meer R L J Predehl P amp Brinkman A C2002b AampA 393 911

Ness J-U amp Wichmann R 2002 Astron Nachr 323 (2002) 2129

Pease D O et al 2000 Proc SPIE 4012 700Perryman M A C et al 1997 AampA 323 L49Phillips K J H Mathioudakis M Huenemoerder D P

Williams D R Phillips M E amp Keenan F P 2001 MNRAS325 1500

Pradhan A K amp Shull J M 1981 ApJ 249 82Rosner R Tucker W H amp Vaiana G S 1978 ApJ 220 643Schmitt J H M M Collura A Sciortino S Vaiana G S

Harnden F R Jr amp Rosner R 1990 ApJ 365 704Schmitt J H M M 1997 AampA 318 215Schmitt J H M M amp Ness J-U 2003 AampA submittedSchrijver C J Mewe R amp Walter F M 1984 AampA 138 258Shirai T Sugar J Musgrove A amp Wiese W L 2000 J Phys

and Chem Ref Data Monograph 8

Smith R K Brickhouse N S Liedahl D A amp Raymond J C2001 ApJ 556 L91

Vaiana G S amp Rosner R 1978 ARAampA 16 393Vedder P W amp Canizares C R 1983 ApJ 270 666Wargelin B J Beiersdorfer P amp Kahn S M 1993 PRL 71

2196Wargelin B J 1993 A Study of Diagnostic X-Ray Lines in

Heliumlike Neon Using an Electron Beam Ion Trap Ph DThesis U C Berkeley

Young P R Dupree A K Redfield S Linsky J L Ake T Bamp Moos H W 2001 ApJ 555 L121


A key plasma parameter for inferring the size of X-rayemitting regions is therefore the plasma density Directspectroscopic information on plasma densities at coro-nal temperatures on stars other than the Sun first be-came possible with the advent of ldquohighrdquo resolution spec-tra (λ∆λ sim 200) obtained by EUVE that were capableof separating individual spectral lines Even with thisresolution the available diagnostics have often tended tobe less than definitive owing to the poor signal-to-noiseratio of observed spectra or to blended lines This hasbeen especially the case for the active stars Studies ofdensity-sensitive lines of Fe XIX to Fe XXII in EUVEspectra of RS CVn stars revealed tempting evidence forhigh densities of ne sim 1012 to 1013 cmminus3 at coronal tem-peratures near 107 K (Drake 1996 Dupree et al 1993)Such high densities suggest that emitting structures arecompact static loop models such as those described byRosner Tucker amp Vaiana (1978) would have heights ofsim 1000km with confining field strengths of up to 1 kGand surface filling factors of 1 to 10 In the case ofthe evolved active binary Capella Brickhouse (1996) ob-tained ne sim 1012 cmminus3 from Fe XIX to Fe XXII butne sim 109 cmminus3 based on Fe XII to Fe XIV suggestingthat the cool and hot plasma are from distinctly differ-ent structures In contrast to the high densities reportedby Dupree et al (1993) only upper limits were found byMewe et al (2001) (ne lt 2 minus 5 times 1012 cmminus3) using thesame line ratios of Fe XX to Fe XXII but with the higherspectral resolution of Chandra LETGS spectra Theypoint out that the Fe XIX to Fe XXII line ratios are onlysensitive above 1011 cmminus3 such that no tracer for lowdensities for the hotter plasma component is available

The high spectral resolution of Chandra and XMM-Newton has significantly changed the situation regard-ing plasma diagnostics for stellar coronae High spec-tral resolution coupled with large effective area allowsapplication of line-based diagnostic techniques at X-raywavelengths Line ratios of the strong H-like Lyman se-ries which are sensitive to the Boltzmann factor and theHe-like triplets which exploit the competition betweencollisional excitation and recombination-driven cascadescan be useful temperature diagnostics Simply examin-ing the presence or absence of principle lines of differ-ent ionic states of different elements gives an indicationof the emitting plasma temperatures The He-like sys-tems also provide density diagnostics based on the low-lying metastable level 1s2s 3S1 above the 1s2 1S0 groundstate The He-like oxygen diagnostic has been used tosupport the previous EUVE result for Capella show-ing that the lower temperature plasma (sim2 MK) alsohas lower density (Audard et al 2001 Brickhouse 2002Brinkman et al 2000 Canizares et al 2000 Ness et al2001)

We compare the capabilities of this current generationof high spectral resolution X-ray instruments with a de-tailed study of the Ne IX triplet as a density and tem-perature diagnostic We focus our study on the Capellabinary system (HD 34029 α Aurigae G8 III + G1 III)Capella has been extensively studied from X-ray to radiowavelengths and is the brightest steady coronal sourcein the X-ray sky As such it has been a key calibrationtarget for the Chandra high-energy and low-energy trans-mission grating spectrometers (HETGS and LETGS re-spectively) as well as the XMM-Newton reflection grat-

ing spectrometers (RGS) and has been observed on sev-eral occasions by both satellites (Audard et al 2001Brinkman et al 2000 Canizares et al 2000) The emis-sion measure distribution of Capella shows a steep butnarrow enhancement at 6 MK (Brickhouse et al 2000Dupree et al 1993) making it ideal for studying theblending of Ne IX with high temperature lines

The purpose of this study is threefold (i) assess theaccuracy and reliability of state-of-the-art plasma radia-tive loss models for describing the emission spectrum ofCapella in the region of the He-like complex of Ne (ii)determine how well these models might describe the spec-tra of stars both more and less active than Capella andwith different coronal temperatures and (iii) gain furtherinsight into the plasma density in the Capella coronae inthe key temperature range 4 to 6MK

2 HE-LIKE IONS IN HIGH RESOLUTION STELLARSPECTRA

21 The Use of the He-like Triplet Diagnostics

He-like ions are produced at temperatures ranging be-tween sim 2MK for C V and N VI and 10MK for Si XIIIThe densities over which the diagnostics are sensitive in-crease with increasing element number Z such that thelower Z ions C V N VI and O VII with temperaturesof peak emissivity at 10 16 and 20MK respectivelyprovide diagnostics for densities up to sim 1012 cmminus3 Thehigher temperature (gt 6 MK) ions Mg XI and Si XIIIare sensitive at densities above sim 1012 cmminus3 Ne IXtypically formed at sim 4MK is sensitive to densitiesbetween sim 1011 and 1013 cmminus3 In many cases mea-surements of Ne IX are the only chance to fill the gapbetween the cooler plasma and the hotter plasma Inaddition some stars show a neon overabundance (egBrinkman et al 2001 Drake et al 2001 Gudel et al2001 Huenemoerder Canizares amp Schulz 2001) makingNe IX lines more easily detectable in those cases How-ever blending with Fe lines in the Ne IX triplet spectralregion can compromise the diagnostic utility when ana-lyzing plasmas hot enough to produce them (especiallyFe XIX at sim 6 MK)

The theory of the He-like triplets was originally devel-oped by Gabriel amp Jordan (1969) The density diagnos-tic is often referred to as the R-ratio where R = fidenoting the forbidden line flux with f and the inter-combination line flux with i The relation can be param-eterized as

R =R0

1 + neNc

(1)

with low-density limit R0 and electron densityne The critical density Nc is the density atwhich R = 12R0 Pradhan amp Shull (1981) andBlumenthal Drake amp Tucker (1972) calculated theo-retical values of R0 and Nc and determined the Z-dependence of the density-sensitivity as shown in Fig-ure 1 Larger R-ratios mean relatively weaker intercom-bination lines since the i is intrinsically weak at lowdensities it may be difficult to measure accurately

Strong ultraviolet radiation fields can also de-populatethe metastable levels of the He-like triplets and changethe R-ratios For the case of Capella and Ne IX suchfields can be neglected (Ness et al 2001)

For an optically thin plasma a temperature diagnostic



is

G =i + f

r (2)
















23 Spectral Models
















4 DATA ANALYSIS







Γ )β





















λbobs λc


model





122p3








323d32 0325plusmn0018 02008




13551 0005 Fexix 2p4 3P2 ndash 2p122p2








323d52 0085plusmn0010 00133






323d52 0089plusmn0010 00573



323d52 0234plusmn0015 01778


13839 0005 Fexix 2p4 3P2 ndash 2p2


13839 0005 Fexix 2p4 1D2 ndash 2p2


























323d52


323p12 42796 plusmn 722 0136 plusmn 0023 2037


















ksminus1




for Capella
















64 66 68 70 72Log T (K)

000

010

020

030

040

050

060

Con

trib

utio

n F

ract

ion









6 CONCLUSIONS


60 62 64 66 68 70 72 74Log T(K)

0

20

40

60

80

100

Em

issi

vity

(x

10minus

17 p

h cm

3 sminus

1 )

Ne IXminus

Fe XVIIminus

Fe XVIII

Fe XIXminus

Fe XXminusFe XXI








REFERENCES
















































is

G =i + f

r (2)
















23 Spectral Models
















4 DATA ANALYSIS







Γ )β





















λbobs λc


model





122p3








323d32 0325plusmn0018 02008




13551 0005 Fexix 2p4 3P2 ndash 2p122p2








323d52 0085plusmn0010 00133






323d52 0089plusmn0010 00573



323d52 0234plusmn0015 01778


13839 0005 Fexix 2p4 3P2 ndash 2p2


13839 0005 Fexix 2p4 1D2 ndash 2p2


























323d52


323p12 42796 plusmn 722 0136 plusmn 0023 2037


















ksminus1




for Capella
















64 66 68 70 72Log T (K)

000

010

020

030

040

050

060

Con

trib

utio

n F

ract

ion









6 CONCLUSIONS


60 62 64 66 68 70 72 74Log T(K)

0

20

40

60

80

100

Em

issi

vity

(x

10minus

17 p

h cm

3 sminus

1 )

Ne IXminus

Fe XVIIminus

Fe XVIII

Fe XIXminus

Fe XXminusFe XXI








REFERENCES



















































23 Spectral Models
















4 DATA ANALYSIS







Γ )β





















λbobs λc


model





122p3








323d32 0325plusmn0018 02008




13551 0005 Fexix 2p4 3P2 ndash 2p122p2








323d52 0085plusmn0010 00133






323d52 0089plusmn0010 00573



323d52 0234plusmn0015 01778


13839 0005 Fexix 2p4 3P2 ndash 2p2


13839 0005 Fexix 2p4 1D2 ndash 2p2


























323d52


323p12 42796 plusmn 722 0136 plusmn 0023 2037


















ksminus1




for Capella
















64 66 68 70 72Log T (K)

000

010

020

030

040

050

060

Con

trib

utio

n F

ract

ion









6 CONCLUSIONS


60 62 64 66 68 70 72 74Log T(K)

0

20

40

60

80

100

Em

issi

vity

(x

10minus

17 p

h cm

3 sminus

1 )

Ne IXminus

Fe XVIIminus

Fe XVIII

Fe XIXminus

Fe XXminusFe XXI








REFERENCES



















































4 DATA ANALYSIS







Γ )β





















λbobs λc


model





122p3








323d32 0325plusmn0018 02008




13551 0005 Fexix 2p4 3P2 ndash 2p122p2








323d52 0085plusmn0010 00133






323d52 0089plusmn0010 00573



323d52 0234plusmn0015 01778


13839 0005 Fexix 2p4 3P2 ndash 2p2


13839 0005 Fexix 2p4 1D2 ndash 2p2


























323d52


323p12 42796 plusmn 722 0136 plusmn 0023 2037


















ksminus1




for Capella
















64 66 68 70 72Log T (K)

000

010

020

030

040

050

060

Con

trib

utio

n F

ract

ion









6 CONCLUSIONS


60 62 64 66 68 70 72 74Log T(K)

0

20

40

60

80

100

Em

issi

vity

(x

10minus

17 p

h cm

3 sminus

1 )

Ne IXminus

Fe XVIIminus

Fe XVIII

Fe XIXminus

Fe XXminusFe XXI








REFERENCES

































































λbobs λc


model





122p3








323d32 0325plusmn0018 02008




13551 0005 Fexix 2p4 3P2 ndash 2p122p2








323d52 0085plusmn0010 00133






323d52 0089plusmn0010 00573



323d52 0234plusmn0015 01778


13839 0005 Fexix 2p4 3P2 ndash 2p2


13839 0005 Fexix 2p4 1D2 ndash 2p2


























323d52


323p12 42796 plusmn 722 0136 plusmn 0023 2037


















ksminus1




for Capella
















64 66 68 70 72Log T (K)

000

010

020

030

040

050

060

Con

trib

utio

n F

ract

ion









6 CONCLUSIONS


60 62 64 66 68 70 72 74Log T(K)

0

20

40

60

80

100

Em

issi

vity

(x

10minus

17 p

h cm

3 sminus

1 )

Ne IXminus

Fe XVIIminus

Fe XVIII

Fe XIXminus

Fe XXminusFe XXI








REFERENCES































































323d52


323p12 42796 plusmn 722 0136 plusmn 0023 2037


















ksminus1




for Capella
















64 66 68 70 72Log T (K)

000

010

020

030

040

050

060

Con

trib

utio

n F

ract

ion









6 CONCLUSIONS


60 62 64 66 68 70 72 74Log T(K)

0

20

40

60

80

100

Em

issi

vity

(x

10minus

17 p

h cm

3 sminus

1 )

Ne IXminus

Fe XVIIminus

Fe XVIII

Fe XIXminus

Fe XXminusFe XXI








REFERENCES


















































ksminus1




for Capella
















64 66 68 70 72Log T (K)

000

010

020

030

040

050

060

Con

trib

utio

n F

ract

ion









6 CONCLUSIONS


60 62 64 66 68 70 72 74Log T(K)

0

20

40

60

80

100

Em

issi

vity

(x

10minus

17 p

h cm

3 sminus

1 )

Ne IXminus

Fe XVIIminus

Fe XVIII

Fe XIXminus

Fe XXminusFe XXI








REFERENCES
























































64 66 68 70 72Log T (K)

000

010

020

030

040

050

060

Con

trib

utio

n F

ract

ion









6 CONCLUSIONS


60 62 64 66 68 70 72 74Log T(K)

0

20

40

60

80

100

Em

issi

vity

(x

10minus

17 p

h cm

3 sminus

1 )

Ne IXminus

Fe XVIIminus

Fe XVIII

Fe XIXminus

Fe XXminusFe XXI








REFERENCES
















































6 CONCLUSIONS


60 62 64 66 68 70 72 74Log T(K)

0

20

40

60

80

100

Em

issi

vity

(x

10minus

17 p

h cm

3 sminus

1 )

Ne IXminus

Fe XVIIminus

Fe XVIII

Fe XIXminus

Fe XXminusFe XXI








REFERENCES














































Modeling the Ne IX Triplet Spectral Region of Capella with the Chandra and XMM-Newton Gratings

Documents

Transcript of Modeling the Ne IX Triplet Spectral Region of Capella with the Chandra and XMM-Newton Gratings