Neon Insights from Old Solar X-rays: a Plasma Temperature Dependence of the Coronal Neon Content
aa r X i v : . [ a s t r o - ph . S R ] S e p To appear in the Astrophysical Journal
NEON INSIGHTS FROM OLD SOLAR X-RAYS: A PLASMATEMPERATURE DEPENDENCE OF THE CORONAL NEONCONTENT
Jeremy J. Drake Smithsonian Astrophysical Observatory, MS-3,60 Garden Street,Cambridge, MA 02138, USA [email protected]
ABSTRACT
An analysis using modern atomic data of fluxes culled from the literature forO VIII and Ne IX lines observed in solar active regions by the P78 and SolarMaximum Mission satellites confirms that the coronal Ne/O abundance ratiovaries by a factor of two or more, and finds an increase in Ne/O with increasingactive region plasma temperature. The latter is reminiscent of evidence for in-creasing Ne/O with stellar activity in low-activity coronae that reaches a “neonsaturation” in moderately active stars at approximately twice the historicallyaccepted solar value of about 0.15 by number. We argue that neon saturationrepresents the underlying stellar photospheric compositions, and that low activitycoronae, including that of the Sun, are generally depleted in neon. The implica-tion would be that the solar Ne/O abundance ratio should be revised upward bya factor of about two to n (Ne) /n (O) ∼ .
3. Diverse observations of neon in thelocal cosmos provide some support for such a revision. Neon would still be ofsome relevance for reconciling helioseismology with solar models computed usingrecently advocated chemical mixtures with lower metal content.
Subject headings:
Sun: abundances — Sun: activity — Sun: corona — X-rays:stars
1. Introduction
Interest in Neon, the newcomer, has been strongly piqued recently by two possibly re-lated puzzles concerning the value of the Ne abundance in the Sun and stars. In a study 2 –of 21 mostly magnetically active stars observed by the
Chandra high resolution X-ray spec-trometers, Drake & Testa (2005) found Ne/O abundance ratios to be consistently higher bya factor of 2.7 than the recommended solar value of the time, Ne/O=0.15 by number . Thelatter value is traceable to the assessment of Anders & Grevesse (1989).Drake & Testa (2005) suggested that the same ratio might also be appropriate for theSun and help solve the “Solar Model Problem”. Models employing a recently advancedsolar chemical composition based on 3-D non-LTE hydrodynamic photospheric modelling(Asplund et al. 2005) led to predictions of the depth of the convection zone, helium abun-dance, density and sound speed in serious disagreement with helioseismology measurements(Basu & Antia 2004; Bahcall et al. 2005a). The Asplund et al. (2005) mixture contained lessof the elements C, N, O and Ne that are important for the opacity of the solar interior by25-35 % compared to earlier assessments (e.g. Anders & Grevesse 1989; Grevesse & Sauval1998). Antia & Basu (2005) and Bahcall et al. (2005b) suggested the uncertain solar Neabundance might be raised to compensate. The estimated Ne/O ratio required was in agree-ment with that found in stellar coronae by Drake & Testa (2005).More detailed investigations of helioseismology models now suggest that raising the Neabundance alone cannot fully reconcile models and observations (Delahaye & Pinsonneault2006; Lin et al. 2007; Delahaye et al. 2010). Nevertheless, the Ne abundance still remainsan important individual ingredient, not only for the Sun but also for nucleosynthesis andgalactic chemical evolution.Ne exhibits no lines in the visible light spectra of late-type stars and is not retainedin meteorites. Consequently, the solar abundance is based largely on transition region andcoronal lines, and energetic particle measurements, supplemented with local cosmic estimates(e.g. Meyer 1985; Anders & Grevesse 1989; Asplund et al. 2009). Two studies of transitionregion EUV and coronal X-ray lines find consistency with the ratio Ne/O=0.15 (Young 2005;Schmelz et al. 2005) and conclude that this ratio represents that of the bulk of the Sun.However, based on UV lines observed during a flare Landi et al. (2007) obtained an absoluteabundance Ne/H a factor of 1.9 larger than the Asplund et al. (2005) value. Subsequently,the latter authors have revised their recommended Ne and O abundances back up towardolder values by 0.13 and 0.07 dex, respectively (Asplund et al. 2009), raising their favouredabsolute logarithmic Ne abundance to Ne/H= 7 . ± . Throughout we use the notation Ne/O to refer to the ratio by number of Ne to O nuclei, rather thanthe implicit logarithmic abundance ratio sometimes used.
Solar Maximum Mission (SMM) Flat Crystal Spectrometer (FCS)observations obtained between 1986 May 20 and 1987 December 18 are two exceptions thatform the basis of this work.
2. Analysis2.1. Methods and Atomic Data
The analysis presented here is similar to that originally applied to Ne and O linesby Acton et al. (1975), and duplicated in successive papers including McKenzie & Feldman(1992) and Schmelz et al. (2005) (see also Drake & Testa 2005): Ne/O abundance ratios areinferred from observed ratios of the intensities of the transitions Ne IX λ .
47 1 s p P → s S and O VIII λ .
97 2 p P / , / → s S / and comparison with theoretical predic-tions.Line intensities were taken from Table 2 of McKenzie & Feldman (1992) and Table 2of Schmelz et al. (2005). One possible complication to the analysis of the Ne IX λ .
47 line isthe potential presence of Fe XIX blends (see, e.g. Ness et al. 2003); both McKenzie & Feldman(1992) and Schmelz et al. (2005) take pains to note this and assess that for their non-flaringspectra such blends are negligible. This assessment is verified below when specific accountof the Fe XIX contribution to Ne IX λ .
47 is taken.The theoretical Ne IX/O VIII line ratio has significant temperature dependence andderivation of an abundance ratio from a measured flux ratio requires an estimate of temper-ature (see Drake & Testa 2005, for a more temperature-insensitive ratio using both Ne IX and 4 –Ne X lines). McKenzie & Feldman (1992) utilised the ratio of Fe XVIII λ .
21 2 p P / − p ( D )3 d D / , P / and Fe XVII λ .
01 2 p S − p d P as a temperature diagnostic.They list the line fluxes which we use here to re-derive temperatures using the atomic datanoted below. Schmelz et al. (2005) used the same Fe XVIII lines combined with Fe XVII λ .
68 2 p S − p s P but omit measured fluxes and only list their derived tempera-tures which we employ directly. Questions have been raised in the past regarding possibleoptical depth of Fe XVII λ .
01 due to resonance scattering: Brickhouse & Schmelz (2006)have recently concluded the line is consistent with being optically thin in FCS observationsof active regions, and attribute apparent quenching to Fe XVI blends in the neighbouringFe XVII λ .
26 line often used as an optically thin comparison.Spectral line fluxes were analysed using the PINTofALE IDL software (Kashyap & Drake2000), employing collisional excitation rates and energy levels from the CHIANTI databasev6.0.1 (Landi et al. 2006; Dere et al. 1997, and references therein), together with the ioniza-tion fractions as function of temperature from Bryans et al. (2009). The sensitivity of theresults to the adopted atomic data was also investigated. We first illustrate in Figure 1 the McKenzie & Feldman (1992) and Schmelz et al. (2005)Ne/O line ratios (in photon units) as a function of the isothermal plasma temperature. Un-certainties in the temperatures, both from Schmelz et al. (2005) and derived here using theMcKenzie & Feldman (1992) Fe XVII and Fe XVIII line fluxes and measurement uncer-tainties, amount typically to no more than 0.05-0.1 dex and are omitted for clarity. TheMcKenzie & Feldman (1992) and Schmelz et al. (2005) Ne/O flux ratios are consistent withone another and follow a steep, approximately linear, trend of increasing ratio with increasingtemperature.We now examine whether the data simply reflect the temperature dependence of theNe IX/O VIII line ratio and are consistent with a single Ne/O abundance ratio. To do this,we first compare the observed flux ratios with the theoretical ratio computed as a functionof the isothermal plasma temperature. This comparison assumes the individual observationsalso correspond to isothermal plasma; this is strictly not likely to be the case and we return Interactive Data Language, Research Systems Inc. minimum theoretical photon intensity ratio reached matches the minimum observed ratio, excluding the most extreme observed point. The abundance ratio for the up-per curve (Ne/O=0.17) is such that the maximum theoretical photon intensity ratio reachedmatches the maximum observed ratio, again excluding the most extreme observed point.Note that there are still observed photon intensity ratios that lie below the lower theoreticalratio curve and one that lies slightly above the upper one: in principle, these could be madeto fit within the two curves were there to be some source of additional error in the derivedplasma temperatures which allowed the points to be shifted arbitrarily to the left or right.In this way, the two curves represent the minimum possible spread in Ne/O abundance ratio(Ne/O=0.12 and 0.17) that can match the data. The separation of the two curves is muchgreater than the statistical errors in the data, and we conclude that the observed intensityratios cannot be explained by a single Ne/O abundance ratio.The trend of the observed intensities with temperature is also much steeper than thetheoretical ratio. If the observed intensities are from isothermal plasma, this indicates atrend of increasing Ne/O abundance ratio with temperature.
The fields of view of the SOLEX and FCS instruments from which the observations anal-ysed here were obtained were 1 arcmin and 15 arcsec across, respectively (McKenzie & Feldman1992; Schmelz et al. 2005). It is likely than in regions of this size the observed plasma is notisothermal. Any deviation from isothermality tends to flatten out the theoretical line inten-sity ratio curves. We investigate this quantitatively using a model continuous differentialemission measure distribution (DEM).For the DEM model, we adopted the form Φ( T ) = n e ( T ) dV ( T ) dT , where V is the volumeoccupied by material with electron density n e at temperature T . Φ( T ) was approximated 6 –by a combination of two power laws bridged by a constant flat top:Φ( T ) = φ α T α + φ β T β + rect(( T max − T ) /δT ) , (1)where the DEM peaks at a temperature between T = T max ± δT / φ α = 1 /T αmax for T ≤ T max − δT /
20 for
T > T max − δT / φ β = 0 for T ≤ T max + δT / /T βmax for T > T max + δT / , (2)and rect( T ) is the rectangular functionrect(( T max − T ) /δT ) = 1 for | T − T max | < δT /
20 for | T − T max | ≥ δT / . Here, α and β are constants describing the steepness of the rise of the DEM for T < T max and its subsequent decay for
T > T max . Constant conductive loss models and quasi-static,constant cross-section uniformly heated loop models have the well-known power law slope α = 3 / T ); e.g. Drake et al. (2000) found α ∼ ξ Boo A and ǫ Eri.The observed line intensity for a transition ij in species X with an abundance A in aplasma with an DEM Φ T max ( T ) characterized by a peak temperature T max can be written F ( T max ) ij = A Z ∞ ε ij ( T )Φ( T ) dT, (3)where ε ij is the effective line emissivity. We can use theoretical line fluxes as a function ofthe characteristic temperature, T max , calculated using Equation 3 in a similar way to theuse of isothermal fluxes described above in Section 2.2.1. As an illustrative case, we adopted α = 3 , β = 5, with a small flat-topped maximum of width δT = 0 . T . Using thismodel DEM and the McKenzie & Feldman (1992) Fe XVII and FeXVIII line intensities, wecomputed the DEM peak temperature, T max , corresponding to all the different observations,as for the isothermal case. The McKenzie & Feldman (1992) Ne/O line intensity ratios areillustrated as a function of this T max as grey points in Figure 1. Overlaid are two dashedcurves representing the multi-thermal DEM theoretical Ne IX/O VIII line intensity ratiosas a function of the DEM peak temperature T max . These two curves correspond to twoNe/O abundance ratios (Ne/O=0.09 and 0.22) chosen in the same way as for the isothermal 7 –case described in Section 2.2.1: these dashed curves are the multi-thermal equivalent of theisothermal ones illustrated as solid curves. They represent the minimum spread in Ne/Othat can explain the observations and demonstrate that under multi-thermal conditions thedata cannot be explained by a single Ne/O abundance ratio.The multi-thermal intensity ratio curves are relatively insensitive to the adopted valuesof α , β and δT : as slopes α and β tend toward much higher values, the curve approaches theisothermal case; for much shallower slopes and larger values of δT , the curve simply becomesflatter, tending toward the constant ratio of line emissivities integrated over all temperatures.The smoother multi-thermal theoretical intensity ratios require a much larger spread in Ne/Oabundance ratio than the isothermal ones. Again, the observed Ne/O intensities rise muchmore steeply with peak DEM temperature than the theoretical curve, indicating a generaltrend of increasing Ne/O abundance ratio with plasma temperature.The conclusion that the active region Ne/O line ratios are inconsistent with a singleNe/O abundance ratio was made earlier by McKenzie & Feldman (1992), who found Ne/Ovaried by a factor of 2.4. What is new here is the trend of increasing Ne/O with increasingplasma temperature.We note in passing that indications of Ne/O abundance variations of a factor of about 2,consistent with those found here and by McKenzie & Feldman (1992), are also apparent inthe earlier analysis of Ne, O and Fe lines in some SMM FCS spectra by Strong et al. (1988).Their Figure 4 shows observed Ne/O intensity ratios as a function of plasma temperaturethat lie between theoretical curves spanning Ne/O abundance ratios of 0.33 to 0.17. Thatwork was focused on temperature diagnostics, and the authors drew attention to probableabundance errors but not variations. The Ne/O abundance ratio for each of the observed line intensity ratios was obtained forisothermal and DEM cases using the theoretical ratios like those illustrated in Figure 1. Inthese calculations, specific account of the possible contribution of Fe XIX blends at 14.423and 14.462 ˚A to the Ne XI flux was taken. The largest contribution found amounted to15%, but was less than 10% for the great majority of the lines, confirming the assessment ofMcKenzie & Feldman (1992) that Fe XIX blends are not significant for these observations.The abundance ratios as a function of temperature for both isothermal and DEM cases areshown in Figure 2, together with their error-weighted linear best fits. Best-fits accounted forerrors both in the Ne/O flux ratios, and in the temperatures derived from the Fe XVII and 8 –Fe XVIII lines.The temperature dependence in the derived Ne/O abundance ratios is again obvious,with both DEM and isothermal cases showing a rise to higher Ne/O with rising tempera-ture. The mean Ne/O ratio is slightly lower than values proposed by Asplund et al. (2005,Ne/O=0.15 by number, or − .
82 in logarithm), Asplund et al. (2009, Ne/O=0.17 by num-ber, or − .
76 in log) and Grevesse & Sauval (1998, Ne/O=0.18 by number or − .
75 in log).The data of Schmelz et al. (2005) did not include the Fe fluxes we require here to deriveDEM peak temperatures and are therefore not shown. However, it is clear from the overlapof Schmelz et al. (2005) and McKenzie & Feldman (1992) line intensity ratios shown in Fig-ure 1, and comparison of Figures 1 and 2, that those data also correspond to a mean slightlylower than these values.
3. Discussion3.1. Atomic Data
Can errors in the underlying atomic data be responsible for the apparent trend of Ne/Owith temperature? The He-like and H-like ions present the simplest cases for computing bothion balances and collisional excitation rates, and atomic data for these species should in gen-eral be more accurate than for more complex ions. The shape of the theoretical Ne IX/O VIIIemissivity ratio in the temperature range of interest here is determined primarily by the in-creasing excitation rate with temperature for both species and the ramp-down of the Ne IXand O VIII ion populations toward higher temperatures. It is difficult to imagine incurringan error of a factor of ∼ R -Matrix collision strengths for Ne IX within a few percent of early data. We have never-theless repeated the calculations reported here for CHIANTI versions 4 and 5, and find onlyvery small differences of less than 10% in derived abundances.The rising Ne abundance with increasing temperature observed in active regions couldbe mimiced by an ion population error in Ne IX, but the required errors are again of order afactor of 2 at temperatures of log T ∼ .
6, or a shift in the Ne IX ion population curve towardhigher temperatures by 25% or so. Such a shift would also require commensurate changes inthe Ne X population, and it seems difficult to introduce such a change in Ne without invokingsimilar changes in the O ion balance. The calculations presented here employ the most 9 –recent ion balance assessment currently available (Bryans et al. 2009), and it seems highlyunlikely that residual errors of such magnitude remain. We have also repeated the abundancecalculations for the ionization equilibria of Mazzotta et al. (1998) and Arnaud & Rothenflug(1985). The latter result in slightly lower temperatures from Fe XVII and Fe XVIII lines byabout 0.1 dex, and systematically higher Ne/O abundance ratios by a similar amount, butthe abundance trends are unaffected (see also § λ .
21 resonance line strength predicted byCHIANTI v4.2 and other databases appeared lower by 25% relative to the λ .
92 resonance2 p − s line compared with the ratio observed in Capella. More recent Fe XVIII electroncollisional excitation calculations have been published by Witthoeft et al. (2007). Using thesedata, Del Zanna (2006) finds good agreement between observed and predicted Fe XVIII linestrengths. It is these data that are included in CHIANTI v6.0.1 used here.We conclude that atomic data errors are a very unlikely explanation for the observedtemperature trend in Ne/O, and that it is the underlying abundance ratio itself which isresponsible. We noted in § ∼ .
15 dex or so. In this regard,the ionization balance of Arnaud & Rothenflug (1985) provides better agreement with thesolar assessments than that of Bryans et al. (2009). However, the trend with temperaturecannot be erased by plausible adjustments of the estimated temperatures: the scatter inthe observed Ne/O ratios is simply much larger than variations in the theoretical line ratio.The main result we emphasise here is this temperature dependence in the abundance ratio,rather than its absolute value.By comparison of Fe/O and Ne/O ratios, McKenzie & Feldman (1992) showed that it isthe abundance of Ne that varies rather than that of O. This is also expected on other grounds:the almost identical ionization potentials of neutral O and H, and the consequently largecharge-exchange cross-section between their neutral and ionized species, should couple theseelements quite efficiently such that the coronal O abundance is expected to follow that ofH. Echoing McKenzie & Feldman (1992), we conclude that the coronal Ne abundance varies 10 –by more than a factor of 2. We also conclude here that the mechanism that fractionates Neoperates so as to either enhance hotter plasma with Ne, or to deplete Ne in cooler plasma (orperhaps both). Until this fractionation mechanism can be identified and understood, it isnot obvious which case applies. This has profound consequences for the solar Ne abundancesince it is not yet possible to ascertain whether the cooler or hotter regions represent thetrue Ne abundance, or whether any region of the solar outer atmosphere has a Ne contentrepresentative of that of the bulk of the Sun . The well-known chemical fractionation basedon first ionization potential (FIP) observed in the solar corona is thought to originate inthe chromosphere (e.g. Meyer 1985; Feldman 1992). If the Ne fractionation operates in asimilar region, it is then quite possible that the Ne abundance is modified in all regions ofthe corona, and in the solar wind and energetic particles. Assessment of the true solar Necontent must then be undertaken deeper into the atmosphere (see Drake & Ercolano 2007,for a method based on photospheric X-ray fluorescence).Based on the constancy of the Ne/O abundance ratio derived from
Chandra
X-ray spec-tra of a sample of mostly quite magnetically active stars, Drake & Testa (2005) concludedthat the data likely represented the true underlying stellar Ne/O abundance ratios. Theirvalue is ∼ . . ± . . . In this regard,the increasing Ne/O ratio we see toward hotter coronal temperatures is quite conspicuous.Using plasma temperature as an activity proxy, the Ne/O ratio increases as a function ofsolar region “activity”. This activity-dependent Ne content of the solar corona fits in wellwith a recent study of Ne/O ratios in low-activity stars by Robrade et al. (2008), who findevidence for a trend of higher Ne/O ratios with increasing stellar activity level (but assumedthe lower stellar activities represented photospheric abundances). A similar trend is alsopossibly present in the compilation of G¨udel (2004).Based on existing results, we echo the earlier conclusions of Drake & Testa (2005) andsuggest that active stars represent coronal “neon saturation”, with Ne content reachingphotospheric levels. It seems otherwise unlikely that the Ne/O ratio can be so well controlledby a chemical fractionation mechanism when in the same stars fractionation varies the Fe/Oratio by an order of magnitude (e.g. G¨udel 2004; Garc´ıa-Alvarez et al. 2008, 2009). In thisscenario, the solar corona, and the coronae of similar “low activity” stars, is depleted in Neby factors ∼ . −
4; the currently recommended solar Ne abundance underestimates thetrue abundance by a factor of about 2, or possibly more. A ratio Ne/O= 0 . ± . . is alsoformally consistent with the valued 0 . ± . . preferred by Delahaye et al. (2010) based onhelioseismology constraints. 11 –This picture of depleted Ne in the solar corona is not unprecedented: the element witha FIP closest to that of Ne is He, which is depleted in the solar corona and wind by afactor of 2 or so (e.g. Laming 2009, and references therein). Laming (2009) presents afractionation model based on the ponderomotive force resulting from the oscillating electricfield of Alfv´en waves propagating through the chromosphere and corona. This model predictscoronal depletion of He to the different degrees required by observations of slow and fast solarwinds, and also of Ne by a factor of ∼
4. Conclusions
An analysis using modern atomic data of solar active region X-ray spectra obtained inthe 1970s and 1980s confirms that the coronal Ne content varies by a factor of 2 or more,and reveals a trend of increasing Ne abundance with increasing plasma temperature. Thistrend seems to reflect the emerging picture of Ne/O abundances in late-type stellar coronae,with Ne/O appearing to increase with stellar activity until quickly reaching a fairly constant“neon saturation” in only moderately active stars. We argue this latter value representsphotospheric ratios. Instead, the Sun is more capricious and withholds neon (as well ashelium), from its outer atmosphere: the true solar Ne abundance cannot yet be inferredwith any degree of certainty from any existing observations of the outer solar atmosphere orwind. Under the “neon saturation” hypothesis it is more abundant than currently assessed,perhaps by as much as a factor of two.The author is gratefu to the anonymous referee for helping clarify and improve themanuscript. JJD thanks the NASA AISRP for providing financial assistance for the de-velopment of the PINTofALE package. JJD was funded by NASA contract NAS8-39073to the
Chandra X-ray Center during the course of this research and thanks the Director,H. Tananbaum, for continuing support and encouragement. 12 –
REFERENCES
Acton, L. W., Catura, R. C., & Joki, E. G. 1975, ApJ, 195, L93Anders, E., & Grevesse, N. 1989, Geochim. Cosmochim. Acta, 53, 197Antia, H. M., & Basu, S. 2005, ApJ, 620, L129Arnaud, M., & Rothenflug, R. 1985, A&AS, 60, 425Asplund, M., Grevesse, N., & Sauval, A. J. 2005, in ASP Conf. Ser. 336: Cosmic Abundancesas Records of Stellar Evolution and Nucleosynthesis, 25Asplund, M., Grevesse, N., Sauval, A. J., & Scott, P. 2009, ARA&A, 47, 481Bahcall, J. N., Basu, S., Pinsonneault, M., & Serenelli, A. M. 2005a, ApJ, 618, 1049Bahcall, J. N., Basu, S., & Serenelli, A. M. 2005b, ApJ, 631, 1281Basu, S., & Antia, H. M. 2004, ApJ, 606, L85Brickhouse, N. S., & Schmelz, J. T. 2006, ApJ, 636, L53Bryans, P., Landi, E., & Savin, D. W. 2009, ApJ, 691, 1540Chen, H., Gu, M. F., Beiersdorfer, P., Boyce, K. R., Brown, G. V., Kahn, S. M., Kelley,R. L., Kilbourne, C. A., Porter, F. S., & Scofield, J. H. 2006, ApJ, 646, 653Cunha, K., Hubeny, I., & Lanz, T. 2006, ApJ, 647, L143Del Zanna, G. 2006, A&A, 459, 307Delahaye, F., & Pinsonneault, M. H. 2006, ApJ, 649, 529Delahaye, F., Pinsonneault, M. H., Pinsonneault, L., & Zeippen, C. J. 2010, ArXiv e-printsDere, K. P., Landi, E., Mason, H. E., Monsignori Fossi, B. C., & Young, P. R. 1997, A&AS,125, 149Dere, K. P., Landi, E., Young, P. R., Del Zanna, G., Landini, M., & Mason, H. E. 2009,A&A, 498, 915Desai, P., Brickhouse, N. S., Drake, J. J., Dupree, A. K., Edgar, R. J., Hoogerwerf, R.,Kashyap, V., Wargelin, B. J., Smith, R. K., Huenemoerder, D. P., & Liedahl, D. A.2005, ApJ, 625, L59 13 –Drake, J. J., & Ercolano, B. 2007, ApJ, 665, L175Drake, J. J., Peres, G., Orlando, S., Laming, J. M., & Maggio, A. 2000, ApJ, 545, 1074Drake, J. J., & Testa, P. 2005, Nature, 436, 525Feldman, U. 1992, Physica Scripta Volume T, 46, 202G¨udel, M. 2004, A&A Rev., 12, 71Garc´ıa-Alvarez, D., Drake, J. J., Kashyap, V. L., Lin, L., & Ball, B. 2008, ApJ, 679, 1509Garc´ıa-Alvarez, D., Drake, J. J., & Testa, P. 2009, in American Institute of Physics Confer-ence Series, Vol. 1094, American Institute of Physics Conference Series, ed. E. Stem-pels, 796–799Grevesse, N., & Sauval, A. J. 1998, Space Science Reviews, 85, 161Jordan, C., Ayres, T. R., Brown, A., Linsky, J. L., & Simon, T. 1987, MNRAS, 225, 903Kashyap, V., & Drake, J. J. 2000, Bulletin of the Astronomical Society of India, 28, 475Kilian, J. 1994, A&A, 282, 867Kilian-Montenbruck, J., Gehren, T., & Nissen, P. E. 1994, A&A, 291, 757Laming, J. M. 2009, ApJ, 695, 954Landi, E., Del Zanna, G., Young, P. R., Dere, K. P., Mason, H. E., & Landini, M. 2006,ApJS, 162, 261Landi, E., Feldman, U., & Doschek, G. A. 2007, ApJ, 659, 743Lanz, T., Cunha, K., Holtzman, J., & Hubeny, I. 2008, ApJ, 678, 1342Lin, C.-H., Antia, H. M., & Basu, S. 2007, ApJ, 668, 603Liu, Y., Liu, X.-W., Barlow, M. J., & Luo, S.-G. 2004, MNRAS, 353, 1251Magrini, L., Stanghellini, L., & Villaver, E. 2009, ApJ, 696, 729Mazzotta, P., Mazzitelli, G., Colafrancesco, S., & Vittorio, N. 1998, A&AS, 133, 403McKenzie, D. L., & Feldman, U. 1992, ApJ, 389, 764Meyer, J.-P. 1985, ApJS, 57, 151 14 –Morel, T., & Butler, K. 2008, A&A, 487, 307Ness, J.-U., Brickhouse, N. S., Drake, J. J., & Huenemoerder, D. P. 2003, ApJ, 598, 1277Perinotto, M., Morbidelli, L., & Scatarzi, A. 2004, MNRAS, 349, 793Pottasch, S. R., & Bernard-Salas, J. 2006, A&A, 457, 189Przybilla, N., Butler, K., Becker, S. R., & Kudritzki, R. P. 2006, A&A, 445, 1099Robrade, J., Schmitt, J. H. M. M., & Favata, F. 2008, A&A, 486, 995Rubin, R. H., Simpson, J. P., O’Dell, C. R., McNabb, I. A., Colgan, S. W. J., Zhuge, S. Y.,Ferland, G. J., & Hidalgo, S. A. 2011, MNRAS, 410, 1320Schmelz, J. T., Nasraoui, K., Roames, J. K., Lippner, L. A., & Garst, J. W. 2005, ApJ, 634,L197Sigut, T. A. A. 1999, ApJ, 519, 303Stanghellini, L., Guerrero, M. A., Cunha, K., Manchado, A., & Villaver, E. 2006, ApJ, 651,898Strong, K. T., Claflin, E. S., Lemen, J. R., & Linford, G. A. 1988, Advances in SpaceResearch, 8, 167Tsamis, Y. G., Barlow, M. J., Liu, X.-W., Danziger, I. J., & Storey, P. J. 2003, MNRAS,345, 186Wang, W., & Liu, X.-W. 2008, MNRAS, 389, L33Witthoeft, M. C., Del Zanna, G., & Badnell, N. R. 2007, A&A, 466, 763Young, P. R. 2005, A&A, 444, L45
This preprint was prepared with the AAS L A TEX macros v5.2.
15 –Fig. 1.— Theoretical ratios of photon fluxes from the transitions O VIII 2 p P / , / → s S / (18.97 ˚A) and Ne IX 1 s p P → s S (13.45 ˚A) as a function of both isothermalplasma temperature, T , and model DEM peak temperature T max , compared with obser-vations of solar active regions analysed by McKenzie & Feldman (1992) and Schmelz et al.(2005). Each pair of solid (isothermal) and dashed (DEM) curves corresponds to the mini-mum spread in Ne/O abundance ratios that can possibly bracket the observations—the lowerlimit to the variation in observed Ne/O where the lowest flux ratios originate from regionswith temperature of minimum theoretical ratio, and vice versa. These limits are denotedby black solid horizontal lines. If regions are isothermal, the minimum range is 42%, fromNe/O=0.12 (lower solid curve) to 0.17 (upper solid curve) by number; for a typical DEM tem-perature structure (see text), the range is more than a factor of two, from Ne/O=0.09 (lowerdashed curve) to 0.22 (upper dashed curve). For clarity, error bars for McKenzie & Feldman(1992) intensity ratios are only plotted on the grey points with DEM-based temperatures. 16 –Fig. 2.— Ne/O abundance ratios derived from the McKenzie & Feldman (1992) solar activeregion line fluxes for both isothermal temperatures and using the model DEM (see text),compared with the recommended solar Ne/O ratio of Asplund et al. (2005), the “superseded”solar ratio of Grevesse & Sauval (1998, the value recommended by Asplund et al. 2009 liesin the middle of these two), the ± σσ