Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://solarwind.cosmos.ru/txt/2011/presentations/Sun/Kuznetsov/JASR2006-S-06-01000.pdf Äàòà èçìåíåíèÿ: Fri Apr 15 15:32:24 2011 Äàòà èíäåêñèðîâàíèÿ: Mon Oct 1 21:04:41 2012 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: ï ï ï ï ï ï ï ï ï ï

Elsevier Ed itorial System(tm) for JASR 2006

Manuscript Draft

Manuscript Number: Title: Determination of K, Ar, Cl, S, Si and Al flare abundances from RESIK soft X-ray spectra Article Type: Contributed Paper Section/Category: E2.4 - Solar Magnetic Field and Activities Keywords: solar physics; X-ray flares; spectroscopy; abundances Corresponding Author: Dr. J. Sylwester, Corresponding Author's Institution: Space Research Center First Author: J. Sylwester Order of Authors: J. Sylwester; Barbara Sylwester, Dr.; Enrico Landi, Dr.; Kenneth J Phillips, Dr.; Vladimir D Kuznetsov, Prof. Abstract: The RESIK is a high sensitiv ity, uncollimated bent crystal spectrometer which successfully operated aboard Russian CORONAS-F solar mission between 2001 and 2003. It measured for the first time in a systematic way solar soft X-ray spectra in the four wavelength channels from 3.3 A to 6.1 A. This range includes characteristic strong lines of H- and He-like ions of K, Ar, Cl, S, Si and Al in the respective spectral channels. A distinguishin g feature of RESIK is its possib ility of making reliable measurements of the continuum radiation in flares. Interpretation of line and the continuum intensities observed in vicinity of these triplets provides diagnostics of plasma temperature and absolute abundances of K, Ar, Cl, S, Si and Al in several flares. We analyzed the observed intensities of spectral lines and the nearby continuum using the CHIANTI v5.2 atomic data package. A specific, so-called "locally isothermal" approach has been used in this respect


allowing us to make not only flare-averaged abundance estimates, but also to look into a possible variability of plasma composition during the course of flares.


Manuscript

Determination of K, Ar, Cl, S, Si and Al flare abundances from RESIK soft X-ray sp ectra
J. Sylwester a,, B. Sylwester a,
a

Space Research Centre, Polish Academy of Sciences,ul. Kopernika 11, PL-51-622 Wroclaw, Poland

E. Landi b,
b

Artep, Inc. at Naval Research Laboratory, USA

K.J.H. Phillips c,
c

Mul lard Space Science Laboratory, University Col lege London, U.K.

V.D. Kuznetsov d,
d

IZMIRAN, Russian Academy of Sciences, Troitsk, Russia

Abstract The RESIK is a high sensitivity, uncollimated b ent crystal sp ectrometer which successfully op erated ab oard Russian CORONAS-F solar mission b etween 2001 and 2003. It measured for the first time in a systematic way solar soft X-ray sp ectra in ° the four wavelength channels from 3.3 ° to 6.1 A. This range includes characteristic A strong lines of H- and He-like ions of K, Ar, Cl, Si, S and Al in the resp ective sp ectral channels. A distinguishing feature of RESIK is its p ossibility of making reliable measurements of the continuum radiation in flares. Interpretation of line and the continuum intensities observed in vicinity of these triplets provides diagnostics of plasma temp erature and absolute abundances of K, Ar, Cl, S, Si and Al in several flares. We analyzed the observed intensities of sp ectral lines and the nearby continuum using the CHIANTI v5.2 atomic data package. A sp ecific, so-called "locally isothermal" approach has b een used in this resp ect allowing us to make not only flare-averaged abundance estimates, but also to look into a p ossible variability of plasma comp osition during the course of flares. Key words: solar physics, X-ray flares, sp ectroscopy, abundances

Corresp onding author Email address: js@cbk.pan.wroc.pl (J. Sylwester).

Preprint submitted to Elsevier Science

1 Novemb er 2006


1

RESIK Bent Crystal Spectrometer

Fig. 1. An example of a calibrated RESIK sp ectrum. This sp ectrum was accumulated over entire 2003 February 22 flare which took place around 09:30 UT. In gray, several sp ectral bands are marked, selected to study the absolute abundance of indicated resp ective elements.

The RESIK consists of two double-channel X-ray spectrometers equipped with bent crystals, a design similar to the BCS spectrometers aboard SMM and Yohkoh. RESIK was designed to observe hot solar plasmas. It was included in the scientific payload of the Russian solar satellite mission CORONAS-F. The detailed description of the RESIK instrument, its operation and calibration are presented in the paper by Sylwester et al. (2005). Here we summarize a number of instrument characteristics important for the present study. The nominal wavelength coverage of RESIK was 3.35 ° - 6.1 ° This range conA A. tains several spectral features useful for the X-ray plasma diagnostics. The observed line intensities can be used in order to study the physical conditions in the flaring plasma as well as to investigate its relative (element-to-element) composition. Inclusion of the continuum, also reliable measured, allows us to measure the absolute (relative to hydrogen) abundances, as substantial contribution to the continuum comes from bremsstrahlung emission. In the RESIK spectra one can find lines belonging to the elements with substantially different values of FIP: from the lowest being 4.34 eV (K), the highest being 15.75 eV (Ar). Other lines are identified by Sylwester et al. (2006) and Kepa et al. (2006). They include the triplets of He-like ions (K xviii, Ar xvii, Cl xvi and S xv) as well as lines corresponding to (n > 3) (1) transitions in Si and Al. 2


2

Spectra observed and new abundance analysis approach

For the purpose of the present analysis we have selected 7 typical flares, well observed by RESIK. The flares selected represent short and longer duration events seen on the disk and/or at the limb. All flares were observed early

Fig. 2. Nine examples of the fit b etween observed (histogram) and synthetic sp ectra. In each row, the abundance of the element of interest (Ar) is kept constant and the plasma temp eratures corresp ond to three different values. In each column, the plasma temp erature is constant and the abundance of Ar is set to three different values. In the central panel, the optimum fit is shown, characterized by the smallest value of 2 parameter. Such an optimum fit has b een obtained for each of the sp ectra recorded throughout the flare.

3


in 2003, when the instrument settings were optimised (cf. Table 1). We have carefully calibrated the observed spectra, establishing an absolute wavelength scale and determining the absolute spectral photon fluxes. We incorporated all known corrections which essentially remove the instrumental fluorescence radiation from the RESIK bent crystals material. This leads us to believe that the level of the continuum seen in the longer wavelength channels of RESIK is not contaminated. For the two shorter wavelength bands, the contribution of this fluorescence to the continuum is found not to be a problem as it amounts to less than one percent.

Fig. 3. left: p otassium Contours of min(2 )+ 1 for the 2003 February 22, 09:30 UT flare. Individual shades of the contours represent the time-in-flare, with dark shades representing fits to the sp ectra observed earlier on in the rise phase. The light gray contours represent the decay phase of the flare. The "phot. " and "cor." horizontal lines designate the photospheric and coronal abundance levels for p otassium as taken from Asplund et al. (2005) and Feldman & Laming (2000) resp ectively. right: argon A corresp onding diagram shown for more abundant element Ar. Here the min(2 )+ 1 contours are more compact which indicates that resp ective abundance uncertainties are much smaller. The inset in the upp er right shows the actual shap e of the 1/2 surface at one time corresp onding to flare maximum.

In the analysis, we have used a novel "lo cally isothermal" (LI) approach in order to separate the impact of abundance variability from the temperature (multitemperature-DEM) effects on the spectra as much as possible. In the LI approach we make "implicit use" of somewhat similar temperature dependence of the line and the nearby continuum emission functions in the region of the effective line formation. This similarity causes the abundance effects to dominate over DEM effects in the Line-to-Continuum (L/C) ratio variability. Our non-standard determination technique is based on converging iterative scheme, where we start from the measured value of a total (wavelength integrated) flux in entire selected wavelength band including the lines of the element of interest and substantial part of near-by continuum. A counterpart, 4


Fig. 4. Time dep endence of derived flare abundances. Two flares are shown in the figure in resp ective columns: a fast evolving flare on 2003 February 22 (left), and a "double" flare of 2003 January 9. In the upp er part, the flare flux in the Ar sp ectral band is plotted for each flare. The uncertainties plotted corresp ond to the vertical extension of the iso-contours shown in Figure 3. Resp ective coronal and photospheric abundance levels are given for each element as dotted and dashed lines resp ectively.

theoretically predicted fluxes in the same selected band depend on the assumed plasma temperature, emission measure and the composition. By varying values of the two parameters within the calculation cube, i.e. for a set of selected 5


temperatures and particular element abundances (40 linearly separated values from "zero" up to four times the coronal abundance value of Feldman & Laming, 2000), we determine respective emission measure values from the measured band fluxes. These values are next used to calculate the shape of synthetic spectrum in each of the observed individual spectral bins contained in the analyzed band. Such synthetic spectra are then compared with the observed spectra in terms of the 2 statistics. This comparison makes use of the measurement uncertainties in each spectral bin separately. By performing the calculations for each of the mesh values of the 2D parameters temperature­ abundance (T - AEl ) grid, we construct the 2 surface, identify the position of its lo cal minimum in the parameter space, and determine the 1 contours according to the classical approach of Bevington (1969). Example result of the described pro cedure is illustrated in Figure 2. As mentioned, the quality of the spectral fits between the observed and synthetic spectra was characterized by the value of 2 statistics which in the temperature (T - AEl ) parameter space is represented as a surface. The position of the minimum and the shape of the 2 contour provide a realistic estimate of the instantaneous value of average temperature of formation for the line and nearby continuum for each measured spectrum, as it evolves throughout the flare. It is also possible to estimate the uncertainty of both the temperature and abundance. For purposes of the spectral synthesis task, we pre-calculated an extensive 10 A, GB-database consisting of a grid of theoretical spectra, with = 0.001° for 101 temperatures between 1 - 100 MK, for 6 elements: K, Ar, Cl, S, Si and Al, and for 41 abundance value of each element. The calculations were based on CHIANTI v5.2 spectral co de, a part of SolarSoft. The important free-free and free-bound pro cesses have all been included in the calculations of the continuum. As checked by Chifor et al. (2006) the continuum calculations following from CHIANTI are in very go o d agreement (few percent) with the earlier results of Mewe et al. (1986). As these two sets of calculations depend on independent approximations for the basic atomic cross-sections, such a go o d agreement increases our confidence in derived values of the abundances discussed in the present study.

3

The Results

In Figure 4 we show examples of time dependence of best-fit abundance determinations for two flares observed by RESIK in contiguous manner. One flare was of a fast rise and decay and have fallen entirely within the 20 min observation window. However it was a rather faint event, and therefore the statistically reliable spectra collection time was necessarily longer for the early rise and late decay phases. The other flare was much stronger with the two pronounced maxima. It was possible to study its spectral variability each few 6


Table 1 Average absolute flare abundances [10-6 ] FIP [eV] Flare Element 2003-Jan-04 06:30 B9.2 S23 E50 2003-Jan-07 07:50 M1.0 S24 E08 2003-Jan-07 23:30 M4.9 S11 E08 2003-Jan-09 01:39 C9.8 S09 W25 2003-Jan-21 15:26 M1.9 S07 E90 2003-Jan-25 18:55 C4.4 N13 W27 2003-Feb-22 09:30 C5.8 N16 W05 AEl photospherica AEl coronalb
a b

4.34 K 0.53 ±0.4 0.52 ±0.4 0.66 ±0.4 0.65 ±0.4 0.43 ±0.3 0.70 ±0.5 0.76 ±0.4 0.115 0.468

15.76 Ar 2.28 ±0.5 2.26 ±0.6 2.38 ±0.9 2.53 ±0.4 2.69 ±0.7 2.79 ±0.8 2.60 ±0.8 1.51 3.80

12.97 Cl 0.52 ±0.2 0.35 ±0.3 0.41 ±0.3 0.25 ±0.3 0.36 ±0.4 0.32 ±0.3 0.28 ±0.2 0.170 0.316

10.36 S 8.96 ±1.6 10.1 ±1.4 9.46 ±1.5 9.12 ±1.2 11.3 ±2.1 9.81 ±3.6 9.41 ±1.5 14.45 18.62

8.15 Si 17.5 ±6.1 30.5 ±17 19.8 ±6.6 23.8 ±4.2 26.2 ±13 24.9 ±23 23.0 ±6.0 32.36 125.9

5.99 Al 13.3 ±5.8 7.26 ±9.4 11.3 ±10 5.04 ±4.7 7.45 ±8.5 9.68 ±9.1 13.3 ±7.1 2.69 10.96



- from Asplund, Grevese and Sauval, (2005) - from Feldman and Lamming, (2000)

seconds. In the Figure we plot the abundance values for each element in question together with respective ±1 uncertainties. These uncertainties represent thesizeofthe min(2 )+1 contour from Figure 3, as pro jected onto the abundance axis. It is worth noting that the results presented in Figure 3 show for the first time the plasma composition variability during flares, as determined from spectroscopic data. Examination of Figure 4 and similar plots constructed for the other flares studied, allows us to draw a number of conclusions. · For many events the plasma composition is approximately constant in time. · For some events like 2003 February 22, AAr appears to anticorrelate with AK ; these two elements have the largest FIP contrast among those studied. · For the events shown in Figure 4, it is highly improbable that the time variability pattern observed o ccurs by chance, and therefore, we conclude that substantial plasma composition variability takes place during flares. If the above conclusions are confirmed in a following larger study covering many tens of flares observed by RESIK, the physical consequences are expected to be profound for considerations concerning the origin (source) of matter 7


contributing to soft X-ray flaring plasma. It is possible, that such matter may be characterized by a highly non-uniform composition probably influenced by plasma-magnetic field mechanisms acting selectively on semi-ionized gas over longer perio ds. Based on the observed composition variability pattern for the analyzed flares, it was possible to determine seven flare-averaged plasma abundance sets for the six elements studied. These average values are given in Table 1, together with formal ±1 uncertainties. The uncertainties were calculated taking into account the scatter about the mean value. The uncertainties are large for cases where the abundance appears to be time dependent. 4 Summary and Conclusions

In this study we present analysis of RESIK bent crystal spectrometer soft X-ray spectra for seven well-observed flares. The novel nature of the spectral analysis applied, as well as unprecedented quality of the spectra measured, have allowed us to study time-variability of absolute elemental composition for the following elements: K, Ar, Cl, S, Si and Al. The results shown in Figure 4 indicate possible changes in flaring plasma composition as the flare progress. The flare-averaged composition for seven analyzed flares (presented in Table 1) is seen to vary from event-to-event. The amplitude of variability is the largest for the low-FIP elements like potassium. Derived average abundances of sulphur and silicon are significantly below, at the level of (60%) of the photospheric values taken from Asplund et al. (2005). Abundances of Ar are between the photospheric and coronal (Feldman & Laming , 2000) values, while for potassium and aluminum the derived values are significantly above the assumed coronal level. Notwithstanding of a careful analysis of the instrumental fluorescence contribution, there is still a possibility of some uncounted contamination of fluorescence in the RESIK channel where S, Si and Al lines are observed. Work is in progress and we hope to present completely unbiased estimates for these elements in the subsequent extended study following the outline presented here. Acknowledgements RESIK is a common pro ject between NRL (USA), MSSL and RAL (UK), IZMIRAN (Russia) and SRC (Poland). B.S and J.S. acknowledge support from the Polish Ministry of Education and Science grant 1.P03D.017.29.

References Asplund, M., Grevesse, N., & Sauval, A.J., Cosmic Abundances as Records of Stellar Evolution and Nucleosynthesis, 2005, ASP Conf. Ser., 336, 25-38. 8


Bevington, P. R., Data Reduction and Error Analysis for the Physical Sciences, McGraw-Hill Bo ok Company, (New York, 1969). Chifor, C., Del Zanna, G., Mason, H. E., Sylwester, J., Sylwester, B., & Phillips, K. J. H., A benchmark study for CHIANTI based on RESIK solar flare spectra, 2006, to be published in Astron. and Astrophys.. Dere, K.P., Landi, E., Mason, H.E., & others,, An Atomic Database for Emission Lines, 1997, Astron. and Astrophys. Suppl., 125, 149-173. Feldman, U. & Laming, J. M., Elemental abundances in the upper atmospheres of the Sun and stars: Update of observational result, 2000, Physica Scripta, 61, 222-252. Kepa, A., Sylwester, J., Sylwester B., Siarkowski, M., Phillips, K.J.H., & Kuznietsov, V.D., Observations of 1s2 - 1snp and 1s - np lines in RESIK soft X-ray spectra, 2006, to be published in Adv. Space Res. Mewe, R., Lemen, J. & van den Oord, G. H., J. Calculated X-radiation from optically thin plasmas. VI - Improved calculations for continuum emission and approximation formulae for nonrelativistic average Gaunt factors, 1985, Astronomy and Astroph. Suppl., 65, 511-536. SolarSoft ­ http://sohowww.nascom.nasa.gov/solarsoft/ Sylwester, J., Gaicki, I., Kordylewski Z. et al., RESIK: A bent crystal X-ray spectrometer for Studies of solar coronal plasma composition, 2005, Solar Physics, 226, 45-72. Sylwester, B., Sylwester, J., Siarkowski, M., & 5 others, Lines in the range 3.2 ° - 6.1 ° observed in RESIK spectra, 2006, to be published in Adv. A A Space Res.

9