Elemental Abundances in Stellar Coronae with XMM-Newton

8 Pages·0.18 MB·

by  Marc Audard

1
ELEMENTAL ABUNDANCES IN STELLAR CORONAE WITH XMM-NEWTON
M.Audard and M.Gu¨del
Paul ScherrerInstitut,Wu¨renlingen and Villigen, 5232 Villigen PSI, Switzerland
2
Abstract 1996). EUVE spectra either indicated the absence of any
0
0 The XMM-Newton Reflection Grating Spectrometer FIP-related bias (Drake et al. 1995), or a solar-like FIP
2 effect (Drake et al. 1997) in inactive stellar coronae. The
Teamhasobtainedobservationsofalargenumberofcoro-
newX-rayobservatoriesXMM-Newton andChandra com-
n nal sources of various activity levels, ages, and spectral
a types. In particular, X-ray bright RS CVn binary sys- bine the high spectral resolution with moderate effective
J areas to routinely obtain data allowing to measure the
temsdisplaysaturatedcoronalemissionwithspectrallines
1 abundances in stellar coronae.
characteristic of hot (10-30 MK) plasma. Furthermore,
3
we have obtained XMM-Newton data from young solar Recently, Brinkman et al. (2001) showed a trend to-
1 analogsbothwithinandoutsidetheX-raysaturationreg- wardsenhancedhigh-FIPelementalabundances,whilelow-
v ime.WehavesimultaneouslyanalyzedtheEPICMOSand FIP abundances are depleted; this effect was dubbed the
0 RGS data from these objects and have obtained coronal “Inverse FIP” (IFIP) effect. Other active stars showed a
3 abundances of various elements (e.g., C, N, O, Ne, Mg, similar trend (Gu¨del et al. 2001a; Gu¨del et al. 2001b),
5
Si, Fe). We show that there is evidence for a transition except the intermediately active Capella (Audard et al.
1
0 from an Inverse First Ionization Potential (FIP) effect in 2001a).Notehoweverthatstellarcoronalabundanceshave
2 mostactivestarstoa“normal”solar-likeFIPeffectinless often been normalized to the solar photospheric abun-
0 active stars. We discuss this result with regard to photo- dances, while they should better be normalized to the
h/ spheric abundances. stellar photosphericabundances.Thelatteraredifficultto
p measure.Nevertheless,forsomestars,photosphericabun-
- Key words: Missions: XMM-Newton – stars: abundances dances are known. The uncertainty introduced by photo-
o
r – stars: coronae spheric abundances can then be removed. We will show
t that there is a transition from an IFIP to a normal FIP
s
a effectinthelong-termevolutionofthecoronaefromactive
v: toinactivesolaranalogs.We willusedataofbrightactive
i 1. Introduction RSCVn binarysystems tocomplementthe study aswell.
X
Finally,thevariationofstellarcoronalabundancesduring
r High-resolution X-ray spectra of stellar coronae obtained
flares will be discussed.
a
by XMM-Newton and Chandra now allow us to study in
detail the rich forest of X-ray lines emitted by elements
abundant in stellar coronae, such as C, N, O, Ne, Mg, Si, 2. Observations and Data analysis
S, Ar, Ca, Fe, and Ni. In the past, stellar coronal abun-
danceshavefrequentlybeen determinedusingthe moder- XMM-Newton observed several coronal sources as part
ate spectral resolution of CCD spectra from ASCA (e.g., of the RGS stellar Guaranteed Time Program. Obser-
Drake 1996; Gu¨del et al. 1999) or from the low sensitiv- vations of active bright RS CVn binary systems (e.g.,
ityspectrometersonboardEUVE (e.g.,Drakeetal.1995; HR1099,Capella,UXAri,λAnd)providedexcellenthigh-
Lamingetal.1996;Schmittetal.1996;Drakeetal.1997). resolutionRGSspectra.The solarpasthas alsobeenpro-
The abundance pattern in stellar coronae is complemen- bed with observations of young solar analogs of different
tarytotheabundancepatternintheSun:thesolarcorona, ages and activity levels (e.g., AB Dor, EK Dra, π1 UMa,
the solar wind, and solar energetic particles (and proba- χ1 Ori).TheRGS1,RGS2,andEPICMOS2spectrahave
bly also galactic cosmic rays) display a so-called “First beensimultaneouslyfitted(exceptforCapellawherethere
Ionization Potential” (FIP) effect, for which abundances are no EPIC data available) in XSPEC 11.0.1aj (Arnaud
of low-FIP (< 10 eV) elements are enhanced relative to 1996) using the vapec model (APEC code with variable
theirrespectivephotosphericabundances,whiletheabun- abundances). Because of the inaccuracy and incomplete-
dances of high-FIP (> 10 eV) elements are photospheric ness of atomic data for non-Fe L-shell transitions, signifi-
(Feldman 1992, see also Meyer 1985). Stellar coronal ob- cant parts of the RGS spectra had to be discarded above
servationshoweveroftenshowedadeficiencyofmetalsrel- 20 ˚A. Furthermore, some Fe L-shell lines with inaccurate
ativetothesolarphotosphericabundances(Schmittetal. atomic data were not fitted. For additional information
Proc. Symposium ‘New Visions of the X-ray Universe in the XMM-Newton and Chandra Era’ 26–30 November2001,
ESTEC, The Netherlands
2
onthe data analysis,we refer to Audardetal.(2002)and displayasimilarabundancebias,closetothatobservedin
Gu¨del et al. (2002). theSun.NotethatthetransitionintheFIPbiasoccursfor
low-FIP elements while the high-FIP elements appear to
3. Results have similar coronal abundances (normalized to oxygen)
in all stars.
3.1. Coronal abundances of active stars
Based on the emission measure distributions obtained
The RGS spectra of four RS CVn-type systems shown in
frommulti-temperaturefits,wehavecalculatedalogarith-
Figure 1 display bright H-like and He-like emission lines
mic averagetemperature anddefined the latter as the av-
(mostly from C, N, O, Ne, Mg, Si) and numerous Fe L-
eragecoronaltemperature.Whilemostinactivestarshave
shell lines. Although the overallspectralfeatures are sim-
temperatures between 4 and 6 MK, RS CVn binary sys-
ilarin HR 1099,UX Ari, andλ And, there aredifferences
tems display average quiescent temperatures around 10–
in terms of line intensities or ratios. This can suggest dif-
20 MK. Previous analysis showed a correlation between
ferences in the emission measure distributions, or in the
the X-ray luminosity of solar analogs and their average
elemental composition in their coronae. A hint pointing
coronal temperatures (Gu¨del et al. 1997). The average
toward the latter interpretation comes from the different
coronaltemperature can therefore be considered to be an
ratiosofthe Fe xvii (at 15˚A)andNe ix lines, whichhave
activity indicator. The abundances of low-FIP elements
similar maximum formation temperatures. The line ratio
correlate with the coronal temperature. While they are
in Capella definitely points toward an abundance effect,
depletedin the mostactivestars,atransitionoccurswith
withahighFe/Neratio.Wehaveperformedfits(see§2)to
decreasing temperature, and their abundances drastically
the data and modeled the X-ray spectra. Coronal abun-
increase(relativetohigh-FIPelements).Ontheotherside,
dances (normalized to the oxygen abundance to remove
the abundances of high-FIP elements stay constant. In
the uncertainty that might be introducedby the determi-
Figure5wegiveexamplesforFeandNe,representinglow-
nation of the underlying continuum) relative to the solar
FIP elements and high-FIP elements, respectively. Note
photosphericabundancesareshowninFig.2asafunction
thatthisbehaviorisreproducedinotherlow-FIPelements
of the first ionization potential of the element. A similar
(e.g., Mg, Si) and high-FIP elements (e.g., C, N).
trend (IFIP effect) as found by Brinkman et al. (2001) is
seeninUXAri,withlow-FIPelementsdepletedrelatively
3.2. Coronal abundance during flares
tothehigh-FIPelements,andconfirmedinthisnewanal-
ysis of HR 1099. In contrast, the FIP bias in λ And is
lessclear.TheabsenceofanyFIPbiasis suggestedinthe Intheprevioussection,coronalabundancesofactivestars
intermediately active binary Capella. inquiescence havebeenreportedtoshowatransitionfrom
Since reliable determinations of stellar photospheric under-tooverabundantlow-FIPelementswithdecreasing
abundances in active stars are rare, we cannot normal- activity (or coronal temperature), while the high-FIP el-
ize the derived coronal abundances by their photospheric emental abundances stay constant (relative to oxygen).
counterparts. Hence, it is possible that the FIP bias ob- However,previousdatashowedthatthe averagemetallic-
served in HR 1099 and UX Ari is simply a reflection of ity Z or the Fe abundance canincreaseduring largeflares
their photospheric composition. However,it is possible to (e.g., Ottmann & Schmitt 1996). Gu¨del et al. (1999) ob-
use stars with known photospheric abundances to deter- tainedtime-dependentmeasurementsofseveralelemental
mine whether a FIP effect (or its inverse) exists in active abundances during a large flare in UX Ari with ASCA.
stars. We have used spectra from solar analogs (Fig. 3) The abundance of low-FIP elements increased more sig-
for which the photospheric composition is close to solar. nificantly than those of high-FIP elements. Recently, Au-
These stars represent the Sun in its past and probe its dardetal.(2001b)found asimilarbehaviorinthe XMM-
activity in its infancy. While AB Dor is strictly speaking Newton data of a flare in HR 1099. We have redone the
notasolaranalog(spectraltypeK0),itsactivityissimilar analysis of this flare using a more recent calibration. Fig-
to that of a very active young Sun. Similarly to the RS ure 6 shows the Fe/O and Ne/O ratios versus the coro-
CVn binary systems, the differences in the spectral fea- naltemperatureofHR1099,beforetheflare(quiescence),
tures do not depend solely on the emission measure dis- duringtheflarerise,andatflarepeak(nocompletedecay
tribution,butalsoonintrinsicvariationsofcoronalabun- available). Consistently with our previous analysis (Au-
dances.Spectralfitsconfirmthatthecoronalcomposition dard et al. 2001b), we have found that the absolute Fe
ofoursampleisdifferentforeachstar.Figure4showsthe abundance increasesduring the risingpartof the flare.In
coronalabundancesasafunctionoftheFIP,likeinFig.2. contrast,theabsoluteNeabundancestaysconstant.Other
The FIP bias appears to correlate with the activity level low-FIP elements and high-FIP elements show similar re-
(orage):anIFIPeffectisfoundinthemostactivestarAB spective behavior.Note, however,that the signal-to-noise
Dor, like in the most active RS CVn binaries, while the ratioin the RGStime-dependent spectra did notallowus
intermediately active EK Dra shows no pronounced bias. to better sample the flare event. Longlasting strong flares
Finally, the oldest, less active stars π1 UMa and χ1 Ori are needed to obtain high-quality spectra of a flare.
3
4. Conclusions Acknowledgements
WeacknowledgesupportfromtheSwissNationalScienceFoun-
High-resolutionX-rayspectraofmagneticallyactivestars dation(grant2100-049343).Thisworkisbasedonobservations
have been investigated with the Reflection Grating Spec- obtainedwithXMM-Newton,anESAsciencemissionwithin-
trometers on board XMM-Newton. The high-energy data struments and contributions directly funded by ESA Member
(> 1.5 keV) of the EPIC CCD spectra were used to bet- States and the USA(NASA).
ter constrain the high-temperature part of the emission
measure distributions and to profit from the presence of References
H-like and He-like transitions of Si, S, Ar, and Ca. It was
found that the most active stars, such as the bright RS Anders,E.,&Grevesse,N.1989,Geochim.Cosmoschim.Acta,
CVnbinarysystems,showa markeddepletionoflow-FIP 53, 197
Arnaud, K. A. 1996, in ASP Conf. Ser. 101, Astronomical
elements (e.g., Fe, Mg, Si) relative to high-FIP elements
DataAnalysisSoftwareandSystemsV,ed.G.Jacoby&J.
(e.g., C, N, O, Ne), opposite to the FIP effect observed
Barnes (San Francisco: ASP),17
in the solar corona. This “inverse FIP” effect is however
Audard, M., Behar, E., Gu¨del, M., et al. 2001a, A&A, 365,
not observed in the intermediately active RS CVn binary
L329
Capella. Since their photospheric abundances are mostly Audard,M., Gu¨del, M., & Mewe, R.2001b, A&A,365, L318
unknownornotreliable,onecanhypothesizethattheob- Audard,M., Gu¨del, M., et al. 2002, A&A,in preparation
servedFIPbiasissimplyareflectionoftheirphotospheric Brinkman,A.C.,Behar,E.,Gu¨del,M.,etal.2001,A&A,365,
composition. L324
To remove the uncertainty of surface abundances, we Drake, J. J., Laming, J. M., & Widing, K.G. 1995, ApJ,443,
393
have analyzed high-resolution X-ray spectra of solar ana-
Drake, J. J., Laming, J. M., & Widing, K.G. 1997, ApJ,478,
logs of known photospheric composition (close to solar).
403
These solar-likestars spana wider rangeof coronalactiv-
Drake, S. A. 1996, in Proceedings of the 6th Annual October
ity (from inactive to active) and represent the evolution
AstrophysicsConference inCollege Park,eds.S.S.Holt&
of the solar corona in time. We have found a transition
G. Sonneborn,(San Francisco: ASP),215
from a depletion of low-FIP elements (relative to high- Feldman, U. 1992, Physica Scripta, 46, 202
FIP elements) in the most active stars toward a marked Grevesse, N., & Sauval, A.J. 1999, A&A,347, 348
enhancementoftheirabundancesintheinactivestars.On Gu¨del, M., Audard, M., Briggs, K., et al. 2001a, A&A, 365,
the other hand, the abundances of high-FIP elements do L336
not vary with the activity level (or coronal temperature), Gu¨del, M., Audard, M., Magee, H., et al. 2001b, A&A, 365,
relativetoO.TheIFIPeffectfoundintheactiveRSCVn L344
Gu¨del,M.,Audard,M.,Sres,A.,Wehrli,R.,&Mewe,R.2002,
binary systems fit well into this transition, under the as-
ApJ, submitted
sumptionthattheirphotosphericcompositionisalsoclose
Gu¨del,M.,Guinan,E.F.,&Skinner,S.L.1997,ApJ,483,947
to solar. Similarly, the solar FIP effect (enhancement of
Gu¨del, M., Linsky,J. L., Brown, A., & Nagase, F. 1999, ApJ,
low-FIP elements by factors of 4–8) fits into this picture.
511, 405
However,althoughthe scenarioof correlatingthe activity
Laming, J. M., Drake, J. J., & Widing, K.G. 1996, ApJ,462,
levelwiththe FIPbiasistempting,itmaybe toosimplis- 948
tic; indeed, such scenario does not explain the absence of Meyer, J.-P. 1985, ApJS,57, 173
any FIP bias in the corona of the old, inactive Procyon Ottmann,R., & Schmitt, J. H. M. M. 1996, A&A,307, 813
(Drake et al. 1995). Schmitt, J. H. M. M., Stern, R. A., Drake, J. J., & Ku¨rster,
M. 1996, ApJ, 464, 898
We have put forward first ideas to explain the inverse
FIPeffectseeninactivestars:downwardpropagatingelec-
tronsdetectedbytheirgyrosynchrotronemissioninactive
stars could prevent chromospheric ions (mostly low-FIP
elements) from escaping into the corona by building up a
downward-pointing electric field (Gu¨del et al. 2002). As
the density of high-energy electrons decreases with de-
creasing activity, the inverse FIP effect is quenched. Dur-
ing large flares,however,the high-energy electrons heat a
significantportionofthechromospheretobringupanear-
photosphericmixtureofelementsintothecorona,andthis
effect has indeed been observed (Gu¨del et al. 1999; Au-
dardet al.2001b).The new results byXMM-Newton and
Chandra have opened a new field of research relevant to
thephysicsofheatinganddynamicsofouterstellaratmo-
spheres.
4
αSi XIII He αMg XII LyαMg XI HeβNe X LyFe XXIVαNe X LyαNe IX He Fe XVIIFe XVIIIFe XVII αO VIII Ly αO VII He αN VII Ly αC VI Ly
25
20
15 HR 1099
10
5
0
)
1 15
−
Å

2
− 10
m UX Ari
c

1
− 5
s

h
p
0
3 12
−
0
10
1
(
8
x λ
And
u
l 6
f

n 4
o
ot 2
h
0
P 50

40
30 Capella
20
10
0
10 15 20 25 30 35
Wavelength (Å)
Figure1. RGS spectra of bright active RS CVn binary systems. The sources have been ordered with decreasing activity
levels (or average coronal temperature) from top to bottom. Major emission lines have been labeled.
5
10.0
Ca Ni Si S C O N Ar Ne
e
F
g
M
1.0
HR 1099
0.1
10.0

c
i
r
e
h
p 1.0
s
o
t
o
h UX Ari
p
r
a
ol 0.1
s10.0
o
t
e
v
i
t
a
l
e 1.0
r
O,
A/ λ And
0.1
10.0

1.0
Capella
0.1
7 10 20
First Ionization Potential [eV]
Figure2. Coronal abundance normalized to oxygen in RS CVn binaries as a function of the First Ionization Potential.
Solar photospheric abundances from Anders & Grevesse (1989) were used, except for Fe (Grevesse & Sauval 1999).
6
αSi XIII He αMg XII LyαMg XI HeβNe X Ly αNe X LyαNe IX He Fe XVIIFe XVIIIFe XVII αO VIII Ly αO VII He αN VII Ly αC VI Ly
12
10
AB Dor
8
6
4
2
0
2.0

)
1
1.5
−
Å EK Dra

1
− 1.0
s

2
−
0.5
m
c

h 0
p

3 2.0
−
0 π1
1 UMa
( 1.5

x
u
fl 1.0

n
o
ot 0.5
h
P
0

5

4 χ1 Ori

3

2

1

0
10 15 20 25 30 35
Wavelength (Å)
Figure3. RGS spectra of solar analogs. Their order is set similarly to Fig. 1. The arrows designate lines with similar
maximum formation temperature; hence different line ratios of the Fe xvii line at 15˚A and the Ne ix line suggest
differences in coronal abundances in each star.
7
3
MgFeSi S C O N Ar Ne
2

1
)
r
a
l AB Dor
o
s 0

o
t 8 EK Dra
e
v
i
t 6
a
l
e
(r 4

O
o 2
t

d
e 0
liz 8 π1 UMa
a
m
r 6
o
n

e 4
c
n
a
d 2
n
u
0
b
A 1
8 χ Ori

6

4

2

0

7 10 20

First Ionization Potential (eV)

Figure4. Coronal abundance normalized to oxygen in solar analogs as a function of the First Ionization Potential.
Note that the activity level decreases (their age increases) from top to bottom. Similar photospheric abundances have
been taken as in Fig. 2.
8
c 10.0 c 10.0
eri eri
h h
p p
s s
o o
ot ot
h h
p p
ar ar
sol 1.0 sol 1.0
o o
e t e t
v v
ati Low−FIP ati High−FIP
el el
O, r O, r
Fe/ 0.1 Ne/ 0.1
2 4 6 8 10 12 14 2 4 6 8 10 12 14
Coronal Temperature (MK)
Figure5. Coronal abundances (normalized to O) as a function of the average coronal temperatures, for Fe (low-FIP;
left) and Ne (high-FIP; right). The data include solar analogs and RS CVn binaries.
c 2.0 c 5
eri eri
h h
sp sp 4
o 1.5 o
ot ot
h h
p p
ar ar 3
ol 1.0 ol
s s
o o
e t e t 2
v v
ati 0.5 ati
el el 1
O r Q R P O r Q R P
e/ e/
F 0.0 N 0
5 10 15 20 25 30 35 5 10 15 20 25 30 35
Coronal Temperature (MK)
Figure6. Coronal abundances (normalized to O) as a function of the average coronal temperature during a large flare
in HR 1099. Left panel gives Fe/O ratios, while the right panel gives Ne/O ratios. ‘Q’ stands for quiescent, ‘R’ for
flare rise, and ‘P’ for flare peak.