X-ray flares in Orion young stars. I. Flare characteristics

Abstract

Pre-main sequence (PMS) stars are known to produce powerful X-ray flares which resemble magnetic reconnection solar flares scaled by factors up to 10^4. However, numerous puzzles are present including the structure of X-ray emitting coronae and magnetospheres, effects of protoplanetary disks, and effects of stellar rotation. To investigate these issues in detail, we examine 216 of the brightest flares from 161 PMS stars observed in the Chandra Orion Ultradeep Project (COUP). These constitute the largest homogeneous dataset of PMS, or indeed stellar flares at any stellar age, ever acquired. Our effort is based on a new flare spectral analysis technique that avoids nonlinear parametric modeling. It can be applied to much weaker flares and is more sensitive than standard methods. We provide a catalog with >30 derived flare properties and an electronic atlas for this unique collection of stellar X-ray flares. The current study (Paper I) examines the flare morphologies, and provides general comparison of COUP flare characteristics with those of other active X-ray stars and the Sun. Paper II will concentrate on relationships between flare behavior, protoplanetary disks, and other stellar properties. Several results are obtained. First, the COUP flares studied here are among the most powerful, longest, and hottest stellar X-ray flares ever studied. Second, no significant statistical differences in peak flare luminosity or temperature distributions are found among different morphological flare classes, suggesting a common underlying mechanism for all flares. Third, comparison with the general solar-scaling laws indicates that COUP flares may not fit adequately proposed power-temperature and duration-temperature solar-stellar fits. Fourth, COUP super-hot flares are found to be brighter but shorter than ... ABRIDGED Comment: Accepted for publication in ApJ (07/11/08); 63 pages, 16 figures, 4 tables

Full text

Page 1

arXiv:0807.3005v1 [astro-ph] 18 Jul 2008
Accepted for publication in the Astrophysical Journal
07/11/08
X-ray flares in Orion young stars. I. Flare characteristics
Konstantin V. Getman1, Eric D. Feigelson1, Patrick S. Broos1, Giuseppina Micela2, Gordon
P. Garmire1
gkosta@astro.psu.edu
ABSTRACT
Pre-main sequence (PMS) stars are known to produce powerful X-ray flares
which resemble magnetic reconnection solar flares scaled by factors up to 104.
However, numerous puzzles are present including the structure of X-ray emitting
coronae and magnetospheres, effects of protoplanetary disks, and effects of stellar
rotation. To investigate these issues in detail, we examine 216 of the brightest
flares from 161 PMS stars observed in the Chandra Orion Ultradeep Project
(COUP). These constitute the largest homogeneous dataset of PMS, or indeed
stellar flares at any stellar age, ever acquired. Our effort is based on a new flare
spectral analysis technique that avoids nonlinear parametric modeling. It can
be applied to much weaker flares and is more sensitive than standard methods.
We provide a catalog with > 30 derived flare properties and an electronic atlas
for this unique collection of stellar X-ray flares. The current study (Paper I)
examines the flare morphologies, and provides general comparison of COUP flare
characteristics with those of other active X-ray stars and the Sun. Paper II will
concentrate on relationships between flare behavior, protoplanetary disks, and
other stellar properties.
Several results are obtained. First, the COUP flares studied here are among
the most powerful, longest, and hottest stellar X-ray flares ever studied. Peak
luminosities are in the range 31 < logLX,pk< 33 erg s−1; rise (decay) timescales
range from 1 hour to 1 day (few hours to 1.5 days); many peak temperatures
exceed 100 MK. The scale of their inferred associated coronal structures is
1Department of Astronomy & Astrophysics, 525 Davey Laboratory, Pennsylvania State University, Uni-
versity Park PA 16802
2INAF, Osservatorio Astronomico di Palermo G. S. Vaiana, Piazza del Parlamento 1, I-90134 Palermo,
Italy

Page 2

– 2 –
0.5 − 10 R⋆. Second, no significant statistical differences in peak flare luminos-
ity or temperature distributions are found among different morphological flare
classes, suggesting a common underlying mechanism for all flares. Third, com-
parison with the general solar-scaling laws indicates that COUP flares may not fit
adequately proposed power-temperature and duration-temperature solar-stellar
fits. Fourth, COUP super-hot flares are found to be brighter but shorter than
cooler COUP flares. Fifth, the majority of bright COUP flares can be viewed as
enhanced analogs of the rare solar “long-duration events”.
Subject headings: open clusters and associations: individual (Orion Nebula Clus-
ter) - stars: flare - stars: pre-main sequence - X-rays: stars
1.INTRODUCTION
All solar-type stars exhibit their highest levels of magnetic activity during their pre-
main sequence (PMS) phase (Feigelson et al. 2007). This includes ‘superflares’ with peak
luminosities logLx? 32 erg s−1in the 0.5−8 keV band, 104more powerful than the strongest
flares seen in the contemporary Sun (e.g. Tsuboi et al. 1998; Grosso et al. 2004; Favata et al.
2005). PMS stars thus join RS CVn binary systems (e.g. Osten et al. 2007) as laboratories
to study the physics of the most powerful magnetic reconnection events. PMS stars are more
distant and fainter than the closer RS CVn systems, but hundreds of flaring PMS stars can
be simultaneously studied due to their concentration in rich clusters.
The magnetic field structure of PMS stars, and thus the nature of their reconnection
and flaring, may (or may not) qualitatively differ from other stars due to the presence
of a protoplanetary disk during the early PMS stages. The intense high energy radiation
from these PMS reconnection events may affect the physical and chemical properties of
the surrounding circumstellar environment and play an important role in the formation
of planets (Glassgold et al. 2005; Feigelson et al. 2007). A consensus has emerged during
the past decade that PMS accretion is funneled by magnetic field lines linking the disk
inner edge to the stellar surface (e.g. Hartmann 1998; Shu et al. 2000). However, while early
theory assumed a dipolar field morphology, recent studies point to a complex multipolar field
structure similar to the Sun’s (Jardine et al. 2006; Donati et al. 2007; Long et al. 2008).
It is also unclear whether the X-ray flares occur primarily in large loops with both
footprints anchored on the stellar surface, or in loops linking the stellar photosphere with
the inner rim of the circumstellar disk (Isobe et al. 2003; Favata et al. 2005). The first
case may suffer instability due to centrifugal force (Jardine & Unruh 1999) while the second

Page 3

– 3 –
case may load the loop with cool accreting material so that X-rays may not be produced
(Preibisch et al. 2005).
The 13-day nearly continuous observation of ∼ 1408 PMS stars in the Orion Nebula, the
Chandra Orion Ultradeep Project (COUP; Getman et al. 2005a), enables both studies of in-
dividual flare properties and statistical studies of flaring from Orion stars (Wolk et al. 2005;
Flaccomio et al. 2005; Stassun et al. 2006; Caramazza et al. 2007; Albacete Colombo et al.
2007). COUP also provided a unique opportunity to study relatively rare superflares and
long-duration flares. Favata et al. (2005) have analyzed the strongest 32 flares in the COUP
dataset using a long-standing method of time resolved spectroscopy (TRS) modeled as cool-
ing plasma loops. They concluded that at least 1/3 of these are produced by magnetic
reconnection in very long coronal 5 − 20 R⋆ structures. Such structures were predicted
in magnetospheric accretion models (e.g Shu et al. 1997) but not clearly identified before
COUP. Favata et al. (2005) recognized that their sample was too small to quantitatively
probe the relationship between long coronal flaring structure and disks or accretion.
The aim of the current study is to extend the flare sample of Favata et al. (2005) utilizing
a more sensitive technique of flare analysis, the “method of adaptively smoothed median
energy” (MASME) introduced by Getman et al. (2006). We combine this method with the
astrophysical cooling loop models of Reale et al. (1997) to trace the evolution of the flare
plasma in temperature-density diagrams and derive flaring loop sizes. The method allowed us
to examine 216 of the brightest flares from 161 brightest COUP PMS stars. These constitute
the largest homogeneous sample of powerful stellar flares ever acquired in the X-ray band.
In Getman et al. (2008) (Paper II), we use these results to study in detail the relationships
between PMS X-ray flares, stellar properties, protoplanetary disks, and accretion.
Our flare analysis and the derived flare properties and classifications are presented in
§2. Properties of the stars themselves are also provided. Global properties of our flares are
considered in §3 and compared to published studies of older stars.
2.FLARE ANALYSIS
We analyze 216 of the brightest X-ray flares from 161 brightest COUP young stars.
This is a > 5-fold increase from the bright flare sample of Favata et al. (2005). The major
steps of our analysis are presented below.

Page 4

– 4 –
2.1. Flare Sample
Among 1616 COUP X-ray sources, 1408 have been associated with the stellar members
of the Orion region (Getman et al. 2005b). With the exception of 10 hot OB-type stars,
1398 are identified as cool young members of the Orion region (Feigelson et al. 2005).
We start with the sub-sample consisting of the brightest 188 cool Orion stars, those with
net counts NC ≥ 4000 as tabulated by Getman et al. (2005a, Table 4). All of them show
signs of variability, i.e. their X-ray lightcurves are characterized by at least two Bayesian
Block segments (Table 7 of Getman et al.) during the 13.2-day COUP observation. The
complete set of 1616 COUP lightcurves are available in the Source Atlas provided as a
figure set in the electronic edition of Getman et al. Individual examples of strong flares are
published in Favata et al. (2005) and Wolk et al. (2005).
Using the interactive software graphical program function 1d in the IDL-based TARA
package1, we identify flare-like events in which the peak count rate is ≥ 4 times that of
the “characteristic” level. The characteristic level is the typical pre-flare or inter-flare; as
explained by Wolk et al. (2005).We avoid the designation “quiescent” level because it
probably arises from the integrated effect of many weaker flares. This results in 216 flares
from 161 cool stars with median of ∼ 1500 counts. Only 20 flares have < 600 counts.
2.2.MASME Spectral Modeling
Our analysis of flare spectral evolution is based on the method of adaptively smoothed
median energy (MASME) introduced by Getman et al. (2006). We compare its results to
the long-standing method of time-resolved spectroscopy (TRS), which involves statistical
fitting of multiparameter spectral models, in Appendix A. The MASME method is simpler,
statistically more stable, and computationally quicker than TRS. It employs an adaptively
smoothed estimator of the median energy2of flare counts and count rate to infer the evo-
lution of plasma temperatures and emission measures during the flare. This procedure is
similar to the analyses of Reale & Micela (1998); Wargelin et al. (2007) which avoid nonlin-
1Description and code for the “Tools for ACIS Review & Analysis” (TARA) package developed at Penn
State can be found at http://www.astro.psu.edu/xray/docs/TARA/.
2X-ray source median energy is an observed quantity (Getman et al. 2005a) and can be effectively used
as an indicator of plasma temperature if absorbing column is known, or indicator of column density at
median energies > 1.7 keV (Feigelson et al. 2005). Other researches have also found that median energies
are effective spectral estimators in X-ray CCD spectroscopy (Hong et al. 2004).

Page 5

– 5 –
ear parametric flare modeling using hardness ratios.
The smoothing kernel is a rectangle (“boxcar”) of variable width; as it moves through the
time series its width is adjusted so that it encompasses a specified number of X-ray counts.
Kernels are evaluated on a time grid with bin width ∆t = 1.14 ks, chosen to divide the
total COUP observation span of 1140 ks into 1000 bins. Flux and median energy estimates
are computed from the counts found in each of these overlapping kernels, forming smoothed
flux and median energy time series which then used for spectral modeling. Similar flux and
median energy time series are produced by the ACIS Extract software package3. For each
flare, the target number of counts in the kernel is chosen such that the resulting smoothed
light curve closely matches the binned lightcurve given in the COUP Source Atlas of Getman
et al. (2005a). This generally results in ∼ 100 to ∼ 600 counts included in each kernel for
the weakest to the brightest flares, respectively. The typical width of the smoothing kernel
is a few kiloseconds at the peak of the flare and ∼ 10 ks at the base of the flare.
PMS X-ray spectra are typically modeled as emission by hot plasmas at one or two
temperatures subject to absorption by a column density, NH. If we adopt the NHinferred
from the time-integrated spectral fits of Getman et al. (2005a) and assume that NH does
not change during the COUP observation, then we can calibrate median energies and count
rates to plasma temperatures (kT) and emission measures (EM). Figure 1 shows the median
energy to plasma temperature calibration based on simulations similar to those described by
Getman et al. (2006). For each COUP source, the fakeit command in XSPEC (Arnaud 1996)
is used to simulate a grid of high signal-to-noise ACIS-I spectra with absorbed plasma mod-
els (WABS×MEKAL) and fixed column density. Simulated spectra are then passed through
ACIS Extract to perform photometric analysis, including calculation of fluxes (observed
count rates) and median energy. At each time point of interest comparison of simulated
count rates with that of the observed flare provides an estimate of flare emission measure;
comparison of simulated and observed median energies provides an estimate of flare plasma
temperature4. Due to the curved dependency of the true plasma temperature on observed
median energy (Figure 1), estimated plasma temperatures of ? 200 MK (? 17 keV) are ex-
pected to be quite reliable, while temperature values of ? 200 MK are much more uncertain.
The accuracy of our temperature estimates is discussed in detail in Appendix B. We find the
3ACIS
oped
http://www.astro.psu.edu/xray/docs/TARA/ae users guide.html.
Extract
Penn
isa
Chandra-ACIS
was
source
the
extraction
COUP
and
analysis
analysis package
is described
devel-
atState. Itused for dataand at
4XSPEC fitting of simulated spectra using the χ2statistic tends to overestimate plasma temperature. In
order to avoid this systematic bias in our simulations, we adopt the input plasma temperature rather than
the temperature derived from XSPEC fit of the simulated spectrum.