A deep X-ray view of the Class I YSO Elias 29 with XMM-Newton and NuSTAR
I. Pillitteri, S. Sciortino, F. Reale, G. Micela, C. Argiroffi, E. Flaccomio, B. Stelzer
AAstronomy & Astrophysics manuscript no. elias29_manuscript c (cid:13)
ESO 2019January 24, 2019
A deep X-ray view of the Class I YSO Elias 29 with
XMM-Newton and
NuSTAR . (cid:63) I. Pillitteri , , S. Sciortino , F. Reale , , G. Micela , C. Argiro ffi , E. Flaccomio , and B. Stelzer , INAF-Osservatorio Astronomico di Palermo, Piazza del Parlamento 1, 90134 Palermo, Italye-mail: [email protected] Harvard-Smithsonian Center for Astrophysics, 60 Garden St., 02138 Cambridge MA, USA Università degli Studi di Palermo, Piazza del Parlamento 1, 90134 Palermo, Italy Eberhard Karls Universität, Institut für Astronomie und Astrophysik, Sand 1, 72076 Tübingen, GermanyReceived ; accepted
ABSTRACT
X-ray emission is a characteristic feature of young stellar objects (YSOs) and the result of the interplay between rotation, magnetismand accretion. For this reason high energy phenomena are key elements to understand the process of star formation, the evolutionof their circumstellar disks and eventually the formation of planets. We investigated the X-ray characteristics of the Class I YSOElias 29 with joint
XMM-Newton and
NuSTAR observations of total duration 300 ks and 450 ks, respectively. These are the firstobservations of a very young ( < . − XMM-Newton , and ≈ −
80 keV in
NuSTAR ). The quiescent spectrum is well described by one thermal component at ∼ . N H ∼ . × cm − . In addition to the hot Fe complex at 6.7 keV, we observed fluorescent emission from Fe at ∼ . ≈ . − − . XMM-Newton and
NuSTAR . For this flare, we used its peak temperature and timing as diagnostics to infer a loop size of about 1 − (cid:12) in length, whichis about 20% −
30% of the stellar radius. This implies a relatively compact structure. We systematically observed an increase of N H during the flares of a factor five. This behavior has been observed during flares previously detected in Elias 29 with XMM-Newton and ASCA. The phenomenon hints that the flaring regions could be buried under the accretion streams and at high stellar latitudes,as the X-rays from flares pass through gas denser than the gas along the line of sight of the quiescent corona. In a di ff erent scenario,a contribution from scattered soft photons to the primary coronal emission could mimic a shallower N H in the quiescent spectrum.In the spectrum of the full NuSTAR exposure, we detect hard X-ray emission in the band ≈ −
80 keV in excess with respect tothe thermal emission and significant at a level of ≥ σ . We speculate that the hard X-ray emission could be due to a population ofenergetic electrons accelerated by the magnetic field along the accretion streams. These particles could concur to pumping up the Fefluorescence when hitting cold Fe of the circumstellar disk along with X-ray photons with E > .
11 keV.
Key words.
Stars: activity – Stars: flare – Stars: formation – Stars: coronae – Stars: pre-main sequence – Stars - individual: Elias 29
1. Introduction
X-ray observations of star-forming regions (SFRs) have estab-lished young stars as bright X-ray sources, from the Class Istage, when a thick envelope shrouds the central object, throughClass II, when a thick disk has been fully formed and is visi-ble, to the Class III stage, where very little, if any, circumstel-lar disk or envelope remains, the accretion process has ceased,proto-planets may have formed and the photosphere of the disk-less star is hardly distinguishable from that of more maturestars (Montmerle 1990; Feigelson & Montmerle 1999; Favata& Micela 2003)Extensive and deep surveys of SFRs in X-rays have beenobtained with
Chandra and
XMM-Newton (eg. COUP, XEST,DROXO, CCCP, Getman et al. 2005; Güdel et al. 2007; Pillitteriet al. 2010; Townsley et al. 2011). From these data we assessedthat a large fraction of the X-ray emission of Class I and II Young (cid:63)
Based on observations obtained with XMM-Newton, an ESA sci-ence mission with instruments and contributions directly funded byESA Member States and NASA
Stellar Objects (YSOs) is of coronal origin as clearly shown, forexample, by impulsive activity similar to the flares observed inthe solar corona. The magnetic structures that form the stellarcoronae of YSOs sometimes can create rotationally modulatedemission (Flaccomio et al. 2005). Another component of the X-ray emission likely arises from to the interaction of the centralstar and its circumstellar disk. This can be due to infalling mat-ter heated by the accretion process (e.g. Kastner et al. (2002)). Acoronal activity a ff ected by the accretion process has been pro-posed for explaining the soft X-ray excess observed in youngaccreting stars (Güdel & Telleschi 2007). Another phenomenonis the fluorescent emission, mostly in the neutral Fe of the diskat 6.4 keV and likely stimulated by coronal X-rays with ener-gies > .
11 keV (Imanishi et al. 2001; Tsujimoto et al. 2005)The YSO YLW16A in the ρ Ophiuchi Dark Cloud is where forthe first time during a large flare Imanishi et al. (2001) detectedprominent Fe K α line at 6.4 keV in its Chandra spectrum. Theyexplained the feature as the result of the excitation of neutral Fefrom hard X-ray photons produced during the flare. In the spec-tra obtained with a 850 ksec long continuous ACIS observation
Article number, page 1 of 15 a r X i v : . [ a s t r o - ph . S R ] J a n & A proofs: manuscript no. elias29_manuscript dubbed the
Chandra
Orion Ultradeep Pointing (COUP) Tsuji-moto et al. (2005) reported the detection of a Fe K α ρ Oph
XMM-Newton
Observation (DROXO) of the core F region revealed 61 ρ Ophiuchi YSO members (Pillitteri et al. 2010). In nine of these61 YSOs, specifically in 4 Class I, 4 Class II, 1 Class III objects,Stelzer et al. (2011) detected the Fe K α α / II YSO in the Rho Oph Dark Cloudwhere Giardino et al. (2007) (hereafter Paper I) detected signif-icant variability in the equivalent width (EW) of the Fe K α ∼
30 eV) and appeared at its maximum strength90 ks after Elias 29 underwent a flare with an EW ∼
800 eV. Thethermal X-ray emission was the same in the two time intervals,while variability of the 6.4 keV line was significant at a 99.9%confidence level.As for the excitation mechanisms, photo-ionization alonecould not be su ffi cient to explain strong fluorescent emissionwith EW in excess of ∼
150 eV and other mechanisms like col-lisional excitation are invoqued. Drake et al. (2008) analyzed theFe K α fluorescent line emission in a few stars concluding thatthere was not compelling evidence for a collisionally excited flu-orescence from high energy electrons. On the one hand a simpledisk illuminated geometry cannot produce EW in excess of 150eV and thus the origin of the strong emission observed in Elias29 is still not clear. More in general Drake et al. (2008) has con-sidered four di ff erent possible explanations for the case of FeK α with EW >
150 eV, namely: 1) high Fe abundance of the diskmaterial that could increase line intensity, but this rapidly satu-rates at EW ∼
800 eV (Ballantyne et al. 2002), 2) disk flaringthat, thanks to a favorable geometry, can increase line intensityby a factor two or three; 3) emission induced by an "unseen"flare obscured by the stellar disk implying that the evaluationof the exciting continuum is grossly underestimated; 4) excita-tion due to high energy non-thermal electrons that however re-quires a substantial amount of energy stored in the impingingparticles (Ballantyne & Fabian 2003). Since the presence of theFe K α fluorescent line with EW >
150 eV is a quite common fea-ture among YSOs, explanations based on ad-hoc geometry of thesystem or peculiar conditions of the systems seem still unsatis-factory. The extraordinary example of Fe fluorescence of V 1486Ori (Czesla & Schmitt 2007), where an EW of ∼ ∼ . / disk system. There large EW (0 . < EW < H > cm − (e.g. Fukazawa et al. 2011). The soft (0 . −
10 keV) andthe hard ( >
10 keV) X-rays spectrum of a YSO with disk show-ing Fe fluorescence can reveal the presence of a non-thermalpopulation of electrons responsible for at least part of the flu-orescence. In this context we obtained a joint and simultaneous
XMM-Newton and
NuSTAR observation of Elias 29 devoted to
Table 1.
Log of the observations . We will refer to the
XMM-Newton ob-servations as first, second and third
XMM-Newton observation respec-tively. Analogously we will refer to the first, second and third
NuSTAR observation for simplicity.
Satellite ObsID Start (UT) Net Exposure (ks)
XMM-Newton
XMM-Newton
XMM-Newton
NuSTAR
NuSTAR
NuSTAR
Note: for
NuSTAR the science time per orbit is about 55% of the orbitduration. The third
NuSTAR observation was obtained about 10 monthsafter the joint
XMM-Newton and
NuSTAR observations. acquiring spectra from soft (
XMM-Newton band 0.3-8.0 keV) tohard (
NuSTAR band 3-80 keV) X-rays. We conceived this pro-gram in order to detect any non-thermal hard X-ray emissionfrom Elias 29, study the time variability and relate these featuresto the fluorescent emission, and eventually explain its origin.We present the characteristics of the new X-ray observationsand the adopted analysis in Sect. 2, we illustrate the results ofthe time-resolved spectral analysis in Sect. 3, discuss the resultsin Sect. 4, finally we draw our conclusions in Sect. 5.
2. Observations and data analysis
Elias 29 ( α = h m . s , δ = − d m . s , other identifiers:[GY92] 214, 2MASS J16270943-2437187, ISO-Oph 108) is themost IR luminous Class I YSO in the Rho Ophiuchi Dark Cloud(Bontemps et al. 2001; Natta et al. 2006). Its accretion rate isabout 1 . × − M (cid:12) yr − , the circumstellar disk has an estimatedmass of about 0.012 M (cid:12) , its inner radius is ∼ .
36 AU which isabout 13 stellar radii ( R (cid:63) ∼ . − . R (cid:12) ), and the outer radiusof the disk is about 600 AU (Boogert et al. 2002; Miotello et al.2014; Rocha & Pilling 2015). The system is tilted in a way thatthe line of sight partly crosses the envelope, the star is visiblethrough the outflow cavity as well as a portion of the disk (cf.Fig. 14 in Rocha & Pilling 2015).The XMM-Newton and
NuSTAR observations were acquiredas part of a large, joint program (PI: S. Sciortino). The total ex-posure time was ∼
300 ks for
XMM-Newton and ∼
450 ks for
NuSTAR , but due to the low orbit of
NuSTAR about 250 ks of sci-ence exposure were obtained with
NuSTAR . The surveyed regioncovered most of the dense core F of LDN 1688, approximately 6 (cid:48) north of the previous pointing of the 500 ks long
XMM-Newton observation called DROXO (Pillitteri et al. 2010). Basic infor-mation on the
XMM-Newton and
NuSTAR observations are re-ported in Table 1.
XMM-Newton observations
The three
XMM-Newton observations have been carried out onthree subsequent satellite orbits (orbits 3238 to 3240), withnomimal aim point at Elias 29 and very little variation of theposition angle among the three orbits. We will refer to the
XMM-Newton observations as first, second and third
XMM-Newton ob-servation respectively. A log of the observations is reported inTable 1. The
XMM-Newton
EPIC ODF data were processed withSAS software (version 16.1.0) and the latest calibration files inorder to produce full field of view lists of events calibrated inboth energy and astrometry (Fig. 1). See http://xmm.vilspa.esa.es/sas
Article number, page 2 of 15. Pillitteri et al.: A deep X-ray view of the Class I YSO Elias 29 with
XMM-Newton and
NuSTAR . Fig. 1.
Left: color-coded image of the EPIC data integrated over the entire observation (red: 0 . − . . − . . − . NuSTAR observation. Right: color-coded image of the
NuSTAR data integrated over the entireobservation (red: 3 −
10 keV; green 10 −
40 keV; blue 40 −
60 keV). In both images a circle marks the position of Elias 29.
We have subsequently filtered these photons and retainedonly those with energy in the 0 . − ==
0; PATTERN <= XMM-Newton observa-tions. The chosen energy band limits ensure a good overlap withthe
NuSTAR band and the best EPIC calibration.High background variability was present during the first partof each
XMM-Newton exposure. This has the e ff ect to increasethe noise in light curves and spectra when subtracting the back-ground. However, for Elias 29 we preferred to use the full expo-sure time rather than excising the intervals with high backgroundin order to have a continuous monitoring. We used a circular re-gion of radius 30 (cid:48)(cid:48) centered on the source centroid to extract theevents for both MOS and PN. This region should contain about80% of the encircled energy of the XMM-Newton point spreadfunction (PSF).The background events have been extracted from a nearbycircular region of 40 (cid:48)(cid:48) radius without sources from the same chipand, for the PN, at the same distance from the read out node, asprescribed by the SAS guide. In order to produce the spectra weused a more strict selection (PATTERN <=
4) as recommendedin the SAS guide. With SAS we obtained light curves and spec-tra for both source and background events, response matrices(RMF) and e ff ective area files (ARF) for the spectral analysis.The spectra were grouped to have at least 25 counts per bin be-fore the analysis with the XSPEC software. NuSTAR observations
Two
NuSTAR observations were taken simultaneously with the
XMM-Newton observations, while a third exposure of duration ∼
195 ks was obtained in June 2018. This third exposure withoutan
XMM-Newton counterpart was not initially planned as part of the campaign. Due to the low satellite orbit, the total of the sci-ence exposures amounts to ∼
250 ks out of a total exposure of ∼
450 ks. The
NuSTAR data were processed with the heasoft suite (version 6.22.1), the
NuSTAR dedicated pipeline and thelatest calibration files (CALDB ver. 4.8) in order to produce fullfield of view lists of events calibrated in both energy and astrom-etry for the two cameras FPMA and FPMB. The resulting imagein the 3 −
78 keV band is shown in Fig. 1 (right panel) whereabout 10 sources are recognized by eye. The sources in the FOVare rather weak, with Elias 29 being the strongest one.We adopted the standard thresholds for the rejection of par-ticle background and the cut-o ff threshold at the SAA passage.For Elias 29 we extracted the spectra from a circular region cen-tered on the source centroid and of radius of 40 (cid:48)(cid:48) , this regionshould contain about 40% of the total source counts (Harrisonet al. 2013); a circular area of 80 (cid:48)(cid:48) radius was used to extract thebackground events from a nearby region. The tasks nuproducts and nupipeline were used to extract events in di ff erent energybands and time intervals, and to create spectra, light curves andrelated calibration files like response matrices and arf for thespectral analysis with xspec . Spectra from the events of
XMM-Newton
MOS1, MOS2 andPN and
NuSTAR
FPM A and B were accumulated in di ff erenttime intervals for time resolved spectroscopy (see Sect. 3.2).The spectra (energy band 0.3–8.0 keV) were modeled with anabsorbed thermal APEC component in order to derive the prop-erties of the emitting plasma, specifically the temperature, theemission measure (EM), the global abundances ( Z / Z (cid:12) ) and theflux. In addition, we used a Gaussian line with intrinsic widthequal to zero to model the fluorescent emission in the 6 . − . Article number, page 3 of 15 & A proofs: manuscript no. elias29_manuscript keV range . The EPIC spectral resolution is the main factor ofthe broadening of the Gaussian line width. In principle we couldexpect a variation of the lines contributing to the blend of fluo-rescent emission and their relative strengths, however we do notexpect velocity fields that can increase the line width to a de-tectable level. The global abundance was derived from the bestfit modeling to the spectrum of the third XMM-Newton obser-vation (ObsId 080031001) with a value Z / Z (cid:12) = .
54. This isconsistent with sub-solar Z / Z (cid:12) often derived from the analysisof low resolution spectra of young coronae (Maggio et al. 2000;Güdel 2003; Maggio et al. 2007). For the other time intervals weused a fixed Z / Z (cid:12) = . . − . xxv line and the neutralFe line at 6.4 keV. The plasma temperature is found in excessof 50 MK ( ∼ . xxv around 6.7 keV. The blendbetween the 6.7 keV Fe line and the fluorescent line at ∼ . ff ected by the low spectral resolutionof EPIC, the limited count statistics of the spectra, the strengthof the line, but also by the gas absorption, the temperature andFe abundance for the estimate of the continuum and the blendwith the 6.7 keV line. For this reason we also performed a bestfit modeling to the spectra in the energy range 5 − N H absorption is less constrained,however the continuum of the line is determined by the APECtemperature and its normalization, and the abundance value Z / Z (cid:12) constrains the intensity of the 6.7 keV Fe line. The best fit valueof the temperature found in the full band produce an accept-able fit in the narrow band, however only for the purpose of thebest evaluation of the continuum around 6.4 keV we let free tovary the temperature free and the global abundances. The resultsfrom such fits are listed in Table 4 and we will refer to these re-sults for discussing the fluorescence (Sect. 3.4). The N H valuewas kept fixed to the value estimated from the quiescent level( N H = . × cm − ). Approximately, half of the counts inthe full band spectrum are present in the 5-8 keV band. For thefirst and third XMM-Newton observations (ObsId 08030801 and08031001) we considered the full exposure in order to keep anadequate level of count statistics, while for the second observa-tion (ObsId 08030901) we analyzed the spectra in the same timeintervals used for the full energy band. The choice of the timeintervals is detailed in Sect. 3.2.
3. Results
Visual inspection of the
XMM-Newton images shows the pres-ence of more than 100 sources, while in the smaller
NuSTAR
FOV we can easily recognize about ten sources. We defer thestudy of the remaining X-ray sources to a separate paper, whilein this paper we focus on Elias 29 which is the strongest of the
NuSTAR sources (cf. Fig. 1).
XMM-Newton and
NuSTAR light curves of Elias 29
The PN and FPM light curves or Elias 29 are shown in Fig. 2.Two major flares occurred during the exposures, but only the In XSPEC terminology TB abs (APEC + G aussian ) first flare was observed simultaneously with XMM-Newton and
NuSTAR . The first flare had a duration of about 20 ks and a espo-nential decay time of about 7.6 ks, the second flare had durationwas about 50 ks with an exponential decay time of about 9.3ks. Before the second flare possibly the final decay of anotherflare was recorded with
NuSTAR . The lack of
XMM-Newton si-multaneous coverage limits the information we can obtain aboutthe second flare, however its detection allows to infer that ap-proximately every 200-250 ks a flare of intensity similar to thoseobserved in the present data can occur on Elias 29. A detailedanalysis of the first flare is given in Sect. 3.2.3. Other than thetwo main flares, we notice that in the first
XMM-Newton obser-vation (ObsID 080003081) the rate smoothly increased and thendecreased on a time scale of about 50 ks, it also showed a veryshort spike in the same time interval.In order to study the time variability and identify any statis-tical change in the count rate, we compared two distinct tech-niques: the
Bayesian change point (bcp) analysis (Wang &Emerson 2015; Erdman & Emerson 2007, 2008) and the PruneExact Linear Time (PELT) analysis (Killick et al. 2012). Thefirst method (bcp) uses a bayesian approach to determine thechange points in a time series. For each data point it derives aposterior mean of the rate and a probability of change of the rateat each data point. The second method, PELT, uses a competitivealgorithm that minimizes a cost function while guarding againstoverfitting the data by means of a penalty function. Fig. 3 showsthe posterior mean and the posterior probability of a change ateach light curve bin obtained from the bcp analysis. In the prob-ability panel we can decide which threshold to use to identifyvariability at some level of significance. For example, we canpick the values at P = . P = .
1, respectively. Above P = . P = . P = . NuSTAR light curves,however the lower count statistics of the data introduces morespurious peaks of posterior probability above 0.3 and that do notlook related to real variability.Figs. 4 and 5 show the results from PELT analysis: the inputbackground-subtracted PN and FPM light curves are shown withover-plotted the time segments and individual segment averagerate. We identified change points based on changes of mean rateand its variance ( cpt.meanvar function). We used an asymptotictype penalty and the default value of 0.05 (corresponding to a95% statistical significance level at each change point). We fur-ther checked the results of the number of intervals identifiedby using a manual value for the penalty function and produc-ing a plot of the number of change points as a function of thepenalty . Small values of the penalty produce more spuriouschange points, their number flattens out rapidly with increas-ing values of the penalty. The elbow corresponds to the numberof expected change points. Compared to bcp PELT seems lesssensible to small variations of the rate while bcp analysis seemsmore capable to find short duration change of rate of smaller am-plitude. On the other hand, the time segments found with PELTare more adequate for performing a robust time resolved spectralanalysis as they include more counts overall. We performed thesame analysis on the narrow energy band of 6 . − . . − . Implemented in the R package "bcp" Implemented in the R package "changepoint" See, e.g.,
Article number, page 4 of 15. Pillitteri et al.: A deep X-ray view of the Class I YSO Elias 29 with
XMM-Newton and
NuSTAR . pntime − t0 R a t e ( c t/ k s ) R a t e ( c t/ k s ) lll TotalbackgroundNet rate
XMM PN ( nu r a t e A + nu r a t e B ) * MJD − 57978.7 (days) R a t e ( c t/ k s ) lllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllll lllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllll lllllllllllllllllll lllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllll lllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllll lllllllllllllllllll llllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllll lll FPMAFPMBTotal
NuSTAR
Fig. 2.
XMM-Newton
PN (top panel) and
NuSTAR (bottom panel) light curves of Elias 29. High background variability a ff ects the first part ofeach XMM-Newton exposure and during the main flare, but its e ff ect can be adequately corrected as it is shown by the background-subtracted(red) light-curve. The origin of time axes is set to the time of start of the first PN exposure. The time gap between the second and third NuSTAR exposures is marked with the gray area and amounts to about 300 days. peak segment and a flare decay segment in agreement with thefull energy band analysis. For the time resolved spectroscopy weused the time segments identified with PELT on the light curveof the full energy band of
XMM-Newton (0.3-8.0 keV). The to-tal net counts in the di ff erent intervals vary between ∼
600 and ∼ NuSTAR exposure is divided intofive segments by PELT. We can recognize an initial partial decayphase, likely from an unseen flare, then a quiescent segment be-fore the flare, a peak and a decay segment for the flare and thepost-flare quiescent phase.
XMM-Newton time resolved spectroscopy
The third
XMM-Newton observation shows a low PN rate forabout 100 ks, and it appears as a continued quiescent phase afterthe flare registered in the second
XMM-Newton exposure. De-spite PELT identifies two time intervals with di ff erent rate vari-ance during the third exposure, we considered the full exposure as a whole for producing the MOS and PN spectra. The PN spec-trum is shown in Fig. 6 with the best fit model composed by anAPEC plus Gaussian line. The best fit parameters of the modelare shown in Table 2. The average plasma temperature is ∼ . − . N H is 5 . × cm − (90% confidence range3 . − . × cm − ). These values are similar to those found byGiardino et al. (2007) and by Favata et al. (2005), and thus weconclude that the X-ray coronal emission of Elias 29 has beenstably hot over a time scale of ∼
12 years. The quiescent unab-sorbed flux in 0 . − . . × − erg s − cm − which corresponds to L X ∼ . × erg s − at 120 pc.Furthermore, we accumulated a PN spectrum of the exposureencompassing the quiescent period after the flare in the second XMM-Newton exposure (segment 5) and the whole third expo-sure. This is justified by the similar count rate in both segmentsthat suggests similar spectral characteristics of the plasma. Theresulting PN spectrum had about 3400 counts, the best fit with anabsorbed APEC component had values (90% confidence rangein braces): N H = . . − . × cm − , kT = . . − . Z / Z (cid:12) = . . − . EM = . . − .
79) cm − Article number, page 5 of 15 & A proofs: manuscript no. elias29_manuscript lllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllll . . . . P o s t e r i o r M ea n P o s t e r i o r P r ob a b ilit y . . . . llll ll ll l lllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllll . . . . P o s t e r i o r M ea n P o s t e r i o r P r ob a b ilit y . . . . l llllllllllll l ll l lllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllll . . . . P o s t e r i o r M ea n P o s t e r i o r P r ob a b ilit y . . . . l l l lll Fig. 3.
BCP analysis of the
XMM-Newton
PN light curves in the 0.3-8.0 keV bandpass. Top panel shows the light curve (gray dots) with theposterior mean (solid line). The scale of the Y-axis is logarithmic and with the same range of values across the 3 panels for ease of comparison.Bottom panel shows the posterior probability at each point. We indicated the probabilities P > . P = . P = .
1, respectively.
ObsID 0800030801 lllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllll . . . . ObsID 0800030901 lllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllll . . . . ObsID 0800031001 lllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllll . . . . R a t e ( c t/ s ) Time (ks)
Fig. 4.
PELT analysis of the
XMM-Newton
PN light curves in the 0.3-8.0 keV bandpass subtracted of background rate. Bin size is 600 s andthe rate is displayed in a log scale. Panels have the same range on they-axis. Horizontal segments and shaded areas mark the average countrate and variance in each time interval (numbers on top of the intervals).The semi-log scale evidences the regular exponential decay of the flare. and unabsorbed flux in 0 . − F APEC = − .
21 (-12.23 – -12.19) erg s − cm − (see Table 2). Fluorescence is firmlydetected in the quiescent phase, as the modeling of the quiescentspectrum with a thermal APEC component alone shows a sharpexcess of emission around 6.4–6.5 keV (Fig. 6). The Gaussianline gives a best fit centroid of 6.49 (6 . − .
60) keV, EW = . − .
38) keV and flux of the line of log F Gau = − .
17 (-14.27 – -13.93) erg s − cm − . In the first
XMM-Newton observation (ObsId 0800030801) thePN rate of Elias 29 showed a slow increase of the rate followedby a similar smooth decrease. A spike of duration ≤ N H gas absorptionvary in the range 5 . − . × cm − within the five segments,however these values are still consistent with each other and withthe N H derived from the quiescent spectrum at a 90% confi-dence level. The best fit temperature, kT , varies in the 3 . − . XMM-Newton observation at a confidence level of 90%. Also,the segment that contains the short spike has a somewhat hightemperature (5.4 keV, 90% range ∼ . − . XMM-Newton observation compared to the averagespectrum of the last exposure which represents the quiescentemission. While the gas absorption N H and plasma temperaturekT were found similar in the two spectra ( N H ∼ . − . × cm − , kT ∼ − . Article number, page 6 of 15. Pillitteri et al.: A deep X-ray view of the Class I YSO Elias 29 with
XMM-Newton and
NuSTAR . . . . . Time (ks) R a t e ( c t/ s ) lllllllllll llllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllll lllllllllllllllll . . . . Time (ks) R a t e ( c t/ s ) llllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllll . . . . Time (ks) R a t e ( c t/ s ) lllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllll Fig. 5.
As in Fig. 4 for
NuSTAR light curves in the 3 −
80 keV. −3 −3 −3 no r m a li z ed c oun t s s − k e V − data and folded model r a t i o Energy (keV)
Fig. 6.
PN spectrum of the quiescent phase (top panel) and ratiodata / model. The spectrum was accumulated from events collected dur-ing the quiescent phases after the flare and in the third XMM-Newton observation. The model is an absorbed APEC thermal component. Asharp excess of emission is visible at ∼ . the region could show flaring activity like the short spike weobserved, as a result of the complex dynamics of the magneticstructures in it. The passage of such a region lasted about 80 ksor ∼ .
93 days, consistent with a rotation period of about twodays, typical of a very young Class I YSO like Elias 29. The factthat it did not appear again in the second
XMM-Newton exposuresets a lifetime to the order of one day. However, the region itselfcould have hosted the main flare observed in the second exposurewhen the region did appear again in view. The flare could havealso destroyed or heavily reshaped the active region as this latterdid not appear again in the third
XMM-Newton observation.
Two main flares were observed in Elias 29 but we discuss indepth the first flare because of the simultaneous coverage with
XMM-Newton and
NuSTAR coverage. This flare showed a quite −3 −4 −3 −3 no r m a li z ed c oun t s s − k e V − data and folded model ( da t a − m ode l ) / e rr o r Energy (keV)
Fig. 7.
PN spectra and best fit models for the first
XMM-Newton ex-posure (red symbols) and the third one (black symbols). Lower panelshows the residuals (data - model values). The model for both spec-tra is an absorbed thermal APEC component plus a Gaussian to takeinto account the fluorescent emission from partially ionized Fe lines at6 . − . ff erence between the two spectra is due to a di ff er-ence of EM. regular decay phase well modeled with an exponential decay.The decay of the flare appears faster in the hard band (5–8 keV)than in the soft band (0.3–5.0 keV), with e-folding times τ ∼ . τ ∼ . NuSTAR data with a decay timeequal to ∼ . ± . ∼ . ± . ≈ L X = . ρ Ophiuchi itself (Pillitteri et al. 2010). A flarewith similar duration and peak rate was observed by Giardinoet al. (2007). Taking into account the past X-ray observations, ina total exposure time of ∼
800 ks Elias 29 has shown flares with
Article number, page 7 of 15 & A proofs: manuscript no. elias29_manuscript duration of less than one day and peak rate about 10 times thequiescent rate. The second flare, observed only with
NuSTAR ,had total duration of about 50 ks (almost 14 hours), quite shorterthan the day long lasting flares seen in the Orion Nebula and in ρ Ophiuchi.The modest brightness of the flare implies a modest countstatistics. This fact reduces the detail and accuracy of the timeresolved spectroscopy we can perform on it. The PELT algo-rithm divided the flare roughly in a peak segment (number 2),two decay segments (marked 3 and 4) and two quiescent seg-ments (1 and 5) before and after the flare, respectively. Table2 lists the best fit parameters of the flare segments. In orderto improve the statistics and better constrain the model param-eters, we made a simultaneous fit of the spectra of segments1 and 5 as they are representative of the quiescent phases be-fore and after the flare. In these time intervals we measured N H ∼ × cm − , kT ∼ . ∼ . Flux ∼ − . XMM-Newton exposure.During the flare there is an increase of both the temperatureand the gas absorption. The temperature rises to about 11.1 keV( ∼
130 MK) and the absorption reaches values of N H ∼ . × cm − which is about a factor of four higher than the N H of thequiescent phase. The di ff erence of N H between quiescence andflare states is significant at a level > σ . A similar increase of N H was noticed by Giardino et al. (2007) in the flare observedin DROXO and by Kamata et al. (1997) in a flare observed withASCA. Such a behavior suggests that the X-rays from the flaringregion cross material optically thicker than the gas crossed byX-rays coming from the quiescent corona. Kamata et al. (1997)attributed the increase of N H to the disk and envelope geometrysurrounding Elias 29. They proposed that the flaring sites arepreferentially at a low latitudes and their lines of sight crosses thedisk. This explanation however remains at odds with the face-on geometry of the disk inferred from far IR observations (cf.Boogert et al. 2002).We remark that the flare temperature peaks at segment 2, butthe EM is detected at its maximum during segment 3. The timedelay between the temperature and the EM peaks is predicted bymodels of flaring loops (e.g., Reale 2007): the flare heat pulsedrives a strong plasma flow from the chromosphere upwardsalong the magnetic tube, and the flow continues to fill the tubefor some time after that the heat pulse has stopped (and the cool-ing starts). It is then reasonable to work in the assumption thatthe flare occurs in a single flaring loop, and to use the relateddiagnostics to determine the characteristics of the flaring loopbased on hydrodynamic simulations and calibrations on X-raysolar flare observations (Reale 2007). In particular, we may inferthe semi-length of the loop L from the decay time of the flare,the peak temperature and the slope of the decay in the density-temperature diagram, by using equations A.1, A.2 and A.3 inReale (2007). We derive a maximum temperature at the peak of ∼
325 MK from the kT at segment 2 (11.1 keV ∼
130 MK) andlog EM[cm − ] ∼ .
24. Because of the large uncertainties in thetemperature we cannot derive a reliable value of the slope in thedensity-temperature diagram. Thus, we assumed the maximumvalue of the slope determined by Reale (2007). The maximumslope corresponds to the case of absent sustained heating duringthe flare decay, i.e., consistent with the pure cooling of a singleflaring loop. A shallower curve in the density- temperature dia-gram would instead suggest the progressive involvement of moreand shorter loops, like in arcade flares. Our assumption impliesthat we are deriving an upper limit for the length L of the flar- ing loop(s). We measured an e -folding decay time τ ∼ .
65 ksfrom the light curve in the soft band 0.3–5 keV (Fig. 8) and weestimated L ≤ . × cm or L ≤ < . R (cid:12) (or < . L ≈ ψ T / ∆ t R (1)where L is in cm , T =
332 MK is the loop maximum tempera-ture at the flare peak, and ∆ t R , in ks , is the time range betweenthe flare start and its peak (we assume that the peak of the lightcurve is a good proxy of the emission measure peak). From thelight curve we measure ∆ t R = ± . ψ is theratio between the maximum temperature and the temperature atthe density maximum. This is unconstrained in our case, and thewhole possible range 1 . < ψ < . × < L < . × cm, i.e, 0 . R (cid:12) < L < . R (cid:12) (still L < . × erg released in about20 ks. Still in the framework of a single flaring structure, andassuming a representative semi-length L ≈ cm ( ∼ . R (cid:12) ), atypical cross section radius of R L = . L (see, e.g., Golub et al.1980 and Klimchuk et al. 2000) and considering the values ofEM derived from the spectral analysis, we can push our analy-sis to infer a value of the electron density in the loop during theflare from EM ∼ n e V , where V = π R L L is the total volumeof the loop. Although this is to be taken with care, for a peakvalue of EM ∼ × cm − and V ∼ × cm we obtain n e ∼ × cm − , which is similar to the typical values foundfor solar flaring loops (e.g., Reale 2014).From the density and temperature, we can in turn infer a min-imum strength of the magnetic field ( B ) capable to confine theplasma inside the loop (e.g., Maggio et al. 2000), and this is oforder of B ≥ B = (16 π n e k B T ) / ∼
500 G, similar to otherflares of active stars and compatible with average fields of kG onthe stellar surface, as found in other YSOs.For the second flare, observed only with
NuSTAR , we hadeven more limited information. From the analysis of the FPM Aand B spectra from the flare interval as a whole and from the riseplus peak segments we derived a plasma temperature of about 5keV (90% confidence level: 4.3–6.5 keV) and a N H ∼ . × (0.8–1.7 × ) cm − . The estimate of N H is less precise than theone inferred from XMM-Newton spectra as the band below 3.0keV is not observed by
NuSTAR . However, in agreement withthe first flare and its previous flares, this second flare of Elias 29showed once again a N H value higher than the one measuredduring the quiescence. We will discuss this finding in Sect. 4speculating about the location of the flaring regions in Elias 29. NuSTAR spectroscopy
In Fig. 9 we show the time averaged
NuSTAR spectra of theFPMA and FPMB instruments in three di ff erent time intervals.The spectra refer to the total exposure, the sum of the quies-cent intervals and the sum of the flaring intervals recorded in NuSTAR data. We obtained about 4200 net counts in the to-tal spectra, about 2400 net counts in the quiescent spectra andabout 2000 net counts in the flare spectra. We plot also a best
Article number, page 8 of 15. Pillitteri et al.: A deep X-ray view of the Class I YSO Elias 29 with
XMM-Newton and
NuSTAR . lllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllll . . . . . . . Time (s) R a t e ( c t/ s ) lllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllll Fig. 8.
XMM-Newton
PN light curves in the 0 . − . − fit model where, in this case, we added a power law to theAPEC + Gaussian model for modeling the spectra above ∼ NuSTAR
FPM instruments, a free centroid of the Gaussianline does not improve the model results and thus we kept fixedthe line centroid at 6.4 keV. A good fit with a model composedby a thermal APEC component and a Gaussian line is satisfac-tory up to ∼
20 keV. A joint fit of
NuSTAR and PN spectra founda temperature similar to that found with the best fit to the PNspectrum alone. At energies above 20 keV a systematic resid-ual emission is observed in excess of the thermal emission. Thespectrum has a low statistics in this spectral range, neverthelessthe excess is significant above 2 σ and represents to date the bestexample of hard X-ray spectrum of a YSOs. Adding a secondAPEC or a Bremsstrahlung components does not improve thefit above 30 keV as evaluated with the χ statistics. Adding apower law component improves the fit and gives a spectral index γ ∼ .. − ∼ −
80 keV. However, thelow count statistics above 50 keV after background subtractionmakes the best fit procedure and the χ test not applicable in the50 −
80 keV range. We speculate that the emission in 20 −
50 keVand modeled with a power law could be of non-thermal naturefrom a population of high energy particles that can contributeto pumping up the fluorescent emission as discussed by Emslieet al. (1986) for energetic solar flares.The excess of hard X-ray emission with respect to the ther-mal emission is detected not only during the flare but also duringthe quiescent phase. This means that the non-thermal componentis weak but present also during the quiescent phase rather thanbeing emitted exclusively during the flare. The flux in 10 − . × − − . × − erg s − cm − between thequiescent and flaring phases. Table 4 reports the values of the centroids, the equivalent widthsand the fluxes of the Gaussian line for the fluorescent emission at ∼ . ∼ . − α and K β lines from neutral to multiply ionizedFe (see Kallman et al. 2004). In a few cases the centroid is at 6.5keV with a 90% confidence range of ∼ . − . ff erent levelsof count statistics. The spectra are generated from a model com-posed by a thermal APEC component at 4 keV absorbed by a gascolumn density of N H = . × cm − , with a global abun-dance Z ∼ . Z (cid:12) and a Gaussian line at 6.4 keV with equivalentwidth in the set: 0, 0.15, 0.3, 0.6 and 0.8 keV. The abundance waskept fixed in one set of 1000 simulations and variable in anotherset of 1000 simulations. Fig. 10 shows the cumulative distribu-tions of the line centroid as a function of the count statistics ofthe input spectra and di ff erent line intensities. From these sim-ulations we infer that in the case of spectra with more than 500counts and with EW ≥ . P < .
05) to determine the centroid at energies well above 6.5keV. In the data, the most significant case where we measuredthe centroid at 6.51 keV before the flare (second
XMM-Newton exposure, segment 1), when the number of counts in the spectraare 476, 76 and 50 for PN, MOS1 and MOS2, respectively, andthe EW is about 0.47 keV (0.25–0.8 keV 90% confidence range).Here we possibly caught fluorescent emission from partially ion-ized Fe just at the beginning of the flare. Fig. 11 shows the PNspectra during the segment 1 of the second
XMM-Newton expo-sure and the third exposure. The best fit model is shown and thecentroid in one case is found at ∼ .
53 keV and in another case isfound at ∼ .
42 keV. These values are marginally compatible atthe 90% significance level as determined from the uncertaintiescalculated with XSPEC. Our simulations suggest that these twovalues of the centroid are di ff erent at a 95% significance levelgiven the counts of the spectra are ≥ XMM-Newton ob-servations in absence of evident flaring activity. The EW is foundto be between ∼ ∼ XMM-Newton spectrawith best fit parameters in Table 2 and in the 7.11-80.0 keV bandfor the
NuSTAR spectra with best fit parameters in Table 3. Asystematic excess of the
NuSTAR fluxes in the 7.11-80.0 keVband is present with respect to the 7.11-10.0 keV
XMM-Newton fluxes due to the larger bandwidth of
NuSTAR fluxes. A linear fitbetween the two fluxes gives a slope of 0 . ± .
13 and interceptof − ± .
6. Fitting a relationship of the type y = Ax gives aslope of 1 . ± .
005 The three
NuSTAR points gives a slopeof 0 . ± .
02 and 1 . ± .
005 when fitting without intercept.The correlation between Gaussian flux and flux above 7.11 keVcan be understood by the fact that the photons at energies above7.11 keV can induce Fe fluorescence. However, it is presumablethat very hard X-ray photons will not be absorbed by materialwith N H < cm − , thus the photons that concur to excitefluorescence have energies well below 80 keV. Article number, page 9 of 15 & A proofs: manuscript no. elias29_manuscript −10 −9 −8 −7 −6 −5 −4 −3 no r m a li z ed c oun t s s − k e V − data and folded model
102 5 20 50−202 ( da t a − m ode l ) / e rr o r Energy (keV) 10 −10 −9 −8 −7 −6 −5 −4 −3 no r m a li z ed c oun t s s − k e V − data and folded model
10 1002 5 20 50−202 ( da t a − m ode l ) / e rr o r Energy (keV) 10 −9 −8 −7 −6 −5 −4 −3 no r m a li z ed c oun t s s − k e V − data and folded model
10 1002 5 20 50−202 ( da t a − m ode l ) / e rr o r Energy (keV)
Fig. 9.
NuSTAR
FPM A and B spectra in three di ff erent time intervals, black is FPMA, red is FPMB. The best fit models (dashed lines) and the χ terms in units of σ (bottom panels) are also shown. Left panel: average NuSTAR spectra accumulated on the total exposure ( ≈
260 ks), Centralpanel: spectra during the quiescent phase. Right panel: spectra during the two flares.
Fig. 10.
Cumulative distributions of the best fit centroid positions of the Gaussian line from the simulations at di ff erent levels of count statistics(values in the plots) and for three values of the equivalent width of the Gaussian line used in the starting model (indicated in the title of the plots).
4. Discussion
Fluorescent emission is a feature of Elias 29, both in the qui-escent and flaring states. An EW in excess of 0.15 keV is de-tected in almost all the the time intervals. This result makes un-realistic a simple model made of an irradiated disk, and hintsthat other mechanisms of reverberation and / or a more complexgeometry that takes into account the cavity where Elias 29 sitscan have a role for explaining such high EWs. Fe fluorescencein YSOs of ONC has been investigated by Czesla & Schmitt(2010), they remarked how explaining the origin of the fluores-cent line at 6.4 keV in a few case of quiescent sources is still anopen issue. Their sample of COUP sources span a range of N H in 2 × − × cm − and EWs between ∼ . ≥ . NuSTAR spectra we detected an excess of hard X-rayemission in Elias 29 likely of non thermal origin. The countingstatistics do not allow to perform in depth a time resolved analy-sis, yet there is no evidence of an increase or a concentration ofsuch a hard emission during flares only as the hard X-ray emis-sion seems produced ubiquitously during the entire observation.We speculate that the excess of hard X-ray emission is associ-ated to a population of accelerated particles moving along theaccretion streams and varying with stochastic frequency in timedue to a highly structured magnetic field. The average strengthof the magnetic field is expected to be of order of a few hundreds of G in order to constrain plasma at an average temperature of afew keV, while, in comparison, the average coronal field in theSun is of order of 2 G. Locally the magnetic field of YSOs likeElias 29 can reach up to few kG of strength in the core of activeregions and during flares. Still, it is possible to use the Sun as atemplate for the corona to build up magnetic fields with valuesin excess of a kG (see Orlando et al. 2003). In such a scenariopart of the flux of the Fe K α > .
15 keV seems di ffi cultto obtain with a simple irradiated disc model. In the Sun, Parmaret al. (1984) found that the fluorescence observed during solarflares is produced almost exclusively by photons at E > . Article number, page 10 of 15. Pillitteri et al.: A deep X-ray view of the Class I YSO Elias 29 with
XMM-Newton and
NuSTAR . Fig. 11.
Top panel: PN spectrum and best fit model (solid line) duringsegment 1 of the second
XMM-Newton exposure. Dotted lines show theAPEC component, dashed lines show the Gaussian component, respec-tively . Bottom panel: the same plot for the PN spectrum during the third
XMM-Newton exposure. Vertical dashed lines in both panels mark thecentroid positions and the 90% confidence range. ll lll ll lll
Fig. 12.
Flux of the Gaussian spectral component vs. the flux above7.11 keV. The di ff erent symbols refer to the first XMM-Newton obser-vation (circles), the second one during the flare (triangles), and the last
XMM-Newton observation (cross). The filled symbols refer to the fluxesderived from
NuSTAR best fit models (see Table 3). The lines are thelinear best fit to the
XMM-Newton and
NuSTAR data respectively. l l l l l l l
Fig. 13.
Flux of the APEC thermal component (large symbols) and ofthe Gaussian line (small dots) as a function of the time. Error bars referto 90% significance level of uncertainty. rescence in the flares they have analyzed. High energy electronsare e ffi cient at stimulating fluorescence when their energies are <
25 keV, whereas the e ffi ciency of hard X-ray photons to stim-ulated fluorescence has a cut o ff at around 20 keV. With the datain hand we cannot detect any delay between the increase of thefluorescence with respect to the overall coronal flux during theflare. In principle one can expect a delay if the emitting region isthe inner disk and the excitation of the fluorescence takes sometime to reach its maximum and to fade out after the flare.We find suggestion that the centroid of the fluorescent linecould vary in time. From our simulations we estimated a signifi-cance of such variation at a 95% confidence level. The change ofline centroid can be explained in a scenario where the emissionarises from excited material at di ff erent ionization stages whoserelative contributions (and associated energy of excited emissionlines) to the overall emission in the 6 . − . XMM-Newton and high spectral resolution (2 . − NuSTAR band, its weakhard X-ray flux remains still too faint to allow for a more detailedtime-resolved analysis.Isola et al. (2007) found a significant correlation between thesoft X-ray emission in the GOES band 1 . − . −
40 keV and 60 −
80 keV (mostly of non-thermal na-ture) during solar flares. This correlation holds up to the most en-ergetic events, spanning about four orders of magnitude in peakflux. They showed that the same scaling law holds for the hand-ful of available hard X-ray observations of intense stellar flaresobserved with Beppo
SAX in active stellar binaries or zero-agemain sequence stars. If the X-ray emission in Class I / II YSOsis just a scaling up of solar phenomena, we expect such a cor-
Article number, page 11 of 15 & A proofs: manuscript no. elias29_manuscript relation to be valid for very young pre-main sequence stars. InElias 29 the flux at the peak of the flare from the PN spectrumof segment 3 in the 1 . − . F S = . × − W / m ; from the NuSTAR spectrum of the whole flare the fluxin 20 −
40 keV is 1 . × − erg s − cm − corresponding to F H = . × − photons cm − s − keV − . For a direct com-parison to the Isola et al. (2007) results we rescaled these twoquantities to a distance of 1 AU obtaining a flux of F S ≈ .
97W m − (1.6-12.4 keV band), and F H ∼ × photons cm − s − keV − (20-40 keV band). The relationship of Isola et al. pre-dicts F S ∼ .
22 W m − , which is within a 10% of uncertaintyfrom our measurement of F S . It is evinced that in a Class I ob-ject like Elias 29 the soft and hard fluxes at the flares show thesame scaling law empirically found for the Sun and active stars.The flare of Elias 29 can be considered a scaled-up version ofpowerful solar flares and similar to those of active stars on themain sequence.A further test for the analogy between the flare in Elias 29and the solar flares was based on the thermal flux estimated byIsola et al. (2007). We used the flux in the 20 −
40 keV bandmeasured from
NuSTAR spectra and the coe ffi cient of scaling m given in Table 3 of Isola et al. (2007) and corresponding to thetemperature of 6 keV (the closest to the one observed in Elias 29)to estimate the thermal flux in the 1 . − . F S ∼ . × − ∗ m = . × − to be comparedto 5 . × − photons cm − s − keV − which is within 30% un-certainty from the value estimated from Isola et al. relationship.This again corroborates the analogy between the flare of a veryyoung corona of a Class I object (Elias 29) and the flares of moreevolved active stars.The Coordinated Synoptic Observations (CSI) observed theYSOs in the ∼ C OROT,
Chandra and S pitzer. Light curvesof tens of Class I, II and III YSOs have been obtained duringthe ∼ E flare , can easily reach values up to ∼ erg, i.e. up to 1-2 dex above the energy released in thestrongest (X100 class) flares ever observed on the Sun. The flarein Elias 29 released an energy of about 8 × erg and this is inthe median of the energies of the detected flares in NGC 2264.At the same level of flare energy from studies on other SFRs likeONC (Caramazza et al. 2007) and Cyg OB2 (Albacete Colomboet al. 2007), a frequency of flares of about 1 − − is expected.When considering also the DROXO exposure, Elias 29 showedtwo similar flares detected in a global XMM-Newton exposure of800 ks which is in good agreement with the flaring rate derivedin other SFR regions.The energy released during the flare is high even when com-pared to the energies released by the most powerful solar flares,but it is of the same order of magnitude of the energies of flaresof similar YSOs. On the other hand, the peak temperature is quitehigh and on top of an already hot temperature during quiescence(185 MK vs. ∼
50 MK). A flare with similar duration was alsoobserved in DROXO (Giardino et al. 2007). A compact size ofthe hosting loop(s) is suggested.A peculiar feature of the flares in Elias 29 is the increase ofgas absorption of about five times during the flares as describedin Sect. 3.2.3. The increase of gas absorption was also observedin past flares observed with
XMM-Newton and ASCA (Giardinoet al. 2007; Kamata et al. 1997). Kamata et al. (1997) speculatedthat, under the hypothesis that the disk is edge on, the flares oc-cur at low latitudes and the X-rays pass through the disk being thus heavily absorbed, while the X-rays emitted during quiescentphases are coming from the rest of the corona and encounterless gas along the line of sight. However, while plausible, thisinterpretation conflicts with the geometry of the source extrap-olated from sub millimeter and FIR observations (e.g. Boogertet al. 2002; Ceccarelli et al. 2002; Miotello et al. 2014; Rocha& Pilling 2015, see also Fig. 14). An alternative explanation thatcan reconcile the system geometry and the higher gas absorp-tion during flares is that the flares are generated near the feetof the accretion streams (see Fig. 14. The feet could be locatedpreferentially at high stellar latitudes around the stellar poles asthe streams follow the large scale dipolar geometry of the mag-netic field and likely these are the sites of frequent flares. TheX-rays generated during the flares at the stream feet travel a por-tion of path across the dense accreting gas before arriving to theobserver. As a result the gas absorption measured during flaresis found systematically larger than the one measured from thequiescent corona.Finally, a di ff erent scenario can in principle explain both the N H enhancement associated to flares and the large EW observedfor the 6.4 keV line. Elias 29 displays a hard X-ray emission( E > keV ), possibly of non-thermal origin, in addition to itsthermal X-ray spectrum ( E < keV ) of the corona. At theseenergies the number of X-ray photons that undergo to Comptonscattering, instead of photo-absorption, could be non negligible.Compton scattering diminishes the energy of scattered photonsand cause a global softening of the X-ray spectrum. Depend-ing on the system geometry, Compton scattering could thereforemimic a N H lower than that experienced by the primary X-rayphotons emitted by the corona on our line of sight. In fact, theline of sight toward Elias 29 passes approximately on the edgeof the inner cavity. Therefore X-ray photons scattered toward usby the disk surface and the inner cavity wall will experience an N H lower than that su ff ered by primary photons emitted by thecentral star. Hence, the total X-ray spectrum reaching us wouldbe approximately composed of highly-absorbed primary X-rayphotons, and less-absorbed scattered photons. In this scenario,an increase of the thermal emission of the corona (i.e. a flare)implies an increase of the highly-absorbed primary-photon com-ponent only. That would therefore explain why the X-ray flaresobserved on Elias 29 show N H systematically higher than thatof the quiescent phases. In addition, assuming that the real N H between us and the central star is that observed during flares (i.e. ∼ × cm − ), the EW of the fluorescent line at 6.4 keV willbe larger than the model predictions simply because, at these N H values, photons at ∼ . ∼ N H (e.g. Fukazawa et al. 2011). Thecase of Elias 29, i.e. EW ∼ . N H ∼ cm − , wouldneatly fit the EW vs N H pattern observed for the AGNs.
5. Summary
We presented the results of a joint
XMM-Newton and
NuSTAR simultaneous observation of Elias 29, the IR brightest Class Iobject in the Rho Ophiuchi Dark Cloud core F (LDN 1688).The full EPIC image contains about 100 X-ray sources while
NuSTAR shows about ten sources among which Elias 29 is thebrightest. We observed a flare of duration of about 20 ks witha regular exponential-like decay characterized by an e-foldingtime of about 7.6 ks in the 0 . − . Article number, page 12 of 15. Pillitteri et al.: A deep X-ray view of the Class I YSO Elias 29 with
XMM-Newton and
NuSTAR . Fig. 14.
Cartoon of the proposed scenario to explain the increase ofN H during flares. The flares originate at the base of the accretion streamlocated around the pole and, in a face-on geometry, the emitted X-rayswould cross a portion of the accretion stream with N H absorption largerthan the average N H value. of about 3 ks. Through time resolved spectroscopy we inferredthe properties of the quiescent and the flaring plasma. We de-termined that the flaring structures are relatively compact with alength of about 1 − (cid:12) , which suggest that likely the loop isanchored to the stellar surface. A magnetic field of at least 500 Gis required to confine the plasma within the loop. A second flarewith a duration of 50 ks was observed with NuSTAR only andfor which we inferred a loop size similar to the one that hostedthe first flare. During these flares we observed an increase of N H that suggests a specific location of the flaring sites at the base ofthe accretion streams of the star.Fluorescent emission from neutral or partially ionized Fe isobserved both during quiescence and during the flares. Fluo-rescence is modeled with a Gaussian line varying both in thecentroid position and in strength. There is a clear correlationbetween fluorescent emission and coronal emission during theflare. However, there is still significant Fe fluorescence outsidethe flare that cannot be explained exclusively with the contribu-tion of photons at E > .
11 keV. We detect a hard X-ray emis-sion from Elias 29 above ∼
20 keV which is not explained bya thermal emission. We argue that a non-thermal population ofelectrons accelerated from the coronal magnetic field could beresponsible for this emission. We speculate that the same popu-lation could contribute to the fluorescent emission of Elias 29.
Acknowledgements.
We thank the anonymous referee for her / his comments andsuggestions which improved the manuscript. The authors acknowledge mod-est financial contribution from the agreement ASI-INAF n.2017-14.H.O. IP ac-knowledges support from the ASI and the Ariel Consortium. We made use of R , a language and environment for statistical computing, the XMM-SAS suite,HEASOFT fv, XSPEC and DS9. Article number, page 13 of 15 & A proofs: manuscript no. elias29_manuscript T a b l e . P a r a m e t e r s fr o mm od e l b e s t fi ti n t h e s p ec t r a fr o m t h e ti m e i n t e r v a l s i d e n ti fi e d w it h t h e P ELT a l go r it h m ( F i g . ) . O b s S e g m e n t C oun t s N H k T l og E M l og F l ux l og L X C e n t r o i d l og F G au E W l og F E > . ke V χ r e d P r ob D . o . F . c m − k e V c m − e r g s − c m − e r g s − k e V e r g s − c m − k e V e r g s − c m − F i r s t + + . ( . ) ( . . ) . ( . . )- . (- - . ) . . ( . . )- . (- . - . ) . - . . . + + . ( . . ) . ( . ) . ( . . )- . (- . - . ) . . ( . . )- . (- . - . ) . - . . . + + . ( . . ) . ( . . ) . ( . . )- . (- . - . ) . . ( . . )- . (- . - . ) . - . . . + + . ( . . ) . ( . ) . ( . . )- . (- . - . ) . . ( . . )- . (- . - . ) . - . . . + + . ( . . ) . ( . . ) . ( . )- . (- . - . ) . . ( . . )- . (- . - . ) . - . . e - S ec ond11800 + + ( . . ) . ( . . ) . ( . . )- . (- . - . ) . . ( . . )- . (- . - . ) . - . . . + + . ( . . ) . ( . ) ( . . )- . (- . - . ) . . ( . . )- . (- . - . ) . - . . . + + . ( . . ) . ( . . ) . ( . . )- . (- . - . ) . . ( . . )- . (- . - . ) . - . . e - + + . ( . . ) . ( . . ) . ( . . )- . (- . - . ) . . ( . . )- . (- . - . ) . - . . . a nd52919 + + ( . . ) . ( . . ) . ( . . )- . (- . - . ) . . ( . . )- . (- . - . ) . - . . . + + . ( . . ) . ( . . ) . ( . . )- . (- . - . ) . . ( . . )- . (- . - . ) . - . . . T h i r d12040 + + . ( . . ) . ( . . ) . ( . . )- . (- . - . ) . . ( . . )- . (- . - . ) . - . . . Q u i e s c . ( . . ) . ( . . ) . ( . . )- . (- . - . ) . . ( . . )- . (- . - . ) . ( . . )- . . . N o t e :t h e m od e li s a n a b s o r b e d A P E C t h e r m a l c o m pon e n t p l u s a G a u ss i a n t h a t m od e l s t h e fl uo r e s ce n cea t ∼ . − . e V . R a ng e s a tt h e % c on fi d e n ce l e v e l a r e i nd i ca t e d i nb r ace s , i n s o m eca s e s t h e h i gh li m it o f t e m p e r a t u r e un ce r t a i n t y i s no t c on s t r a i n e d . A v a l u e o f Z / Z (cid:12) o f . a s b ee nd e r i v e d fr o m t h e t h i r d e xpo s u r e ( qu i e s ce n t ph a s e ) a ndu s e d f o r t h e b e s t fi t o f t h e s p ec t r a o f t h e fi r s t a nd s ec ond e xpo s u r e s . T h e c on fi d e n ce l e v e l un ce r t a i n t yo f t h ee qu i v a l e n t w i d t h s o f t h e G a u ss i a n li n e i s o f o r d e r o f . - . e V . U n a b s o r b e d fl ux e s f o r t h e A P E C a nd t h e G a u ss i a n c o m pon e n t s a r e g i v e n i n t h ee n e r gyb a nd0 . . e V , t h e fl ux e s a t E > . e V a r ea l s o li s t e d . X -r a y l u m i no s iti e s a r eca l c u l a t e d fr o m fl ux e s u s i ng a d i s t a n ce o f c . T a b l e . P a r a m e t e r s o f t h e b e s t fi t m od e l s t o t h e N uS T AR s p ec t r a s ho w n i n F i g . . S p ec t r u m N H k T Z / Z (cid:12) l og E M γ E W l og F − ke V l og F . − ke V c m − k e V c m − k e V e r g s − c m − e r g s − c m − T o t a l . ( . . ) . ( . . ) . ( . . ) . ( . . ) . ( . . ) . ( . . )- . - . Q u i e s ce n t . ( . . ) . ( . . ) . ( NA – NA ) . ( . . ) . ( . . ) . ( . . )- . - . F l a r e . ( . ) . ( . . ) . ( NA – NA ) . ( . . ) . ( . . ) . ( . . )- . - . T a b l e . P a r a m e t e r s fr o mm od e l b e s t fi t o f t h e s p ec t r a i n5 . − . e V b a nd . O b s S e g m e n t C oun t s k TE M Z C e n t r o i d N G au E W F APE C F G au χ r e d P r ob ( P N , M O S , M O S ) k e V c m − k e V − pho t on s c m − s − k e V e r g s − c m − e r g s − c m − F i r s t a ll + + . ( . . ) . ( . . ) . ( . . ) . ( . . ) . ( . ) . ( . - . )- . (- . - . )- . (- . - . ) . . S ec ond1463 + + . ( . . ) . ( . . ) ( . ) . ( . . ) ( . ) . ( . - . )- . (- . - . )- . (- . - . ) . . + + . ( . . ) . ( . . ) . ( . . ) . ( . . ) ( ) . ( . - . )- . (- . - . )- . (- . - . ) . . + + ( . . ) . ( . . ) . ( . . ) . ( . . ) ( ) . ( . - . )- . (- . - . )- . (- . - . ) . . + + . ( . . ) . ( . . ) . ( . . ) . ( . . ) ( ) . ( . - . )- . (- . - . )- . (- . - . ) . . a + + . ( . . ) . ( . . ) . ( . . ) . ( . . ) . ( . . ) . ( . - . )- . (- . - . )- . (- . - ) . . & + + . ( . . ) . ( . . ) . ( . . ) . ( . . ) ( . ) . ( . - . )- . (- . - . )- . (- . - . ) . . T h i r d a ll + + . ( . . ) . ( . . ) . ( . . ) . ( . . ) . ( . . ) . ( . - . )- . (- . - . )- . (- . - . ) . . N o t e : F o r t h e fi r s t a nd t h i r d X MM - N e w t on ob s e r v a ti on s w ec on s i d e r e d t h e s p ec t r a o f t h e f u ll e xpo s u r e , f o r t h e s ec ondon e w ec on s i d e r e d t h e s p ec t r a i n t h e ti m e i n t e r v a l s i d e n ti fi e d w it h P ELT ( s ee F i g . ) . U n ce r t a i n t y r a ng e s a t % c on fi d e n ce l e v e l a r e i n r e po r t e d . G a s a b s o r p ti onh a s b ee n fi x e d t o N H = . × c m − . F l ux e s o f t h e A P E C a nd t h e G a u ss i a n c o m pon e n t s a r eca l c u l a t e d i n t h e . - . e V b a nd , t h e s e fl ux e s a r e p l o tt e d i n F i g . a s a f un c ti ono f t h e ti m e . Article number, page 14 of 15. Pillitteri et al.: A deep X-ray view of the Class I YSO Elias 29 with
XMM-Newton and
NuSTAR . References
Albacete Colombo, J. F., Caramazza, M., Flaccomio, E., Micela, G., & Sciortino,S. 2007, A&A, 474, 495Ballantyne, D. R. & Fabian, A. C. 2003, ApJ, 592, 1089Ballantyne, D. R., Fabian, A. C., & Ross, R. R. 2002, MNRAS, 329, L67Barret, D., Lam Trong, T., den Herder, J.-W., et al. 2016, in Proc. SPIE, Vol.9905, Space Telescopes and Instrumentation 2016: Ultraviolet to GammaRay, 99052FBontemps, S., André, P., Kaas, A. A., et al. 2001, A&A, 372, 173Boogert, A. C. A., Hogerheijde, M. R., Ceccarelli, C., et al. 2002, ApJ, 570, 708Caramazza, M., Flaccomio, E., Micela, G., et al. 2007, A&A, 471, 645Ceccarelli, C., Boogert, A. C. A., Tielens, A. G. G. M., et al. 2002, A&A, 395,863Czesla, S. & Schmitt, J. H. H. M. 2007, A&A, 470, L13Czesla, S. & Schmitt, J. H. M. M. 2010, A&A, 520, A38Drake, J. J., Ercolano, B., & Swartz, D. A. 2008, ApJ, 678, 385Emslie, A. G., Phillips, K. J. H., & Dennis, B. R. 1986, Sol. Phys., 103, 89Erdman, C. & Emerson, J. W. 2007, Journal of Statistical Software, 23, 1Erdman, C. & Emerson, J. W. 2008, Bioinformatics, 24, 2143Favata, F., Flaccomio, E., Reale, F., et al. 2005, ApJS, 160, 469Favata, F. & Micela, G. 2003, Space Sci. Rev., 108, 577Feigelson, E. D. & Montmerle, T. 1999, ARA&A, 37, 363Flaccomio, E., Micela, G., Sciortino, S., et al. 2005, ApJS, 160, 450Flaccomio, E., Micela, G., Sciortino, S., & the CSI Collaboration. 2018, A&A,submittedFukazawa, Y., Hiragi, K., Mizuno, M., et al. 2011, ApJ, 727, 19Getman, K. V., Flaccomio, E., Broos, P. S., et al. 2005, ApJS, 160, 319Giardino, G., Favata, F., Pillitteri, I., et al. 2007, A&A, 475, 891Golub, L., Maxson, C., Rosner, R., Vaiana, G. S., & Serio, S. 1980, ApJ, 238,343Güdel, M. 2003, Advances in Space Research, 32, 2045Güdel, M., Briggs, K. R., Arzner, K., et al. 2007, A&A, 468, 353Güdel, M. & Telleschi, A. 2007, A&A, 474, L25Harrison, F. A., Craig, W. W., Christensen, F. E., et al. 2013, ApJ, 770, 103Imanishi, K., Koyama, K., & Tsuboi, Y. 2001, ApJ, 557, 747Isola, C., Favata, F., Micela, G., & Hudson, H. S. 2007, A&A, 472, 261Kallman, T. R., Palmeri, P., Bautista, M. A., Mendoza, C., & Krolik, J. H. 2004,ApJS, 155, 675Kamata, Y., Koyama, K., Tsuboi, Y., & Yamauchi, S. 1997, Publications of theAstronomical Society of Japan, 49, 461Kastner, J. H., Huenemoerder, D. P., Schulz, N. S., Canizares, C. R., & Wein-traub, D. A. 2002, ApJ, 567, 434Killick, R., Fearnhead, P., & Eckley, I. A. 2012, Journal of the American Statis-tical Association, 107, 1590Klimchuk, J. A., Antiochos, S. K., Norton, D., & Watko, J. A. 2000, inAAS / Solar Physics Division Meeting / Solar Physics Division Meet-ing, 01.44Maggio, A., Flaccomio, E., Favata, F., et al. 2007, ApJ, 660, 1462Maggio, A., Pallavicini, R., Reale, F., & Tagliaferri, G. 2000, A&A, 356, 627Miotello, A., Testi, L., Lodato, G., et al. 2014, A&A, 567, A32Montmerle, T. 1990, in Reviews in Modern Astronomy, Vol. 3, Reviews in Mod-ern Astronomy, ed. G. Klare, 209–233Nandra, K., Barret, D., Barcons, X., et al. 2013, ArXiv e-prints[ arXiv:1306.2307 ]Natta, A., Testi, L., & Randich, S. 2006, A&A, 452, 245Orlando, S., Peres, G., & Reale, F. 2003, Advances in Space Research, 32, 955Parmar, A. N., Culhane, J. L., Rapley, C. G., et al. 1984, ApJ, 279, 866Pillitteri, I., Sciortino, S., Flaccomio, E., et al. 2010, A&A, 519, A34Reale, F. 2007, A&A, 471, 271Reale, F. 2014, Living Reviews in Solar Physics, 11, 4Rocha, W. R. M. & Pilling, S. 2015, ApJ, 803, 18Sciortino, S., Rauw, G., Audard, M., et al. 2013, ArXiv e-prints[ arXiv:1306.2333 ]Stelzer, B., Flaccomio, E., Pillitteri, I., Argiro ffi , C., & Sciortino, S. 2011, inAstronomical Society of the Pacific Conference Series, Vol. 448, 16th Cam-bridge Workshop on Cool Stars, Stellar Systems, and the Sun, ed. C. Johns-Krull, M. K. Browning, & A. A. West, 1279Townsley, L. K., Broos, P. S., Corcoran, M. F., et al. 2011, ApJS, 194, 1Tsuboi, Y., Yamazaki, K., Sugawara, Y., et al. 2016, Publications of the Astro-nomical Society of Japan, 68, 90Tsujimoto, M., Feigelson, E. D., Grosso, N., et al. 2005, ApJS, 160, 503Wang, X. & Emerson, J. W. 2015, Working PaperWolk, S. J., Harnden, Jr., F. R., Flaccomio, E., et al. 2005, ApJS, 160, 423, C., & Sciortino, S. 2011, inAstronomical Society of the Pacific Conference Series, Vol. 448, 16th Cam-bridge Workshop on Cool Stars, Stellar Systems, and the Sun, ed. C. Johns-Krull, M. K. Browning, & A. A. West, 1279Townsley, L. K., Broos, P. S., Corcoran, M. F., et al. 2011, ApJS, 194, 1Tsuboi, Y., Yamazaki, K., Sugawara, Y., et al. 2016, Publications of the Astro-nomical Society of Japan, 68, 90Tsujimoto, M., Feigelson, E. D., Grosso, N., et al. 2005, ApJS, 160, 503Wang, X. & Emerson, J. W. 2015, Working PaperWolk, S. J., Harnden, Jr., F. R., Flaccomio, E., et al. 2005, ApJS, 160, 423