1: %\documentclass[12pt,aas2pp4]{aastex}
2: \documentclass[preprint2]{aastex}
3:
4: \usepackage[]{natbib}
5: \usepackage{graphicx}
6: \usepackage{amssymb}
7: \usepackage{color}
8: \usepackage{ulem}
9:
10: \input{macro.inp}
11:
12: \newcommand{\myemail}{Kenji.Hamaguchi@nasa.gov}
13:
14: \shorttitle{Resolving a Protostar Binary System}
15: \shortauthors{Hamaguchi et al.}
16:
17: \slugcomment{arXiv}
18:
19: \begin{document}
20:
21: \title{
22: Resolving a Class I Protostar Binary System with Chandra
23: }
24:
25: \author{Kenji Hamaguchi\altaffilmark{1,2}, Minho Choi\altaffilmark{3}, Michael F. Corcoran\altaffilmark{1,4}, Chul-Sung Choi\altaffilmark{3}, Ken'ichi Tatematsu\altaffilmark{5}, Rob Petre\altaffilmark{6}}
26:
27: \altaffiltext{1}{CRESST and X-ray Astrophysics Laboratory NASA/GSFC,
28: Greenbelt, MD 20771}
29:
30: \altaffiltext{2}{Department of Physics, University of Maryland, Baltimore County,
31: 1000 Hilltop Circle, Baltimore, MD 21250}
32:
33: \altaffiltext{3}{International Center for Astrophysics, Korea Astronomy and Space Science
34: Institute, Hwaam 61-1, Yuseong, Daejeon 305-348, South Korea}
35:
36: \altaffiltext{4}{Universities Space Research Association,
37: 10211 Wincopin Circle, Suite 500, Columbia, MD 21044}
38:
39: \altaffiltext{5}{National Astronomical Observatory of Japan, 2-21-1 Osawa, Mitaka,
40: Tokyo 181-8588, Japan}
41:
42: \altaffiltext{6}{Astrophysics Science Division, NASA Goddard Space Flight Center,
43: Greenbelt, MD 20771}
44:
45: \altaffiltext{}{{\tt Mail to:} \myemail}
46:
47: \begin{abstract}
48: Using a sub-pixel event repositioning technique,
49: we spatially resolved X-ray emission from the infrared double system IRS~5
50: in the R Corona Australis molecular cloud with $\sim$0\FARCS8 separation.
51: As far as we know, this result - obtained from 8 Chandra archival observations between 2000 and 2005 - is the first X-ray study of individual sources in a Class I protostar binary system with a projected separation of less than 200~AU.
52: We extracted light curves and spectra of the individual sources using a
53: two-dimensional image fitting method.
54: IRS~5a at the south,
55: the source which was brighter in the near-infrared, showed three X-ray flares lasting $>$20~ksec,
56: reminiscent of X-ray flares from pre-main sequence stars,
57: while the northern source (IRS~5b) was quiescent in X-rays in all the observations
58: except for a 2005 August 9 observation with a factor of $\sim$2 flux enhancement.
59: In quiescence, these sources showed almost identical X-ray spectra, with
60: \NH\ $\sim$4$\times$10$^{22}$~\UNITNH, \KT\ $\sim$2~keV,
61: and log \LX\ $\sim$30.2$-$3~\UNITLUMI.
62: IRS~5a showed plasma at temperatures up to \KT\ $\sim$5$-$6~keV during flares,
63: while the column density of IRS~5b increased by a factor of 2
64: during an observation on 2005 August 9.
65: We discuss the evolutionary stages and variation of the X-ray activity of these sources.
66: \end{abstract}
67:
68: \keywords{techniques: high angular resolution ---
69: binaries: visual --- stars: magnetic fields --- stars: pre--main-sequence --- X-rays: stars}
70:
71: \section{Introduction}
72: \label{sec:introduction}
73:
74: Many, if not most, stars are born as a member of a multiple system through fragmentation
75: of the parent molecular cloud. Because young stellar objects (YSOs) that form as a
76: binary or multiple system are of the same age and chemical composition, any differences in the
77: level of their stellar activity should be due to differences of stellar initial parameters such as
78: mass, net angular momentum and/or the presence of the companion star
79: (See reviews about multiplicity of young stars in \citealt{Duchene2007,Monin2007},
80: and magnetic activity on various stellar parameters
81: in ex. \citealt{Chabrier2006,Dobler2006}).
82: YSO binaries are therefore a useful probe to test models of the dependence of the
83: activity on stellar parameters for stars of the same age.
84:
85: The R Corona Australis (R CrA) molecular cloud has rich association of YSOs
86: with various evolutionary stages \citep{Taylor1984}.
87: Among them, IRS~5 (a.k.a. TS 2.4) suffers strong extinction (\AV\ $\sim$45 mag)
88: and is classified as a Class~I protostar
89: \citep[][see therein for the spectral classification of YSOs]{Wilking1986,Wilking1992}.
90: High angular resolution near-infrared imaging separated IRS~5
91: into two stellar components,
92: IRS~5a ($K \sim$10.9$^{m}$) and IRS~5b ($K \sim$11.5$^{m}$). The
93: separation of the two stars is less than an arcsecond \citep{Chen1993,Nisini2005}. In what follows the term IRS~5 means the entire binary system,
94: or is used when the emitting source was unknown due to limited imaging capability.
95: The southern source, IRS~5a, has a stellar spectral type of K5-K7V
96: and is located in a low mass protostar phase on the HR diagram
97: \citep{Nisini2005}.
98: Stellar parameters of IRS~5b are poorly known due to its faintness.
99:
100: IRS~5 is also known as one of the rare Class I protostars
101: showing strongly variable cm continuum radio emission \citep{Suters1996}
102: with significant circular polarization \citep{Feigelson1998}.
103: Such gyrosynchrotron radio emission is
104: believed to be the major radio emission mechanism
105: of weak-lined T-Tauri Stars \citep[wTTSs, e.g.][]{Carkner1997},
106: but the emission from protostars is supposed to be absorbed
107: by the surrounding ionized materials heated by mass accretion.
108: IRS~5 does not currently show strong mass accretion activity,
109: but there is evidence of an outflow possibly driven by IRS5 \citep[HH~731,][]{Wang2004}.
110: The near-infrared spectrum of IRS~5a shows small continuum excess
111: above the photospheric flux,
112: which also suggests a low accretion rate \citep{Nisini2005}.
113: The infrared spectrum also shows characteristics of scattered emission.
114: From these results, \citet{Nisini2005} questioned
115: the idea that the IRS~5 system is a bona-fide protostar,
116: but it may be close to the TTS phase.
117:
118: IRS~5 also showed strongly absorbed thermal X-ray emission
119: \citep[\NH\ $\sim$4$\times$10$^{22}$~\UNITNH, \KT\ $\sim$2~keV,][]
120: {Koyama1996,Neuhaeuser1997,Forbrich2006,Forbrich2007,Forbrich2007b}.
121: The emission was highly variable, accompanied by occasional rapid flares.
122: These X-ray characteristics are consistent with plasma heating by magnetic activity
123: in the stellar corona, so that the
124: X-ray and radio emission from IRS~5 is suspected to arise from the same root cause.
125: Similarly,
126: \citet{Forbrich2007} did not find a clear correlation between the X-ray and near-infrared
127: flux variation from IRS~5 and other cluster members, which
128: tends to support a coronal origin for the X-ray activity
129: instead of accretion driven activity.
130:
131: To understand the relation between the X-ray and radio activity of IRS~5 and
132: study the evolution of the activity,
133: we need to resolve the IRS~5 binary system in both wavelengths.
134: \citet{Choi2008} successfully resolved IRS~5 into two cm radio sources
135: in a VLA radio observation.
136: Resolving X-ray sources with a sub-arcsecond separation is challenging even
137: with the \CHANDRA\ observatory with its unprecedented X-ray angular resolution.
138: However, encouraged by a marginal resolution of IRS~5 in a \CHANDRA\ image
139: \citep[see Figure 5 of][]{Forbrich2007},
140: we revisited \CHANDRA\ archival observations of IRS~5
141: and clearly resolved the individual binary components using enhanced spatial information
142: obtained with the Sub-pixel Event Repositioning (SER) technique
143: \citep{Tsunemi2001,Li2003,Li2004}.
144: Using this technique we were also able to extract light curves and spectra of each source
145: with a two-dimensional image
146: fitting method
147: and measured the plasma parameters of individual binary stars for the first time.
148:
149:
150: The distance to the cloud is controversial.
151: \citet{Marraco1981} estimated it to be 129~pc from uvby$\beta$ photometry
152: of three stars in the cloud.
153: Assuming \AV~=3.1 mag,
154: \citet{Casey1998} derived the same distance (129$\pm$11~pc) to the eclipsing binary TY~CrA,
155: which is suspected to be a cluster member.
156: However, \citet{Knude1998} measured $d =$ 170~pc
157: from the interstellar reddening of stars in the cloud having \HIPPARCOS\ distances and Tycho
158: B - V colors, and whose distance we use for IRS~5.
159:
160: \input{tab1}
161:
162: \section{Observations}
163:
164: We found eight archival pointings with \CHANDRA\ \citep{Weisskopf2002}
165: that include IRS~5 in the {\FOV} (Table~\ref{tbl:obslogs}).
166: These observations occurred between 2000 October and 2005 August
167: and used the Advanced CCD Imaging Spectrometer with
168: the Imaging array (ACIS-I) at the prime focus with exposure times between 15$-$40~ksec.
169: During the observations,
170: IRS~5 was put at 1\ARCMIN$-$2\ARCMIN\ off-axis, where the \PSF\ quality is negligibly
171: different from the on-axis position.
172: We use the convention that individual \CHANDRA\ observations are designated CXO,
173: subscripted with the year, month and day of the observation.
174: We downloaded the data from the HEASARC archive\footnote{http://heasarc.gsfc.nasa.gov/W3Browse/} and reprocessed them with the CIAO\footnote{http://cxc.harvard.edu/ciao/} analysis software
175: version 3.4 (CALDB ver. 3.3.0.1).
176: We reprocessed the data, removed the pixel randomization and
177: utilized the SER algorithms to improve the spatial resolution.
178:
179: Photon pile-up can cause source count rates to be underestimated, spectral shapes to appear harder,
180: and degrades event position determination in the SER method.
181: Fortunately,
182: observed X-ray photon counts of IRS~5 ranged between 0.02$-$0.08 \UNITCPS,
183: corresponding to small pile-up fractions of $\sim$2$-$8~\%\footnote{http://heasarc.gsfc.nasa.gov/Tools/w3pimms.html}.
184: We thus did not correct for any pile-up effect in our analysis.
185:
186: \section{Analysis and Results}
187: \label{sec:anaresult}
188:
189: \subsection{Absolute Astrometry}
190:
191: This study crucially requires absolute astrometry of the X-ray images
192: with a few tenths of arc-second precision.
193: We therefore measured sky positions of X-ray sources around IRS~5
194: and cross-correlated them with sources in the Point Source Catalog of the
195: 2MASS All-Sky Data Release.
196:
197: There is up to 7~year time span between the interval when the 2MASS observations of the southern sky
198: (1998 March $-$ 2001 February) and the \CHANDRA\ observations (2000 Oct $-$ 2005 Aug).
199: The cluster members can move significantly within 7~years if their proper motions are suitably large.
200: The \HIPPARCOS\ catalogue has 4 sources with proper motion measurements \citep{Perryman1997b}.
201: Among them, R CrA and V709~CrA have proper motion of $\sim$80$-$100~mas yr$^{-1}$,
202: but they have large uncertainty in the parallax measurement ($\sim$68, 140 mas)
203: and almost opposite directions in their proper motions.
204: This would suggest that their measurements are less reliable,
205: possibly due to contamination of surrounding nebular emission.
206: Two other sources, TY CrA and HD~176386, with better parallax measurements
207: have small proper motions ($\lesssim$20~mas~yr$^{-1}$).
208: Our tentative measurement of the proper motion of cluster members
209: using VLA data between 1997 and 2005 also
210: derived similarly small proper motions of $\lesssim$28 mas~yr$^{-1}$\citep{Choi2008}.
211: These results suggest that the proper motion of the cluster members
212: is small, less than $\sim$150~mas in 7 years ($\lesssim$1~pixel in the Figure 1 image).
213:
214: \begin{figure*}
215: \begin{center}
216: \epsscale{1.3}
217: \plotone{f1a.eps}
218: \caption{Magnified images of IRS~5 (raw: {\it left},
219: smoothed: {\it right}), sequentially in time from above.
220: The gray scales are shown in a linear scale for all figures (but saturated
221: above 10 cnts for the left panels).
222: Crosses show X-ray source positions determined from our analysis.
223: Circles at the bottom left of the left panels show the half of the
224: peak intensity of the \PSF.
225: Raw image bins are 0\FARCS123 pixel$^{-1}$.
226: Gray scales and contours in the right panels
227: are normalized by the effective area and exposure time in CXO$_{050808}$.
228: \label{fig:image}}
229: \end{center}
230: \end{figure*}
231:
232: \begin{figure*}
233: \epsscale{1.3}
234: \plotone{f1b.eps}
235: \centerline{Fig. 1. --- Continued.}
236: \end{figure*}
237:
238: \begin{figure}[t]
239: \begin{center}
240: \epsscale{1.0}
241: \plotone{f2.eps}
242: \caption{Composite of images in CXO$_{001007}$, CXO$_{050808}$,
243: CXO$_{050812}$ and CXO$_{050813}$, when both of IRS~5a and IRS~5b
244: were in the quiescent phase.
245: Markers --- cross: X-rays, circle: cm radio \citep{Choi2008},
246: diamond: 2MASS, star: infrared source positions in \citet{Nisini2005},
247: shifted to match for the cm radio source positions.
248: Error ranges of the X-ray source positions from the 2MASS frame (200~mas),
249: radio beam size \citep{Choi2008} and IR image pixel size \citep{Nisini2005}
250: are shown at the bottom left.
251: \label{fig:imgcomposite}
252: }
253: \end{center}
254: \end{figure}
255:
256: To measure X-ray source positions in each observation,
257: we ran the {\it wavdetect} source detection tool in the CIAO package for
258: a 0.35$-$8~keV image binned at 0\FARCS5 pixel$^{-1}$ and
259: only considered sources at above the 4$\sigma$ significance.
260: The 2MASS positions of IRS~1, IRS~2, R~CrA, IRS~13 and HBC~667 had
261: corresponding X-ray counterparts
262: with similar offsets
263: ($\lesssim$0\FARCS1 between the sources,
264: see source positions in Figure~1 of \citealt{Hamaguchi2005b} and \citealt{Forbrich2006}),
265: while there was no counterpart to the X-ray source corresponding to IRS~6
266: in the 2MASS catalogue within 1\ARCSEC.\footnote{It was in the ``reject" table.}
267: We therefore used the former 5 sources for the positional references.
268:
269: The 2MASS positions of IRS~1, IRS~2, R~CrA, IRS~13 and HBC~667 ($K \sim$7.0--10.5$^{m}$)
270: have possible position inaccuracy of up to 100~mas RMS, except for the brightest source R CrA
271: ($K$ =2.9$^{m}$),
272: which had a slightly larger position uncertainty of 160~mas (RMS)\footnote{http://www.ipac.caltech.edu/2mass/releases/allsky/doc/}.
273: The uncertainty of the mean position of all the sources is $\sim$50~mas.
274: Combined with the cluster proper motion noted above,
275: we can determine the absolute source position better than $\sim$200~mas for these sources.
276: Table~\ref{tbl:obslogs} lists the shifts of the X-ray sky coordinate frame
277: from the original X-ray data.
278: Earlier observations in 2000$-$2003 needed relatively large shifts ($\sim$0.3$-$1\ARCSEC),
279: while observations in 2005 needed almost negligible shifts ($\lesssim$0.2\ARCSEC).
280: Table~\ref{tbl:obslogs} also lists the standard offset deviation of the X-ray source positions
281: from the 2MASS source positions.
282: The result suggests that X-ray source positions in all the \CHANDRA\ observations
283: were determined at $\lesssim$200~mas from the 2MASS
284: frame, and hence $\lesssim$400~mas in the absolute coordinate frame.
285:
286: \begin{figure*}
287: \begin{center}
288: \epsscale{2.1}
289: \plotone{f3.eps}
290: \caption{Light curves between 1$-$8~keV ({\it top}) of IRS~5a ({\it black}) and IRS~5b ({\it red}).
291: Each bin has 4~ksec and tic marks on the horizontal axis are shown by 10~ksec
292: from the beginning of each observation.
293: The error bars show 1$\sigma$ confidence range.
294: In CXO$_{030626}$, CXO$_{040617}$, CXO$_{050810}$, and CXO$_{050813}$,
295: IRS~5b emitted X-rays comparable to the other observations, but {\tt wavdetect} did not detect it.
296: This is probably because enhanced X-ray emission from IRS~5a contaminated strongly at the
297: IRS~5b position, and thus increased the apparent background level (see \S\ref{subsec:timspec}).
298: Labels at the top depict
299: observations which showed significant X-ray variations in either of IRS~5a nor IRS~5b
300: (FO: Flare Observation) or in neither of them (QO: Quiescent Observation).
301: \label{fig:lightcurves}}
302: \end{center}
303: \end{figure*}
304:
305: \subsection{Image Analysis}
306: \label{subsec:anaimage}
307:
308: Figure~\ref{fig:image} shows X-ray images of IRS~5 between 1$-$8~keV
309: after the absolute position correction.
310: They were binned with 0\FARCS123 pixel$^{-1}$ to resolve sources having sub-arcsecond separation.
311: In all the images, X-ray emission was clearly elongated toward the NE-SW direction.
312: We checked in two ways if this structure is real. First, we created
313: simulations of
314: point spread functions (\PSF) at the location of IRS~5 with {\tt ChaRT} \citep{Carter2003} and
315: {\tt MARX}\footnote{http://space.mit.edu/CXC/MARX/} version 4.2.1, assuming a spectrum
316: \NH\ =4.5$\times$10$^{22}$~\UNITNH, \KT\ =3~keV and $Z$ =0.3~solar,
317: (based on our preliminary fits of the IRS~5 spectra using the standard method). Second, we made images of
318: IRS~1 with a similar off-axis angle to IRS~5
319: to determine if the SER method might introduce asymmetries in the \PSF.
320: We found no such extension from either of the half maximum \PSF\ circles
321: (Figure~\ref{fig:image}) nor the IRS~1 images.
322: We conclude that the deduced elongation of the X-ray image of IRS~5 is real.
323:
324: In CXO$_{030626}$, CXO$_{040617}$ and CXO$_{050810}$
325: the south-west part of the elongation was remarkably bright,
326: while in CXO$_{050809}$ the north-east part looked slightly brighter than the south-west part.
327: In the other observations, we recognized apparently two peaks with similar intensities.
328: We re-ran the {\tt wavdetect} source detection tool for these high resolution images
329: and detected the south-west source in all observations, and detected the north-east source in
330: CXO$_{001007}$, CXO$_{050808}$, CXO$_{050809}$ and CXO$_{050812}$.
331: The north-east peak was not detected
332: when the south-west peak was very bright or images lacked sufficient exposure.
333: Both peaks were located at, on average,
334: ($\alpha_{2000}$, $\delta_{2000}$) = (19$^{h}$1$^{m}$48$.^{s}$07, $-$36\DEGREE57$'$21\FARCS8)
335: designated using the \CHANDRA\ naming convention as CXOU J190148.1-365722 and
336: ($\alpha_{2000}$, $\delta_{2000}$) = (19$^{h}$1$^{m}$48$.^{s}$02, $-$36\DEGREE57$'$22\FARCS4)
337: as CXOU J190148.0-365722 and the positions of the individual detections differed by
338: 0\FARCS09 for the former and 0\FARCS075 for the latter in RMS.
339: These sources are separated by only $\sim$0\FARCS8$\pm$0\FARCS12 (1$\sigma$), corresponding
340: to a $\sim$140~AU projected separation at $d =170$~pc.
341:
342: A composite image of
343: CXO$_{001007}$, CXO$_{050808}$, CXO$_{050812}$ and CXO$_{050813}$
344: clearly showed two peaks at the north-east and south-west sides (Figure~\ref{fig:imgcomposite}).
345: These X-ray peaks have corresponding cm radio counterparts
346: within $\sim$0\FARCS1 for each \citep{Choi2008}.
347: They have offsets from the infrared sources IRS~5a and IRS~5b of $\sim$1\ARCSEC\
348: according to their absolute coordinates in the $H$ and $K$ band maps in Figure 1 of
349: \citet{Nisini2005}.
350: The 2MASS survey with an absolute astrometric accuracy of $\sim$100~mas
351: detected an infrared source very close to the north-east X-ray source
352: though it did not resolve the two infrared sources probably due to the limited spatial resolution of the CCD (2\ARCSEC).
353: Moreover,
354: the two X-ray sources and the IR sources in \citet{Nisini2005} have similar position
355: angle and separation, and therefore match their positions well by shifting the IR image
356: frame (see Figure~\ref{fig:imgcomposite}).
357: We therefore identified the south-west X-ray source (CXOU J190148.0-365722) as IRS~5a and
358: the north-east source (CXOU J190148.1-365722) as IRS~5b.
359:
360: \begin{figure*}
361: \begin{center}
362: \plotone{f4.eps}
363: \caption{Quiescent spectra of IRS~5a ({\it top left}) and IRS~5b ({\it top right}) and flare spectra of IRS~5a
364: in CXO$_{030626}$ ({\it middle left}), CXO$_{040617}$ ({\it middle right}), CXO$_{050808}$ ({\it bottom left}) and of IRS~5b in CXO$_{050809}$ ({\it bottom right}), overlaid on the quiescent spectra (grey).
365: \label{fig:spectra}}
366: \end{center}
367: \end{figure*}
368:
369: \input{tab2}
370:
371: \subsection{Timing and Spectral Analysis with 2-Dimensional Image Fits}
372: \label{subsec:timspec}
373:
374: Because the X-ray emission from IRS~5a and IRS~5b is strongly blended,
375: the standard method to generate light curves and spectra ---
376: counting photons
377: within user defined source and background regions --- cannot be used straightforwardly.
378: Therefore
379: we utilized the 2-dimensional image fitting method in which we generated multiple images from each observation in restricted time and energy ranges and fit each
380: 2-dimensionally by the combined individual model \PSF\ images for IRS~5a and IRS~5b.
381: We created images
382: with a size of 80~pixel~$\times$~80~pixel
383: binned by 0\FARCS123~pixel$^{-1}$.
384: For the model images, we used the {\tt ptsrc2d} function in the CIAO package,
385: assuming two point source components at the position of the X-ray sources,
386: and used the {\tt sherpa} tool to carry out the fitting procedure
387: (We should note that a part of the X-ray emission may come from different
388: and/or extended regions.)
389: We did not include background in the model images, since the background is negligibly small
390: ($\sim$ 3$\times$10$^{-4}$~cnts / (1 pixel=0\FARCS123)$^{2}$ / 10~ksec).
391: The {\tt ptsrc2d} function loads an image as a template model \PSF.
392: Using these model PSFs, we generated photon events at the position of IRS~5
393: using both {\tt ChaRT} \citep{Carter2003} and {\tt MARX}.
394: For the light curve analysis, we assumed a \PSF\ to have the typical spectrum
395: of IRS~5 based on our preliminary analysis using the standard method,
396: i.e. \NH\ =4.5$\times$10$^{22}$~\UNITNH, \KT\ =3~keV and 0.3~solar elemental abundance.
397: In both the timing and spectral analyses,
398: we generated \PSF\ images in the same energy band as the observed images
399: using large numbers of photon events to minimize statistical fluctuations.
400: In the {\tt sherpa} fits,
401: we fixed the centroids of two point source components at the X-ray source positions of
402: IRS~5a and IRS~5b and only allowed their normalization parameters to vary.
403: We used the Cash statistic appropriate for Poisson-limited data and used the Powell method to find the
404: best-fit values.
405: We derived upper and lower boundaries of the 1$\sigma$ confidence range
406: using the {\tt projection} function.
407:
408: Figure~\ref{fig:lightcurves} shows individual light curves of IRS~5a and IRS~5b
409: in 4~ksec bins in the energy range 1$-$8~keV.
410: We normalized the individual lightcurves to the 3~keV effective area at the position of IRS~5a,
411: resulting in small correction factors ($\lesssim$1~\%).
412: Their flux variations look to be uncorrelated, suggesting no significant interference
413: with each other in our 2-dimensional fits.
414: To further check the consistency of the image fitting results,
415: we summed the individual light curves of IRS~5a and IRS~5b
416: and compared them to a light curve extracted from the entire IRS~5 complex, including photon
417: counts within a 3\ARCSEC\ radius circle
418: centered on between IRS5a and IRS5b, which
419: include $\sim$90$-$93\% of photons from both IRS~5a and IRS~5b.
420: We found that the standard method obtained 93$-$98\% of the photon counts
421: we derived by combining the individual IRS~5a and IRS~5b photons.
422: We had similar results for the split energy bands between 1$-$3.5~keV and 3.5$-$8~keV,
423: which we used for a hardness ratio analysis described in the next paragraph.
424: This suggests that our image fitting method produces a result that is consistent with the standard method
425: within $\lesssim$10\% discrepancy.
426:
427: In CXO$_{030626}$, CXO$_{040617}$, and CXO$_{050810}$,
428: IRS~5a showed significant time variations, reminiscent of stellar X-ray flares.
429: The fast rise ($\sim$4~ksec in CXO$_{040617}$) and
430: slow decay (a half decay time of 15$-$30~ksec) time scales based on visual inspection
431: were similar to those of relatively strong X-ray flares from YSOs.
432: The flux level did not change remarkably during the other observations ($\lesssim$50\%).
433: IRS~5b was quite stable in almost all the available observations.
434: The non-detection of IRS~5b with {\tt wavdetect} in 4 observations would be therefore
435: caused by increased contamination from IRS~5a by flares, or from limited photon statistics.
436: IRS~5b showed a small flux enhancement by a factor of 2 in CXO$_{050809}$,
437: and an apparent decrease at the middle of CXO$_{040617}$, but this decline
438: could be an artifact produced by strong contamination from the IRS~5a flare
439: in this observation along with a possible offset of
440: the south-west emission peak from IRS~5a.
441: We split the energy band into two at 3.5~keV, which is near the median of
442: the photon distribution against energy in a spectrum,
443: and measured the hardness ratio ($HR$) of each observation,
444: which we have defined as $HR = (H-S)/(H+S)$,
445: where the $H$ and $S$ are count rates of the 3.5$-$8~keV and 1$-$3.5~keV bands, respectively.
446: The average $HR$s of both IRS~5a and IRS~5b
447: in CXO$_{001007}$, CXO$_{050808}$, CXO$_{050812}$ and CXO$_{050813}$,
448: when neither showed significant variations
449: (and which we refer to as the ``quiescent observations''),
450: were $HR\sim-$0.32$\pm$0.15,
451: $-$0.29$\pm$0.18 (1$\sigma$), respectively.
452: The $HR$s were $\sim-$0.02$\pm$0.09 (1$\sigma$) during flares from IRS~5a
453: (CXO$_{030626}$, CXO$_{040617}$, CXO$_{050810}$)
454: and $\sim$0.04$\pm$0.08 (1$\sigma$) during a small flux enhancement of IRS~5b
455: (CXO$_{050809}$).
456: The $HR$ was slightly higher in the flare than during the quiescent observations.
457:
458: We generated individual spectra of IRS~5a and IRS~5b in key phases
459: and analyzed them with {\tt xspec} version~11.
460: One technical issue was that {\tt xspec} requires integer photon counts,
461: while the {\tt ptsrc2d} function outputs the normalization values (i.e. photon counts) in floating point.
462: Thus, to avoid any round off error,
463: we multiplied photon counts and exposure time by 1000 when creating spectral files.
464: The {\tt xspec} format also assumes a single error value for each spectral bin,
465: while the {\tt projection} function outputs upper and lower error ranges.
466: We therefore used the larger value output from {\tt projection} as the {\tt xspec} error for each spectral bin. Since
467: we set most of the spectral bins to have $\gtrsim$15~photons
468: the upper and lower error ranges typically have less than $\sim$10\% differences.
469: We generated response matrices and auxiliary files corresponding to the spectral files
470: using {\tt mkacisrmf} and {\tt mkarf} in the CIAO package.
471:
472: Since no quiescent observation had
473: enough photon statistics to generate a spectrum with sufficient signal-to-noise,
474: we combined images of these observations in each spectral bin.
475: We combined their \PSF\ images and spectral responses as well for our 2-dimensional
476: image fits and spectral model fits.
477: The result is shown in Figure~\ref{fig:spectra}.
478: The quiescent spectra of IRS~5a and IRS~5b were almost identical with
479: significant emission up to $\sim$8~keV and a cut-off due to absorption at $\sim$1.5~keV.
480: We modeled both spectra using an absorbed optically thin, 1-temperature (1T) plasma model
481: (WABS, \citealt{Morrison1983}; APEC\footnote{http://cxc.harvard.edu/atomdb/},
482: Table~\ref{tbl:specres}).
483: The best-fit models had \NH\ $\sim$4$\times$10$^{22}$~\UNITNH\ and \KT\ $\sim$2~keV
484: for both sources.
485: The absorption corrected log \LX\ was 30.2$-$3 (0.5$-$10~keV).
486:
487: We compared the IRS~5a spectra during flaring states
488: (CXO$_{030626}$, CXO$_{040617}$ and CXO$_{050810}$)
489: to its quiescent spectrum (Figure~\ref{fig:spectra}).
490: All flare spectra showed stronger emission in the 1$-$8~keV energy band
491: than the quiescent spectrum.
492: We fit the flare spectra by an absorbed 1T plasma model plus the best-fit model of
493: the quiescent IRS~5a spectrum, assuming that a new plasma component appeared
494: during the flares.
495: The best fit to the flare spectra seems to have higher \KT\
496: ($\sim$5$-$7~keV) than the quiescent spectrum but similar \NH\ ($\sim$5$\times$10$^{22}$~\UNITNH).
497: However,
498: IRS~5b showed similar increase in hard band flux in CXO$_{050809}$
499: compared to the quiescent phase, while the flux below $\sim$2~keV was apparently
500: unchanged (Figure~\ref{fig:spectra}).
501: The best fit, assuming an absorbed 1T model plus the best-fit model of
502: the quiescent IRS~5b spectrum showed significantly larger
503: \NH~$\sim$7.8$^{+4.2}_{-1.6}\times$10$^{22}$~\UNITNH\ (1$\sigma$),
504: which does not overlap with \NH\ in the quiescent spectrum (4.5$^{+0.4}_{-0.5}\times$10$^{22}$~\UNITNH, 1$\sigma$).
505: The best-fit model also showed slightly higher \KT\ $\sim$4~keV than the quiescent spectrum.
506: None of these spectra had enough photon statistics to show the Fe line profile,
507: which appeared in some flare spectra of the entire IRS~5a complex.
508:
509: The values of \NH\ and \KT\ for IRS~5a and IRS~5b are similar
510: in all the observations and are consistent with the values in \citet{Forbrich2006},
511: which were derived from the combined spectra of IRS~5a and IRS~5b.
512: The \KT\ and \LX\ values are typical of those from TTSs and Class~I protostars,
513: while the derived \NH\ is typical of Class I protostars \citep[e.g.,][]{Imanishi2001}.
514:
515: \input{tab3}
516:
517: \section{Discussion}
518: \label{sec:discussion}
519:
520: We still have a discrepancy between the positions of the near-infrared sources in \citet{Nisini2005}
521: and the sources detected by 2MASS, VLA and \CHANDRA.
522: IRS~5a was a factor of two brighter than IRS~5b in \citet{Nisini2005},
523: while the 2MASS source, whose position should be biased toward the brighter source (IRS~5a),
524: was instead closer to IRS~5b (see \S\ref{subsec:anaimage}).
525: The 2MASS source is unlikely to be mis-positioned due to surrounding contamination
526: since both IRS~5a and IRS~5b are sufficiently bright in the images shown in Figure~1 of \citet{Nisini2005}.
527: IRS~5a or IRS~5b might have experienced an increase in its near-infrared flux
528: during the observations in \citet{Nisini2005} or during the 2MASS observations.
529: We should note that
530: IRS~5 only showed variation of less than $\pm$0.03mag during monitoring observations
531: in 2005 \citep[see Figure 6 of][]{Forbrich2007}.
532:
533: It is possible that the X-ray sources we have identified with either IRS~5a or IRS~5b could be a background red giant star or an active galactic nucleus, though this is unlikely.
534: However,
535: X-ray emission from red giant stars are not usually as bright
536: as, nor as hot as, emission from IRS~5a or IRS~5b \citep{Pizzolato2000}.
537: Though some AGNs are observed with fluxes similar to those from IRS~5a or IRS~5b,
538: the X-ray spectra are generally harder
539: \citep[spectral photon index $\Gamma \lesssim$2, ex.][]{Ueda1998}
540: than those observed for either IRS~5a or IRS~5b ($\Gamma \sim$3.5 assuming an
541: absorbed power-law model).
542: On the other hand,
543: \citet[][]{Pontoppidan2003} showed CO ice absorption lines of IRS~5a and IRS~5b
544: have similar systemic velocities to the other YSOs in the R~CrA cloud ($\sim-$20~\UNITVEL)
545: though this feature can be produced by an absorber physically unrelated to the infrared sources.
546: The X-ray spectra are typical X-ray emission from Class~I protostars.
547: These results suggest that IRS~5a and IRS~5b form a protostar binary system within the R CrA cloud.
548:
549: Table~\ref{tbl:xraybin} shows binary systems with projected separations
550: less than 200AU that have been spatially resolved in X-rays.
551: According to this table, IRS~5 is the first X-ray binary system believed to be comprised of Class~I
552: protostars\footnote{Although the infrared spectral slope,
553: which defines the YSO classification,
554: has not been measured for IRS~5b,
555: we think this star is a Class~I protostar based on its high
556: absorption in X-rays.
557: }
558: and is a factor of $\gtrsim$10 younger than the other systems.
559: Binary systems with linear separations of less than 140~AU have very little disk mass,
560: perhaps because of the gravitational influence of the binary companion \citep{Beckwith1990}.
561: The IRS~5 binary separation is near this boundary and the separation may be smaller
562: if the distance is $d\sim$130~pc.
563: The separation is so large that the binary stars would not have direct interaction
564: by magnetic field, for example
565: (see \citealt{Uchida1985} as an example of the system with magnetic interaction),
566: but they may have an influence as discussed by \citet{Beckwith1990}.
567: IRS~5 may be an important example of the influence on X-ray activity of binary
568: interaction in a very early phase.
569:
570: The column densities of IRS~5a and IRS~5b derived from their X-ray spectra
571: ($\sim$4$\times$10$^{22}$~\UNITNH)
572: were typical of Class~I protostars \citep[c.f.][]{Imanishi2001}.
573: These values of \NH\ were a factor of 2 smaller than the \NH\ value estimated from \AV\ for IRS~5a ($\sim$45~mag),
574: assuming an empirical \NH$-$\AV\ relation appropriate to the $\rho$ Oph cloud \citep{Imanishi2001}.
575: This result was consistent with X-ray absorption measurements
576: of IRS1, 2 and 5 \citep{Forbrich2006} and
577: qualitatively consistent with a small \NH/A$_{\rm J}$ ratio of the R CrA cloud \citep{Vuong2003}.
578: The moderate \NH\ may not suggest the edge-on geometry discussed by \citet{Nisini2005}
579: since X-rays from the stellar core would suffer stronger extinction in the edge-on disk
580: than infrared emission from the stellar core and inner disk.
581:
582: The \KT\ and \LX\ values of the quiescent spectra of both IRS~5a and IRS~5b
583: are typical of Class~I protostars and TTSs \citep{Imanishi2001}.
584: Assuming the mass of IRS~5a between 0.4$-$0.9~\UNITSOLARMASS\ \citep{Nisini2005},
585: the \LX\ $\sim$10$^{30.2}$ \UNITLUMI\ corresponds to a $\sim$1~M-year-old pre-main sequence star
586: based on the \LX-age relation of young stars in the Orion nebula though the relation has a large scatter \citep{Preibisch2005}.
587: The \KT s during the flares of IRS~5a are typical of YSOs
588: \citep[e.g.,][]{Imanishi2001,Wolk2005}.
589:
590: IRS~5a exhibited 3 prominent flares during the 8 \CHANDRA\ observations.
591: Since the observations ended at the middle of these flares,
592: we only know lower limits to the flare durations ($\gtrsim$37.6, 19.9 and 15.0 ksec)
593: and total flare energies ($\gtrsim$9.2, 8.7 and 9.3 $\times$10$^{34}$~ergs),
594: respectively.
595: These flares were similar in brightness to that of the brightest flares
596: from classical TTSs and weak-line TTSs
597: observed in the XEST project \citep[Fig. 9 of][]{Stelzer2007}.
598: Curiously, we did not see any flare-like event from IRS~5a fainter than those flares
599: though such flares would have been detectable.
600: IRS~5a had 3 flares during the 152~ksec exposure, that is, one flare per $\sim$50~ksec.
601: This is more than an order of magnitude more frequent than the average of the XEST
602: \citep[1 flare in 770~ksec for flares with the flare energies above 10$^{35}$~ergs,][]{Stelzer2007}
603: and COUP \citep[1 flare in 650~ksec,][]{Wolk2005},
604: though we did not consider possible detections of flares that start outside the observing window.
605: IRS~5 showed remarkable variation in the other \XMM\ observations too \citep{Forbrich2006},
606: for which either of IRS~5a or IRS~5b would be responsible.
607: IRS~5a seems to be an actively flaring source,
608: though the accretion rate was measured to be low \citep{Nisini2005}.
609:
610: IRS~5b did not show apparent variation during the \CHANDRA\ observations
611: except for a flux enhancement in CXO$_{050809}$.
612: The column density of IRS~5b during the flux enhancement
613: increased by a factor of $\sim$2 from the quiescent phase.
614: Such variable X-ray emission in CXO$_{050809}$ might have been produced from a plasma
615: on or around the stellar surface that appeared behind the circumstellar disk,
616: or perhaps IRS~5b may have another deeply embedded companion.
617:
618: Except for the flare activity,
619: X-ray properties of IRS~5a and IRS~5b are strikingly similar.
620: This may strengthen a general assumption that stars in a similar mass range experience
621: similar evolution of their quiescent X-ray spectra.
622: This result also raises a caution in measurements of the \LX\ function of YSOs ---
623: \LX\ can be significantly overestimated if a YSO has an unresolved companion.
624:
625: \section{Conclusion}
626:
627: We spatially resolved X-ray emission from each component of the infrared double system
628: IRS~5 in the R CrA molecular cloud with a 0\FARCS8 projected separation in 8 \CHANDRA\ observations.
629: Their X-ray brightnesses and spectra suggest that
630: they are a Class~I protostar binary system.
631: Separations of binary YSOs are typically less than $\sim$10\ARCSEC\ in nearby star
632: forming regions.
633: \CHANDRA\ enabled study of X-ray properties of such young binary systems
634: for the first time.
635:
636: We derived light curves and spectra of these objects using a 2-dimensional image
637: fitting method.
638: During the 8 \CHANDRA\ observations,
639: the southern source IRS~5a showed three X-ray flares lasting more than
640: 15~ksec and with flare energies of $\gtrsim$10$^{35}$ ergs.
641: These flares were relatively powerful compared to flares from classical TTSs and weak-line TTSs.
642: The flare frequency (1 flare per $\sim$50~ksec) was higher than flares from
643: PMSs in the Taurus and Orion cloud.
644: IRS~5a might be in an active phase, or may have a different mechanism which drives its X-ray flares.
645: IRS~5b, on the contrary, did not show strong variation through the observations,
646: except for a flux increase by a factor of two in 2007 Aug. 9 accompanied by
647: an apparent \NH\ increase.
648:
649: Though IRS~5a and IRS~5b had clear differences in their X-ray variability,
650: their quiescent X-ray spectra were almost identical.
651: This may suggest that both stars have gone through a similar evolutionary phase
652: in X-ray activity.
653: Detailed infrared spectroscopy of IRS~5b is crucial to
654: determine its stellar parameters and to understand nature of the IRS~5 system.
655: Measurement of proper and radial motions of both IRS~5a and IRS~5b
656: would be also important to understand whether they are a gravitationally bound system.
657:
658: \acknowledgments
659:
660: This work is performed while K.H. was supported by the NASA Astrobiology Program
661: under CAN 03-OSS-02.
662: This publication makes use of data products from the Two Micron All Sky Survey, which is a joint project of the University of Massachusetts and the Infrared Processing and Analysis Center/California Institute of Technology, funded by the National Aeronautics and Space Administration and the National Science Foundation.
663: This research has made use of data obtained from the High Energy Astrophysics Science Archive Research Center (HEASARC), provided by NASA's Goddard Space Flight Center.
664:
665: Facilities: \facility{CXO(ACIS-I)}
666:
667: \bibliographystyle{apj}
668: \bibliography{inst,sci_AI,sci_JZ,scibook}
669:
670: \end{document}
671:
672: