A&A 446, 155-170 (2006)
J. Forbrich1,2 - Th. Preibisch1 - K. M. Menten1
1 - Max-Planck-Institut für Radioastronomie, Auf dem Hügel 69, 53121 Bonn, Germany
2 - Astrophysikalisches Institut und Universitäts-Sternwarte Jena, Schillergäßchen 2-3, 07745 Jena, Germany
Received 11 February 2005 / Accepted 7 September 2005
The Coronet cluster in the nearby R CrA dark cloud offers the rare opportunity to study at least four "class I'' protostellar sources as well as one candidate "class 0'' source, a Herbig Ae star, and a candidate brown dwarf within a few square arcminutes - most of them detected at radio- and X-ray wavelengths. These sources were observed with the Very Large Array (VLA) at cm on nine occasions in 1998, spread over nearly four months. The source IRS 5, shown earlier to emit circularly polarized radio emission, was observed to undergo a flux increase accompanied by changes in its polarization properties. Comparison with VLA measurements taken in January 1997 allows analysis of longer-term variability. In addition to this radio monitoring, we analyze archival Chandra and XMM-Newton X-ray data of these sources. Three class I protostars are bright enough for X-ray spectroscopy, and we perform a variability analysis for these sources, covering a total of 154 ks spread over more than two and a half years. Also in X-rays, IRS 5 shows the most pronounced variability, whilst the other two class I protostars IRS 1 and IRS 2 have more stable emission. X-ray data is also analyzed for the recently identified candidate class 0 source IRS 7E, which shows strong variability as well as for the Herbig Ae star R CrA for which we find extremely hot X-ray-emitting plasma. For IRS 1, 2 and 5, the hydrogen column densities derived from the X-ray spectra are at about half the values derived with near-infrared techniques, a situation similar to what has been observed towards some other young stellar objects.
Key words: stars: pre-main sequence - stars: individual: R CrA - radio continuum: stars - X-rays: stars
The energetic processes in and around pre-main sequence stars lead to considerable variability over a wide spectral range. The study of radio and X-ray emission allows to probe energetic processes in the coronae of these stars (see Feigelson & Montmerle 1999) in all their evolutionary stages: Low-mass stars evolve through a series of stages termed class 0 to class III (Lada 1987, and e.g. André et al. 2000). In the class 0 stage, gravitational collapse has started, but most of the material is still in the circumstellar envelope. Soon after the onset of accretion, powerful bipolar outflows develop. The luminosity of class I sources is still mainly due to accretion, although a considerable fraction of the final matter has already been accreted. The canonical age of such objects is assumed to be about 105 years. The class I phase is followed by the class II (=classical T Tauri star) and class III (=weak-line T Tauri star) phases, during which the circumstellar material plays a consecutively diminishing role in the protostar's evolution.
X-ray emission from classical and weak-line T Tauri stars has been known for some time and is thought to be produced in magnetically confined plasma in the coronae (Feigelson & Montmerle 1999; Favata & Micela 2003). During the 1990s, X-ray emission towards a few even younger class I protostars was discovered (reviewed by Neuhäuser 1997). With the launches of Chandra and XMM-Newton, X-ray observatories with higher spatial resolution and better sensitivity to harder photons (which are less affected by extinction) became available, allowing better studies of star-forming regions, such as the Orion Nebula Cluster (Feigelson et al. 2003; Garmire et al. 2000; Feigelson et al. 2002; Getman et al. 2005), Ophiuchi (Ozawa et al. 2005; Imanishi et al. 2001) or IC 348 (Preibisch & Zinnecker 2002,2001). However, the processes giving rise to magnetic activity in young stars in general and class 0/I protostars in particular are still poorly understood. For a review of the associated high-energy processes, see Feigelson & Montmerle (1999), Preibisch (2004a), or, in the framework of stellar radio and X-ray astronomy, Güdel (2002,2004).
Many X-ray emitting class I protostars have X-ray luminosities which are more than ten times higher than those of typical T Tauri stars while their X-ray spectra are harder and show stronger absorption at lower energies due to circumstellar extinction. Powerful flares have been observed frequently (e.g. Imanishi et al. 2001). Until now, however, only relatively few X-ray-detected class I protostars are known. A ROSAT survey of young stellar objects yielded only eleven class I sources (Carkner et al. 1998). Searching for more X-ray detections among class I protostars yields very different detection rates. While Imanishi et al. (2001), studying the Oph cloud, detect 70% of the class I sources in X-rays, a result confirmed by Ozawa et al. (2005), Preibisch (2004b) detects four out of 19 class I sources in the Serpens dark cloud. Imanishi et al. (2001) and Preibisch (2004b) suggest that class I sources might be intrinsically more variable at X-ray wavelengths than protostars in more evolved evolutionary stages, i.e., class II and class III young stellar objects.
The first tentative X-ray detection of two candidate class 0 sources (in OMC-3) was reported by Tsuboi et al. (2001). However, the identification as class 0 sources is ambiguous, Tsujimoto et al. (2004) concluding that in one case the X-ray emission is shock-induced by jets from a nearby class I source. The detection of an X-ray flare from IRS 7E, a candidate class 0 source in the Coronet cluster, was first presented by Hamaguchi et al. (2005) and is discussed further in this paper.
When it comes to physical explanations for the hard X-ray emission from class I protostars, often a magnetic interaction of the accretion disk and the stellar magnetosphere is put forward. In the model of Hayashi et al. (1996), closed magnetic loops connecting the central star and the disk are twisted by the rotation of the disk (see also e.g. Birk et al. 2000; Isobe et al. 2003; Shu et al. 1997). The loops expand with increasing twist until reconnection takes place, hot plasma giving rise to the X-ray and, conceivably, nonthermal radio emission. This mechanism predicts flare periodicity on the order of the rotation period of the innermost parts of the disk and is compatible with observed flare durations, i.e. several hours to days (e.g. Koyama et al. 1996). Montmerle et al. (2000), basing their conclusions on the occurence of quasi-periodic flares on YLW 15, propose an evolutionary scenario explaining X-ray properties of protostars as due to their rotation velocity. In this picture, young class I sources in which the rotation of star and disk is not yet synchronous, would constitute the strongest X-ray emitting population of star formation. This phase quickly comes to an end (after a few 105 years) and coronal activity becomes the dominant source for X-ray emission. Thus, the following part of the class I stage is already comparable to the class II phase with synchronous rotation, however with much more extinction due to circumstellar material than in the later evolutionary stages. Until now, with no further periodicities found, observational evidence for star-disk magnetic coupling in these objects is inconclusive (Montmerle 2003). However, Favata et al. (2005) find evidence for such coupling from the analysis of intense X-ray flares necessitating large magnetic structures.
Thermal as well as nonthermal radio emission has been observed towards pre-main sequence stars. Distinctive features of nonthermal radio emission include rapid variability, circular polarization, negative spectral indices and exceedingly high brightness temperatures (at high spatial resolution). Identified as gyrosynchrotron emission from mildly relativistic electrons gyrating in magnetic fields in this context, nonthermal radio emission is a direct tracer of circumstellar magnetic activity. Thermal radio emission is thought to be produced by ionized material e.g. at the base of outflows. The radio properties of young stellar objects are reviewed by e.g. André (1996); for a discussion of gyrosynchrotron radio emission from a young B star, see André et al. (1988).
A number of X-ray emitting classical and weak-line T Tauri stars are sources of nonthermal radio emission produced in large-scale magnetic structures in the coronae of these objects also giving rise to thermal X-ray emission from coronal plasmae. In particular, circularly polarized radio emission indicative of gyrosynchrotron emission has been observed towards several YSOs (e.g. Rodríguez et al. 1999; André et al. 1988; Phillips et al. 1993; Skinner 1993; André et al. 1992; White et al. 1992).
Class I protostars have mostly been observed to be thermal radio
sources. Shang et al. (2004) model the centimetric free-free emission of
class I protostars based on the X-wind model (Shu et al. 1997) with X-ray
emission as the main ionization source, and indeed some of the class I
sources showing X-ray emission are known radio sources as well (e.g.
Grosso et al. 1997; Smith et al. 1999; Feigelson et al. 1998). However, already at the base of such outflows,
the material becomes optically thick so that emission from the immediate
surroundings of the star - e.g. nonthermal radiation caused by magnetic
fields - can easily be hidden. Gibb (1999) proposes a scenario where only
the earliest and latest stages of star formation emit at radio
wavelengths: With fading accretion, the (thermal) jet emission decreases while
at the same time presumably a stellar magnetic field builds up,
giving rise to nonthermal emission.
|Figure 1: Optical (DSS2-Red, left) and near-infrared (K', right) views of the Coronet cluster. T CrA, seen at left, is not covered by the radio data presented here. It is marginally detected in X-rays (Skinner et al. 2004). The K' image is from Hodapp (1994), coordinates are J2000.|
|Open with DEXTER|
Among the three class I sources discussed here in detail, one (IRS 5) is observed to emit nonthermal radio emission (as reported earlier by Feigelson et al. 1998). This source is in fact the only class I protostar known to emit such radiation. Already its circular polarization is a clear sign, additionally its short-term variability is difficult or impossible to reconcile with thermal emission.
At present, little is known about the temporal correlation of the X-ray and radio emission, especially when nonthermal radio emission is concerned. Here, we study these phenomena separately, based on non-simultaneous observations. The only possibility to learn more about these sources consists of simultaneous multi-wavelength studies looking for correlations in the variability at different wavelengths, especially in the radio and X-ray regimes. Until now, only few systematic attempts of such observations of young stellar objects have been made: Feigelson et al. (1994) observed a magnetically active T Tauri star, Guenther et al. (2000) targeted pre-main sequence stars in the Taurus-Auriga region, specifically V773 Tau, and Gagné et al. (2004) observed pre-main sequence stars in the Oph cloud complex. However, only in the latter case two class I protostars were targeted but not detected at both radio and X-ray wavelengths. Simultaneous observations like these can be very valuable, as underlined by the serendipitous discovery of an enormous flare at radio and X-ray wavelengths, apparently from a highly obscured weak-line T Tauri star in Orion (Bower et al. 2003).
Table 1: VLA Coronet observation dates.
Table 2: Archival X-ray data used in this study.
The Corona Australis (CrA) molecular cloud complex is a nearby region with ongoing star formation, at a distance of only 150 pc, which is similar to that of the Taurus and Oph dark cloud complexes (Finkenzeller & Mundt 1984). Taylor & Storey (1984) found a compact cluster of infrared sources in the densest part of the cloud complex close to the Herbig Ae star R CrA, calling it the Coronet cluster. Figure 1 shows the DSS2-Red and K'-band views of its central region, clearly showing a number of deeply embedded sources.
The Coronet cluster is an association of young stellar objects mostly identified at radio as well as X-ray wavelengths. It is comprised of at least five deeply embedded class I protostars (Wilking et al. 1997,1992,1986), four of which are covered by the observations presented here. Koyama et al. (1996) reported the detection of hard X-ray emission towards these five protostars. Three of these sources could be verified at higher angular resolution by ROSAT (Neuhäuser & Preibisch 1997), more ROSAT-detected young stellar objects were reported by Neuhäuser et al. (2000). First results from Chandra-ACIS observations of this region were presented by Garmire & Garmire (2003) who detected more than 70 sources in a 20 ks exposure.
The Coronet cluster has been analyzed at radio wavelengths already several times. Brown (1987) carried out VLA observations at cm in 1985, identifying five sources associated with young stellar objects. Suters et al. (1996) analyze the radio variability of pre-main-sequence stars using archival VLA (1985-1987) and Australia Telescope (1992) data, while Feigelson et al. (1998) report on VLA observations at 3.5 cm, taken in 1997. They detect circular polarization toward the class I protostar IRS 5.
Studying dust emission from the region at mm, Chini et al. (2003) report several peaks, with the brightest one (called MMS 13 and already detected by Henning et al. 1994) in the vicinity of IRS 7 in the Coronet cluster. Millimeter emission is associated with IRS 1, 2 and 5 as well. They conclude that the MMS 13 peak probably is a class 0 source not related to any of the sources observed at other wavelengths. Based on millimeter continuum and molecular line data, Groppi et al. (2004) reached a similar conclusion. Nutter et al. (2005) present higher-resolution submillimeter continuum data resolving MMS 13 into two protostellar and one prestellar source where gravitational collapse has not yet started. This prestellar source is at the position of the class 0 source surmised earlier and is estimated to have a mass of 6 to 11 , while the envelopes of the protostars considered in this work all have sub-solar mass dust masses. The two protostellar sources are associated with unresolved radio and X-ray sources.
Using near-infrared spectroscopy, Nisini et al. (2005) constrain the accretion rates and evolutionary stages of some Coronet protostars. They conclude that IRS 1 and IRS 2 appear to be mostly accretion-powered (accounting for 80% to 65% of , respectively), while IRS 5 - a binary source - appears to be a deeply embedded, more evolved object with accretion accounting for only 20% of the bolometric luminosity of one of the components, IRS 5a. IRS 3 and IRS 6a show no signs of accretion. Comparing their derived stellar luminosities and effective temperatures to different models of pre-main sequence evolution, IRS 1 emerges as the youngest object with an age of about 105 years, while IRS 2, 5a and 6a apparently have about the same age, closer to 106 years in spite of their different accretion properties.
The VLA data were analyzed using NRAO's Astronomical Image Processing System (AIPS). An absolute flux density scale was established by observations of 3C286, the nearby phase calibrator 1924-292 was observed every ten minutes. At a declination of nearly , the Coronet cluster is observable with the VLA only at very low elevations (<), increasing problems with phase instabilities caused by the atmosphere. Here, however, by using self-calibration, these effects could be minimized. All observation epochs were phase-only self-calibrated, as the low flux densities did not warrant self-calibration of, both, phases plus amplitudes. Following this, Stokes-I and Stokes-V were imaged separately, with the input uv-data split into different time intervals. The polarization of the secondary calibrator 1924-292 is 1%. Source positions and fluxes were determined, using the AIPS task SAD, by fitting Gaussians to each detection. The light curves thus derived are shown in Fig. 3. The root mean square (rms) noise level in the Stokes I and V maps for the different epochs is shown in Table 1. For the detection of weak sources, the five AM596 observations were combined, yielding a map with an rms of <Jy. The synthesized beam size for this combined data is at position angle .
Chandra X-ray data were analyzed using the Chandra Interactive Analysis of Observations (CIAO) 3.1 software package together with CALDB 2.28. Apertures used were three or five (IRS 1, 2, and 5) arcseconds in diameter. Spectra were prepared with the task psextract, including the creation of response matrices and ancillary response files. Finally, spectra bin sizes were determined by a minimum event number per bin of fifteen.
XMM-Newton data were analyzed using the Science Analysis System (SAS) 6.0.0. Due to the superior signal-to-noise ratio among XMM imaging instruments, we only make use of the EPIC-PN data here. We generally used apertures with a diameter of 32 , but in cases of close neighbouring sources, we defined smaller (14 diameter) non-overlapping extraction regions. These apertures encircle approximately 66% and 40% of the point source flux, so that corresponding correction factors had to be taken into account in the luminosity determination. Spectroscopy involved the creation of response matrices and ancillary response files using the tasks rmfgen and arfgen, respectively. Finally, the spectra were binned with the same minimum event number criterion as was used for the Chandra data.
Models were fitted to all background-subtracted spectra with Sherpa from the CIAO 3.1 software package, using Monte Carlo multi-parameter fits. The spectral model consists of the Astrophysical Plasma Emission Code (APEC) with an absorption factor. This code calculates the emission spectrum from collisionally ionized diffuse gas. The main fit parameters are the plasma temperature, the absorbing hydrogen column density and the element abundances compared to solar values.
After a determination of elemental abundances from the Fe K line at 6.7 keV from one spectrum per source, an absorbed APEC emission model was fitted to the spectra. It turned out that the spectra can be well described with a single-temperature model; there was no need to assume two temperature components. However, due to the high absorbing column densities toward the sources, softer emission components might easily be extincted. The uncertainties of the best-fit parameter values were estimated with the Sherpa task "uncertainty''. Based on these fits, the count rates in the lightcurves were converted into unabsorbed flux using the PIMMS software (Mukai 1993) and finally, into luminosity units (energy range 0.5-10 keV), assuming a distance of d=150 pc.
After a general discussion of the 1998 VLA monitoring and the analysis of archival X-ray data, individual source properties will be summarized.
|Figure 2: Upper panel: image made from the combined uv data of all five VLA observation epochs. Contours shown are 5, 10 and 100 in the main plot and -3 (dashed), 3, 5, 50 and 100 in the inset showing IRS 7. Coordinates are J2000.0. Lower panel: the same view, showing the merged Chandra ACIS-I data. The inset shows approximately the same view in the merged XMM- Newton data (MOS+EPN, X1-X3).|
|Open with DEXTER|
In Stokes-I, eight sources can be detected throughout all five AM596
observation epochs, taking a 5
as the criterion for source existence. These sources are also detected
throughout the AK469 data with the exception of IRS 2, not detected
due to primary beam attenuation far from the observation's phase center.
An image created from the combined AM596 uv data is shown in Fig. 2.
Five weak additional sources only emerge from this
integration of the five AM596 observation epochs, including the T Tauri star IRS 6.
In Stokes-V only IRS 5 was detected, showing epoch-averaged circular polarization
degrees between 0% and 23%. IRS 5 at the same time is the most
spectacularly variable source in this dataset: Starting at a relatively high emission level,
but interestingly without detected circular polarization, the source dims, then rises sharply
towards two peaks at the end of the period covered, accompanied by changes in its polarization.
Analyzing the data in shorter intervals, the Stokes-V emission of
IRS 5 remains variable at timescales of 0.5 h while this is not the
case for any of the sources in Stokes-I.
The remaining sources show diverse degrees of variability, but to a lesser extent
(Fig. 3). All sources detected in single epochs have been
detected at radio wavelengths before. The source fluxes derived from the VLA
observations are summarized in Table 4,
together with the 3.5 cm fluxes from Feigelson et al. (1998) for comparison.
|Figure 3: Radio flux curves of sources identified in the 1998 VLA data. Error bars indicate the uncertainty estimates given by the AIPS task "sad'', they are based on actual noise () rather than the quality of the fit. Filled triangles denote upper limits. Also shown is Stokes-V for IRS 5.|
|Open with DEXTER|
At X-ray wavelengths, nine sources are clearly detected, including an unidentified weak source (No. 3) only detected by Chandra (see Tables 3 and 5). Figure 2 shows the combined Chandra data, with the combined XMM-Newton data shown as an inset. It is interesting to compare this X-ray view with the radio view in Fig. 2. The five weak radio sources only emerging in the complete VLA data all remain undetected in X-rays together with the Radio Sources 5 and 9 (Brown 1987). However, the class I protostar IRS 9 - clearly detected in X-rays, even with a flare in the C1 observation - remains undetected at radio wavelengths, together with the weak X-ray source No. 3 not showing any counterpart at other wavelengths. Comparing the count rates for these sources derived from the five X-ray datasets shows that there is considerable variability in many cases (Table 5). For the three class I protostars IRS 1, 2 and 5, this is corroborated by the luminosity curves (Fig. 7) and the corresponding spectra (Fig. 5). All three sources have X-ray luminosities on the order of several 1030 erg/s. IRS 5 emerges again as the most variable source.
Table 3: Positions and identification of Coronet radio and X-ray sources.
The elemental abundances for the three sources, as derived from the 6.7 keV Fe K line emission, are (IRS1, X1), (IRS2, X3) and (IRS5, X3) relative to solar abundances. There is no evidence for fluorescent emission of weakly ionized or neutral iron at energies below 6.7 keV. Fitting absorbed APEC emission models to the spectra leads to the results depicted in Fig. 6 and listed in Table 6. All three sources show high absorbing column densities (several 1022 cm-2) and hot plasma emission (several 107 K). There are signs of long-term variability in the absorbing column densities, especially towards IRS 2 and IRS 1. Note that the X2 and X3 observations are consecutive while a total range of about 2.5 years is covered. Tentatively, there is a trend for the emission being harder when the sources are brighter, although the situation is inconclusive.
Interestingly, though, the hydrogen column densities as derived from the X-ray spectra for IRS 1, 2, and 5 are at around half the values derived by Nisini et al. (2005) from near-infrared colours (see Table 7). In order to check the significance of this discrepancy, we tried to fit the spectra while freezing the absorbing column density to the value derived from NIR colours. This lead to very poor fits (large ) in all cases, demonstrating that the large values are inconsistent with the X-ray spectra. Additionally, by using a Monte Carlo fitting scheme, the parameter space was checked for multiple minima, but again no acceptable model was found for large values. The empirical relation is (Ryter 1996; Vuong et al. 2003). A similar discrepancy has been observed previously towards the young stellar objects L1551 IRS 5 (Bally et al. 2003), EC 95 (Preibisch 2003a), and towards SVS 16 (Preibisch 2003b). Possible solutions to this "extinction problem'' are configurations in which the protostar and its X-ray emission become detached, so that the absorbing column densities for the infrared and X-ray emission become different. Scenarios include X-ray emission from jet shocks close to the protostar, X-rays scattered towards the observer by circumstellar material, and huge coronal structures.
Table 4: Coronet sources, radio data.
Table 5: Coronet sources, X-ray-data.
Table 6: X-ray spectral fitting results for IRS 2, 1, 5, and R CrA: Absorbing column density and plasma temperature, with error margins.
For six of the young stellar objects studied here, it is possible to reliably determine X-ray luminosities based on their spectra. The radio and X-ray luminosities of these sources have been plotted into Fig. 8 together with an empirical relationship for active late-type stars from Benz & Guedel (1994). The underlying physics of this relation is that the acceleration of synchrotron-emitting electrons and the heating of the coronal plasma leading to X-ray emission are accomplished by the same mechanisms. The Coronet YSOs appear to be compatible with this relationship.
Comparing the X-ray luminosities to the bolometric luminosities (see Table 8), it turns out that IRS 2, 5, 1, and R CrA are all in the unsaturated regime (Fleming et al. 1995), but a lot more active than the sun ( ), IRS 5 being the most active source with a ratio of log ( .
IRS 2 does not show significant radio variability in any time interval covered by our observations (Fig. 3 and Table 4). The flux is, however, at about half the value determined by Feigelson et al. (1998). IRS 2 is one of the strongest X-ray sources among the objects studied here. The overall count rate in the C1 Chandra observation is twice as high as during the C2 observation which is clearly visible in the luminosity curves (Fig. 7). The XMM-Newton count rates are less diverse. Variability is seen also within observations, especially X1, showing a continuous increase, and X3, showing a sudden increase in the second half of the observation. Already the ROSAT data presented by Neuhäuser & Preibisch (1997) show variability on timescales of months. The X-ray spectra (Fig. 5), can be explained as the emission of hot plasma ( K), absorbed by a column density of cm-2. Apparently, the absorbing column density towards IRS 2 has continuously diminished over two and a half years, the 3 errorbars of the first and the last measurements are clearly separated (Fig. 6 and Table 6). This could be caused by a clumpy circumstellar medium, probably involving the circumstellar disk. Nisini et al. (2005) find that accretion accounts for 60% of this object's bolometric luminosity.
IRS 5 shows several flares in our VLA dataset (Figs. 3 and 4, Table 4) and has been observed flaring already once before at 4.9 GHz (Suters et al. 1996). In our data, the source starts at a relatively high flux, dims, and then sharply rises towards two peaks, separated by about nine days. This source was previously reported to emit nonthermal radio emission (Feigelson et al. 1998). Interestingly, the initial high flux of IRS 5 is accompanied by undetectable Stokes-V emission. Also when looking at the double peak towards the end of the period covered here, the Stokes-V flux appears to be largely uncorrelated to the Stokes-I flux: only in the second peak there is a rise in Stokes-V flux. Possibly this different behaviour is at least partly due to an averaging effect, given the faster variability in Stokes-V: contrary to the Stokes-I emission, Stokes-V in our data is variable down to timescales of 0.5h, a clear sign of nonthermal emission. In the epoch with the highest Stokes-V flux (R8), for example, Stokes-V emission was only detected in the first quarter. Two epochs earlier, at the sharp rise in Stokes-I, the polarization degree drops from 23% to 10% because only the Stokes-I emission increases. In the subsequent epoch, the polarization degree is back at 20% while the total flux increased even further. Feigelson et al. (1998) reported a polarization degree changing between 10% to 37% during a day.
Table 7: Comparison of column densities towards IRS 2, 5, and 1 derived from NIR and X-ray emission.
Table 8: Bolometric and X-ray luminosities (derived from X-ray spectra).
|Figure 4: IRS 5 epoch-averaged polarization degrees in percent. Errors are propagated from those shown in Fig. 3.|
|Open with DEXTER|
In X-rays, IRS 5 is the most variable object when compared to IRS 1
and IRS 2, within our datasets as well as on longer time scales. The
overall count rate in 2003 (C2) is nearly three times as high as in
2000 (C1). In the two consecutive observations X2 and X3, the X3 count rate
is nearly twice as high as the X2 count rate. The X3 luminosity curve
(Fig. 7) starts with an increase followed by a slight diminution in
emission towards the end of the observation. The high-luminosity points at the
beginning of the X2 luminosity curve remain unexplained.
The X-ray spectra (Fig. 5) show signs of high absorption
cm-2), the temperature of the emitting
plasma is at around
K. There appears to be no
significant temporal evolution in these fit parameters (Fig. 6 and Table 6).
Nisini et al. (2005) confirm earlier observations by Chen & Graham (1993), finding that IRS 5 is
a binary object with only 90 AU separation (scaled to d=150 pc). Thus, the attribution of
X-rays to any of the components, or both is difficult. For IRS 5a, they find that
accretion accounts for below 20% of the object's bolometric luminosity.
While this suggests a deeply embedded object in a later evolutionary stage,
Nisini et al. (2005), using standard evolutionary tracks, conclude that IRS 5a has about
the same age as IRS 2 (
Already Feigelson & Montmerle (1999) note that the radio properties of this source
rather resemble a weak-line T Tauri star. Due to the compact grouping
of protostars surrounding it, it remains unclear whether IRS 5
powers an outflow (e.g. Wang et al. 2004). The weak neighbouring radio
source (No. 5 in Tables 3 and 4) has
a very faint 2MASS counterpart within the nebulosity surrounding IRS 5.
|Figure 5: X-ray spectra of IRS 2, 5 and 1 derived from the five X-ray datasets. The line shows the result of fitting an absorbed APEC model.|
|Open with DEXTER|
|Figure 6: Results of fitting XSWABSXSAPEC, fit parameters and kT. Abundances set to values determined from best spectra. Errorbars are 3. The X-ray datasets are numbered according to Table 2. The NIR-determined column densities from Nisini et al. (2005) are cm-2 (IRS 1), cm-2 (IRS 2) and cm-2 (IRS 5).|
|Open with DEXTER|
IRS 1 was not found to be variable at radio frequencies of 4.9 GHz (Suters et al. 1996). In our data, this source does not
show significant radio variability either (Fig. 3 and Table 4). The observed flux is compatible with
the value measured by Feigelson et al. (1998). There are, however, clear signs
of X-ray variability in this relatively bright X-ray
source - within the datasets as well as on longer time scales.
The C2 count rate is 35% greater than the C1 count rate, and the
XMM-Newton count rates also suggest variability even though the X2 and X3 count rates are affected by a chip gap in the source region, which
causes a flux loss af about 20%. The X-ray luminosity curves (Fig. 7)
show only slight variability within the single observations.
The X-ray spectra (Fig. 5)
can be modeled by highly absorbed
emission of hot plasma
K). There is a significant
difference in the absorbing column densities derived from the X1 and C2 observations.
No such signs of differences in the fitted plasma temperatures are seen
(Fig. 6 and Table 6).
According to Wang et al. (2004), IRS 1 could be the powering source of four
Herbig-Haro objects. Nisini et al. (2005) find that 80% of this object's
bolometric luminosity is due to accretion (see also Nisini et al. 2004, who find
evidence for an accelerating wind in the vicinity of IRS 1).
|Figure 7: Kilosecond-binned, background-subtracted X-ray luminosity curves in temporal order for the three class I protostars IRS 2,5, and 1. In the X2 and X3 observations, around 20% of the flux from IRS 1 is lost due to a chip gap in the source region.|
|Open with DEXTER|
|Figure 8: X-ray vs. radio luminosity of identified young stellar objects for which X-ray luminosities can be derived from spectra. The errorbars correspond to the standard deviation within the 1998 VLA and archival X-ray data, except for IRS 7E/W, where the X-ray luminosities are from Hamaguchi et al. (2005). For IRS 7E, the X-ray errorbars reflect the flare seen in X2/X3 while for IRS 7W, only a single luminosity was determined from the X2 data. The line shows the empirical relationship for active late-type stars from Benz & Guedel (1994).|
|Open with DEXTER|
IRS 9 is detected as a weak X-ray source throughout all datasets studied here, shows an X-ray flare in the C1 Chandra data (not shown), however, was not detected at radio wavelengths (S<0.035 mJy). This is consistent with previous radio observations.
Radio Source 5, initially observed by Brown (1987), has not been detected at X-ray wavelengths. In our data, its radio intensity is variable on timescales of days to months (Fig. 3 and Table 4). Feigelson et al. (1998) report a radio flux on a similar scale. Variability of this source at radio frequencies (4.9 GHz) was also reported by Suters et al. (1996) who argue against an extragalactic origin of the emission. Given its faint infrared counterpart, this source could be a brown dwarf (Wilking et al. 1997; Feigelson et al. 1998). If confirmed, the radio data presented here together with the data analyzed by Suters et al. (1996) would constitute an interesting dataset of long-term variability of brown dwarf radio emission.
IRS 6, detected as a weak X-ray source by Chandra as well as XMM-Newton, was only found in the combined AM596 uv-data: its emission is very weak at 0.15 mJy. No radio emission from this source has been reported before. While this source is too weak for an analysis of short-timescale X-ray variability, the emission is somewhat weaker in the 2003 (C2) data when compared to the 2000 (C1) level. The X2 and X3 count rates for IRS 6 are affected by a chip gap in the source region. Wang et al. (2004), looking for optical outflows, argue that IRS 6 could be the driving source of four Herbig-Haro objects. Nisini et al. (2005) resolve this source into a binary object with a separation of only 112 AU (scaled to d=150 pc), making the attribution of X-rays difficult. However, they find no sign of ongoing accretion in the near-infrared towards IRS 6a. Thus, IRS 6 appears to contain at least one T Tauri star.
Two main sources can be detected in the IRS 7 region, at both radio and X-ray wavelengths, namely IRS 7E and IRS 7W. Another major radio source is located five arcseconds north of IRS 7W (Radio Source 9, as defined by Brown 1987). Three additional weak radio sources accompany IRS 7E/W.
There has been confusion about possible protostellar sources detected at different wavelengths around IRS 7, although some clarification was brought by Nutter et al. (2005), presenting submillimeter data where the single emission peak observed before at millimeter wavelengths (Groppi et al. 2004; Henning et al. 1994; Chini et al. 2003) is resolved into three subsources. A molecular outflow and a disk were detected in the IRS 7 region and analyzed by Anderson et al. (1997) and Groppi et al. (2004), although the latter showed that it is not yet possible to attribute these to any of the sources seen at higher angular resolution. IRS 7 is also the driving source of Herbig-Haro objects (Wang et al. 2004). Harju et al. (2001) find evidence for a jet emanating from IRS 7E at cm, using the ATCA radio interferometer. Choi & Tatematsu (2004) analyze the IRS 7 region at mm with the VLA. Interestingly, they only marginally detect IRS 7E while there is elongated emission around IRS 7W and Radio Source 9, although the connection remains unclear. The emerging picture is that of IRS 7W being an infrared-detected, deeply embedded protostar, probably a class I or II source, while IRS 7E could be a class 0 source although the source was detected at 4.8 m by Pontoppidan et al. (2003). This would be one of the first class 0 sources detected in X-rays.
IRS 7E is quite variable on all timescales covered in the 1998 VLA data (Fig. 3 and Table 4), at a flux level mostly below the value reported by Feigelson et al. (1998). The weak radio source detected in the eastern vicinity of IRS 7E (source 15) might be a sign for extended radio emission. The radio sources detected around IRS 7E/W are shown in an inset of Fig. 2.
IRS 7E is the most variable X-ray source among the objects studied here. While IRS 7E was only marginally detected in the C1, X1, and C2 observations, the 2003 (X2, X3) datasets show IRS 7E as a strong and very hard X-ray source. This flaring has been studied by Hamaguchi et al. (2005) who argue that this might be an X-ray detected class 0 source. They do not find any near-infrared counterpart down to a K-band magnitude of 19. However, Pontoppidan et al. (2003) detect both sources, IRS 7E/W, at 4.8 m using VLT-ISAAC, arguing that IRS 7E remained undetected at 10 m (Wilking et al. 1997) due to silicate absorption. It is tempting to attribute also the enormous flare observed by Koyama et al. (1996) to this source, although that remains unclear due to the limited angular resolution of their ASCA data. Since spectra of IRS 7E could only be taken from the XMM-Newton data, an estimated luminosity curve for this source was produced by assuming the same spectral characteristics, taken from Hamaguchi et al. (2005), of kT=3 keV (corresponding to a temperature of 35 MK, rather a lower limit) and cm-2 for all five datasets, then determining the fluxes and luminosities following the procedure outlined above. To minimize contamination, Hamaguchi et al. (2005) excluded IRS 7W from the IRS 7E source region in their analysis. The plot shows the enormous increase in luminosity (Fig. 9). In the two Chandra observations, only 19 and 13 - however highly energetic - photons were recorded from IRS 7E respectively. Given the high temperature of the plasma emission as well as the quick variability, especially when comparing the subsequent X2 and X3 datasets, it appears unlikely that the emission is due to an accretion shock. These characteristics rather point towards circumstellar magnetic activity.
IRS 7W is the brightest radio source in the sample considered here, mostly emitting at around 4.2 mJy, although in two epochs the source is detected at >5 mJy (Fig. 3 and Table 4). This source shows only moderate radio variability within the 1998 monitoring, especially on timescales of days. For Radio Source 9, Feigelson et al. (1998) report an 8.4 GHz flux density of 1.58 mJy, considerably higher than all fluxes measured in 1998 which are <1.1 mJy. Close to Radio Source 9, there are two weak radio sources (sources 10 and 13) which do not have counterparts at other wavelengths.
In the C1/C2 data, IRS 7W is detected as a weak X-ray source emitting energetic photons, similar to IRS 7E. Due to contamination caused by the flaring nearby source IRS 7E, a meaningful spectrum could only be extracted from the X2 dataset, leading to kT=4.7 keV (55 MK) and cm-2 (Hamaguchi et al. 2005).
|Figure 9: IRS 7E X-ray luminosity. Countrates for the five X-ray datasets were converted into luminosities without background subtraction, assuming a single-temperature spectrum of kT=3 keV and cm-2 (Hamaguchi et al. 2005). The errorbars denote errors of 30%. Note that the source is only marginally detected in C1 and C2.|
|Open with DEXTER|
R CrA is one of the optically brightest ( mag) objects in the CrA cloud and it illuminates the reflection nebulosity NGC 6729. The stellar parameters are somewhat uncertain, especially because the optical spectrum seems to be variable and spectral types ranging from B8 to F7 have been reported (e.g. Bibo et al. 1992). Here we adopt the values listed in Lorenzetti et al. (1999), i.e. spectral type A5e, , and mag. Comparison of these parameters to the pre-main sequence models of Palla & Stahler (1999) suggest a mass of and an age of slightly more than 1 Myr for R CrA.
The spectral energy distribution of R CrA rises very steeply between m and m (see, e.g. Acke & van den Ancker 2004), demonstrating the presence of large amounts of hot circumstellar material, presumably in the form of an optically thick circumstellar disk. This very large infrared excess suggests that R CrA is in a very early evolutionary state and thus we discuss it here together with the lower-mass protostars.
In the VLA data (Fig. 3 and Table 4), R CrA was detected as a weak and rather constant radio source with an average flux of 0.23 mJy, the same flux level as measured by Feigelson et al. (1998). However, R CrA remained undetected at radio wavelengths several times before: in September 1985 (VLA, 6 cm, 3 mJy, Brown 1987), in February 1990 (VLA, 3.6 cm, 3 mJy, Skinner et al. 1993) as well as with the VLA and the AT in 1985-87 and 1992, respectively (Suters et al. 1996). Even though R CrA is quite close to the respective 5 limits, this is suggestive of long-term variability.
|Figure 10: R CrA spectra, taken from an ellipse region avoiding the neighbouring source IRS 9. The results of fitting an absorbed APEC model are shown. In the X3 fit, the column density was frozen to the values from X1 and X2.|
|Open with DEXTER|
|Figure 11: R CrA X-ray luminosity curves from the same elliptical region also used for the spectra. Luminosity calculated assuming spectral properties of the best spectrum (X2). Data is kilosecond-binned and background-subtracted, datasets are in temporal order as labeled.|
|Open with DEXTER|
X-ray observations of the Coronet region with EINSTEIN and ROSAT did not detect X-ray emission from R CrA (Zinnecker & Preibisch 1994; Damiani et al. 1994; Walter et al. 1997; Neuhäuser & Preibisch 1997). Koyama et al. (1996) detected hard X-ray emission near the location of R CrA with ASCA. However, due to the poor spatial resolution of the ASCA data ( ) the contributions of the individual young stellar objects R CrA, IRS 7, and IRS 9 could not be resolved. Nevertheless, since the intensity peak in the ASCA image was located close to the position of R CrA, it appears likely that R CrA produced a significant fraction of the observed hard X-ray emission. During this ASCA observation a strong flare was seen. The X-ray spectrum was found to be very hard and could be fitted with a keV plasma model and an absorbing hydrogen column density of . During the flare, the spectrum showed a remarkable broadened or double emission line structure between 6.2 and 6.8 keV. However, it remained unclear which of the sources caused the flare.
X-ray emission from R CrA is clearly detected in all the Chandra and XMM-Newton data sets used in this paper. The detection in the Chandra observation C1 was already noted by Skinner et al. (2004). During both Chandra observations R CrA is only seen as a rather weak source, yielding no more than 100 counts, too few for detailed spectral analysis. During the first XMM-Newton observation the observed count rate for R CrA is well consistent with that in the Chandra observations, but during the XMM-Newton observations X2 and X3 the count rates were about a factor of two larger (see lightcurve in Fig. 11). The lightcurves show no indications of a flare, it appears rather as if the source has "switched'' to a higher activity level duringX2 and X3. A sudden drop of the count rate (by about a factor of two) is seen in the second half of observation X3. A possible interpretation for the general shape of the lightcurve would be the temporal presence of a dominant active region during X2 and X3. The sharp drop in X3 may then be explained by the occultation of such an active region when the stellar rotation moves it behind the stellar disk.
The X-ray spectra of R CrA extracted from observations X1, X2, and X3 show a very strong high-energy tail and a pronounced 6.7 keV emission line, providing clear evidence for very hot plasma. The spectral fits yield a plasma temperature of about 100 MK, the highest value of all sources in the Coronet region. The hydrogen column densities derived from the spectral fits are and correspond to an optical extinction of mag (Ryter 1996; Vuong et al. 2003). This is higher than the extinction estimate derived from the optical properties ( mag), but given the uncertainty of the optical parameters of R CrA, both values may well be in agreement.
The X-ray luminosity of R CrA derived from the spectral fits ranges from erg/s in the C2 observation to erg/s in the X2 observation. The fractional X-ray luminosity is thus .
The observed X-ray emission of R CrA is particularly interesting since, according to our current understanding of stellar X-ray emission mechanisms, one would not expect to see any X-ray emission emission from this star. Being an intermediate-mass A-type star, R CrA should neither possess a magnetically driven corona, which is the source of the X-ray emission from late type (F to M) stars, nor can it have a strong radiation driven stellar wind, in which shocks cause X-ray emission observed in the more massive O- and early B-type stars (see e.g. Favata & Micela 2003; Pallavicini 1989). The theoretical expectation that late B and A stars should not be instrinsic X-ray emitters is well confirmed in the case of main sequence stars (e.g. Cassinelli et al. 1994; Schmitt et al. 1993). However, it is still unclear whether the Herbig Ae/Be stars, i.e. the class of very young intermediate-mass pre-main sequence stars, to which R CrA belongs, also fit to this scenario. In EINSTEIN, ROSAT, ASCA, and Chandra X-ray observations of samples of Herbig Ae/Be stars (Zinnecker & Preibisch 1994; Skinner et al. 2004; Preibisch & Zinnecker 1996; Damiani et al. 1994; Hamaguchi et al. 2005) some 30-50% of the observed stars were detected as X-ray sources. Several possible explanations have been proposed for these findings (see Skinner et al. 2004; Preibisch & Zinnecker 1996; Caillault et al. 1994), but no final explanation has been found. There exist some ideas how an AeBe star may produce X-ray emission (e.g. the scenario of a nonsolar dynamo powered by rotational shear in very young intermediate mass stars; see Tout & Pringle 1995), but no really convincing evidence has yet been found for this. A possible explanation for the observed X-ray emission may be that the X-rays originate not from the AeBe stars themselves, but from close, unresolved, late type companions.
For our case of R CrA, we note that Takami et al. (2003) presented spectro-astrometric observations and found some evidence for the presence of a companion at a separation of about , i.e. only 10-15 AU. If such a companion actually exists, it is probably a lower-mass young stellar object, i.e. a T Tauri star or a low-mass protostar, which could easily explain the observed X-ray emission.
What remains quite remarkable, however, is the extremely high plasma temperature of 100 MK inferred from the X-ray spectra. Typical plasma temperatures in young stellar objects range from 10 MK up to 50 MK, reaching higher values usually only during large X-ray flares. From other X-ray observations of star forming regions it seems that class I protostars show systematically higher plasma temperatures than the more evolved T Tauri stars (see, e.g., Imanishi et al. 2001, for the case of Oph). This may be an indication that the companion of R CrA is a very young object (i.e. more likely a class I protostar than a T Tauri star). In this case, some (perhaps the largest) fraction of the infrared excess emission may come from the companion and not from the Ae star R CrA itself.
We presented the results of radio monitoring of protostars in the Coronet cluster together with archival X-ray data of these sources covering 154 ks. The main results are:
The National Radio Astronomy Observatory is a facility of the National Science Foundation operated under cooperative agreement by Associated Universities, Inc. This study is partly based on observations obtained with XMM-Newton, an ESA science mission with instruments and contributions directly funded by ESA Member States and NASA.