we reexamine the origin of the radio emission from these two components ... ern source is most likely the result of free-free radiation related to stellar winds.
The Astrophysical Journal, 657:916Y 924, 2007 March 10 # 2007. The American Astronomical Society. All rights reserved. Printed in U.S.A.
ON THE NATURE OF THE EXTENDED RADIO EMISSION SURROUNDING T TAURI SOUTH Laurent Loinard, Luis F. Rodrı´guez, Paola D’Alessio, Mo´nica I. Rodrı´guez,1 and Ricardo F. Gonza´lez Centro de Radiostronomı´a y Astrofı´sica, Universidad Nacional Auto´noma de Me´xico, Michoaca´n, Mexico Received 2003 December 17; accepted 2006 November 16
ABSTRACT At centimeter wavelengths, the young stellar system T Tauri is known to be composed of two sources, the northern one associated with the optical star T Tau itself, and the southern one related to the infrared companion T Tau S. Here we reexamine the origin of the radio emission from these two components using archival 2 cm, 3.6 cm, and 6 cm VLA observations. The emission from the northern member is confirmed to be largely dominated by free-free radiation from an ionized wind, while the southern radio source is confirmed to consist of a compact component of magnetic origin, surrounded by an extended halo. Only moderately variable, the extended structure associated with the southern source is most likely the result of free-free radiation related to stellar winds. However, its flat spectral energy distribution, its extent, and the lack of variation of its size with the frequency of observation are incompatible with the classical picture of a fully ionized wind with constant velocity and mass-loss rate leading to an electron density distribution of ne (r) / r2 . Instead, we propose a model in which the ionization results from the impact of a supersonic wind driven by T Tau Sb onto dense surrounding material, possibly associated with the circumbinary disk recently identified around the T Tau Sa /T Tau Sb pair. The timescales for cooling and recombination in such a situation are in good agreement with the observed morphological changes undergone by the extended structure as its driving source moves through the environment. Subject headings: binaries: general — ISM: jets and outflows — radiation mechanisms: general — radio continuum: stars — stars: formation
1. INTRODUCTION
nents of the relative motion strongly suggests that the orbit is nearly in the plane of the sky ( Ducheˆne et al. 2006 and references therein). Infrared observations have shown that T Tau Sb has the infrared spectrum of a normal but very obscured (AV 15) preYmain-sequence M1 star, whereas the spectrum of T Tau Sa is generally featureless, with the exception of Br in emission ( Ducheˆne et al. 2002, 2005). On the basis of a detailed analysis of near- and mid-infrared observations and of orbital motions, Ducheˆne et al. (2005, 2006) convincingly argued that T Tau Sa is an intermediate-mass (2.5Y3 M) young star surrounded by a small nearly edge-on disk (see also Schaefer et al. 2006). In addition to this circumstellar disk, T Tau Sa and T Tau Sb are apparently surrounded by an edge-on circumbinary torus seen as an absorption ultraviolet feature ( Walter et al. 2003), which is responsible for the large extinction toward T Tau S. The exact size of this structure is still poorly constrained, but it is likely on the order of 1 00 ; 0:5 00 (Walter et al. 2003; Ducheˆne et al. 2005). To be stable against the orbital motion of the Sa/Sb system, the inner radius of this structure must be larger than about 0.35 00 (Ducheˆne et al. 2005). Two outflows have been identified around T Tau (Solf et al. 1988; Bo¨hm & Solf 1994; Solf & Bo¨hm 1999). One is oriented roughly east-west ( P:A: 65 ) and points almost exactly to the observer (i 80 ), while the other is approximately northsouth ( P:A: 15 ) and almost exactly in the plane of the sky (i 10 ). The western side is the most prominent part of the east-west jet; it terminates as a series of Herbig-Haro objects ( known as HH 155) in NGC 1555 ( Hind’s Nebula), about 3500 west of T Tau. The north-south jet also defines a series of HH knots, including Burnham’s Nebula, 800 south of T Tau, which is generically known as HH 255, and that apparently connects at larger scale with the giant flow HH 355 ( Reipurth et al. 1997). Solf & Bo¨hm (1999) showed that each side of these two flows produces well-defined spatiokinematical structures, and they
1.1. Optical and Infrared Observations Ever since it was first identified as the prototype of a new class of variable stars (Joy 1945), T Tauri has been the subject of intense scrutiny. It is now known to be a 2 M preYmainsequence star of spectral type K1, with a bolometric luminosity of 8 L (Koresko et al. 1997). T Tau appears to be surrounded by a 4 ; 103 M accretion disk whose plane is apparently inclined by 20 Y40 with respect to the plane of the sky (Akeson et al. 2002), and it appears to suffer about 1.5 mag of visual extinction ( Koresko et al. 1997) caused by material along the line of sight. High-resolution near-infrared observations ( Dyck et al. 1982) revealed early that it had a companion (hereafter T Tau S) located 0.700 to its south, whose nature remains somewhat mysterious. T Tau S has about twice the bolometric luminosity of the optical star T Tau (which we will refer to as T Tau N in the rest of the paper) and is very significantly more obscured (e.g., Koresko et al. 1997; Stapelfeldt et al. 1998); indeed, its spectral energy distribution peaks around 3 m (Ghez et al. 1991). The relative motion between T Tau N and T Tau S (Ghez et al. 1991; Roddier et al. 2000; Ducheˆne et al. 2002, 2005) strongly suggests that they are physically bound to each other. In addition, the comparison between the transverse (5 km s1; Beck et al. 2004) and radial (2.5 km s1; Ducheˆne et al. 2005) components of the relative orbital motion shows that the orbit is close to being in the plane of the sky (i 30 ) and is, therefore, very likely to be roughly coplanar with the circumstellar disk surrounding T Tau N. Recently, T Tau S was itself found to be composed of two infrared sources ( T Tau Sa and T Tau Sb; Koresko 2000; Ko¨hler et al. 2000) in rapid relative motion ( Ducheˆne et al. 2002). Here too, the comparison between the transverse and radial compo1
Currently at Space Telescope Science Institute, Baltimore, MD.
916
917
RADIO EMISSION FROM T TAU S
orientation, which is almost along the line of sight, of the jet powered by T Tau N. The circumbinary disk around T Tau Sa / T Tau Sb and the circumstellar disk around T Tau Sa, on the other hand, appear to be almost exactly perpendicular to the plane of the sky, in agreement with the orientation of the jet powered by T Tau Sa, which is almost exactly in the plane of the sky. In this situation, it is very likely that the disk around T Tau Sb is also either in the plane of the sky or perpendicular to it. Its jet will accordingly be directed either along the line of sight or perpendicular to it. 1.2. Radio Observations
Fig. 1.—Sketch of the T Tauri system.
attribute the east-west flow to T Tau itself and the north-south one to the infrared companion T Tau S. The recent discovery of a nearly edge-on disk around T Tau Sa (Ducheˆne et al. 2005) strongly suggests that the flow originating from T Tau S and oriented north-south in the plane of the sky is powered by T Tau Sa. If this is correct, then T Tau Sb would be the only one of the three known members of the T Tau system to have no clearly identified associated jet, although it most probably does power one, since it is actively accreting. That this jet has not yet been identified may plausibly result from the overall complexity of the region surrounding T Tau S. The summary presented above (see Fig. 1 for a sketch of the region) shows that there are two ‘‘privileged’’ planes in the T Tau system. The T Tau N/T Tau S and T Tau Sa / T Tau Sb orbits, as well as the circumstellar disk around T Tau N, appear to be almost in the plane of the sky. This is consistent with the
At radio frequencies, T Tau is a double source separated by about 0.700 in the north-south direction (Schwartz et al. 1986). While the northern radio source corresponds to the optical star T Tau, the southern component is clearly associated with the infrared companion. Detailed comparisons between radio and infrared data obtained almost simultaneously in late 2000/early 2001 suggest that it is associated with T Tau Sb ( Loinard et al. 2003; Johnston et al. 2003). However, the exact relation between the infrared sources and the radio emission remains disputed. Furlan et al. (2003), for instance, have suggested that the southern radio source is not associated with any of the known infrared objects, but instead traces a fourth stellar source in the system. Johnston et al. (2004a, 2004b), on the other hand, argued that the southern radio source, although associated with T Tau Sb, is not coincident with it. To avoid confusion, we refer to the southern radio source as T Tau Sr (for T Tau S radio) in the rest of this paper. As we will see below, because of the composite nature of the southern radio source in T Tau, it is in any case convenient to use different names for the radio and infrared sources. The northern radio source ( T Tau N ) has a positive spectral index, as would be expected for free-free emission from a dense stellar wind. However, it is also moderately variable and was reported to exhibit strong circular polarization at various epochs (Johnston et al. 2003), two properties that suggest the existence of a component associated with an active magnetosphere. The southern radio source (T Tau Sr), on the other hand, is strongly variable on timescales from hours to years (Skinner & Brown 1994; Phillips et al. 1993; Johnston et al. 2003; Smith et al. 2003), has consistently exhibited strong circular polarization (Johnston et al. 2003), and has a nearly flat spectral energy distribution (Johnston et al. 2003). While pure gyrosynchrotron emission would be a natural emission mechanism if the source was unresolved, the presence of an extended halo of radio emission (White 2000; Johnston et al. 2003) suggests the existence of an additional component. Indeed, very long baseline interferometry ( VLBI ) observations, which effectively filter out any extended component, only recover about 35% of the total 3.6 cm flux of T Tau Sr (Smith et al. 2003), showing that emission associated with an active magnetosphere only accounts for part of the total radio flux. This conclusion was further confirmed by multiepoch Very Long Baseline Array ( VLBA) observations ( Loinard et al. 2005; L. Loinard et al. 2007, in preparation) which also showed that the motion of the compact, nonthermal radio source associated with T Tau Sr is essentially identical to that of the infrared source T Tau Sb ( Loinard et al. 2005). This demonstrates that the compact radio component can indeed be identified with T Tau Sb. The origin of the extended radio halo in T Tau Sr and its relation with the infrared sources, however, remains poorly understood. In an attempt to better constrain the emission mechanisms at work in the two radio sources in T Tau, and particularly in the extended component of T Tau Sr, we reanalyze here archival
918
LOINARD ET AL.
Vol. 657
TABLE 1 List of Observations and Results for ROBUST = 0 Synthesized Beam T Tau N Date of Observation (1) 2001 1988 1988 1988 1989 1989 1990 1990 1990 1992 1995 1998 1998 2001 2003 1998 1998
Jan 19 (2001.053)..................... Oct 30 (1988.831) .................... Nov 12 (1988.866) ................... Nov 15 (1988.874) ................... Feb 3 (1989.093) ...................... Feb 5 (1989.098) ...................... Mar 14 (1990.200) ................... Apr 3 (1990.257)...................... Apr 6 (1990.263)...................... Nov 17 (1992.880) ................... Jul 14 (1995.534) ..................... Mar 6 (1998.178) ..................... Mar 22 (1998.222) ................... Jan 20 (2001.052)..................... Jun 23 (2003.477) .................... Mar 6 (1998.178) ..................... Mar 22 (1998.222) ...................
k (cm) (2) 2 3.6 3.6 3.6 3.6 3.6 3.6 3.6 3.6 3.6 3.6 3.6 3.6 3.6 3.6 6 6
T Tau S
max ; min (arcsec ; arcsec) (3)
P.A. (deg) (4)
rms (5)
Peak (6)
Integrated (7)
Peak (8)
Integrated (9)
; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ;
82 89 +8 52 +73 +68 46 +45 +42 7 33 6 +52 +82 +90 1 +40
40 23 16 29 22 41 16 18 35 29 42 32 20 24 13 27 19
1.24 0.64 0.55 0.49 0.60 0.60 0.64 0.60 0.66 0.63 0.57 0.75 0.57 0.68 0.48 0.57 0.47
1.93 0.86 0.87 0.74 1.09 0.94 1.07 0.76 0.95 0.75 0.90 1.04 0.94 1.05 0.72 0.60 0.54
2.32 8.35 2.53 2.54 3.04 3.29 3.02 6.52 5.38 1.69 3.41 2.34 2.54 2.54 3.27 2.25 2.39
5.43 10.06 4.36 4.16 4.98 5.18 4.90 8.10 7.61 3.10 5.76 4.57 4.67 4.15 5.04 3.71 3.86
0.13 0.23 0.20 0.22 0.23 0.24 0.21 0.21 0.21 0.23 0.23 0.22 0.21 0.26 0.23 0.36 0.36
0.12 0.21 0.20 0.20 0.21 0.21 0.20 0.20 0.20 0.21 0.23 0.20 0.21 0.22 0.21 0.33 0.34
Notes.—The rms noise level given in col. (5) is in units of Jy beam1. The peak intensities in cols. (6 ) and (8) are in units of mJy beam1 and are affected by typical uncertainties of 0.08 mJy beam1. The integrated flux densities in cols. (7) and (9) are in units of mJy and are affected by typical uncertainties of 0.12 mJy.
VLA observations obtained in the last two decades at 2 cm, 3.6 cm, and 6 cm. 2. OBSERVATIONS In this article, we make use primarily of a series of 14 archival 3.6 cm observations obtained between 1988 and 2003 ( Table 1) with the Very Large Array (VLA) of the National Radio Astronomy Observatory (NRAO).2 This data set corresponds to all the existing VLA observations of T Tau at 3.6 cm, obtained in the most extended (A) configuration of the VLA, with an integration time on source of more than about 1 hr. It is complemented by a deep 2 cm observation from 2001 and two 6 cm observations from 1998, also obtained in the A configuration of the VLA. All observations were recorded in both circular polarizations, with an effective bandwidth of 100 MHz. The calibration was made within AIPS following the standard procedure in use at the VLA. The flux density scale was established using observations of the primary flux calibrators 3C 48 and 3C 286, obtained at the beginning or end of each session. The resulting visibilities were imaged and CLEANed using the IMAGR algorithm, also within AIPS. To obtain relatively high angular resolutions without losing much sensitivity, all observations were initially restored using a weight intermediate between uniform and natural ( ROBUST parameter set to 0 in AIPS). The resulting angular resolution (see Table 1) is on the order of 0.1300 for the 2 cm data, 0.2200 for the 3.6 cm data, and 0.3400 for the 6 cm observations. Self-calibration was applied to the 3.6 cm and 6 cm data in order to produce maps with increased dynamical ranges, but not to the 2 cm data, which have a somewhat lower signal-tonoise ratio. The 2001 and 2003 data were collected while Pie Town, a VLBA antenna that can be connected to the VLA to improve the 2
NRAO is a facility of the National Science Foundation operated under cooperative agreement by Associated Universities, Inc.
angular resolution, was also in use (Johnston et al. 2003). To obtain a homogeneous data set, this additional antenna was ignored in the processing of the data as reported in Table 1. However, to allow an easy comparison at relatively high angular resolution, the 3.6 cm observations from 2001 were also imaged including the Pie Town antenna, and with ROBUST ¼ 0, while the 2 cm data are imaged without Pie Town, but with natural weighting ( ROBUST ¼ 5). With this processing, the restored beams are on the order of 0.2000 for both data sets. Indeed, the restoration process was eventually forced to produce circular synthesized beams with a FWHM of 0.200 ( Table 2). 3. RESULTS 3.1. Northern Radio Source As reported in the past (e.g., Johnston et al. 2003; Skinner & Brown 1994), the northern radio source of the T Tau system shows relatively little evidence of temporal variability. The present series of 3.6 cm observations amply confirms this result ( Table 1); averaged over the 14 observations, we find the integrated flux density to be 0:91 0:12 mJy, and we find the two extrema to be within 0.19 mJy of this mean. This implies that the temporal variability at 3.6 cm never exceeds 20%. The average value of the spectral index of T Tau N obtained using TABLE 2 T Tau Sr 2 cm and 3.6 cm Observations from 2001.052 at 0.200 Angular Resolution Compact Component
Extended Component
k (cm)
Flux Density (mJy)
Flux Density (mJy)
max ; min (arcsec ; arcsec)
P.A. (deg)
2................ 3.6.............
2.82 0.02 2.07 0.02
2.23 0.20 2.20 0.12
0.51 ; 0.36 0.53 ; 0.30
+65 +47
No. 2, 2007
RADIO EMISSION FROM T TAU S
the data published by Johnston et al. (2003; see their Table 2) is on the order of 1.0. Using the 2 cm and 3.6 cm data obtained in 2001 ( Table 1), we get a value of 1:1 0:1, whereas we get 1:0 0:1 when using the 3.6 cm and 6 cm data obtained in 1998 (also in Table 1). Combined, the lack of strong temporal variability and this value of the spectral index suggest that the emission mechanism is free-free radiation in a dense ionized wind. A spectral index of 0.6 is expected for a fully ionized stellar wind with constant velocity and mass-loss rate ( Panagia & Felli 1975), but higher values are expected if, for example, the electron density drops faster than r 2 ( Wright & Barlow 1975). At all epochs and wavelengths, the numerical value of the peak intensity of T Tau N is found to be smaller than the numerical value of its integrated intensity (Table 1). This provides a good indication that the centimeter emission from T Tau N is somewhat resolved, since equal numerical values are expected for a point source. In the 2001 2 cm data restored with ROBUST ¼ 0 (the option that provides the best compromise between angular resolution and sensitivity), the emission is found to be elongated in a direction at a position angle of 62 5 , which is remarkably similar to the position angle of the east-west flow powered by T Tau (65 ; Bo¨hm & Solf 1994). This strongly suggests that the radio emission from T Tau N traces the base of the large-scale east-west flow driven by T Tau, and it confirms that this flow is indeed driven by T Tau N itself. The possibility that the entire radio flux from T Tau N would result from free-free emission associated with stellar winds appears to be seriously challenged, however, by the detection by Johnston et al. (2003) of significant levels of circular polarization. To explain such a polarized component, a magnetically driven process is required, so the radio emission from T Tau N would have to be the superposition of two components. In the 2 cm data obtained in 2001, we do indeed find about 25% of right circular polarization. However, for all other epochs and wavelengths, we find the level of circular polarization to be smaller than about 5%Y10%. This would suggest that the magnetic activity was relatively somewhat higher during the 2001 observation than at all other epochs. Interestingly, however, neither the total flux density nor the spectral index appear to be significantly different in 2001 from their values at other times. Consequently, the magnetically driven process required to explain the polarization in the 2 cm data observed in 2001 must strongly affect the polarization level without contributing much to the integrated fluxes. The most natural way to obtain such a behavior is through a radiation process that intrinsically produces levels of polarization on the order of 100%. As in T Tau Sr (Smith et al. 2003; see below), an electron cyclotron maser associated with accretion funnels onto a highly magnetized central star might be at work. Indeed, the magnetic field at the surface of the optical star T Tau is known to be a few kG (Guenther et al. 1999), as would be required for this mechanism to produce emission at centimeter wavelengths. The lack of polarization in the 3.6 cm observations of 2001 would be naturally explained in this scheme, since cyclotron masers are expected to affect only relatively limited frequency ranges. 3.2. Southern Radio Source The radio emission from T Tau Sr has long been known to be strongly variable on timescales from hours ( Phillips et al. 1993; Smith et al. 2003) to years (Johnston et al. 2003), and significantly circularly polarized (Skinner & Brown 1994; Phillips et al. 1993; Johnston et al. 2003). Only a few emission mechanisms, all magnetically driven, are able to produce large degrees of circular polarization. Gyrosynchrotron radiation from
919
mildly relativistic electrons is usually considered the most attractive mechanism because it naturally produces polarization over a wide range of frequency. For reasonable values and geometries of the magnetic field, it is expected to produce degrees of circular polarization of up to about 50% (Dulk 1985). Two coherent emission processes could also be invoked: electron cyclotron masers and plasma masers. These two mechanisms can create even larger values of the circular polarization (indeed, up to 100%), but they are expected to affect only rather specific frequencies (respectively, the cyclotron frequency c and the plasma frequency p, or their first harmonics), although a varying angle between the line of sight and the emitting structures might broaden the frequency range over which strong polarization is produced. The existence of a magnetically driven component associated with T Tau Sr was further confirmed by Smith et al. (2003), who, using VLBI observations at 3.6 cm obtained in 1999 December, detected a compact (diameter < 15 R), time-variable, circularly polarized radio source at the position of T Tau Sr. The radio emission from this compact source appears to be the superposition of a relatively steady component, which exhibits fairly modest levels of circular polarization, and sporadic bursts, which appear on timescales of hours and are characterized by extremely high levels of circular polarization (up to 100%). As pointed out by Smith et al. (2003), the most natural process able to explain these extreme levels of circular polarization is an electron cyclotron maser associated with accretion funnels onto a highly magnetized star. The relatively steady component, on the other hand, is most naturally explained as gyrosynchrotron radiation associated with magnetic recombination loops in an active magnetosphere. Interestingly, the VLBI observations presented by Smith et al. (2003) recovered only about 35% of the total 3.6 cm flux density measured simultaneously with the VLA. Hence, a major fraction of the radio flux density emitted by T Tau Sr must emanate from an extended component that was filtered out in the VLBI observations. The existence of an extended halo of radio emission associated with T Tau Sr has indeed been noted in the past (e.g., White 2000; Johnston et al. 2003 and references therein), although it was never discussed in detail. Its existence can be confirmed by comparing the VLA observations reported here with the 12 VLBA observations of T Tau obtained by Loinard et al. (2005; L. Loinard et al. 2007, in preparation) at 3.6 cm between 2003 September and 2005 September. Indeed, the mean value of the flux density recovered in the VLBA observations is only about 1.7 mJy (L. Loinard et al. 2007, in preparation), whereas the 3.6 cm VLA flux density has consistently been about 4Y5 mJy in the last decade (Table 1). Since the magnetically driven radio component will remain totally unresolved at the VLA, the total 3.6 cm emission can be seen as the superposition of a point source and an extended component. To characterize the extended component, we fitted each 3.6 cm image with the sum of two two-dimensional Gaussian ellipsoids, one forced to have the size and orientation of the synthesized beam, and the other left to have entirely free parameters. Clearly, the true structure of the extended emission is unlikely to be a perfect Gaussian ellipsoid; a more complex structure would undoubtedly be revealed with higher resolution observations. However, at the present angular resolution, Gaussian ellipsoids appear to provide reasonably good fits to the data, and they allow us to characterize the position and structure of the extended component. The results of these fits are given in Table 3, where we report the flux density of the compact and extended components for each epoch, as well as the size and orientation of the extended component (cols. [4] and [5]) and the offset from the compact
920
LOINARD ET AL.
Vol. 657
TABLE 3 Properties of the Compact and Extended Components in T Tau Sr at 3.6 cm Extended Component Angular Size
Compact Component
Date of Observation (1) 1988.831............................................. 1888.866............................................. 1988.874............................................. 1989.093............................................. 1989.098............................................. 1990.200............................................. 1990.257............................................. 1990.263............................................. 1992.880............................................. 1995.534............................................. 1998.178............................................. 1998.222............................................. 2001.052............................................. 2003.477.............................................
Flux Density (mJy) (2) 7.57 1.63 1.75 2.33 2.62 2.38 6.00 4.47 1.25 2.75 1.94 2.20 2.15 2.98
0.02 0.01 0.02 0.02 0.03 0.01 0.01 0.02 0.02 0.03 0.02 0.01 0.02 0.01
Flux Density (mJy) (3) 2.33 2.69 2.46 2.66 2.47 2.44 2.00 3.18 1.90 3.00 2.43 2.32 2.13 1.91
component to the center of the extended component (cols. [6] and [7]). An immediate conclusion that can be drawn from this table is that the compact component is indeed very variable, having a minimum flux density of 1.25 mJy in 1992, compared to a maximum of 7.6 mJy (6 times higher) in 1988. The average flux density obtained considering the entire 15 yr time span covered by the 3.6 cm observations reported here is about 3 mJy. In the last decade (1992Y2003), it has remained fairly steady at around 2.2 mJy. Interestingly, this is somewhat higher than the 1.7 mJy mean flux density recovered in the 12 recent VLBA observations reported by L. Loinard et al. (2007, in preparation). This discrepancy could have several causes. In principle, it could be that because of its erratic temporal variability, the source just happened to be brighter during the VLA observations than during the VLBA ones. That possibility does not appear to be particularly likely, however, because the fairly large number of observations considered here should limit such stochastic effects. An alternative possibility is that our two-dimensional Gaussian fits systematically attribute to the compact source some of the emission that actually belongs to the extended structure. Finally, the discrepancy may reflect the existence of an additional component at an intermediate spatial scale (0.0500 Y0.1500 ) that would be filtered out by the VLBA but would appear pointlike to the VLA. A natural candidate for such a structure would be a compact thermal jet. In the absence of observations with a resolution intermediate between those of the VLA and VLBA, it is impossible to decide which of these possibilities is correct. The extended component is found to be somewhat variable, although much less so than the compact one. Averaged over the 14 epochs, we get a mean integrated flux of 2:42 0:36 mJy for this structure, and a maximum deviation from that mean of 0.76 mJy. Thus, the temporal variability of the extended component is about 30%. According to Smith et al. (2003), the difference between the total flux density of T Tau Sr measured at the VLA and the flux density of the compact component measured simultaneously using their VLBI data (in principle a direct measure of the flux density of the extended component) has remained constant at about 4.5 mJy during the time span cov-
0.08 0.06 0.10 0.09 0.15 0.06 0.07 0.12 0.14 0.19 0.10 0.06 0.11 0.07
Offset from Compact Component
max ; min (arcsec ; arcsec) (4)
P.A. (deg) (5)
(arcsec) (6)
P.A. (deg) (7)
; ; ; ; ; ; ; ; ; ; ; ; ; ;
+70 +72 +71 +74 +36 +75 +77 +83 +79 +30 +35 +17 +51 +54
0.038 0.052 0.061 0.064 0.072 0.051 0.060 0.027 0.047 0.063 0.196 0.193 0.118 0.060
+55 80 74 89 81 80 74 +84 22 +15 +62 +59 +33 +7
0.44 0.30 0.29 0.35 0.34 0.34 0.33 0.35 0.47 0.47 0.33 0.34 0.54 0.52
0.16 0.23 0.26 0.29 0.30 0.27 0.28 0.24 0.34 0.34 0.19 0.22 0.32 0.37
ered by their VLA/VLBI observations. Interestingly, this is about 2 mJy more than the flux density found here. Part (perhaps about 25%) of this difference clearly reflects the fact mentioned in the previous paragraph that the flux density recovered by the VLBI observations is systematically smaller than the VLA flux attributed to the compact component by our two-dimensional Gaussian fits. The remainder of the difference can be traced to an extended, very diffuse component, which the B-configuration observations of Smith et al. (2003) recovered better than our A-configuration data. We note, indeed, that when the 3.6 cm data considered in the present article are restored with natural weighting, and when relatively large regions centered on T Tau are considered, total flux densities that are typically 2 mJy larger than those reported in Table 1 are obtained for the entire system ( T Tau N + T Tau Sr). This very diffuse component is distributed over a region about 1 arcsec 2 in size, and its surface brightness is much weaker than that of both the compact and extended components detected in the observations presented here. It is, therefore, most certainly a distinct physical entity. Since it is poorly reconstructed with the present A-configuration data, this very diffuse component will not be discussed further in this paper. The spectral indices of the compact and extended components identified in the present A-configuration data can be estimated by applying the same decomposition in two two-dimensional Gaussians to the 2 cm and 3.6 cm observations obtained in 2001 when they are restored with similar beams, as described in the last paragraph of x 2 (see Table 2). The resulting spectral indices are 0:0 0:2 for the extended component and 0:5 0:1 for the compact one. Also, while (as mentioned before) the compact component exhibits a significant degree of circular polarization, the extended structure does not show any signs of polarization. An additional piece of information related to the structure of T Tau Sr can be obtained from the last two columns of Table 3. From 1988 to 1995, the offset between the center of the extended structure and the position of the compact radio component has remained rather modest (on average, about 50 mas) and was only marginally significant. However, in 1998 and 2001, the offset increased dramatically to 100Y200 mas, well above our uncertainties of about 20 mas (see also Fig. 1). Interestingly, this change in
No. 2, 2007
921
RADIO EMISSION FROM T TAU S
Fig. 2.—Images of the T Tau system at 3.6 cm, obtained from 1992 to 2003. The first contour is at 0.25 mJy beam1, and the contour interval is 0.2 mJy beam1. All images have been registered so as to have T Tau N at the origin of the coordinates. The crosses indicate the positions of the compact component and the center of the extended structure.
offset is accompanied by a notable change in morphology. While from 1988 to 1995, the 3.6 cm emitting structure remained fairly circular, it became quite elongated in 1998 and 2001 (Table 3; Fig. 2). In the last epoch considered here (2003), the offset between the compact and extended components has returned to a typical pre-1998 value (60 mas), and the extended source is again rather circular. Note, finally, that the extended component is most likely driven by T Tau Sb, since it ‘‘follows’’ it. While it was centered about 60 mas to the east of T Tau N in the early 1990s, it was about 140 mas to its west in 2003; this is very similar to what is observed for the compact radio source and the infrared star T Tau Sb with which we associate it. In contrast, during the same time range, T Tau Sa is expected to have moved by only about 100 mas ( Ducheˆne et al. 2006). The extended component is, therefore, most certainly associated with T Tau Sb and is unrelated to T Tau Sa. 4. DISCUSSION OF THE EXTENDED COMPONENT IN T TAU Sr As reported in the previous section, the extended component in T Tau Sr appears to be moderately variable, exhibits no signs of polarization, and has a flat spectral energy distribution. All these characteristics suggest optically thin free-free emission. To explain the observed flux density (2.5 mJy) given the distance to T Tau (149 pc; L. Loinard et al. 2007, in preparation), and if we assume an electron temperature of 10 4 K, a flux of ionizing photons of larger than 4 ; 1042 s1 would be required if the ionization were to be maintained through photoionization. A B4 zero-age main sequence (ZAMS) star would be needed to produce such a flux of ionizing photons. However, a B4 star has a luminosity of 500 L (Thompson 1984), much larger than the observed luminosity of T Tau S (11 L; Koresko et al. 1997). Indeed, since the stellar source associated with the compact radio component in T Tau Sr is the infrared source T Tau Sb, the only available star type is an M star, with an exceedingly weak ionizing flux. Thus, the ionization has to be provided by other means, most likely through shocks induced by supersonic stellar winds. However, the classical picture of a fully ionized, supersonic stellar wind with constant velocity and mass-loss rate, leading to an electron density distribution of ne (r) / r2 , as described by Panagia & Felli (1975), does not provide a satisfactory interpretation either. In the first place, in this scheme the angular size of the emitting structure should decrease with frequency roughly as 0.7, whereas we find the emitting region to have the same
size at 2 cm and 3.6 cm. In the second place, the observed radio flux densities and sizes imply brightness temperatures of only about 500 K, much lower than the value of about 104 K traditionally expected for optically thick ionized gas. For a temperature of 10 4 K, a 2.5 mJy source should be much more compact (0.0700 instead of the observed 0.300 ). Finally, a dense wind should have a spectral index on the order of 0.6, rather than 0 as observed. Thus, although the ionization must come from shocks, the ionization mechanisms and the exact processes leading to the production of the extended free-free radiation are not a priori obvious. According to Loinard et al. (2003), the compact radio source directly associated with T Tau Sb has accelerated westward to about 20 km s1, starting around 1996, after being on a relatively slow (5Y10 km s1) elliptical orbit from 1983 to 1995. That both the offset from the compact source to the center of the extended component and the morphology of this extended component have changed just when this acceleration occurred is unlikely to be a coincidence. It is, indeed, all the more unlikely that both the morphology and the offset have returned to typical pre-1998 characteristics in 2003, when the velocity of the compact component has also returned to a more modest value (Johnston et al. 2004b). This further points to T Tau Sb as the driver of the extended component and suggests that the extended component reacts with a certain delay to a change of position of T Tau Sb. An estimate of this delay can be obtained by dividing the average offset between the compact source and the center of the extended component in 1998Y2001 (typically 160 mas ¼ 24 AU) by the average space velocity of the compact radio source at these times (20 km s1, averaged from 1995 to 2003; see L. Loinard et al. 2007, in preparation). The resulting timescale is about 6 yr. Since our proposal is that the radio emission from the extended component is free-free radiation produced by shock-ionized material, we must compare this timescale to that expected for full recombination of the material. In the case of ionization through shocks produced by winds with velocities of a few hundred km s1, the ionized gas reaches temperatures of the order of 105Y10 6 K (e.g., Ghavamian & Hartigan 1998). Since the gas cannot recombine efficiently at such high temperatures, it must first cool to about 104 K. According to Spitzer (1978), the cooling function at T ¼ 104 Y106 K is approximately constant at 3 ; 1022 n 2 ergs s1 cm3, where n is the number density. This yields a cooling time from 106 K to 104 K of about cool
22;000 yr: n
922
LOINARD ET AL.
Vol. 657
Fig. 3.— (a) Flux and (b) spectral index for different models of internal winds: a ¼ 1:5 (solid lines), a ¼ 2 (dashed lines), and a ¼ 2:5 (dotted lines). In each case, a model with no mass-loss rate change (b ¼ 1) and one with a mass-loss rate increase of a factor of 2 (b ¼ 2) have been considered. The models with b ¼ 2 always produce slightly higher fluxes and spectral indices. The spectral indices for b ¼ 1 and b ¼ 2 are so similar that they do not appear as two separate curves.
Once at 104 K, the gas will recombine in a typical timescale of (Osterbrock 1989) recomb
100;000 yr: n
For a density of n ¼ 2 ; 104 cm3, cool 1 yr, and recomb 5 yr, in good agreement with the measured timescale of our process. Hence, in our proposal, the ionized gas emitting the bulk of the free-free radiation must have a density of the order of 2 ; 104 cm3. A more detailed analysis in which the dependence of the cooling time on the wind speed is taken into account (Ghavamian & Hartigan 1998; Gonza´lez 2002) gives a similar estimate for the required density. T Tau Sb is known to have an accretion rate of 9:2 ; 108 M yr1 ( Ducheˆne et al. 2005), so its wind must carry a mass-loss rate of about 108 M yr1. The wind speed, on the other hand, must be typical of low-mass T Tauri stars: 300 km s1. With these parameters, it is possible to estimate the wind density n at a distance R from the central star, assuming a biconical wind of aperture max (here max is measured from the axis of symmetry of the wind; the total aperture is thus 2max ). For each polar cap of such a wind, ˙ ¼ 2(1 cos max ) R2 Vnmp : M
ð1Þ
Here mp is the mass of the proton, and is the mean molecular weight (1.4 for neutral gas). For an isotropic wind, the radius at which the density is 2 ; 104 cm3 is R ¼ 6 AU. For a strongly collimated wind with max ¼ 10 , however, the radius at which the density is 2 ; 104 cm3 is R ¼ 52 AU. Such values are in reasonable agreement with the size scale of the extended component: 0:35 00 ¼ 50 AU (Table 3). To generate free-free emission on the correct size scales, the wind powered by T Tau Sb must produce strong shocks at several tens of AU from the driving source, and we still have to elucidate the very nature of these shocks. In principle, they could either be internal to the wind or result from the impact of the wind onto circumstellar or circumbinary material. The former possibility follows from the analysis of time-variable winds presented by Gonza´lez & Canto´ (2002). If the velocity of the
wind powered by a young star suddenly increases from an initial value v0 to a final faster value av0 (with a > 1), then a propagating working surface (two-wave structure; Raga et al. 1990) is created that can produce significant free-free emission. The change in velocity may be accompanied by a modification in ˙ 0 to a final value the mass-loss rate, from an initial value of M ˙ 0 . The intensity of the emission generated by this mechof bM anism depends on the initial and final velocities, as well as on the initial and final mass-loss rates. To test the plausibility of this mechanism as the origin of the extended radio emission seen in T Tau Sr, we ran six different models, three with no change in mass-loss rate (b ¼ 1) and with velocity increase factors of a ¼ 1:5, 2, and 2.5, and three with an increase in the massloss rate of a factor of b ¼ 2 and the same velocity increase factors. These models ( Fig. 3) predict that the free-free emission will become optically thin very rapidly after the velocity increases (this would be in agreement with the observed spectral index of the free-free emission in the halo of T Tau Sr), but provide insufficient flux even for fairly extreme cases (0.9 mJy for a ¼ 2:5 and b ¼ 2, against the observed value of 2Y3 mJy). Moreover, for a ¼ 2:5 and b ¼ 2, the internal shocks generated by the velocity change travel at about 600 km s1. At that velocity, it takes only a few months for the traveling working surface to reach a distance of 50 AU (0.3500 ), comparable to the size of the extended structure in T Tau Sr. Thus, in that scheme, one would expect T Tau Sr to undergo significant morphological changes on very short timescales. As described earlier, however, both the morphology and the flux density of the extended structure in T Tau Sr are only moderately time-variable, and the variability occurs on somewhat longer timescales. An a priori viable alternative compatible with internal shocks would be if the wind powered by T Tau Sb had undergone a succession of somewhat more moderate (e.g., a ¼ 1:5 and b ¼ 1) acceleration events. Each of these events would generate a flux density of about 0.02 mJy ( Fig. 3), so more than 100 of them would have to be present at any given time to explain the total 2Y3 mJy flux of the extended component in T Tau Sr. For a ¼ 1:5 and b ¼ 1, the working surface travels at about 400 km s1 and would take about a year to reach a distance of 50 AU. Thus, for this scheme to provide a reasonable interpretation of the overall extended radio emission in T Tau Sr, one would need the wind
No. 2, 2007
RADIO EMISSION FROM T TAU S
speed to change significantly with a periodicity of days to weeks, much shorter than the orbital period of the T Tau Sa/T Tau Sb system. It is quite unclear what mechanism(s) could produce such rapid wind parameter variability. The alternative possibility is that the shocks are produced when the wind powered by T Tau Sb interacts with circumstellar or circumbinary material. Given the substantial amount of material associated with T Tau S, it is not unexpected that the winds powered by T Tau Sb would encounter dense clumps of gas. We mentioned earlier that, given the general morphology of the system, the wind powered by T Tau Sb is likely to be either almost exactly in the plane of the sky or almost exactly along the line of sight. The former possibility is perhaps the least likely, since winds exactly in the plane of the sky are the most easily identified. In the latter case, the wind would naturally collide with the circumbinary structure surrounding the T Tau Sa / T Tau Sb binary. If that happened, the physical size of the radio source produced would be about 2 R sin max , where R ¼ 50 AU is the distance from T Tau Sb to the inner radius of the torus. Seen from the Earth, this structure would have an angular size of ¼ (2R /d )max . Given the mean deconvolved size of the extended radio structure in T Tau Sr (0.2400 ; Table 3), an opening angle of max 20 would be required. It is interesting to note that for max ¼ 20 and the wind parameters used before, equation (1) gives n ¼ 2:1 ; 104 at R ¼ 50 AU, almost exactly the value needed for the gas to recombine on a timescale compatible with the observations. Also, given the observed properties of the circumbinary disk (inner radius of 0.3500 , outer radius 0.500 , opacity AV ¼ 15 mag), its mean density must be in excess of 107 cm3. This value is so much larger than the density of the wind itself that the circumbinary structure would act as a ‘‘wall’’ for the wind, and the wind would not easily punch a hole through the torus, especially because the orbital motion of T Tau Sb around T Tau Sa would change the position of the impact with time. In summary, while photodissociation, ionized winds as per Panagia & Felli (1975), and internal shocks related to timevariable winds cannot easily explain the properties of the extended structure in T Tau Sr, the impact of a wind powered by T Tau Sb with dense gas located within 50 AU provides a natural explanation for it. Interestingly, such an interaction would naturally occur if the wind driven by T Tau Sb were oriented along
923
the line of sight, one of the two orientations ‘‘permitted’’ by the overall structure of the T Tau system. 5. CONCLUSIONS AND PERSPECTIVES The present analysis of archival radio observations of the T Tau multiple stellar system confirmed that the radio emission from the northern member of the system is largely dominated by free-free radiation from an ionized wind at a position angle between 60 and 65 , comparable to that of the large-scale eastwest flow driven by T Tau. A nonthermal component may also be occasionally present in T Tau N. The southern radio source ( T Tau Sr), on the other hand, is composed of at least two components. One is compact and produces time-variable, circularly polarized radio emission. As pointed out in the past (e.g., Phillips et al. 1993; Johnston et al. 2003; Smith et al. 2003), this implies a magnetic origin. The other component is extended, unpolarized, only moderately variable, and with a spectral index typical of optically thin free-free radiation. It is presumably produced by the interaction of a supersonic stellar wind driven by T Tau Sb and dense surrounding material, possibly associated with the circumbinary structure recently identified around the T Tau Sa/ T Tau Sb pair. It could be argued that the existence of this extended halo of radio emission might affect the astrometries reported by Johnston et al. (2003) and Loinard et al. (2003). A detailed analysis of this issue will be presented in a forthcoming article. Here we merely mention (1) that the existence of this structure may indeed explain some of the discrepancies between radio (particularly at the longest wavelengths) and infrared data noted by various authors and (2) that the effect on the 2 cm data used by Johnston et al. (2003) and Loinard et al. (2003) appears to be limited, thanks first to the positive spectral index of the compact component compared to the flat energy distribution of the extended one; and second to the better resolution and more effective spatial filtering at 2 cm than at 3.6 cm.
We acknowledge the financial support of DGAPA, UNAM, and CONACyT, Me´xico.
REFERENCES Akeson, R. L., Ciardi, D. R., van Belle, G. T., & Creech-Eakman, M. J. 2002, Johnston, K. J., Gaume, R. A., Fey, A. L., de Vegt, C., & Claussen, M. J. 2003, ApJ, 566, 1124 AJ, 125, 858 Beck, T. L., Schaefer, G. H., Simon, M., Prato, L., Stoesz, J. A., & Howell, R. R. Joy, A. H. 1945, ApJ, 102, 168 2004, ApJ, 614, 235 Ko¨hler, R., Kasper, M., & Herbst, T. M. 2000, in IAU Symp. 200, Birth and Bo¨hm, K.-H., & Solf, J. 1994, ApJ, 430, 277 Evolution of Binary Stars, ed. B. Reipurth & H. Zinnecker ( Potsdam: AsDucheˆne, G., Beust, H., Adjali, F., Konopacky, Q. M., & Ghez, A. M. 2006, trophys. Inst. Potsdam), 63 A&A, 457, L9 Koresko, C. D. 2000, ApJ, 531, L147 Ducheˆne, G., Ghez, A. M., & McCabe, C. 2002, ApJ, 568, 771 Koresko, C. D., Herbst, T. M., & Leinert, Ch. 1997, ApJ, 480, 741 Ducheˆne, G., Ghez, A. M., McCabe, C., & Ceccarelli, C. 2005, ApJ, 628, 832 Loinard, L., Mioduszewski, A. J., Rodrı´guez, L. F., Gonza´lez, R. A., Rodrı´guez, Dulk, G. A. 1985, ARA&A, 23, 169 M. I., & Torres, R. M. 2005, ApJ, 619, L179 Dyck, H. M., Simon, T., & Zuckerman, B. 1982, ApJ, 255, L103 Loinard, L., Rodrı´guez, L. F., & Rodrı´guez, M. I. 2003, ApJ, 587, L47 Furlan, E., Forrest, W. J., Watson, D. M., Uchida, K. I., Brandl, B. R., Keller, L. D., Osterbrock, D. E. 1989, Astrophysics of Gaseous Nebulae and Active Galactic & Herter, T. L. 2003, ApJ, 596, L87 Nuclei ( Mill Valley: University Science Books) Ghavamian, P., & Hartigan, P. 1998, ApJ, 501, 687 Panagia, N., & Felli, M. 1975, A&A, 39, 1 Ghez, A. M., Neugebauer, G., Gorham, P. W, Haniff, C. A., Kulkarni, S. R., Phillips, R. B., Lonsdale, C. J., & Feigelson, E. D. 1993, ApJ, 403, L43 Matthews, K., Koresko, C., & Beckwith, S. 1991, AJ, 102, 2066 Raga, A., Canto´, J., Binette, L., & Calvet, N. 1990, ApJ, 364, 601 Gonza´lez, R. F. 2002, Ph.D. thesis, Univ. Nac. Auto´noma Me´xico Reipurth, B., Bally, J., & Devine, D. 1997, AJ, 114, 2708 Gonza´lez, R. F., & Canto´, J. 2002, ApJ, 580, 459 Roddier, F., Roddier, C., Brandner, W., Charissoux, D., Ve´ran, J.-P., & Courbin, Guenther, E. W., Lehmann, H., Emerson, J. P., & Staude, J. 1999, A&A, 341, F. 2000, in Birth and Evolution of Binary Stars, Poster Proc. of IAU Symp. 768 200, ed. B. Reipurth & H. Zinnecker ( Potsdam: Astrophys. Inst. Potsdam), 60 Johnston, K. J., Fey, A. L., Gaume, R. A., Claussen, M. J., & Hummel, C. A. Schaefer, G. H., Simon, M., Beck, T. L., Nelan, E., & Prato, L. 2006, AJ, 132, 2004a, AJ, 128, 822 2618 Johnston, K. J., Fey, A. L., Gaume, R. A., Hummel, C. A., Garrington, S., Schwartz, P. R., Simon, T., & Campbell, R. 1986, ApJ, 303, 233 Muxlow, T., & Thomasson, P. 2004b, ApJ, 604, L65 Skinner, S. L., & Brown, A. 1994, AJ, 107, 1461
924
LOINARD ET AL.
Smith, K., Pestalozzi, M., Gu¨del, M., Conway, J., & Benz, A. O. 2003, A&A, 406, 957 Solf, J., & Bo¨hm, K.-H. 1999, ApJ, 523, 709 Solf, J., Bo¨hm, K.-H., & Raga, A. 1988, ApJ, 334, 229 Spitzer, L., Jr. 1978, Physical Processes in the Interstellar Medium ( New York: Wiley)
Stapelfeldt, K. R., et al. 1998, ApJ, 508, 736 Thompson, R. I. 1984, ApJ, 283, 165 Walter, F. M., et al. 2003, AJ, 126, 3076 White, S. M. 2000, in Radio Interferometry: The Saga and the Science, ed. D. G. Finley & W. M. Goss ( Washington, D.C.: Associated Universities, Inc.), 86 Wright, A. E., & Barlow, M. J. 1975, MNRAS, 170, 41