CONFIRMATION OF A RECENT BIPOLAR EJECTION IN THE VERY YOUNG HIERARCHICAL MULTIPLE SYSTEM IRAS 16293–2422

In the two new 3.6 cm observations, component B is at its expected position, with its usual morphology (Figure 1) and flux (Section 2). The structure of component A, however, has changed significantly since early 2006—when the last VLA observation prior to those presented here was obtained (Figure 1(c)). In the observation obtained in mid-2007 (Figure 1(d)), component A of IRAS 16293–2422 appears to contain four radio sources (see particularly the zoom on this region shown in Figure 2). Although somewhat blended with the rest of the emission, A1 is still clearly discernable to the east of the system (Figure 2). A radio source is also visible again at the expected position of A2. Note that this was not the case in the 2006.11 observation, where the emission associated with A2 was blended with that of A2α. The two additional sources in the system (indeed, the two brightest ones) are located on each side of A2, and we identify them with A2α and A2β. They are clearly not at the same positions as in the 2006.11 observation. Instead, they have moved away from A2 in a roughly symmetrical manner, A2α toward the southwest, and A2β toward the northeast. In the observation obtained at the end of 2008, component A is (again, but fortuitously) composed of three sources (Figures 1(e) and 2). Source A2 is still clearly identified, while source A2α has moved further from A2 toward the southwest. Component A2β has also kept moving away from A2 (toward the northeast), but now appears to be blended with A1 (Figure 2).

Zoom In Zoom Out Reset image size

To further investigate the nature and properties of A2α and A2β, it is interesting to compare their relative positions at the various epochs. In 2006.11, A2α and A2β were separated by 0166 ± 0003 at a P.A. of 62° ± 2°. In the 2007.62 image, however, the separation has increased to 0455 ± 0011, but the P.A. has not changed significantly: it is now measured to be 61° ± 3°. From the change in their separation, we can estimate the velocity at which A2α and A2β are moving away from one another to be 109 ± 4 km s⁻¹. The errors quoted here and later in the paper only account for the positional uncertainties of the ejecta; they do not include the errors on the distance d to the source. If the velocity since ejection has been constant at the value estimated above, the ejection must have occurred 0.87 ± 0.04 years before the 2006.11 observation (or 2.38 ± 0.11 years before the 2007.62 observation), i.e., in 2005.24 ± 0.04. This is—as expected—between the epochs of the 0.7 and 1.3 cm observations shown in Figures 1 and 2, but only very shortly after the 0.7 cm data were gathered.

Since A2 and A2α are well separated in the last two 3.6 cm observations, one can also consider the evolution of the relative position between these two sources. In 2007.62., they were separated by 0300 ± 0015 at a P.A. of 49°± 4°, whereas in the 2008.95 observation, the separation was 0494 ± 0014 and the P.A. was 61°± 3°. Thus, A2α appears to be moving away from A2 at 82 ± 9 km s⁻¹. Assuming again that the velocity has not changed appreciably, the ejection must have occurred 3.41 ± 0.37 years before the 2008.95 observation, i.e., in 2005.54 ± 0.37. This is in good agreement with the date estimated above from the relative motion between A2α and A2β, suggesting that the assumption of constant velocity is reasonable. Note that the relative velocity between A2 and A2α derived here is somewhat larger than half of the velocity between A2α and A2β calculated above. This shows that A2α is moving away from A2 somewhat faster than A2β.⁸ Indeed, one can estimate the relative velocity between A2β and A2 combining the present result and the relative velocity between A2α and A2β calculated earlier. We obtained (assuming again constant velocities) 26 ± 10 km s⁻¹. Thus A2β appears to move away from A2 about three times faster than A2α. Bipolar ejections usually produce somewhat more symmetric patterns. It should be mentioned, however, that the northeast and southwest lobes of the molecular outflow driven by A2 have long been known to be very asymmetric. For instance, SiO emission is very strong in the direction of the northeast lobe and nearly absent toward the southwest counterpart (e.g., Castets et al. 2001; Hirano et al. 2001). Since SiO is a good tracer of shocks between jets and circumstellar material, this most likely indicates that the region to the southwest of component A contains relatively little dense gas capable of decelerating the ejecta.

In summary, A2α and A2β appear to behave kinematically exactly as would be expected if they were ejecta from A2: they are moving (in projection) at 30–80 km s⁻¹ away from A2 along the direction (P.A. ∼ 60°) of the outflow known to be powered by A2. The true velocity of the jet must be of the order of the escape velocity from A2. As we will see in Section 4, A2 is likely to be a ∼1.5 M_☉ protostar. The radius at which jets are launched is usually believed to be a few stellar radii (∼3 R_*), and very young stars are a few times larger than their main-sequence counterparts of the same mass. It is, therefore, reasonable to assume that the radius of the protostar associated with A2 is about 3 R_☉, and that the escape velocity should be calculated at ∼10 R_☉. Under these assumptions, we obtain V_esc ≈ 240 km s⁻¹. To obtain a rough estimate of the orientation of the jet, we assume that this value provides a reasonable estimate of the true current velocity of the jet (this would require, in particular, that the jet has suffered little deceleration since it was launched). The projected velocity of the ejecta is only 30–80 km s⁻¹, so the jet powered by A2 must be oriented along a direction only 10°–15° from the line of sight. We conclude that A2 drives a flow oriented almost along the line of sight, and that A2α and A2β are ejecta along that flow. Episodic bipolar mass ejections are known to occur in young stars (e.g., Marti et al. 1995). To our knowledge, this is the first time, however, that an ejection is actually observed from the very beginning: we seem to have witnessed the very birth of a Herbig–Haro (HH) pair.

The centimeter emission produced by winds and ejecta from low-mass stars is thought to be nearly entirely of free–free origin (e.g., Anglada 1995; Shang et al. 2004). As detached clumps, A2α and A2β are likely less dense than the so-called thermal jets associated with the central regions of winds driven by young stars. As a consequence, the free–free emission from A2α and A2β is likely to be optically thin. In the absence of simultaneous multi-frequency observations (see Section 2), it is somewhat hazardous to estimate their spectral index and ascertain the characteristics of the emission. We note, however, that our data are fully consistent with optically thin free–free emission. The source A2β was well resolved in the 2006.11 1.3 cm data and in the 2007.62 3.6 cm observations. The spectral index derived from these two observations is α = −0.09 ± 0.05, in excellent agreement with the expected value (α = −0.1) for optically thin free–free emission.

To further constrain the properties of the ejecta, we will concentrate on A2α, because it is well resolved from the other sources in both of our 3.6 cm data sets. Within the errors, the 3.6 cm flux of A2α does not appear to have changed much between the two observations (0.62 ± 0.09 mJy in 2007.62 and 0.93 ± 0.10 in 2008.95—Table 2). The angular size of the emission (deconvolved from the primary beam) was found to be 021 × 008 in the 2008.95 data, whereas the emission was only resolved in one direction in the 2007.62 observations. In the resolved dimension, the angular size was 014, whereas in the other direction, the emission came from a region smaller than 017. Thus, the mean angular size of the emission was about 014 in 2008.95 and less than about 015 in 2007.62.

Table 2. Source Positions and Fluxes

Epoch	ν	Source	α (J2000.0)	δ (J2000.0)	Fν
	(GHz)		16h32m	−24°28'	(mJy)
2003.65	8.46	A1	228823 ± 00004	36301 ± 0011	0.81 ± 0.08
		A2	228544 ± 00002	36348 ± 0007	1.35 ± 0.08
2006.11	22.46	A1	228829 ± 00002	36373 ± 0004	0.78 ± 0.09
		A2β	228527 ± 00001	36438 ± 0004	0.64 ± 0.05
		A2+A2α	228594 ± 00006	36398 ± 0006	3.58 ± 0.27
2007.62	8.46	A1	228799 ± 00006	36393 ± 0013	0.50 ± 0.08
		A2β	228657 ± 00003	36371 ± 0009	0.70 ± 0.07
		A2	228531 ± 00004	36394 ± 0011	0.60 ± 0.07
		A2α	228366 ± 00006	36592 ± 0015	0.62 ± 0.09
2008.95	8.46	A1+A2β	228803 ± 00004	36414 ± 0009	0.93 ± 0.09
		A2	228548 ± 00008	36481 ± 0014	1.26 ± 0.13
		A2α	228232 ± 00005	36723 ± 0011	0.93 ± 0.10

Download table as:  ASCIITypeset image

Assuming optically thin free–free emission, the mass of ionized gas can be calculated from the radio flux as (e.g., Rodríguez et al. 1980)

$$\begin{eqnarray} {M_i \over M_\odot } &=& 3.39 \times 10^{-5} \left({S_\nu \over 1 \;\mbox{mJy}} \right)^{0.5} \left({\nu \over 1 \;\mbox{GHz}} \right)^{0.05} \left({T_e \over 10^4 \;\mbox{K}} \right)^{0.175}\nonumber\\ &&\times \left({\theta \over 1^{\prime \prime }} \right)^{1.5} \left({d \ \over 1 \;\mbox{kpc}} \right)^{2.5}. \end{eqnarray} \tag{ 1 }$$

Using the numbers above and T_e = 10⁴ K, we obtain, for A2α, M_i ≈ (1.00 ± 0.05) × 10⁻⁸ M_☉ using the 2008.95 observations (once again, the quoted uncertainty does not include the errors on the distance to the source), and M_i ≲ 0.9 × 10⁻⁸ M_☉ using the 2007.62 data. Assuming that both ejecta have similar masses, the bipolar ejection event reported here corresponds to a total mass of about 2 × 10⁻⁸ M_☉.

From the observed radio flux, one can also calculate the electron density n_e of the ejecta (e.g., Rodríguez et al. 1980):

$$\begin{eqnarray} {n_e \over 1 \;\mbox{cm}^{-3}} &=& 7.8 \times 10^{3} \left({S_\nu \over 1 \;\mbox{mJy}} \right)^{0.5} \left({\nu \over 1 \;\mbox{GHz}} \right)^{0.05} \left({T_e \over 10^4 \;\mbox{K}} \right)^{0.175}\nonumber\\ &&\times \left({\theta \over 1^{\prime \prime }} \right)^{-1.5} \left({d \ \over 1 \;\mbox{kpc}} \right)^{-0.5}. \end{eqnarray} \tag{ 2 }$$

With the observed parameters of the emission, we get n_e = 4.4 × 10⁵ cm⁻³ for A2α. Interestingly, the recombination timescale at that density is about 6 months. Since the ejecta have remained ionized at least since 2006.11 (when they were first detected) and most certainly since their creation around 2005.3 (see above), some mechanism must provide energy to keep them ionized. The most likely candidates are shocks either with the surrounding medium or internal to the jets. To remain ionized, the ejecta require an ionization rate of ∼10⁴² s⁻¹. Assuming that 13.6 eV of energy are required ionization, a power of ∼2 × 10³¹ erg s⁻¹ is needed. If this power is produced by the kinetic energy of the ejecta, we expect that the ejecta should decelerate at a rate of about 12.5 km s⁻¹ yr⁻¹ (for an initial velocity of 240 km s⁻¹ and a mass of 10⁻⁸ M_☉). The true deceleration would be somewhat smaller if the ejecta were only partially ionized as seems to be commonly the case (e.g., Podio et al. 2009). In any case, since the jet appears to be only 10°–15° from the line of sight, the projected deceleration would only be 2–3 km s⁻¹ yr⁻¹ and would be undetectable with the existing observations.

Detached clumps have long been known to exist in the jets driven by young stars. They can be created in (at least) two ways. One possibility is if the driving source experiences episodic increases in its mass loss rate. This hypothesis is often the preferred one to explain the presence of symmetric pairs of well-defined HH knots in evolved outflows (e.g., Arce & Goodman 2002). An alternative possibility is if the mass loss rate remains constant, but the ejection velocity increases abruptly from an initial value v_i to a final (larger) one v_f. In this situation, a working surface where material accumulates is created at the interface between the two winds. This naturally creates a gas condensation which can eventually become a detached clump (Masciadri & Raga 2001). Both of these mechanisms are plausible scenarios for the creation of A2α and A2β, and it would be interesting to be able to distinguish between them.

Ejection and accretion are believed to be intimately linked in protostars, so if A2α and A2β were created during an episode of increased mass loss, one would expect that increased accretion would also have been occurring. Assuming that the mass ejection rate is about 10 times smaller than the accretion rate (e.g., Hartmann & Kenyon 1996), then the total mass accreted during this episode must have been about 2 × 10⁻⁷ M_☉. Very young protostars derive much of their luminosity from accretion, so one could wonder if such an episode of increased accretion might have produced a detectable increase in the total luminosity of IRAS 16293–2422. To try and answer that question, one must characterize the timescale of the ejection/accretion episode.

From Figure 2, it is clear that by 2007.62, the ejecta had become detached from A2. As a consequence, a conservative upper limit on the timescale of the ejection event is 2.34 years, the time elapsed between the ejection and the 2007.62 observation (Section 3.1). The corresponding lower limit on the accretion rate during the event would be $\dot{M}_{\rm min}$ ≈ 8.5 × 10⁻⁸ M_☉ yr⁻¹, and the corresponding lower limit on the excess accretion luminosity L_acc,min ≈ 1.3 L_☉ (here we have assumed that all the accretion energy is released on the stellar surface and that the stellar radius is 3 R_☉). The bolometric luminosity of IRAS 16293–2422 is about 15 L_☉ (Section 1), so the increased accretion should have produced a ∼10% increase in the total luminosity of the source during the assumed 2.34 year duration of the event.

Of course, the duration of the event might have been significantly shorter than 2.34 years. An alternative way of estimating the characteristic timescale is the following. The ejecta are currently about 014 (≡ 17 AU) across (they were likely smaller in the past since A2β appeared to be unresolved in the 1.3 cm observations obtained in 2006.11). A clump of that size moving at ∼240 km s⁻¹ would become fully detached from its ejecting star in about 1.06 × 10⁷ s (≡ 0.34 years, just about 4 months). If this provides a good estimate of the true duration of the ejection event, then the mass accretion rate during the event would have been $\dot{M}$ ≈ 6 × 10⁻⁷ M_☉ yr⁻¹, and the corresponding excess accretion luminosity L_acc ≈ 9 L_☉. This would have produced a 60% increase in the total luminosity of the source during the 4 month duration of the event.

We conclude that if the creation of the A2α/A2β pair is the result of an increase in the mass accretion/ejection rates of component A2, their birth should have been accompanied by a 10%–60% increase in the total luminosity of IRAS 16293–2422. Most of the luminosity of IRAS 16293–2422 is radiated in the far-infrared and submillimeter parts of the electromagnetic spectrum, so the corresponding luminosity increase should be sought there. In particular, there have been several submillimeter observations of IRAS 16293–2422 obtained with the Sub-Millimeter Array (SMA) in the last few years (T. Bourke 2009, private communication). Note that IRAS 16293–2422 is a multiple system (Section 1) and that the only member of the system which should have experienced an increase in luminosity is component A2. While components A1 and A2 are not resolved in SMA observations, component A is well resolved from component B (e.g., Chandler et al. 2005). Moreover, components A and B happen to have similar submillimeter fluxes (e.g., Chandler et al. 2005). As a consequence, the total submillimeter luminosity increase for component A in SMA images ought to be at least 20%–120% and should be detectable.

We note that an ejection/accretion event such as the one considered in the present discussion would remain fairly modest in comparison with the more spectacular FUOr events which produce increases in the luminosity, mass accretion rates, and mass ejection rates of several orders of magnitude (Hartmann & Kenyon 1996). In this sense, we would only have witnessed a "mini-outburst." The very fact that such a mini-outburst would have been observed, however, would indicate that they most likely occurred quite frequently. They may, therefore, offer more tractable evidence of the relation between accretion and outflow phenomena than the more dramatic, but much less common FUOr events.

If the creation of the A2α/A2β pair is related to a modification of the jet ejection speed without an associated increase in mass accretion rate, no change in the luminosity of IRAS 16293–2422 is expected. Thus the presence or absence of an associated luminosity increase would be a good discriminant between the two possibilities. We note that a ∼30% increase in the velocity of a fairly modest underlying jet (with $\dot{M}$ of a few 10⁻⁷ M_☉ yr⁻¹) lasting for a few months would be sufficient to accumulate a mass of order 10⁻⁸ M_☉ in a working surface.