A&A 413, 993-1007 (2004)
J. K. Jørgensen1 - M. R. Hogerheijde2 - E. F. van Dishoeck1 - G. A. Blake3 - F. L. Schöier1,4
1 - Leiden Observatory, PO Box 9513, 2300 RA Leiden, The Netherlands
2 - Steward Observatory, The University of Arizona, 933 N. Cherry Avenue, Tucson, AZ 85721-0065, USA
3 - Division of Geological and Planetary Sciences, California Institute of Technology, MS 150-21, Pasadena, CA 91125, USA
4 - Stockholm Observatory, AlbaNova, 106 91 Stockholm, Sweden
Received 18 June 2003 / Accepted 1 October 2003
This paper investigates small-scale (500 AU) structures of dense gas and dust around the low-mass protostellar binary NGC 1333-IRAS2 using millimeter-wavelength aperture-synthesis observations from the Owens Valley and Berkeley-Illinois-Maryland-Association interferometers. The detected mm continuum emission from cold dust is consistent with models of the envelope around IRAS2A, based on previously reported submillimeter-continuum images, down to the 3 , or 500 AU, resolution of the interferometer data. Our data constrain the contribution of an unresolved point source to 22 mJy. The importance of different parameters, such as the size of an inner cavity and impact of the interstellar radiation field, is tested. Within the accuracy of the parameters describing the envelope model, the point source flux has an uncertainty by up to 25%. We interpret this point source as a cold disk of mass 0.3 . The same envelope model also reproduces aperture-synthesis line observations of the optically thin isotopic species C34S and H13CO+. The more optically thick main isotope lines show a variety of components in the protostellar environment: N2H+ is closely correlated with dust concentrations as seen at submillimeter wavelengths and is particularly strong toward the starless core IRAS2C. We hypothesize that N2H+ is destroyed through reactions with CO that is released from icy grains near the protostellar sources IRAS2A and B. CS, HCO+, and HCN have complex line shapes apparently affected by both outflow and infall. In addition to the east-west jet seen in SiO and CO originating from IRAS2A, a north-south velocity gradient near this source indicates a second, perpendicular outflow. This suggests the presence of a binary companion within (65 AU) from IRAS2A as driving source of this outflow. Alternative explanations of the velocity gradient, such as rotation in a circumstellar envelope or a single, wide-angle () outflow are less likely.
Key words: ISM: individual objects: NGC 1333-IRAS2 - stars: formation - ISM: molecules - ISM: jets and outflows
|Figure 1: a) and b) SCUBA maps of IRAS2 at 450 and 850 m, respectively, centered on IRAS2A. c) Observed SED (symbols) and model fit (solid curve). d) and e) Brightness profiles at 450 and 850 m (symbols) and the model fit (solid curve). The dashed curves show the beam profiles. f) Density (dashed line) and temperature (solid line) distributions of best-fit model. See Paper I for details.|
|Open with DEXTER|
The deeply embedded ("class 0''; André et al., 1993; Lada, 1987) young stellar system NGC 1333-IRAS2 (IRAS 03258+3104; hereafter IRAS2) has been the subject of several detailed studies. It is located in the NGC 1333 molecular cloud, well known for harboring several class 0 and I objects, and was first identified from IRAS data by Jennings et al. (1987). Quoted distances to NGC 1333 range from 220 pc (Cernis, 1990) to 350 pc (Herbig & Jones, 1983); here we adopt 220 pc in accordance with jorgensen02. At this distance the bolometric luminosity of IRAS2 is . Submillimeter-continuum imaging (Sandell & Knee, 2001, and Fig. 1 below) and high-resolution millimeter interferometry (Blake, 1996; Looney et al., 2000) have shown that IRAS2 consists of at least three components: two young stellar sources 2A and, 30 to the south-east, 2B; and one starless condensation 2C, 30 north-west of 2A. The sources 2A and 2B are also detected at cm wavelengths (Reipurth et al., 2002; Rodríguez et al., 1999).
Maps of CO emission of the IRAS2 region show two outflows, directed north-south and east-west (Engargiola & Plambeck, 1999; Knee & Sandell, 2000; Sandell et al., 1994; Liseau et al., 1988). Both flows appear to originate to within a few arcsec from 2A (Engargiola & Plambeck, 1999), indicating this source is a binary itself although it has not been resolved so-far. The different dynamical time scales of both flows suggests different evolutionary stages for the binary members, which lead Knee & Sandell (2000) to instead propose 2C (30 from 2A) as driving source of the north-south flow. It is unclear how well dynamic time scales can be estimated for outflows that propagate through dense and inhomogeneous clouds such as NGC 1333. Single-dish CS and HCO+ maps also show contributions by the outflow, especially for CS (Ward-Thompson & Buckley, 2001). The north-south outflow may connect to an observed gradient in centroid velocities near 2A, but the authors cannot rule out rotation in an envelope perpendicular to the east-west flow.
jorgensen02 determined the physical properties of the IRAS2 envelope using one-dimensional radiative transfer modeling of Submillimeter Common User Bolometer Array (SCUBA) maps and the long-wavelength spectral energy distribution (SED). Assuming a single radial power-law density distribution, , an index p=1.8 and a mass of 1.7 within 12 000 AU was found (see Table 1 and Fig. 1). Monte-Carlo modeling of the molecular excitation and line formation of C18O and C17O observations yield a CO abundance of with respect to H2, a factor 4-10 lower than what is found in local dark clouds (e.g. Frerking et al., 1982; Lacy et al., 1994).
Table 1: The parameters for IRAS2 from jorgensen02.
This paper presents mm interferometric observations of IRAS2 in a range of molecular emission lines probing dense gas and continuum emission tracing cold dust. It builds on the modeling of Paper I by using it as a framework to interpret the small-scale structure revealed by the aperture-synthesis data. Section 2 describes the observations and reduction methods. Section 3 analyzes the continuum emission, and compares it to the previously derived models. Section 4 presents the molecular-line maps and discusses the physical and chemical properties of the gas in the proximity of IRAS2. Section 5 reports a pronounced north-south velocity gradient around IRAS2A and explores rotation or outflow as possible explanations. Section 6 concludes the paper by summarizing the main findings. A companion paper (Jørgensen et al., 2003a) presents a detailed study of the bow shock at the tip of the east-west jet from IRAS2 based on single-dish and interferometric (sub)millimeter observations.
The millimeter interferometer of the Berkeley-Illinois-Maryland Association (BIMA) observed IRAS2 on November 4-5, 2000, and January 20-21, February 20, and June 5-6, 2001. The array B-, C-, and D-configurations provided projected baselines of 1.7-68 k. The lines of HCO+ J=1-0, HCN 1-0, N2H+ 1-0, and C34S 2-1 were recorded in 256-channel spectral bands with a total width of 6.25 MHz (20 km s-1). The complex gain of the interferometer was calibrated by observing the bright quasars 3C84 (4.2 Jy) and 0237+288 (2.3 Jy) approximately every 20 min. The absolute flux scale was bootstrapped from observations of Uranus. The rms noise levels are 0.14 Jy beam-1 in the 24 kHz channels, with a synthesized beam size of FWHM. The data were calibrated with routines from the MIRIAD software package (Sault et al., 1995).
In the reduction, data points with clearly deviating phases or amplitudes were flagged. The maps were cleaned down to 3 times the rms noise using the MIRIAD "clean'' routine. The strong continuum of the two central point sources allowed self-calibration, which was applied and used to correct the spectral line data. The naturally weighted continuum observations typically had rms noise better than Jy beam-1 with half power beam widths (HPBW) of 3 for the OVRO observations and 8 for the BIMA data (see Table 3). Table 2 lists the details of the line observations.
Table 2: Line data of IRAS2 discussed in this Paper.
|Figure 2: Maps of continuum emission at 86-89 GHz from OVRO a) and BIMA b). Offsets are with respect to the pointing center of and . Contours are shown at 3, 6, 12, 24 and 48, where is the rms noise level of Table 3. The filled ellipses in the lower left corner of the panels indicate the synthesized beam sizes; the large circles show the 50% sensitivity levels of the primary beams.|
|Open with DEXTER|
Table 3: Results of fits to the visibilities.
|Figure 3: Visibility amplitudes of the observed continuum emission from the BIMA ( upper panel) and OVRO ( lower panel) observations as a function of projected baseline length in k centered at the position of IRAS2A. The data are plotted as filled symbols with 1 error bars, the continuous lines indicate the predictions from the continuum model of jorgensen02 with the same (u,v) sampling as the observations - respectively with (solid) and without (dashed) an unresolved compact source of 22 mJy added to the model. The dotted histogram indicates the zero-expectation level: the expected amplitude signal due to noise alone in the absence of source emission.|
|Open with DEXTER|
|Figure 4: Visibility amplitudes of the observed BIMA continuum emission as in Fig. 3 compared to various input models centered at the position of IRAS2A. Upper panels: a) models with changing steepness of the density profile; b) test of different values of the outer radius and inclusion of the interstellar radiation field. Models with an outer radius 3 times larger than the model from Paper I (i.e. 36 000 AU) are shown. Lower panels: c) fit to the inside-out collapse model of Shu (1977) with parameters constrained independently by molecular line observations and SCUBA continuum observations; d) models with changing size of the inner radius.|
|Open with DEXTER|
The inferred point source flux is consistent with Looney et al. (2003), who find a 20 mJy source at 2.7 mm associated with IRAS2A. For reasonable assumptions about the spectral index of the point source (-4 if thermal) this component does not contribute significantly to the flux in the much larger SCUBA beams, and therefore does not invalidate the SCUBA-based envelope models. The thermal nature of the point source is supported by detection of 2A at a flux of 0.22 mJy at 3.6 cm and a resolution of 0.3 with the VLA (Reipurth et al., 2002). This yields a spectral index of 1.9 between 3.6 cm and 3.3 mm, consistent with optically thick thermal emission. A similar conclusion was reached by Rodríguez et al. (1999) based on the spectral index from VLA observations of IRAS2A at 3.6 and 6 cm.
Assuming that the point source emission is optically thin and thermal, the inferred flux of 22 mJy corresponds to a dust mass of if we adopt an average temperature of 30 K and a emissivity per unit (dust) mass at 3.5 mm of cm2 g-1 from extrapolation of the opacities by Ossenkopf & Henning (1994) for grains with thin ice mantles as was assumed in the envelope models in jorgensen02. With a standard gas-to-dust ratio of 100, the total mass is 0.33 . If the emission is optically thick as the spectral index indicates, this is in fact a lower limit to the mass. The favored explanation for this compact mass distribution is a circumstellar disk.
Harvey et al. (2003) find that the uncertainty in the central point source
flux dominates over uncertainties in other model parameters such as
external heating by the interstellar radiation field (ISRF),
wavelength dependence of the dust emissivity, outer radius of the
envelope, and deviations from spherical symmetry (e.g., an evacuated
outflow cavity). Because our interferometer data only sample the inner
regions, they are not sensitive to variations in the outer radius or
inclusion of heating by the ISRF. The latter hardly affects the
temperature structure because the source is relatively luminous and
dominates the heating as is seen in Fig. 5.
|Figure 5: Temperature profile in the outermost region of the envelope without (solid line) and with (dashed line) contributions from the interstellar radiation field. The dotted line indicates the temperature of 10 K corresponding to the envelope outer radius jorgensen02.|
|Open with DEXTER|
In similar studies to that presented in jorgensen02, Shirley et al. (2002) and Young et al. (2003) modeled the SEDs and brightness profiles from SCUBA observations of protostellar sources using 1D radiative transfer, assuming power-law density profiles and solving for the temperature structure. Two differences exist, however, in the approaches taken in these two papers and our jorgensen02: Shirley et al. and Young et al. included contributions to the heating of the envelope by the external interstellar radiation field and adopted outer envelope radii significantly larger than those set by the 10 K boundary used in jorgensen02. For the sources common to the two samples, Shirley et al. found on average steeper density profiles than ours for the class 0 objects whereas Young et al. found similar density profiles to ours for the class I objects. Young et al. suggested that the disagreement for the class 0 objects and agreement for the class I objects was due to a combination of neglect of the ISRF and an underestimate of the sizes of the envelopes in jorgensen02: while inclusion of the ISRF will indeed tend to steepen the derived density profile, an overestimate of the outer radius (by factors of 2 or more) will tend to flatten the derived profile. As illustrated above, however, these parameters have negligible impact on the IRAS2 envelope structure. It is therefore interesting to note the agreement in slope between the interferometer and SCUBA continuum observations, in contrast to the discussion of B335 by Harvey et al. (2003). Comparing to the results of Shirley et al. (2002), Harvey et al. found a slightly flatter density profile when modeling the interferometer observations. While uncertainty in the outer radius and ISRF may lead to only small departures for the interferometry data, it can lead to systematic changes in the slope of derived power-law density profile from the SCUBA observations of 0.2. This could explain the differences between the density profiles from the interferometry and SCUBA data for B335.
Our envelope model is entirely based on SCUBA data, and the
interferometer fluxes serve only to constrain any point source
flux. The robustness of that constraint depends on the assumption that
the envelope model can be extrapolated down to scales much smaller
than the SCUBA resolution (4
= 900 AU). In
jorgensen02 the inner radius is fixed at a temperature of
250 K which occurs at R=22 AU, but it was argued that this is not
determined by the data. The size of any inner cavity is expected to
affect the interferometer data since these sample small scales: a
larger adopted cavity would result in a higher inferred point source
flux to compensate for the reduced small-scale
emission. Figure 6 plots this "required''
point source flux against inner cavity size. The point source flux
increases from 22 mJy for cavities <25 AU to 27 mJy for
cavities 200 AU (1
). The interferometer data would
resolve cavities larger than this and a point source could no longer
compensate for the removed emission.
|Figure 6: Derived point source flux plotted against size of the inner envelope cavity.|
|Open with DEXTER|
Turning this reasoning around, the inferred point source could be due
to an increase in envelope density on small scales, as opposed to a
circumstellar disk in an envelope cavity. Assuming a temperature of
150 K appropriate for the envelope scales unresolved by the
interferometers instead of the 30 K assumed for the disk, one derives
a mass of 0.06 .
For comparison the mass of the single
power-law density model within 150 AU is only 0.008 .
explain the detected flux an increase in density by almost an order of
magnitude is needed, which seems unlikely.
|Figure 7: Changes of the emerging SED due to inclusion of a 200 AU outer, 0.3 disk: in the upper panel the SEDs from the star + envelope (solid line) and star + envelope + disk models (dashed line) are compared. In both models the central star is represented by a 5000 K blackbody. In the lower panel the ratio between the models are shown - the typical 20% error-level in the flux calibration is illustrated by the solid rectangle.|
|Open with DEXTER|
The model from jorgensen02 assumes that the envelope is heated by a stellar blackbody of 5000 K at the center. If, as is argued above, the star is surrounded by a disk that reprocesses a significant fraction of the stellar light, the input spectrum shifts to longer wavelengths. To investigate the effect on the envelope's temperature structure, Fig. 7 compares the SEDs of the original model and a model where the central star is surrounded by a 200 AU outer radius, 0.33 disk. The disk follows the descriptions of Chiang & Goldreich (1997) and Dullemond et al. (2001), and the envelope's inner cavity has been increased to 200 AU in radius so that it encompasses the disk. As a result, the temperature at the inner edge of the envelope drops from the original 250 K (at 22 AU) to 75 K (at 200 AU). The radiative transfer code DUSTY produces the envelope's temperature distribution and emergent SED using the star+disk spectrum as heating input, similar to the calculations of jorgensen02 for the star-only spectrum. The comparison in Fig. 7 shows that the SEDs between 60 m and 1.3 mm are unchanged. Our derived envelope parameters are therefore unaffected by the exact form of the input spectrum. The departures grow larger at the shorter wavelengths (2-20 m) and may be observable with, e.g., SIRTF. It is not surprising that the SEDs are most different at these wavelengths. Flared disk models such as those of Chiang & Goldreich (1997) are specifically invoked to explain so-called "flat-spectrum'' sources. Their superheated surface layers "flatten'' the SED of these star+disk systems by boosting the 2-20 m emission. It is not obvious that such a description of the disk is valid for early, deeply embedded objects, such as IRAS2A. Still, the important point here is that the influence of the disk on the observed SED is likely to be negligible at the wavelengths where the envelope model is constrained.
Fitting the Shu (1977) inside-out collapse model to the SCUBA data
for IRAS2 gives best fit values of a=0.3 km s-1 and
yrs (see Fig. 8) with the quality
of the fits essentially identical to those of the single power-law
models. This collapse model also fits the BIMA and OVRO data if a
point source of 25 mJy is introduced.
|Figure 8: Fits to the SCUBA observations with a inside-out collapse model with an isothermal sound speed, a, of 0.3 km s-1 and an age of yrs.|
|Open with DEXTER|
The integrated CS and C34S line intensities were fitted independently
with the collapse model using detailed radiative transfer as in
Schöier et al. (2002) for a constant fractional abundance with radius. The
uncertainties in the line fluxes are assumed to be 20% and for each
model the -estimator is used to pick out the best model and
estimate confidence levels for the derived parameters. Interestingly,
the fits to the CS and C34S lines (Fig. 9) give
identical parameters to those derived from the dust modeling, in
contrast with the other sources (e.g. Schöier et al., 2002; Shirley et al., 2002).
|Figure 9: Constraints on the inside-out collapse model derived from the CS and C34S line intensities, assuming CS and C34S abundances of and respectively found in the single power-law density model. The inferred age and sound speed agree well with those derived from the continuum data.|
|Open with DEXTER|
The identical fits to the line intensities and continuum observations and success of both collapse and power-law density models illustrates the low age inferred for IRAS2: the collapse expansion radius in the inside-out collapse model is for the envelope parameters located at and therefore not directly probed by the SCUBA continuum maps. Likewise the CS lines predominantly probe the outer regions of the envelopes where the density profile in the collapse model for the low age of IRAS2 essentially is a power-law. This also explains why a slightly higher point source flux is obtained with the collapse model than the power-law density model: inside the collapse expansion radius the density profile flattens in the collapse model, lowering the mass and thereby the flux towards the unresolved center of the envelope. This is compensated by increasing the point source flux when modeling the interferometer observations.
In summary, the interferometer continuum data are well described by the same 1.7 envelope models that fit the SCUBA data. Power-law descriptions for the density and inside-out collapse model fit the data equally well. They indicate the presence of a 22 mJy point source, presumably a 0.33 circumstellar disk. Uncertainties associated with the envelope model are reflected in the accuracy of the point source flux, which may vary by up to 25% from the quoted value. The next section describes the line emission in this context.
|Figure 10: Integrated line emission from the BIMA observations for a) HCN, b) HCO+, c) N2H+ and d) C34S plotted over the 3 mm continuum maps (grey-scale). The outflow axes have been marked with straight lines with the red part being solid and blue part being dashed. For HCN and HCO+ the emission has been integrated over the red and blue parts of the line (3 to 7 km s-1 and 9 to 13 km s-1) shown as the dashed and solid lines, respectively. For N2H+ and C34S the total integrated emission is presented, in the case of N2H+ integrated over the main group of hyperfine lines. For C34S the contours are presented in steps of 3, for HCN and HCO+ in steps of 5 and for N2H+ in steps of 7.|
|Open with DEXTER|
The intensity ratio of 1.5 : 2.7 : 4.7 of the N2H+ hyperfine lines
suggest that the emission is optically thin and close to LTE, where
the ratio would be 1 : 3 : 5. Relative to the 450 m emission from
SCUBA that traces cool dust, the N2H+ emission is strongest
around 2C, lower around 2A, and absent toward 2B as illustrated in
Fig. 13. In a study of dark cloud cores,
Bergin et al. (2002,2001) find that N2H+ has a large abundance
deep inside the clouds and a lower abundance in the exterior
regions. This trend is opposite to that of CO, which is often highly
depleted deep inside cores. Bergin et al. argue that N2H+ is effectively destroyed through reactions with CO, and
therefore is only present at high abundance where CO is depleted. This
scenario can also explain the relative distribution of N2H+ in 2C, 2A, and 2B: in the starless core 2C temperatures are low and CO is
highly depleted, resulting in strong N2H+ emission; in 2A the
star has already heated the material and some CO has been released,
reducing the N2H+ abundance and emission; the evolved core 2B
has been thoroughly heated by the star, and most N2H+ has been
effectively destroyed by the released CO. This hints at triggered star
formation with the sources lining up from southeast to northwest in
evolutionary order, with 2B older than 2A, and 2A older than 2C.
|Figure 11: Integrated line emission from the OVRO data, showing a) CS, b) H13CO+, c) SO and d) CH3OH. CS is integrated over blue (5 to 9 km s-1; dashed contours) and red (9 to 13 km s-1; solid contours) parts of the line with contours in steps of 3. For the other molecules lines have been integrated from 7 to 11 km s-1 and contours are given in steps of 2. The grey-scale indicate the 3 mm continuum maps.|
|Open with DEXTER|
|Figure 12: First order moment (velocity) maps of the a) HCN, b) HCO+(BIMA) and c) CS emission (OVRO). Each map has been overplotted with the total integrated emission in steps of 3 (solid line contours). The outflow axes have been marked by lines and IRAS2A and IRAS2B continuum sources by stars.|
|Open with DEXTER|
|Figure 13: The contrast between the N2H+ and SCUBA emission: the N2H+ emission divided by the 450 m continuum emission (normalized) along a straight line with a position angle of 45through IRAS2A (solid line) and the 450 m continuum emission in the same positions (dashed line).|
|Open with DEXTER|
Several molecules only show emission in a very narrow velocity range
around the systemic velocity of the cloud of 8 km s-1:
H13CO+, SO, CH3OH, N2H+, and C34S. Others show
pronounced gradients in a north-south direction within 20
from IRAS2A (HCN, HCO+, CS) and along the
east-west outflow (most clearly in HCO+). Whether the north-south
gradient around 2A is rotation in a circumstellar envelope or is
related to the north-south outflow seen on larger scales is addressed
in Sect. 5. To asses the amount of recovered flux as
function of velocity, Fig. 14 compares single-dish
spectra with interferometer spectra averaged over the single-dish beam
size and converted to the antenna-temperature intensity scale. The
interferometer recovers at most 17% of the emission, and much less in
many cases. Apart from a scaling factor, the line shapes of C34S,
N2H+, and H13CO+ are similar in the interferometer
and single-dish spectra, implying that although the
interferometers picked up only the more compact emitting structures,
only small velocity gradients can be present within the envelope
itself. Deep self-absorption apparent in HCO+, HCN, and CS near
the systemic velocity indicates that surrounding cloud material is
optically thick and entirely resolved out. The velocity structure seen
in these lines therefore only reflects material at relatively extreme
red- and blue-shifts.
|Figure 14: Comparison between the single-dish observations (dark) and corresponding spectra from the interferometer observations restored with the single-dish beam (red). The spectra from the interferometry observations have been scaled by the factors indicated in the upper right corner (factors 5-12) to include all spectra in the same plots.|
|Open with DEXTER|
Figure 15 compares the observed and modeled line emission for C34S. The upper panel shows a comparison between the model and the single-dish observations (upper spectrum) and the interferometry data convolved with the single-dish beam (lower spectrum). In the lower panel the visibilities are plotted as a function of projected baseline length. Both comparisons show that the model works very well in describing the interferometry and single-dish line observations simultaneously and reproduces the emission distribution at the observed scales. This has two implications. First, that optically thin species such as C34S trace material in the envelope and are well described by the model derived from the continuum observations. Second, for species such as C34S the chemistry is homogeneous at the observed radial scales in the envelope, so that a constant fractional abundance is sufficient to describe the chemistry within the assumptions and uncertainties in the modeling.
For H13CO+ the situation is a bit more complex. The modeling of the single-dish lines reveal a picture similar to that of the CO isotopic species in jorgensen02; while a constant fractional abundance can describe the line intensities of higher J lines, the intensity of the J=1-0 line is underestimated by the model. In jorgensen02 it was suggested that this was due to ambient cloud material being picked up by the larger single-dish beam. The same problem may be an issue for H13CO+. Fitting the H13CO+ J=1-0 single-dish line alone gives an abundance of . With this abundance, the model can describe the intensity of a spectrum convolved with a beam equivalent to the single-dish observations as illustrated in the upper panel of Fig. 16.
Table 4: Abundances in the 1D static envelope model derived from of single-dish line observations. For details see Jørgensen et al. (2003b).
|Figure 15: Upper panel: comparison between the C34S emission from the single-dish observations using the IRAM 30 m telescope with the BIMA interferometry observations (lower spectrum), offset along the axis by -0.8 K and restored with a 25 beam similar to the single-dish data. The prediction from the 1D static model of the emission brightness distribution from jorgensen02 with abundances derived from single-dish line observations given in Table 4 and sampled at the relevant (u,v)grid has been overplotted on the spectrum in red. Lower panel: visibility amplitude plotted as function of projected baseline length. The predictions from the envelope model with C34S abundance constrained by the single-dish line observations have been overplotted as the solid line. The zero-expectation level is indicated as the dotted histogram.|
|Open with DEXTER|
The model, however, cannot fit the line profiles simultaneously for the single-dish and interferometry spectra when a constant turbulent broadening is adopted. This is likely caused by the larger single-dish beam picking up the more extended cloud where the velocity distribution may be different. We emphasize, however, that this is a significantly smaller effect than what is seen for, e.g., CS, HCN and HCO+(Fig. 14). It is also seen that the model cannot describe the emission on smaller scales when plotting the visibilities vs. projected baseline length, as illustrated in the lower panel of Fig. 16. If the lower H13CO+ abundance found from fitting the higher J lines is adopted, the model perfectly reproduces the observed H13CO+ emission distribution. This discrepancy suggests that the single-dish beam of 44 picks up the ambient cloud and the envelope around IRAS2B as it also is seen in the H13CO+ 1-0 interferometry maps. This will contribute to the spectrum extracted from the interferometry cube when convolved with the single-dish beam (e.g. the upper panel of Fig. 16). The good fit to the visibility curve in the lower panel of Fig. 16 indicates that the abundance constrained by the higher excitation single-dish line observations of H13CO+ is representative of the actual envelope abundance.
It is not possible to account for the observed CS, HCN and HCO+emission within the envelope models. As can be seen in
Fig. 17, the CS line intensity is, for example
reproduced only at intermediate baselines where also the single-dish
line observations are sensitive. On longer baselines the model clearly
breaks down and underestimates the observed emission. It is also
evident that the pronounced double peak structure seen in
interferometry spectra cannot be explained with a simple collapse
model alone. This problem will be further explored in the next
|Figure 16: As in Fig. 15, but for H13CO+ emission as traced by single-dish observations from the Onsala 20m telescope (44 beam) and interferometry with the OVRO millimeter array. The H13CO+ interferometer spectra have been scaled by a factor 3 in order to be able to visualize them together with the single-dish observations. The solid line in both panels indicate the model with H13CO+ abundance constrained by the high J line observations. The dashed line indicate the model constrained by the H13CO+ J=1-0line. In the lower panel models both with a turbulent broadening of 0.5 and 1.5 km s-1 have been shown for the abundance constrained by the 1-0 line. The integrated intensities are the same, however.|
|Open with DEXTER|
|Figure 17: Comparison between interferometry observations and envelope model for CS.|
|Open with DEXTER|
The inferred north-south velocity gradient of 1.10 km s-1 arcsec-1, or km s-1 pc-1, is two orders of magnitude larger than that inferred by Ward-Thompson & Buckley (2001) from single-dish CS and HCO+observations. This increase in velocity gradient from the spatial scales of the single-dish to the interferometer data is too large to be explained by differential rotation in either a Keplerian structure (expected increase of a factor ) or a magnetically braked core (Basu, 1998) (expected increase of a factor 10). Even if the gradients on single-dish and interferometer scales are unrelated, Keplerian rotation cannot explain our observed velocities since it requires an unrealistically large central mass of 31 .
An alternative explanation for the north-south velocity gradient around IRAS2A is that it is part of the north-south outflow detected on larger scales in CO (Engargiola & Plambeck, 1999; Knee & Sandell, 2000; Liseau et al., 1988). Engargiola & Plambeck (1999) concluded that the origin of this flow lies within a few arcsec from 2A, while Hodapp & Ladd (1995) infer a north-south jet that passes within a few arcsec from 2A from H2images. Neither this Paper nor Looney et al. (2000) find evidence for continuum emission from a second source, although it could be below the detection limit or be unresolved (< AU). The different levels to which HCO+, HCN, and CS trace the east-west and north-south flows may reflect differences in shock chemistry as the flows progress through the inhomogeneous cloud environment of IRAS2. This also serves as caution in interpreting differences in the spatial extent of, e.g., CO line wing emission as differences in "dynamic time scales''; this presupposes similar environments in which the flows propagate.
Instead of two perpendicular outflows, a single, wide-angle (),
northwest-southeast flow could also explain the
observed gradients. In this interpretation, what appear to be two
independent flows actually trace the interaction of the wide-angle
flow with the surrounding material along the sides of the cavity. This
scenario is reminiscent of the wide-angle outflow of B5-IRS1
(Velusamy & Langer, 1998). However, in this scenario the jet-like morphology
of the shocked region east of IRAS2
(Blake, 1996; Bachiller et al., 1998; Jørgensen et al., 2003a) is difficult to explain.
|Figure 18: Position-velocity diagrams for a) HCN, b) HCO+ and c) CS. The solid line indicates a linear gradient fitted to the centroids for the velocity channel. The hyperfine splitting of HCN is seen as the extension of emission along the velocity axis.|
|Open with DEXTER|
Table 5: Velocity gradient derived through a linear fit to the centroid of the brightness for each velocity channel as indicated in Fig. 18.
The success of the envelope model in describing the optically thin species, such as C34S and H13CO+ makes IRAS2 a promising candidate in order to study the relation between the envelope chemistry and the spatial distribution of molecular species. In particular, studies of a larger sample of optically thin molecular lines at arcsec scale resolution may probe differences in the radial distributions of molecules reflecting the chemistry. IRAS2A is for this purpose a promising target due to the relative simplicity of the central envelope component. High angular resolution, high sensitivity maps may also allow for a more detailed comparison to models for the protostellar collapse in order to possibly address the evolution of low-mass protostars in the earliest stages.
The authors thank Kees Dullemond for use of the CGPLUS program and discussions of disk models. The research of J.K.J. is funded by the Netherlands Research School for Astronomy (NOVA) through a network 2 Ph.D. stipend and research in astrochemistry in Leiden is supported by a Spinoza grant. This paper made use of data from a range of telescopes among them the Owens Valley Radio Observatory and Berkeley-Illinois-Maryland-Association millimeter arrays, Onsala Space Observatory 20 m telescope and the James Clerk Maxwell Telescope. The authors are grateful to the staff at all these facilities and their host institutions for technical support, discussions, and hospitality during numerous visits.