4. Emission lines
Several narrow emission lines were detected in the L1551 IRS 5 spectrum, specifically [Fe II] 17.94 and 25.99, Si II 31.48, O I 63.2, OH 84.4 and C II 157.7. These can be seen in Fig. 4, and are now discussed by species.
4.1. [Fe II]
[Fe II] line emission is often detected towards starburst galaxies, where it has been interpreted as having been excited by collisional excitation in supernova remnant shocks (Moorwood & Oliva 1988; Lutz et al. 1998). [Fe II] lines have been reported towards several supernova remnants (Oliva et al. 1999a [RCW 103], b [IC 443]), galactic nuclei (Lutz et al. 1996 [SgrA*], 1997 [M82]), and in regions known to have energetic outflows (Wesselius et al. 1998 [S106 IR and Cepheus A]). Optical echelle spectra towards the jet and working surface where the jet interacts with the surrounding medium (Fridlund & Liseau 1988) show linewidths of 100 and 200 km s-1 respectively. These results unambiguously show that shock excitation is occurring, although the position of the working surface in L1551 IRS 5 lies outside the SWS field of view. From the raw data, the linewidth of the 26.0 [Fe II] line, deconvolved from the instrumental resolution, is 230 km s-1. The ratios [Fe II] 35.3 / 26.0 and [Fe II] 24.5 / 17.9 should be density sensitive, although their variation over a wide range of densities is only a factor of two. The transitions have high critical densities 106 cm-3 and high excitation temperatures 400 K. As minor coolants, they do not significantly affect the thermal structure of the cloud. To test the possibility that the gas could be photoionised, we ran the photoionisation code CLOUDY (Ferland et al 1998) over a wide range of values. The [Fe II] 24.5 line is only efficiently excited under conditions of high density ( cm-3) and high UV illumination ( 106) (see also Hollenbach et al. 1991) - however, it is impossible to excite the [Fe II] 17.9 line for reasonable values of ionising flux near L1551 IRS 5, due to the low effective temperature of the star ( 5500 K). A shocked environment seems a more likely environment to excite the lines. Such a shocked environment could be associated with the optical jet emanating from L1551 IRS 5. We detect emission from [Fe II] 17.9 and 26.0 of 9.44 1.510-16 W m-2 and 2.04 0.310-15 W m-2 respectively, with a marginal detection of [Fe II] 35.3 of 1.62 0.610-15 W m-2. There are further [Fe II] lines at 5.34, 51.3 and 87.4 for which we set 2 upper limits of 1.0510-15, 6.5210-16 and 3.610-16 W m-2 respectively. The 5.34 and 17.9 lines can be used to constrain density, and the 26 line can be used with other lines as a temperature estimator. However, as pointed out by Greenhouse et al. (1997), Lutz et al. (1998) and Justtanont et al. (1999), other excitation mechanisms such as fluorescence or photoionisation may be important in certain environments.
4.1.1. Observed transitions
A diagram of the lowest energy levels of [Fe II] is shown in Fig. 17 and the major transitions in the range of the ISO spectrometers are listed in Table 5. The observed line fluxes are also given, along with the aperture sizes of the data. From these it is clear that for extended emission several of the line intensity ratios are aperture dependent. Fortunately, some of the potentially most important lines, viz. the 26, 17.9 and 24.5 µm lines (the 2-1, 7-6 and 8-7 transitions in Fig. 17, respectively), were observed with the same aperture.
Table 5. SWS and LWS aperture sizes at various [Fe II] transitions, observed and dereddened fluxes
The table also lists intrinsic [Fe II] line fluxes, , i.e. after a correction for the attenuation by dust extinction has been applied. The values of the dust opacities used for this correction were obtained from the best fit model of the overall SED of IRS 5 (see Sect. 3 and below) and are displayed in Fig. 8 and Fig. 18. Contrary to naive expectations, extinction has a considerable effect on the line ratios in IRS 5 even in the mid- to far-infrared. Not only does the temperature diagnostic ratio 17.9/26 become slightly altered, but is actually inverted, changing from the observed value of 0.5 to the ´de-reddened´ one of 2.
Emission lines of [Fe II] detected in the SWS spectrum of L1551 IRS 5 include transitions at 26 and 17.9 µm. Further, the high excitation lines (cf. Fig. 17) [Fe II] 1.64 (Sect. 3.2) and 0.716 (Fridlund & Liseau, in preparation) are also clearly present in this source, as well as the [Si II] 35 line. Therefore, in the absence of any nearby bright source of UV radiation, excitation of these lines by shocks presents the only known, feasible alternative. However, the comparison of the line intensity ratios for various lines deduced from shock models (Hollenbach & McKee 1989; Hollenbach et al. 1989) with those of the observations makes no immediate sense. In fact, results obtained from these line ratios (including upper limits to, e.g., the hydrogen recombination lines) lead to diverging conclusions regarding the shock speeds and pre-shock densities in the source. When based on different chemical species, these inconsistencies could possibly be accounted for by differences in the abundances between the source and the models (the models use highly depleted abundances, e.g. for iron A (Fe)= 10-6, relative to hydrogen nuclei).
At first, an abundance mismatch could also be thought of capable explaining the remarkable strength of the observed mid-infrared emission lines. For instance, the observed flux of the [Fe II] 26 line, 10- 12 erg cm-2 s-1 (Table 5), would imply an intensity of the non-extinguished line 10-2 erg cm- 2 s-1 sr-1, assuming the jets from IRS 5 to be responsible for the shock excitation ( sr within the field of view of the SWS; Fridlund et al. 1997). This is, however, significantly larger than the maximum intensity from the published shock models (Hollenbach & McKee 1989, which is 10-2 erg cm- 2 s-1 sr-1 and which pertains to the extreme parameter values of the models, viz. cm-3 and km s-1. Therefore, matching the observations would need still higher pre-shock densities ( cm-3) and would thus be indicative of post-shock densities of the order of 108 cm-3. Such high densities are nowhere observed in or along the jets. The fact that the jet emission in the density sensitive [S II] 0.6717 to 0.6734 line ratio is nowhere saturated implies that post-shock jet-densities never exceed a few times 103 cm-3 (Fridlund & Liseau 1988). Finally, the dense-and-fast-jet scenario can be ruled out since the expected H I recombination line emission (e.g., Br ) that should be excited in this scenario is not observed. In conclusion, it seems obvious that the hypothesis that the lines are excited by one or both of the jets encounters major difficulties.
An alternate model, presented below, provides not only a satisfactory explanation of the observed [Fe II] spectrum, but also a coherent picture of the central regions of the IRS 5 system. An [Fe II] source of dimension a few times 1015 cm, i.e. twice the size of the central binary orbit ( 90 AU), with densities of the order of 109 cm-3 and at average gas temperatures of about 4000 K is capable of explaining the observed line fluxes. This putative source of emission is situated at the centre of L1551 IRS 5 and seen through the circumstellar dusty material, which attenuates the radiation both by extinction and by scattering (see Fig. 8 and Fig. 18). This configuration presumably constitutes the base of the outflow phenomena from IRS 5. The precise nature of the heating of the gas remains unknown, although one obvious speculation would involve the interaction of the binary with the surrounding accretion disc.
4.1.2. [Fe II] excitation and radiative transfer
The model computations of the [Fe II] spectrum made use of the Sobolev approximation. This seems justified, since (a) velocity resolved observed [Fe II] lines have widths exceeding 200 km s-1 (e.g., 0.716 µm; Fridlund & Liseau, in preparation) and (b) according to the above discussion, high velocity shocks are needed to meet the energy requirements of the observed emission. The energies of the [Fe II] levels, Einstein A-values and the wavelengths of the transitions were adopted from Quinet et al. (1996). The number of radiative transitions included in the calculation was 1438. These lines are distributed from the FUV to the FIR spectral regions (0.16 to 87 µm) and the level energies span the range = 0 - 9.1104 K (the ionisation potential of [Fe II] corresponds to nearly 1.9105 K). The collision rate coefficients were calculated from the work by Zhang & Pradhan (1995), who provide effective collision strengths for 10 011 transitions among 142 fine structure levels in [Fe II]. These are Maxwellian averages for 20 temperatures in the range 1000 K to 105 K.
No lines of or (or of higher ionisation for that matter) have been detected from IRS 5, so that our assumption that essentially all iron is singly ionised seems reasonably justified. We further assume that iron is undepleted in the gas phase with solar chemical abundance, i.e. A (Fe)=3.210-5 (Grevesse & Sauval 1999). At temperatures significantly below 8000 K the gas would be only partially (hydrogen) ionised. In this case, we assume that the electrons are donated by abundant species with similar and/or lower ionisation potentials. In particular, primarily by Fe and Si plus other metals such as Mg, Al, Na, Ca etc., so that .
The line intensities were calculated for a range of gas kinetic temperatures and hydrogen densities and an example of the results is presented in Fig. 19. In the figure, intensity ratios for [Fe II] 5.34, 17.9 and 26.0 µm lines are shown. These lines are connected (see Fig. 17): the 17.9 µm and 5.34 µm lines originate from the same multiplet, a 4Fe. The upper level of the 17.9 µm line is at 3500 K above ground and its lower level is the upper level of the 5.34 µm line, which connects to the ground, and so does the 26 µm groundstate line (a 6De). The corresponding SWS data are also shown in the graph, where the emitting regions have been assumed to be much smaller than any of the apertures used for the observations. These data indicate source temperatures to be somewhere in the range of 3000 to 5000 K and gas densities to be above 3107 cm-3.
4.1.3. [Fe II] model calculations and results
The model spectrum, ´tuned´ to the SWS observations with the parameters of Sect. 4.1.1, is shown in Fig. 20. The upper panel of the figure displays most of the 1438 lines of the computed intrinsic [Fe II] spectrum, stretching from the far-UV to the far-IR. This spectrum suffers extinction by the circumstellar dust before it reaches the outside observer and is shown in the middle panel. Scattering by the circumstellar dust is of only low significance for the emission in the mid-infrared but is probably important at near-infrared and shorter wavelengths. In the lower panel of Fig. 20, a simple scattering model has been applied: the displayed spectrum corresponds to the point (about 1" off the binary centre) where about 1 % of the emitted spectrum is scattered at the efficiencies shown in Fig. 8. The intrinsically strongest transitions (see also Table 6) fall into the visible and near infrared regions of the spectrum and the scattered fraction of this emission may be detectable. Polarimetric line imaging would be helpful in this context, providing valuable insight into the source geometry. This would be of particular relevance to modelling of the H2 line emission (cf. Sect. 4.5), and may help to constrain the grain properties, which would in turn reduce any uncertainty in the computed extinction curve. It is clear that future modelling of this source will need to address a wider parameter space, particularly of grain properties - however this is beyond current computational capabilities.
Table 6. [Fe II] line optical depths and intensities (Model: = 4000 K, n(H) = 109 cm- 3, )
The spectrum of the adopted model (Table 6) is shown superposed onto selected regions of the observed SWS scans in Fig. 21. The theoretical fit is acceptable for most lines, except perhaps for the 24.52 µm transition which appears too strong in comparison with the observed line. All of the major [Fe II] lines are thermalised and optically thin, implying that the emission model can be easily scaled by keeping the parameter constant. For instance, models with higher velocity and lower density (e.g. 300 km s-1 and 5108 cm-3) or vice versa (e.g. 15 km s-1 and 1010 cm-3) would still yield the same [Fe II] intensities (but would otherwise disagree with the presence or absence of other lines in the spectrum of IRS 5). The cooling of the gas in all [Fe II] lines amounts to a few times 10-3 , which is comparable to the [O I] 63 µm luminosity, 710-3 . For comparison, the total luminosity generated by the shocks can be estimated as , where we have assumed a gas compression . This amounts to about 20 % of the total radiative luminosity of IRS 5 ( 45 , Table 3). The largest uncertainty lies in the value of , the dependence of which is cubic. However, the radiative luminosity of the IRS 5 system can be expected to be dominated by accretion processes, whereas the shock luminosity is probably generated by mass outflows. To order of magnitude, these estimates would then seem reasonable.
As to how the degree of ionisation of the gas, albeit low, is generated and maintained we have essentially no information. Dissociation, if initially molecular, and subsequent ionisation through shocks seems a likely option. In any case, one would expect the partially ionised gas to produce free-free continuum emission, with a flux density (in mJy) . From the [Fe II] model, the emission measure of the gas is , the free-free Gaunt factor and the source solid angle . Consequently, we find at 1.4 GHz (21 cm wavelength) , which is not far from what has actually been observed. The Very Large Array (VLA) measurements taken in August 1992 by Giovanardi et al. (2000) obtained mJy are closest in time to the ISO observations.
Recent observations with the SUBARU telescope by Itoh et al. (2000) have shown that the optical jet is dominated by [Fe II] lines, and suggest that the extinction to the jet is on average 7 mag. Fridlund et al. (1997) provide observed H fluxes for the entire jet (ground based and HST), viz. = 4.210-14 erg s-1 cm2 (the working surface, knot D, alone radiates 50 of this flux; this is not contained inside the observed fields of view of either the ISO-SWS or SUBARU data). Applying the extinction value estimated by Itoh et al. (2000) results in an H (0.6563 µm) extinction of 165, so that the intrinsic H flux of the observed jet would be = 3.510-12 erg s-1 cm2.
The shock models of Hollenbach & McKee (1989) predict that over the range = 40-150 km s-1 and = 103-106 cm-3; the intensity ratio of H/[Fe II 26 µm] should be 30 - 500. Taking the observed value of the 26 µm line, = 210-12 erg s-1 cm2, would then imply a `predicted' H flux F(H)0 610-11 - 10-9 erg s-1 cm2. This exceeds by more than one order of magnitude, the value inferred for the jet putatively extinguished by 7 magnitudes of visual extinction.
Since the intrinsic line ratio (H/H)0 is equal to or larger than 3, dust extinction with = 7 mag would result in a line ratio 35 (see, e.g. Appendix B of Fridlund et al. 1993). The observed ratio is = 15 (Cohen & Fuller 1985), which is significantly smaller than the predicted lower limit to the line ratio. This would increase the discrepancy even more.
In summary, it is again concluded that the [Fe II] emission observed by the ISO-SWS is not dominated by the jet, but its source is of different origin. That the jet is emitting in [Fe II] lines has been known for some time and is not new, but the level of emission is not sufficient to explain the SWS observations.
4.2. Si II
The sole [Si II] line detected is the - ground state magnetic dipole transition at 34.8. It should be one of the major coolants in hot ( 5000 K) gas. It has been previously suggested to be a shock tracer (Haas et al. 1986). We detect a flux of 2.35 0.310-15 W m-2 from this line. It is of interest to understand to what extent this flux is consistent with the prediction from our model of the central source.
with obvious notations. The fractional population of the upper level, , is obtained from
where the collision rate constants, (cm3 s-1), are related to the respective Maxwellian average of the collision strength, , by
At the level, upper limits can be set for F[Si I] 68.5µm W cm-2 and F[Si III] 38.2µm W cm-2), whence we can safely assume that essentially all silicon is singly ionised. can therefore be expressed as , where is the abundance of silicon with respect to hydogren nuclei.
Energy levels and Einstein-A values were adopted from Wiese et al. (1966) and Kaufman & Sugar (1986) and collision strengths from Callaway (1994). As for iron, silicon is assumed to be undepleted in the central core regions of L 1551 IRS 5 and we adopted the solar value of the silicon abundance, (Asplund 2000).
For the values of the model parameters, viz. n(H) = 10 cm-3, cm-3 and K, the fractional population becomes . Further, in conjunction with sr and AU, the intrinsic model flux then becomes ([Si II ] 35 µm) erg cm-2 s- 1. Taking the dust extinction by the intervening disk into account, Fig. 18, leads to the predicted estimation of the observable flux, i.e. ([Si II ] 35 µm) erg cm- 2 s-1.
In this undepleted case, the model flux is only slightly larger, by a factor of 1.5, than the actually observed value. We conclude therefore that our model of the central regions in L 1551 IRS 5 is capable of correctly predicting the flux in the [Si II ] 35 µm line. It is likely that most of this emission also originates in these central regions.
The excitation of OH has been modelled by Melnick et al. (1987), who studied OH emission towards the Orion Nebula. The line which at best detected at a low significance level, corresponds to a blend of the J= 7/2+ - 5/2- and J= 7/2- - 5/2+ rotational lines at 84.42 and 84.59 µm respectively (see Fig. 4). The integrated flux contained in these two blended lines is 1.5110-19 W cm-2. Other OH lines within the ISO spectral bands are the J = 9/2 - 7/2 transitions at 55.9 µm, the J = 5/2 - 3/2 lines at 119.3 µm and the J= 3/2 - 1/2 transitions at 163 µm, for which we set 3 upper limits of 3.110-20 W cm-2, 2.910-20 W cm-2 and 5.910-20 W cm-2 respectively.
Water emission has already been observed towards a number of molecular outflows with the ISO spectrometers (Liseau et al. 1996; Ceccarelli et al. 1999; Nisini et al. 1999). It is already well understood that the presence or absence of H2O in a cloud, may be used to trace the shock activity of the gas (Bergin et al. 1998, 1999). We have been unable to identify any H2O emission from IRS 5 (see Fig. 4), although we set a 2 upper limit on the 40.69 µm 432-303 line of 3.62 1.5510-15 W m-2. This is the most intense water line seen in W Hya (Neufeld et al. 1996). No lines were detected towards L1551 IRS 5 at 29.8, 31.8, 174.6, 179.5 and 180.5 µm, which correspond to strong lines that have been detected from other sources, to upper 2 upper limits of 21, 26, 4, 6 and 6 10-20 W cm-2 respectively (uncorrected for extinction - see Fig. 18).
No emission lines were detected from any of the rotational lines of H2 in the SWS spectrum towards IRS 5. A single detection of the = 1-0 vibrationally excited S (1) line at 2.122 µm has previously been reported by Carr et al. (1987), along with the Q-branch lines at 2.407, 2.414 and 2.424 µm, in a 3 aperture. Their reported S (1) flux of 2.8 0.310-17 W m-2 lies below the SWS 2 sensitivity limit at slightly longer wavelengths (the SWS spectrum starts at 2.4 µm) of 5.610-16 W m-2. In view of the low S/N of the Carr et al. (1987) data, we searched the UKIRT archive for a 2 µm spectrum to confirm the Carr et al. result. This spectrum is shown in Fig. 22.
The fluxes in the UKIRT observations are summarised in Table 7. From the upper limits to the = 0-0 lines, it is possible to set limits on the beam averaged H2 column densities using an ´excitation diagram´. The extinction corrected intensity of an H2 line I () is related to the column density of the line, N() by the relationship:
where are the wave-numbers (Dabrowski 1984) and A () are the transition probabilities for the various transitions, taken from Turner et al. (1977). The column densities are then compared with those predicted for a thermal distribution characterised by a rotational temperature since the rotational populations of a given vibrational level can be approximated by a thermal distribution:
Table 7. Summary of IRS 5 H2 line fluxes.
In this relationship the g terms are the degeneracies of the transitions, = (2J + 1) for even J (para-H2) and = 3 for odd J (ortho-H2), and the E () terms represent the energy of the () level. Thus the rotational temperature can be estimated from the inverse of the slope of a plot of against , correcting for an ortho/para ratio of 3:1.
It is clear from Fig. 23 that the = 0-0 lines are far less affected by extinction than the = 1-0 S (1) lines. We can infer from the 2 upper limits that the extinction to the = 1-0 S (1) emitting gas must be 80 - otherwise we should have detected emission in the = 0-0 S (6) or S (7) lines, assuming that the rotational temperature is characteristic of many sources, 2000 K. However, this limit is dominated by the S/N of the UKIRT spectra, and is not usefully stringent alone.
Shock models of the H2 lines have been calculated by a number of workers. For J-shocks, 1D models have been presented by Brand et al. (1988), Hollenbach & McKee (1989), Burton et al. (1992), Neufeld & Hollenbach (1994), and for C-shocks, by Draine et al. (1983), Smith (1991), Kaufman & Neufeld (1996).
The ISOCAM CVF observations of Cabrit et al. (1999) made towards a number of molecular outflow sources show that the mid-IR H2 lines dominate the cooling, and probe rotational temperatures in the range 300-2000 K. In the cases they studied, the observations were consistent with excitation in low-velocity C-shocks (10-30 km s-1), based on the non-detection of J-shock tracers predicted to be present in the standard shock models of Hollenbach & McKee (1989), for shock velocities 50 km s-1. From the L1551 IRS 5 data set, we set 2 upper limits of 2.510-15 W m-2 and 3.810-15 W m-2 on the Ni II 6.64 and Ne II 12.81 lines respectively.
To understand the excitation mechanism of the H2, the = 2-1 S (1) 2.248 to = 1-0 S (1) 2.122 ratio has often been used as a diagnostic. This ratio has values typically 0.1 in shocked regions and molecular outflows, and 0.6 for pure fluorescence. However, Draine & Bertoldi (1996) show that in dense PDRs, thermal collisions can transfer the lower-level population ( 2) towards LTE conditions, so that the line ratios approach those in shocked regions. Observational studies (Usuda et al. 1996; Takami et al. 1999) seem to confirm that this may indeed be the case. In one example, the case of the shocked region close to Orion KL, the ratio varies between 0.1 and 0.2 over a wide range of = 1-0 S (1) intensity, whereas in a typical dense PDRs such as the Orion Bright Bar, and the reflection nebulae NGC 2023 and NGC 7023, the ratio varies between 0.2 and 0.6, and shows a clear anti-correlation with the = 1-0 S (1) intensity. In L1551, the 2 upper limit on the = 2-1 S (1) 2.248 line is 9.110-19 W m-2, and the = 2-1 S (1) 2.248 to = 1-0 S (1) 2.122 ratio is 0.11. Thus the ratio appears to be inconsistent with fluorescent excitation, even allowing for thermal collisions. This conclusion is consistent with the lack of an obvious ionising source in the centre of L1551. Although no firm conclusions can be reached on the basis of the line ratio data shown in Table 7, we can rule out (a) low density ( 104 cm-3) shock models and (b) that the H2 emission is seen through more than two or three magnitudes of visual extinction (because the S (2) and S (3) lines, which suffer from the highest extinction, would become too bright relative to any of the models). It thus seems most likely that the = 1-0 H2 emission is seen in reflection.
4.6. CO vibrational bands
The CO absorption against IRS 5 has been previously studied by Carr et al. (1987) although the new UKIRT data presented here have higher sensitivity and spectral resolution. The data shown in Fig. 24 are fitted by a gas temperature of 2500 K, Doppler width of 5 km s-1, and CO gas phase column density of 61020 cm-2.
4.7. CO rotational lines
It was not possible to detect rotational molecular CO line emission in any of the spectra. For the lowest-J transitions towards L1551 IRS 5, we set 2 upper limits for the J =14-13 and J =15-14 lines of 4.310-17 W m-2 and 1.3710-16 W m-2 for L1551 IRS 5 and HH 29 respectively. L1551 appears unlike many other outflow sources, which are rich in shock excited CO line emission.
4.8. O I
Towards L1551 IRS 5 the OI 63um flux is observed to be more intense than along the flow, which indicates that the emission is intrinsic and not due to the diffuse PDR. If the emission was due to the jet, then the mass loss rate is 7.610-7 yr-1 (using the relationship from Hollenbach 1985), which is very similar to the 10-6 yr-1 derived from this source by other measurements. Along the flow it is likely that the OI emission comes from shocks in the outflows more than from diffuse PDR, since the OI lines are frequently brighter than the C I 157 µm, particularly towards HH29, L1551 NE and on the b1,b2 and b3 positions, which are associated with peaks of C emission in the outflow map by Rainey et al. (1987) and Moriarty-Schieven & Snell (1988).
4.9. HH 29
A deep (2 hour) observation with the LWS was also made centred on the location of HH 29. The baseline subtracted (fitting a low order polynomial to the continuum level) spectrum is shown in Fig. 24, and is devoid of line emission, except for the O I 63.2 and C II 157.7 lines as listed in Table 1.
4.10. L1551 NE
A red 6 source has been reported lying close to IRS 5, which has become known as L1551 NE (Emerson et al. 1984). This source has been imaged by Draper et al. (1985) and Campbell et al. (1988). Moriarty-Schieven et al. (1995) first suggested that L1551 NE was the source of a second outflow close to IRS 5, and subsequent observations by Devine et al. (1999) indicated that L1551 NE may be responsible for driving the Herbig-Haro flow HH454. Observations of the continuum towards L1551 NE were also made, and are shown in Fig. 26. This spectrum shows little evidence of any line emission, other than reported in Table 1.
4.11. Other locations along the outflow
Observations at a number of other locations along the molecular outflow, revealed no emission apart from weak C II and O I lines, and will not be further discussed here.
© European Southern Observatory (ESO) 2000
Online publication: January 29, 2001