A high-resolution long-slit spectroscopic study of the various bipolar outflow components in M 2-9 ( Butterfly Nebula )

Transcription

1 Astron. Astrophys. 354, (2) ASTRONOMY AND ASTROPHYSICS A high-resolution long-slit spectroscopic study of the various bipolar outflow components in M 2-9 ( Butterfly Nebula ) J. Solf Thüringer Landessternwarte Tautenburg, Sternwarte 5, 7778 Tautenburg, Germany (solf@tls-tautenburg.de) Received 9 September 1999 / Accepted 12 November 1999 Abstract. High-resolution long-slit spectrograms of the line emission from the bipolar nebula M 2-9 obtained at various slit positions are presented. The data are used to study in detail both the kinematic and morphological properties of the various components of the bipolar outflow in M 2-9. Three main regions of outflow have been distinguished: a compact inner region represented by the central core, an extended intermediate region represented by the bright bipolar lobes, and an outer region represented by the faint outer loops. All three regions show a remarkably high bipolar symmetry with a uniform inclination of the bipolar axes ( 73 ), whereas the deduced outflow velocities and kinematical ages are largely different from each other. In the central core region, two physically distinct gas components, a high-velocity component and a low-velocity component, have been identified. The fast gas is of relatively high excitation and represents a highly collimated bipolar outflow system (micro-jets) with velocities of up to 195 km s 1. The kinematical age of the micro-jets is extremely small (<1 years). The observations suggest that the outflow source is surrounded by a dense equatorial disk obscuring the inner portions of the receding jet. The slow gas is of lower excitation and is suggested to represent a wind either from the equatorial disk or from the evolved stellar component of the presumed central binary. In each of the bipolar lobes, a (co-axial) double-shell structure has been identified, consisting of an inner shell of fast hot gas and an outer shell of slow cool gas/dust. The hot gas, traced by the narrow line component representing in situ emission, shows outflow velocities of 46 km s 1 indicating a kinematical age of 13 years. The highest velocities are found near the bipolar axis. The cool gas/dust, traced by the broad line component representing dust-scattered emission, exhibits outflow velocities of 17 km s 1. The kinematical age of the cool-gas shell is about three times as large as that of the hot-gas shell. The faint outer loops, traced by dust-scattered H α line emission, present large redshifts in both loops indicating outflow velocities of 141 km s 1. The deduced kinematical age is 13 years, comparable to that of the bipolar lobes, suggesting that both the outer loops and the bipolar lobes were formed at the same time. Based on observations collected at the German Spanish Astronomical Center, Calar Alto, Spain Our results confirm that M 2-9 exhibits many properties which have little in common with those of planetary nebulae. In particular, the detection of fast bipolar jets and of a dense disk in the core region strengthen the hypothesis that M 2-9 probably belongs to a class of close mass-exchanging binary systems, like symbiotic novae, which are sources of collimated fast bipolar outflows. Key words: ISM: planetary nebulae: individual: M 2-9 ISM: jets and outflows ISM: kinematics and dynamics stars: circumstellar matter ISM: dust, extinction 1. Introduction M 2-9 (PK ) is one of the most exciting bipolar nebulae and has attracted the attention of many researchers in the field of collimated mass outflows from evolved stars. Basically the nebula is composed of three main components, namely two bright, narrow bipolar lobes oriented approximately along the north-south direction and extending up to about ±25 from the center; a bright, compact central core ( nucleus ) of about 1 extent; and two faint outer loops oriented along the bipolar axis of the lobes and extending up to about ±1 north and south of the center. The general structure at optical wavelengths of the bipolar nebula and of its various components has been described in detail, e.g., by Kohoutek & Surdej (198) and Schwarz et al. (1997). Within each lobe several bright condensations have been identified which appeared to change their relative positions in time (Allen & Swings 1972, van den Bergh 1974, Kohoutek & Surdej 198, Hora & Latter 1994, Goodrich 1991). Highresolution near-infrared images and spectra of M 2-9 by Hora & Latter (1994) revealed a double-shell structure of the lobes, consisting of an inner shell of hot (ionized) gas and an outer shell of cool (molecular) gas. Recently, a high-resolution narrow-band image of the core region and the bipolar lobes of M 2-9, obtained with the HST WFPC2, has been made available via the Internet (Balick, Icke & Mellema 1997). The HST image of M 2-9 (see Fig. 1) clearly indicates the presence of a double-shell structure in the bipolar lobes and reveals remarkable details of the various nebular condensations within each lobe. Near-infrared color images of M 2-9 by Aspin et al. (1988) suggest that the central

2 J. Solf: A study of the various bipolar outflow components in M PA 24 PA 9 δ +1" PA 9 δ " PA 9 δ 1" PA N2 N1 S2 S1 N3 PA -24 S3 M2-9 HST WFPC2 Fig. 1. Direct image of the planetary nebula M 2-9 derived from combined HST WFPC2 images obtained in the light of [N ii] λ6583, [O i] λ63, and [O iii] λ57 (Balick, Icke & Mellema 1997). The central core region (C) and the bright nebular condensation in the bipolar lobes (N1 to S3) are marked. The various slit positions of the obtained long-slit spectrograms are indicated. C E N core of M 2-9 is surrounded by a dense extended equatorial disk of luminous dust. Interferometric mapping of the 12 CO J =1 emission from M 2-9 by Zweigle et al. (1997) indicates the existence of an expanding equatorial torus of molecular gas of about 6 diameter around the center of M 2-9. The evolutionary status of M 2-9 is still controversial. M 2-9 has been identified with objects which bear the characteristics indicative for very young planetary nebulae, such as M 1-91, M 1-92 (Solf 1994a), CRL 618 (Schmidt & Cohen 1981) or proto-planetary nebulae (Walsh 1981). Because of its morphology and nebular spectrum M 2-9 prima facie has been generally considered to be a planetary nebula (PN). Its highly symmetric bipolar aspect lead Balick (1987) to consider M 2-9 as the prototype ( butterfly ) of one of three classes in a series describing a morphological sequence of PNs. As already outlined by Balick (1989, and references therein), M 2-9 exhibits many properties which have little in common with those of typical PNs. In particular, the central star of M 2-9, designated as B IV (Calvet & Cohen 1978) or late O type (Swings & Andrillat 1979), presents (circumstellar) emission lines of very low to high excitation, e.g., permitted and forbidden low-ionisation lines of O i, Fei, Feiii, and high-ionisation lines of [O iii], [Ne iii], and N v. As noted by Balick (1989), the H α emission line profile in the central nucleus exhibits extremely broad wings extending over more than 11, km s 1 at the base. These properties of the central nucleus of M 2-9, unknown in ordinary PNs, appear more typical for very close mass-exchanging binary systems, like other bipolars e.g., MZ-3, CRL 618, M 1-92, or symbiotic novae e.g., RR Tel, V116 Cyg, HM Sge, V1329 Cyg. Although direct evidence is still lacking, it is most plausible that the central object of M 2-9 is a mass-exchanging close binary system. The spatio-kinematic properties of the gas and dust within the various nebular components of M 2-9 were studied in the past using long-slit spectroscopy of the permitted line and forbidden line emissions (see, e.g., Balick 1989; Icke et al.1989). The detailed kinematics of the bipolar lobes was investigated first by Carsenty & Solf (1983) who noted that the H α line emission from the lobes consists of a narrow component (NC) and a redshifted broad component (BC). The authors proposed that the NC of the H α line and the forbidden lines represent intrinsic emission originating from the (hot) gas of the lobes whereas the BC represents reflected line emission originating from the central core and being scattered by dust particles within the gas of the lobes. High-resolution spectropolarimetry of various emission lines from the lobes by Trammell et al. (1995) confirmed the dust-scattering interpretation. Using a simple outflow model for the BC, Carsenty & Solf (1983) derived a flow velocity of 2 km s 1 for the dust/gas mixture in the bipolar lobes. Solf (1993) suggested that the flow velocity of the (ionised) hot gas component of the bipolar lobes can be deduced using the radial velocity and velocity dispersion derived for the observed NC and applying a simple bi-conical outflow model. He obtained a (mean) flow velocity of 46 km s 1 for the hot gas. Deep medium-resolution spectrograms of M 2-9 obtained by Schwarz (199) and Solf (1993) indicated that the faint outer loops present weak H α emission which (surprisingly) is redshifted by large amounts in both loops. Solf (1993) suggested that the observed H α line represents reflected emission originating from the central core of M 2-9 and being scattered by dust particles in the loops moving outwards. With this assumption, he deduced a velocity of 15 km s 1 for the dust/gas in the outer loops. High-resolution spectroscopic observations by Balick (1989) indicated that the [N ii] emission from the central core of M 2-9 consists of a single central component and two fainter satellite components. In this paper, we present new high-resolution long-slit spectroscopic observations of M 2-9 obtained at various slit posi-

3 676 J. Solf: A study of the various bipolar outflow components in M 2-9 tions and position angles. These data permit us to investigate in considerable detail both the kinematic and the morphological properties of the complex outflows in each of the different components of the bipolar nebula. The knowledge of these properties appears to be crucial for any evolutionary and dynamical model the prototype butterfly nebular M Observations and data reduction High-resolution long-slit spectroscopic observations of M 2-9 were obtained during August 14 18, 1992, using the f/12 camera of the coudé spectrograph on the 2.2 m telescope at Calar Alto Observatory (Spain). The f/12 camera (focal length 3.6 m) was equipped with an SITe SI-3 CCD detector with pixels, each of 24 µm in size. A 632 grooves-permm grating was used in the 2nd spectral order providing a reciprocal linear dispersion of 2.2 Åmm 1 and a spectral coverage of about 54 Å on the detector. The entrance slit was set to 117 in length and. 7 in width. The resulting spectral resolution (FWHM) is 7 km s 1. The spatial resolution (FWHM) achieved along the slit direction is The sampling rate on the CCD is 2.3 km s 1 and. 187 per pixel in the directions of the dispersion and of the slit, respectively. Long-slit spectra were obtained in three different spectral regions covering the lines of H α,[nii] λλ6548,6583, [O I] λλ63,6363, and [O III] λ57. The position angle (P.A.) of the slit was set by means of an image de-rotator. Hereafter, we will abbreviate north, east, south, and west as N, E, S, W, respectively. [O iii], [N ii], and [O i] will be used to designate the emission lines of [O iii] λ57, [N ii] λ6583, and [O i] λ63, respectively, unless otherwise noted. Since the bipolar axis of M 2-9 is oriented approximately along the N-S direction, slit positions and P.A. s related to that orientation have been selected: namely a) slit positions centred on the central star and oriented along P.A. s,9, and ±24, and b) slit positions offset from the central star by δ = ±1 and oriented at P.A. = 9. Fig. 1 shows these slit positions superimposed upon an HST image of M 2-9. Spectra of H α and [N II] were obatained at all slit positions mentioned above; spectra of [O III] at P.A. s,9, and ±45 ; and spectra of [O I] atp.a.= only. In most cases, the exposure time was 36 s. It should be noted that due to the limited slit length the obtained high-resolution long-slit spectrograms do not cover the faint outer loops of M 2-9. A single deep, moderate-resolution long-slit spectrogram of M 2-9 at P.A. = was secured on May 27, 1988, using the Cassegrain Twin Spectrograph on the 3.5 m telescope at Calar Alto during its commissioning. In this case, a slit aperture of was used, long enough to reach the faint outer loops of M 2-9 as well. The spectrogram covers the spectral range of Å at a spectral resolution of about 45 km s 1. The evaluation of the CCD data frames (flat-fielding, calibration, stellar continuum subtraction, Gaussian line fits) was performed using a special long-slit data reduction package developed by the author. The two-dimensional long-slit spectroscopic data of the observed emission lines are generally presented as logarithmic isophotic contour diagrams in a positionversus-velocity representation ( PV diagrams ). Throughout this paper, all deduced radial velocities are quoted as relative velocities with respect to the systemic velocity V sys of M 2-9 (see Sect ), and angular positions are quoted as relative positions with respect to the position of the central star of M 2-9 (see Sect. 3.3.). 3. Results Fig. 2 displays the high-resolution PV diagrams of H α and [N ii] for the slit position centred on the central star and oriented at P.A. =. It should be noted that in the central core region of the [N ii] map the contributions from the stellar continuum and from the extended H α line wing (see below) have been eliminated using a special subtraction method (for details, see Solf 1994b). The PV diagrams of [O i] and [O iii] for the same slit position are presented in Fig. 3. Figs. 4 and 5 show the PV diagrams of H α and [N ii] for the slit positions centred on the central star and oriented at P.A. s +24 and 24, respectively. Fig. 6 shows the PV diagrams of [N ii] and [O iii] for the slit at P.A. = 9 and centred on the central star. Fig. 7 presents the PV diagrams of H α and [N ii] for the slit positions at P.A. = 9 and offset from the central star by δ = +1 and δ = 1, respectively. A portion of the deep medium-resolution spectrogram presenting the lines of H α and [N II] λλ6548,6583 is shown in Fig. 8. As already mentioned in Sect. 1., three main regions have been distinguished in the morphology of M 2-9: 1) an extended lobes region including the two prominent bipolar lobes; 2) an outer loops region including the two faint nebular loops at about ±1 N and S of the centre; 3) a compact central core region of about 1 extent. The detailed observational results will be presented following that distinction of the different regions of M Bipolar lobes The PV diagrams at P.A. = (Figs. 2, 3) permit us to study the variations of the spatio-kinematic properties in both lobes along the direction of the main axis of symmetry of the bipolar lobes. The PV diagrams obtained at P.A. s ± 24 (Figs. 4, 5) permit us to study those regions which are off the main axis. On the other hand, the P.A. 9 diagrams at δ = ±1 (Fig. 7) provide information on the variations of the spatio-kinematic properties along directions perpendicular to the main axis. All PV diagrams at P.A. s, ±24, and 9 (Figs. 2, 4, 5, 7) show remarkable differences and similarities between both the kinematics and the morphology of the (permitted) H α line emission, on one hand, and those of the forbidden lines, on the other hand. Generally, the forbidden lines present a single, rather narrow component (NC) only, whereas the H α line presents two clearly distinct components: a narrow component (NC) and a rather broad component (BC). Both the kinematics (velocity, velocity width) and morphology (spatial extent, intensity distribution) presented by the forbidden lines are nearly the same as those presented by the NC of the H α emission. In contrast, the

4 J. Solf: A study of the various bipolar outflow components in M M2-9 PA Hα [N II] 2 Relative Position (arc sec) -> PA S3 N3 C- C S3 N3 C Relative Velocity (km/s) Fig. 2. PV diagrams of the line emission of H α (left hand) and [N ii] λ6583 (right hand) of M 2-9 deduced from an f/12 coudé long-slit spectrogram obtained with the slit centred on the central star and oriented at P.A. = (see Fig. 1). Adjacent isophotic contours represent a factor of 2 1/2 in the intensity. Relative velocities and position are quoted with respect to the systemic radial velocity of the nebula and to the position of the central star, respectively. The bright nebular condensations in the northern and southern lobes (N3, S3) intersected by the slit position as well as the resolved line components in the central core region (C,C,C + ) are indicated. 1 BC of the H α emission exhibits different kinematic and morphological properties compared to those of the forbidden lines and the NC of H α. In order to study the spatial distribution and the kinematics of both the NC and the BC in greater detail, the (relative) intensity I, the (mean) velocity V, and the velocity dispersion (FWHM) V, as a function of the relative position along the directions parallel as well as perpendicular to the bipolar axis, have been deduced for each component using a (single-component or twocomponent) Gaussian line fit of the observed line profiles, row by row, on the data frames. Fig. 9 presents the distributions of I, V, and V derived for the NC along the direction of the bipolar axis using the [NII] data at P.A. = (see Fig. 2). Figs. 1 and 11 show the results derived for [NII], the NC of H α, and the BC of H α along the direction perpendicular to the bipolar axis at offsets δ = +1 and δ = 1, respectively, using the corresponding data of [NII] and H α at P.A. = 9 (see Fig. 7). The presence of two components (NC and BC) in the H α emission of the lobes of M 2-9 was first reported by Carsenty & Solf (1983) who interpreted the observed kinematic differences as due to different origins of the NC and BC. The NC was considered as emission originating in situ from the gas in the bright

5 678 J. Solf: A study of the various bipolar outflow components in M 2-9 Relative Position (arc sec) -> PA M2-9 [O I] PA S3 C N3 C+ C- C- S3 C N3 C+ [O III] Relative Velocity (km/s) Fig. 3. PV diagrams of the line emission of [O i] λ63 (left hand) and [O iii] λ57 (right hand). Otherwise as in Fig. 2. lobes, whereas the BC as emission originating from the central core region and being scattered by dust particles associated with the gas of the lobes. In the following, we will describe in detail the spatial and kinematic properties deduced for the NC and BC from the new observations Narrow component (NC) of the line emission As outlined above, the NC of H α and the forbidden lines represent intrinsic emission of the (hot) gas within the bipolar lobes. This implies that the radial velocities deduced from the NC of H α and the forbidden lines represent the (hot) gas flow in the lobes. The neat bipolar morphology of the lobes of M 2-9 strongly suggests that the bipolar outflow velocity is of the same amount in both lobes. This allows us to derive the systemic velocity of the entire nebula V sys by averaging the radial velocities of the two lobes deduced for the NC of H α and the forbidden lines. As already mentioned above, throughout this paper, the radial velocities deduced for the various nebular components have been quoted as relative velocities with respect to V sys. The spatial extent and intensity distribution of the NC are best studied on the PV maps of [NII]. The (half) extent along the direction of the bipolar axis of the [NII] emission (measured from the central star) (see Figs. 2, 9) is about 27, quite comparable to the extent of the NC of the H α emission (see Fig. 2) and to the N S extent of the nebula visible on the HST image of M 2-9 (Fig. 1). The (half) extent along P.A. s ±24 of the NC is about (Figs. 4, 5). The (half) extent perpendicular to the bipolar axis of the NC at δ = ±1 is about 5 6 (Figs. 7, 1, 11).

6 J. Solf: A study of the various bipolar outflow components in M M2-9 PA +24 Hα [N II] Relative Position (arc sec) -> PA N2 N1 C- N2 N1 C C Relative Velocity (km/s) Fig. 4. PV diagrams of the line emission of H α (left hand) and [N ii] λ6583 (right hand) deduced from a long-slit spectrogram obtained with the slit centred on the central star and oriented at P.A. = +24 (see Fig. 1). The bright nebular condensations in the northern lobe (N1, N2) intersected by the slit position are indicated. Otherwise as in Fig Each of the bright nebular condensations (N1 to S3), marked on the HST image (Fig. 1), can be recognised on the various PV maps, namely N3 and S3 on the maps at P.A. = (Figs. 2, 3), N1 and N2 at P.A. = +24 Fig. 4), S1 and S2 at P.A. = 24 maps (Fig. 5), N2 and S2 at P.A. = 9 (Fig. 7). The velocities V NC derived for the NC (see Figs. 9 11) are nearly constant within each lobe, although some characteristic variations are apparent (see below). Mean velocities of the northern lobe V N NC = +12 km s 1 and of the southern lobe V S NC = 12 km s 1 have been deduced for [NII]. The corresponding values for the NC of H α are ±1 km s 1, respectively. The velocity dispersion V NC presented by the [NII] emission in the lobes is generally quite small, typically 18 2 km s 1 or less. The larger values of V NC observed in the NC of H α are explained as due to the lower atomic weight of H compared to N +. Both (the absolute values of) V NC and V NC are somewhat larger in the nebular condensations N3 and S3 compared to their neighbourhood (Fig. 9). If one neglects these local anomalies, the absolute values of the velocity of the NC present a slight increase along the bipolar axis as one proceeds toward larger distances from the centre (Fig. 9). On the other hand, the distribution of V NC perpendicular to the bipolar axis shows a different behaviour. In both lobes, the absolute values of V NC exhibit their maximum near the bipolar axis and decrease in either direction toward the lateral boundaries of the lobes (Figs. 1,

7 68 J. Solf: A study of the various bipolar outflow components in M M2-9 PA -24 Hα [N II] Relative Position (arc sec) -> PA S1 S2 C C- S1 S2 C Relative Velocity (km/s) Fig. 5. PV diagrams of the line emission of H α (left hand) and [N ii] λ6583 (right hand) deduced from a long-slit spectrogram obtained with the slit centred on the central star and oriented at P.A. = 24 (see Fig. 1). The bright nebular condensations in the southern lobe (S1, S2) intersected by the slit position are indicated. Otherwise as in Fig ). The condensations N2 and S2 (near the eastern edge of each lobe) follow that trend as well (see Figs. 4, 5, 7). The velocities observed in the condensations N1 and S1, covered by the slit positions at P.A. s ±24 (see Figs. 4, 5), show an interesting behaviour. The (absolute) velocity of S1 is unexpectedly large (larger than that of S2), whereas the velocity of N1 is unexpectedly small (smaller than that of N2) and presents the opposite sign (compared to N1). This implies that either N1 and S1 is blueshifted with respect to N2 and S2, respectively. It should be noted that bridges of weak line emission between N1 and N2 and between S1 and S2 are clearly visible on the [NII] maps at P.A. s ±24 (Figs. 4, 5) Broad component (BC) of the H α emission The BC of the H α emission (see Figs. 2, 4, 5, 7) appears to be considerably broader than the NC of H α and the forbidden lines. Typical velocity widths V BC of 7 9 km s 1 have been deduced for the BC, more than a factor of 4 larger than derived for the V NC (see also Figs. 1, 11). Moreover, in both lobes, the BC is always redshifted by a nearly constant amount with respect to the NC (see Figs. 1, 11). Mean values of V N BC = +3.7 km s 1 and V S BC = km s 1 have been deduced for the BC of the H α emission in the northern and southern lobe, respectively.

8 J. Solf: A study of the various bipolar outflow components in M Relative Position (arc sec) M2-9 PA 9 C- C- C C [N II]λ6583 [O III]λ57 C+ C Relative Velocity (km/s) Fig. 6. PV diagrams of the line emission of [N ii] λ6583 (top) and [O iii] λ57 (bottom) deduced from a long-slit spectrogram obtained with the slit centred on the central star and oriented at P.A. = 9 (see Fig. 1). The resolved line components in the central core region (C, C,C + ) are indicated. Otherwise as in Fig. 2. Large differences between the BC and the NC are also apparent in their spatial extents and intensity distributions along the directions of the different P.A. s. As indicated on the H α map at P.A. = (Fig. 2), the (half) extent of the BC (from the centre) along the bipolar axis is about 18 2, considerably less than that of the NC (see above). The (half) extent of the BC along P.A. s ±24 (Figs. 4, 5) is about 15 16, somewhat larger than that of the NC. The (half) lateral extent (perpendicular to the bipolar axis) of the BC at δ = ±1 (Fig. 7) is about 6 7, somewhat larger than that of NC. In all cases, the intensity distribution of the BC along the various P.A. s is nearly constant (see also Figs. 1, 11). It should be emphasised that none of the bright nebular condensation (N1 to S3), which are visible on direct images of M 2-9 (see Fig. 1) and recognised on the PV maps as prominent maxima in the NC of the H α and the forbidden line emission, is observable in the BC of the H α emission Faint outer loops The faint outer loops, located on the bipolar axis of M 2-9 and extending up to about 1 from the central core, have been covered by the slit position of the medium-resolution long-slit spectrogram obtained at P.A. = (Fig. 8). The spectrogram presents weak H α emission at the positions of both the northern loop (OLN) and the southern loop (OLS). Schwarz et al. (1997) reported the presence of very faint [N ii] λ6583 emission in the southern loop which is not present on our spectrogram, probably because of a slightly different P.A. of the slit. As already noted by Schwarz (199) and Solf (1993), both loops present redshifts of large but different amounts. From our data we derive radial velocities of Vloop N = 191 km 1 and Vloop S 1 = 18 km for the northern and southern loop, respectively, in good agreement with the results reported by Solf (1993) and Schwarz et al. (1997) Central core region of M 2-9 The PV maps of the forbidden lines at P.A. = (Figs. 2, 3) and at P.A. s ±24 (Figs. 4, 5) clearly demonstrate the differences between the central core region, dominated by a complex broad line emission, and the region of the bipolar lobes, dominated by a narrow line emission. Compared to the lobes, the [N ii] emission of the core region is weaker (Fig. 2), whereas the [O iii] and, specifically, the [O i] emissions are considerably stronger (Fig. 3). The H α emission detected in the region of the central core of M 2-9 (Figs. 2, 4, 5) is dominated by a rather strong stellar P-Cyg-type line feature. Since that emission is probably originating from the immediate vicinity of the exciting central star, the centroid position X of the spatial distribution (along the slit direction) of the recorded H α emission, determined by a Gaussian fit, is likely to represent the centroid position of the exciting central star of M 2-9. As already mentioned above, in this paper, positions within M 2-9 have been generally quoted as relative positions with respect to X. (In cases the spectrograms do not include the H α line, X has been obtained by applying a Gaussian fit to the spatial distribution of the stellar continuum.) As already noted by Balick (1989) the H α emission presents extremely broad line wings extending up to about ±5,5 km s 1 at the base. Using a two-component Gaussian line fit of the P Cyg line profile, the radial velocity of the emission component V em =.4 km s 1 and that of the absorption component V abs = 11.6 km s 1 with respect to the systemic velocity V sys (see Sect ) have been derived. By integration over the entire H α line profile the effective (centroid) radial velocity of the H α emission V eff = +8.9 km s 1 has been obtained. (It should be noted that, below in Sects and 4.2., the latter will be adopted as the rest velocity of the illuminating source as seen by the dust particles in the outflows.) Since the strong stellar H α emission engulfs any nebular H α emission from the core region, the distibution and kinematics of the nebular gas in the core region cannot be derived from the H α but from the forbidden line emission only. The [N ii] line feature, visible in the central core region on the maps at the various P.A. s (Figs. 2, 4 6), presents a rather complex structure extending over a full width at zero intensity (FWZI) of about 175 km s 1. The line feature has been spectrally resolved into three components: a prominent (cen-

9 682 J. Solf: A study of the various bipolar outflow components in M 2-9 Relative Position (arc sec) -> PA M2-9 H N2 α M2-9 PA 9 PA 9 δ +1" δ +1" M2-9 S2 H α M2-9 PA 9 PA 9 δ-1" δ -1" N2 S2 [N II] [N II] Relative Velocity (km/s) Fig. 7. PV diagrams of the line emission of H α (left hand column) and [N ii] λ6583 (right hand column) deduced from a long-slit spectrogram obtained with the slit oriented at P.A. = 9 and offset by δ = +1 (upper row) and δ = 1 (lower row) with respect to the position of the central star (see Fig. 1). The bright nebular condensations in the northern and southern lobe (N2, S2) intersected by the slit positions are indicated. Otherwise as in Fig. 2. tral) low-velocity component (LVC) and two faint (satellite) high-velocity components (HVCs). The LVC will be designated as C, the redshifted HVC as C +, and the blueshifted HVC as C. It can be noticed that C + is generally somewhat fainter compared to C. A close inspection of the maps at P.A. s and ±24 (Figs. 2, 4, 5) reveals that the HVCs have been spatially resolved as well. C + and C are spatially extended and are pointing (from the centre of the core) into opposite directions. Remarkably, comparable offsets of C + and C (toward the E or W) are not visible on the [N ii] map at P.A. = 9 (Fig. 6). These findings indicate that the centroids of C + and C are offset towards the N and S, respectively. The presence of two satellite components in the [N ii] line emission from the core region has been noted previously by Balick (1989) and Solf (1993). For each PV map at P.A. s, ±24, and 9, threecomponent two-dimensional Gaussian line fits were applied to the observed [N ii] line features of the core region, in order to derive, for each component, C +,C, and C, the (relative) peak intensity I, the (centroid) radial velocity V, the velocity dispersion (FWHM) V, the angular position X, and the angular width (FWHM) X. The results of the fits of the [N ii] line features obtained for P.A. s and 9 are listed in Table 1. V has been quoted relatively to the systemic velocity V sys, and X relatively to the position of the central star X. V and X have been corrected for the spectral and spatial resolution, respectively. It should be noted that the (absolute) values of X derived for C + and C at P.A. s ±24 (not shown in Table 1) are about 3 percent smaller than those quoted for P.A. =. These results clearly indicate that the maximum and minimum offsets X of the HVCs occur along the N-S direction and the E-W direction, respectively. Hence the elongation of the core region of M 2-9 is oriented along the direction (N-S) of the main axis of symmetry of the bipolar lobes. More specifically, the resolved redshifted C + and blueshifted C are pointing into opposite directions and hence are likely to represent the two components a bipolar outflow system (see Sect below).

10 J. Solf: A study of the various bipolar outflow components in M Relative Position (arc sec) -> PA OLN M PA Hα N[II]6548 N[II] OLS -4 4 Relative Velocity (km/s) Fig. 8. A section of a deep moderateresolution long-slit Cassegrain spectrogram of M 2-9 obtained with the slit centered on the central star and oriented at P.A. =. The section presents the spectral range of the H α and [N ii] lines. Adjacent isophotic contours represent a factor of 2 in the intensity. The H α emission from the faint outer loops is marked (OLN, OLS). Deduced radial velocities are quoted in km 1. Otherwise as in Fig. 2. It is noteworthy that the [N ii] results at P.A. = listed in Table 1 permit us a comparison with previous results from a high-resolution spectrogram of M 2-9 at P.A. = obtained by the author in 1983 using the same spectrograph then equipped with an image-tube detector system. The old image-tube spectrogram was re-evaluated and a two-dimensional Gaussian fit applied to the recorded [N ii] line feature, in order to obtain the (centroid) positions X of the components C,C, and C +.A comparison of the 1992 and 1983 results indicates that C and C + have moved away from the centre into opposite directions at a rate of about. 12 yr 1 and. 27 yr 1, respectively. These proper motions imply a total flowing time from the centre (up to the epoch of the 1992 observations) of 27 years and 22 years for C + and C, respectively. The [O iii] line features observed in the core region at P.A. s and 9 (Figs. 3, 6) have been spectrally and spatially resolved as well. They exhibit a similar structure as the [N ii] feature consisting of a LVC (C ) and two HVCs (C + and C ). In this case, however, C + is rather faint, whereas C is rather strong, even stronger than C. Compared to the situation in [N ii], the brightness ratio of C + /C in [O iii] is significantly smaller. Similarly to the case of [N ii], components C + and C observed in [O iii] atp.a.= (Fig. 3) exhibit small spatial offsets toward the N and S, respectively, whereas at P.A. = 9 (Fig. 6), such offsets (toward the E or W) are not present. As the case of [N ii], two-dimensional multi-component Gaussian line fits were applied to the observed [O iii] line features of the core region at P.A. s and 9. The results of the fits, except for X, are also listed in Table 1. ( X has been omitted, because a correction for the actual spatial resolution was not feasible in this case.) The [O i] line feature observed in the core region at P.A. = (Fig. 3) exhibits a different structure compared to [N ii] and [O iii]. In this case, the line feature is dominated by a single bright, rather narrow LVC (C ). Extremely faint extended line wings on either side of the LVC are barely visible on the PV diagram, suggesting that rather weak HVCs (C + and C ) might be present, though a definite claim cannot be made. A centroid velocity V of 7.7 km s 1 and a velocity width V of 32 km s 1 have been deduced for C. The PV diagrams of [N ii] and [O iii] obtained at P.A. = 9 with the slit centred on the central star (Fig. 6) present line emission in the core region only. Outside that region (perpendicular to the bipolar axis of M 2-9) extended nebular emission in the forbidden lines or in the H α line (diagram not shown) has not been detected. 4. Discussion 4.1. Double-shell structure of the bipolar lobes Optical polarisation maps of M 2-9 by King et al. (1981) and Aspin & McLean (1984) revealed the existence of a dust shell associated with the lobes. The polarisation is produced by scattering dust grains illuminated by light from the central core region. Spectropolarimetry of M 2-9 by Schmidt & Cohen (1981) showed that the polarisation of the forbidden lines in the bipolar

11 684 J. Solf: A study of the various bipolar outflow components in M 2-9 Table 1. Peak intensity I, centroid radial velocity V, velocity dispersion (FWHM) V, centroid angular position X, and angular width (FWHM) X of the resolved line components C,C,C + in the central core region. Component Line P.A. = P.A.=9 I V V X X I V V X X km s 1 km s 1 arcsec arcsec km s 1 km s 1 arcsec arcsec C [N II] C [N II] C + [N II] C [O III] C [O III] C + [O III] 26: +34: 4: +.11:... 26: +34: 4: -.3:... Velocity Width (km/s) Intensity Relative Velocity (km/s) M 2-9 [N II]6583 PA S3 C Velocity Width (FWHM) Relative Velocity Relative Intensity Relative Position (arc sec) -> PA Fig. 9. Relative intensity I (bottom), radial velocity V (centre), and velocity dispersion (FWHM) V (top) derived for the [N ii] λ6583 line, as a function of the relative position with respect to the central star, along the direction of the bipolar axis at P.A. = (see Fig. 1). The bright nebular condensations in the northern and southern lobes (N3, S3) and the central core (C) are indicated. N3 lobes is much lower than that of the permitted lines and the stellar continuum. This result indicates that (most of) the observed forbidden line emission arises in situ from the hot gas within the lobes, whereas a major fraction of the permitted line emission and continuum radiation represents dust-scattered light originally produced near the central star. This implies that two gas components in each lobe must be present, namely a component of hot (ionised) gas and a component of cool (molecular) gas and dust. Imaging polarimetry in the H α line by Scarrott et al. (1993) and high-resolution near-infrared imaging by Hora & Latter (1994) revealed a double-shell structure in each lobe, consisting of an elongated cavity-like inner shell of hot gas and a co-axial envelope-like outer shell of molecular gas and dust. High-resolution spectroscopy of M 2-9 by Carsenty & Solf (1983) showed that the permitted lines and the forbidden lines in the lobes exhibit different kinematical properties as well. These results indicate that the two gas components of the double-shell structure can be studied by an analysis of the spectral properties of the permitted lines and the forbidden lines observed in the lobes. As outlined in Sect. 3.1., the H α line emission in the bipolar lobes presents a complex structure, consisting of a narrow component (NC) and a broad component (BC), whereas the forbidden lines present a single NC only. The NC and BC of H α exhibit different kinematic properties (centroid velocities, velocity widths) which have been extracted by means of a twocomponent Gaussian fit. As already shown above, the derived spatial distribution and the kinematical properties of the NC of H α are quite similar to those of the forbidden lines. On the other hand, the BC of H α presents a much larger velocity width compared to and is redshifted with respect to the NC and the forbidden lines. These results suggest that the NC represents line emission originating in the hot (ionised) gas component, whereas the BC represents line emission originating in the core region of M 2-9 and being scattered by dust particles associated with the cool (molecular) gas component of the lobes. In the following we will discuss in detail the different morphological and kinematical properties of the two gas components based on an analysis of the observed BC and NC. These properties allow us to apply bipolar outflow models in order to derive the true bipolar outflow velocities for each of the two gas components Morphology of the cool-gas/dust component and the hot-gas component of the double-shell structure The long-slit spectroscopic results indicate that the cool gas/dust component in the lobes exhibits a spatial distribution which

12 J. Solf: A study of the various bipolar outflow components in M Velocity (km/s) Velocity Dispersion (km/s) Intensity M2-9 [N II] 6583 Hα NC PA 9 δ +1" H α BC Velocity Dispersion Relative Velocity Relative Intensity N2 N Relative Position (arc sec) -> PA 9 Fig. 1. Relative intensity I (lower row), radial velocity V (central row), and velocity dispersion (FWHM) V (upper row) derived for the [N ii] line (left hand column), the narrow component (NC) of the H α line (central column), and the broad component (BC) of the H α line (right hand column),asa function of the relative position with respect to the position of the bipolar axis, along the direction of the slit oriented at P.A. = 9 and offset by δ = +1 (see Fig. 1). The bright nebular condensation in the northern lobe (N2) intersected by the slit position is indicated. Velocity Dispersion (km/s) Intensity Velocity (km/s) M2-9 PA 9 δ 1" Velocity Dispersion Relative Velocity S2 Relative Intensity [N II] 6583 H NC α H α BC Relative Position (arc sec) -> PA 9 S2 Fig. 11. Relative intensity I (lower row), radial velocity V (central row), and velocity dispersion (FWHM) V (upper row) derived for the [N ii] line (left hand column), the narrow component (NC) of the H α line (central column), and the broad component (BC) of the H α line (right hand column) as a function of the relative position with respect to the position of the bipolar axis, along the direction of the slit oriented at P.A. = 9 and offset by δ = 1 (see Fig. 1). The bright nebular condensation in the southern lobe (S2) intersected by the slit position is indicated. in general resembles the bipolar morphology presented by the hot gas component, although some significant differences are apparent. Compared to the hot gas component, the cool gas/dust component a) appears to be less extended along the direction of bipolar axis, b) exhibits a somewhat larger opening angle, and c) does not present regions of enhanced brightness. These findings can be best understood if the cool gas/dust is forming a (relatively thin) envelope-like outer shell which is embedding a cavity-like inner shell filled up with ionised hot gas. Only along directions near the bipolar axis, the hot gas of the inner shell is reaching much beyond the boundary of outer shell. The bright nebular condensations visible in the lobes clearly belong to the (inner) hot-gas component. Some of them, N2 and S2, appear to be located at the boundary layer of the inner shell; others, N3 and S3, appear the be located whithin the inner shell close to the position of the bipolar axis Outflow model of the bipolar cool-gas/dust shell As outlined above, the cool-gas/dust component of the bipolar lobes is traced by the BC of H α. The BC represents emission originating from the central core region and being scattered by

13 686 J. Solf: A study of the various bipolar outflow components in M 2-9 dust particles contained in the outflowing cool gas of the outer shell. Using a simple bipolar outflow model for the scattering particles in the lobes we can relate the outflow velocity V exp dust of the cool gas/dust and the inclination angle Θ dust of the bipolar outflow axis (with respect to the line-of sight) to the radial velocities Vdust N and V dust S of the dust particles observed in the BC of H α in the northern and southern lobe, respectively: Vdust N = V eff + V exp dust (1 + cos Θ dust), (1) Vdust S = V eff + V exp dust (1 cos Θ dust), (2) where V eff denotes the effective velocity (with respect to the systemic velocity V syst ) of the central illuminating H α line source caused by the P Cyg line profile seen by the scattering dust particles. From Eqs. (1) and (2) we obtain: V exp N dust =.5 (Vdust + Vdust) S V eff (3) and Θ dust = arccos( V dust N dust S 2 V exp ). dust (4) Inserting our results Vdust N = VBC N +31 km s 1 and V S dust = V BC S +2 km s 1 (see Sect ), and assuming that the dust particles see virtually the same P Cyg profile as the terrestrial observer, i.e., V eff +9 km s 1 (see Sect. 3.3.), we obtain an outflow velocity of the cool-gas/dust shell V exp dust 17 km s 1 and an inclination angle of the bipolar outflow Θ dust 71. It should be noted that the deduced value of V exp dust is somewhat lower than that reported by Carsenty & Solf (1983), who did not take into account the effective velocity V eff of the central illuminating source. Adopting a distance of 65 pc for M 2-9 (Schwarz et. al. 1997) and using the deduced V exp dust and the (half) extent (along the bipolar axis) of about 2 observed in the BC (see Sect ), we obtain a kinematic age of about 38 years for the cool-gas/dust shell Outflow model of the bipolar hot-gas shell The hot (ionized) gas of the bipolar lobes is traced by the NC of H α and the forbidden lines. In order to derive the outflow velocity V exp ion of the ionized gas and the inclination angle Θ ion of the bipolar axis, we adopt a bi-conical outflow model (see Solf & Böhm 1999) assuming that the velocity dispersion (FWHM) V ion, presented by the NC of H α and the forbidden lines, is caused by the divergence of the velocity vectors of the outflowing gas, filling up the conical volume of each lobe. If Φ ion denotes the (half) aperture angle of the bi-conical shell and V ion the mean (absolute) value of the radial velocity of the ionized gas in the lobes we obtain: V ion Θ ion = arctan( ) (5) 2 V ion tan Φ ion and V exp ion = V ion cos Φ ion cos Θ ion. (6) 46 km s 1 for the hot (ionized) gas of the lobes and an inclination angle Θ ion 73 for the bipolar axis. For the adopted distance of M 2-9 the deduced outflow velocity indicates a (mean) kinematic age of about 13 years for the hot-gas shell (in the considered zone) of the bipolar lobes. As mentioned in Sect , the (absolute) radial velocities in the regions near the (projected) bipolar axis are systematically higher than in those near the lateral edges of the lobes. Moreover, the radial velocities of the bright nebular condensations N3 and S3 (observed near the bipolar axis) are systematically higher than those of the neighboring regions on the axis. These observations suggest that the outflow velocity of the hot gas of the lobes is largest along the polar axis and is decreasing at lower latitude angles. Furthermore, if N3 and S3 are indeed located on the bipolar axis, their true flow velocity must be even higher than the (mean) polar flow velocity of the surrounding material in their neighborhood. On the other hand, the fact (mentioned above) that the line emission from condensation N1 and S1 is blueshifted with respect to that of N2 and S2, respectively, indicates that both N1 and S1 are probably located on the front boundary layer (pointing toward the observer) of the hot-gas shell. Our results indicate that the orientation in space of the bipolar shell of the hot (ionized) gas is approximately the same as that of the shell of the cool gas/dust, i.e., Θ dust Θ ion, suggesting that both shells have the same bipolar axis. On the other hand, the (polar) outflow velocity of the hot-gas shell is much larger (by a factor of 3) than that of the cool-gas/dust shell. This implies that the two shells must be spatially separated in the sense that the (younger) shell of the (fast) hot gas is embedded within the (older) shell of the (slow) cool gas. Only along directions near the bipolar axis, the fast gas is breaking through the slow-gas shell, leading to a much more elongated structure of the hot-gas shell compared to the cool-gas shell. It should be noted that this cocoon-like structure is well compatible with the results from the analysis of the morphology of the double-shell (see Sect ). Because of the (expected) rather large inclination of the bipolar outflow axis, the aperture angle Φ ion of the bi-conical shell can, in principle, be derived from direct images of M 2-9 in the light of the forbidden lines. However, due to the specific morphology of M 2-9, the (effective) aperture angle is not constant if one proceeds along the bipolar axis. Hence it is useful to select a restricted zone in each lobe, where the aperture angle is well determined. Selecting, e.g., the zone between 15 and 2 (from the central core), we derive a (mean) Φ ion of about 13 (see Fig. 1), a (mean) V ion = V NC of about 13 km s 1, and a (mean) V ion = V NC of about 2 km s 1 for [N II] (see Fig. 9). Using these results we obtain a (mean) outflow velocity V exp ion 4.2. Spatio-kinematic structure of the faint outer loops It has been suggested by Solf (1993) and confirmed by Schwarz et al. (1997) that the H α line emission detected in faint outer loops represents dust-scattered light originating from the central core region of M 2-9. Hence we can apply Eqs. (3) and (4) in

14 J. Solf: A study of the various bipolar outflow components in M order to deduce the bipolar outflow velocity of the scattering dust/gas in the outer loops V exp exp loops (instead of Vdust ) and the inclination angle Θ loops (instead of Θ dust ) of the bipolar axis. Inserting the radial velocities Vloop N and V loop S N (instead of Vdust and V S ) derived for northern and southern loop, respectively, dust (see Sect. 3.2.) and taking into account the effective velocity V eff of the illuminating source (see Sect. 3.3.) we obtain V exp loops 141 km s 1 and Θ loops 73. These results are in good agreement with those reported by Solf (1993) and Schwarz et al. (1997). It is remarkable that the deduced orientation in space of the bipolar outflow observed in the outer loops is the same as those of the outflows in double-shell structure of the bipolar lobes, i.e., Θ loops Θ dust Θ ion, whereas the outflow velocity V exp loops of the loops is larger by a factor of 3 than V exp ion of the hot (ionized) gas and larger by a factor of 8 than V exp dust of the cool gas/dust in the double-shell structure of the lobes. For the adopted distance of M 2-9, the derived outflow velocity of the faint loops and their (angular) separation from the central core indicate a kinematic age of about 13 years for the outer loops. It is most remarkable that this age is much less than that of the cool-gas/dust shell but is about the same as that of the hot-gas shell of the bipolar lobes. These findings suggest that the outer loops and the shell of (fast) hot gas of the bipolar lobes are generically related to each other, whereas the shell of (slow) cool gas and dust has been formed in a different process at an earlier stage Structure of the central core region High-velocity bipolar micro-jets As shown in Sect. 3.3., the line features of [N ii] and [O iii] observed in the core region of M 2-9 present a complex structure which has been spectrally and spatially resolved into three components, namely two HVCs (C +,C ) and a single LVC (C ). The results of a two-dimensional Gaussian fit (see Table 1) indicate that the redshifted C + and the blueshifted C are elongated along the N S direction and that their position centroids X are offset (from the position of the central star) toward the N and S, respectively. This result suggests that C + and C are likely to represent the two components of a high-velocity bipolar outflow which is oriented along the N S direction and originating from the center of the core. The deduced spatial characteristics of C + and C (see Table 1) indicate that the outflow is collimated or jet-like. Moreover, the small angular offsets X and small widths X deduced for C + and C suggest that the bipolar outflow system in the core region is rather compact ( micro-jets ). Remarkably, the (projected) orientation (N S) of the microjets is the same as that of the prominent bipolar lobes and that of the faint outer loops. Moreover, the deduced (centroid) radial velocities V of C + and C (see Table 1) indicate that the bipolar axis of the jets Θ jets is inclined in the sense that the southern jet and the northern counter-jet are pointing towards and away from the observer, respectively. A similar orientation of the flow axis is observed in the bipolar lobes and the faint outer loops. Hence it is most likely that the inclination angle of the jets Θ jets is about the same as that derived for the bipolar lobes and the outer loops, i.e., Θ jets 73. Adopting the latter value and using the results of the Gaussian fit (V, V ) obtained for C + and C (see Table 1), we can apply Eqs. (5) and (6) in order to deduce the flow velocity of the jets V exp jets and the (half) opening angle of their collimation Φ jets. In the case of [O iii], we derive V exp jets 137 km s 1 and Φ jets 7 ; the corresponding values for [N ii] are 195 km s 1 and 7, respectively. Hence the flow velocity of the micro-jets is higher by at least a factor of 3 compared to that of the hotgas component of the bipolar lobes, but is of approximately the same magnitude as that of the faint outer loops. The deduced opening angle Φ jets is remarkably small, much smaller than that of the bipolar lobes, suggesting a higher collimation of the high-velocity flow in the core region. It is noteworthy that the opening angle subtended by the bright nebular condensations N3 and S3 on the bipolar axis (see Fig. 1) appears to be of approximately the same magnitude as that deduced for the micro-jets. Therefore it is tempting to suggest that N3 and S3 represent regions of interaction of a highly collimated fast bipolar outflow from the center with slower material within the lobes. The fact that the (absolute) radial velocities deduced for N3 and S3 are somewhat higher than those observed in their immediate neighborhood supports that suggestion. Surprisingly, the flow velocity of the micro-jets deduced for the [N ii] emission is higher than that deduced for the [O iii] emission. It should be noted that the higher and lower velocities of the jets observed in the [N ii] and [O iii] emission, respectively, correspond to the larger and smaller offsets of the centroid positions X of the jets, derived for the respective lines. These findings suggest that some acceleration process may be present in the core region leading to a downstream stratification of the jet material. It seems that the material in the vicinity the source, which is characterized by higher excitation and lower velocity, is headed by the material at larger distances, which is of lower excitation and of higher velocity. The stratification along the outflow direction of the jets resembles a situation generally observed in the shell structure of PNs presenting both an increase of the expansion velocity and a decrease of the excitation if one proceeds to larger distances from the central star. As noted in Sect. 3.3., the HVCs in the core of M 2-9 present rather weak [O i] line emission, if any, but a rather strong [O iii] emission, thus indicating relatively high excitation of the microjet gas. This result and the stratification effect mentioned above are more typical for photo-ionized gas regions, like PNs, but less typical for regions dominated by shock excitation, like jets from young stellar objects. Thus it is likely that the main source of excitation of the jet gas in the core region of M 2-9 is photoionization rather than shock interaction. The high velocity and the compactness of the micro-jets indicate an extremely small kinematic age. For the adopted distance of 65 pc, the resulting kinematic age of the visible micro-jets is 5 years only. This age is significantly less than the flowing time of years obtained from the deduced