planetary nebula NGC 2440
THE DETECTION OF A MOLECULAR BIPOLAR FLOW IN THE MULTIPOLAR PLANETARY NEBULA NGC 2440
Mei-Yan Wang and Tatsuhiko I. Hasegawa
Institute of Astronomy and Astrophysics, Academia Sinica, Taipei 106, Taiwan; [email protected], [email protected]
and Sun Kwok
Department of Physics, University of Hong Kong, Hong Kong, China; [email protected] Received 2007 February 20; accepted 2007 October 1
ABSTRACT
We report the detection of a molecular bipolar outflow in the CO J¼ 3Y2 line in the optical multipolar planetary nebula NGC 2440. The observed CO J¼ 3Y2 emission shows two major velocity components which are blue- and redshifted by 18 km s1relative to the systemic velocity of VLSR¼ 43 km s1. Both components are detected only near the central region and have a size of about 2400. In addition to this bulk motion, weak CO emissions in the central region are detected at higher velocities of20 to 40 km s1. The presence of the fast CO components are in line with the picture that the neutral dense torus is being torn apart by very fast outflowing winds. The spatial distribution of CO emission shows two emission peaks at the systemic velocity that are located 3000to the northeast and 3000to the south-west from the center in point symmetry. The positions of the two CO emission peaks coincide well with the emission knots in the optical, suggesting that the CO outflow corresponds to one pair of the multiple optical bipolar lobes of the nebula. This particular pair of bipolar lobes seem distinctly rich in heavy molecules compared with the other optical bipolar lobes. The extent of CO emission in NGC 2440 is well constrained. The total amount of CO molecules in NGC 2440 is 3:5;1051if the distance to NGC 2440 is 2.19 kpc and the excitation temperature is 25 K.
Subject headinggs: ISM: molecules — planetary nebulae: individual (NGC 2440) Online material: color figures
1. INTRODUCTION
Our understanding of the morphologies of planetary nebu-lae ( PNe) has undergone a significant change in the last decade as the result of deep CCD imaging and observations with the Hub-ble Space Telescope ( HST ). When the effects of projections and ionization are considered, the apparent morphologies of many PNe can be explained by an equatorial torus and two bipolar lobes (Balick1996; Bryce et al. 1994; Zhang & Kwok 1998). This in-trinsic structure of PNe is widely interpreted as the result of a fast outflow interacting with an asymmetric circumstellar enve-lope of the asymptotic giant branch (AGB) progenitor star ( Balick 1987).
In spite of the success of the interacting winds model, obser-vations in the past 10 years have revealed a number of PNe that cannot be explained by this simple model. HST images of young PNe show that a number of PNe have multiple pairs of lobes (Sahai 2000) and that some PNe with apparently typical bipo-lar structures actually have more than one outflow axes. A phe-nomenological interpretation of these PNe is multiple episodic outflows ( Lo´pez et al. 1995), and there are other phenomena such as highly collimated jets associated with the multipolar PNe ( Kwok 2004). NGC 2440 is one of the most dramatic examples of this class of objects ( Kwok 2004).
NGC 2440 is one of representative extreme Type I planetary nebulae ( Kingsburgh & Barlow 1994; Perinotto 1991). This PN is characterized by an N/O ratio greater than 1 (Kingsburgh & Barlow 1994; Zuckerman & Aller 1986), a hot central star with TeA¼ 170; 000Y200; 000 K (Shields et al. 1981; Kwitter & Henry 1996;
tary nebula was once classified as a simple bipolar nebula, but is now considered as a representative multipolar nebula (Cuesta &
Phillips 2000; Lo´pez et al. 1998). The multipolar lobes have been identified in the optical (Cuesta & Phillips 2000; Lo´pez et al. 1998).
The high velocities of order 200 km s1of the ionized gas, detected in optical forbidden lines, are attributed to the multipolar lobes or flows rather than to the heavy mass loss or superwind during the AGB phase.
The optical studies of NGC 2440 with dramatic improvements in sensitivity, angular resolution, and sophistication have not been matched by molecular studies. Many PNe have a significant amount of molecular contents ( Huggins et al. 1996). NGC 2440 is no ex-ception. An observational study of the molecular contents of a PN gives us important insight into the kinematical and physical states of the PN, in particular optically invisible parts of the PN.
Our understanding of molecular distribution in this source has been sketchy. H2observations in the IR only probe excited ( by colli-sion at a few 1000 K or UV radiation) molecular regions. Cold (10Y1000 K ) and dense molecular gas needs to be probed with a rotational line of CO.
CO was first detected in this source by Huggins et al. (1996) and by Dayal & Bieging (1996). The CO( 2Y1) spectra observed by the two groups consistently show a double-peak shape. The CO observations have been limited to seven points, even though the optical size of NGC 2440 is about 8000 and H2observations hint a widespread, if fragmented, presence of molecular gas in this planetary nebula ( Latter & Hora 1997).
We have searched for cold molecular gas in the CO( J¼ 3Y2) line over a 6000; 7000area with a fine sampling interval (600) in A
The Astrophysical Journal, 673:264Y270, 2008 January 20
#2008. The American Astronomical Society. All rights reserved. Printed in U.S.A.
and to detect a bipolar molecular structure associated with one optical bipolar flow.
2. OBSERVATION
The CO(3Y2) observations were made with the James Clark Maxwell Telescope (JCMT) in 2000 October, 2001 October, and 2003 October. The JCMT program ID was M00BC06, M01BC07, and M03BC13. Common user receiver RxB3 was used. RxB3 is a single-sideband, dual-polarization channel receiver with a typical system temperature of about 500 K. The spectrometer was DAS ( Dutch Autocorrelator Spectrometer), which was a standard spec-trometer for common use on the JCMT during the above stated period. The CO(3Y2) spectra were originally obtained with a fre-quency resolution of 378 kHz (a channel interval of 312.5 kHz) and an instantaneous bandwidth of 250 MHz. Sky subtraction was made in a beam switching mode with an azimuth beam throw of 12000and at a cycle of 1 Hz.
In total, 121 positions were observed, covering a region of about 6000; 7000 centered at ( ; )¼ (07h41m55:41s;1812030:700) (J2000.0). Each position was observed as a part of a grid map of a typical size of 5; 5 or 3 ; 9 grid points with a common grid in-terval of 600. The integration time was 30 or 60 s at a time for each position. The grid mapping was repeated 5Y12 times, so that the to-tal integration time for each position was about 6 minutes for each
SCUBA calibration and imaging document,1OH231.8 is not re-solved at 850 or 450 m with the SCUBA on the JCMT. There-fore, OH231.8 can be regarded as a pointlike source for pointing calibration at 345 GHz for the JCMT.
At the frequency of CO(3Y2), 345.7960 GHz, the JCMT has a half-power beam width (HPBW ) of 1400. We adopt a main-beam efficiency Bof 0.63, which was a JCMT standard facility value at 345 GHz (Matthews et al. 2004). All intensities are expressed in main-beam temperature TMB¼ TA/B in this paper.
During data reduction, the spectra from the two polarization channels were averaged, and eight channels were binned. The frequency resolution of the CO(3Y2) spectra presented here is 2.5 MHz (2.17 km s1). Only linear baselines were removed from the individual spectra, where the adopted baseline regions were (1) VLSR¼ 60Y0 km s1, and (2) VLSR¼ 80Y140 km s1. The rms noise levels MBof the individual spectra presented in the present work are typically about 0.048 K (0.03 K in TA) at the frequency resolution of 2.5 MHz.
3. RESULTS
3.1. CO(3Y2) Spectra and Maps
The CO(3Y2) emission was detected at about 45 positions, where the peak intensities are 0.16Y0.65 K in TMB. Representative spectra are shown in Figure 1 for positions ( ; ) ¼ (0; 0), (+1800, +2400), (1800,2400), (+1200, +600), and (1200,600). The peak intensities, noise levels (MBin TMB), and integrated inten-sities are given in Table 1 for the five positions.
The strongest CO(3Y2) emission is found at the (0, 0) position.
The spectrum at the center has (1) a full-width at zero intensity (FWZI) of 75 km s1, (2) a line center velocity of 43 km s1in VLSR, and (3) two narrow components at VLSR¼ 26 and 61 km s1. In addition, the spectrum shows two weaker but distinct narrow components at 42 and 51 km s1. These features are consistent with the previously reported observations by Huggins et al. (1996) and by Dayal & Bieging (1996). The systemic velocity for the molecular component from the present observation is thus VLSR¼ 43 km s1. The 26 and 61 km s1components in Figure 1 are blue- and red-shifted by 18 km s1from the systemic velocity.
The double-horn profile at the central position is indicative of an unresolved, expanding torus or ring. Very similar double-horn CO profiles have been observed in NGC 6072, NGC 6563, and IC 4406, where the dark lanes ( likely molecular regions) in op-tical images show obvious torus morphologies (Cox et al. 1991).
For examples of model double-horn spectra of an unresolved ring, see Ford et al. (2003). Based on these empirical relations be-tween the CO profiles and the morphologies, we assume an un-resolved, expanding molecular torus for the CO-emitting region in NGC 2440. The assumed torus is probably highly fragmented
Fig.1.— Representative CO(3Y2) spectra in NGC 2440. Position offsets ( , ) (arcseconds) are indicated to the left of individual spectra. The spec-tra have been offset by ( from bottom to top) 0, 0.4, 0.8, 1.2, and 1.6 K.
and the torus morphology may not be readily recognizable. No existing optical image of NGC 2440 shows a dark lane indicative of a torus. A similar suggestion of a torus morphology has been made from H2observations by Latter et al. (1995) and by Kastner et al. (1996). A similar suggestion has also been made for the ion-ized central gas of NGC 2440 ( Lo´pez et al. 1998), even though a torus morphology is hardly recognizable in any optical ( line or continuum) image.
If we assume an expanding molecular torus at the central region of NGC 2440, then the velocity shifts of 18 km s1of the two prominent spikelike features from the systemic velocity can be interpreted as the expansion velocity Veof the bulk molecular gas in the central region. The expansion velocity likely reflects the mass-loss velocity during the late AGB phase. The expansion ve-locity could be as large as 21 km s1, if the expansion velocity is estimated from the full width at half maximum of the CO spec-trum at the central position in Figure 1. The ionized part of the cen-tral, probably toroidal, structure is expanding at 22 km s1( Lo´pez et al. 1998). The molecular gas in the central region is thus expand-ing at almost the same velocity as the ionized gas in the central region.
Figure 2 shows the distribution of integrated intensity over a 9 km s1interval centered at different velocities (channel maps).
The distributions of the 26 and 61 km s1components appear in the 24 and 60 km s1maps, respectively, in Figure 2. These pro-minent velocity components are concentrated to the central re-gion with a size of about 2400, although no detailed structures are resolved in the present observation.
Beyond the expansion velocity, fast moving CO emission is clearly recognizable in the spectrum at the central position. The
The fast blue component (0 < VLSR<20 km s1) is detected only at the central nine positions with confidence. Likewise, the fast red component (65 < VLSR<75 km s1) is detected only at several positions near and at the center. The distribution of the fast blue emission can be seen in the 6 and 15 km s1maps in Fig-ure 2. The distribution of the fast red component can be seen in the 69 km s1map in Figure 2. The fast components are con-centrated to the central region that is bright in optical and H2
observations. The presence of the fast CO components are in line with the often suggested idea that the neutral, dense central torus (still containing significant molecular material) is being blown away by very fast outflowing winds. With the angular resolution and S/ N levels in the present observation, the details of the fast molecular components are not well constrained.
Near the systemic velocity, the CO emission has an elon-gated distribution as can be seen in the 33 and 42 km s1maps in Figure 2. The extent of the CO emission near the systemic velocity is about 2000 (northwest-southeast) by 7000 (northeast-southwest) in the 42 km s1map. The emission distribution is nearly point symmetric about the central position, and shows a bi-polar morphology. The northeast and southwest emission peaks (hereafter the northeast and southwest peaks, respectively) are 3000away from the central position. There is no detectable velocity difference between the northeast and southwest emission peaks [see the spectra at positions (+1800, +2400) and (1800,2400) in Fig. 1].
The locations of the two CO emission peaks coincide well with those of the optical emission knots NK and SK described by Lo´pez et al. (1998). (See Fig. 4 by Lo´pez et al. [1998] for NK and SK.) The distribution of the CO emission near the systemic
veloc-Fig.2.— Channel maps of CO( 3Y2 ). The central velocity in VLSRis indicated at the top left corner in each frame. Each map covers a velocity interval of 9 km s1. Both contours and gray scale show the integrated TMB. The lowest contour is 0.4 K km s1, and the contour interval is 0.4 K km s1, corresponding to a 2 level for the velocity in-terval. Position offsets are from ( ; )¼ (07h41m55:41s;1812030:700½ J2000:0) ((; ) ¼ (07h39m41:50s;1805024:000[J1950 ).
WANG, HASEGAWA, & KWOK
266 Vol. 673
in Figures 3 and 4 are both from 35 to 55 km s1. ( The velocity range for the 42 km s1map in Figure 2 is from 38 to 47 km s1.) The overall distributions of CO emission in Figures 2 and 3 are very similar to each other, and are not affected by the choice of veloc-ity interval. Low-level emission of CO(3Y2) is detected above a 3 level in the regions between the central peak and the north-east and southwest peaks [for example, at positions (+1200, +1200), (+600, +1800), (600,1800), and (1800,1200)]. The CO emission in Figures 3 and 4 (and in the 42 km s1map in Fig. 2) roughly traces the well-collimated optical bipolar lobes, one ex-tending from the center to the northeast and the other extend-ing from the center to the southwest, with a position angle ( P.A.) of 35( hereafter the P:A:¼ 35bilobal structure following the notation by Lo´pez et al. 1998). The P:A:¼ 35bilobal structure is also seen in the [ N ii] image by Cuesta & Phillips (2000).
3.2. CO Column Densities
The beam averaged column density of CO, N(CO), at each position is estimated with N (CO) (cm2)¼ 4:64;1012Texexp (þ33:19/Tex)R
TMBdV ( km s1) ( Hasegawa & Mitchell 1995).
In deriving the above equation, we ignored the cosmic background radiation and line photon trapping, and assumed the same exci-tation temperature Texfor all rotational transitions. A dipole mo-ment of 0.112 D (Lovas & Tiemann 1974) is assumed. We assume Tex¼ 25 K as a representative value rather than estimating from CO line ratios (see below). If the actual Texis 10 or 200 K, the as-sumption of 25 K will lead to an underestimate of the column density by a factor of 2.9 or 2.4, respectively. The beam averaged CO column densities are given in Table 1 for five representative positions.
Currently it is difficult to determine Texreliably for a particular PN from CO line ratios. Dayal & Bieging (1996) obtained a large ratio of 3.9 from the integrated intensities of CO(2Y1) and CO(1Y0) in NGC 2440 and adopted a lower limit of 150 K for Tex. Huggins et al. (1996) obtained a CO(2Y1)/(1Y0) line ratio of 3.07 in NGC 2440, which, without a correction for different beam coupling
1.5Y2.5). The theoretical CO(2Y1)/(1Y0) ratio at 25 K is 2.57.
Bachiller et al. (1997 ) find Tex in the range 25Y60 K for ob-served molecules in several PNe and assume a representative Texof 25 K.
The total number of CO molecules N(CO)S (S is the pro-jected area of the CO-emitting region) is obtained by summing Ni(CO) dSiover detected positions i, where Ni(CO) is the beam averaged CO column density at observed grid position i and dSi
is the projected area corresponding to 600; 600at position i. For the central 2400; 2400region N (CO) S ¼ 6:0; 1050D2( kpc).
For the northeast peak region within 1400from (+1800, +2400), N (CO) S¼ 8:1; 1049D2( kpc). For the southwest peak region within 900from (1800,2400), N (CO)S¼ 5:4; 1049D2( kpc).
In the above estimates, D (kpc) is the distance in kpc to NGC 2440. Distance estimates for NGC 2440 range from 1.0 kpc to 2.19 kpc ( Hyung & Aller 1998; Gathier et al. 1986).
If the relative number of CO molecules with respect to hydrogen nuclei is f (¼ N (CO)/½N (H i) þ 2N ( H2) ), then the hydrogen mass MHassociated with CO is given by MH¼ N (CO) SmH/ f (mHis the hydrogen mass). Following Huggins et al. (1996), we assume f ¼ 3; 104. For the central CO emitting region of NGC 2440, MH¼ 1:67; 103D2(kpc) M.
Huggins et al. (1996) estimate MHþHe¼ 2; 102Mfor NGC 2440 with He / H¼ 0:1, D (kpc) ¼ 2:19, and Tex¼ 77 K. Under the same assumptions for D( kpc) and Texas by Huggins et al.
(1996) our CO observation translates to MHþHe¼ 1:4; 102M for the central 2400; 2400 region. If the northeast and southwest regions are included, all of CO emission in the 6000; 7000region
Fig. 3.— CO(3Y2) emission in contours is overlaid with the Digitized Sky Survey image ( gray scale) of NGC 2440. The VLSRinterval for CO emission is 35Y55 km s1. The lowest contour and the contour interval are both 0.63 K km s1, which corresponds to about 2 . [See the electronic edition of the Journal for a color version of this figure.]
Fig.4.— CO(3Y2) emission in contours is overlaid with the [ N ii] image ( gray scale) of NGC 2440 obtained with the CFHT ( Kwok 2004). North is to the top, and east is to the left. The VLSRinterval for CO emission is 35 to 55 km s1. The lowest contour and the contour interval are both 0.63 K km s1. which corresponds to about 2 . [See the electronic edition of the Journal for a color version of this figure.]
CO(3Y2) IN NGC 2440 267
No. 1, 2008
with many sampling positions, significantly reduced the uncer-tainty in source size. The remaining uncertain parameters in con-verting CO observations are now Tex, D ( kpc), and f.
4. DISCUSSION
The complex morphology of the ionized component of NGC 2440 has been well documented (Cuesta & Phillips 2000). So far, at least two bilobal structures (with P:A:¼ 85and P:A:¼ 35) and a possible third one ( P:A:¼ 60) have been identified from nebular line observations ( Lo´pez et al. 1998). The P:A:¼ 35 bilobal structure is the brightest of the three, and is often re-ferred to as the main lobes. It is identifiable when NGC 2440 is imaged and displayed in a particular nebular line with a special filter and a care ( Lo´pez et al. 1998; Cuesta & Phillips 2000).
The H , He ii 4686, and [S iii] images by Cuesta & Phillips (2000), for example, selectively show the P:A:¼ 35bilobe.
Even though the P:A:¼ 35bilobal structure is identified based on intensity distribution and velocity, it does not show any dis-tinct nature from the rest of the optical nebula in electron tem-perature or electron density ( Fig. 3 in Cuesta & Phillips 2000).
The image in the H2line by Latter & Hora (1997) shows even more complex and extended distributions of molecular material.
Figure 5 ( gray scale) shows an H2image of NGC 2440 overlaid with the CO (3Y2) emission (contours). This H2 image is very similar to the one by Latter & Hora (1997 ). The H2images ( Fig. 5 and one by Latter & Hora (1997 ) show two bright spots to the northeast and southwest coincident with the CO emission peaks and with the optical knots NK and SK. Additional H2emission features extend to north-northwest, northwest, south-southeast, and southeast. In the present search, CO was not detected with confidence in these extended, fragmental H2emission regions.
The CO(3Y2) emission in the present work closely traces the P:A:¼ 35bilobal structure, and turns out to be an effective probe in disentangling the complex morphology of NGC 2440. The de-tection of CO in the P:A:¼ 35 bipolar lobes (other than the central region) clearly indicates that this bilobal structure is rich in heavy molecules. The molecular rich property makes the P:A:¼ 35 bipolar lobes distinct from the rest of the extended optical structures of NGC 2440. The lack of a velocity gradient along the P:A:¼ 35 bilobal structure in CO (3Y2) is consistent with the optical line observations by Lo´pez et al. (1998). The lack of a velocity gradient is explained as the bipolar axis being nearly perpendicular to the line of sight ( Lo´pez et al. 1998). The NK and SK optical emission knots are interpreted as working interfaces (shocked regions) between a fast outflowing medium and a slow-moving dense gas by Lo´pez et al. (1998). The NK and SK optical knots have associated H2-emitting knots ( Latter & Hora 1997;
Hora et al. 1999), as discussed above. The CO line emission at the NK and SK peaks most likely originates from the postulated slow-moving, optically invisible dense molecular gas adjacent to the optical knots.
The origins of CO in the optical NK and SK regions are not clear.
Although a formation of CO through shock- or photochemistry on a timescale of order 10Y100 yr cannot be ruled out, more
Although a formation of CO through shock- or photochemistry on a timescale of order 10Y100 yr cannot be ruled out, more