The outflows and 3D structure of NGC 6337, a planetary nebula with a close binary nucleus

NGC 6337 is a member of the rare group of planetary nebulae where a close binary nucleus has been identified. The nebula's morphology and emission line profiles are both unusual, particularly the latter. We present a thorough mapping of spatially resolved, long-slit echelle spectra obtained over the nebula that allows a detailed characterization of its complex kinematics. This information, together with narrow band imagery is used to produce a three dimensional model of the nebula using the code SHAPE. The 3-D model yields a slowly expanding toroid with large density fluctuations in its periphery that are observed as cometary knots. A system of bipolar expanding caps of low ionization are located outside the toroid. In addition, an extended high velocity and tenuous bipolar collimated outflow is found emerging from the core and sharply bending in opposite directions, a behavior that cannot be accounted for by pure magnetic launching and collimation unless the source of the outflow is precessing or rotating, as could be expected from a close binary nucleus.


Introduction
To explain the complex morphologies and collimated outflows often observed in planetary nebulae the presence of toroidal magnetic fields and binary nuclei are commonly invoked (e.g., García-Segura & López 2000;Soker & Rappaport 2000). However, from the observational standpoint it has proven rather difficult to detect the firm supporting evidence for these theories. Recently, García-Díaz et al. (2008) have discussed the case of the planetary nebula NGC 1360 where evidence of a strong stellar magnetic field has been detected and magneto-hydrodynamical modeling has successfully reproduced its key kinematic and morphological features. In this work we now explore in detail the kinematic behavior and morphological structure of NGC 6337, a nebula where convincing evidence of the presence of a close binary nucleus exists. NGC 6337 (PN G349.3-01.1) appears as a thick ring with radial filaments and knots, a faint elliptical shell surrounds the ring or toroid and a conspicuous pair of condensations bright in [N II] are located at P.A. −43 • on opposite sides of the ring. The core of this nebula has been identified from time-resolved CCD (V-band) photometry (Hillwig 2004) as a close binary nucleus with a period of 0.173 days. Hillwig (2004) assumes a primary mass of 0.6 M ⊙ to derive a mass for the companion star M 2 ≤ 0.3 M ⊙ and a binary separation a ≤ 1.26 R ⊙ , these characteristics indicate the possibility that the binary core underwent a common envelope phase.
A kinematic analysis of NGC 6337 has previously been made by Corradi et al. (2000) via two long-slits across the nebula with position angles, P.A. = −39 • and P.A. = −75 • .
From these data they interpreted the ring-like structure as the waist of a pole-on bipolar PN. They also identified the expanding caps, which they labeled features A and B, located to the north-west (NW) and south-east (SE) and detected limited regions of high velocity that they tentatively identified as corresponding to a point-symmetric bipolar outflow.
Here we extend that study by using 12 long-slit positions that fully sample the extent of -4the nebula and allow us to analyze in detail the strange structure of the line profiles and combine them with the morphological information to produce a morpho-kinematic 3-D model with SHAPE (Steffen & López 2006) (http://www.astrosen.unam.mx/shape/) of the main nebula and its complex outflows that resolve the aspect-oriented complexity of this object.
The paper is organized as follows. In §2 we present the observations, in §3 we discuss the results, §4 describes the 3-D SHAPE model of NGC 6337 and finally in §5 we summarize the conclusions of this study.

Observations and Results
High-resolution spectroscopic observations and monochromatic images of NGC 6337 were obtained at the Observatorio Astronómico Nacional at San Pedro Mártir, (SPM), México, on two observing runs 2006, July 20 and 2007 June 20 -23. These observations were taken using the Manchester Echelle Spectrometer (MES-SPM) (Meaburn et al. 2003) on the 2.1 m telescope in a f /7.5 configuration. This instrument was equipped with a SITE-3 CCD detector with 1024 × 1024 square pixels, each 24 µm on a side. We used a 90Å bandwidth filter to isolate the 87th order containing the Hα and [N II] nebular emission lines. Two times binning was employed in both the spatial and spectral directions.
Consequently, 512 increments, each 0. ′′ 624 long gave a projected slit length of 5. ′ 32 on the sky. We used a slit of 150 µm wide (≡ 11 km s −1 and 1. ′′ 9) oriented to a P.A. of −43 • for the majority of the observations and a P.A. = 0 • for one pointing across the center of the nebula. All spectra and images were acquired using exposure times of 1800 s. The spectra were calibrated in wavelength against the spectrum of a Th/Ar arc lamp to an accuracy of ±1 km s −1 when converted to radial velocity. Deep images of the field were also obtained with MES in its imaging configuration in three different filters: [N II]  We reduced the data using standard IRAF 1 tasks to correct bias, remove cosmic rays and calibrate the two-dimensional spectra based on the comparison lamp spectra. All spectra presented in this paper are corrected to heliocentric velocity (V hel ).  Figure 3, and from g -l in Figure 4.

Discussion
The images in Figure 1 show the bright main ring, its filamentary and knotty structure and the bright, low ionization filaments located to the northwest and southeast of the ring and that are referred to here as caps. In addition, the Hα+[N II] image shows a faint elliptical outer shell which is also observed in the [O III] λ 5007Å image where it displays a brightness distribution that resembles a slight "S" shape, sometimes associated with bipolar envelopes. A tenuous, high-speed, bipolar outflow that is clearly present in the long-slit spectra (see below) cannot be identified from these images and is revealed by the SHAPE The toroid shows a very slow projected radial expansion, in the order of only 1 -2 km s −1 on average. This very slow radial expansion together with the nearly perfect circular shape of the toroid indicates that it has a very small tilt with respect to the plane of the sky (≤ 10 o ) and that we are looking at it nearly face on. A crude estimate of the deprojected expansion velocity for the ring yields about 11.5 km s −1 , assuming a 10 o tilt with respect to the plane of the sky. The angular outer radius of the ring is 24 ′′ , adopting a distance D = 1.3 pc (Stanghellini et al. 1993) to NGC 6337, its linear radius becomes 0.15 -7pc ≡ 4.66 × 10 17 cm. These values yield a kinematic age for the ring of 1.2 × 10 4 years.
The toroid and its inner region contain high excitation gas, revealed by the presence of the He II λ 6560 line emission. Positions e, h, and i show a bipolar type structure in the line emission of this ion. It is however curious that the regions closest to the star, see slits a, f, and g, do not show projected He II emission, as if a cylindrical cavity perpendicular to the plane of the toroid is indeed fairly void of material close to the core. A cavity produced by an isotropic free flowing wind region would be expected to contain the high excitation gas at its edge and observable in projection over the star, which is not the case. The presence of this cavity is perplexing and to discern its origin will require additional information and modeling (out of the scope of the present work).
The caps are readily recognized in the spectra as the knotty extensions located immediately outside the ring emission regions, prominent in slit positions e to i. They are reminiscent of FLIERS (Balick et al. 1993)

SHAPE modeling
In order to disentangle the 3-D geometry and kinematic structure of NGC 6337, we used the program SHAPE (Steffen & López 2006). SHAPE is a morpho-kinematic modeling tool that allows the user to reconstruct the 3D structure and observed spectral line profiles using expanding geometrical forms. SHAPE uses as reference monochromatic images and observed position -velocity diagrams to reproduce the 3-D structure and kinematics of the object. Particles are distributed over a surface, or throughout a specified volume, and are assigned a specific velocity law and relative brightness. Several particle -9systems can be used and each can be assigned different velocity laws to form a complex object. The resultant 2-D image and spectral information are then rendered from the 3-D model and compared with the real data.
To model NGC 6337, we used the [N II] image and position-velocity spectra since all the main kinematic components are present here. Our model was built with a torus and conic surfaces for its knotty structure. Segments of spheres were used to model the caps.
The bipolar collimated outflows were built from elongated cylinders, flattened and bent.
The observations indicate that the projected width of the redshifted section of the bipolar outflow is slightly wider than its blue counterpart and it has been modeled accordingly.
This difference in widths may result from projection effects along the line of sight such as a slightly twisted bipolar outflow, as may be indicated by the kinematic analysis of the northwest cap (see section 3) however, since a projected twisting effect of the outflows cannot be disentangled from the emission line profiles of the collimated outflows, this potential twisting has not been considered in the model. Figure  -10 - The first frame is shown with the north rotated 43 • counter-clockwise, equivalent to having slits b -l aligned vertically. The next two panels are rotated on the y-axis by 45 • and 90 • , clockwise, respectively. As an additional test of the overall goodness of our model we have produced synthetic line profiles from slits located at P.A. = −39 • and P.A. = −75 • , corresponding to those observed by Corradi et al. (2000), obtaining also a good match with their observations. The present SHAPE model is able to reproduce the basic 2-D morphology and the set of emission line profiles that provide a representation of the third dimension of the nebula through the radial velocity component. Slightly different geometric forms could have been used to build up the final model, but in the end there is only a very limited set of solutions that are able to replicate the complex P -V diagrams of this nebula. A key advantage of SHAPE for objects like NGC 6337 is its ability to model independent structures, each with its own velocity law, and then merge them into a single product. We have used velocity laws of the type v = k · r/r o where k is a constant, r is the distance from the source and r 0 is the distance at which the velocity k is reached. The values for k and r 0 have been chosen to match the observed velocities for the various components of NGC 6337 and to provide reasonable distance scales along the line of sight. However, the model cannot place restrictions on, for example, the length of the bipolar collimated outflows or its detailed structure nor on the thickness of the toroid or the precise distance of the caps from the toroid. Nevertheless, the present model allows a good understanding of the complex structure and outflows of this object that otherwise are rather difficult to visualize.

Conclusion
A thorough analysis of the kinematic structure and morphology of the planetary nebula NGC 6337 has been carried out. The nebula is composed of a conspicuous thick ring or Considering the current parameters derived by Hillwig (2004) for the binary core, it is likely that this underwent a common envelope episode that may have influenced the formation of the equatorial density enhancement into a thick ring. The region close to the core appears as a cylindrical cavity void of material; it is unclear how this cavity may have formed. The rich, knotty structure apparent in the thick ring has most likely resulted as a consequence of instabilities produced by the erosive interaction with the radiation field and winds from the core throughout the evolution of the system. We confirm the suggestion by Corradi et al. (2000) of the presence of a point-symmetric, bipolar, collimated outflow; SHAPE modeling of our data has revealed the structure of such a jet for the first time. Although the binary interaction may have spun up the white dwarf central star, favoring a magnetic launching and collimation of the outflows, it is unlikely that a magneto-hydrodynamic mechanism can be solely responsible of the extreme bending