1 Introduction

The southern planetary nebula (PN) SuWt 2 (PN G311.0+02.4) is a particularly exotic object. It appears as an elliptical ring-like nebula with much fainter bipolar lobes extending perpendicularly to the ring, and with what appears to be an obvious, bright central star. The inside of the ring is apparently empty, but brighter than the nebula's immediate surroundings. An overall view of this ring-shaped structure and its surrounding environment can be seen in the H$ \alpha $ image available from the SuperCOSMOS H$ \alpha $ Sky Survey (SHS; Parker et al., 2005). West (1976) classified SuWt 2 as of intermediate excitation class (EC; $ p=6$-$ 7$; Aller & Liller, 1968) based on the strength of the HeII $ \lambda $4686 and [OII] $ \lambda $3728 doublet lines. The line ratio of [N II] $ \lambda $6584 and H$ \alpha $ illustrated by Smith et al. (2007) showed a nitrogen-rich nebula that most likely originated from post-main-sequence mass-loss of an intermediate-mass progenitor star.

Over a decade ago, Bond (2000) discovered that the apparent central star of SuWt 2 (NSV 19992) is a detached double-lined eclipsing binary consisting of two early A-type stars of nearly identical type. Furthermore, Bond et al. (2002) suggested that this is potentially a triple system consisting of the two A-type stars and a hot, unseen PN central star. However, to date, optical and UV studies have failed to find any signature of the nebula's true ionizing source (e.g. Exter et al., 2010; Bond et al., 2002; Bond et al., 2003; Exter et al., 2003). Hence the putative hot (pre-)white dwarf would have to be in a wider orbit around the close eclipsing pair. Exter et al. (2010) recently derived a period of 4.91 d from time series photometry and spectroscopy of the eclipsing pair, and concluded that the centre-of-mass velocity of the central binary varies with time, based on different systemic velocities measured over the period from 1995 to 2001. This suggests the presence of an unseen third orbiting body, which they concluded is a white dwarf of $ \sim 0.7 {\rm M}_{\bigodot}$, and is the source of ionizing radiation for the PN shell.

There is also a very bright B1Ib star, SAO 241302 (HD 121228), located 73 arcsec northeast of the nebula. Smith et al. (2007) speculated that this star is the ionizing source for SuWt 2. However, the relative strength of HeII$ \lambda $4686 in our spectra (see later) shows that the ionizing star must be very hot, $ T$ $ >$ 100,000K, so the B1 star is definitively ruled out as the ionizing source.

Narrow-band H$ \alpha $+[NII] and [OIII]5007 images of SuWt 2 obtained by Schwarz et al. (1992) show that the angular dimensions of the bright elliptical ring are about $ 86.5$arcsec$ \times $$ 43.4$arcsec at the 10% of maximum surface brightness isophote (Tylenda et al., 2003), and are used throughout this paper. Smith et al. (2007) used the MOSAIC2 camera on the Cerro Tololo Inter-American Observatory (CTIO) 4-m telescope to obtain a more detailed H$ \alpha $+[NII] image, which hints that the ring is possibly the inner edge of a swept-up disc. The [NII] image also shows the bright ring structure and much fainter bipolar lobes extending perpendicular to the ring plane. We can see similar structure in the images taken by Bond and Exter in 1995 with the CTIO 1.5 m telescope using an H$ \alpha $+[NII] filter. Fig.1 shows both narrow-band [NII] 6584Å and H$ \alpha $ images taken in 1995 with the ESO 3.6 m New Technology Telescope at the La Silla Paranal Observatory using the ESO Multi-Mode Instrument (EMMI). The long-slit emission-line spectra also obtained with the EMMI (programme ID 074.D-0373) in 2005 revealed much more detail of the nebular morphology. The first spatio-kinematical model using the EMMI long-slit data by Jones et al. (2010) suggested the existence of a bright torus with a systemic heliocentric radial velocity of $ -25\pm5$ kms$ {}^{-1}$ encircling the waist of an extended bipolar nebular shell.

In this paper, we aim to uncover the properties of the hidden hot ionizing source in SuWt 2. We aim to do this by applying a self-consistent three-dimensional photoionization model using the MOCASSIN 3D code by Ercolano et al. (2003a); Ercolano et al. (2008); Ercolano et al. (2005). In Section 2, we describe our optical integral field observations as well as the data reduction process and the corrections for interstellar extinction. In Section 3, we describe the kinematics. In Section 4, we present our derived electron temperature and density, together with our empirical ionic abundances in Section 5. In Section 6, we present derived ionizing source properties and distance from our self-consistent photoionization models, followed by a conclusion in Section 7.

Ashkbiz Danehkar