3. The CO maps
Fig. 1 shows the velocity-channel maps of 12CO(1-0) emission towards the nucleus of NGC 3593, displayed from v=796.9kms-1 to v=484.1kms-1. Although CO emission is highly concentrated in the nucleus, the central channel maps (from 725.2kms-1 to 529.7kms-1) adopt the shape of a spider web diagram. This is the characteristic signature of a projected rotating disk which has been spatially resolved. The eastern (western) side of the CO disk is red (blue)-shifted if we refer all radial velocities to a systemic velocity of =630kms-1. This value is taken from HI and is confirmed in this work (see below). A comparison with the kinematics of the stars studied by B96 indicates that molecular clouds in NGC 3593 are counterrotating with respect to the primary stellar disk (disk I, adopting B96 's notation). Disk I contains the bulk of the stellar mass (MI=1.21010) and therefore it will define what we call hereafter direct rotation. On the other hand, the molecular clouds corotate with the ionized gas, and with the secondary less massive stellar disk denoted as disk II (MII=2.7109). According to Fig. 1, there is no evidence of molecular gas in direct rotation.
With a velocity-integrated CO intensity of 5103K kms-1, and a CO-to-H2 conversion factor X=N(H2)/ICO=2.31020cm- 2K-1km-1s (Strong et al. 1988), the total H2 mass derived from the 12CO(1-0) interferometer map is M(H2)=9.2108. For this, we assumed the distance to be D=12.4 Mpc, and have integrated the CO flux within a rectangular area of dimensions 7018" centered on the position (0,0). Including the mass of helium, the total molecular gas mass in the Bure field is Mgas=M(H2+He)=1.2109. Taking into account that the shortest spacing measured by the interferometer is 20m, we expect to filter scales 20-25" at 115GHz. We have derived the fraction of the single-dish 12CO(1-0) 30m flux measured by Wiklind & Henkel 1992 included in the Bure maps. We detect nearly 70 of the 30m-flux within our primary beam.
The over-all morphology of CO emission is best displayed in Fig. 2a, which represents the velocity-integrated intensity contours. For the sake of simplicity we will distinguish two regions in our maps, as detailed below. The division into these two regions is far from arbitrary, as it is based on morphological and kinematical criteria.
3.1. The circumnuclear disk
Half of the emission in the nucleus of NGC 3593 is due to an elongated circumnuclear source of Mgas=M(H2+He)=4.7108. The source has a projected diameter of 20", defined by the separation of two CO peaks, at (,)= (10",0") in Fig. 2a. The average gaussian size (FWHM) of the circumnuclear source along its minor axis is 6". So defined, the circumnuclear source is centrally peaked: the CO maximum is identified in Fig. 2a as (,)=(0",0"). To maximize the symmetry of the 2D-velocity field of Fig. 1 requires the assumption that the (0,0) offset of Fig. 2a is the dynamical center of the galaxy (, 49´05.6"), and take =630kms-1 (see also Fig. 5a,b). Moreover, the CO-based dynamical center reassuringly coincides within 2" with the peak intensities of the NGC 3593 images obtained, in the near-infrared ()-bands (Moriondo et al. 1998a), in the 1.49GHz radio continuum (Condon et al. 1990), and in the optical continuum (Corsini et al. 1998; note however that coordinates of their Fig. 1 are wrong).
The central source can be described as a ringed circumnuclear disk of radius r10"(600 pc) (hereafter denoted as CND). If we assume an inclination angle i=70o, the expected size of the CND along its minor axis would be 6-7", namely, in rough agreement with the value reported for . Therefore the CND is barely resolved along its minor axis by the present CO observations. The 12CO(1-0) peak brightness intensity map of Fig. 2b, more sensitive to contrasted structures, highlights also the ringed disk structure of the CND. The edges of the limb brightened ring at =10" stand out in Fig. 2b, whereas the central maximum of Fig. 2a breaks up into a series of clumps which are comparatively weaker. The near-infrared (NIR) color map of Moriondo et al. (1998a), a fair tracer of dust extinction, underlines the existence of a dusty nuclear disk of projected diameter 20", in good agreement with what we derive from CO (see Fig. 3b).
The sizes of the molecular CND and the stellar disk II, both in counterrotation with respect to the outer stellar disk, are comparable: 2 rII, where rII10" is the exponential scale length of the counterrotating stellar disk. B96 obtained the value of rII (and its counterpart for disk I : rI40") by a photometric decomposition of the NGC 3593 disk. According to those authors, and from a purely kinematic point of view, the fit on the observed stellar velocities would be much improved if disk II was truncated beyond r15-20"; this underlines the link between disk II and the gaseous CND.
3.2. The one-arm spiral
Besides the CND, molecular gas emission is also detected along a ridge which starts on the eastern side of the map (=15"-20"). The ridge extends south of the CND, going across the minor axis at -5" and crossing the major axis at =-35" (see Fig. 2a,b). Hereafter this Southern Ridge will be referred as S-R. The ridge fades progressively in the northwest quadrant and it is linked with the beginning of a strong dust lane that stretches out on the northern side of the optical image (see Fig. 3c). Although CO emission becomes fainter on the dust lane, there is a good correspondence of the latter with a maximum of the color map (see Fig. 3b). This is suggestive of a maximum in the neutral gas column density (maybe atomic rather than molecular hydrogen). The northern dusty arc seen in the map, hereafter referred as N-R, continued in the S-R, forms a gaseous arc which opens unevenly in the plane of the sky. According to the B96 data, there is no old stellar component corotating with molecular gas along this gaseous arc. However a simple inspection of the H and P maps indicates that new stars are being formed along this feature.
It is important to define the geometry of the CO arc, paying attention to the N-R and S-R regions. Fig. 2c shows the peak brightness intensity map deprojected onto the galaxy plane (assuming i=70o). We use (Loge(R),)-coordinates, where R is the galaxy deprojected radius in ", and is the azimuthal angle in degrees, measured counterclockwise from the western side of the major axis at PA=-90o. Any spiral logarithmic feature would appear as a straight line with a non-zero slope in Fig. 2c. Using this representation, the bulk of the S-R develops between 2.9Loge(R)3.4 and 225360o. The S-R, which shows a 2-periodicity along , appears as a straight line. We can derive an average pitch angle (p) of the spiral from tan(p)Loge(R)/ along the arm, p being the angle between the circular and the spiral arc at a given radius, measured counterclockwise. This gives p168o for the S-R. On the other hand, the bulk of the N-R develops between 3.4Loge(R)3.0 and 0140o. A similar calculation gives p9o for the N-R.
Therefore we have indications that the pitch angle of the one-arm spiral changes in NGC 3593. Although the dominant feature characterising the morphology of the CO disk outside the CND is the S-R, the presence of secondary compressions, revealed by the N-R, strongly indicates a complex picture, in which a mixture of distinct modes in the nucleus of NGC 3593 might be at play.
From a theoretical viewpoint, it is crucial to establish whether these one-arm spiral arcs are leading or trailing with respect to the gas flow. A simple-minded approach consists of assessing which side (N or S) of the NGC 3593 disk is closer to us, using de Vaucouleurs' criterion (de Vaucouleurs 1958). The dust lane is a projected foreground layer situated on the northern side of the disk (see Fig. 3c). The absorbing dust lies embedded in the plane of the disk (at an average deprojected radius of 1kpc), and is attenuating the emission of the inner nucleus (bulge+disk). This identifies unambiguously the northern side of the galaxy (0) as the near side.
According to the observed over-all sense rotation of the gas, we can conclude that the one-arm spiral is leading with respect to the gas across the S-R, whereas it is trailing with respect to the stars of disk I . The secondary gas compression, denoted by N-R, would be a trailing arm with respect to the gas. Moreover, the gas flow shows the presence of non-circular motions related to the one-arm spiral. Their detailed analysis will serve to cross-check the precedent conclusion on the nature of the modes in NGC 3593.
© European Southern Observatory (ESO) 2000
Online publication: December 5, 2000