HST STIS UV Spectroscopic Observations of the Protoplanetary Nebula Hen3-1475^{∗ †}

The UV morphology of Hen 3-1475 is studied in conjunction with HST WFC3 images in the F656N and F658N filters that show the Hα and [N ii] λ6583 emission, mostly from the jets and knots, which are presumably shock excited (Borkowski & Harrington 2001); the broadband images (e.g., F555W and F814W) show the continuum radiation, mostly scattered starlight. We created UV monochromatic images by first selecting in the STIS 2D spectrum regions centered on UV emission lines, and then carefully registered them with WFC3 images. The overall morphology of Hen 3-1475's brightest regions (NW1/SE1 and the central reflection nebula), as outlined in optical emission, is visible in several UV lines (Figure 2). In the Mg ii λ2800 spectral-line image, the NW1 knot is resolved into four components, hereafter named A, B, C, and D (from the inside out; Figures 3(a), (b)), with angular sizes of ∼012–014 along the jet axis; in the other UV lines, generally fainter than Mg ii, only the three outer components, B, C, and D, are seen (Figures 2(b)–(f)). The SE1 knot seen in the UV could be the counterpart of the outermost D component.

Zoom In Zoom Out Reset image size

The optical counterparts (in Hα) of the four UV components A, B, C, and D are located 249 (173-W, 179-N), 267 (193-W, 185-N), 291 (207-W, 204-N), and 330 (247-W, 219-N) from the central star, respectively. B, C, and D are elongated (in the dispersion direction) in Mg ii; this elongation is obvious in B and C, which are compact in Hα. D comprises two subcomponents in Hα that are ∼015 apart and stretch by 032, similar to the size in the UV. The NW conical shock "converges" on A (Figures 4(a), (b)), the most compact (radius ∼006) among the four UV components of NW1. Elongation of the NW1 components in the UV may be caused by large, internal negative radial velocity gradients (Borkowski & Harrington 2001).

Zoom In Zoom Out Reset image size

Along the (spectral) dispersion direction, the four components of NW1 seem to be displaced with respect to their optical counterparts; we discuss this in Section 4.2. Along the slit direction, positions of A, B, and C in the UV are unchanged compared to their counterparts in Hα, which is consistent with the very little expansion of NW1 that we found through comparison of the 1999/2009 F658N images (see Section 2). Component D and SE1 in the UV are slightly outside their Hα counterparts (Figure 3(b)) due to jet expansion/propagation since 2009. The position of D in Mg ii is shifted outward (i.e., toward NW) by 0069 from its position in Hα, which corresponds to a proper motion of 11.8 mas yr⁻¹, or ∼280 km s⁻¹. Similarly, SE1 in Mg ii is displaced from its position in Hα by 0091, corresponding to ∼15.5 mas yr⁻¹, or ∼360 km s⁻¹.

In the UV spectrum, <2'' from the central star, there are four "stripes" (or striations) that are more prominent near the red end of the STIS NUV-MAMA spectrum (Figure 2(b)). These striations are only detected in the NW lobe, and seem to spatially align with the optical features at corresponding locations: the hollow U-shaped "bowls" near the nucleus of Hen 3-1475 also seem striated in the optical (Figures 4(b), (c)). The central region of Hen 3-1475 is dominated by the scattered stellar light as shown in the HST STIS G750M spectrum (Borkowski & Harrington 2001, Figure 2 therein). Along the jet axis, the four UV striations in the NW lobe are ∼013, 05, 095 and 143 from the central star (Figure 3, bottom).

Whereas the positions of A, B, and C in the UV along the slit direction are consistent with those of their counterparts in Hα, there is a noticeable shift along the dispersion direction of the UV spectrum (Figure 3(b)). Given that the NW jet of Hen 3-1475 has high approaching speeds (Borkowski & Harrington 2001; Riera et al. 2003), it is plausible to attribute these displacements to Doppler shift.⁹ Indeed, A, B, and C are shifted blueward in the UV spectrum by 9.4, 7.2 and 6.7 pixels which, if entirely due to Doppler shift, correspond to radial velocities of −1550 ± 160, −1200 ± 330 and −1100 ± 400 km s⁻¹, respectively. The systemic radial velocity of Hen 3-1475 (+40 km s⁻¹, Borkowski & Harrington 2001) is within the velocity errors. Although the outermost component D, as well as SE1 in the UV, seems to align with its optical counterpart along the spectral direction, we cannot rule out the possibility that D also has a high approaching speed, given its extension and the low spectral resolution of G230L. This sets an upper limit of ≲310 km s⁻¹ for the approaching speed of D. Adopting this radial speed and the sky-projected velocity of D (∼280 km s⁻¹, see Section 4.1), we deduce a jet inclination (i, with respect to the line of sight) ≳42°, which is consistent with the estimate of (Borkowski & Harrington 2001, i = 40°). Adopting i = 40°, we corrected the velocities of A, B, C, and D to be ∼2000, 1600, 1400, and ≲400 km s⁻¹, respectively.

Both the [N ii] and Hα emission lines in the HST STIS G750M optical spectroscopy (in 1999; Borkowski & Harrington 2001) show an increase in radial velocity from the base region of the NW conical shock, reaching a maximum blueshift of −910 km s⁻¹ at the cone tip (i.e., component A), and followed by a small drop (by ∼110 km s⁻¹) near the cone tip, and continuous decrease in velocity through C to D. This small drop in velocity near the cone tip is reflected by the velocity difference between A and B in the UV. Our radial velocities derived for the UV components of NW1, although higher than those found in the STIS G750M spectrum, generally follow the same trend (i.e., decreasing from the inside out), indicating a sequence of negative speed changes within NW1. The jet might have been accelerated since 1999 possibly because the central star has become hotter.

If the NW1 components are moving clumps along the jet, the sky-projected velocities of A, B, C, and D should be ∼1300, 1000, 900, and ≲260 km s⁻¹ (adopting the above radial velocities and i = 40°), corresponding to proper motions of ∼55, 42, 38 and ≲12 mas yr⁻¹, respectively. However, A, B, and C are actually unchanged in positions, whereas D and SE1 show small radial displacements from their positions in the 2009 Hα image (Figure 3(b); see Section 4.1). Such a spatio-kinematic pattern of NW1 seems to resemble quasi-stationary shocks: A, B, C, and D may actually be shock interfaces where the outflowing gas moves through with high speeds, and have no (or very small) measurable proper motions; each of these clumpy shocks may brighten or fade in time owing to variations in density or speed of the gas flow. If A, B, C, and D are indeed the locations of shocks, the gas flow should be slowed down at each of these shocks as some kinetic energy is converted into heat that locally excites line emission (B. Balick 2018, private communication). The negative radial velocity gradient in NW1 may thus be explained as a series of speed losses of the (possibly unstable) flow as it propagates outward along the jet. This interpretation, although plausible, is still speculative and needs careful assessment with comprehensive hydrodynamic simulations.

The outer pair of knots (NW2/SE2 and NW3/SE3) in Hen 3-1475 was previously proposed to be fast-moving "bullets"; but their almost tangential motion along the S-shaped arms (Figure 1, bottom; see also Section 2) suggests that this is probably not the case. One sensible interpretation is that these outer knots are actually bright spots along the curved arms, which may be faint lobe walls projected on the sky; these bright spots are produced due to the Kelvin–Helmholtz instabilities as excited by the fast streamlines flowing over the surface of lobe walls.

We analyzed the archival HST STIS G750L spectrum and found that the optical line ratios (e.g., [O ii]/Hβ, [O iii]/Hβ, [S ii]/Hα) from the NW1 knot are roughly consistent with the predicted values of a bow-shock model at 200 km s⁻¹ (Hartigan et al. 1987); this analysis will be presented separately (X. Fang et al. 2018, in preparation). The high-velocity jet impinges on the slower AGB wind, creating shocks as delineated by the [N ii]/Hα ratio (Figure 4(e)): the three pairs of knots, as well as the two limb-brightened inner cones, are enhanced in [N ii] emission (although we are aware that at least for SE1, the F658N filter may be contaminated by redshifted Hα). Within the NW lobe, there are filaments enhanced in the Hα/Hβ ratio (Figure 4(f)); they may not be due to higher local extinction but actually related to variation in shock speed that causes locally enhanced temperature and/or density, which results in departure from the Case B recombination of hydrogen. The NW conical shock is slightly curved before it converges (Figures 4(b), (c)), an effect possibly due to interaction with the lobe walls, which are bright in scattered lights as seen in F555W (Figure 4(d)). The SE conical shock is also curved in optical line emission.

For each of the four striations within the NW lobe, its 1D NUV spectrum is composed of a smooth continuum overlaid by several emission features, one of them being Mg ii λ2800 (see Figure 4(g) for 1D spectra of the inner three, namely Striations 1, 2, and 3). The smooth continuum could be the scattered starlight, while the broad UV emission features may come from fast stellar outflows. The innermost UV striation is not located exactly on the central star, but offset by 013 (probably due to obscuration by the central dusty torus), corresponding to ∼9.7 × 10¹⁵ cm; the half-width (FWHM/2) of Mg ii emission is ∼22 Å, corresponding to a radial velocity gradient up to 2360 km s⁻¹. These two quantities are consistent with both location (∼10¹⁶ cm) and velocity (∼2300 km s⁻¹) of the ultra-fast "pristine" post-AGB outflow found by Sánchez Contreras & Sahai (2001). We cannot rule out the possibility that such UV spread/broadening might also be (at least partially) due to spatial extent; but the broadening of emission features on the innermost layer (Striation 1) can be due to kinematics because its Mg ii emission seems to extend beyond the U-shaped boundary of the NW lobe (Figure 4(a)).

The exact location of X-ray emission in NW1 is still unclear; although it seems to peak around the cone tip (Figure 2(a)), this might be an effect of heavy smoothing of the low-photon counting Chandra data. If X-ray emission comes from the component A, the simultaneous presence of X-ray, UV, and optical emission suggests that A is a cooling region of shocked gas with very strong temperature gradients. However, A is spatially unresolved in our UV monochromatic imaging, given that its angular size is comparable to the angular resolution of the STIS/NUV-MAMA detector (∼006; Table 1). The physical conditions within the NW1 knots will be assessed by a re-analysis of the X-ray data and substantiated with the UV and optical emission line ratios to derive shock properties (J. A. Toalá et al. 2018, in preparation).

Using Equation (1) of Gruendl et al. (2004) and the observed Mg ii λ2800 line flux (5.2 × 10⁻¹⁴ erg cm⁻² s⁻¹), we estimate a density of ∼10⁵ cm⁻³ for component A. Here an extinction of A_V ∼ 3.7 magnitude (Riera et al. 1995) and the solar abundance (Mg/H = 3.98 × 10⁻⁵, Asplund et al. 2009) were adopted; we also assume a maximum ionization fraction Mg⁺/Mg = 0.69 at T_e ≈ 15,800 K in ionization equilibrium (Shull & Van Steenberg 1982), and a filling factor of 1. Using the same assumptions above, we estimate densities of ∼50,000–60,000 cm⁻³ in B, C, and D. These estimates of the density from the UV-emitting region in NW1 are much higher than the electron density derived from the [S ii] λ6716/λ6731 ratio (∼3000 cm⁻³, Riera et al. 2006), suggesting different physical conditions in the UV, optical, and X-ray emission regions (especially in component A).