JSTOR

1. INTRODUCTION

Martin et al. (2006) found a variable emission line centered at 6307 Å in multiple spectra of η Car, recorded by the Hubble Space Telescope Space Telescope Imaging Spectrograph (HST/STIS) and by VLT/UVES. They were unable to identify the origin of the emission line, but demonstrated that the line was present across the high state (defined by presence of forbidden lines of doubly-ionized elements, see Damineli et al. 2008 and references therein) and disappeared during the low state.

Recently Gull et al. (2009), using the same spectra, focused on the spatially-extended forbidden line emission both from high-ionization (herein defined as >14 eV) and low-ionization (8–13 eV). We found that the forbidden emission originated from (1) the Weigelt condensations (Weigelt & Ebersberger 1986), as narrow lines centered on -43 km s^-1; (2) the boundaries of the primary wind (η Car A) as ropelike [Fe ii], photoionized by mid-UV and collisionally excited at densities around N_c = 10⁷ cm^-3; and (3) the wind interaction region, as high-ionization emission from [N ii], [Fe iii], [Ar iii], [Ne iii], and [S iii], photoionized by far-UV and collisionally excited for densities, n_c ranging from 10⁵ to 10⁸ cm^-3. The high-ionization emission lines are present both for the Weigelt condensations and the wind interaction region during the 5 yr high state, but every 5.5 yr, disappear during the low state. X-ray models (Pittard & Corcoran 2002; Parkin et al. 2009) place the binary periastron event near the onset of the low state with a 3 to 6  month recovery.

We applied a three-dimensional (3D) SPH (smoothed particle hydrodynamics) model (Okazaki et al. 2008) extended out to 1700 AU (0.67″) to match the spatial structure seen in these forbidden lines, and realized that the bulk of the high-ionization emission structure originated from the wind interaction region in the outer portion of the massive wind structure. We found that portions of the wind interaction structure, moving ballistically outward, are directly illuminated by the far-UV radiation of the hot secondary, η Car B, leading to highly ionized, collisionally excited gas and hence the high-ionization extended emission.

Further examination of the HST/STIS longslit spectra showed that the previously unidentified emission at 6307 Å is blue-shifted [S iii] λ6313 emission from the interacting wind structure. The evidence is subtle, but convincing. We summarize the observations in § 2. A description of how the spectroimages were produced is in § 3. The forbidden emission structures are described in § 4. Discussion in § 5 provides insight on the ionization and excitation leading to [S iii] emission and the potential for monitoring changes with orbital phase, including mapping temperature with density dependence. We conclude with a summary in § 6. Throughout this article, all wavelengths are in vacuum, the velocities are heliocentric, directions are compass points (N–north, etc.), and the phase of the binary orbit is referenced to the X-ray minimum beginning at 1997.9604 (Corcoran 2005).

2. THE HST/STIS OBSERVATIONS

The spectra discussed here are a portion of the Eta Carinae Treasury observations accessible through the STScI archives2 as reduced by a special reduction tool developed by K. Ishibashi and K. Davidson. For brevity we focus on two sets of observations, recorded in 2002 July (ϕ = 0.820) and 2003 July (ϕ = 1.001).

Observations were recorded with the HST/STIS moderate dispersion gratings and CCD detector through the 52^″ × 0.1^″ aperture. We wanted to monitor the change in both η Car and the Weigelt condensations (Weigelt & Ebersberger 1986), which drop in excitation during the low state. However, the range in aperture position angle (P.A.) is limited by the required orientation of HST solar panels and changes throughout the year. Arbitrary orientation of the long, narrow aperture can prevent inclusion of any one of the three Weigelt condensations within the aperture when centered on η Car (Fig. 1). For observations centered around the X-ray minimum, predicted to be around 2003 July 1, we scheduled a visit one year earlier, 2002 July 2 (orbital phase, ϕ = 0.820), at a preselected P.A. = 69°, that would be accessible just before (ϕ = 0.995) and after the X-ray minimum (ϕ = 1.001). During all three visits, separate observations centered on η Car and Weigelt D were obtained with the aperture placed as shown in Figure 1. Additional information on the observations is presented in Martin et al. (2006) and Gull et al. (2009).

3. SPATIALLY-RESOLVED EMISSION

The HST/STIS spatial-resolution (0.1″ at Hα) separates the spectrum of η Car’s core from extended structures, especially the narrow line emission that originates from the Weigelt condensations (Davidson et al. 1995), located between 0.1″ to 0.3″ in the NW quadrant relative to η Car (see Fig. 1). As described by Gull et al. (2009), we found considerable differences between many broad forbidden emission line profiles of η Car as recorded by the VLT/UVES and the HST/STIS. Extractions with a 0.127″ high slice of the STIS spectra (five half rows in the reduced spectroimages) yielded broad wind line profiles for H i, He i, and Fe ii lines that compared favorably with those recorded by VLT/UVES. In contrast, broad profiles of forbidden lines recorded by VLT/UVES were nearly absent in the HST/STIS extractions centered on η Car. Examination of the HST/STIS long aperture spectra revealed faint structure in these emission lines extending out to 0.7″. However, the observed emission was highly variable both with aperture P.A. and orbital phase, ϕ. Clearly the forbidden line emission is spatially extended on scales resolved by HST/STIS but not by VLT/UVES.

We examined individual lines in more detail and attempted to enhance visibility of the extended line emission by several reduction procedures. While various spatial filters were tried, the best results were obtained by subtraction of measured continuum on a spatial row-by-row basis. We used spectral plots of the Weigelt condensations (Zethson 2001) to identify 10 to 20 Å intervals with no obvious presence of narrow or broad line emission. At each position along the aperture, we measured and subtracted the average continuum in that spectral interval. Examples of the resulting spectroimages (intensity images with x = velocity and y = angular size) are presented in Figure 2.

These spectroimages provide only a qualitative view of the line profile. We caution the reader that quantitative measures require much greater precision for the following reasons:

1.	The STIS utilized a three-axis mechanism to select a grating and to set the correct tilt angle for the spectral interval of choice. While return to that grating position is within a few CCD pixels, variations in the tilt are not fully reproducible. A tilt of 1/20 pixel along the 1024 element row led to significant photometric variation in an attempt to extract spectra at the 0.1″ spatial resolution.
2.	The standard calibration for the STIS photometry is properly referenced to extractions of a stellar spectrum with a 2″ wide aperture, allowing for full capture of flux from a point source. Apertures with widths comparable to the diffraction limit of HST sample the point-spread function of the telescope, which changes dynamically even within an orbit.
3.	Charge transfer inefficiency (CTI) leads to a trail in the direction of CCD columns and with on-orbit cosmic radiation exposure, increases. For complex sources like extended structures, a proper extraction is not available.

These problems complicate attempts to show extended structure in the vicinity of a bright star, which is exactly the situation with η Car. This leads to the obvious linear striations at the star position in the spectrally dispersed (velocity) coordinate. These variations do affect measures of the extended emission closest to the stellar position. However, for offsets to Weigelt D, the stellar flux is blocked by the aperture, and quantitative measures are then possible. We note that the stellar spectrum scattered from the direction of Weigelt D is very different from the direct spectrum of η Car. The lack of P Cygni absorption in Hα across the high state indicates fully-ionized hydrogen in the region spatially located between η Car and Weigelt D (Gull et al. 2009). On the side of caution, we limit this discussion to a description of the spatial and velocity structure of the lines. Even with qualitative descriptions, we gain much insight on the spatially resolved wind interactions and the source of the 6307 Å emission.

4. DESCRIPTION OF THE EMISSION STRUCTURES

We refer the reader to Figure 2 for the following descriptions of the forbidden line emission. The first two columns of spectroimages are extracted from spectra centered on η Car in the high state (ϕ = 0.820) and early in the low state (ϕ = 1.001). Likewise, the last two columns are centered on Weigelt D (a more complete summary on variation of the forbidden emission structures with ionization potential and orbital phase is presented by Gull et al. 2009).

Four basic structures contribute to these spectroimages:

1.	The central core of η Car, not resolved by HST at 0.1″, which contributes the bulk of the continuum and P Cygni wind lines, notably of H i and Fe ii.
2.	Weigelt D and other, lesser condensations that contribute many narrow emission lines centered at -40 km s-1.
3.	Ropelike structures of high-ionization forbidden emission lines with velocity components extending from +200 to -500 km s-1.
4.	Noticeably more diffuse, ropelike structures of low-ionization forbidden emission lines, specifically [Fe ii].

Spectroimages of [Fe ii] λ4815 (Fig. 2, row 1) show narrow ropelike features extending to 0.7″ at -500 km s^-1 and other, more diffuse structures closer to the star extending ± 500 km s^-1. The narrow emission at -40 km s^-1 originates from extended structure WSW of η Car, not noted by Weigelt & Ebersberger (1986), but present throughout the observational period from 1999 to 2004 whenever the STIS aperture sampled this position. The narrow [Fe ii] emission centered on Weigelt D extends 0.5″ E and W of Weigelt D, but at about 0.25″ E of Weigelt D, a diffuse emission extends to -400 km s^-1 and away from η Car. During the low state, the outer [Fe ii] emission drops, becomes more diffuse, and is located closer to η Car.

The [Fe iii] λ4703 (Fig. 2, row 2) is interior to the ropelike [Fe ii] λ4815. A series of highly filamentary loops extend from η Car to the east at velocities from -40 to -500 km s^-1. No redshifted velocity components are seen at this P.A. However, Gull et al. (2009) find that at P.A. = -28°, observed 6 times from 1998.0 to 2004.3 (ϕ = 0.000 to 1.122), redshifted components extend to +200 km s^-1 at early phases, but fade late in the high state. All [Fe iii] λ4703 disappears during the low state.

The [N ii] λ5756 (Fig. 2, row 3) has very similar structure to that of [Fe iii] λ4703, with a higher signal-to-noise ratio (S/N). A narrow emission line, [Fe ii] λ5748, appears at the -550 km s^-1 position and persists in the low state, along with weak [N ii] λ5756. The structure of [N ii] λ5756 extends from -40 to -500 km s^-1 in the spectroimage centered on Weigelt D during the high state, but also disappears in the low state.

Three [S iii] lines are shown in Figure 2, rows 4–6, as each is important in accounting for the 6307 Å emission. Unfortunately, the [S iii] λ6313 (row 4) was recorded only at ϕ = 0.820, but the other two lines were observed at both phases. Most noticeable in the spectroimage of η Car at ϕ = 0.820 is a knot of emission, centered on the stellar position at -400 km s^-1. The effective wavelength is 6307 Å. That spectral interval was recorded at other phases, but at other position angles, during the 2003.5 minimum with no [S iii] λ6313 present either at the positions of η Car or Weigelt D. The extended structure is less apparent in the spectroimage of η Car at ϕ = 0.820, but is well defined in the spectroimage centered on Weigelt D. Narrow lines of [O i] λ6302 and Fe ii λλ6307, 6309, and 6319 contaminate the spectroimage and persist during the minimum.

The [S iii] λ 9071 (Fig. 2, row 5) weak emission extends off of η Car, but several narrow lines (N i λ9063, Fe ii λ9073) also contribute to the spectroimage. The high-velocity arc of [S iii] λ 9071 extends blueward from Weigelt D.

The [S iii] λ9533 emission (Fig. 2, row 6) is quite similar to that of [Fe iii] λ4703 (Fig. 2, row 2) and confirms that [S iii] emission extends to -500 km s^-1. The bright emission to the red of [S iii] λ9533 is H i Pa 8 λ9548, which originates primarily from the central core.

5. DISCUSSION

We associate the low-ionization structure with the massive, slow-moving wind of η Car A. The high-ionization emission is from the interacting wind region piled up by the fast-moving, less massive wind of η Car B (Pittard & Corcoran 2002). The bulk of the interacting wind, by its velocity, appears to be mostly ionized wind of η Car A. The higher velocity side of the shock is likely less dense and more highly ionized by η Car B.

Using the 3D SPH models of Okazaki et al. (2008) and simple geometric models, we determined that the high-ionization emission originates from a distorted paraboloidal structure lying in the skirt of the Homunculus. Based upon the blueshifted velocities and near symmetry for P.A.s ranging from +22 to +38°, the paraboloid points in our general direction, with axis of rotation projecting onto the sky at P.A. ≈ -25°.

Martin et al. (2006) performed a very complete analysis on the unidentified 6307 Å line in both the HST/STIS and VLT/UVES spectra, finding very similar behavior with orbital phase. Their search of possible line identifications focused primarily on singly-ionized species such as Fe⁺, V⁺, and S⁺, although they do list [S iii] as a narrow nebular line identified by Zethson (2001) in the spectrum of the Weigelt condensations. Their candidate of greatest interest appeared to be Fe iii] λ6306.43 with unknown atomic data for the transition.

Most important in their analysis was the tracking of the strength of the emission throughout the 5.5 yr orbit. They found that the line disappeared during the low state, but might be anticorrelated with Fe ii λ5529. Both suggest a high-ionization source. Nielsen et al. (2007) analyzed the behavior of the He i absorption, finding an anticorrelation with Fe ii absorption.

Salient are three facts:

1.	On the star, both HST/STIS and VLT/UVES see the same emission bump with similar strengths.
2.	The emission correlates with high-ionization variations, not the behavior of the low-ionization emission of Fe ii.
3.	The extended emission of [S iii] correlates very well with the extended emission identified with [S iii] λλ9071 and 9533.

6. CONCLUSIONS

We have presented conclusive evidence that the emission line at 6307 Å, noted in the spectra of η Car by Martin et al. (2006) is blueshifted emission of [S iii] λ6313 originating from the distorted paraboloidal interaction region located between the massive binary members. While the massive primary, η Car A, provides the dominant wind ejecta, 10^-3M_⊙ yr^-1 at 500 km s^-1, the hotter secondary provides a less massive, faster wind, 10^-5M_⊙ yr^-1 at 3000 km s^-1, and far-UV photons that ionize iron, neon, argon, and sulfur to doubly-ionized states. Thermal collisions, mid-UV photons, and possible charge exchange excite the doubly-ionized species to upper states with forbidden transitions leading to forbidden line emission in regions with densities close to n_c. Specifically, [S iii] λλ9533, 9071, and 6313 have extended spatial structures. The intensity ratio (Flux (λ9533) + Flux (λ9071))/Flux(λ6313) leads to density estimates ranging from 10⁷–10⁸ cm^-3 on the scale of 0.1″, the limit of HST/STIS spatial capabilities. Mapping in these and other doubly-ionized lines will provide powerful measures for models of wind interactions using various 3D hydrodynamical codes.