COLLISIONALLY EXCITED FILAMENTS IN HUBBLE SPACE TELESCOPE Hα AND Hβ IMAGES OF HH 1/2

HH 1 and 2 were the first Herbig–Haro (HH) objects discovered (Herbig 1951; Haro 1952), and they have played an important role in the field of HH objects (see the historical review of Raga et al. 2011). For example, Hubble Space Telescope (HST) images (Schwartz et al. 1993; Hester et al. 1998), proper motions (ground-based: Herbig & Jones 1981; HST: Bally et al. 2002; IR: Noriega-Crespo et al. 1997; radio: Rodríguez et al. 2000), and detections in radio continuum (Pravdo et al. 1985), UV (Ortolani & D'Odorico 1980), and X-rays (Pravdo et al. 2001) were first obtained for HH 1 and 2.

The HH 1/2 system has a central source detected in radio continuum (see, e.g., Rodríguez et al. 2000) and a bipolar jet system with a NW jet (directed toward HH 1), which is visible optically, and a SE jet (directed toward HH 2), which is visible only in the IR (see Noriega-Crespo & Raga 2012). HH 1 has a "single bow shock" morphology, and HH 2 is a collection of condensations, some of which also have bow-shaped morphologies (see, e.g., Bally et al. 2002). The emission line structure of these objects has been studied spectroscopically, with one-dimensional (Solf et al. 1988) and two-dimensional (Solf et al. 1991; Böhm & Solf 1992) coverage of the objects. It should be pointed out that the HH 1/2 outflow lies very close to the plane of the sky, so that projection effects do not have to be considered when interpreting the observations of these objects.

The spatial structure of the optical line emission has been studied at higher angular resolution with HST images. Schwartz et al. (1993) obtained Hα, [S ii] 6716+6730, and [O i] 6300 images. Later images of HH 1 and 2 were all taken with filters isolating the Hα and the red [S ii] lines (Bally et al. 2002; Hartigan et al. 2011).

In this Letter, we describe a pair of new HST images of HH 1 and 2 obtained with filters isolating the Hα and Hβ lines. These images were obtained in consecutive exposures, so that they are not affected by proper motions (which become evident in HST observations of the HH 1/2 complex separated by more than a few weeks) or by differences in the pointing, and they therefore yield an accurate depiction of the spatial distribution of the Hα/Hβ ratio in these objects. These images show effects that have not been detected previously in ground-based studies of the emission line structure of HH 1 and 2 (see, e.g., Solf et al. 1991 and Böhm & Solf 1992) or in HST images of other HH objects (since HST Hβ images of HH objects have not been previously obtained).

This Letter is organized as follows. The new HST images are described in Section 2. The spatial distribution of the Hα/Hβ ratio, the line ratios as a function of Hβ intensity, and the distribution functions of the line ratios are presented in Section 3. Finally, an interpretation of the results is presented in Section 4.

The region around HH 1 and 2 was observed with the Hα (F656N) and Hβ (F487N) filters on 2014 August 16 with the Wide Field Camera 3 (WFC3) on the HST. The Hα image was obtained with a 2686 s exposure and the Hβ image with a slightly longer 2798 s exposure. The images were reduced with the standard pipeline, and a simple recognition/replacement algorithm was used to remove the cosmic rays. The final images have 4130 × 4446 pixels, with a pixel size of 003962.

The images contain only two stars: the Cohen-Schwartz star (Cohen & Schwartz 1979) and "star No. 4" of Strom et al. (1985). These two stars have been used to determine astrometric positions in CCD images of the HH 1/2 region since the work of Raga et al. (1990), yielding better positions for HH 1 (which is closer to the two stars) than for HH 2. We have carried out paraboloidal fits to the point-spread functions of these two stars, and find no evidence for offsets and/or rotation, which shows the excellent tracking of the HST during the single pointing in which the two images were obtained. Also, we have analyzed the Hα −Hβ difference images of the two stars and find no offsets between the two frames.

The full Hα frame is shown in Figure 1, as well as zoomed in views of the regions around HH 1 and HH 2 in both Hα and Hβ. As seen in the top frame, the Hα map shows the extended collection of HH 2 knots (to the SE) and the more compact distribution of the HH 1 knots (toward the NW). The central frames of Figure 1 show the Hα emission of HH 2 (left) and HH 1 (right) on a larger scale. In the fainter Hβ emission (bottom panel of Figure 1), only the brighter regions of HH 1 and 2 are detected. We have defined two boxes (labeled A and B in the bottom frame of Figure 1) enclosing the regions of the two objects that are detected in Hβ. In the following section, the regions within these two boxes are used in order to study the spatial dependence of the Hα/Hβ ratio.

Zoom In Zoom Out Reset image size

As discussed in detail by O'Dell et al. (2013), the F656N filter is contaminated by emission from the [N ii] 6548 line, and both the F656N and F487N filters have contributions from the nebular continuum. Using the fact that at all measured positions within HH 1 and 2, the [N ii] 6548/Hα ratio does not exceed a value of ≈0.35 (see, e.g., Brugel et al. 1981a and Solf et al. 1988) and the transmission curve of the F656N filter (see O'Dell et al. 2013 and the WFC3 Instrument Handbook), one then finds a peak contribution of ≈2% to the measured flux. For estimating the effects of the continuum in the F656N and F487N images, one can use the continuum and line fluxes obtained by Brugel et al. (1981a, 1981b) and the bandpasses of the filters to obtain estimates of ≈0.4 and 5% (for the F656N and F487N filters, respectively). Therefore, when interpreting the Hα/Hβ ratios obtained from our HST images, it is necessary to keep in mind that there is an uncertainty of ∼5% due to a possible spatial dependence in the Hβ line to continuum ratio within the F487N filter. As this uncertainty is ∼1 order of magnitude smaller than the Hα/Hβ ratio variations described below, we do not discuss it further.

Figure 2 shows the Hα map (right) and Hα/Hβ ratio map (left) for HH 2. To avoid having extended regions dominated by noise, in order to calculate the line ratio map, it is necessary to place a lower bound on the line fluxes. We have chosen to calculate the ratios only in regions in which the observed Hβ flux is larger than I₀ = 5.4 × 10⁻¹⁸ erg s⁻¹ pixel⁻¹.

Zoom In Zoom Out Reset image size

For calculating the intrinsic Hα/Hβ ratios, we have applied the following reddening correction. We first calculate the observed ratios for all of the pixels with Hβ intensities larger than I₀ (see above) for the A and B boxes (shown in the bottom frames of Figure 1). For HH 2, we obtain a mean line ratio 〈Hα/Hβ〉_obs = 3.82 and for HH 1 we obtain an almost identical 〈Hα/Hβ〉_obs = 3.79 value. Considering an observed line ratio of 3.8 for both objects, comparing with the case B recombination cascade intrinsic Hα/Hβ ratio of 2.8 and using the average Galactic extinction curve, we obtain an E(B − V) = 0.27 color excess. This value is somewhat lower than the E(B − V) ≈ 0.35 value deduced for HH 2 by Brugel et al. (1981a), using the method of Miller (1968), based on the fixed ratios between the auroral and transauroral lines of [S ii] (i.e., not assuming a recombination cascade Hα/Hβ ratio). In order to calculate the dereddened Hα/Hβ ratios, we therefore multiply the observed ratios by a factor of 2.8/3.8, basically assuming that the extinction toward HH 1 and 2 is position-independent.

The dereddened Hα/Hβ ratios of HH 2 (see Figure 2) have values in the 2 → 6 range, with the regions of higher values corresponding to filamentary structures in the leading edge of the emitting knots (i.e., in the edges directed away from the outflow source). In order to illustrate the positions of these "high Hα/Hβ" regions, we have superimposed an Hα/Hβ = 4 contour on the Hα emission map (purple contour in the right frame of Figure 2).

Figure 3 shows the Hα map (bottom) and dereddened Hα/Hβ ratio map (top) for HH 1. We have calculated the ratios only for pixels with an observed Hβ flux larger than I₀ = 5.4 × 10⁻¹⁸ erg s⁻¹ pixel⁻¹ (i.e., the same cutoff used for HH 2; see above). The region with Hα/Hβ > 4 is a thin filament on the E side of the leading edge of HH 1 (see the purple contour on the Hα emission map in the bottom frame of Figure 3). It is clear that HH 1 shows a strong side-to-side asymmetry with respect to the outflow axis, as the SW region of the leading edge does not show high Hα/Hβ ratios (see the top frame of Figure 3). The Hα emission also shows a strong side-to-side asymmetry.

Zoom In Zoom Out Reset image size

Figure 4 shows the dereddened Hα/Hβ line ratio as a function of the (observed) Hβ flux for all of the pixels with I_Hβ > I₀ (see above) for HH 1 (top frame) and HH 2 (bottom frame). It is clear that for low values of the Hβ intensity in both HH 1 and 2 we have a relatively broad distribution of line ratios (the width of this distribution representing the relatively large errors of the line ratio at low intensities) centered on the Hα/Hβ = 2.8 recombination cascade value. For pixels with brighter Hβ intensities, we see a distribution of Hα/Hβ ratios extending from ≈3 to larger values of ∼5 (for HH 1) or ∼6 (for HH 2).

Zoom In Zoom Out Reset image size

This result is seen more clearly in Figure 5, where we show the normalized distributions of the line ratios of pixels with I₀ < I_Hβ < I₁ = 2.5 × 10⁻¹⁷ erg s⁻¹ pixel⁻¹ (distribution f₁, top frame), of pixels with I₁ < I_Hβ < I₂ = 4.7 × 10⁻¹⁶ erg s⁻¹ pixel⁻¹ (distribution f₂, center), and of all pixels with I₂ < I_Hβ (distribution f₃, bottom frame of Figure 5, with appropriate pixels found only in HH 2). For both HH 2 (left column) and HH 1 (right column of Figure 5), we see that the distribution f₁ of the lower intensity pixels is approximately symmetrical, centered at an Hα/Hβ ≈ 2.8 line ratio. The distributions for higher intensity pixels (f₂ and f₃; see above and the middle and bottom frames of Figure 5) start at values of Hα/Hβ ∼ 2–2.5, have a peak at a line ratio of ≈3.3 and have a wing extending to Hα/Hβ ∼ 5 for HH 1 and ∼6 for HH 2. In the following section, we show that these high Hα/Hβ ratios coincide with the values expected for collisional excitation of the n = 3 and 4 levels of H.

Zoom In Zoom Out Reset image size

From our new Hα and Hβ HST images, we can compute dereddened Hα/Hβ maps for the brighter regions of HH 1 and 2. For the reddening correction, we assume that the mean value of the Hα/Hβ ratio coincides with the recombination cascade value of 2.8, as found previously by Brugel et al. (1981a), who calculated the reddening correction with Miller's method, based on the ratios of auroral to transauroral [S ii] lines.

We find that in limited spatial regions the (dereddened) Hα/Hβ ratio has values of ∼4 → 6, which are inconsistent with the recombination cascade value. These high Hα/Hβ regions are filaments along the leading edges (i.e., the edges away from the outflow source) of the brighter emitting regions of HH 1 and 2 (see Figures 2 and 3).

Raga et al. (2015) show that the r_αβ = Hα/Hβ ratio for a "case B" cascade fed by collisional excitations from the ground state of hydrogen has the approximate form:

$$\begin{equation} r_{\alpha \beta }=3.40\,e^{E_{43}/(kT)}+\frac{0.95}{(1+{\rm 5\times 10^4\,K}/T)^4}, \end{equation} \tag{ 1 }$$

where k is Boltzmann's constant and E₄₃ is the energy difference between the n = 4 and n = 3 energy levels (so that E₄₃/k = 7680 K). The first term of this functional form has a temperature dependence derived from the ratio of the n = 1 → 3 and n = 1 → 4 collisional excitation coefficients (assuming temperature-independent collision strengths), and the second term is a correction necessary to match the results of a five-level, collisionally fed cascade matrix description of the hydrogen atom in the T = 10³ → 10⁶ K temperature range (see Raga et al. 2015). It is clear that the functional form of r (see Equation (1)) has high values for low temperatures and has an asymptotic value of 4.35 for T → ∞.

From Equation (1), one obtains r(T = 1.5 × 10⁴ K) = 5.6 and r(T = 10⁵ K) = 3.9. Therefore, the wing of the line ratio distributions extending to Hα/Hβ ∼ 4 → 6 (see Figure 5) can straightforwardly be explained as produced in regions with temperatures in the 1.5 → 10 × 10⁴ K range emitting collisionally excited Balmer lines.

This clear evidence that we are observing collisionally excited Balmer lines together with the fact that the high Hα/Hβ regions are restricted to the leading edges of the outward moving condensations of HH 1 and 2 is quite conclusive evidence that we are observing the region of collisional excitation of H lines right after the shock waves driven into the surrounding medium by the condensations. Most of the Hα emission, however, comes from a region further away from the shock, in which the Balmer lines are produced through the standard recombination cascade (as evidenced by the Hα/Hβ ∼ 3 ratios; see Figures 2 and 3).

The theoretical prediction of these two regions of Balmer line emission (a collisionally excited Balmer line region immediately after the shock, and the recombination region with Balmer lines dominated by the recombination cascade) in HH shock wave models is already mentioned by Raymond (1979), and the Hα emission from the two regions was studied in more detail by Raga & Binette (1991). These two regions are of course present in all shock models (for example, in the plane-parallel, time-dependent shock models of Teşileanu et al. 2009).

In non-radiative shocks observed in some supernovae remnants or in pulsar cometary nebulae, the observed emission comes exclusively from the region of collisional excitation right behind the shocks (see, e.g., the review of Heng 2010). In HH objects, the only previous observational evidence of the emission from the immediate post-shock region (as opposed to the emission from the recombination region) were the Hα filaments seen in HST images of some bow shocks, notably in the HST images of HH 47 (Heathcote et al. 1996), HH 111 (Reipurth et al. 1997), and HH 34 (Reipurth et al. 2002). However, as only Hα was observed, it was not possible to guarantee that these filaments did correspond to the region of collisionally excited Balmer lines.

Our new Hα and Hβ images for the first time show in a quite conclusive way that we have a detection of the immediate post-shock region of HH objects (in which H is being collisionally ionized and the levels of H are being collisionally excited). The detection of this region provides a clear way forward for developing models of HH bow shocks, in which the position of the shock wave relative to the recombination region is directly constrained by the observations.

We should note that throughout this Letter we have assumed that the extinction is uniform over the emission regions of HH 1 and 2. In principle, it could be possible that foreground structures in the vicinity of the objects might produce changes in the extinction on angular scales comparable to the size of the objects. However, estimates of the density of the pre-bow shock material of HH 1 and 2 (based on observations of the post-shock density and on plane-parallel shock models; see, e.g., Hartigan et al. 1987) give values ∼100–300 cm⁻³. Clearly, such a low-density environment will not produce appreciable extinction on spatial scales comparable to the size of the HH objects. Because of this, if one wants to attribute the observed changes in the Hα/Hβ ratio to an angular dependence of the extinction, it is necessary to assume that still undetected, sharp-edged, high-density regions are present in the immediate vicinity of HH 1 and 2.

Support for this work was provided by NASA through grant HST-GO-13484 from the Space Telescope Science Institute. A.R. and A.C.R. acknowledge support from the CONACyT grants 101356, 101975, and 167611 and the DGAPA-UNAM grants IN105312 and IG100214.