HiRes DECONVOLVED SPITZER IMAGES OF 89 PROTOSTELLAR JETS AND OUTFLOWS: NEW DATA ON THE EVOLUTION OF THE OUTFLOW MORPHOLOGY^*

The sample of Class 0 protostars, H₂ jets, and outflow sources we selected for HiRes deconvolution of Spitzer images are listed in Table 1. The majority of our target protostellar objects were selected from "The Youngest Protostars" webpage hosted by the University of Kent (http://astro.kent.ac.uk/protostars/old/), which are based on the young Class 0 objects compiled by Froebrich (2005). In addition to these objects, our sample includes some Herbig-Haro (HH) sources and a few well known jet outflow sources. Our sample also includes one high-mass protostar (IRAS 20126+4104; cf. Caratti o Garatti et al. 2008) to demonstrate the use of HiRes for such sources. Our choice for target selection was primarily based on the availability of Spitzer images in IRAC and MIPS bands in the archives and the feasibility for reprocessing based on the published Spitzer images wherever available.

To maximize the scientific return of the Spitzer images, we use the HiRes deconvolution processing technique that makes optimal use of the spatial information in the observations. The algorithm, "HiRes" and its implementation have been discussed by Backus et al. (2005) and its performance on a variety of astrophysical sources observed by Spitzer is presented by Velusamy et al. (2008). The HiRes deconvolution algorithm is based on the Richardson-Lucy algorithm (Richardson 1972; Lucy 1974) and the Maximum Correlation Method employed by Aumann et al. (1990) for IRAS data. As demonstrated by Velusamy et al. (2007, 2008), the HiRes deconvolution on Spitzer images retains a high fidelity and preserves all the main features. HiRes deconvolution improves the visualization of spatial morphology by enhancing resolution and removing the contaminating side lobes from bright sources. The benefits of HiRes include (1) enhanced resolution of ∼0.6–08 for IRAC bands, ∼18 and ∼6'' for MIPS 24 μm and 70 μm images, respectively, (2) the ability to detect sources below the diffraction-limited confusion level, (3) the ability to separate blended sources and thereby provide guidance to point-source extraction procedures, and (4) an improved ability to show the spatial morphology of resolved sources. We reprocessed the images at all the IRAC bands and the MIPS 24 and 70 μm bands using all available map data in these bands in the Spitzer archives containing the selected prostellar objects.

We used the pipeline-processed basic calibrated data (BCD) and post-BCD (pBCD) downloaded from the Spitzer Science Center archives. Our nominal data processing produced reprocessed images at each band by applying the HiRes deconvolution on BCDs following the steps outlined by Velusamy et al. (2008). The images containing 43 objects, as noted by "bcd" in Column 7 of Table 1, were processed in this mode using BCDs as input to HiRes. All of the HiRes images using the BCD images as input were obtained after 50 iterations. Although HiRes was originally designed to take BCD images as input, we have found that it works equally well when mosaic images are used (cf. IRAS 20293+3952 and IRAS 05358+3843 by Kumar et al. 2010; M51 by Dumas et al. 2011). For 46 protostellar objects processed after 2012, we used the pBCD images as input to HiRes deconvolution, taking advantage of the improved data products in the Spitzer archives. These objects are identified in Table 1 (Column 7 as "pbcd"). The pBCD data are mosaic images made using the BCD images after some resampling applying convolutions. Therefore, in comparison to the BCD images as input, the mosaic images represent already smoothed input images. Thus, the HiRes deconvolution of such images converges more slowly and typically requires about 150 iterations to match the same level as with BCDs as input.

The overall performance of HiRes deconvolution on the Spitzer images presented here is in good agreement with that of the previously published examples (cf. Velusamy et al. 2008). The HiRes deconvolution interpolates well across the "bad" (NANed) pixels in the input images such as due to muxbleed and saturation. All bands are relatively free of artifacts except for the IRAC channels 3 and 4 (in particular at 8 μm), which, in some cases, show streaks due to muxstripping and a spurious secondary component adjacent to very bright sources due to uncorrected muxbleed. Such spurious secondary components seldom occur with HiRes in the BCD input mode. However, in the pBCD input mode, our algorithm can miss muxbleed pixels because of the convolutions used in the pBCDs. Such spurious components, if present in the maps, are identified in Figures 3.1(b)–3.53(b) in the online journal. As a note of caution, when interpreting multiple components in the vicinity of bright objects in 8 μm images, one should examine the sources more closely to check whether they might be artifacts in the data or are real.

The angular resolutions in the deconvolved images are typically in the range of 0.6–08 in the 3.6 μm & 4.5 μm IRAC bands, 0.8–1'' in the 5.8 μm & 8 μm IRAC bands, and ∼2'' and ∼6'' in the MIPS 24 μm and 70 μm bands, respectively, which are a factor of 2 better than those in the mosaic images. The resolution enhancement achieved depends on the signal-to-noise-ratio, the image coverage (redundant sets of BCD or pBCD images used as input), and the background levels for the point sources. The angular resolution in each HiRes image can be obtained by examining Gaussian fits to the "point" sources in each image.

Figure 3.1 is an example of how our HiRes images are displayed in Figures 3.1–3.53. In each figure, there are two panels: (a) the 24 μm emission is overlaid as contours on the 4.5 μm image and (b) an RGB three-color representation of IRAC 3.6 μm (blue), 4.5 μm (green), and 8 μm (red). We chose to display a 4.5 μm image since it traces equally well both the scattered light and H₂ line emission in the jet outflow. The 24 μm emission contours identify the driving source (protostar/disk) associated with the jets and outflows. Furthermore, in cases where the atomic jet features, traced by the atomic and/or ionic line emission within the MIPS 24 μm band, are present, they are identified by the 24 μm contours overlaid on the 4.5 μm grayscale image. To highlight such features more clearly, where it is useful, we show a zoomed view of the selected regions (indicated by the boxes in the left panel) on the right-hand side of the figure. The three-color representation in the lower panel (b) helps to highlight the color differences among various features, especially the wide-angle outflow traced by blue representing the scattered light (predominantly at 3.6 and 4.5 μm) and the H₂ jets/bow shocks in green and red. The protostellar object and the jet outflow features are also identified in the figures in panels (a) and (b).

HH1/HH2 is a well-studied system (the first detected HH object); see, for example, Noriega-Crespo & Raga (2012) for a detailed anatomy of the jets and counter-jets as seen in Spitzer images. These data are a good example to illustrate the significant features that can be traced in the HiRes Spitzer images. Furthermore, as far as we know, this object is the only one for which there exists a deconvolved image using other techniques. Note that the HiRes image at 4.5 μm in Figure 3.1(a) compares well with a recently deconvolved image (Noriega-Crespo & Raga 2012) obtained by using a different algorithm, the deconvolution software A WISE Astronomical Image Co-Adder (AWAIC), developed by the Wide Field Infrared Survey Explorer(WISE) for the creation of their Atlas images (see, e.g., Masci & Fowler 2009). Of particular importance in our results, as seen in the enlarged view (in the right panels in Figure 3.1(a)), is the association of 24 μm features with the optical knots in HH2 (southeast), as identified by Raga et al. (1990) and Noriega-Crespo & Raga (2012), as well as with the jet and counter jet. These 24 μm features trace the atomic and ionic jet features, as discussed in the cases of HH46/HH47 and Cep E (Velusamy et al. 2007, 2011). Another new feature in our results in Figure 3.1(b) is the detection of the wide-angle outflow cavity, implying that the protostar HH1-2 VLA is driving simultaneous wide-angle outflow and collimated jets.

As reported in the case of Cep E (Velusamy et al. 2011), the MIPS images have small pointing differences (∼1'') with respect to the IRAC images. This offset is also evident in some of the maps shown Figures 3.1–3.53. The difference in the positions of the point sources between the IRAC and MIPS images is, in part, due to the fact that MIPS used a scan mirror to change the field of view during their observations and the mechanism itself suffered a hysterisis effect that increased the positional uncertainty by 06–10. This uncertainty is a small fraction of the MIPS beam at 24 μm in the mosaic images, but it becomes more obvious when comparing MIPS and IRAC HiRes images. Note that the MIPS images shown in the figures or the FITS image data are not corrected for any pointing offset with respect to the IRAC images. Therefore, in case detailed positional matching is required, the pointing offset can be obtained by comparing the "point" sources in the MIPS 24 μm image with their counterparts in the IRAC bands using the HiRes FITS images. The MIPS handbook gives a 1σ radial uncertainty of 14, compared with ∼02 for IRAC.

One problem with HiRes deconvolution is that it tends to resolve out very smooth, extended low surface brightness features. Optimal-performance HiRes requires that the background emission is fully subtracted out in each BCD prior to applying the deconvolution. Furthermore, the positivity criteria implicit in the deconvolution algorithm makes it insensitive to negative intensities. In rare cases (as in Figure 3.53), although the HiRes-deconvolved image is free from side lobe contamination from bright sources in the image, its resolution enhancement and its lack of preserving the background can make it harder to detect extremely extended low surface brightness emission.

Figures 3.1–3.53 highlight the prominent features in the HiRes images of the protostar regions. The HiRes FITS images contain more details over the full extent of the processed maps and will be available in the Web site given above. Spitzer's coverage of a broad range of infrared emission in the IRAC and MIPS bands with high sensitivity and photometric stability, along with the IRS data in some cases, provides sufficient information for a comprehensive modeling of the spectral energy distribution (SED) to derive the physical characteristics and the evolutionary stages of protostars (cf. Robitaille et al. 2007; Forbrich et al. 2010). However, by applying the HiRes reprocessing, we can extract even more information from the Spitzer data. As demonstrated in Figure 3.1, the sensitivity, the resolution enhancement, and removal of the diffraction lobe confusion have led to visualizing and characterizing more clearly the following.

• 1.  
  Very high dynamic ranges in the maps in which the protostar is well resolved on a subarcsecond scale from the surrounding jets and outflows.
• 2.  
  The protostar-disk itself is detectable in the IRAC bands in many cases (with relatively low obscuration) and almost always in the MIPS bands.
• 3.  
  Wide-angle outflow cavities in the IRAC 3.6 and 4.5 μm images are identified by the scattered photospheric starlight; although they are visible in the Mosaic standard pBCD products, they are brought out more clearly in the HiRes images.
• 4.  
  Jets and bow shocks in H₂ molecular emission within the IRAC bands.
• 5.  
  The hottest atomic and ionic gas in the jet and jet head traced by the [Fe ii]/[S i] line emission within the MIPS 24 μm band.

The Spitzer IRAC and MIPS bands offer a unique resource to study a wide range of components simultaneously: protostars, protostellar disks, outflows, protostellar envelopes, and cores. At the short-wavelength IRAC bands, the outflow cones are observable in scattered light from the protostar through the cavity created by the jets and outflows (cf. Tobin et al. 2007; Velusamy et al. 2007, 2011). A significant fraction of the emission in the IRAC bands is also considered to contain emission from the H₂ rotational lines (e.g., Noriega-Crespo et al. 2004a, 2004b; Smith & Rosen 2005; Neufeld & Yuan 2008) and therefore is an excellent tracer of H₂ emission in the protostellar jets. Recently, Velusamy et al. (2007, 2011) have shown that molecular jets and molecular gas in the bow shocks are readily identifiable in the IRAC bands, while the hottest atomic and ionic gases in the bow shocks are also identifiable in the MIPS 24 μm band, which covers a few atomic and ionic emission lines. The dust emission from the protostar in the MIPS bands is a good diagnostic of the circumstellar disks.

With the exception of HH 1-2 MMS 3, all the other protostars are detected in one or more bands in the processed images. In at least three cases, BHR 71, CG 30, and CB 230, we clearly detect the binary components and associated H₂ jets and outflows. The binary components (A,B) in L1448 IRS3 (cf. Tobin et al. 2007), IRAS 16293-2422 (cf. Takakuwa et al. 2007), and L723 (cf. Girart et al. 2009) are marginally to well resolved in the 24 μm HiRes images. We clearly detect 31 wide-angle outflows in scattered light in the IRAC bands (there may be others that are less obvious due to confusion with other features). Out of these, there are 12 outflows, including B5 IRS1, which is known to have parsec-scale HH flows (Yu et al. 1999), in which no detectable H₂ jets or bow shocks have been observed. In about 14 H₂ jet/bow-shock systems, atomic/ionic emission in the jet/jet heads is detected in the MIPS 24 μm images. Prominent atomic jets are evident, for example, in Figures 3.1 (HH 1-2), 3.16 (HH 211), 3.30 (HH 92), 3.31 (HH 212), 3.37 (HH 46/47) 3.42 (Serp-SMM1), and 3.52 (Cep E).

Of the wide-angle outflows, 31, representing a large fraction of the sample, show a high efficiency of scattered light in the Spitzer images as a tracer of outflow cavities. In Figure 4, we show an image gallery of all 31 outflows. Although the wide-angle outflows are observed by their CO emission, a comparison between the outflow cavities traced in the Spitzer images of HH46 and in the ALMA maps of CO emission show that the former is broader than the latter (Arce et al. 2013). In the case of CB 26 (Launhardt et al. 2008), a narrower outflow is also detected in CO than is seen in the scattered light cavity. Arce et al. argue that the narrower outflow seen in ¹²CO is a consequence of opacity effects. In Figures 5 and 6, we show two examples of a comparison between the Spitzer scattered light cavity and the CO outflows. In the case of L 483, shown in Figure 5, we overlay the redshifted and blueshifted ¹²CO (1-0) intensity contours on the HiRes Spitzer image at 4.5 μm (see Figure 3.41). The ¹²CO (1-0) maps are from our unpublished OVRO observations made in 1997. The redshifted lobe is narrower than the blueshifted lobe and both show narrower opening angles compared with those in scattered light. The narrower opening angle for the CO outflow may be evidence for ¹²CO opacity effects close to the source. In the second example shown in Figure 6, we compare the scattered light in the HiRes Spitzer image at 4.5 μm of Cep E (left panel) with the OVRO ¹³CO (1-0) maps (right panel) obtained by Moro-Martín et al. (2001). Clearly, the ¹³CO outflow seems to trace the full extent of the scattered light cavity (also shown in Figure 3.52), as may be expected for ¹³CO, which has a lower opacity compared with ¹²CO in the previous example. Thus, compared with CO, the Spitzer images offer a more powerful tool to study the morphology of wide-angle outflows detected in scattered light. In addition to the scattered light cavities, the H₂ molecular jets, bow shocks, and, in some cases, the atomic/ionic jet components, when present, are also detected in the Spitzer images, thus providing a more complete morphology of the protostellar outflows and jets. The predominance of the simultaneous presence of collimated jets and wide-angle outflows in our sample (19 out of 31) is consistent with unified jet outflow models such as those presented by Machida et al. (2008) in which the accretion in the compact protostellar disk drives the high-velocity jets and the accretion from the extended infall envelope drives the wide-angle outflows.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Our sample contains a large number of H₂ jet/bow shocks (partly due to the selection of many known HH objects). The data presented here provide a comprehensive database to study the number, distribution, and their separation from the protostar of the H₂ knots and bow-shocks, which are indicators of the episodic ejections and/or the inhomogeneities in the surrounding interstellar medium.

Protostellar jets and outflows carve cavities and inject energy, momentum, and turbulence into the surrounding medium (e.g., Shu et al. 2000; Bally 2007). They have a profound effect on the parent core, which is the reservoir for material accreted by the forming star, and thus on the final properties of the newborn star as well. Outflows originate close to the surface of the forming star (e.g., Pudritz et al. 2007) and such a clearing out of the gas and dust along with a widening of the outflow with time can lead to the termination of the infall phase (Velusamy & Langer 1998). A possible evolutionary trend in the outflow morphology (widening with time) as a function of age was seen in a sample observed in CO (Arce & Sargent 2006) and is predicted in the simulations of outflow evolution (Offner et al. 2011). A broadening of outflow with age was also found by Seale & Looney (2008), who used the scattered light in the outflow cavities in the unprocessed Spitzer images of 27 young stellar objects combined with their SEDs to quantify the shapes of the outflow.

In Table 2, we list all 31 of the wide-angle outflows detected in our sample (see Figure 4), along with the opening angles measured, as discussed below. The outflow shapes are typically parabolic; the opening angles are the broadest at the base and narrower at larger distances from the vertex of the cavity. As discussed in Velusamy & Langer (1998), to study the evolution of the outflow, the opening angle measured near the base is most relevant. When bipolar outflows are clearly seen, we use only the widest lobe for measuring the opening angle. We do not use the average of the opening angles of the outflow and the counter outflow because, in addition to the evolutionary status of the outflow, its observed characteristics may vary with the viewing geometry (inclination) and/or the environmental conditions of the protostellar and circumstellar envelopes. For the purpose of tracing the outflow evolution with age, the most critical outflow characteristic is the opening angle near the vertex, at the start of the outflow (cf. Velusamy & Langer 1998) before it is modified by any environmental effects. Therefore, we can assume that the widest opening angle in any of the lobes is the most representative of the opening angle for correlating with age. The opening angles are not corrected for inclination and are listed in Column 2 of Table 2. Note that the arrows marked in Figures 3.1(b)–3.53(b) are meant to draw attention to the presence of wide-angle outflow and they do not represent the exact opening angles. The values of the opening angles in Table 2 are measured between the boundaries of the widest outflow lobe in each object.

Following the approach of Arce & Sargent (2006), we use T_bol as a measure of the protostellar age. In Table 2, we list the values for T_bol collected from the literature, as indicated in Column 3. When a range of values are given, we use their mean value for estimating the ages. In a few cases, we estimated the T_bol values using the SEDs available in the literature and these are noted in the table. The ages were estimated from T_bol using the empirical relationship given by Ladd et al. (1998) in their Appendix (Equation (C9)). The ages and their uncertainties are listed in Column 5 in Table 2. The uncertainty in the estimated ages corresponds to the uncertainty in the empirical fit, as given by Ladd et al. In Figure 7, we show the plot of opening angle versus protostellar age for all 31 outflows in our sample (denoted by crosses). The uncertainty in the ages is also indicated. The error in the measured opening angles is small (<5°). We also included in this plot the data from Arce & Sargent (2006) represented by filled triangles. The scatter in this plot is large, but a trend in the opening angle, increasing with age, is evident. For example, we can fit power laws as shown in Figure 7: opening angle, θ(deg) = 6.7×[t(yr)]^0.32 for ages <8000 yr and θ(deg) = 88.5×[t(yr)]^0.02 for ages >7000 yr. (For the fits, we use only the data in our sample, all observed by scattered light in the processed Spitzer images). At ages ⩾ 8000 yr, the plot suggests a slower rate for the increase of opening angle with time, which would be consistent with the opening angles approaching 180° asymptotically. The scarcity of data at higher ages is due to the fact that our sample is primarily Class 0 objects with just two Class 1 objects (B5 IRS1 & HH46/47). It should be noted that we do not correct the opening angle for inclination. Although a widening trend is evident in our data, any interpretation of fit to the data is subject to the large uncertainty in their ages.

Zoom In Zoom Out Reset image size

Our results are broadly consistent with the plots in Arce & Sargent (2006) and Seale & Looney (2008). However, our values for the opening angles seem higher than those in their results. The "intensity weighted" opening angles estimated by Seale & Looney using azimuthal intensity in their circular cuts are also likely to be narrower than those measured geometrically from high spatial resolution images. Arce & Sargent used CO outflow data. Due to opacity effects, the ¹²CO may not trace the entire extent of the outflow cavity, as opposed to scattered light (Figure 5; also see Arce et al. 2013). Thus, the data from the HiRes processed images clearly provide the most consistent set of apparent opening angles for the entire extent of the cavity. Nevertheless, the observed trend in the opening angle versus age is subject to the large uncertainties in estimating the ages. Reliable age estimates combined with a more detailed characterization of the morphological shape of the cavity should provide observational constraints on the theories of outflow and star-formation processes.