THE ANATOMY OF THE YOUNG PROTOSTELLAR OUTFLOW HH 211

We observed the protostellar outflow HH 211 with Spitzer/IRS (Houck et al. 2004) on 2007 March 12, 2008 October 9, and 2009 March 7. We mapped the complete outflow including the region ahead of the SE terminal shock using the IRS short–high (SH, 9.9–19.6 μm) and long–high (LH, 18.7–37.2 μm) modules. We also obtained short–low module observations (SL2/SL1, 5.2–8.7/7.4–14.5 μm) of the SE terminal shock and the region ahead of it complementary to the existing Spitzer archival data covering the rest of the outflow. Finally, we combined our mapping observations with the single pointing observation toward the HH 211 SE terminal shock taken on 2008 October 6 (PI: B. Nisini, PID 50364, Spitzer data archive) to further improve the signal-to-noise ratio (S/N). We observed the protostellar outflows HH 111, HH 212, HH 240, HH 2, and NGC 2264 G with Spitzer/IRS (Houck et al. 2004) between 2008 October 10 and 2009 April 13. We mapped selected outflow regions using the IRS SL, SH, and LH modules. The nominal spectral resolution is R = 64–128 as a function of wavelength for the low-resolution and R = 600 for the high-resolution IRS modules. We also obtained Infrared Array Camera (IRAC; Fazio et al. 2004) image mosaics from the Spitzer data archive.

We performed the IRS data reduction consisting of background subtraction using off-source slit positions, masking of bad pixels, extracting the spectra, and generating the spectral line maps with CUBISM version 1.7 (Smith et al. 2007a). The individual steps of our data reduction are as follows.

• 1.  
  Start with the basic calibrated data as provided by the Spitzer/IRS pipeline.
• 2.  
  Import observations taken sequentially in time with each IRS module into a CUBISM project.
• 3.  
  Select and subtract an average background in CUBISM by using all slit positions devoid of source emission.
• 4.  
  Identify and exclude remaining bad pixels using CUBISM's automated statistical algorithms.
• 5.  
  Build the data cube using CUBISM's slit loss correction function (SLCF) for extended sources.
• 6.  
  Extract spectra for each source in CUBISM using custom apertures for each source (see polygons in Figures 1 and 12; nominal pixel sizes for the IRS SL/LL/SH/LH modules are 18, 51, 23, and 45, respectively).
• 7.  
  Import spectra into SMART⁴ (Higdon et al. 2004) for merging of the spectral modules and unit conversion.

Zoom In Zoom Out Reset image size

We applied scaling factors to seamlessly merge the spectra from the LH and the SL modules with the SH module to produce combined 5–36 μm spectra. The order mismatches are due to calibration issues inherent to slit spectroscopy of inhomogeneous, extended sources, which are partially corrected by CUBISM's SLCF. To address the remaining mismatches in a consistent way, we scaled the LH spectrum so that the continuum flux and the adjacent OH spectral lines in the two high-resolution modules match. The SL module spectra were scaled to match the H₂ S(2) 12.28 μm line flux in the SL2 and the SH module. The mismatches between the different modules that we corrected in this way were less than 30%. In the case of HH 211, for example, the scaling factors applied to the LH and SL modules were 1.28 and 1.15, respectively.

Observations toward the SE terminal shock of the HH 211 jet were carried out with the SMA on 2008 November 9 and 16 in the compact configuration. During the first track, the main aim was to observe the HCO⁺ 3–2 line at 267.5 GHz and the nearby HCN 3–2 line at 265.9 GHz while the second track contained the CO 2–1 line at 230.5 GHz for reference. During the first track, the zenith opacity was excellent with τ_225 GHz ∼ 0.05. During the second track, conditions were foggy with a higher opacity of τ_225 GHz ∼ 0.2–0.3 during the first few hours of the track and a stable value of τ_225 GHz ∼ 0.1 later on. Phases were very stable during the whole track. The correlator configuration was set to a frequency resolution of 0.4 MHz across the 2 GHz wide band. This resulted in a velocity resolution of ∼0.5 km s⁻¹ at the observed frequencies. The primary beam sizes for HCO⁺ 3–2 and HCN 3–2 are ∼47'' and ∼54'' for CO.

We have used the MIR software package, with quasars 3C454.3, 3C84, and 3C111 as passband calibrators, 3C84 and 3C111 as gain calibrators, and Callisto and Titan as flux calibrators. The resulting estimated absolute flux uncertainty is ∼20%. The calibrated visibilities were subsequently processed with the MIRIAD software package. The resulting synthesized beam sizes are 23 × 19, P.A. −56° for HCN and HCO⁺, and 25 × 20, P.A. −47° for CO. The rms noise level in the CO data is 37 mJy  beam⁻¹ at a velocity resolution of 1 km s⁻¹ and 23 mJy  beam⁻¹ in both HCN and HCO⁺, again at velocity resolutions of 1 km s⁻¹.

Figures 1 and 2 show the Spitzer/IRS spectrum extracted from a small region near the center of the terminal shock of HH 211 (see the dashed box in Figure 3). The main features of the spectrum, in particular the unique sequence of superthermal OH rotational lines, have been discussed earlier by Tappe et al. (2008). Here, we focus on the aspects of the spectrum that relate to the spatial structure and excitation of the gas in the SE terminal shock as well as the underlying outflow jet in the SE lobe of HH 211. In addition to the spectral features discussed by Tappe et al. (2008), we now detect 5σ emission lines of the ν₂ fundamental vibrational band Q-branches of HCO⁺ at 12.07 μm and CO₂ at 14.97 μm in the SE terminal shock. These weak lines were not seen earlier due to a lower S/N. Another notable and unusual feature of the terminal shock spectrum is the rising infrared continuum emission toward 5 μm. We discuss this aspect of the spectrum in detail in Section 4.1.1.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Figure 3 shows a selection of emission line contour maps generated with CUBISM from our Spitzer/IRS mapping observations of the SE outflow lobe (see Section 2). The contour maps are plotted on a recent near-infrared H₂ 1–0 S(1) 2.12 μm image of HH 211 (Hirano et al. 2006). The spatial resolution of the Spitzer/IRS observations at the shortest wavelength of the SL module at 5.3 μm is good enough to partially resolve the structure of the terminal shock as seen in the near-IR H₂ image (cf. FWHM beam sizes in Figure 3, middle and bottom panels).

In the top panel of Figure 3, we show the integrated emission from the superthermal OH with nearly complete coverage including the region ahead of the shock as well as improved S/N compared to our earlier work (Tappe et al. 2008). We can now clearly state that there is no discernible superthermal OH emission ahead of the terminal shock as traced by vibrationally excited H₂. Instead, the OH emission maximum seems to be just ahead of the Hα, [Fe ii] 5.34 μm, and H₂ 1–0 S(1) 2.12 μm emission maxima (see the middle panel in Figure 3). Furthermore, we note that the observed infrared 5.3 μm continuum (bottom panel of Figure 3) is spatially extended along the central jet axis and has a weak maximum coinciding with the secondary H₂ 2.12 μm maximum at the very head of the terminal shock. We discuss this interesting fact in more detail in Section 5.

The top four panels in Figure 4 show archival Spitzer/IRAC images of the SE, blueshifted outflow lobe of HH 211 (pipeline S18.18, observation program 50596 by G. Rieke, 2009 March) in comparison to H₂ 2.12 μm contours (Hirano et al. 2006). The nominal IRAC detector pixel size is 12 in all channels, and the pipeline product images are sampled to a pixel size of 06. The rising background in channels 3 and 4 is due to emission from polycyclic aromatic hydrocarbons (PAHs) at 6.2 and 7–9 μm. We measured the IRAC fluxes of the terminal shock by performing simple aperture photometry with a source annulus diameter of 12'' centered at 3^h43^m5945,  32°00'360 (J2000). The flux shows a distinct peak in IRAC channel 2, with background-subtracted fluxes for channels 1, 2, 3, and 4 of 5.2, 26.0, 13.0, and 9.6 mJy, respectively.

Zoom In Zoom Out Reset image size

We also note that the spatial flux distribution within the terminal shock is notably different in channel 2. The peak flux is located in an unresolved, point-like source at the head of the terminal shock. In the other channels, the IRAC flux peak coincides with the H₂ 1–0 S(1) 2.12 μm flux maximum, which is expected since the IRS mid-infrared spectrum of HH 211 is dominated by H₂ pure rotational lines. This difference suggests that another atomic or molecular species is responsible for the flux peak in IRAC channel 2.

To further analyze the nature of the channel 2 flux, we can use our Spitzer/IRS SL mapping data to generate IRAC channel 3 and 4 maps with CUBISM. These maps are shown in the bottom two panels of Figure 4. We have added a constant background flux to those images to account for the fact that the surrounding background emission was previously subtracted in the IRS maps. Doing the same kind of aperture photometry on the CUBISM generated maps yields background-subtracted fluxes of 8.9 and 8.5 mJy for channels 3 and 4. This includes an approximation for a few missing pixels at the edge of the aperture due to incomplete IRS coverage (<5% of the total flux).

The CUBISM-predicted channel 4 flux of 8.5 mJy is about 10% below the measured IRAC channel 4 value, which is within the expected uncertainty due to the IRS and IRAC absolute flux calibration accuracy. However, the predicted channel 3 flux is about 30% below the measured IRAC flux. Even when including an approximate upper limit flux estimate for the H₂ S(8) pure rotational line at 5.05 μm, which is not covered by the IRS spectral range, the flux is still at least 20% below the observed IRAC flux. This "missing" flux must come from emission in the 5.0–5.3 μm range, which is not covered by the IRS but within the IRAC channel 3 bandpass. We hypothesize that the missing flux in channel 3 as well as part of the channel 2 emission are due to CO v = 1–0 fundamental band lines. We discuss the reasoning behind this hypothesis in Section 4.1.1.

Figure 5 shows our SMA CO 2–1, HCO⁺ 3–2, and HCN 3–2 data. We chose the velocity intervals and color coding to highlight the different jet structures and overlaid the HCO⁺ and HCN as well as the H₂ 1–0 S(1) data contours for comparison. Despite the decreasing sensitivity beyond the FWHM of the SMA primary beam, our observations trace the CO jet all the way to the YSO. We detect several HCN knots that coincide with H₂ knots visible in the H₂ 1–0 S(1) 2.12 μm image (see white contours in Figure 5). The terminal shock also shows HCN as well as HCO⁺ emission coinciding with maxima of vibrationally excited H₂ emission. We note that HCN and HCO⁺ trace different regions of the terminal shock. The peak of the HCO⁺ emission coincides with the primary and secondary peaks of H₂ 1–0 S(1), whereas the HCN peak coincides with a small H₂ protrusion upstream of the main maxima. This protrusion is also visible in our CO 2–1 data.

Zoom In Zoom Out Reset image size

Note that the CO 2–1, HCO⁺ 3–2, and HCN 3–2 lines have different critical densities but similar upper energy levels between 15 and 26 K. The CO 2–1 line has the lowest critical density of ∼7 × 10³ cm⁻³, whereas HCN and HCO⁺ 3–2 have critical densities ∼8 × 10⁷ cm⁻³ and ∼4 × 10⁶ cm⁻³, respectively (Tielens 2005). Lines with high critical densities cause an observational bias favoring regions with gas densities similar to the respective critical density. In this case, we expect that HCN and HCO⁺ trace the denser regions of the CO jet. HCO⁺ is also a tracer of ionization, whereas HCN predominantly traces neutral or low-ionization gas in comparison. We further discuss the HCO⁺ and HCN emission in Section 5.2.

4.1.1. Rising Infrared Continuum toward 5 μm and CO Model Spectrum

Figure 6 compares the observed Spitzer/IRS SL spectrum integrated over the entire HH 211 SE terminal shock area with a local thermodynamic equilibrium (LTE) CO model spectrum at the resolution of the Spitzer/IRS SL module, R = 90, and with plausible parameters for CO column density and temperature (see the legend of Figure 6). The CO model spectra were generated with the Spectrafactory tool⁵ (see Cami et al. 2010 for further details on the model) using the Goorvitch CO line list, which contains all the rotation-vibration transitions of the fundamental, first, and second overtone bands up to v = 20 and J = 149 of the CO X ¹Σ⁺ state (Goorvitch 1994). We also show a CO model spectrum at a higher resolution of R = 2000 to illustrate the individual ro-vibrational lines as well as the four IRAC channel bandpasses. We scaled the CO model spectrum to reproduce the estimated, dereddened IRAC channel 2 excess flux that we attribute to CO (green bar in Figure 6; see further analysis below). The scaling of the LTE spectrum is necessary because the very high critical densities of the ro-vibrational CO v = 1–0 fundamental band transitions cause a departure from LTE energy level populations at the expected densities in HH 211. For that reason, we caution that the temperature and the column density assumed for the LTE CO model represent only a rough but physically plausible estimate until further revised by a non-LTE CO model (see Section 4.1.2 for further discussion). Nevertheless, we expect that the relative ro-vibrational line intensities within the CO v = 1–0 fundamental band are approximately correct and sufficiently accurate to demonstrate that the rising pseudo-continuum toward 5 μm can plausibly be attributed to CO.

Zoom In Zoom Out Reset image size

To obtain an estimate of the channel 2 excess flux beyond pure H₂ emission, we subtracted a scaled IRAC channel 1 image from channel 2. The channel 1 scale factor of 2.4 is a fairly typical value for warm H₂ (cf. Neufeld & Yuan 2008, Figures 1 and 7) and in the case of HH 211 subtracts most of the H₂ emission along the outflow. The resulting H₂-subtracted IRAC image is shown in the top panel of Figure 7. For comparison, we show the CO 2–1 SMA jet backbone and Spitzer 5.3 μm continuum contours (cf. Figures 3 and 5) as well as the optical Hα contours from Walawender et al. (2006). It is striking that the estimated channel 2 excess follows the observed Spitzer 5.3 μm continuum flux very closely. Furthermore, the channel 2 excess/5.3 μm continuum seems to be a spatial continuation of the SMA-observed CO jet backbone with the Hα emission located at the transition point between the two. These are further indications that the IRAC excess flux is indeed due to CO fundamental band emission. The prerequisites for CO fundamental band emission, high density CO gas and suitable excitation conditions, are both met in the SE terminal shock of HH 211.

Zoom In Zoom Out Reset image size

We can exclude that the excess flux is due to continuum emission from hot dust grains. Dust emission with a flux maximum shortward of 5 μm would be from hot dust and would show a strong silicate feature in emission near 10 μm. This is not observed in our Spitzer/IRS spectrum. Instead, the dust continuum peaks longward of 20 μm and can be fitted by thermal dust emission at modified blackbody temperatures of 30–170 K (Tappe et al. 2008). The H i Brα line could be another potential contribution to IRAC channel 2 emission. However, only a small area of the HH 211 SE terminal shock shows H i Hα emission and it does not spatially correlate with the IRAC channel 2 excess emission (see Figure 7, top panel).

It was previously suggested that the observed HH 211 2.14 μm near-infrared continuum, see the solid white contours displayed in the bottom panel of Figure 7, is partially due to scattered light that originates close to the protostar and escapes along the outflow path (Eislöffel et al. 2003; O'Connell et al. 2005). O'Connell et al. (2005) pointed out that the near-infrared emission nearer the YSO source traces predominantly scattered light while the emission further out mostly traces compact, H₂ line-emission features.

We can measure the near-infrared continuum flux from a 2.14 μm narrowband continuum filter image provided by Eislöffel et al. (2003). We estimate an upper limit of 1.5 × 10⁻¹⁷ W m⁻² for the dereddened 2.14 μm continuum flux of the SE terminal shock of HH 211. This upper limit includes a 35% flux uncertainty due to the calibration and low S/N. For the dereddening, we assumed a visual extinction toward the terminal shock of 10 mag and the R_V = 3.1 extinction law by Weingartner & Draine (2001) and Draine (2003).

It is very likely that most of flux observed with the 2.14 μm narrowband continuum filter is actually due to H₂ 1–0 S(1) 2.12 μm line emission. The observed, dereddened H₂ 1–0 S(1) 2.12 μm line flux for the SE terminal shock of HH 211 is 1 × 10⁻¹⁵ W m⁻² (Caratti o Garatti et al. 2006, Table 10). The 2.14 μm continuum filter has approximately a 2% transmission at the wavelength of the 2.12 μm H₂ line, which therefore results in a flux ∼2 × 10⁻¹⁷ W m⁻² leaking into the continuum filter due to the H₂ 2.12 μm line. This is approximately equal to the upper limit 2.14 μm continuum flux derived above.

A caveat for the comparison of narrowband image photometry and slit spectroscopy are the different apertures as well as calibration issues inherent to slit spectroscopy of inhomogeneous, extended sources. To avoid this issue, we can also compare to photometry from the H₂ 2.12 μm filter image by Hirano et al. (2006). Using the same aperture as for the continuum image above, we obtain a dereddened H₂ 1–0 S(1) 2.12 μm line+continuum flux of 6.8 × 10⁻¹⁶ W m⁻². Again, assuming a continuum filter transmission at 2.12 μm of 2%, we get a flux of ∼1.4 × 10⁻¹⁷ W m⁻² leaking into the continuum filter. We conclude that the SE terminal shock flux observed in the 2.14 μm continuum filter is consistent with being mostly due to H₂ 1–0 S(1) 2.12 μm line flux. This also explains why the 2.14 μm continuum filter flux contours follow the H₂ 1–0 S(1) 2.12 μm emission very closely (see Figure 7, bottom panel).

In contrast to that, the inner jet shows 2.14 μm continuum emission that does not follow the H₂ flux and this indicates the presence of scattered light. This is consistent with "negative flux" at that location in our H₂-subtracted IRAC channel 2 excess image (Figure 7, top panel). The scattered light is present in both IRAC channels 1 and 2, and hence the channel 1 scale factor chosen to subtract pure H₂ emission from channel 2 oversubtracts scattered continuum flux. We observe the same "negative" flux for the inner region of the opposite NW lobe of HH 211, which is not shown in this work. We conclude that scattered light is only present in the inner regions of the jet lobes, whereas the terminal shock emission seen in the 2.14 μm narrowband filter is entirely due to H₂ 2.12 μm line emission.

4.1.2. IRAC Color–Color Ratios: Comparison to H₂ and CO Models

Another way to quantitatively analyze the CO emission in IRAC channel 2 is to compare the observed IRAC fluxes to predictions by non-LTE models of H₂ and CO line emission. Unfortunately, detailed data on collisional rate coefficients for the CO v = 1–0 fundamental band at different temperatures up to a few thousand kelvin and up to high rotational J-levels are currently unavailable. Due to the very high critical densities of the ro-vibrational CO v = 1–0 fundamental band transitions, a non-LTE modeling approach is necessary to correctly predict the absolute line strengths. The best possible approach with the existing molecular data is described by Neufeld & Yuan (2008), who calculated the predicted IRAC color–color ratios for pure H₂ emission using non-LTE models and estimated the fractional contribution of CO to IRAC channel 2 (see Figures 9 and 11 in Neufeld & Yuan 2008).

We can measure the IRAC color–color ratios of the SE terminal shock in HH 211 and compare this to H₂ non-LTE models in a color–color diagram (see Figure 8). For this purpose, we consider extended H₂ non-LTE models up to densities of 10⁹ cm⁻³ with 51 H₂ energy levels up to about 20,000 K and collisional excitation by H₂/He followed by spontaneous decay (based on Neufeld & Yuan 2008, extended calculations by Y. Yuan 2011, private communication). UV excitation is not taken into account. The models are displayed in the IRAC color–color diagram in Figure 8. Red lines represent the expected IRAC band ratios for H₂ emission at different densities, with n(H₂) varying from 10⁴ to 10⁹ cm⁻³. We adopt a power-law temperature distribution as described in Neufeld & Yuan (2008), with the observed column density of gas at temperatures between T and T + dT proportional to T^−b and a temperature range of 100 to 5000 K. It is expected that both decreasing the power-law index b, i.e., having a larger fraction of hot gas, and increasing the density n(H₂) would raise the 4.5 μm/8.0 μm and the 5.6 μm/8.0 μm band ratios. In Figure 8, the solid red lines are for models with b in the range of 6–2.5 from bottom left to top right. The dashed lines indicate the extensions of b from 2.5 to the negative zone, and smaller values of b mean a larger fraction of hot H₂ gas closer to the assumed maximum temperature of T_max = 5000 K.

Zoom In Zoom Out Reset image size

The dereddened IRAC colors of HH 211 are indicated by the open squares. We assumed a visual extinction of 10 mag toward HH 211. We show the measured colors of the SE terminal shock both with and without subtraction of our estimated CO fundamental band contribution and the colors of the entire outflow integrated over both lobes and the central YSO. We discuss the additional data displayed in this figure in Section 5.1. Note that the x-axis IRAC 4.5 μm/8.0 μm ratio is not affected by extinction to good approximation. The effect of 20 mag of visual extinction on the y-axis IRAC 5.6 μm/8.0 μm ratio is indicated quantitatively by the solid arrows in the bottom right corner of Figure 8. The dashed arrows qualitatively indicate the shift that higher H₂ excitation and density as well as the presence of CO fundamental band emission would cause in the color–color diagram.

The dereddened IRAC log(4.5 μm/8.0 μm) and log(5.6 μm/8.0 μm) ratios toward the terminal shock of HH 211 are 0.44 and 0.10, respectively. These colors are an indicator that H₂ emission alone is an unlikely cause of the observed IRAC channel ratios even when considering very high H₂ densities ∼10⁹ cm⁻³. How does this picture change if we subtract the estimated CO fundamental band emission contributing to the IRAC channels 2 and 3? We can obtain a fairly accurate value for the log(5.6 μm/8.0 μm) ratio excluding CO by using the CUBISM maps generated from our Spitzer/IRS spectra. If we assume the H₂ S(8) pure rotational line flux at 5.05 μm to be equal to the H₂ S(6) line flux, we derive an IRAC log(5.6 μm/8.0 μm) ratio of 0.08 ± 0.05. This ratio is dominated by H₂ emission.

The log(4.5 μm/8.0 μm) IRAC ratio excluding CO is harder to estimate since we cannot directly measure the H₂ lines that fall in the IRAC channel 2 bandpass. In the following, we use the H₂-subtracted "IRAC channel 2 minus 2.4 ×channel 1" image in Figure 7, top panel, as a surrogate for the CO fundamental band contribution until direct spectroscopic measurement and/or non-LTE modeling of CO fundamental band emission become available. Subtracting this CO fundamental band flux contribution from the observed IRAC channel 2 flux yields an approximate H₂ line-emission flux of 12.5 mJy in that channel, i.e., 2.4 times the IRAC channel 1 flux as we assumed above (see Section 4.1.1). With that, we derive a log(4.5 μm/8.0 μm) IRAC ratio of 0.12 and an uncertainty range of −0.2 to 0.3. The latter assumes a 50% uncertainty for the estimated CO contribution to the IRAC channel 2 flux. Within the estimated errors, the IRAC color–color ratios excluding CO are consistent with H₂ emission from molecular gas with densities of ∼10⁶ to ≳ 10⁹ cm⁻³ (see open square with error bars in Figure 8).

To further test the reliability of our estimated CO fundamental band contribution, we can compare to model predictions of the CO fractional contribution to the IRAC channel 2 by Neufeld & Yuan (2008, Figure 11). A 50% CO contribution to IRAC channel 2 with a 50% error as we estimated above corresponds to a logarithmic fractional CO contribution of −0.3 with an uncertainty range of −0.1 to −0.6. Depending on the assumed H/H₂ abundance ratio, this would require H₂ densities of at least 10⁶ (H/H₂ = 1) and up to approximately 10⁹ cm⁻³ (H/H₂ = 0) according to the model of Neufeld & Yuan (2008, Figure 11). These densities are in good agreement with the estimated H₂ gas densities above, so we conclude that our derived CO contribution to channel 2 of 50% is a reasonable value supported by both theoretical models and our observations.

The very high inferred molecular gas densities of about 10⁶ to ≳ 10⁹ cm⁻³ are also plausible in light of the very high critical densities ≳ 10¹¹ cm⁻³ of the CO v = 1–0 ro-vibrational transitions as well as the Q-branches of the ν₂ v = 1–0 fundamental vibrational bands of HCO⁺ at 12.07 μm and CO₂ at 14.97 μm (see Section 3.1). The high critical densities of these transitions, i.e., a very large ratio of the respective Einstein A_ul value over the sum of all collisional rate coefficients out of the upper energy level u, mean a strong observational bias toward high density gas. Interestingly, we did not detect the Q-branch of the ν₂ v = 1–0 fundamental vibrational band of HCN near 14.0 μm, which should have comparable critical densities to the detected Q-branches of the HCO⁺ and CO₂ ν₂ fundamental bands. The HCN band is partially blended with high-J OH lines at 14.08 μm but should be clearly distinguishable. Since we clearly detected the HCN 3–2 pure rotational transition with the SMA at locations spatially offset from HCO⁺, we surmise that the observed HCN emission originates from cooler, non-ionized gas, consistent with our non-detection of vibrationally excited HCN gas in our Spitzer spectra (see Section 5.2 for further discussion).

Finally, the volume density of the CO knots in the inner region of the jet was estimated to be n ≳ 2.8 × 10⁶ cm⁻³ by Lee et al. (2010). It is likely that the density is even larger in the terminal shock due to gas compression. Since HH 211 is a very young outflow with clearly discernible, dense knots along the jet, it is also likely that multiple dense knots stack up or "slam" into each other in the terminal shock region where the jet speed slows down drastically.

A decisive observational test for the presence of the CO v = 1–0 fundamental band in HH 211 would be to obtain a medium-to-high-resolution 4–5 μm M-band infrared spectrum of the SE terminal shock. This is achievable with high-resolution spectrographs at large, ground-based telescope facilities, see, for example, the recent survey of the CO v = 1–0 fundamental band in protoplanetary disks with the Very Large Telescope (Bast et al. 2011), and also airborne or space-based telescopes like Sofia or AKARI.

Cyganowski et al. (2008) identified a large number of extended sources with excess 4.5 μm IRAC channel 2 emission in the Spitzer Galactic Legacy Infrared Mid-Plane Survey Extraordinaire. These sources were subsequently named EGOs or "green fuzzies," for the common coding of the 4.5 μm IRAC channel 2 as green in three-color composite images (Cyganowski et al. 2008; Chambers et al. 2009).

The origin of the excess 4.5 μm IRAC channel 2 emission is still a matter of investigation because the Spitzer/IRS coverage does not extend to wavelengths shorter than 5 μm. A known observational fact is that EGOs are for the most part associated with YSOs and outflows in star-forming regions (e.g., Cyganowski et al. 2008, 2011; Chambers et al. 2009). De Buizer & Vacca (2010) presented a direct spectroscopic investigation of two EGO sources, which observationally demonstrated that pure H₂ emission can be the origin of some of the EGOs. However, Takami et al. (2010) noted that CO fundamental band emission is the most likely candidate for the excess 4.5 μm IRAC emission observed in several HH objects, including HH 211. This is consistent with our analysis of HH 211 and supported by the observed rise in continuum toward 5 μm in our Spitzer spectrum.

Figure 8 shows a comparison of EGO candidates from Cyganowski et al. (2008), the HH objects from Takami et al. (2010), and HH 211 from our work in an IRAC color–color diagram. We note that many EGO candidate sources are consistent with the model predictions for pure H₂ emission from high density gas. For example, G19.88-0.53, the EGO source that was spectroscopically identified by De Buizer & Vacca (2010) as pure H₂, has a log(4.5/8.0 μm) ratio of −0.15 and a log(5.6/8.0 μm) ratio of 0.06. For the EGO sources that are consistent with H₂ model predictions, we cannot unambiguously distinguish between pure H₂ emission and H₂ plus fundamental CO band emission. However, CO contributions larger than ≈10% to IRAC channel 2 are unlikely if the H₂ density is below 10⁶ cm⁻³ (see Neufeld & Yuan 2008, Figure 11). Thus, we would expect significant CO fundamental band emission only for EGOs lying to the right of the H₂ models for n = 10⁶ and 10⁹ cm⁻³ in Figure 8. This is the case for approximately 20% of the EGO candidates from the Cyganowski et al. (2008) sample. We surmise that these sources are the best candidates for significant contributions of CO emission to the IRAC channel 2 flux.

It is likely that many of the extended EGO source photometries in the Cyganowski et al. (2008) sample contain continuum emission from YSOs due to confusion and the insufficient spatial resolution of IRAC. Therefore, we indicated the IRAC color–color location of YSO sources from the survey by Gutermuth et al. (2009) in Figure 8. Continuum-dominated Class II/II* YSOs could explain the high log(4.5/8.0 μm) ratios observed for some of the EGO candidates. For example, TW Hya (see the filled triangle symbol in Figure 8) is a well-studied nearby T Tauri star with a transition disk. The archival and published Spitzer spectra as well as the published near-infrared 3–5 μm spectra of TW Hya show that its IRAC color–color ratios are dominated by YSO continuum emission (Najita et al. 2010; Vacca & Sandell 2011) but the spectra also show H i Brα and Pfβ and CO fundamental band emission in the 4–5 μm region (Salyk et al. 2007; Vacca & Sandell 2011). The Spitzer SH spectrum presented by Najita et al. (2010) shows some of the same molecular lines observed in HH 211, i.e., high-J OH, the Q-branches of the HCO⁺ and CO₂ ν₂ v = 1–0 fundamental vibrational bands but only weak H₂ lines. The non-detection of the 14.0 μm Q-branch of the HCN ν₂ v = 1–0 fundamental band is also notable, which is also the case in HH 211 (see Section 4.1.2). The mid-infrared spectrum of TW Hya is dominated by dust continuum emission as well as hydrogen recombination lines and [Ne ii] 12.8 μm (see, e.g., Gorti et al. 2011 for an overview of detected emission lines).

We can exclude YSO continuum emission for the spatially resolved HH objects in the sample of Takami et al. (2010) as well as for our detailed investigation of HH 211. The group of EGOs with high log(5.6/8.0 μm) ratios larger than 0.5 in the top right of Figure 8 are most likely deeply embedded YSOs with visual extinctions greater than 20 mag. This is supported by a visual inspection of their IRAC and MIPS images, which shows that these objects are bright 24 μm sources located in or adjacent to patches that are dark at 8 μm. This is most likely due to the high extinction from the wing of the 10 μm silicate dust absorption feature. A dereddening of these sources would move them downward toward the dashed section of the H₂ models in Figure 8, so it is possible that there is a contribution from highly excited H₂ emission in these sources.

In summary, EGO sources whose IRAC channel 2 excess is dominated by CO fundamental band emission are most likely characterized by the following physical conditions in the CO emitting region: gas densities larger than about 10⁷ cm⁻³, temperatures ≳ 1000 K that are hot enough to excite the CO fundamental band with upper energy levels larger than 3000 K, and potentially the presence of atomic H gas/UV radiation. Another condition that could further increase CO fundamental band emission is a CO/H₂ abundance ratio larger than the standard value of 10⁻⁴, e.g., due to spatially localized ice evaporation from interstellar grains due to shocks and/or UV irradiation. These processes could play a role in the terminal shock of HH 211 (Tappe et al. 2008).

Figure 9 shows the major structural features of the HH 211 SE outflow lobe. We can distinguish between three different regions: (1) the jet origin including the point of emergence from the surrounding protostellar envelope and the inner jet region, (2) the outer jet region where the jet expanded into the surrounding medium, and (3) the terminal shock, where the jet currently interacts with the surrounding medium. The inner jet region including the origin and emergence from the protostellar envelope is not the focus of this paper and we refer to the work by Hirano et al. (2006), Lee et al. (2007, 2009, 2010), and Tanner & Arce (2011) for further discussion.

Zoom In Zoom Out Reset image size

After the jet emerges from the protostellar envelope, we note the presence of reflected light, which probably originated close to the YSO and escaped along the path that the outflow carved into the surrounding envelope. Further along the jet, we note a number of dense knots that are traced by vibrationally excited H₂ and the high critical density HCN 3–2 line. About two-thirds along the way from the YSO toward the terminal shock, the jet seems to be deflected northward from its otherwise relatively straight path. The deflection point shows HCN 3–2 emission as well as a wall of vibrationally excited H₂ south of the jet. The dense H₂/HCN knots are either due to irregular outflow events, also known as "pulsed" outflows, or due to internal shocks in the jet. We note that the knots do not show any notable [Fe ii] or HCO⁺ emission lines. Thus, the gas is hot enough to show vibrationally excited H₂ and dense enough to emit HCN 3–2 but not ionized. As noted in Section 4.1.2, the volume density of the knots in the inner region of the jet was estimated to be ≳ 10⁶ cm⁻³ (Lee et al. 2010).

The terminal shock is a very complex region because it is here that the jet hits the surrounding interstellar medium, gets shocked, and rapidly slows down in velocity. If the outflow is pulsed or has dense knots as HH 211, dense, fast-moving gas parcels that approach the slowed-down terminal shock gas may cause sporadic shock bursts upon impact. At the time of our observations, the point in the terminal shock where most of the jet's energy is currently released is presumably the Hα emitting region that coincides with the H₂ 1–0 S(1) primary maximum, the HCO⁺ 3–2 maximum, and the [Fe ii] 5.3 μm maximum. Kinematically speaking, this is also the spot where the collimated, blueshifted CO jet slows down and splits up into new velocity components (see Figure 5).

The Hα, HCO⁺, and [Fe ii] emission indicate the presence of ionizing UV radiation, which is consistent with the fact that the observed OH emission maximum lies just upstream of this region. In our earlier paper, we described how UV dissociation of H₂O is the likely cause of the unique high-J OH emission in HH 211 (Tappe et al. 2008). It is clear that the OH emission does not originate in cool post-shock gas. The origin of the H₂O as the precursor to the OH emission can be chemistry in the dense, shocked gas, and/or photoevaporation of grain ice mantles. We will explore these possibilities further using detailed non-LTE models and comparisons to other outflows in a subsequent paper.

The gas in the terminal shock gets heated and compressed. The densest areas, presumably part of the collimated outflow backbone, show fundamental band emission from vibrationally excited CO (gray area in the terminal shock in Figure 9). The observed maximum of that emission coincides with the secondary H₂ and HCO⁺ maximum near the head of the terminal shock. Finally, the HCN 3–2 maximum coinciding with the small H₂ protrusion at the very head of the terminal shock is a bit of a mystery. This gas is presumably dense but not ionized, since it lacks HCO⁺ and [Fe ii] emission. It may stem from a prior outflow event or it may be a specific feature of the turbulent terminal shock gas.

The outflow from the NGC 1333 IRAS 4B protostar (hereafter IRAS 4B) is an ideal candidate for a comparison to HH 211. Both outflows are dynamically young, highly collimated, and part of the Perseus molecular cloud complex at approximately the same distance of 260 pc. The inclination angle of HH 211 is near the plane of the sky (5° to 10°; see O'Connell et al. 2005; Lee et al. 2007), and the dynamical age is ∼500 years based on the distance between the protostar and the terminal shock and an assumed average outflow velocity of 100 km s⁻¹. The inclination angle of IRAS 4B is not well constrained. Therefore, we assume a wide range of inclinations between 10° and 80°, an average outflow velocity of 100 km s⁻¹, and thus infer a dynamical age between 100 and 500 years.

IRAS 4B has a rich mid-infrared H₂O emission line spectrum, which was initially interpreted as a spatially unresolved accretion shock in its circumstellar disk (Watson et al. 2007). However, additional interferometric observations suggested that the H₂O emission originates spatially offset from the protostellar source and is instead associated with the outflow of IRAS 4B (Jørgensen et al. 2007; Jørgensen & van Dishoeck 2010). Furthermore, Kristensen et al. (2010) observed broad H₂O line profiles spanning ≈70 km s⁻¹ at their base toward IRAS 4B with the Herschel/HIFI spectrograph. They concluded that the bulk of the observed H₂O emission originates in shocked gas where H₂O is presumably released from grains by means of sputtering. The shock origin is further supported by subarcsecond SMA observations of CS, CH₃OH, and H₂CO, which prominently trace the blueshifted southern and redshifted northern outflow lobes of IRAS 4B (Jørgensen et al. 2007). Jørgensen et al. (2007) noted that these species are located on the tips of the CO outflows and concluded that they likely probe shocks where the outflow impacts the surrounding envelope.

In this paper, we reanalyzed the available Spitzer archival data for IRAS 4B to comparatively discuss the origin of OH and H₂O emission in HH 211 and IRAS 4B. We used the same data reduction routine as described for HH 211 in Section 2. The top panel of Figure 10 shows the archival IRAC channel 2 image of IRAS 4B (pipeline S18.18, observation program 30516 by L. Looney, 2007 February) together with CO, H₂, and N₂H⁺ contours. Figure 11 shows the reanalyzed Spitzer/IRS spectrum extracted from the southern lobe of IRAS 4B. Unfortunately, there is no Spitzer SH data that cover the spatial position of the outflow, so we only show the SL and LH data extracted from the same spatial outflow position (see the top panel of Figure 10 for the extraction region). Our reanalysis demonstrates that the rich H₂O emission observed in the LH spectrum spatially coincides with the southern outflow lobe maximum (see Spitzer LH H₂O emission line map in Figure 10, bottom panel). In an independent study performed contemporaneously to our work, Herczeg et al. (2011) confirm that the highly excited H₂O emission from IRAS 4B is produced in the outflow rather than the envelope-disk accretion shock.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Interestingly, the northern lobe counterpart does not show any detectable H₂O, IRAC channel 2, or near-infrared H₂ 1–0 S(1) emission (for a near-IR H₂ map of the IRAS 4 region, see Choi et al. 2006). However, the northern lobe is clearly observed with SMA in the same molecular lines as the southern lobe by Jørgensen et al. (2007), which indicates that high extinction blocks the infrared emission from the northern lobe. The line of sight toward the northern, redshifted lobe of the outflow crosses a large part of the protostellar core (see N₂H⁺(1–0) contours in Figure 10, top panel), which is likely to cause very large visual extinctions. The path of the southern, blueshifted lobe leads toward the boundary of the core facing the observer and thus has a much lower extinction in comparison to the northern lobe. For the southern lobe, we estimated a visual extinction of A_V ∼ 15 mag from a best-fit model to the observed Spitzer H₂ emission lines (95% confidence interval of A_V ⩽ 38 mag assuming 20% flux errors and using the same H₂ non-LTE model as described in Section 4.1.2). This is consistent with the estimated large-scale foreground extinction toward IRAS 4B of A_V ∼ 9 mag (see extinction map by Gutermuth et al. 2008).

5.3.1. Shock Speed and Emission Lines

To our knowledge at the time of this writing, HH 211 is the only protostellar outflow showing strong high-J OH emission lines. Therefore, we designed a follow-up Spitzer survey with the aim of finding further examples of highly excited OH occurring together with H₂O and H₂ in protostellar outflows (see the Appendix). The focus of the survey was to select a sample of outflows with a range of properties that supports high excitation and to determine the physical conditions that lead to the formation of high-J OH. As shown in the Appendix, we successfully detected high-J OH in two more outflows, HH 240 and HH 2, albeit at much lower line intensities and S/N compared to HH 211 (see the Appendix, Figure 19 and Table 2). None of these two outflows shows any detectable H₂O lines in our spectra, which is consistent with the notion of high-J OH production via H₂O photodissociation and the absence of grain ice evaporation in those sources. HH 240 and HH 2 show a higher degree of ionization compared to HH 211, indicating faster shocks and stronger UV radiation fields (see, e.g., the atomic Fe, Ne, and S lines listed in the Appendix, Table 2). Thus, we would expect that the chemical OH/H₂O balance is shifted even more strongly toward OH in those cases (cf. Bethell & Bergin 2009), which makes the detection of H₂O in those cases unlikely.

We note in this context that the Lyα photodissociation cross section of H₂O is about a factor of 10 larger compared to OH (Kalyanaraman & Sathyamurthy 1994; Vatsa & Volpp 2001) and several orders of magnitude larger at energies >12 eV (e.g., Budzien et al. 1994, excluding photoionization). As a consequence, the far-UV (FUV) photodissociation of H₂O is faster compared to OH on a per molecule basis, and increased FUV radiation tends to produce a temporary OH abundance enhancement with respect to H₂O. The presence of high-J OH emission lines is an indicator of this process, which continues to take place as long as there is a sufficient supply of H₂O, e.g., due to shock chemistry or ongoing grain ice evaporation. The high-J OH emission is also independent of the thermal excitation conditions because OH is formed directly in the high-J state via a non-thermal excitation process, namely the FUV photodissociation of H₂O (Tappe et al. 2008). This explains, for example, the ubiquitous detection of high-J OH in the inner disks of T Tauri stars (see, e.g., Carr & Najita 2011), where both UV radiation and H₂O are present. Fogel et al. (2011) pointed out that although increased UV/Lyα radiation normally dissociates both H₂O and OH, it may actually increase the H₂O and OH abundance due to grain ice UV photodesorption. This is consistent with the fact that many of the T Tauri stars in the sample of Carr & Najita (2011) show also C₂H₂, HCN, and CO₂ fundamental band emission.

As described in the previous section, grain ice evaporation is probably the reason for the very high H₂O abundance in IRAS 4B. In this case, however, the lack of strong UV radiation inhibits high-J OH emission in favor of low excitation OH lines (see the cross-ladder transitions in Figure 11 and Table 1). The detail of the IRAS 4B H₂O modeling presented by Herczeg et al. (2011), who used a combination of Spitzer and Herschel spectra that cover both high and low energy levels, yields a gas phase water column density of 1 × 10¹⁷ to 4 × 10¹⁹ cm⁻². The H₂ column density inferred from the H₂ models of our Spitzer spectra over similar emitting areas is 3 × 10¹⁹ to 3 × 10²⁰ cm⁻². Thus, we can infer an approximate H₂O abundance with respect to H₂ in the range of 5 × 10⁻⁴ to 1, with the most likely value of 8 × 10⁻³. This is a very high abundance and most likely a consequence of grain ice evaporation (cf. Bergin & van Dishoeck 2011, Section 4(c)).

Finally, we can compare our results to the outflow L1157, a particularly illustrative example where extended 179 μm H₂O emission was recently detected with Herschel (Nisini et al. 2010a). Surprisingly, we detect no H₂O or OH emission in the Spitzer spectra of L1157, with an upper limit of 0.05 × 10⁻¹⁷ W m⁻², based on our reanalysis of Spitzer archival data for L1157 knot B1 (cf. Neufeld et al. 2009; Nisini et al. 2010b). This is most likely an abundance as well as an excitation effect. Nisini et al. (2010b) estimated a H₂O abundance of (0.3–3) × 10⁻⁴ L1157, which is at least an order of magnitude lower compared to IRAS 4B. In addition, the OH and H₂O lines in the Spitzer mid-infrared range are only becoming strong and easily detectable at gas densities larger than 10⁶–10⁸ cm⁻³ and preferably large temperatures well above 1000 K, because of the very high critical densities and high energy levels of those lines. We note that this excitation effect can easily be demonstrated quantitatively by excitation models using the RADEX code (van der Tak et al. 2007). As a consequence, the detections of H₂O and OH lines with Spitzer are biased toward high density/high temperature regions. Herschel can detect H₂O in low density/low temperature regions due to its access to lines at much longer wavelengths that originate from lower energy levels. In the case of L1157, Nisini et al. (2010b) quote a temperature of 300–500 K and a density of (1–5) × 10⁵ cm⁻³. As a consequence, the non-detection of strong OH/H₂O lines in L1157 and likewise in many other outflow sources with Spitzer is understandable.