Shocked water in the Cep E protostellar outflow

We observed the line of para-H₂O $3_{13}\rightarrow 2_{20}$ at 183.3 GHz at the IRAM 30m telescope in January 2009, near the transit of Cep E between $61^{\circ}$ and $65^{\circ}$ of elevation. Both receivers at 2mm were used simultaneously. Observations were carried out in wobbler switching mode with a throw of $180\arcsec$. Filterbanks with 1MHz spectral resolution and an autocorrelator providing a spectral resolution of 1.25 MHz (2 km/s) were used as spectrometers. Weather conditions were very good and stable, with an atmospheric optical depth $\approx 1.10$ along the line of sight. Calibration was carefully monitored every 4 minutes between two integration scans on Cep E. Water vapor and system temperatures were found to remain stable, in the range $1650-1800\hbox{\kern 1.99997ptK}$. Any fluctuation of the sky emissivity between the on and off positions would result in a poor spectroscopic baseline and detection of the mesospheric water line, which is not the case in the present observations. Pointing was checked against nearby quasars every hour and has an accuracy of $\approx 2\arcsec$. At the frequency of the H₂O line, the main-beam efficiency of the telescope is 0.63 and the half-power beamwidth is $13.2\arcsec$. We did a small oversampled map at $6\arcsec$ sampling of the southern lobe of the outflow, centered on the nominal position of the protostar $23^{h}03^{m}13.0^{s}$ $+61^{\circ}42\arcmin 21\arcsec$ (J2000). The typical rms is 100mK per velocity interval of $2\hbox{\kern 1.99997ptkm\kern 1.99997pts${}^{-1}$}$.

The SiO $J=8\rightarrow 7$ transition at 347.33006 GHz was observed towards the protostar at the JCMT in August 2005. The weather conditions were good with an average system temperature $\rm T_{sys}\sim 400\hbox{\kern 1.99997ptK}$. Observations were carried out with a nutating secondary, adopting a throw of $60\arcsec$. The DBS was used as spectrometer, providing a spectral resolution of 625kHz. The resolution was degraded to 3.75 MHz, corresponding to a final velocity resolution of about $3.2\hbox{\kern 1.99997ptkm\kern 1.99997pts${}^{-1}$}$. At the frequency of the SiO line, the main-beam efficiency of the telescope is about 0.60 and the half-power beamwidth is $\simeq 14.2\arcsec$, very similar to the telescope parameters of the p-H₂O observations. Complementary interferometric observations at $2\arcsec$ resolution with the Plateau de Bure in the SiO $J=2\rightarrow 1$ transition provide us with the spatial distribution of the emission and its excitation conditions. These observations will be presented and discussed in detail in a forthcoming paper (Lefloch et al. in prep.).

The map of the p-H₂O $3_{13}-2_{20}$ emission in Cep E is displayed in the left panel of Fig. 1. Emission is detected only towards the protostellar core, in a region unresolved by the main beam of the IRAM 30m telescope. In particular, we do not detect any H₂O emission towards HH77, the southern bowshock at offset position ($-5\arcsec$, $-15\arcsec$), associated with bright emission in the lines of SiO and H₂ (L96). To improve the signal-to-noise ratio of the data, we averaged the emission of the water line over the inner core, in the range $\Delta\alpha=$ [$1\arcsec;-5\arcsec$] $\Delta\delta=$ [$1\arcsec;-5\arcsec$]. The averaged spectrum is displayed in the right panel of Fig. 1, and the final rms noise is 30mK ($T_{A}^{*}$) per interval of $2\hbox{\kern 1.99997ptkm\kern 1.99997pts${}^{-1}$}$.

Two bright components ($\sim 1\hbox{\kern 1.99997ptK}$) with broad linewidths ($\sim 10\hbox{\kern 1.99997ptkm\kern 1.99997pts${}^{-1}$}$) are detected at about $v_{lsr}\approx-65$ and $-45\hbox{\kern 1.99997ptkm\kern 1.99997pts${}^{-1}$}$ in the direction of the protostar. Two other, weaker, H₂O components are detected at approximately symmetric (redshifted) velocities $v_{lsr}\approx+34$ and $+50\hbox{\kern 1.99997ptkm\kern 1.99997pts${}^{-1}$}$. We do not detect any emission at the velocity of the protostellar envelope ($v_{lsr}=-10.9\hbox{\kern 1.99997ptkm\kern 1.99997pts${}^{-1}$}$; L96), hence excluding any contribution from the hot corino to the water emission. The presence of both blue- and redshifted components in the water line spectrum indicates that the emission arises from the central protostellar region. This is because there is barely any overlap in the plane of the sky between the blue and redshifted wings of the Cep E outflow, except in the central protostellar region as a consequence of the limited angular resolution of the observations (see Figs. 1-2; also L96).

The spectrum of the SiO $J=8\rightarrow 7$ emission in the direction of the protostar is displayed in Fig. 1 and looks very similar to the p-H₂O $3_{13}-2_{20}$ line spectrum. Resemblance between both emissions is even more striking when comparing the water spectrum with the $J=2\rightarrow 1$ emission observed with the PdBI (Fig. 2). This suggests that the emission of both tracers arises from the same region. Observations of the $J=2\rightarrow 1$ transition with the PdBI of the central protostellar regions (Lefloch et al. in prep) show that the SiO emission arises from two compact clumps, in the blue- and redshifted gas, respectively, located at $\approx 1.5\arcsec$ from the protostar. They have a typical size of $2.5\arcsec$ ($2.3\times 10^{16}\hbox{\kern 1.99997ptcm}$), and a transverse size of $1\arcsec$ (Fig. 2). Their similar physical properties (size, velocity, distance) suggest they are related to an episodic ejection phenomenon. In what follows, we assume that both the H₂O and the SiO emissions have the same size.

The gas kinetic temperature $\rm T_{k}$ in the redshifted, high-velocity jet was previously determined by Hatchell et al. (1999) from JCMT observations of the CO $J=4\rightarrow 3$ and $J=2\rightarrow 1$ lines. Unfortunately, their analysis is biased by a highly overestimated value of the main-beam efficiency $\eta_{MB}$ of the $J=4\rightarrow 3$ observations, poorly known at that time, leading to an underestimate of $T_{k}$ ($\approx 50\hbox{\kern 1.99997ptK}$). Based on their observations and adopting a standard value consistent with the subsequent determinations from point-like calibrators²²2http://www.jach.hawaii.edu/JCMT/spectral$\_$line/Standards/, $\eta_{MB}=0.35$, we obtain the jet kinetic temperature and column density, $T_{k}\simeq 120\hbox{\kern 1.99997ptK}$ and $\rm N(CO)\simeq 2.5\times 10^{17}\hbox{\kern 1.99997ptcm${}^{-2}$}$, respectively (see also L96). Adopting a standard CO abundance in the gas, we then derive the source-averaged gas column-density $\rm N(\hbox{H${}_{2}$})\simeq 2.5\times 10^{21}\hbox{\kern 1.99997ptcm${}^{-2}$}$. We adopt these values of $\rm T_{k}$ and N(CO) for both water-emitting clumps at blue- and redshifted velocities.

The gas density and column density in the water-emitting clumps were derived from modeling the H₂O and SiO line brightness temperatures in the large-velocity gradient (LVG) approach. We adopted the collisional coefficients of Faure et al. (2007) and Dayou & Balança (2006), respectively. The gas density can be constrained reasonably well from the ratio of the SiO $J=8\rightarrow 7$ to $J=3\rightarrow 2$ brightness temperatures. This ratio was computed using the SiO $J=3\rightarrow 2$ observations in the Cep E protostellar core obtained by L96 (see also Fig. 3). We note that the source is unresolved by the single-dish JCMT and IRAM 30m observations; therefore, we approximated the ratio of the line brightness temperatures by the ratio of the main-beam brightness temperatures corrected for the difference of main-beam solid angles $\rm R(8-7/3-2)$. A description of the radiative transfer code and the procedure applied to deal with the masing effect is presented in Cernicharo et al. (1994) and González-Alfonso & Cernicharo (1993).

The top right panel of Fig. 3 displays the variations of $\rm R(8-7/3-2)$. The ratio varies little in the low-velocity gas and is about 0.2. It increases as a function of velocity and reaches values up to 0.7 and 1.5 in the redshifted and blueshifted gas, respectively. The variations of $\rm R(8-7/3-2)$ as a function of $\rm T_{k}$ and $\rm n(\hbox{H${}_{2}$})$ were computed for a typical column density $\rm N(SiO)=2\times 10^{14}\hbox{\kern 1.99997ptcm${}^{-2}$}$ and a linewidth $\Delta V=10\hbox{\kern 1.99997ptkm\kern 1.99997pts${}^{-1}$}$, and they are displayed in Fig. 3. With the constraint $T_{k}\geq 100\hbox{\kern 1.99997ptK}$, it comes out that $n(\hbox{H${}_{2}$})$ must be about $1.0\times 10^{6}\hbox{\kern 1.99997ptcm${}^{-3}$}$ in order to account for the line intensity ratio in the redshifted gas . The density in the blueshifted gas is slightly higher $\approx 2\times 10^{6}\hbox{\kern 1.99997ptcm${}^{-3}$}$. These calculations show that the ratio depends weakly on $T_{k}$ above $100\hbox{\kern 1.99997ptK}$. For these values of $n(\hbox{H${}_{2}$})$ and the kinetic temperature estimated above, we estimate the source-averaged column density necessary to account for the $J=8\rightarrow 7$ line intensity $\rm N(SiO)=(2.0-3.3)\times 10^{14}\hbox{\kern 1.99997ptcm${}^{-2}$}$ for gas at $120\hbox{\kern 1.99997ptK}$ in the blue- and redshifted components. Under such physical conditions both SiO transitions are optically thin, so that R(8-7/3-2) depends weakly on the adopted value of the gas column density. From comparison with N(CO) and assuming a standard abundance $\rm[CO]/[\hbox{H${}_{2}$}]=10^{-4}$, we obtain the relative SiO abundance in the shock : [SiO]/[$\hbox{H${}_{2}$}]\simeq 1.0\times 10^{-7}$, similar to the abundances measured in other Class 0 protostellar outflows (Lefloch et al. 1998).

The para-water (source-averaged) column density $\rm N(p-\hbox{H${}_{2}$O})$ was determined from modeling the line brightness temperature $\rm T_{B}$. The latter was derived from the main-beam brightness temperature $T_{MB}=T_{B}\times ff$, where the filling factor $ff$ was obtained assuming that the H₂O and SiO $J=2\rightarrow 1$ emitting regions have similar sizes, in agreement with Sect. 3.1. We find $T_{B}=60-80\hbox{\kern 1.99997ptK}$ and $15-29\hbox{\kern 1.99997ptK}$ in the blueshifted and redshifted gas, respectively. The brightness temperature of the p-H₂O $3_{13}-2_{20}$ line was computed for a wide range of physical conditions, adopting a typical linewidth $\Delta v=10\hbox{\kern 1.99997ptkm\kern 1.99997pts${}^{-1}$}$ : $\rm T_{k}$ between 50 and $200\hbox{\kern 1.99997ptK}$, $\rm n(\hbox{H${}_{2}$})$ in the range $10^{3}-10^{8}\hbox{\kern 1.99997ptcm${}^{-3}$}$ and $\rm N(p-\hbox{H${}_{2}$O})$ in the range $3\times 10^{16}$ to $10^{19}\hbox{\kern 1.99997ptcm${}^{-2}$}$. The results of the calculations are presented in Fig. 4.

It comes out that the kinetic temperature $\rm T_{k}$ has to be higher than $100\hbox{\kern 1.99997ptK}$ to account for brightness temperatures of $60-80\hbox{\kern 1.99997ptK}$. Our calculations indicate a p-H₂O column density in the range $(0.3-2.0)\times 10^{17}\hbox{\kern 1.99997ptcm${}^{-2}$}$ for $\rm T_{k}\simeq 200\hbox{\kern 1.99997ptK}$. Assuming an ortho-to-para ratio of 3, we derive $\rm N(\hbox{H${}_{2}$O})=(1.2-8.0)\times 10^{17}\hbox{\kern 1.99997ptcm${}^{-2}$}$, and using the H₂ column density derived in the high-velocity clumps, we obtain the water abundance $\rm X(\hbox{H${}_{2}$O})=(0.5-3.2)\times 10^{-4}$. In this range of physical conditions, the line optical depth $\tau$ is small $\approx-1$ and shows evidence for the weak masing amplification of p-H₂O $3_{13}-2_{20}$ line. Suprathermal excitation of the line would require much higher H₂O column densities. This is possible if the emission arises from a small region of subarsec size (see e.g. Cernicharo et al. 1994) and is very sharp. The line $\rm T_{B}$ would then look weaker as a result of the dilution in the $2\hbox{\kern 1.99997ptkm\kern 1.99997pts${}^{-1}$}$ channels of the spectrometer.

Moro-Martin et al. (2001) analyzed the far-infrared emission lines of CO and H₂O in Cep E, as observed with ISO. Like Giannini et al. (2001), they found that the line emission can be accounted for by an extended component of low-density gas ($2\times 10^{4}\hbox{\kern 1.99997ptcm${}^{-3}$}$) at $\sim 1000\hbox{\kern 1.99997ptK}$. They show that the CO and H₂O lines could be reproduced equally well by ”warm” gas at a much lower temperature ($\sim 200\hbox{\kern 1.99997ptK}$) and higher density ($2\times 10^{6}\hbox{\kern 1.99997ptcm${}^{-3}$}$). The physical conditions we derive for the p-H₂O $3_{13}-2_{20}$ emitting region are actually in rather good agreement with this ”warm” solution and inconsistent with the low-density ”hot” solution.

Crimier et al. (2010) determined the physical conditions in the protostellar envelope of Cep E envelope and found $\rm n(\hbox{H${}_{2}$})=1.0\times 10^{7}\hbox{\kern 1.99997ptcm${}^{-3}$}$ and $T_{k}\sim 50\hbox{\kern 1.99997ptK}$, respectively, at the location of the water-emitting clumps, i.e. 1000 AU ($1.4\arcsec$) from the protostar. No p-H₂O $3_{13}-2_{20}$ emission is detected from the envelope at the $2\sigma$ level $=3\hbox{\kern 1.99997ptK}$ ($\rm T_{B}$). From Fig. 4, we conclude that $N(\hbox{H${}_{2}$O})<10^{16}\hbox{\kern 1.99997ptcm${}^{-2}$}$ in the inner protostellar envelope, and the water abundance is enhanced by 1 to 2 orders of magnitude in the high-velocity clumps. The authors estimate a radius of $\approx 220\leavevmode\nobreak\ AU(0.3\arcsec)$ for the hot corino region, and the density is predicted very high $\sim 10^{8}\hbox{\kern 1.99997ptcm${}^{-3}$}$; again, no masing effect is to be expected, in agreement with our observations.

The water-emitting clumps are located along the high-velocity jet, which suggests that they are closely related with the jet (Fig. 2). One possibility is that these clumps are tracing the jet/envelope interaction. However, the density of the protostellar envelope, as estimated by Crimier et al. (2010) at the location of the water emitting-clumps ($n(\hbox{H${}_{2}$})=1.0\times 10^{7}\hbox{\kern 1.99997ptcm${}^{-3}$}$), is about one order of magnitude higher than the density measured in the shocked gas ($10^{6}\hbox{\kern 1.99997ptcm${}^{-3}$}$), in apparent contradiction with this hypothesis. The structure of the envelope could be actually more complex than the simple power-law distribution derived by Crimier et al. (2010), as proposed that multiple protostars are embedded inside the core (Ladd & Hodapp, 1997). Direct observational evidence for several condensations is still missing, however. Alternatively, we speculate that the water emission could arise from internal shocks in the entrained material of the flow. Interferometric observations at subarcsec scale in the inner core ($r<1000$ AU) are needed to determine the gas density and velocity field in order to determine the location and the origin of the water-emitting shocks, as well as their relation with the jet.

Simple calculations in the LVG approximation predict that the H₂O emission from the high-velocity clumps should be easily detected in several ortho- and para-H₂O transitions with the high-resolution spectrometer HIFI onboard Herschel. Such observations will allow us to better constrain the shock physical conditions (see e.g. Flower & Pineau des Forets, 2010).