A&A 399, 1047-1061 (2003)
A. M. S. Boonman 1 - E. F. van Dishoeck 1 - F. Lahuis 1,2 - S. D. Doty 3 - C. M. Wright 4 - D. Rosenthal 5
1 - Sterrewacht Leiden, PO Box 9513, 2300 RA Leiden, The Netherlands
2 - SRON National Institute for Space Research, PO Box 800, 9700 AV Groningen, The Netherlands
3 - Department of Physics and Astronomy, Denison University, Granville, Ohio 43023, USA
4 - School of Physics, University College, ADFA, UNSW, Canberra ACT 2600, Australia
5 - Max-Planck-Institut für Extraterrestrische Physik, Giessenbachstrasse, 85741 Garching, Germany
Received 10 June 2002 / Accepted 22 November 2002
The infrared spectra toward Orion-IRc2, Peak 1 and Peak 2 in the 13.5-15.5 m wavelength range are presented, obtained with the Short Wavelength Spectrometer on board the Infrared Space Observatory. The spectra show absorption and emission features of the vibration-rotation bands of gas-phase CO2, HCN, and C2H2, respectively. Toward the deeply embedded massive young stellar object IRc2 all three bands appear in absorption, while toward the shocked region Peak 2 CO2, HCN, and C2H2 are seen in emission. Toward Peak 1 only CO2 has been detected in emission. Analysis of these bands shows that the absorption features toward IRc2 are characterized by excitation temperatures of 175-275 K, which can be explained by an origin in the shocked plateau gas. HCN and C2H2 are only seen in absorption in the direction of IRc2, whereas the CO2 absorption is probably more widespread. The CO2 emission toward Peak 1 and 2 is best explained with excitation by infrared radiation from dust mixed with the gas in the warm component of the shock. The similarity of the CO2 emission and absorption line shapes toward IRc2, Peak 1 and Peak 2 suggests that the CO2 is located in the warm component of the shock ( K) toward all three positions. The CO2 abundances of 10-8 for Peak 1 and 2, and of a few times 10-7 toward IRc2 can be explained by grain mantle evaporation and/or reformation in the gas-phase after destruction by the shock. The HCN and C2H2 emission detected toward Peak 2 is narrower (-150 K) and originates either in the warm component of the shock or in the extended ridge. In the case of an origin in the warm component of the shock, the low HCN and C2H2 abundances of 10-9 suggest that they are destroyed by the shock or have only been in the warm gas for a short time ( yr). In the case of an origin in the extended ridge, the inferred abundances are much higher and do not agree with predictions from current chemical models at low temperatures.
Key words: stars: formation - ISM: individual objects: Orion IRc2, Peak 1, Peak 2 - ISM: abundances - ISM: molecules - ISM: lines and bands - molecular processes
The Orion-IRc2/KL region ( pc) has traditionally been the prime source for studies in astrochemistry because of its extraordinarily rich spectra. Millimeter and submillimeter single-dish surveys show thousands of lines of nearly a hundred different molecules (e.g., Blake et al. 1987; Sutton et al. 1995; Schilke et al. 1997, 2001), whereas interferometer studies reveal intriguing chemical differentiation over scales of less than 2000 AU (e.g., Wright et al. 1996; Blake et al. 1996). In spite of this wealth of data, molecules such as CO2 and C2H2, which are symmetric and thus lack a dipole moment, cannot be observed through rotational transitions at millimeter wavelengths. Moreover, CO2 cannot be observed from Earth due to its high abundance in our atmosphere. Evans et al. (1991) have shown that important complimentary information can be obtained from vibration-rotation absorption lines toward bright mid-infrared sources. We present here spectra in the 13.5-15.5 m range toward three positions in the core of the Orion molecular cloud, taken with the Short Wavelength Spectrometer (SWS) on board the Infrared Space Observatory (ISO), which are unhindered by the Earth's atmosphere. Absorption and emission features of CO2, C2H2 and HCN are detected, which can be used to constrain the physical structure of this complex region and study the different chemistry of these molecules.
Millimeter studies have revealed a number of different physical and kinematic components in a 30'' region around the infrared source IRc2 (see Genzel & Stutzki 1989 for an overview). A clumpy hot core is located immediately adjacent to IRc2, with the clump exteriors currently being evaporated and/or ablated by the winds from the embedded massive young stellar object(s) (YSOs). This hot core is contained in a cavity, surrounded by a torus of dense, quiescent gas (the extended ridge) in the NE-SW direction. To the NW and SE, two shocked regions - called Peak 1 and Peak 2 - are revealed by bright H2 2 m emission, indicating the positions where the high-velocity plateau or outflow runs into the ambient molecular cloud. Peak 1 is located 25 NW and Peak SE of IRc2 (Beckwith et al. 1978). A cartoon of the core of the Orion molecular cloud indicating these different physical components is shown in Fig. 1.
The ISO-SWS beam ranges from to , so that these different regions can be separated spatially with the SWS. The 2.4-45.2 m ISO-SWS spectrum toward IRc2 has been presented by van Dishoeck et al. (1998) and shows many features including emission lines of ionized species, PAHs, H2, as well as absorption by interstellar ices and gas-phase species (see also Wright et al. 2000; González-Alfonso et al. 1998; Harwit et al. 1998). The full SWS spectrum toward Peak 1 has been presented by Rosenthal et al. (2000), whilst that toward Peak 2 is broadly similar (Wright 2000 and priv. comm. 2002). González-Alfonso et al. (1998) discuss the CO and H2O vibrational emission bands toward Peak 1 and 2. Because of the weaker continuum, the lines are more prominent at these positions than toward IRc2, especially the vibration-rotation and pure-rotational lines of H2.
In this paper, we focus on the ro-vibrational bands of gas-phase CO2, C2H2, and HCN along the lines of sight toward IRc2, Peak 1 and Peak 2. CO2 is predicted to be one of the more abundant carbon- and oxygen-bearing species and is detected ubiquitously in interstellar ices, with abundances of 15% with respect to H2O ice, or 10-5 - 10-6 with respect to H2(e.g. Gerakines et al. 1999). In contrast, the gas-phase CO2abundance is surprisingly low, 10-7, toward massive YSOs (van Dishoeck et al. 1996; van Dishoeck 1998; Dartois et al. 1998; Boonman et al. 2000). Since the abundances of many gas-phase species are enhanced toward IRc2, in particular those of species involved in the gas-grain chemistry (e.g., Blake et al. 1987; Charnley et al. 1992), it is important to investigate whether the CO2 chemistry follows this trend. Observations of C2H2 and HCN are interesting because they are both significant in the carbon- and nitrogen chemistry, and because their excitation provides information on the physical conditions (Lahuis & van Dishoeck 2000). For HCN, rotational transitions in the submillimeter and ro-vibrational transitions in the infrared can be observed. In a number of massive YSOs the HCN abundance derived from submillimeter observations is a factor of 100 lower than that derived from infrared observations, suggesting a jump in its abundance in high temperature regions (Lahuis & van Dishoeck 2000; van der Tak et al. 1999, 2000; Boonman et al. 2001). There is still considerable debate whether such abundance jumps are mainly due to evaporation of ices, to quiescent high-temperature chemistry at a few hundred K or to shock chemistry at a few thousand K. The comparison of the Orion IRc2 and the shocked Peak 1 and Peak 2 results can provide constraints on the different models.
In Sect. 2, the ISO-SWS data reduction methods are discussed. Section 3 will present models for the HCN, C2H2, and CO2absorption toward Orion IRc2. The inferred abundances are compared with those found toward other sources. In Sect. 4, the observations toward the shock positions Peak 1 and 2 are presented, and the excitation of the molecules is analyzed. Section 5 will compare the results for the three different positions and the conclusions are presented in Sect. 6.
|Figure 1: Cartoon of the core of the Orion molecular cloud. The figure represents a cross section in the plane of the sky. The size and orientation of the ISO-SWS beam around 15 m at the approximate positions of IRc2, Peak 1 and Peak 2 are indicated by the rectangular boxes (adapted from van Dishoeck et al. 1998).|
|Open with DEXTER|
The spectra toward Orion IRc2, Peak 1 and Peak 2 from 12.0-16.5 m were made with the ISO-SWS grating. IRc2 was observed on September 6 1997 (revolution 660) using the SWS06 observing mode centered at (1950 , (1950 , which is about 1 S and 3 W of the IRc2 position listed by Gezari (1992). The Peak 1 spectrum was taken at the position (1950 , (1950 in the SWS01 speed 4 observing mode, on October 3, 1997 (revolution 687). Finally, on February 25 1998 (revolution 833) the spectrum toward Peak 2 was taken using the SWS01 speed 4 observing mode at the position (1950 , (1950 , 1 E of the position listed by Beckwith et al. (1978). The beam size in this wavelength range is 14 , and was oriented 7 and 11 in NW-SE direction for IRc2 and Peak 2 respectively. These two beams do not overlap; however, the infrared source BN falls within the IRc2 beam (see sketch of region in Fig. 1). The Peak 1 beam, which has the same size and is orientated 6 in NE-SW direction, partially overlaps with the IRc2 beam, but does not contain the BN object at the observed wavelength range.
Data reduction was done within the ISO-SWS Interactive Analysis System SIA using the ISO Off-line Processing (OLP version 10) software modules and calibration files (see Lahuis et al. 1998 and Wieprecht et al. 2001 for a description of the SIA system and its relation to the ISO OLP system). For the SWS01 observations the Standard Processed Data (SPD) were re-derived to create spectra at full grating resolution with some loss in signal to noise. This software has been developed at the Dutch ISO Data Analysis Centre (DIDAC) and is based on the OLP software and calibration. It will become available within SIA and OSIA for general use. Instrumental fringes have been minimized when applying the Instrumental Spectral Response Function (RSRF) by RSRF matching to allow for offsets in the wavelength calibration and differences in resolution between the data and the RSRF. The remaining fringe residuals after the RSRF calibration were removed using a robust iterative sine fitting method based on an approximated Fabry-Pérot model (see Lahuis & van Dishoeck 2000 and Kester et al. 2001). All spectra have been rebinned to a spectral resolving power of 3500, twice the instrumental resolution. The final spectra have a typical S/N ratio on the continuum of 50-100.
|Figure 2: Comparison of the 13.5-14.3 m spectra toward the massive young stellar objects Orion-IRc2 and AFGL 2591. The HCN and C2H2 Q-branches toward Orion-IRc2 are much narrower than those toward AFGL 2591, and the absence of the hot bands at 13.7 and 14.0 m indicates that less warm gas is probed ( K). The AFGL 2591 spectrum is adapted from Lahuis & van Dishoeck (2000).|
|Open with DEXTER|
The ISO-SWS spectrum of the ro-vibrational band of gas-phase C2H2 at 13.71 m and the ro-vibrational band of gas-phase HCN at 14.05 m toward IRc2 are presented in Fig. 2. The strong Q-branch features of both species are clearly detected in absorption. They are much narrower than the HCN and C2H2 Q-branch features detected toward other massive protostars, such as AFGL 2591 and AFGL 2136 (Lahuis & van Dishoeck 2000), indicating lower excitation temperatures (see Fig. 2). The HCN and C2H2 hot bands at 14.00 m and 14.30 m for HCN, and 13.72 m and 13.89 m for C2H2, seen toward the latter sources are not detected toward IRc2, implying that the excitation temperature of the observed HCN and C2H2 gas is <400 K.
No R- or P-branch lines have been detected for HCN and C2H2. Some R-branch lines at 13.5 m of HCN and C2H2 were previously observed from the ground by Evans et al. (1991) at higher resolving power, . Their observations were made with a 3 beam, which is much smaller than the ISO-SWS beam, allowing IRc2 and BN to be observed separately. A disadvantage of these ground-based data is the need to correct for atmospheric transmission. Since no HCN and C2H2absorption was detected toward BN, Evans et al. (1991) divided their IRc2 spectra by the BN spectra to remove the telluric features. The much stronger Q-branch features seen in the ISO-SWS spectra are difficult to observe from Earth due to telluric interference.
Based on the Evans et al. (1991) data, it is likely that the absorption occurs only toward the area directly surrounding IRc2. This area includes the infrared sources IRc2, IRc7, and IRc4, which are the dominant sources between 12 and 20 m (Gezari et al. 1992). However, Evans et al. (1991) show that most of the absorbing C2H2 gas lies in front of IRc2 and IRc7 and that the HCN and C2H2 column densities toward IRc2 are a factor of 2 higher than toward IRc7. Therefore, we assume in our analysis that the absorption occurs only toward IRc2, although some contribution from the line of sight toward IRc7 may be present. The fraction of the continuum flux due to BN at 13.5 m in the ISO-SWS beam has been estimated to be % from mid-infrared maps by Downes et al. (1981) and Wynn-Williams et al. (1984) and is corrected for the beam profile by comparing our ISO-SWS spectrum centered at IRc2 with that centered at BN (Cernicharo et al., unpublished results). The BN continuum flux has been subtracted from the total flux in the ISO-SWS beam, resulting in a spectrum in which the absorption lines are superposed on the IRc2 continuum only. This spectrum has then been divided by the IRc2 continuum to get the relative absorption spectrum. The percentage of BN continuum has been varied from 55% to 68% and the corresponding values have been included as error bars in Table 1.
The normalized spectra have been analyzed using synthetic absorption spectra as described by Lahuis & van Dishoeck (2000). These model fits depend only on the excitation temperature , the total column density of the molecule N and the Doppler broadening parameter b. Because of the high densities in Orion, the C2H2and HCN excitation is likely to be close to thermal, and can be described by a single excitation temperature. Recently, new high-resolution ground-based observations of IRc2 around 13.3 m with the Texas Echelon-Cross-Echelle Spectrograph (TEXES) have been performed at a spectral resolving power of / (Lacy et al. 2002). In these spectra the individual R-branch lines of C2H2 and HCN are resolved and have a range of Doppler b parameters of 3-10 km s-1 (Lacy, priv. comm), compared to 3 km s-1 found by Carr et al. (1995). Therefore different b-values between 3 and 10 km s-1 have been explored in the models. Because of the low spectral resolution of the ISO-SWS, the absorption seen in our spectra is expected to be dominated by the components with the larger line widths.
The best fit model for C2H2 has an excitation temperature of K and a column density of cm-2, whereas that for HCN has K and cm-2. These values are listed in Table 1 along with the error bars obtained from fits. The C2H2 excitation temperature corresponds well with values derived by Carr et al. (1995), who re-analyzed the data from Evans et al. (1991). The HCN excitation temperature corresponds well to the K derived for the hot core (Blake et al. 1987; Wright et al. 1996), but not with that of 132 K listed in Carr et al. (1995). Based on kinematic grounds combined with their excitation temperatures, Evans et al. (1991) concluded that the HCN and C2H2 absorption probably originates in the plateau gas. However the HCN excitation temperature found here is higher than their value. This may be partly due to the fact that the lower spectral resolution ISO-SWS data are less sensitive to the colder, less turbulent gas, with small line widths. In addition, a high-resolution map of H13CN toward IRc2 by Wright et al. (1996) reveals the presence of several clumps with temperatures 150 K. These warm clumps fall outside the 3 beam used by Evans et al. (1991), but within the ISO-SWS beam. If these clumps are in front of an infrared source, this probably explains the higher HCN excitation temperature derived from the ISO-SWS spectra. Higher-resolution spectra such as those by Lacy et al. (2002, in prep.) can help to disentangle the different components present along the line of sight.
It should be noted that a significant fraction of cold HCN and C2H2 gas could still be hidden in the low-resolution ISO-SWS spectra, which is picked up in higher-resolution ground-based spectra (Evans et al. 1991; Lacy et al. 2002, in prep.). This could explain why both the HCN and C2H2 column densities are a factor of 4-5 lower than the values N(HCN cm-2 and N(C2H cm-2 previously derived by Carr et al. (1995).
The models used so far assume pure absorption and do not include possible emission in the line itself. In order to examine its effect on the derived column density, an excitation model has been set-up including levels up to J=21 in the ground and vibrational states for HCN and up to J=24 for C2H2. A blackbody with K has been adopted as background source, corresponding to the temperature found by Gezari et al. (1998) for IRc2. Excitation temperatures ranging from 125 to 175 K for C2H2 and 220 to 330 K for HCN have been explored, corresponding to the values listed in Table 1, assuming that the populations in both the and state follow a Boltzmann distribution at these temperatures. For C2H2 the derived column densities are 25% higher if emission in the line is included, which is not significant. Because of the higher excitation temperatures, this effect does play a role for HCN, where it becomes important whenever is close to the temperature of the background radiation, which is the case for K. The derived column densities can be a factor of up to 8 higher than derived from the pure absorption models for b=3-10 km s-1. This is still consistent with the non-detection of H13CN which gives an upper limit of N(H13CN cm-2.
|Figure 3: a) Normalized spectrum of HCN and C2H2 toward IRc2 with 62% BN continuum subtracted (see text). The best fitting model for b=5 km s-1 (see Table 1) shifted upward by 0.2, is shown for comparison. b) Normalized spectrum of CO2 toward IRc2 with no continuum subtracted, together with the best fitting model spectrum for b=5 km s-1 shifted upward by 0.2 (Table 1).|
|Open with DEXTER|
As shown in Fig. 3, the gas-phase CO2 -0 band at 15.0 m is clearly detected in absorption toward IRc2. The individual R- and P-branch lines are also detected in absorption. The CO2 -1 ro-vibrational band has not been detected, nor any lines of 13CO2. In this case, both results with and without correction for the continuum of BN have been determined, since there are no prior data which indicate that the CO2 absorption occurs only toward IRc2. Modeling of the CO2 band toward IRc2 without subtraction of the BN continuum results in K and cm-2, assuming pure absorption, thermal excitation and a Doppler parameter of b=3-10 km s-1. Table 1 shows that these results are not sensitive to the adopted line width. When 62% of the BN continuum is subtracted, like in the case for HCN and C2H2, the ro-vibrational lines start to become optically thick for b=3 km s-1, resulting in a much larger column density at this line width. The excitation temperature in both cases lies in between those of HCN and C2H2, suggesting it might also originate in the plateau gas.
The effect of emission in the line has also been tested for CO2, using an excitation model similar to that for HCN and C2H2, including levels up to J=40 in both the and 1 vibrational state. For excitation temperatures equal to those in Table 1 emission in the line itself results in column densities that are 20% higher, which is within the error bars listed and therefore is not significant. This effect only becomes important for CO2 for K.
Evans et al. (1991) and Wynn-Williams et al. (1984) argue from NH3 maps that the line of sight toward IRc2 passes through the edge of the hot core, so that at most a small part of the hot core can be probed in absorption. In combination with the derived excitation temperatures, this suggests an origin in the plateau gas for all three molecules. Therefore the abundances are computed using 0.8-3 cm-2 corresponding to the line of sight column density found by Evans et al. (1991) from the depth of the silicate feature. Since the optical depth of the continuum between 13 and 15 m rapidly becomes very high for larger H2 column densities, these values are appropiate for (colder) gas probed in absorption in front of the location.
It should be noted that the HCN and C2H2 abundances are derived from pure absorption models, without including dust mixed with the gas. Given the complex nature of the IRc2/BN complex, determining a realistic physical structure to be used in radiative transfer models is not trivial. At present, the accuracy of the derived HCN and C2H2 abundances is best determined by comparison to observations at high spectral and spatial resolution, e.g. with TEXES (Lacy et al., in prep.).
The inferred HCN and C2H2 abundances are a few times 10-7 (Table 2). These are of the same order of magnitude as the values found for other massive protostars (Lahuis & van Dishoeck 2000). They also agree with those found by Blake et al. (1987) and Schilke et al. (2001) for HCN in the hot core and plateau gas. Stutzki et al. (1988) find similar abundances for HCN in the plateau gas from the J=9-8 transition. The abundances are listed in Table 2.
Wright et al. (1996) mapped the HCN emission in the IRc2 region at 6'' resolution using the BIMA interferometer. It shows no significant HCN emission at the position of BN, consistent with the absence of HCN absorption toward BN. They divide the plateau gas near IRc2 in three different chemical regions, an SiO-rich zone, an expanding shell rich in SO and SO2 emission and a zone with high-velocity HCN emission extending over more than 50 km s-1, which they attribute to evaporation of ices and ablation from pre-existing dense clumps. This could be a possible explanation for the high HCN abundance found from the absorption lines, which is an order of magnitude higher than found in the more extended Orion ridge (Bergin et al. 1997). The HCN and C2H2 abundances also agree well with predictions from hot core models by Rodgers & Charnley (2001). Abundances taken at a single temperature/density point from envelope models following Doty et al. (2002) are in agreement with the low end of the observed range (see Table 2 and Sect. 5.1 for details). Thus, the observed HCN and C2H2 probably originate in gas that has been blown away from the hot core clumps near IRc2 and now resides in the plateau gas in the swept-out cavities between the hot core clumps (see Fig. 1). This is in agreement with Carr et al. (1995), who ascribe their observed ro-vibrational HCN and C2H2 absorption lines to the plateau gas.
Based on the fact that CO2 absorbs in the same wavelength region as HCN and C2H2 and has the same excitation temperature, it is likely that the observed CO2 does not originate in the hot core itself, although it still can contain large amounts of hidden CO2. Since no previous CO2 observations with smaller beam sizes exist, it is difficult to determine whether the CO2 is confined to a region close to the hot core like HCN and C2H2 or whether it is more widespread. The gas-phase CO2 abundance, assuming it is seen both in the direction of IRc2 and BN, is of the same order as those of HCN and C2H2 (Table 2). It is somewhat higher than the values found for some other massive YSOs (van Dishoeck et al. 1996; van Dishoeck 1998; Boonman et al. 2000). Assuming CO2 is observed only in the direction of IRc2, like HCN and C2H2, the derived abundances are (0.3-3 for b=10 km s-1 and 10-5 for b=3 km s-1. This is of the same order of magnitude as the CO2 ice abundance of N(CO2)/N(H toward IRc2 and agrees also with the CO2 ice abundances found toward other massive protostars (Gerakines et al. 1999). The same is true if CO2 is seen both in the direction of IRc2 and BN. This suggests that most of the CO2 ice in the neighborhood of IRc2/BN must have evaporated off the grains. In both cases the CO2 abundances also show reasonable agreement with both the models from Charnley (1997) and Doty et al. (2002) listed in Table 2. This indicates that the observed gas-phase CO2 abundances are explained best by a combination of pure gas-phase chemistry and grain-mantle evaporation.
Contrary to what is observed for IRc2, the CO2 -0 band is seen in emission toward both Peak 1 and 2 (Fig. 4). HCN and C2H2 are also seen in emission but only toward Peak 2, although the non-detection toward Peak 1 could be due to the higher noise level there. The shape of the CO2 Q-branch toward Peak 1 and Peak 2 is very similar to that seen in absorption toward IRc2 (Fig. 5). In order to investigate whether the similarities and differences between both positions and IRc2 are true physical and/or chemical differences or whether they are due to geometrical and radiative transfer effects, we follow the same analysis as in González-Alfonso et al. (2002; hereafter GA02). They discuss the CO and H2O ro-vibrational bands toward the same three positions and find the same striking similarities between the Peak 1 and Peak 2 spectra as is found here for CO2. GA02 also investigate in detail the excitation mechanisms and show that radiative pumping, rather than collisional excitation, likely dominates. The comparison with CO may allow accurate relative abundances in the shocked gas to be determined. The resulting CO2/CO ratio can then be compared with predictions from chemical models.
|Figure 4: Continuum subtracted spectra of CO2, C2H2 and HCN toward a) Orion-IRc2, b) Peak 1, and c) Peak 2. d) Blow-up of the 13.6-14.15 m region toward Peak 2.|
|Open with DEXTER|
|Figure 5: Comparison of the CO2 band shape toward IRc2, Peak 1 and Peak 2. The CO2 absorption toward IRc2 has been inverted to an emission spectrum. All continuum subtracted spectra have been normalized with respect to their own peak flux. The Peak 2 spectrum is shifted by -0.6 and the Peak 1 spectrum by -1.2.|
|Open with DEXTER|
A radiative transfer model has been constructed to investigate optical depth effects of the emission toward Peak 1 and Peak 2. In this model the source is represented as a slab of thickness d, divided in a number of smaller sheets, each with the same constant H2 density and gas temperature. For CO2, levels up to J=40 in the ground and vibrational states have been included. For HCN and C2H2, levels up to J=21 and J=24 respectively, are included in both vibrational states. Once the level populations are known, the radiative transfer can be calculated, including emission and absorption by both lines and continuum. In the next paragraphs, three different scenarios will be discussed. Scenario 1 considers radiative pumping of the molecules by the IRc2/BN complex. Scenario 2 assumes that the molecules are excited by collisions only, which would be the case if the shock itself is the dominant excitation mechanism. Finally, scenario 3 discusses the possibility of excitation by infrared radiation from dust mixed with warm gas. Following the discussion in GA02, first the excitation rates in the case of scenario 1 and 2 are compared, in order to determine the relative importance of radiative versus collisional excitation.
|Figure 6: Schematic representation of the scenarios discussed in Sect. 4.|
|Open with DEXTER|
If it is assumed that the CO2 gas toward Peak 1 and
Peak 2 is excited predominantly by radiation from a source of temperature
and effective radius
located in the
IRc2/BN complex and subsequently emits
observer, the radiative excitation rate from the
state to the (,J1) level (in cm-3 s-1) is
given by (cf. Eq. (6) in GA02)
For the populations of the levels in the v=0 state, a Boltzmann distribution has been adopted at T=300 K, the best-fit temperature for the warm component toward Peak 1 and Peak 2 as determined from the CO ro-vibrational band at 4.7 m by GA02. Using the CO2 excitation temperature found for IRc2 (Table 1) does not change the results significantly.
For Peak 1 it is assumed that BN is the dominant excitation source and that the extinction along the line of sight from BN to Peak 1 at 15.0 m is zero. For , a value of 500 K is adopted consistent with the derived color temperature for BN from Gezari et al. (1992). The radius of the stellar source has been adjusted so that the flux at 15 m at a distance of 450 pc is 1300 Jy, corresponding to the interpolation of the dereddened 12 m and 20 m fluxes listed by Gezari et al. (1992). The exact choice of and is not important as long as the emerging flux at 450 pc is fixed to 1300 Jy. A distance between the CO2 molecules and BN of cm is adopted, corresponding to the angular distance of 15 between Peak 1 and BN. Other infrared sources in the Orion BN/KL region might also contribute to the radiative pumping of Peak 1. Since these sources are less luminous than BN at this wavelength and located farther from Peak 1 their contribution will change the total radiative pumping rate from the to the state by at most a factor of 2.
Although BN is the dominant infrared source in the 13.5-15.5 m range (Gezari et al. 1992), for Peak 2 IRc2 is much closer. Therefore it is assumed that the exciting source for Peak 2 is located at the position of IRc2 at cm. The dereddened flux of IRc2 at 15 m is 650 Jy, but in order to account for a possible contribution of other sources close to IRc2 with comparable infrared fluxes, a flux of twice this value is used, corresponding to a BN type source at the position of IRc2 (Gezari et al. 1992). Therefore the same values for and used for Peak 1 are used for Peak 2. Assuming instead a flux of 3400 Jy, equal to the observed ISO-SWS flux toward IRc2/BN at a distance in between BN and IRc2, does not change the total pumping rate by more than a factor of 2.
Alternatively, if collisions excite the state of CO2 the excitation rate (in s-1) is given by the product n(X) , where n(X) is the density of the collision partner X and k0-1 the collisional excitation rate coefficient from the to the state. A discussion on the collisional rate coefficients for CO2 is given in Appendix A. Based on that discussion only H2 will be considered as a collision partner.
Using fractional populations in Eq. (1) and adopting a hydrogen density of n(H cm-3, corresponding to the density in the warm component of the shock (GA02), the radiative and collisonal pumping rates (in s-1) can be compared directly. Figure 7 shows the ratio of these rates for CO2 in Peak 1 and Peak 2, but the curves can easily be scaled to other densities. It can be seen that for kinetic temperatures -300 K radiative pumping is comparable to collisional excitation, for both Peak 1 and Peak 2. However, the line of sight between the exciting source and Peak 1/2 may not be in the plane of the sky, thus underestimating the actual distance. In that case the radiative pumping rate used is too high. Also, the extinction along the line connecting the exciting source and Peak 1/2 is probably small, but not zero, again indicating that the radiative pumping rates used are too high. Together this suggests that, for K radiative excitation by infrared sources in the IRc2/BN complex is not the dominant excitation mechanism for CO2, contrary to what is the case for CO (GA02). As will be shown below, radiative pumping by warm dust mixed with the warm gas at the Peak 1 and 2 positions may be still significant. If the CO2 is located in the hot component at K, collisions become more important, because of the higher kinetic temperature of the gas.
The HCN-H2 and C2H2-H2 collisional rates are not known. Because the vibrational energy and reduced mass of HCN and C2H2 are close to those of CO2, their collisional rates will be approximately the same as those for CO2. Adopting the same collisional de-excitation rates as for CO2 the ratio of the radiative and collisional rates are calculated for HCN and C2H2 in the same way as for CO2. Figure 7 shows that also for HCN and C2H2 the radiative and collisional excitation rates are comparable for T=150-300 K, although for C2H2 the ratio is higher than for CO2 and HCN, due to its larger Einstein-A coefficient.
|Figure 7: Estimated ratio of the radiative to collisional excitation rates for CO2, HCN and C2H2 in Peak 1 and Peak 2 as functions of the kinetic temperature T. The radiative excitation rates per molecule are computed from Eq. (1), assuming zero extinction and summing over all rotational levels. Only H2 is considered as collision partner with an adopted density of cm-3, corresponding to the density of the warm component of the shock (GA02).|
|Open with DEXTER|
Using Eq. (1) and adopting the same rotational excitation temperature for
state as in Sect. 4.1, the density of the (,
level n1,J1 can be calculated from
Using Eqs. (1) and (2), the CO2 vibrational excitation temperature for Peak 1 varies from 65-110 K within the ISO-SWS beam and for Peak 2 from 63-80 K (Figs. 8 and 9). Assuming a constant vibrational temperature, corresponding to a constant radiative pumping rate within the ISO-SWS beam, together with a constant CO2 abundance and H2 density and a constant rotational excitation temperature in the state, the level populations are known and thus a synthetic spectrum can be calculated. It should be noted that in this case the level populations are not calculated self-consistently with the radiative transfer. An H2 density of cm-3 has been adopted, whereas the thickness of the slab has been chosen such that the integrated column density across the slab is N(H cm-2 for Peak 1 and N(H cm-2 for Peak 2, corresponding to the values for the warm component of the shock listed by Rosenthal et al. (2000) and Wright (2000), respectively.
The CO2 abundance and the rotational excitation temperature in the state have then been varied to find a good match to the observed CO2 emission. These calculations have been done for different values of within the SWS beam. The results are shown in Figs. 8 and 9.
For Peak 1 it is found that the Q-branch is well fit with a rotational temperature of 150-225 K, as long as the lines are not too optically thick (i.e. ). For line widths in the range of b=3-10 km s-1 this is the case for K, corresponding to cm. Lower vibrational temperatures require lower rotational temperatures to fit the CO2 Q-branch, but then huge CO2 column densities cm-2 are needed to match the observed flux and the P- and R-branches then exceed the 3 upper limit derived from the SWS spectra.
Since Peak 2 is located further from the IRc2/BN complex the vibrational temperatures within the ISO-SWS beam due to radiative excitation are somewhat lower. Following the same approach as for Peak 1, the CO2 Q-branch toward Peak 2 is fit best with a rotational temperature of K for , which is the case for K and b=3-10 km s-1. This is outside the observed range of within the ISO-SWS beam (Fig. 9). This figure also shows that the CO2 emission toward Peak 2 cannot be explained by radiative excitation by the IRc2/BN complex, unless very large column densities are involved. The rotational temperature of K is in excellent agreement with the temperature derived by Wright (2000) for the H2 lines in the warm component of the shock. This suggests that CO2 and H2 are co-located.
The same analysis has been performed for CO in the warm component of the shock allowing the determination of the CO2/CO ratio for comparison with chemical models. Here, the CO vibrational excitation temperatures from GA02 are used, as a function of the distance from the exciting source. The CO column density found from these models at the position of Peak 1 ( cm) is consistent with that found by GA02. Comparison of the CO2 column densities for Peak 1 with those of CO gives a CO2/CO ratio of 0.3 as long as the optical depth of the CO2 lines is comparable to that of the CO lines (i.e. for ). These conditions are met for cm, i.e. between BN and Peak 1. Since for Peak 2 the CO2 emission cannot be fit well for K due to high optical depths, no CO2/CO can be derived within the SWS beam in the case of radiative excitation by the IRc2/BN complex. For K, which is outside the SWS beam, a ratio of CO2/ is found (Table 3).
In order to investigate optical depth effects in the case of collisional excitation of CO2, the same radiative transfer model has been used, but the level populations in both vibrational states are calculated with the Accelerated Monte Carlo method by Hogerheijde & van der Tak (2000), using the collisional excitation rates from Fig. A.1 for the same slab model as described before and including no infrared pumping. Kinetic temperatures of T=150-200 K for Peak 1 and T=150 K for Peak 2 have been used, corresponding both to the rotational temperatures derived from the CO2 emission and the temperature of warm H2 gas as derived by Wright (2000) for Peak 2, and estimated for Peak 1 by Rosenthal et al. (2000). The resulting level populations indicate a vibrational excitation temperature of 78 K at T=150 K and 87 K at T=200 K. The CO2 abundances with respect to H2, n(CO2)/n(H2), found for collisional excitation are shown in Table 3. The optical depth in this case is 1. These abundances are lower than in the case of radiative excitation by the IRc2/BN complex at the same vibrational temperature.
A third, preferred possibility is that infrared radiation from warm dust mixed with the shocked gas can excite the molecules. In order to investigate this effect, the level populations are calculated using the same method as in the case of collisional excitation, but now including infrared pumping by dust, using grain opacities from Ossenkopf & Henning (1994) and assuming the dust temperature is close to the kinetic temperature. For Peak 2 also models with a dust temperature of 100 K have been investigated, corresponding to the color temperature around 15 m toward Peak 2. For Peak 1 the color temperature is 150 K. The resulting abundances with respect to H2 are shown in Table 3. These abundances are significantly lower than those for collisional excitation only, indicating that if dust is mixed with the gas, the CO2 will be radiatively excited. Since the color temperatures of the dust around 15 m towards both Peak 1 and 2 are comparable to the kinetic temperature of the warm H2 gas and the rotational temperature of the CO2 emission, it is likely that dust is mixed with the gas in the warm component of the shock. The pumping rates by this warm dust are much higher than those by the IRc2/BN complex resulting in much lower abundances.
|Figure 8: Peak 1 vibrational excitation temperature (left axis) versus distance from BN for the physical model described in Sect. 4.2. The column densities of the best fitting models for CO2 (right axis) are shown by the dots, assuming no background radiation, b=10 km s-1 and a constant vibrational excitation temperature in the SWS beam as a function of radius. The x-coordinates of the dots refer to the radii at which the value of listed between parentheses occur. The optical depth of the most optically thick CO2 ro-vibrational transition of the band is also listed. The position of the SWS beam w.r.t. BN is indicated.|
|Open with DEXTER|
|Figure 9: Same as Fig. 8 but now for Peak 2. The source that is radiatively exciting the CO2 molecules is the same as described in Sect. 4.2.|
|Open with DEXTER|
The same three scenarios discussed for CO2 have been investigated for the HCN and C2H2 emission detected toward Peak 2. In the case of radiative excitation by the IRc2/BN complex (scenario 1), it is found that the vibrational excitation temperature within the SWS beam ranges from 67 K to 83 K for HCN. For the width of the HCN Q-branch is well-fit with a rotational temperature of 50-125 K for b=3-10 km s-1. This is the case for K. For lower the rotational temperature decreases to 20 K and large column densities ( cm-2) are required to match the Q-branch, since the optical depth increases rapidly. Such high column densities are not observed in the extended ridge or plateau gas, which match the derived rotational temperature. Therefore it is likely that the HCN emission is not very optically thick. Since no HCN has been detected toward Peak 1, only upper limits have been determined. Comparison of the HCN column densities with those of CO in the case of scenario 1 results in an HCN/CO ratio of 0.01-0.1 for Peak 2 and 0.1 for Peak 1 (Table 3). For C2H2 the vibrational temperature within the SWS beam ranges from 73 K to 90 K. The Q-branch is matched best with a rotational temperature of 50-175 K for b=3-10 km s-1. As for HCN, this is the case for K and the same conclusions are found. The results are listed in Table 3.
Scenarios 2 and 3, considering collisional excitation in the warm component of the shock and excitation by dust mixed with the warm gas respectively, have also been investigated. The results are included in Table 3.
Since both the HCN and C2H2 Q-branches toward Peak 2 are much weaker than that of CO2, their rotational temperature is less well-constrained. The derived temperature range of 50-175 K indicates that an origin in the colder extended ridge is also possible. Assuming collisional excitation at K, corresponding to the temperature of the extended ridge, the observed emission for both HCN and C2H2 could not be reproduced. Similarly, including dust at 50 K could also not reproduce the observed emission for both molecules.
As a final possibility, the slab model for the warm component of the shock has been used, with the assumption that no HCN and C2H2 is present. Then, a slab of cold gas at T=50 K has been placed in front of it, such that the total column density across the whole slab is N(H cm-2 corresponding to the value derived by Sutton et al. (1995) for the extended ridge. In this case the warm dust pumps the cold HCN and C2H2, resulting in abundances of a few times 10-7 for both HCN and C2H2. These are similar to those derived in the case of collisional excitation in the warm component of the shock. They are also similar to those in the case of radiative excitation by the IRc2/BN complex, using the same H2 column density. But they are much higher than in the case of radiative excitation by dust mixed with gas in the warm component of the shock (Table 3).
Thus, if HCN and C2H2 originate in the warm component of the shock, the most likely excitation mechanism is by infrared radiation from dust mixed with the warm gas (scenario 3), since the pumping rates in this case are much higher than in the other cases. If they originate in the extended ridge they are probably excited by radiation from the IRc2/BN complex and/or dust of the warm component of the shock lying behind the extended ridge.
The previous sections show that the CO2 emission toward Peak 1 and Peak 2 does not arise in the hot component of the shock at K, but probably arises in gas of -200 K, consistent with conditions in the warm component of the shock. Three different scenarios have been calculated, with the inferred CO2 abundances differing by more than four orders of magnitude (Table 3).
Comparing the derived CO2/CO ratios for Peak 1 and Peak 2 of 0.3 in the case of radiative excitation by the IRc2/BN complex (scenario 1) with the shock models by Charnley & Kaufman (2000) shows that this ratio is much higher than their post-shock value of CO2/ for a shock that destroys the CO2 without reformation in the gas-phase (i.e. km s-1, cm-3). These results are listed in Table 3. If the CO2 is totally destroyed in either a shock or a flaring event then the CO2 will be reformed through gas-phase reactions increasing to abundances CO2/CO of 10-2 at yr for T=200 K. (Charnley & Kaufman 2000; Doty et al. 2002). This is still at least one order of magnitude lower than the CO2/CO ratio found above. Additionally, the time since the passage of the shock is estimated to be much shorter than 105 yr, about 103-104 yr (Genzel & Stutzki 1989; Wilson et al. 1986). This shows that the abundances found in the case of radiative excitation by the IRc2/BN complex cannot be explained by reformation in the gas-phase after destruction of the CO2 in the shock. In fact, such high CO2 abundances have not been reported in any other star-forming regions, making it unlikely that the IRc2/BN complex is the exciting source.
In the preferred case of scenario 3 where the CO2 is radiatively excited by warm dust mixed with the gas, the derived abundances are in agreement with predictions for quiescent warm gas at yr (Doty et al. 2002). Here the Doty et al. (2002) models are used as generic high-temperature models. They are also in good agreement with predictions from Charnley & Kaufman (2000) for a shock that destroys the CO2, with little or no reformation through gas-phase reactions on time scales of yr. In the case of collisional excitation of CO2 in the warm component of the shock (scenario 2), the abundances with respect to H2 are in agreement with the predictions from the chemical models by Doty et al. (2002) at yr. But, as argued in Sect. 4.3.3, it is likely that dust is present in the warm component of the shock, making scenario 3 the most likely excitation mechanism for CO2.
Contrary to CO2, the origin of the HCN and C2H2 gas is more difficult to establish. In the case of radiative excitation by the IRc2/BN complex (scenario 1) the HCN/CO and C2H2/CO ratios are very high compared to predictions from chemical models of HCN/ and C2H2/ for T<200 K (Doty et al. 2002). On the other hand, comparison with H2, using N(H cm-2 from Sutton et al. (1995) for the extended ridge, abundances of 10-7 for HCN and C2H2 are found. Similar abundances are obtained when it is assumed that dust in the warm component of the shock excites the molecules in the much cooler extended ridge. These are higher than predicted by gas-phase models at K, which predict abundances of at late times. For K HCN and C2H2 abundances of 10-7 are easily produced by gas-phase models (Rodgers & Charnley 2001; Doty et al. 2002). The HCN abundance is also higher than the values of 10-8 derived from submillimeter observations by Blake et al. (1987) and Schilke et al. (1992) for the extended ridge. Together this makes both scenarios less likely excitation mechanisms. Moreover, as noted in Sect. 4.3.3 and above, warm dust is likely mixed with the gas.
The HCN and C2H2 abundances found in the case of the preferred scenario 3 are in agreement with predictions from gas-phase models for T<200 K and yr (Doty et al. 2002) (Table 3). The HCN abundance is also in good agreement with the predicted abundances from chemical models for the Orion shock of a few times 10-9 by Schilke et al. (1992) which also explains their value derived from HCN J=1-0 observations in the direction of Peak 1. This may indicate that HCN and C2H2 are destroyed by the shock. On the other hand, Blake et al. (1987) find HCN abundances of for the extended ridge, similar to what is found when HCN and C2H2 originate in the warm component of the shock. Therefore, it is also possible that HCN and C2H2 have been heated to temperatures K only recently ( yr), so that enhanced gas-phase formation has not yet taken place.
In summary, scenario 3 seems to be the preferred scenario for all three molecules, but in the case of HCN and C2H2 an origin in the extended ridge, based on the derived excitation temperatures, cannot be completely excluded. Mapping of the 13-15 m continuum at high angular resolution could provide further evidence for the presence of warm dust toward Peak 1 and 2, and thus strengthen the case for scenario 3. In addition, infrared observations of HCN and C2H2 toward Peak 1 and 2 at high spectral resolution, e.g. with TEXES (see Sect. 3.1), will be useful to discriminate between an origin in the extended ridge or the warm component of the shock.
HCN abundances as high as 10-6 as found toward IRc2 (Table 2) cannot be explained adequately by steady-state low-temperature gas-phase chemical models (Evans et al. 1991). As argued by van der Tak et al. (1999), not only grain-mantle evaporation but also high-temperature gas-phase chemistry plays a role in producing high HCN abundances in hot cores. The observed abundances toward IRc2 are consistent with predicted HCN abundances of up to 10-6 from pure gas-phase chemistry at T>200 K (Doty et al. 2002; Rodgers & Charnley 2001).
The observed C2H2 abundance of 10-7 toward IRc2 is higher than the predicted abundances of 10-8 for pure gas-phase chemistry at K by Doty et al. (2002), but their formation route for C2H2 is through reactions of water with C2H3+ instead of dissociative recombination. Pure gas-phase models by Rodgers & Charnley (2001) do predict C2H2 abundances up to 10-7 for T=100-300 K. The observed C2H2 abundance is also consistent with the upper limits found in interstellar ices of <10-5 (Boudin et al. 1998) and its detection in cometary ices at an abundance of 0.1-0.9% with respect to H2O (Brooke et al. 1996; Bockelée-Morvan et al. 2000), corresponding to abundances with respect to H2 of 10-7. Therefore evaporation from grain mantles could also explain the observed C2H2 abundance toward IRc2.
The widths of the HCN and C2H2 ro-vibrational bands toward Peak 2 are somewhat smaller than those toward IRc2, suggesting a different origin. Also, much lower abundances are found in the preferred scenario 3, consistent with destruction of these molecules in the shock or heating on short time scales (see Sect. 4.5). Therefore, the HCN and C2H2 toward IRc2 probably originate in gas that has been blown away from the hot core clumps and now resides in the plateau gas in the swept-out cavities between these clumps, probing hot-core chemistry (see Sect. 3.3), whereas the HCN and C2H2 toward Peak 2 probably probe either shock chemistry or quiescent gas-phase chemistry at K.
The excitation temperature estimated from the CO2 emission indicates that the CO2 toward Peak 1 and Peak 2 does not originate in the hot component of the shock at K. The results of Sect. 4.3 suggest an origin in the warm component of the shock. If the CO2 toward Peak 1 and Peak 2 originates in this so-called plateau gas, then it is likely that the CO2 absorption toward IRc2 also arises in this plateau gas, given the similar excitation temperatures and line widths. In that case the CO2 absorption will be seen toward both IRc2 and BN contrary to what is observed for HCN and C2H2. Therefore the CO2 ro-vibrational absorption band probably probes different gas and thus a different chemistry than the HCN and C2H2 ro-vibrational absorption bands toward IRc2.
Charnley & Kaufman (2000) show that C-shocks with speeds above 30 km s-1 in regions with cm-3 can efficiently destroy CO2 after it is sputtered off the grains by the shock, converting it into CO. These physical conditions are approximately met in the Orion outflow (Chernoff et al. 1982; Schilke et al. 1992). The CO2 emission toward Peak 2 is stronger than toward Peak 1, whereas the opposite is the case for CO in the hot component of the shock (GA02). This could suggest that the CO2 gas has been destroyed more efficiently toward Peak 1 than toward Peak 2.
If the CO2 is destroyed completely in the shock, the predicted abundances are lower than the observed gas-phase CO2 abundances toward the IRc2/BN complex and Peak 1/2. Then the molecule may be reformed by high-temperature gas-phase chemistry in the post-shock gas (Charnley & Kaufman 2000). The primary gas-phase formation route for CO2 is: CO + OH CO2 + H. This reaction requires a temperature above 100 K. Since CO is abundant throughout the warm gas, the formation of CO2 is limited by the amount of available OH and it is likely that the OH and CO2 molecules are co-located. Abundant OH is observed both toward IRc2 and Peak 1 and Peak 2 (Watson et al. 1985; Melnick et al. 1987; Cernicharo et al. 1999), consistent with the widespread CO2detection. The OH is produced by high-temperature reactions of O with H2. However, above -300 K all O and OH are driven into H2O, leaving no OH to form CO2 (Charnley 1997). This could explain why the rotational excitation temperature toward all three positions does not exceed this value. Since the observed CO2 abundance toward IRc2/BN is higher than those toward Peak 1 and 2 for scenario 3, this suggests that reformation of CO2 must have been taken place here for a longer time than toward Peak 1 and Peak 2. The abundances toward Peak 1 and 2 suggest that reformation has taken place on time scales of yr for K (Charnley & Kaufman 2000).
If on the other hand, the CO2 is at most partly destroyed in the shock, then grain-mantle evaporation starts to play a role. In Sect. 3.3 it is shown that grain-mantle evaporation can account for the observed gas-phase CO2 abundances toward IRc2/BN. Since the temperature of the warm component of the shock is -400 K, most of the CO2 ice will already be evaporated here, and the observed CO2 ice toward Peak 1 and 2 originates in the extended ridge. This results in CO2 ice abundances of N(CO2)/N(H and respectively, as determined from the SWS spectra, using N(H cm-2 corresponding to the value derived by Sutton et al. (1995) for the extended ridge. Comparison with the CO2 gas-phase abundances in the case of scenario 3 (Table 3) shows that the ice abundances are much higher than the observed gas-phase CO2 abundances toward Peak 1 and Peak 2. Assuming that the pre-shock CO2 ice abundances toward Peak 1 and 2 are similar to those in the extended ridge, this suggests that toward Peak 1 and 2 grain-mantle evaporation probably does not play a dominant role. Thus, the CO2 gas likely probes the same physical component toward all three positions, but the chemical origin may be different. Toward IRc2 the inferred abundances can be explained by both grain-mantle evaporation and reformation in the gas phase on time scales of yr after destruction by the shock or a heating event. The inferred CO2 abundances toward Peak 1 and 2 are best explained by reformation in the gas phase on time scales yr after destruction by the shock.
This work was supported by the NWO grant 614-41-003, and the Research Corporation (SDD). CMW acknowledges support from an ARC Australian Postdoctoral Fellowship. The authors wish to thank Do Kester for his contribution to the high-resolution AOT1 software. The authors would also like to thank John Black, Eduardo González-Alfonso, Teije de Jong, and Xander Tielens for useful discussions and Willem Schutte for providing column densities for solid CO2.
Little is known about collisional rate coefficients for vibrational excitation, even for the simplest molecules (see discussion in GA02). However, some experimental and theoretical values exist for the vibrational de-excitation rate of the CO2 to the state by collisions with H2 and He for temperatures between 150 K and 300 K (Allen et al. 1980; Banks & Clary 1987; Lepoutre et al. 1979). These values are summarized in Fig. A.1. The corresponding excitation rates can be found by multiplying with the factor , with and the statistical weight and energy of the state, respectively. The collisional de-excitation rates for CO2-H2 are at least a factor of 1000 higher than the CO-H2 rates in the same temperature range (Reid et al. 1997). Since the results of Sect. 4.3 show that the observed CO2 originates in gas at -200 K, this suggests that collisions are much more important for CO2 than for CO.
|Figure A.1: Collisional de-excitation rates for CO2 from the to the state, with collision partner X. Only collisions with H2 and He are considered. Triangles and plus-signs denote experimental data from Lepoutre et al. (1979) and Allen et al. (1980), respectively. Theoretical calculations for the CO2-He rates from Banks & Clary (1987) are shown by the thin and dashed lines, showing good agreement with the experimental data.|
It should also be noted that the ratio between the CO2-H2 and the CO2-He vibrational de-excitation rates per collision is at least 50 between 150-300 K (Fig. A.1). This is much higher than the factor of 1.348 based on the difference in reduced mass, which is generally used to convert rotational de-excitation rates of a molecule with He to that with H2. This may be due to resonances with energy levels in the H2 molecule (Allen et al. 1980), leading to enhanced vibration-rotation energy transfer and vibrational de-excitation rates for collisions with H2 compared to those with He. The same large difference is seen between the CO-H2 and CO-He vibrational de-excitation rates (GA02). Nevertheless, further experimental and theoretical studies of the CO2-H2 vibrational de-excitation rates are warranted to confirm the large values adopted here. Since the He abundance is much lower than that of H2, collisions with He are not important. No data could be found for CO2-H collisions but assuming that the empirical formula from Millikan & White (1963) used in GA02 is also valid for CO2-H collisions, we find de-excitation rates a factor of at least 5 lower than the CO2-H2 rates between T=150 K and 300 K. In addition, the abundance ratio H/H2 is expected to be <1. Together these suggest that collisions with H2 dominate.