5.1. CH4 abundance
The ISO-SWS observations toward W 33A and C 7538 : IRS9 indicate that the abundance of interstellar CH4 is low: . This value is determined with respect to the integrated hydrogen column density derived from the depth of the silicate bands (Table 2). However, it is an average along the line of sight, and strong abundance variations may occur locally. Notably, the rather high temperature ( K) of the gas phase CH4 toward C 7538 : IRS9 indicates that it is not associated with the large amount of cold CO gas along this line of sight (Mitchell et al. 1990). Contrary, the kinetic temperature of the warm CO gas ( K) is significantly higher than the CH4 gas temperature, and the gaseous CH4 toward C 7538 : IRS9 then must reside in a separate volume. However, if we assume a two temperature model for the CH4 gas, at least 70% of the CH4 could have a temperature of 200 K (Sect. 4). Then, assuming the warm CH4 and CO gas are in the same volume, we find that , which is an order of magnitude larger than the average along the line of sight (using a conversion factor ; Lacy et al. 1994).
The temperature of the warm CO component toward W 33A ( K; Mitchell et al. 1990) is comparable to the CH4 gas temperature, and these molecules may be present in the same volume. In this line of sight, the warm and cold CO gas components are equally abundant, and the local gas phase CH4 abundance probably does not vary significantly from the average along the line of sight (Table 2).
5.2. Chemistry of interstellar CH4
Toward both W 33A and C 7538 : IRS9, the CH4 ice is embedded in a matrix of polar molecules, and the CH4 gas is warm, K. The combined CH4 gas and ice abundance is . The CH4 gas/solid state abundance ratio () is very low compared to CO, but higher than for H2O. These results allow us to constrain the models for the origin of interstellar CH4.
CH4 may have been formed through grain surface reactions involving accreted atomic C and H, similar to the reaction that converts atomic O into H2O ice (e.g. Brown et al. 1988). This reaction is very efficient. When starting from an atomic gas, the abundance will be , which is two orders of magnitude larger than the observed abundance (Table 2). Hence, if the observed CH4 was formed on grain surfaces, the initial atomic C abundance must have been low, i.e. CH4 was formed during a cold, dense cloud core phase. At the same time, most of the atomic C is locked up in CO, which will, at sufficiently low temperatures, stick to the grains as well. However, CH4 is absent in non-polar ices (Boogert et al. 1996). This could imply that during CH4 formation, CO accreted at a high atomic H abundance, and is efficiently hydrogenated to CH3OH. Using the model calculations on CH3OH formation by Charnley et al. (1997), we find that this conclusion is supported by the low CO/CH3OH ratio in the polar solid phase ( toward C 7538 : IRS9; Allamandola et al. 1992; Tielens et al. 1991).
The low observed gas/solid state ratio for CH4, contrary to CO, is a natural consequence of surface chemistry models. At low temperatures, no CH4 is expected in the gas phase. The high derived temperature for gaseous CH4 for both W 33A and C 7538 : IRS9 supports the location of the CH4 in a hot core region near the protostar. Although pure CH4 ice sublimates at K at low interstellar pressures, it sublimates at temperatures up to 90 K in polar ices, depending on the relative amount of CH4 in the ice (see Sandford & Allamandola (1988) for a discussion on sublimation of H2O:CO ices). The width and peak position of the interstellar CH4 ice band indicate that the ratio of CH4 ice with respect to polar molecules in the ice (H2O and CH3OH) is at most 10% (Boogert et al. 1996). Thus the CH4 sublimation temperature is probably close to 90 K. The observed CH4 excitation temperature is 70 and 110 K for C 7538 : IRS9 and W 33A respectively, and is expected to be close to the gas kinetic temperature. At the high densities in the hot core, the gas and dust temperatures are closely coupled (Ceccarelli et al. 1996), and therefore the observed warm gas phase CH4 may indeed result from out-gassing of H2O-rich ices. The somewhat higher gas/solid CH4 ratio toward W 33A is then probably related to the larger abundance of warm CO gas in this line of sight. Note that the CH4 gas/solid ratio is significantly higher than the H2O gas/solid ratio (Table 2), perhaps indicating that the CH4 molecules diffuse out off the H2O ice matrix at temperatures less than 90 K, well before the H2O ice itself sublimates. We conclude that the presence of CH4 in a polar ice and the low gas/solid ratio are naturally explained in the grain surface models of CH4. These models are restricted to have a high initial CO/C ratio, and an efficient CH3OH formation at a high atomic H abundance.
Rather than grain surface chemistry, an origin of solid CH4 in UV photolysis of CH3OH containing ices has sometimes been suggested as well (Allamandola et al. 1988; Gerakines et al. 1996). Both W 33A and C 7538 : IRS9 contain about 5% solid state CH3OH (Allamandola et al. 1992). However, judging from the strength of the 4.62 µm XCN feature, which is often ascribed to FUV photolysis of interstellar ices, FUV photolysis has been much more important towards W 33A than towards C 7538 : IRS9. Yet, the abundances of solid CH4 are very similar and the solid state CH4/H2O ratio is much less toward W 33A than toward C 7538 : IRS9. Therefore, it seems unlikely that photo-processing of CH3OH-rich ices is an important production mechanism of interstellar CH4.
Alternatively, CH4 may have been formed by low temperature gas phase chemistry (e.g. Millar & Nejad 1985; Helmich 1996) and preserved through accretion in ice mantles. In these models, the observed average abundance along the line of sight is produced in a narrow time interval early in the collapsing phase. The high local abundance of toward C 7538 : IRS9, if it were located in the same volume as the warm CO gas (Sect. 5.1), can never be reproduced by gas phase models. The length of the interval with large CH4 abundances depends strongly on the assumed initial atomic C abundance, i.e. yr when starting from an atomic gas, and yr when starting from CO/C=10 (translucent clouds; Helmich 1996). The models with large initial CO/C ratio's, and thus a lower CH4 production, are probably more realistic. This is because the free fall time to form molecular clouds from diffuse, atomic clouds ( yr; e.g. Elmegreen 1987) is much larger than the time assumed in gas phase models to form hot cores ( yr). Thus, at the start of the collapse, the gas is no longer atomic. After this short peak in the gas phase CH4 abundance, the CH4 `burns' to CO, and consequently the observed CH4 ice must originate from accretion during a very narrow time interval at yr. Furthermore, at this (or any) stage of the collapse, little H2O is present (), and pure gas phase models cannot explain the formation of polar ice mantles. Grain surface formation of H2O is needed to explain the presence of interstellar CH4 in a polar ice mantle (Boogert et al. 1996). Other molecules that, in this model, would be formed in the gas phase (CO, CH4) will then co-condense with the atomic O and H and are trapped in the H2O-rich ice. However, inevitably any accreted atomic C will react rapidly to CH4 on the grain surface as well. Thus, if the observed CH4 ice originates primarily from the gas, the CO/C ratio must have been very high during accretion. This contradicts the low CO/C ratio required to explain the observed CH4 abundance. Finally, at later times ( yr) these gas phase models predict the absence of (cold) gas phase CH4, since it is easily converted to CO (e.g. Helmich 1996). Since star formation has occurred in the cores of C 7538 : IRS9 and W 33A, they are probably older than 0.5 million years, and indeed no cold CH4 gas was detected toward these sources. The observed hot CH4 gas is most likely located in a hot core near the protostar, where the gas phase composition reflects evaporated ice mantles. The species released from the grains in the hot core survive on a time scale of yr (Brown et al. 1988).
Concluding, the observations impose strict conditions to the models of gas phase formation of CH4. Contrary to the grain surface models, a low, perhaps unrealistic, initial CO/C abundance is required to explain the observed CH4 abundance, and the time window for CH4 production and accretion on the grains is narrow in any case. Additionally, the presence of CH4 in a polar ice cannot be explained by pure gas phase models. Additional grain surface formation of H2O and CH3OH is required. We conclude that formation of CH4 on grain surfaces is a more likely explanation.
© European Southern Observatory (ESO) 1998
Online publication: July 7, 1998