4. EDA complex formation
4.1. Study of complex formation in the laboratory
To test the donor-acceptor complex hypothesis discussed above, we have performed many experiments at the IAS. The experiments we performed use the classical techniques of matrix spectroscopy. Gas mixtures are slowly deposited onto a cold (4-100 K) substrate transparent to infrared wavelengths (CsI window). Infrared spectra are recorded with an IFS66v Bruker Fourier Transform Spectrometer (FTS). Details of such experiments can be found elsewhere (e.g. Allamandola 1987).
We chose to prepare ice samples containing in roughly equal proportions CO2 and another molecule with lone electron pair(s). To interact with CO2, we first chose molecules that could be abundant in interstellar space such as H2O, HCOOH, NH3. As methanol seems to be a good candidate to reproduce the astronomical spectrum (Ehrenfreund et al. 1998), we then decided to perform the same analysis with molecules from the alcohol group: CH3OH, C2H5OH, C3H7OH (2 isomers, propan-1-ol and propan-2-ol), C4H9OH (butan-1-ol). In a subsequent step, to ensure ourselves that the process was not specific to the alcohol group molecules, but to lone electron pairs, we also use C3H6O (acetone) and C4H12O (diethylether). In the last experiments, to fully investigate our EDA hypothesis based on interactions with lone electron pair(s), we find necessary to choose other molecules with another kind of atom presenting free electron doublet(s), here the nitrogen atom. Our choice was made on (C2H5)3N (triethylamine, tertiary amine), (C2H5)2NH (diethylamine, secondary amine) and (C3H7)NH2 (propylamine, primary amine) for experimental reasons (a vapor pressure of a few tens of a millibar at ambiant temperature is preferable to obtain significant mixing ratios with the CO2 molecule).
Selected results from these experiments are shown in Fig. 3 (oxygen containing molecules) and Fig. 4 (amines). We clearly see in these figures the three component splitting of the CO2 bending mode when CO2 is mixed with methanol, ethanol, propanol, butanol, diethylether and acetone. In the stretching mode region, the shape of the band is not altered too much as can be seen on Fig. 3 (middle panel), even when the substructure appears in the 15.2 µm band.
The temperature at which the bending mode shows a triple substructure is roughly related to the strength of the interaction between the acid and the base and the evaporation temperature of the less volatile species but stay for the molecules shown in this panel from 65 to 110 K.
The mixtures that do not exhibit the triple substructure can nevertheless show a double peaked structure at a temperature near the CO2 evaporation one (Fig. 3. and Sandford & Allamandola, 1990). This structure is then not related to the complex formation but to the particular CO2 crystalline interactions when a layer of pure CO2 forms by migration of this molecule on top of the mantle. This stage can be very rapid as it is because we approach the CO2 evaporation temperature.
When the experiments are performed with molecules from the alcohol group, the interaction is always present but with the propan-2-ol the effect is less pronounced. This is certainly due to the steric environment of this molecule compared to the propan-1-ol, which reduces the process. This picture agrees with our general ideas on the complex formation.
Experiments with H2O, NH3 and formic acid did not lead to the same observations althought these molecules also possess lone electron pairs. In these cases, the complex formation is probably inhibited by the possibility of such molecules to form strong hydrogen bonds. This is confirmed by other experiments dedicated to the H2O-CO2 complex formation in a nitrogen matrix. It has been shown (Fredin et al. 1975) that a 1:1 complex of the EDA type effectively forms when these molecules are diluted in a nitrogen matrix. The C2 axis of the water molecule is then perpendicular to the axis of the CO2 one, the oxygen atom pointing toward the carbon atom of carbon dioxide (Jonsson et al. 1975). The same occurs for NH3 that forms a geometrically similar complex, the C3 axis perpendicular to the axis of the CO2 molecule (Fredin & Nelander 1976). When these molecules are codeposited in an ice mixture, the hydrogen bonding interaction will dominate the ice structure, hindering the complex formation.
In the last step of our study, we mix the CO2 with molecules containing a nitrogen atom. These molecules are not astrophysically relevant but help to constrain the CO2 EDA complex hypothesis as the same interactions should occur with the nitrogen atom lone pair. Their infrared spectra are very complex as they have many atoms. To distinguish transitions associated with CO2 from those of the amines we have proceeded by depositions increasing each time the carbon dioxide to amine ratio (Fig. 4). After normalisation on the bands already known to pertain to the amines, we expect, at first approximation, to see the CO2 related bands grow. The small deviations in the amine bands normalisation come from slight phase change as the carbon dioxide concentration changes. As can be seen on Fig. 4, an additional band appears at 15.5-16 µm. We again interpret this bending mode splitting in terms of the complex formation.
4.2. EDA complex formation and CO2 segregation
The CO2-X complex formation (where X is a Lewis base) is able to break (in two) the degeneracy of the CO2 bending mode at low temperature (Fig. 3) but a warm-up of the ices is necessary to reproduce more precisely the interstellar observations (three subpeaks). During the annealing of the sample, an additional structure appears in the bending mode in the subpeak initially at 15.2 µm, becoming more and more alike the astronomical spectra of embedded objects.
Of particular interest is then to look at the isotope absorptions to constrain the mechanism responsible for this as it is much less abundant and therefore diluted in the ice matrix. Thus, it behaves in the same way as the first isotope concerning the complex interactions but does not agglomerate, therefore being more appropriate to reveal the effects experienced during the warm-up. Fig. 5 presents the warm up of the methanol-CO2 ice experiment.
The temperatures indicated in the right part of the left panel are indicative of the range of temperature in which the spectra were recorded to ensure a good signal to noise ratio. The temperature was increased in steps of about one hour between each spectrum, except for the last two spectra that showed a rapid variation. Because we approach the CO2 sublimation temperature, in the ice matrix the carbon dioxide molecule can then almost freely migrate to the surface whereas for the spectra at lower temperature, there is no evolution at the laboratory timescale (a day). This reflects the fact that the activation energies for the processes are exponentially dependent.
In space, we expect to obtain the same behavior, but at lower temperatures due to the much longer timescale of evolution implied. The exact temperature will also vary with the presence of other less volatile constituent in the mantle like water (which is expected from our fit with equal proportions of water, methanol and carbon dioxide, Fig. 6.) and the fact that in space the ambiant pressure is much lower than the one encountered in the laboratory, even in the so called "dense cores".
The 13CO2 stretching mode absorption, located at 4.395 µm just after deposition is progressively transfered into another component at 4.38 µm, revealing the rearrangement of the ice mantle as the sample is annealed. In the main isotope (12CO2) bending mode region, the initial two broad features goes trough a three-peak structure. The bending mode feature evolution is entirely correlated to the 13CO2 stretching mode progressive shift and ends in a two sharp peak feature at high temperature (T70-80 K). Below the warm-up sequence (Fig. 5) is plotted a pure CO2 sample deposited and recorded at 10 K. Comparing the warm spectra to the pure CO2 spectrum, there is evidence that during warm-up some CO2 migrates into the matrix. It then aggregates either at the surface or in the bulk to form a pure carbon dioxide mantle or cristallites. The isotope absorption feature around 4.39 µm is then a sensitive and practical probe of the temperature evolution of this ice.
The broadness of the 13CO2 profile observed toward RAFGL7009S could then be a combination of the different line absorption positions of the isotope stretching mode but require a small contribution by another mixture like a CO2:H2O 1:1 to fill the gap at 4.385 µm. Given the profile observed at 15.2 µm (where we would not observe the triple substructure with such a mixture), as well as the fits on methanol transitions obtained in the short wavelength part of the spectrum (Dartois et al. 1999), this should not be the major contribution.
The observation of asymetric profiles in various protostars would be the proof of the complex formation and provide a second temperature constrain as the dependance can be measured in the laboratory.
As a conclusion, we have shown that CO2 interacts with some Lewis bases to form an EDA complex. The complex formation is betrayed by the splitting of the CO2 bending mode. Warm-up of the resulting ice can provide additional structure in the band as a combination of both the complex and some pure CO2 ice segregation.
© European Southern Observatory (ESO) 1999
Online publication: November 16, 1999