Technical Comments

Comment on “Experimental Test of Self-Shielding in Vacuum Ultraviolet Photodissociation of CO”

See allHide authors and affiliations

Science  19 Jun 2009:
Vol. 324, Issue 5934, pp. 1516
DOI: 10.1126/science.1167717


Chakraborty et al. (Reports, 5 September 2008, p. 1328) demonstrated very large, wavelength-dependent mass-independent isotopic effects during carbon monoxide (CO) photodissociation and argued that self-shielding in CO was not responsible. We suggest that variations in band oscillator strengths and linewidths among CO isotopologs are responsible for most of the wavelength dependence observed and that the reported experiments confirm the importance of self-shielding during CO photodissociation.

Chakraborty et al. (1) reported vacuum ultraviolet (VUV) photodissociation experiments on CO gas using the Advanced Light Source (Lawrence Berkeley National Laboratory) as the radiation source. In these experiments, O produced during photolysis combined with CO to form CO2, which was collected in LN2-cooled cold traps. Oxygen isotope analysis of the CO2 revealed very large enhancements of 17O and 18O, with δ17O and δ18O values ~1000s of per mil (‰). The objective of the experiments was to provide a laboratory test of CO self-shielding, a photochemical process that is believed to be essential to understanding the distribution of oxygen isotopes throughout the solar system (24). Chakraborty et al. concluded that because their experiments did not produce δ17O/ δ18O ≈ 1.0, as predicted by self-shielding models and as observed in highly refractory mineral phases in primitive meteorites (5), CO self-shielding was not an important process in the solar nebula. Although we vigorously support the type of experiments carried out by Chakraborty et al., we strongly disagree with their interpretation of their results.

Isotope effects during photolysis can arise from variations in the absorption cross sections among isotopologs and also from self-shielding for molecules with line-type absorption spectra. Such rovibronic spectral features often accompany long-lived predissociation states. Photolysis of a given isotopolog in a column of gas occurs until the lines of that isotopolog are saturated, so abundance-dependent isotope fractionation occurs. For CO in the solar nebula (or parent molecular cloud), all CO isotopologs are optically thick everywhere, except in the outer reaches of the nebula where CO freezes out. In the experiments of Chakraborty et al. (1) the CO gas column densities are within the range of 4 to 12 × 1017 cm−2. The peak cross sections for the bands they analyzed are ~ 3 × 10−16 to 10−15 cm2, implying that C16O was optically thick (optical depth ~100 to 1000) in all experiments reported. The optical depths for C18O are 0.2 to 2, and C17O is optically thin in all experiments. This makes the expected isotopic shifts complex, but consistent with δ17O > δ18O. Thus, the conditions for C16O self-shielding were clearly met in the experiments, consistent with the large delta values measured (δ17O(CO2) ~ 1000 to 5000 ‰). Large delta values (~2000 ‰) are also obtained in the CO self-shielding models (4). Although the experiments were in a regime where C16O self-shielding occurs, Chakraborty et al. concluded that self-shielding was not the principal process responsible for the observed fractionation. They reached this conclusion because they observed marked wavelength dependence in the δ17O/ δ18O ratio of photoproduct O (measured as CO2), with ratios (using the logarithmic definition in Chakraborty et al.) varying from ~0.65 at 94.12 nm to 1.45 at 105.17 nm and 107.61 nm, whereas self-shielding models (4) using shielding functions (6) obtain δ17O/ δ18O ≈ 1.0.

Chakraborty et al. attribute this difference in δ17O/ δ18O ratios to CO dissociation dynamics, focusing in particular on accidental predissociation in CO isotopologs. However, there are several reasons for the wavelength dependence they measure. First, the variation of oscillator strengths and predissociation probabilities among the CO isotopologs undoubtedly contributes to isotope-selective photodissociation. Such differences are evident in the few CO isotopolog absorption bands that have been measured (7). The shielding function formulation used in self-shielding models (4, 6) assumes identical oscillator strengths, linewidths and mutual shielding for all isotopologs. Second, the bandwidth of the synchrotron beam is ~2 nm full width at half maximum (FWHM) with a Gaussian beam shape. This means that several bands (~3 to 10) are involved in photolysis at each of the four wavelengths analyzed. At 94.12 nm, ~10 bands fall within the synchrotron beam. Several of these bands are diffuse [e.g., 13, 15 and 19, using the band numbering of van Dishoeck and Black (6)] and will yield mostly mass-dependent fractionation, which may explain the small slope measured for this case. At 97.0 nm, four bands are dominant within the synchrotron beam (Fig. 1). Three of these bands undergo shifts of 10s cm−1 upon isotope substitution (8) and will certainly undergo self-shielding. Band 24 at 97.03 nm has a band origin shift of 0.2 cm−1, comparable to the Doppler broadening width of ~0.05 cm−1 (not pressure broadening, as suggested by Chakraborty et al.) and may not contribute to self-shielding because of line overlap. Nevertheless, large non-mass-dependent fractionation is measured at this wavelength because of the synchrotron beamwidth. At both 105.17 and 107.6 nm, band 32 at 106.3 nm contributes strongly to self-shielding. In addition, computed isotopolog spectra (9) demonstrate that self-shielding will occur in the P and R branches of band 33 at 107.6 nm. Third, the experiment uses a continuous flow of CO. One of us (J.R.L.) has observed in self-shielding calculations that large transient δ17O/ δ18O ratios (e.g., ~1.30) occur before reaching (lower) steady-state values. This is a result of a relatively higher rate of C17O photolysis due to C18O self-shielding. In the experiments of Chakraborty et al., the gas flow rate is 10 SCCM (standard cubic centimeters per minute at standard temperature and pressure) and the reaction chamber volume is 300 cm3, so CO resides in the chamber for 0.9 s for an experimental pressure of 400 mtorr. We estimate the dissociation time scale for the CO gas in the reaction chamber to be >>104 s, which means that <<0.01% of the CO gas in the chamber is photolyzed before replacement with new tank gas. By contrast, in the solar nebula at 30 astronomical units, the radial mixing time scale is ~105 years and the time scale to dissociate the complete column of CO is ~105 years (4), so a substantial fraction of the gas will be photolyzed.

Fig. 1

The absorption spectrum (black) of 12C16O at 295 K (10), with Gaussian beam from synchrotron (red, in arbitrary units) centered on 97.03 nm (FWHM = 1.88 nm). The four bands shown dominate the contribution to CO photodissociation. At the low pressures of the experiment, all four of these bands will undergo self-shielding. In an H2-rich astrophysical environment (also at 295 K for comparison with CO), absorption by H2 (blue) alters the weighting of the bands.

Because the reaction chamber and first differential pumping stage are separated by a large orifice rather than a narrow tube, there is a large rate of CO flow into the first pumped stage. The reaction chamber and first pumped stage are separated by a 3-mm-diameter orifice (1), which has a conductance ~0.8 to 1.4 l s−1 in a thin-wall approximation for molecular flow to transition flow regimes. For a conductance value of 1 l s−1 and a pressure of 0.5 mbar, the CO flow rate through the orifice is 0.5 mbar l s−1 = 30 SCCM, implying that ~ 75% of inlet CO is lost to the first turbomolecular pump. In addition, the pressure ratio between the reaction chamber and the first pumped stage is ~200, assuming the maximum turbo pump rate and the shortest reasonable vacuum manifold. This yields a pressure of 2.6 microbar in the first pumped stage and maximum optical depths of ~1 to 14 (depending on wavelength); the range of optical depths is a factor of 4 lower for the lowest pressure experiments performed. Thus, 12C16O dissociation and self-shielding occurred in the first pumped stage for several of the reported experiment runs, which yields massive 17O and 18O enrichments in the reaction chamber.

In summary, the experiments of Chakraborty et al. (1) are very much in a self-shielding regime, which accounts for the large enrichments measured. Variations in band oscillator strengths and linewidths among the rare CO isotopologs likely contributed to the wide range of δ17O/ δ18O values measured. CO photodissociation experiments under optically thin conditions could be used to separate self-shielding effects from those due to variations of oscillator strengths and predissociation probabilities among the isotopologs. The temperature dependence of the δ17O/ δ18O ratio, when summed over all dissociating bands, will likely place an important constraint on the location of self-shielding in the solar nebula and in molecular clouds.


View Abstract

Navigate This Article