Technical Comments

Response to Comments 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.1167919


We address the comments by Lyons et al., Federman and Young, and Yin et al. regarding the interpretation of our carbon monoxide photodissociation experiments and provide further experimental data analysis, including measured synchrotron beam profiles. The experimental data do not support existing self-shielding models that attempt to explain observed meteoritic oxygen isotopic compositions because they rely on previously untested theoretical assumptions.

We reported that an anomalously enriched isotopic reservoir can be generated through carbon monoxide photodissociation without self-shielding (1). Here, we respond to comments by Lyons et al. (2), Federman and Young (3), and Yin et al. (4) on critiquing our results. Important aspects of all self-shielding models (57) are the basic assumptions that (i) the quantum yield of dissociation is unity (e.g., each photon absorption after light-shielding leads to immediate dissociation with no other fractionation process), and (ii) the oscillator strengths and predissociation probability for different isotopomers are identical to those for 12C16O (e.g., no isotope selectivity). We address both of these issues. These assumptions lead to the conclusion that in self-shielding models, isotope selectivity arises only due to photo-shielding. The self-shielding model for the solar nebula (7) was based in large part on the self-shielding model of (5), and those assumptions lead to a slope value of unity (equal 17O, 18O in oxygen three-isotope plot) in the product atomic oxygen reservoir to produce the observed meteoritic values. Our experiments (1) provided the first test of the model. Yin et al. (4) claimed that an arbitrary criterion of a slope equal to 1 was adopted. It is actually the self-shielding model (5) that assumes this, and nebular self-shielding models (6, 7) require it to explain the calcium-aluminum-rich inclusion (CAI) data; hence, the slope = 1 value is not our adoption.

The issues addressed in the comments (24) in general are (i) observation of a slope value >1 in a delta-delta plot, contrary to shielding predictions and associated optical thickness; (ii) overlap of discrete CO absorption bands within single synchrotron bandwidths and band-head separation among different CO isotopologs; and (iii) CO residence time in the reaction chamber and self-shielding in the differentially pumped (DP) sectors (prior to the reaction chamber). Although the comments are based on theory and assumptions, our response is based on experiments and is assumption-free.

Shielding isotopolog-specific spectral lines quantitatively depends on optical depth. The optical depths for the two extreme column density experiments [runs 8 and 7 in table 1 in (1), lowest and highest column densities, respectively] are optically thick for C16O but thin for the minor isotopologs (C17O and C18O), ideal for testing self-shielding. Lyons et al. (2) and Federman and Young (3) suggest that the optical depth of the experiments is sufficiently high to shield C18O lines, and thus that the observed slope values support self-shielding. For the experimental conditions, the optical depths of these two extreme densities are 1.07 and 2.90 for C16O, with corresponding values for C18O of 2.13 × 10−3 and 5.79 × 10−3, respectively (8, 9). Confusion may arise regarding the use of integrated absorption cross sections versus the peak absorption cross section to calculate the optical depth. The peak absorption cross sections are a few orders of magnitude higher, and using peak absorption cross sections may increase the optical depth compared with the integrated cross sections. The CO absorption band is structured with spikes over a continuum (9), and to calculate the total absorption by a band, line-by-line integration using peak cross-section over the entire band is used. A recent line-by-line (using peak absorption cross section) model calculation (10) of our experiment (1) yields a slope value of 1.05 for the 105-nm synchrotron band. In the model calculation, a slope value of unity is obtained when C18O and C17O lines are optically thin, and the slope value diverges rapidly with increasing optical opacity of C18O lines because of the increase in preferential dissociation of the C17O isotopolog. Comparing these model results (10) with that shown in figure S1 in (1) indicates that the use of the integrated cross section underestimates the effective cross section by ~3 times, which is inadequate to make C18O line optically thick (the upper limit of optical depth of C18O increases to 1.7 × 10−2 from 5.79 × 10−3). Using the effective integrated cross section, the absorption (I/Io) of C18O for these two experiments (runs 8 and 7) are 0.997 and 0.993, respectively, and hence no shielding of C18O lines occurs within the reaction chamber that could account for the measured slope in excess of 1.0, as noted by Lyons et al. (2). Therefore, the high slope values measured in our experiment are not due to optically thick C18O lines, consistent with the initial premise shown in figure S1 in (1). Figure 1 shows a comparison in a three-isotope plot of experimental data with the model calculation (10) for the two extreme column density experiments [runs 8 and 7 in (1)]. The model values fall on a slope 1 line (predicted by self-shielding) but severely overestimate 18O and underestimate 17O [by a few 1000 per mil (‰)] compared with the measured values. The observations thus require the intervention of another isotope selective mechanism during photodissociation subsequent to self-shielding.

Table 1

Measured and calculated pressures of the reaction chamber and three differentially pumped (DP) chambers for all experimental runs described in (1). The effective optical depths for these DP sectors are shown, which indicate negligible absorption.

View this table:
Fig. 1

Comparison of the measured isotopic composition of two experiments with the highest- and lowest-column density for the 105.17-nm synchrotron band, i.e., runs 7 (red triangle) and 8 (red diamond) (1), respectively, with the corresponding values derived by a self-shielding model (10) shown in blue, in a three-isotope oxygen plot. The model values lie over a slope value of 1.05, showing that the experimental conditions are correct for testing self-shielding. To explain the discrepancy between the measured and model values, another post-shielding, isotope selective process is required in addition to the photodissociation process.

Federman and Young (3) comment, based on a self-shielding model (5), that for an astrophysical environment, our experimental column densities are sufficiently high to saturate both minor isotopolog lines and are not appropriate for such conditions. It should be noted that the CO self-shielding model (5) fixes the column density of hydrogen [N(H)] at 5 × 1020 cm−2 and computes the column densities of H2 and CO. There is no upper limit of CO column density for self-shielding, and under the experimental condition, the basic physical-chemical process and nebular environment is appropriately tested.

Regarding the CO2 production yield, it is 0.57 μmole (for run 7), which is only ~0.5% of the expected yield (120 μmole) from the input photons and for a quantum yield of 1 for photodissociation. This extremely low yield is not due to inefficient trapping of CO2 in the experiment because the 2-m-long stainless steel spiral submersed in LN2 (during the entire length of the experiments) was used to collect the product CO2. This apparatus typically collects >99% of CO2. Therefore, the low yield of CO2 production requires another physical chemical process subsequent to the photo-shielding and during the actual bond scission process.

As noted in (24), there are a few CO absorption bands under the broad spectral spread of the synchrotron beam, and their contributions collectively depend on each band’s absorption cross section and corresponding photon intensity. The actual experimentally measured synchrotron beam energy profile centered at 105.17 nm is shown in Fig. 2 along with the contributing CO bands and their relative photolysis contributions (run 7). Within the 105.17-nm synchrotron band, CO absorption bands at 105.17, 106.31, 107.61, and 108.79 nm are present. The 108.79-nm band does not lead to dissociation (11). The 105.17-nm CO absorption band receives the peak photon intensity of the synchrotron beam, whereas the 106.31-nm absorption band receives 0.75 times the peak photon intensity. The absorption bands at 107.61 and 108.79 nm receive only 0.45 and 0.30 times the peak photon intensity, respectively. Self-shielding is only possible for a particular CO absorption band when the band heads of different isotopologs are separated. Based on experimental data (8, 9), the bands at 105.17 and 106.31 nm may self-shield (the peak head separations between C16O and C18O are 49.26 and 51.18 cm−1 for 105.17 and 106.31 nm, respectively). The band heads are closely spaced for the 107.61-nm absorption band, and their positions for 12C16O, 13C16O, 12C17O, and 12C18O isotopologs (the experiments were done with CO of natural isotopic abundance, and all isotopolog specific lines must be considered) are 92929.912 ± 0.008, 92929.709 ± 0.008, 92929.670 ± 0.004, and 92929.510 ± 0.002 cm−1, respectively (12). The spacing of all these lines varies from 12C16O: 0.203 ± 0.011, 0.242 ± 0.009, and 0.402 ± 0.008 cm−1 for 13C16O, and 12C17O, 12C18O, respectively. The effective broadening of the spectral lines of ~0.17 cm−1 results in irresolvable lines for 13C16O, 12C17O, and 12C18O, which render the 107.61-nm band a nonshielding band, contrary to the claim of Yin et al. (4). It is calculated that for the 105.17-nm synchrotron band ~77% dissociation was through the shielding bands (105.17 and 106.31 nm) and only 23% was through the nonshielding band (107.61 nm). In contrast, for the 107.61-nm synchrotron band, ~72% of dissociation is produced through a nonshielding band. Therefore, the 105.17- and 107.61-nm synchrotron bands are representative of shielding and nonshielding bands, respectively.

Fig. 2

Measured synchrotron beam profile centered at 105.17 nm. Of the four CO absorption bands within this synchrotron band, three take part in dissociation and are shown in the figure. The 105.17- and 106.31-nm bands are shielding bands and together are responsible for about 77% of CO dissociation; the remaining dissociation occurs at the 107.61-nm nonshielding band. The 105.17-nm synchrotron band is therefore considered a representative shielding band.

The measured individual slope values (δ17O/ δ18O) for the 107.61-nm synchrotron band [runs 14 and 15 in table 1 of (1)] are 1.85 and 2.13, respectively (where 72% dissociation is through the nonshielding band), significantly higher compared with that measured in all five experiments (runs 2, 3, 5, 6, and 8) for the 105.17-nm synchrotron band (mean slope value = 1.68 ± 0.05, where 77% dissociation was through the shielding bands). The observations at shielding-dominated versus nonshielding synchrotron bands are inconsistent with a pure shielding hypothesis and require the involvement of another process.

The objective of the photolysis experiments at 97.03 and 94.12 nm was to compare the effects of higher electronic states with the upper electronic state E1Π. There is a very good correlation between the synchrotron photon energy and the observed slope values. The slope value decreases from about 2.0 to 0.62 with an increase of photodissociation energy (11.52 to 13.17 eV), which indicates an electronic state–dependent fractionation. Photon self-shielding only alters incoming photon energy distribution for initiation of bond dissociation. The actual photochemical fractionation process is controlled by the predissociation process, post-light absorption, and it is curve-crossing factors that dominate initial self-shielding. The upper electronic state for the bands at 107.62 and 105.17 nm are the same, but with different vibrational states (v = 0 and 1, respectively). The similar slope value (1.38) for these two bands [shown in figure 1 in (1)] in a three-isotope oxygen plot depicts the role that accidental predissociation plays in these bands—a dominant effect that controls the final isotopic ratios. Dissociation at 97.03 and 94.12 nm arises from two different upper states (C1Σ and 1Π, respectively) and consequently yields a different fractionation as a result of a dissociation effect and not photo-shielding. We agree with Lyons et al. (2) and Federman and Young (3) that the CO absorption band systems within the 97.03-nm synchrotron band are complex and require careful analysis. Nevertheless, experiments reveal that the isotope effect in a direct predissociation to a repulsive state is different from that of electronic predissociation at a E1Π state.

Lyons et al. (2) claim that the observed slope value of 0.64 for the 94.12-nm synchrotron band is due to a contribution from interfering diffuse CO absorption bands within this synchrotron band, which fractionate mass dependently and lower the slope value from unity (which arises from self-shielding at the central band at 94.12 nm). Self-shielding can enhance 18O/16O a few hundred to thousands of ‰, whereas the observed photodissociation (using a broad light source) process fractionates molecules by only tens of ‰ (1315), unless a complex dissociation process (e.g., curve cross effect as seen for CO2 dissociation) is involved (16). Thus, it is unlikely to reduce the value of the slope of unity (with very high δ values generated through self-shielding) to a value of 0.6 by contribution of a mass-fractionated composition (of small delta values), for which there is otherwise no supporting experimental or theoretical basis.

Lyons et al. also question the effect of CO pressures in differentially pumped (DP) sectors upon opacity. In Table 1, the experimental pressures in the DP sectors and associated optical depths for the experiments are presented. None of the values are close to the range (1 to 14) theoretically suggested by Lyons et al. Hence, there is nearly zero absorption in the DP sectors (I/Io for C16O is ~0.998). The measured flow rate in the experiments is <1 SCCM (standard cubic centimeter per minute); thus, the CO residence time within the chamber is ~300 min, compared with 0.9 s calculated by Lyons et al., an error in their calculation of ~104. Experiments are performed under standard steady-state flow conditions in the reaction chamber [figure S1 in (1)], so the loss of CO through DP sectors is irrelevant for the isotopic measurements.

To avoid the beam overlap problem addressed earlier, Yin et al. (4) suggested conducting CO photolysis experiments using a high-resolution VUV laser system. The beam overlap issue may be avoided with a high-resolution VUV laser system, but a direct self-shielding effect cannot be appropriately studied because the laser resolution (~0.1 cm−1) would be too high to capture a full single CO dissociation band (typically more than 50 cm−1) in one laser setting. Therefore, this system would not be appropriate for high-precision mass-spectroscopic analysis of the product gas. Furthermore, suggestions of performing these experiments under relevant astrophysical conditions are restricted by the available techniques.

To derive a unique isotopic composition of the atomic oxygen pool produced by CO photodissociation through all the CO absorption bands, Yin et al. (4) suggest summing up the yields at different bands that have been normalized by time-integrated VUV photon fluence at different wavelengths and weighted by the solar radiation field. This suggestion requires CO photolysis experiments at all 40 CO absorption bands (a total of 12 synchrotron bands) and could be the subject of future projects. We note, however, that this endeavor would require more than 3000 hours of synchrotron time and would not affect the conclusion of our original study (1).

After CO photodissociation in our experiments, reaction of O and CO generated CO2, the analyte gas for high-precision isotopic measurement. The issue of mass-independent fractionation during CO + O reaction is raised by Yin et al. The CO + O reaction was first measured by one of the authors (M.H.T.) (17) and is specifically addressed in our report (1). The fractionation associated with this process is ~90‰, far less than the fractionation observed in these experiments (up to 4082‰ in δ18O). Inclusion of this factor does not change the critical slope value and may explain the nonzero intercept, as already discussed in (1).

In sum, the solar nebular model of self-shielding requires that the quantum yield in CO photolysis be near unity with no isotope effect in photodissociation to produce the meteoritic slope 1 line. Our experiments demonstrate that these assumptions are incorrect. The experimentally observed fractionation is dominated by the actual fractionation associated with bond breakage, an effect ignored by all models and which gives rise to unprecedented enrichment in 17O [see figure 1 in (1)]. Recent spectroscopic measurements of oscillator strengths and predissociation rates for Rydberg transitions in isotopologs of CO to model the predissociation process (12, 18) also show that these assumptions are incorrect. Therefore, the meteoritic slope 1 line in the oxygen three-isotope plot cannot be due to self-shielding in the solar nebula as illustrated in Fig. 3, fundamentally the most important aspect of the experiments of (1).

Fig. 3

Three-isotope oxygen plot showing meteoritical oxygen isotopic compositions [e.g., CAI line with slope ~1.0 (19)] and the experimental photochemical mixing line [between the solar wind composition (20) and the experimental data] of slope 1.72. This comparison shows that CO self-shielding in the solar nebula cannot explain the observed CAI trend line.


View Abstract

Navigate This Article