Communicated by F. Fleming Crim, University of Wisconsin, Madison, WI, April 4, 2005 (received for review January 21, 2005)
Atmospheric field measurement and modeling studies have long noted discrepancies between observation and predictions of OH and HO2 concentrations in the atmosphere. Novel photochemical mechanisms have been proposed to explain these differences. Although inclusion of these additional sources improves agreement, they are unable to fully account for the observations. We report and demonstrate the importance of weak electronic absorption features, normally ignored or not measured, in contributing to significant OH radical production. Experiments on methyl hydroperoxide, a prototypical organic peroxide in large abundance in the troposphere, highlights how photochemistry in the neglected electronic absorption tail makes an important addition to the tropospheric OH budget. The present results underscore the need to measure absorption cross sections for atmospheric molecules over a wider dynamic range, especially over the wavelength regions where the solar flux is high, to fully quantitate their contributions to atmospheric photochemistry.
Most gases emitted into the atmosphere by natural and anthropogenic activities are removed by their reaction with OH radicals (1–5). Indeed, OH radicals are often called the “detergent” of the atmosphere as reaction of trace gases with hydroxyl radicals leads to the oxidation of these volatile compounds and their ultimate removal from the atmosphere (1). Even though it exists in relatively small concentrations, ≈106/cm3, without the presence of OH radicals the composition of the atmosphere would be vastly different and potentially hazardous as the absence of this radical would lead to a large build-up of toxic atmospheric gases, many of which can also contribute to the greenhouse effect. Hence, accurate modeling of the oxidizing or cleansing capacity of the atmosphere requires knowledge of all significant sources of the hydroxyl radicals. Although several studies have suggested large changes in OH abundances in the atmosphere over the past decades, the direction of the change is apparently unclear. Studies by Krol et al. (6) show a significant global increase in OH, whereas those by Prinn et al. (7) show a decline in OH levels after 1988. Recent studies in the upper troposphere also have reported discrepancies between model predictions and measured HOx (where HOX ≡ OH + HO2) concentrations that strongly depend on solar zenith angles (8–11). These differences between predictions and observation not only highlight the difficulties associated with measuring atmospheric OH concentrations, but also potential variability arising from incomplete accounting of all significant sources and sinks of the OH radical. In this article we present experimental evidence that demonstrates that absorption from often-neglected weak tails of electronic absorption bands can lead to significant amounts of photochemically generated OH radicals that, apparently, are currently unaccounted for in tropospheric models. Methyl hydroperoxide (CH3OOH), which is the prototypical organic peroxide (ROOH) and acts as a sink as well as a temporary reservoir for HOx and ROx species in the atmosphere (12–14), is used to highlight the process under consideration. The relatively large concentration of CH3OOH in the troposphere, the unit quantum yield associated with its dissociation, combined with the fact that this dissociation is initiated by the portion of the solar flux that readily penetrates into the troposphere and hence can initiate photochemistry over a broad range of solar zenith angles, makes this neglected source an important addition to the tropospheric OH budget.
Photodissociation of ozone followed by the reaction of the resulting O(1D) fragments with water and methane is the major mechanism for the production of OH in the troposphere, followed by the photodissociation of hydrogen peroxide (15, 16). Recently there has also been interest in various “unconventional” photochemical sources of atmospheric HOx radicals, arising from the need to explain photochemistry observed at high-solar zenith angles (9, 11). Donaldson et al. (17–20) have argued for the importance of vibrational overtone-induced dissociation as being a significant source of HOx yields in the atmosphere. Vibrational overtone excitation relies on the anharmonicity associated with molecular vibrations. Because light atom stretching motions such as that involving the O—H stretch typically have the largest mechanical anharmonicities, overtone excitation is most efficient for exciting these high-frequency motions (21, 22). Furthermore, because these transitions typically involve excitation at relatively long wavelengths, photochemistry initiated by vibrational overtone excitation is expected to be especially important at high solar zenith angles where absorption and scattering remove much of the energetic short wavelengths present in the solar actinic flux. Researchers investigating the dynamics of vibrationally excited molecules have long used vibrational overtone excitation as a means to prepare and study the spectroscopy of energized molecules (21, 22); nevertheless the role played by these transitions in atmospheric chemistry is just beginning to be recognized. Roehl et al. (23), for example, have demonstrated that the weakly bound species HO2NO2 can be photolyzed in the infrared through vibrational overtone excitation in the vicinity of the 2νOH and 3νOH bands, thus giving rise to HOX radicals. Vaida et al. (24) have argued that vibrational overtone pumping of sulfuric acid and sulfuric acid–water complex can lead to production of SO3 radicals in the upper stratosphere and mesosphere. What we bring attention to in this article is the possibility that excitation in the weak tail region of electronic absorption bands, which are frequently neglected because of the apparent difficulty associated in determining their contributions (25), can give rise to additional atmospheric radical sources whose influence may be stronger than those arising from vibrational overtone-induced dissociation of the same molecule. The contribution from these weak electronic absorption tails can, in many cases, persist over a wide range of solar-zenith angles and hence be significant in modeling tropospheric chemistry. Organic peroxides are one class of molecules where this effect is expected to be important, and we illustrate this effect by using CH3OOH as the prototypical system.
The near-UV photodissociation of CH3OOH is known to be a significant source of OH radicals (10), whereas the reaction CH3OOH + OH is an important removal mechanism for OH in the upper troposphere (14). Current tropospheric models include only the photochemistry associated with UV photodissociation of CH3OOH between 210 and 360 nm (26). Several laboratory studies (27, 28) have shown that CH3OOH absorbs strongly over this wavelength region, resulting in excitation to a dissociative electronic excited state and the concomitant production of photofragments (29–31). Vaghjiani and Ravishankara (29) measured the quantum yields for OH production from the photodissociation of CH3OOH at 248 nm and showed that it is near unity. This finding was also confirmed under molecular beam conditions by the measurements of Thelen et al. (30), who found that at 193 and 248 nm CH3O + OH are exclusively produced. The dynamics of the UV photodissocition were further investigated by Novicki and Vasudev (31), who measured vector correlation associated with the OH fragment arising from the dissociation of CH3OOH at 266 nm. Below we demonstrate that photons in selected regions between 365 and 640 nm can also photodissociate CH3OOH. Photochemistry over the 365- to 640-nm spectral region has not been investigated previously, to our knowledge, and covers excitation within the continuum tail of the molecule's first electronic absorption band and extends down to the structured region of the fourth overtone of its OH stretching vibration, the 5νOH band, which is the lowest OH stretching vibrational overtone state having sufficient energy to undergo unimolecular dissociation with unit quantum yield.
In our experiments we make CH3OOH as outlined in the literature (32) and slowly introduce its vapor into our glass photolysis cell. Tunable photolysis light covering the region between 365 and 405 nm is generated by sum frequency mixing of a dye laser output with that of a Nd:Yag laser's fundamental. The tunable photolysis light is introduced into the cell, and the OH fragments generated from the photodissociation are monitored by laser-induced fluorescence at 308 nm by using a second Nd:Yag pumped dye laser system. For excitation of the CH3OOH 5νOH vibrational overtone band in the 600- to 640-nm range, the dye laser output of the photolysis laser system is used directly. To provide a reference photolysis beam, we also introduce into the cell light at 355 nm generated from the third harmonic of another Nd:Yag laser. The absorption cross section at 355 nm is known from previous measurements (28) to be ≈2.1 × 10–22 cm2 per molecule. The fixed wavelength reference beam enters the cell collinearly with the tunable photolysis beam. The pressure in the cell is typically set to ≈2 Torr by the addition of nitrogen buffer gas to the 30-mTorr CH3OOH sample, and the time delay between the excitation and probe laser is set to ≈2 μs to thermalize the nascent OH fragments. Comparing back-to-back yields of OH at the desired photolysis wavelength with that from the 355-nm reference beam and normalizing for the laser powers gives us an estimate of the absorption cross section at the desired wavelength relative to the known value at 355 nm.
The absorption cross sections for CH3OOH measured at several selected points between 365 and 640 nm are given in Table 1. The corresponding absorption spectrum is shown in Fig. 1, where the black dashed line on the electronic absorption curve represents previous measurements and the black solid line represents present results obtained by interpolating through the data points of Table 1. Ab initio calculations indicate that there is one dominant excited singlet electronic state responsible for this absorption band (J.S.F., unpublished work). As the excitation wavelength is increased toward the visible (λ ≳ 450 nm) in Fig. 1, we enter the spectral region where the CH3OOH absorption is dominated by vibrational overtone excitation, resulting in the dissociation of the molecule on its ground electronic surface. We consider only the 5νOH band of CH3OOH, occurring in the vicinity of 620 nm, as this is the lowest-order OH stretching overtone state having unit dissociation quantum yield. We note that absorption and dissociation from the 6νOH (≈532 nm) and 7νOH (≈ 470 nm) vibrational bands can also occur but these excitations are expected to be substantially weaker as vibrational overtone cross sections decrease roughly a factor of 10 for each unit increase in the order of the overtone (21). Thus by only considering the 5νOH overtone band our present analysis gives a fairly accurate, although strictly speaking a lower limit estimate, of the total contribution coming from these neglected absorption features. Fig. 2 gives a more detailed view of the CH3OOH 5νOH band. This action spectrum of the vibrational band was generated by scanning the wavelength of the vibrational excitation laser (i.e., photolysis laser) while probing the yield of OH fragments in their vibrational ground state corresponding to the N = 2 rotational level. The two strong central peaks in the spectrum shown in Fig. 2 agree fairly well with earlier results from the photoacoustic measurements of Homitsky et al. (33) although with much better signal to noise. Unlike the photoacoustic spectrum, which reveals the excitation wavelengths absorbed by the sample, the action spectrum contains additional information resulting from it being a convolution of the absorption probability and dissociation quantum yield. Thus the 5νOH spectrum in Fig. 2 not only indicates that the CH3OOH molecule absorbs visible light in the vicinity of 620 nm, but in addition shows that after doing so it gains enough energy to dissociate and give OH fragments. Our measured cross sections for the structured 5νOH band, indicated on the vertical axis of Fig. 2, are in good agreement with the results of Homitsky et al. (33), who obtained their values by comparing absorption strengths of CH3OOH with known cross sections of overtone transitions in H2O, appearing in their photoacoustic spectrum roughly over the same spectral region. The action spectrum in Fig. 2 also establishes the purity of our CH3OOH sample. Hydrogen peroxide, a common decomposition by-product, has strong spectral features associated with its 5νOH band that appear at ≈615 nm in the action spectrum. Thus the absence of these features confirms that the level of this impurity is negligible.
The UV electronic absorption data reported in the literature (26–28) are represented by the black dashed line. The black solid line represents interpolation through the present absorption data. The red solid line shows the solar flux over the region of 10 km and zenith angle of 90°. The observed 5νOH vibrational overtone band between 600 and 640 nm is shown in black.
Room temperature action spectrum of the CH3OOH 5νOH band between 600 and 640 nm. The cross section for this band is normalized relative to the electronic absorption cross section at 355 nm (28).
We have investigated the excess energy associated with the 5νOH-initiated dissociation process to determine how close this band is to the CH3O—OH dissociation threshold. This issue is important as dissociation quantum yields near threshold can be affected by collisions, which are common under atmospheric conditions and can dissipate the excitation energy. A careful measurement of the nascent OH fragment Doppler profiles and rotational state distribution in our measurements suggests that 5νOH excitation has at least 1,300 cm–1 of excess energy; the lower limit in this estimate arising from lack of information regarding the internal energy in the partner CH3O fragment. These findings are also consistent with ab initio calculations that we have carried out at the CCSD(T)/CBS limit, which give a D0 value of 43.0 kcal/mol for breaking the O—O bond in CH3OOH (34, 35). As collisions with background gases in the atmosphere, mainly nitrogen, are expected to remove ≈100–200 cm–1 of energy on every collision (26), roughly six or more collisions will be required to remove enough energy to quench the dissociation of a CH3OOH molecule excited to the 5νOH level. Under tropospheric conditions this process will take ≈7 ns. Varying the time delay between the vibrational overtone excitation laser and probe laser while looking at the time evolution of the OH fragment in our experiment reveals that the dissociation from the 5νOH level occurs on a time scale faster than ≈5 ns, which is the temporal resolution limit in our measurements. From these observations we conclude that dissociation of CH3OOH from the 5νOH level occurs sufficiently rapidly and sufficiently above the dissociation threshold to be unaffected by collisions under atmospheric conditions. Hence, we take the dissociation quantum yield associated with the 5νOH excitation to be unity. Finally, to assess the importance of excitation arising from the tail of the CH3OOH electronic absorption between 365 and 405 nm and the 5νOH vibrational overtone band under tropospheric conditions, we have calculated the enhancement of the photolysis rate caused by inclusion of this previously neglected spectral region. As already noted, Fig. 1 shows the full absorption data for CH3OOH, including the 5νOH band. Also shown in Fig. 1, in red, is the solar flux corresponding to an altitude of 10 km and a solar zenith angle of 90° (36). We see from Fig. 1 that, although the absorption cross section for CH3OOH peaks at short wavelengths, the solar flux reaching the troposphere at these wavelengths is very much attenuated by the upper atmosphere. Hence under tropospheric conditions, even relatively weak absorptions features occurring in the near UV and visible region of the spectrum, where the solar flux is high, can contribute significantly toward the overall photolysis rate. Fig. 3 displays the enhancement in the photolysis rate arising from inclusion of the absorption data for the altitudes of 1, 5, and 10 km as a function of solar zenith angles. We find that the photolysis rate enhancement varies between ≈7% and 20% depending on the particular range of solar zenith angle considered. Decomposing this enhancement to extract the relative contribution coming from electronic absorption versus vibrational excitation, we find that dissociation through excitation of the 5νOH vibrational overtone band of CH3OOH contributes a maximum of ≈1.8% to the overall enhancement, this occurring at a solar zenith angle of ≈90°. Thus the major portion of the calculated ≈20% enhancement in the OH yield that occurs at a solar zenith angle of ≈90° is caused by excitation in the electronic absorption tail and illustrates the important influence that these weak electronic absorptions can have on tropospheric radical production.
The percent enhancement in the yield of OH radicals resulting from inclusion of the UV tail absorption between 365 and 405 nm plus the 5νOH vibrational overtone for various solar zenith angles and altitude. J(uv) is the photolysis rate arising from the previous absorption data covering 210–360 nm, and J(tail) is the photolysis rate arising from the present data involving electronic absorption between 365 and 405 nm plus the 5νOH vibrational overtone.
In summary, the present experimental findings demonstrate that excitation of weak absorption features in the region between 365 and 640 nm can photodissociate CH3OOH to give OH fragments with unit quantum yield. The relatively large concentration of CH3OOH in the troposphere, the unit quantum yield associated with the dissociation process, combined with the fact that this dissociation is initiated by the portion of the solar flux that readily penetrates into the troposphere, and hence can initiate dissociation over a broad range of solar zenith angles, makes this “new” OH source important. As CH3OOH is the prototypical organic peroxide molecule, one can expect similar photochemistry and contributions from other organic peroxides as well (37). Because OH radicals are central to determining the oxidative capacity of the atmosphere, photodissociation from these weak absorption features will need to be included in future modeling studies to more completely understand the tropospheric OH budget. On a broader note, the present measurements highlight the need to measure absorption cross sections for molecules over a wider dynamic range, especially over the wavelength regions where the solar flux is high, to fully quantitate their contributions to atmospheric photochemistry.
A.S. thanks the National Science Foundation Chemistry Division for support of this research.
↵ † To whom correspondence should be addressed. E-mail: asinha{at}ucsd.edu.
Freely available online through the PNAS open access option.
