J. Phys. Chem. 1996, 100, 3121-3125
3121
Uptake of Chlorine Dioxide by Model PSCs under Stratospheric Conditions Laura A. Brown and Veronica Vaida*,† Department of Chemistry and Biochemistry, UniVersity of Colorado, Boulder, Colorado 80309-0215
David R. Hanson Aeronomy Laboratory, National Oceanic and Atmospheric Administration, 325 Broadway, Boulder, Colorado 80303
James D. Graham and Jeffrey T. Roberts*,‡ Department of Chemistry, UniVersity of Minnesota, Minneapolis, Minnesota 55455-0431 ReceiVed: June 14, 1995; In Final Form: August 21, 1995X
The uptake of chlorine dioxide by ice at temperatures approaching those needed for PSC formation in the Antarctic stratosphere was studied. The two approaches used in this investigation involved modeling the surface coverage using kinetic parameters obtained in an ultrahigh-vacuum surface experiment and comparing these results to the uptake measured in a flow tube apparatus. For an OClO gas phase concentration of 5 × 1010 molecules/cm3, a surface coverage of ≈2 x 10-4(1 monolayers of OClO was estimated on the ice at 189 K from both methods. For an average OClO concentration in the Antarctic stratosphere of 2 x 109 molecules/ cm3, a surface coverage of 7 × 10-6(1 monolayers of OClO on PSCs is predicted.
Introduction Heterogeneous chemistry in the atmosphere has been intensively studied recently due to the role of polar stratospheric clouds (PSCs) to the formation of the ozone hole over the Antarctic. Chemical reactions occurring on PSC surfaces have been found to convert reservoir chlorine to active forms which contribute to the loss of ozone.1-3 For example, the reaction of ClONO2 with HCl on cloud surfaces has been shown to produce Cl2,1-3 which upon photolysis in the gas phase leads to catalytic ozone destruction. To quantify the effect of these reactions, numerous experiments have measured the uptake of molecules by ice surfaces under conditions that mimic the stratosphere. Uptake of HCl, ClONO2, and HOCl by both nitric acid trihydrate (NAT) and ice solids (representative of type I and type II PSCs) has been investigated.1,4-6 HCl was found to have very large surface coverages, 5 × 1014 molecules/cm2, on both types of surfaces.1 Recent in situ measurements of the HCl concentration in the Antarctic show the quantitative removal of all of the HCl from air parcels that experience temperatures low enough for type II PSCs.7 While this clearly points to heterogeneous processing of HCl, current reaction pathways may not be sufficient to explain the complete depletion of HCl.7 In this context, we have studied the uptake of chlorine dioxide, OClO, by ice surfaces to determine whether its surface coverage on type II PSCs is large enough to have an impact on the heterogeneous chemistry in the Antarctic stratosphere. Since the discovery of the ozone hole, ground based and airborne experiments have measured relatively large concentrations of OClO in the stratosphere over the Antarctic.8-10 Vertical column abundances of (1-20) × 1014 cm2 have been measured over Antarctica11 and (1-10) × 1013 cm2 over the * Authors to whom correspondence should be addressed. † Telephone: (303) 492-8605, fax (303) 492-5894, e-mail vaida@ spot.colorado.edu. ‡ Telephone: (612) 625-2363, fax (612) 626-7541, e-mail roberts@ chemsun.chem.umn.edu. X Abstract published in AdVance ACS Abstracts, January 15, 1996.
0022-3654/96/20100-3121$12.00/0
Arctic12 as compared to an upper limit of ≈5 × 1012 cm2 over Fritz Peak.13 These large concentrations of OClO are directly related to the increased activity of chlorine chemistry in the Antarctic stratosphere. The primary loss mechanism for polar stratospheric ozone is photolysis of the ClO dimer, ClOOCl, which leads to Cl + ClOO.14-16 The asymmetric isomer of chlorine dioxide, ClOO, produced in this reaction quickly falls apart to Cl and O2. The chlorine produced from dimer photolysis reacts with ozone, regenerating chlorine monoxide. Additionally, ClO can react with BrO to produce either OClO, ClOO, or BrCl.17 The ClOO and BrCl produced in this reaction are responsible for 25% of the Antarctic ozone loss.14 As the reaction between ClO and BrO is the only known source for OClO in the atmosphere, the concentration of OClO has been used to estimate the extent of chlorine activation in the polar stratosphere.9,18,19 A complete knowledge of the fate of OClO in the atmosphere is thus important to understanding the chlorine cycles that destroy ozone. The first step in evaluating the atmospheric impact of OClO condensed on ice is to determine the surface coverages that could occur in the stratosphere. The previous paper in this series describes the ultrahigh-vacuum (UHV) experiments that were conducted to determine the sticking coefficient and desorption rate parameters of chlorine dioxide to crystalline and amorphous ice at 100 K. This paper will use these parameters to calculate an equilibrium OClO surface coverage at the temperatures and partial pressures characteristic of the Antarctic spring stratosphere. To establish that the UHV results can be extended and applied to type II PSCs over Antarctica, the OClO uptake by ice was measured in a flow tube under conditions that approach those in the stratosphere. An upper limit to the surface coverage of OClO on type II PSCs was obtained from these experiments and compared to the results of a kinetic model. Experimental Section The uptake experiments described below were conducted in a neutral flow tube apparatus equipped with a chemical ionization mass spectrometer. The experiment has been de© 1996 American Chemical Society
3122 J. Phys. Chem., Vol. 100, No. 8, 1996 scribed in detail elsewhere,1,5 and only a brief summary is given below. The double-jacketed glass flow tube was cooled to temperatures of 189 and 200 K during the course of the experiments. The temperature of the cooling fluid was monitored at both the entrance and exit to the flow tube (∆T < 1 K) and was taken to be the temperature of the ice films. A uniform polycrystalline ice layer was formed by depositing water vapor on the surface of the cooled flow tube. The H2O/He flow used to form the ice substrate was introduced through a sliding injector, allowing the water to be distributed evenly along the entire length of the flow tube. An ice layer of approximately 27 µm was deposited for the study. The ice surface area was assumed to be equal to the geometric area of the inner wall of the cylindrical flow tube. This is only an assumption as the porosity and therefore the surface area of the thick ice film used for the study were not investigated. The OClO was introduced through a separate double-walled movable injector at the upstream end of the flow tube. The injector was maintained at a slightly higher temperature than the flow tube (∼20 K) which decreased the extent of OClO sticking to the inside. The movable injector was used to control the time that the ice surface was exposed to the OClO flow. Chlorine dioxide was produced by flowing a 0.6% mixture of Cl2 in helium through a reaction vessel containing sodium chlorite. Care was taken to prevent a large buildup of OClO, as it is very explosive. The continuous flow of OClO was used directly without further purification. Analysis of the gas mixture using the mass spectrometer showed OClO as the only product with no indication of unreacted Cl2. Chlorine dioxide was detected using chemical ionization mass spectrometry, CIMS. The SF6- ion was found to undergo a charge transfer reaction with chlorine dioxide to produce the OClO- parent ion at mass 67. Due to the delicate ionization method employed, very little fragmentation of the OClO- was observed. To compare with the surface coverages measured in the flow tube experiment, the desorption kinetics and sticking coefficient of OClO on ice surfaces at 100 K were used to calculate an equilibrium coverage. The companion paper of Graham et al.20 describes the UHV methods used to measure these parameters, and a short description follows. The desorption kinetics of OClO from the ice surface were determined using temperatureprogrammed desorption. Chlorine dioxide was found to exhibit first-order desorption kinetics from amorphous and crystalline ice. By varying the heating rate of the ice sample, the Arrhenius preexponential and the activation energy for desorption were determined. The absolute sticking coefficient, S, of OClO on amorphous and crystalline ice was measured using a variation of the King and Wells method.21 The sticking coefficient is defined as the probability that an OClO molecule sticks to the ice after a single collision. In contrast, the net uptake by the surface as measured in the flow tube experiments is the result of OClO having several collisions with the ice surface. In both experiments, the OClO mass signal was monitored before and after exposure to an ice surface with any change being attributed to adsorption. An important difference between the flow tube and UHV experiments is that the UHV experiments were performed at 100 K compared to 189 K for the flow tube studies. The OClO desorption rate at 189 K is significant, whereas it is negligible at 100 K. Neither experiment is currently capable of bridging this temperature gap. The cooling bath of the flow tube experiment cannot go below 189 K, while the ice multilayer in the UHV experiment begins to desorb at temperatures above 150 K. Additionally, for systems with low uptake efficiency,
Brown et al.
Figure 1. OCLO uptake experiment at 189 K and [OClO]0 ) 5 × 1010 molecules/cm3. OClO exposure to the ice is initiated at point A and terminated at point B.
the flow tube is really a thermodynamic rather then kinetic experiment. We show below that the low-temperature, lowpressure UHV kinetic results can be extrapolated to the conditions found in the flow tube and consequently in the polar stratosphere. Results The surface coverage of OClO was determined experimentally by measuring the number of molecules lost to the ice surface. The gas phase concentration of OClO was monitored as the injector was pulled back to expose the ice surface to OClO. The number of molecules adsorbed on the ice was estimated from the change in OClO concentration as a function of time, indicated by the drop in OClO mass signal upon exposure to ice. The surface coverage, Γ (molecules/cm2), is then given by the following equation, where ∆n is the change in gas phase concentration upon exposure to ice for a length of time t, and Vf (cm3/s) is the volume flow rate at the temperature and pressure in the flow tube. ∑ is the surface area of the ice.
Γ ) (Vf∆nt)/Σ
(1)
For these experiments, the change in gas phase concentration, ∆n, was calculated from the difference in the OClO mass signal before and after exposure to ice. The time of exposure, t, was long enough such that a steady state between the adsorbed and gas phase OClO was achieved. This technique has been used previously to study the physical adsorption and uptake of HCl, HOCl, HBr, and HF on ice surfaces.1,5 An uptake experiment with OClO at 189 K is shown in Figure 1. The fluctuation in the signal, evident in the figure, is statistical and is the result of using a small flow of OClO straight from the reactor. Although the signal fluctuation is a large source of error for small uptakes, an upper limit to the uptake of OClO by the ice can be derived. The gas phase concentration of OClO was approximately 5 × 1010(1 molecules/cm3. The error in calculating the gas phase concentration OClO is discussed below. At point A, the injector was pulled out, exposing 12 cm of the ice surface to the OClO flow. The OClO signal decreases by ≈10% immediately following exposure to the ice. Within 2 s the OClO signal has recovered to its initial value. The 2 s decrease in the OClO signal represents the total uptake by the ice. The initial uptake where the kinetic information lies could not be resolved using this technique. At point B in Figure 1, the injector was moved back into the flow tube, no longer exposing the ice to OClO. Desorption of OClO from the ice surface at this point is not evident due to the noise in the signal. OClO uptake at 200 K was also studied, however, no decreases in the signal beyond statistical
Uptake of Chlorine Dioxide by Model PSCs fluctuations were observed. From the data in Figure 1, an upper limit to the surface coverage of OClO was calculated to be 3.3 × 1011(1 molecules/cm2 ([OClO]0 ) 5 × 1010(1 molecules/ cm3). A number of experiments at 189 K with an initial OClO concentration of 5 x 1010(1 molecules/cm3 were performed with an average surface coverage of (2 ( 1) × 1011(1 molecules/ cm2. The OClO signal consistently dropped between 0.3% and 12% upon exposure to ice. The smallest drop of 0.3% was not used in calculating the average as it was not reliably above the noise. While uptake experiments at 189 K always had a decrease in the signal upon exposure to ice, the signal did not increase when exposure was terminated, indicating that the OClO did not desorb. The lack of observation of desorption is discussed further below. There are several sources of uncertainty in the experiment which contribute to the uncertainty in the surface coverage. First, the absolute OClO gas phase concentration is needed to calculate an absolute surface coverage.1 The OClO concentration can be obtained from the CIMS sensitivity or calibration factor which can be estimated from the reaction rate coefficient for SF6- + OClO f OClO- + SF6. This rate coefficient has not been measured; however, it is likely to be in the range of 1.1 × 10-9 to 1.1 × 10-10 cm3 molecule-1 s-1 (the rate coefficients for SF6- charge transfer to O3 and Cl2, respectively).22 A rate coefficient of 3 × 10-10 cm3 molecule-1 s-1 was chosen for OClO. The error introduced by assuming a value for the CIMS sensitivity factor is acceptable for this study since the OClO coverage need only be known to within an order of magnitude for qualitative evaluation of the heterogeneous capabilities of OClO. In addition, the absolute concentration is not necessary for a comparison of the two techniques as surface coverage varies similarly in concentration for both. This can be seen by comparing eqs 1 and 3, which both show a linear dependence on OClO concentration. Second, the lack of observation of any desorption of the OClO calls into question what was actually observed in the adsorption runs: was the OClO adsorbed on the surface, did it dissolve into the ice, or was there some experimental artifact masking the desorption? The uptake was too low, however, preventing a useful discussion of this problem. Note that in the preceding paper it was shown that OClO is incorporated into the bulk of the ice at 100 K and remained until the ice itself desorbed. This may be the situation in the flow tube as well although there was no direct evidence for incorporation into the bulk. The inability to observe the expected desorption indicates a potential limitation with this experimental technique in studying species that have low surface coverages (surface coverage < 10-4 monolayer). While the sources of error discussed above could be reduced with further experiments, the very small uptakes observed did not warrant further investigation. Despite the uncertainty, these experiments give a qualitative idea for the surface coverage of OClO on ice surfaces at temperatures present in the stratosphere. To compare with this experimentally determined coverage, the desorption kinetics and sticking probabilities presented in the previous paper are used to calculate an equilibrium surface coverage. Graham et al.20 reported that the sticking coefficients of OClO on amorphous and crystalline ice at 100 K are 0.8 ( 0.2 and 0.6 ( 0.2, respectively. The sticking coefficient of OClO to crystalline ice at 100 K is used in the following calculations with the assumption that S is a weak function of temperature.23 The desorption kinetics of OClO adsorbed on the ice surface were measured using temperature programmed desorption and were found to follow first order kinetics. The activation energy for desorption is 23 ( 1 kJ/mol with a preexponential of 2 × 109(1 s-1.
J. Phys. Chem., Vol. 100, No. 8, 1996 3123 The desorption kinetics of the surface state of OClO on ice were used to calculate a steady-state coverage in the flow tube. The bulk state of OClO observed in the UHV experiments20 was not included due to the very low surface coverages measured in the flow tube experiment. A much larger uptake would be expected if the OClO was becoming incorporated into the ice bulk. Additionally, there was no direct evidance to support the presence of OClO in the ice at the concentrations of OClO used in the flow tube experiments. The residence time of OClO on the ice surface can be calculated from the desorption rate expression
τ)
() ( ) Ed 1 exp A RT
(2)
where τ is the residence time, A is the Arrhenius preexponential, and Ed is the activation energy for desorption. At 189 K, the residence time of OClO on the ice surface is 1 × 10-3(1 s. The adsorption rate of OClO molecules to the ice surface is also needed to calculate an equilibrium surface coverage. Assuming a gas kinetic rate, the adsorption rate of OClO to the ice surface is given by the equation
adsorption rate )
1 8kT 1/2 S[OClO]0 4 πM
( )
(3)
The sticking coefficient for chlorine dioxide to crystalline ice, S ) 0.6 ( 0.2, is used to calculate an adsorption rate that would apply to the uptake experiments described above. Using an OClO concentration of 5 × 1010 molecules/cm3 the adsorption rate at 189 K is 1.8 × 1014 molecules/(cm2 s). Assuming that the only loss channel of OClO from the surface is desorption, the steady state surface coverage can be calculated as the adsorption rate multiplied by the residence time. For the small surface coverages estimated here, this analysis is identical to the Langmuir model for surface adsorption.24 For the temperatures and pressures in the flow tube uptake experiments, 189 K and [OClO] ) 5 × 1010 molecules/cm3, the calculated OClO surface coverage is 1.8 × 1011(1 molecules/cm2. Although both techniques had errors on the order of a factor of 10, the calculated coverage is in excellent agreement with the uptakes measured in the flow tube. The good agreement supports the assumption that only OClO adsorbed to the ice surface need be considered with this simple kinetic model. Since the above calculations agree very well with the surface coverages measured by experiment, the UHV measurements can be used to calculate the surface coverage at conditions not easily accessible in the flow tube. For example, the concentration of OClO used in the uptake experiments is higher then the ambient OClO concentration found in the stratosphere. The desorption kinetics derived from the UHV experiments can be used to calculate a surface coverage at the more typical OClO nighttime concentration of 2 × 109 molecules/cm3.25 At 189 K under these conditions, the steady state surface coverage on crystalline ice would be 7 × 109(1 molecules/cm2. Due to the efficient photolysis of OClO in the stratosphere, the OClO mixing ratio shows a very pronounced diurnal cycle.25 The decrease in the OClO concentration by 2 orders of magnitude during the day would subsequently reduce the impingement rate of OClO on ice surfaces, resulting in a possible surface coverage of 7 × 107(1 molecules/cm2 at these times. Discussion Heterogeneous reactions occurring on polar stratospheric clouds convert unreactive, reservoir chlorine molecules like HCl and ClONO2 into reactive Cl2 via reactions which proceed at
3124 J. Phys. Chem., Vol. 100, No. 8, 1996 negligible rates in the gas phase.1-3 The impact of these reactions on the atmosphere underscores the importance of understanding the properties and chemistry of PSCs. A number of atmospheric measurements26,27 and modeling28,29 and laboratory studies30-32 have been undertaken for this purpose. Laboratory studies of OClO afford the opportunity to probe chemical properties of ice surfaces. The lack of dissociation or chemical reaction on the clean ice surface allows for the treatment of the OClO adsorption with a simple kinetic model. By comparison, the desorption kinetics of HCl from ice surfaces were measured under conditions similar to those for OClO.30,33 Using the UHV kinetics for HCl, a surface coverage of 10-6 of a monolayer (ML) is predicted on ice at ∼190 K,30 well below the 0.5 ML measured in flow tube experiments.1,2 The discrepancy is thought to be the result of the dissociation of HCl on ice under the conditions present in the flow tube but not in the UHV experiment. Recent molecular dynamics calculations show that it is energetically feasible for HCl to dissociate once it has become solvated by at least one water bilayer.34,35 Due to the dynamic nature of the ice, with desorption and adsorption rates varying between 10 and 1000 ML/s in the polar stratosphere,31 HCl could become solvated with water molecules. The difficulties that dissociation introduces in modeling the surface coverage of HCl on the surface of PSCs are not present in the case of OClO. Since OClO is stable in the dark on the ice surface at both 100 and 189 K, it is an excellent test case for extending the information obtained in UHV experiments to conditions present in the atmosphere. Extension of this technique to a general case, however, would require additional experiments and calculations. For OClO, the comparison between the flow tube studies and the kinetic calculations show that indeed the UHV data can be used to accurately predict the surface coverages of OClO on ice surfaces in the Antarctic spring. The motivation for this study was to investigate the impact that OClO on ice may have in the stratosphere. The implications of chlorine dioxide on polar stratospheric clouds include transport of chlorine, photochemical ozone depletion, and possible reactions with other trace constituents present on the ice. The photochemistry of OClO in the gas phase has been extensively studied.36-38 The predominant photodissociation channel produces ClO and O, which while recycling the ClO for further reaction does not contribute to catalytic ozone loss. A minor dissociation channel leading to Cl and O2 has been studied, but measured quantum yields are low,
