Reactive Uptake of ClONO2 onto Sulfuric Acid Due ... - ACS Publications

May 1, 1994 - to predict reaction probabilities for C 1 0 N 0 2 and HOCl on aerosol particles due to ... heterogeneous chemistry of sulfuric acid part...
0 downloads 0 Views 1MB Size
J . Phys. Chem. 1994, 98, 5728-5135

5728

Reactive Uptake of ClONO2 onto Sulfuric Acid Due to Reaction with HCl and H2O David R. Hanson' and A. R. Ravishankarat NOAA, Aeronomy Laboratory, 325 Broadway, Boulder, Colorado 80303, and Cooperative Institute for Research in Environmental Sciences, University of Colorado, Boulder, Colorado Received: December 21. 1993; In Final Form: March 25, 1994"

The uptake of ClONO2 onto sulfuric acid solutions and HC1-doped sulfuric acid solutions has been measured over temperature and composition ranges relevant to the high-latitude lower stratosphere. The results reveal that the reaction of ClONO2 with H C l in/on sulfuric acid is an important source of active chlorine for the stratosphere. It is suggested that CION02 uptake due to reaction with HC1 is dependent on both bulk and surface concentrations of H C l and that, for present levels of HCl in the stratosphere, the surface-dependent uptake could be a significant contributor to the conversion of CION02 and H C l to Cl2 a t temperatures less than 195 K, if type I PSCs are not formed a t their threshold temperatures. The hydrolysis of ClONO2 and the y for uptake in the absence of H C l have been shown to be dependent on the activity of water. The results suggest a larger reactive uptake a t temperatures near 200 K than has been used previously. The parameters obtained from these uptake measurements will lead to better representations of the CION02 H2O and CION02 HC1 reactions in atmospheric models.

+

+

Introduction The hypothesis that heterogeneous reactions involving chlorine reservoir species leads to enhanced levels of active chlorine has been corroborated by field observations and laboratory experiments.' While these heterogeneous processes are generally recognized to be particularly efficient on the solids that can form in the polar ~tratosphere,~.~ in-depth analysis of these processes in/on liquid sulfuric acid droplets has only recently begun. The sulfate aerosol (the primary form of S(V1) in these droplets is bisulfate, [HS04-], rather than sulfate) in the lower stratosphere is composed of liquid sulfuric acid droplets of -0.l-pm radius under nonvolcanic conditions. Only a few of the chemical reactions in the aerosol particles have been included in atmospheric models. For example, the hydrolysis of ClONO2 and N205 in sulfate aerosol

-

+ H 2 0 HOCl + HNO, N20, + H 2 0 2 H N 0 ,

CIONO,

-

(1) (2)

has been included in atmospheric models by Rodriguez et al.4 and Granier and B r a s ~ e u r .Unlike ~ reactions 1 and 2, which involve HzO already present in the particles, the reactions of HCl requires transfer of both reactants from the gas phase into the liquid. Therefore, incorporation of these reactions ClONO,

+ HCl

HOCl

+ HCl

-

-

C1,

+ HNO,

(3)

C1,

+ H20

(4)

intoatmospheric models is not straightforward. We have recently developed a framework, based on laboratory measured quantities, to predict reaction probabilities for C10N02and HOCl on aerosol particles due to reactions 3 and 4,and hence these reactions can now be included in numerical models.6 For Pinatubo-type conditions, high levels of active chlorine were predicted in a 2-D model that included this framework, and reactions involving HCl were shown to be significant under certain conditions. Even the role of HC1 in direct chlorine activation on background aerosol

* To whom correspondence should be addressed.

t Also associated with Department of Chemistry and Biochemistry,

University of Colorado, Boulder, CO. 0 Abstract published in Advance ACS Abstracts, May 1, 1994.

a t high latitudes in wintertime could not be ignored. We also identified a number of chemical and physical parameters that need to be determined more accurately for a better treatment of heterogeneous chemistry of sulfuric acid particles in atmospheric models. The parameters that determine the uptake of CION02 were some of the most important in that list. By way of background and to show the importance of knowing these quantities, we present the equation for calculating the reaction probability (or uptake coefficient) developed by Hanson et a1.6 The net uptake coefficient for the atmosphere, 7,for a particular process can be calculated from

- = -l +

w

1

4HRfif(lla)

(1)

The physicochemical parameters include the solubility (H), the diffusion coefficient ( D J ,and the first-order loss rate coefficient in liquid sulfuric acid (kl). k1is actually the sum of all first-order loss rate coefficients. R is the gas constant, and w is the mean molecular speed. The functionf ( l / a ) ,equal to [coth(a/l) - I / a ] , takes into account the spherical geometry of the stratospheric aerosol and its small size (compared to the bulk material over which most laboratory measurements are carried out.) a is the droplet radius, and the reacto-diffusive length, 1, is equal to (DJ The mass accommodation coefficient, a,can affect the rate of a heterogeneous process; its value, however, is close to unity, and a does not normally affect the value of the uptake coefficient. This equation can be used for heterogeneous reactions, in sulfuric acid droplets, of species originating in the gas phase, provided the important quantities are known or can be estimated. In eq I, the concentration of one of the reactants is assumed to be constant in the droplet, which is the case when the solubility of that species is high and its uptake can be much faster than that of the second reactant. The reactions of HCl with CION02 and HOCl are two of the most important reactions to be treated using this relation? Since HC1 undergoes near complete dissociation in all but the most concentrated (2-75 wt %) sulfuric acid solutions,7 its reactive form is likely to be C1-. In this paper, we do not distinguish between HCl and C1- in solution, and we often refer to the HCl in the liquid (i.e.,C1 in the -I oxidation state) as either C1- or HC1. We have recently determined values for the solubilities of HOCl and HC1, the diffusion coefficient for HC1 in a 50 wt % H2SO4

0022-365419412098-5728%04.50/0 0 1994 American Chemical Society

The Journal of Physical Chemistry, Vol. 98, No. 22, 1994 5729

Uptake of CION02 onto Sulfuric Acid

Injector To solution, and the valueof the second-order rate coefficient for the / CIMS / reaction of dissolved HOCl with dissolved HCl in 60 wt % SIDE VIEW / solutions.8 We believe these are some of the better determined / / i " quantities for a heterogeneous reaction occurring in sulfuric acid, and, using these values, reaction 4 can be treated accurately in atmospheric models. Reactive uptake of C10N02 due to reaction 1 in sulfuric acid 1 /' solutions has been reported by Tolbert et al.9 and by us.lo There I Flow tube liquid I is fairly good agreement (within a factor of 2) between the two Movable/Rotatable studies for 65-75% acid. The CION02 + HC1 reaction was also glass rod studied over sulfuric acid by both researchers, and no significant Figure 1. Schematicdrawingof the flow tubeused tomeasurethevarious increase in the uptake of ClONOz was observed on solutions with parameters discussed here and rotatable/movableglass rod used for the 165% acid content, suggesting that this reaction was not very application of liquid sulfuric acid. important in the atmosphere. The observed loss of HCl and the generation of Clz product by Tolbert et al. may have been due In this paper we present a reinvestigation of ClONO2 uptake to the reaction of HOCl (produced from ClONO2 Hz0) with onto sulfuric acid solutions with H2S04 content ranging from HCl. In addition, we studied CION02 uptake onto 40 and 60% 46.6 to 65 wt %. Uptake was studied onto neat sulfuric acid acid solutions at -21 5 K and reported values for y for ClONO2 solutions and HC1-doped sulfuric acid solutions. The analytical + H2O and estimated values for ClONO2 HC1. We stated methods for extracting the reaction rate coefficients are described, that ClONO2 + HCl was not likely to be important even in 40and the results of these measurements are discussed in terms of 60% acids based on the HC1 solubilities available at that time. H a n d k*I. From thevariation of reaction probability versus HCl The values of diffusion coefficients and HCl solubilities used in content, the ratio of the first-order rate coefficients for ClONO2 that paper are now known to be in error by factors of 100-1000, loss due to reaction with HCl to that with HzO was estimated. making our earlier claim invalid (see ref 8 and references therein); Using an estimated second-order rate coefficient for the reaction these reactions are now recognized to be much more important of CION02 with HCl, kIIHc1,values for the first-order loss rate in the atmosphere.6 Therefore, a reinvestigation of the reactive coefficient due to hydrolysis, klo, were also obtained. We derive uptake of ClONOz is necessary to place the knowledgeof reactions an estimate for the physical solubility of ClON02, H,based on 1 and 3 on a firmer footing. the measured uptake coefficients and the estimated k1I and Dl. Both HOCl and HCl readily dissolve in H2SO4 solutions, and Finally, we present a table that contains a detailed list of equations they do not irreversibly react with the constituents of the solution. and the values of the parameters to be used in the calculation of Therefore, the solubilities of these species can be directly the uptake of ClONO2 due to reactions 1 and 3 in atmospheric determined without interference from reactive losses. (Note that models. the self-reaction of HOCl to give ClzO and H20 is not likely to Experiment be fast enough to affect laboratory experiments. In our experiments with HOCl gas-phase concentrations C10'2 ~ m - 3 we , ~ saw The experimental apparatus and technique for the uptake no evidence for the occurrence of this reaction. It may occur, measurements are essentially the same as in our previous work. however, if very high HOCl concentrations are present, and this A chemical ionization mass spectrometer (CIMS) monitored the may explain the observation of C120 product by Tolbert et ~ 1 . ~ ) concentration of a gas-phase molecule as it was exposed to a In the case of HCl, the diffusion coefficient can be estimated by liquid residing on the walls of a cylindrical flow tube. Only a comparing the results of direct solubility measurements with those brief discussion of the experiment necessary to understand the from measurements of the time-dependent uptake due to physical current measurements is given here. More details of the solubility.8 In principle, the diffusion coefficient for HOCl could experimental apparatus and techniques can be found in refs 8 also be determined in this way. and 11. In stark contrast to the HOCl and HCl measurements, direct As in our previous work with the HOCl and HC1 molecules,8 determinations of the solubility and diffusivity of CION02 are uptake of a gas-phase molecule onto a quiescent liquid surface not possible. This is because ClONO2 rapidly reacts with water was studied. The liquids were applied onto the inner wall of a in sulfuric acid solutions, making the separation of reactive loss horizontal, cylindrical flow tube (i.d. of 2.2 cm) using a glass rod from physical removal difficult. Solubility and diffusivity can be that extended into the flow tube from a port at the downstream estimated based on analogy to similar molecules, and a combinaend of the flow tube. A schematic diagram of the flow tube and tion of parameters, such as H,DI, and/or k1, can be measured. glass applicator is shown in Figure 1. The amount of liquid in (Liquid-phase kinetics techniques could be employed, perhaps, the flow tube was approximately 1 cm3,and the total surfacearea to directly obtain the hydrolysis rate coefficient.) Because of of the coated flow tube wall (i.e., the liquid) was -40 cm2. The this limitation, the parameters needed to quantify the reactive glass rod could be manipulated from outside the flow tube via a uptake of ClONO2 and to correct the laboratory measurements seal so that the liquid could be reapplied during a run. Since the to the atmosphere are not well defined. Furthermore, the reactive liquid flowed very slowly and gathered at the bottom of the flow uptake of C10N02 when HCl is present will be due to reaction tube, it was necessary to reapply the acid periodically (-3060 with both H20 and HC1. Treatment of these simultaneous min). The thickness of the liquid over most of the flow tube wall reactions in the liquid phase is not as simple as in the gas phase was much less than the average thickness of -0.25 mm, primarily where the rates are additive (see ref 6). because the liquid slowly flowed down and also because it was difficult to uniformly coat the wall. Glass wall that was not the treatment In the framework developed by Hanson et covered with liquid was easily detected visually, and in this case, of ClONO2 uptake onto sulfuric acid requires values for the the liquid was reapplied. At no time during an uptake measuresecond-order rate coefficient for reaction of ClONO2 with ment did the presence of uncoated glass surface area exceed a dissolved HCl, klIHCI, and for the rate coefficient for CION02 few percent of the liquid surface area. The results showed no hydrolysis, either k$ or klIo. Knowledge of these quantities as dependence on time, indicating that the changing liquid thickness, a function of acid content and temperature is essential for a due to the slow flow, did not affect the results. Therefore, the complete treatment. Once these quantities are known, reactive liquid was everywhere thick enough to be considered a thick uptake of C10N02 onto sulfuric acid aerosols can be treated substrate. The first-order rate constants (in the liquid) encounmore accurately in atmospheric models.

I

+

+

5730

The Journal of Physical Chemistry, Vol. 98, No. 22, 1994

tered in the course of this work were > 10SKI.The reacto-diffusive length is 0.3 pm for a species with this loss rate and a D, = 10-8 cm2 s-1, suggesting that our liquids would have to become very thin to affect the measured uptake due to size limitations. The ability of the liquid to ”wet” the flow tube wall was greatly improved by pretreating the glass surface with an HF/HNO3/ H2O solution. A cotton swab soaked with a few drops of this solution was used to treat a -50-cm2 area of glass surface. The flow tube was then ”rinsed” a few times by wiping it with a cotton swab soaked with distilled water. Visible pieces of dust or lint were removed with a foam-tipped swab. After the inner surface was dried with a flow of He, the flow tube wall was prerinsed with -1 cm3 of the sulfuric acid solution under study. The rinse solution was discarded, and another 1-cm3sample of the solution was placed inside the flow tube and cooled to -200 K. During an uptake measurement, the glass applicator rod remained in the measurement region. Its cross-sectional area was about 0.1 5 cm2, and its surface area was negligible compared to the surface area of the sulfuric acid-coated inner wall of the flow tube. The decrease in the cross-sectional area of the flow tube and the concomitant increase in the average flow velocity were taken into account. We used the standard analysis for cylindrical flow tubes in obtaining a value for y.I2J3 A sulfuric acid solution of -60 wt %was prepared by diluting -97% H2SO4 with distilled water while solutions of 46.6-65 wt % were prepared by mixing the 60% solution with either water or 97% acid. The HCl was doped into the solutions by two methods. The first approach was to allow the surface of the acid (and liquid next to the surface) to equilibrate with HC1 vapor introduced with the carrier gas, and the second consisted of mixing an HCl solution with the bulk sulfuric acid solutions. The latter method was used for 46.6%and 5 1% acid solutions where a small amount of the solution containing a known amount of HCl(1 to 0.1 M) was mixed with neat acid, as described in Hanson and Ravishankarae8 For the 5 1% acid and greater solutions, the HCl doping was accomplished by introducing HCl into the carrier gas and allowing the surface of the sulfuric acid to come into equilibrium. The HCl content a t the surface was then calculated from our measured values for the solubility of HC1, H * ~ a . This 8 method requires accurate knowledge of the gas-phase partial pressure of HC1, and the CIMS detector was calibrated for HC1 by measuring the flow rate of a known mixture of HCl in He; the calibration procedure (the dP/dt method) is described in detail in ref 8. Both methods were used for doping 51% solutions with HCl, and the results were comparable (see Figure 4), indicating that these two methods are equivalent. The H2SO4 content of the solutions was determined after the uptake measurements were performed using standard acid-base titrations. The acid content of the solutions was usually not significantly different from that of the initial solutions, within the uncertainty of the titrations (f0.3 wt %). Water vapor was added to the He carrier gas via a calibrated leak valve to reduce the loss of water from the surface of the H2S04 solutions. There was a detectable level of HC1 in the carrier gas when the “pure” sulfuric acid solutions were studied, indicating the presence of a few ppmv of HC1 in the stock -97% sulfuric acid. As discussed below, these small amounts of HCl affected the value of the uptake coefficient for CION02 due to hydrolysis by 110%. The reactants ClONOz and HCl and products C12 and HOCl weredetected with the CIMS using the SF6- reactant ion. Precise quantitative determination of the product yields for HOCl and Cl2 was not possible because HOCl can react with HCl that may be present. The reaction probability for HOCl + HCl is particularly efficient on thick liquids in the laboratory. Only for the lowest amounts of HC1 present were we able to detect product HOC1. Most measurements were carried out at temperatures between 200 and 203 K, although some were performed a t higher

Hanson and Ravishankara

1

.cMn C

Inj. Pos. c m Figure 2. CION02 signal plotted us injector position as it was exposed to 46.6% H2.30, solutions containing differing amounts of HCI. The [HCI], were M over that when no HCl was added implies that CIONOz can directly react with Cl-. This reaction may occur in two steps, Le., reaction 1 followed by reaction 4. This pathway, however, can be shown to have a negligible effect on the overall uptake coefficient for ClONOz based on the previously measured rate coefficient for reaction 4 in sulfuric acid solutions and the very large enhancement in CION02 uptake in the presence of HCl. It is also possible that CION02 uptake is limited by an equilibrium in solution ClONO,

+ H,O F? HOCl + HNO,

(5)

Even if this were the case, we could exclude a two-step mechanism again based on the low rate coefficient for reaction 4. When HC1 was present, virtually all the ClONO2 taken up reacted directly

Uptake of ClONO2 onto Sulfuric Acid

The Journal of Physical Chemistry, Vol. 98, No. 22, 1994 5731 TABLE 1: Parameters for Individual Fits of the Data to Eg 11

lo3

t

I

C

.-

LC

i

198

10‘

1

2

3

4

2.v 5

Inj. P o s c m

Figure 3. Same as Figure 2,except [HCI],, = 10” M was held constant and temperature was varied. The ClONO2 signal for the measurements at 203 and 198 K were divided by factors of 2 and 4,respectively, for clarity.

with dissolved HC1 in the [HCl] 2 M cases; very little was hydrolyzed to produce HOCl. Note that in our bulk liquids even the small amounts of HOCl produced would react quickly with HCl (Le., the two-step mechanism discussed above). This may not be necessarily true for sulfate aerosol particles. The uptake of ClONO2 onto a 46.6% H2SO4 solution doped with 1 X M HCl was studied at 198,203, and 208 K. The ClONO2 signals for these measurements are plotted against injector position in Figure 3. It is clear that the uptake coefficient did not change significantly over this temperature range; its value remained at -0.1. For typical stratospheric conditions, Le., for mixing ratios of H20 and HCl equal to 5 ppmv and 1 ppbv, respectively, and a pressure of 50 mbar, the supercooled sulfate aerosol would contain 46.6 wt % acid a t -192 K and would be expected to contain 1 X 10-3 M HC1 (the same as the solution discussed here). Because there were no large changes in the ClONO2 uptake as the temperature was lowered from 208 to 198 K (at constant [ HCI],,), the temperature d.ependency of the terms in eq I, Le., H, Dj, k”, and 4RT/w, must “work together” to cancel, or render small, the temperature dependence of the overall CION02 uptake coefficient from 208 to 198 K. The uptake onto 46.6% H2SO4 aerosols at 192 K in the stratosphere would be expected to be close to 0.1, assuming that y is independent of temperature from 198 to 192 K. The independence of y with temperature over this range is interesting. One possible explanation is that the factor of 3 decrease in DI at a 10 K lower temperature,*J6J7along with the changes in k l 1 ~ cand l the T/w (proportional to dr)terms, is counterbalanced by an increase in H for CION02 at the lower temperature. (Note that the HCl and H2SO4 contents of the solutions were held constant as the temperature was varied.) Such an increase of a factor of 2-3 in H from 208 to 198 K is possible if the enthalpy of solvation for ClON02 is -6 to -9 kcal mol-’. This value for the enthalpy of solvation is reasonable if the interaction of ClONOz with an HzO-containing liquid involves hydrogen bonding. Note that our suggestion of a coincidental cancellation is one explanation. There may be a more profound reason for this temperature independence, but we are not aware of (nor can we speculate on) such an explanation. CION02 uptake was studied as a function of HCl concentration or, equivalently, gas-phase partial pressure, p(HCl), over 46.6, 51.0, 55.6, 57.5, 58.5, 59.8, and 65.0 wt % H2S04 solutions at 202 f 1 K. Note that HCl was added to solutions of 251% H2SO4 from the gas phase. Even when HC1-doped solutions were prepared by mixing known amounts of HCl into the bulk sulfuric acid, we still present the y as functions of the measured p(HC1). This makes application to the atmosphere easier and also enables a single representation. The ClONO2 uptake coefficient data are shown in Figure 4 as a plot of y us the measured

-

46.6 202-203 1 X lo7 51 202 2 X lo6 55.6 202 3.7 x 105 51.5 202 2.0X lo5 58.5 202 1.6 X lo5 59.6 202 9.6X lo4 65 203 1.8X lo4

0.27 0.19 0.12 0.093 0.08 0.068 0.031

0.038 0.022 0.012 0.0061 0.0051 0.0041 0.00093

0.035 0.02 0.011 0.0058 0.0049 0.0036 0.00088

1 X 104 2.7 X lo4 4.3 x 104 8.6 X lo4 3.5 X lo4 5.3 X lo4 8 X 104

The reaction probabilityfor CION02 measured at the lowestp(HC1) possible. Reaction probability for CION02 + H20 obtainedby applying eq 11 to

YlOWeSt.

partial pressures of HCl. The filled and open symbols are results using the two methods (either mixing bulk solutions or from the gas phase, respectively) for doping the solutions with HCl. It is clear from the figure that the uptake of ClONOl is due to reaction with both HC1 and H20; at low p(HC1) the value of y is independent of p(HC1). These values are close to the previously reported values9J0of ClON02 uptake due to hydrolysis in pure sulfuric acid solutions. It is also evident that the uptake does not vary linearly with the HCl concentration, Le., proportional top(HCl), and thus uptake due to hydrolysis and uptake due to reaction with HCl are not “additive”. This is to be expected when two reactions are competing in the liquid phase, as shown in our previous paper.6 As shown there, the overall uptake coefficient can be expressed as YCIONO,

7 + =YOU1 +

= 70

kHCI/kO

rH*HC!dHC1)

(11) where ki and kkc, are the first-order loss rate coefficients for reaction with water and dissolved HCl, respectively, yo = (4RT/ w)H(DjkI~)~/~is theuptakecoefficient for ClON02+ HzO without HCl present, and r is equal to k i c l / k i . The curves in the figure are least-squares fits of the data to eq I1 performed with a weight equal to (1/yCION02)2. ki is, of course, independent of p(HCl), and kkcl is equal to k~,,H*,,,p(HCl). The fitted values for the ratio r = kEcl/ki are shown in Table 1. Values of H*Ha needed to extract values for r for each solution were taken from ref 8 and are also included in Table 1. The activity of water, aH,O, is also shown in Table 1 and is equal to the vapor pressure of H2O of the solution at 202 Kl*J9 divided by the vapor pressure of supercooled water at 202 K (equal to 4.45 X le3 mbar).20 When the measured uptake coefficient is 20.1, it may be affected by the value of the mass accommodation coefficient a. For the data taken over the 51 and 46.6% acids, t h e y were fit to the equation l / y = 1 / a + 1J y c l 0 ~ 0with , a value of unity for a and ~ C I O N O , given by eq 11. Values for yo for the 55.6, 58.5, 59.6, and 65% acids were obtained by averaging the y’s measured at the lowestp(HC1) and correcting for the small amount of HCl present by using eq I1 and a value of 3 X lo4 for r. This correction was small, ranging between 5 and 12%. Using this procedure, however, to determine the yo for the 46.6, 51, and 57.5% solutions resulted in much larger corrections, up to factors of 2. The measured p(HC1) values in these cases were near the detection threshold and are considered to be upper limits to p(HC1). Therefore, due to the difficulty in determining the HCl concentration, the yo were taken to be 8% less than the lowest measured values of y because 8% is the average of the corrections for the other measurements. In addition, the appearance of the product HOCl and only small, or no, C12 from the C10N02 uptake measurements over these solutions is consistent with only a small contribution of the HC1 reaction. The yo will be discussed in more detail below. Model for Surface Reaction with HCI. Close examination of Figure 4 reveals that eq I1 does not faithfully depict the measured

5732

The Journal of Physical Chemistry, Vol. 98, No. 22, 1994

Hanson and Ravishankara ClONO, uptake at 202 K

C l O N O , u p t a k e a t 202 K 1

3

‘ I 0 1

0

Y 0 01

0 001 u,r

10-11

10-’0

10-~

10-8

10-7

pHCl ( a t m )

10-11

,

. . . , . , . I

,

, , . . . . . I

,

. , . . . , ,I

,

, . , , , .

10-8

10-10

p(HC1) a t m

Figure 4. Measured reaction probability for CION02 plotted against p(HC1). The curves are least-squares fits to the data according to eq 11. The HzS04 contents of the solutions were 46.6% (diamonds), 51%

Figure 5. Same as Figure 4. The curves are the least-squares fits using eq IV.

(squares), 55.6% (upside-down triangles), 57.5% (diamonds), 58.5% (triangles), 59,.6% (circles), and 65% (squares). Results using the two different methods of doping the solutionswith HCI for the 5 1% solution are represented by filled and open squares.

fit the data for all the solutions studied in a global manner: i.e., specifically including the variations of H*HCI,UH,O, and yo with acid content. In addition, we allowed k,,, tovary with acid content by assuming that it was proportional to the activity of water. Therefore, we fit all the data according to the equation

y at high values ofp(HC1). One reason for this breakdown may be due to the assumptions made in deriving eq I1 where it is assumed that the reaction takes place only within the bulk of the liquid, and thus the uptake rate is controlled by the competition between reaction and diffusion in the bulk. This assumption could break down at high p(HC1) because the reaction might take place so fast that a significant number of molecules are lost at the surface before “bulk” processes are in effect, i.e., before diffusion into the bulk has a chance to compete. If the reaction at the surface is significant, then it is incorrect to treat uptake as being only due to a bulk reaction. One way to deal with the contribution of a surface reaction is to assume an Eley-Rideal or Langmuir-Hinshelwood type mechanism2I where the rate of reaction is proportional to the surface coverage of one or both of the reactants. Here we assume that uptake due to a surface reaction is proportional to the surface concentration of HCl, 8HC1, and simply add a linear term, f(8HCl), to (11):

The linear term is a loss rate coefficient proportional top(HC1). We have assumed that the surface coverage of HCl is proportional to its partial pressure and to its concentration in the bulk, Le., equal to the partial pressure of HCl times its solubility. The surface abundance could be different than that in the bulk liquid; however, as long as it is proportional to p(HC1) and therefore to the bulk HCl content, our equation will be valid. Without clear knowledge of the proportionality constant, identifying the derived values for k,,, with a specific process may be erroneous or misleading. The appearance of a surface uptake term as mathematically separate from the bulk (square root) term, as in eq 111, is shown to have a physical basis in the Appendix. By identifying the surface reaction with an uptake coefficient, we have assumed that the rate is proportional top(ClON02). This is consistent with a Langmuir-Hinshelwood mechanism because in general the surface coverage of a species is proportional to its partial pressure. Our data, y us. p(HC1) for each solution, are better fit using eq 111. The parameters rand k,,,, however, varied greatly from solution to solution, making interpretation (i.e., identification with specific physicochemical processes) of these quantities difficult. To “smooth” out these variations and thus to yield a useful parametrization for inclusion in atmospheric models, we

= y o d l + LH*,cp(HCl) ‘H,O

+

and obtainedvalues foctheconstantsp and k’:,,wherer = p/aH,o, k,,, = UH,O~‘:,,, and aH,ois the activity of water. This is equivalent to assuming that kk is dependent upon the activity of water thus thep/aH,o term. In this model, the constant p (Le., not dependent upon H2S04 content) is equal to the ratio of the second-order rate constant for reaction with HCl, k” to the second-order rate constant for reaction with H20, $ ? ‘ h i s is in addition to the assumptions discussed above concerning the surface coverages of HCl and ClON02. We also fit the data to an equation similar to eq IV without including U H ~ Oin the linear term (i.e., assuming k,,, is independent of H2SO4 content.) The resulting fit was not as good as the one obtained using eq IV. The data are again plotted in Figure 5, and the overall fit to eq IV is shown as the solid lines for each set of data. The fit yielded values for the constants p and k’i,, of 2.0 X lo3and 576, respectively, in units of (activity) M-I. The lines are then a set of calculated values using the values of H*HCIand UH,O at 202 K for each solution. The uptake coefficient for ClONO2 for stratospheric conditions of temperature and acid content can be calculated according to eq IV using the known H*Hc~and aH,O. We estimate the uncertainty in the calculated y is on the order of f30%; however, outside the temperature range of 195-215 K and the acid content range of 45-70 wt %, the values of p and k”,,,, may not apply as well, and the calculated y will have a larger uncertainty. The parameter p is equal to k ~ c l / k ~where l’ the prime indicates units of (activity)-’ s-I; thus p is in units of M-l activity. To obtain an approximate ratio for the CION02 rate constants for reaction with HC1 to that with H20, we multiply p by 50 M (activity)-’, which is the approximate [HzO] in a unit activity solution. This dimensionless ratio yields a rate constant for reaction of CION02 with C1- that is 105 times faster than that

-

Uptake of ClONOz onto Sulfuric Acid

The Journal of Physical Chemistry, Vol. 98, No. 22, 1994 5733

for reaction with HzO. This clearly shows why the HCl reaction can compete with hydrolysis even though the HCl abundance is low. The diffusion-controlled rate coefficient, kD, in M-l s-I, is given by2'

where R* is the capture radius, DI is the diffusion coefficient, and N A is Avogadro's number. For an ion-molecule reaction such as C1- + ClON02, the capture radius is likely to be 1-3 nm. At 202 K, the diffusion coefficient for C1- in 50% sulfuric acid is approximately 1e8cm2 S - ' ; ~ thus, the diffusion-controlled rate coefficient from eq V ranges from 107 to 3 X lo7 M-1 s-1. If we assume that the rate coefficient for ClONOz HCl has a value of -3 X lo7 M-1 s-1, the rate coefficient for ClONOl + H 2 0 should be -3 X lo2 M-1 s-I. This reaction is slow enough to be easily measured by liquid-phase kinetics techniques. There is some difference between the fit using eq IV and the data as shown in Figure 5. The systematic deviations arise because it is a global fit using an estimated variation for the surface and bulk reaction parameters as a function of water activity. What matters are the shapes of the curves in Figure 5; inclusion of the linear term (eq IV) better depicts the increase of the reaction probabilities with increasing HCl for each solution. Therefore, although not definitive proof, our uptake data imply a surface mechanism in addition to the bulk processes for the reaction of ClONOz with HC1. Contributionof the Linear Term. The separation of the uptake coefficient into linear and bulk (square root) terms has a physical basis. Even otherwise, it can be viewed as a good representation of the laboratory data. However, it also has some important atmospheric consequences. How much could the linear term contribute to the uptake of ClONO2 in the stratosphere? To answer this question, we separated out the reaction probability for CION02 HC1 into the bulk term (i.e., the part in the square root) and the linear term (the last term in eq IV). The reaction probability due to bulk reactions6 is equal to y a o ~ o , P / ( l+ P) where P is equal to ( P / U H , ~ ) H * H C ~ ( H CThe ~ ) . calculated CION02 HC1 reaction contribution to the total ClONO2 uptake coefficient onto 70, 60, and 50 wt % acids, which are present in the stratosphere at temperatures of -213, 202, and 195 K, respectively, is 1, 20, and 82%. The linear term contributes 12% to the ClONO2 + HCl reaction probability on 70 wt % acid, -21% for uptake onto 60 wt % acid, and 45% for uptake onto 50 wt % acid at 195 K. It is likely that for stratospheric temperatures cold enough to form type I (NAT) PSCs, which is 195 K for typical conditions (Le., for temperatures within -6 K of the ice point), the linear term may dominate the overall uptake of ClONO2 onto sulfuric acid, provided that HC1 gas-phase levels remain high. (We used p(HC1) = 5 X atm for these calculations.) For nonvolcanic conditions, however, we expect that processing on NAT will be more important than on liquid sulfuric acid. The correction for the finite size of the particles,f(l/a),6 needs to be applied only to the first term in eq IV. In practice, however, this distinction is not important becausef(l/a) is essentially unity when this reaction is important in the atmosphere. Therefore, ~ C I O N Ofrom , eq IV can be included in eq I in place of the quantity (4RT/w)H(D/ki) ClONOl Hydrolysis. Shown in Figure 6 are the values of yo, Le., the uptake coefficient due to hydrolysis, plotted against H2SO4 content of the sulfuric acid. The filled circles are previous data for yo, and the line is the fit to those data.10 The data of Tolbert et alS9and Williams et al.22arealsoshown. It is apparent that the results we obtained in this study for 45-55 wt %solutions deviate significantly (up to a factor of 2) from the fit to our previous data. Note that as the atmosphere contains HCl these

+

+

+

-

-

r IC1ONO,+H,OI

1 0 -- 4

40

50

60

70

Acid c o n t e n t

80

Z w/w

+

Figure 6. In yo (Le., for CION02 HzO in the absence of HC1) plotted us acid content. Open circles are the values measured here, the filled circles are from Hanson and Ravishankara,Io the triangles are from Tolbert et al.,* and the symbols are unpublished data from the SRI lab.22

+

0 01

0 1

1

Water a c t i v i t y Figure 7. In yo plotted vs water activity. Symbols are the same as in Figure6. Thedashedlinerepresents thecase that the reaction probability is proportional to activity squared.

yo are not what would be expected for this reaction probability,

especially for 55% and less acid content. Our CION02 + H2O reaction probabilities are plotted as a function of U H , in ~ Figure 7. This is the proper quantity to use because the reaction is believed to be hydrolysis, Le., reaction with H20. Note that when the data are plotted in this way, the uptake is a smooth function of U H , ~ .The data for U H ~ O> 0.02 were taken a t temperatures between 200 and 220 K: the filled circles a t 21 5-220 K and the open symbols at 200-203 K. Since yo does not significantly depend on temperature over this small range, the temperature-dependent terms again must work together to cancel. The strong variation of y with activity, however, can be accounted for in terms of the bulk-phase processes. Folded into the dependence of yo with UH,O are the variations of H , ki, and DIwith water activity. We have assumed that the explicit dependence of k i on water activity is linear; however, as discussed above, ki* lO-5k~; thus ki depends on D/also. Recent measurements of the viscosity of sulfuric acid solutions a t 200-220 KI6.22 reveal that, to first order, the diffusivity of a dissolved species goes as ( U H ~ O ) O . ~ . Therefore, ki is proportional to ( U H ~ ? ) ' . ~ . If we further assume that H for ClONO2 is proportional to the activity of water, then, because yo goes as H(k@J1/', it should vary with water activity to the 2-2.5 power. A dashed line for yo = O.~(UH,O)~ is also shown in Figure 7, and the data lie on this line for U H ~ 1 O 0.05. We believe this variation provides evidence that H for CION02 at a given temperature is proportional to (aHZo)*, where x is between 0.7 and 1. We also expect that for temperatures outside this range yo may show a similar dependence on activity but may have a different scaling factor. Note that for values of yo > 0.1 ( u H ,>~0.4) uptake will

-

5134 The Journal of Physical Chemistry, Vol. 98, No. 22, 1994 TABLE 2

Parameters for Calculating the Reactive Uptake Coefficient for CION02

parameter

function

log p(HzO), mbar H*HCI(T*aHzO),M atm-'

p , M-1 activity M-1

activity

(e6250/T-10.414)(aH

2.0 x 103 576

10

a

)I49

= G = 1 . 4 X 1 O 4 -

rsurl

comment

1.18 x IO-' + 9.1 x ~ O - ~ U H + ~ OO . ~ ( U H , O ) ~ 9.217-2190/( T - 12.7)

YO

k';,,, I, cm

Hanson and Ravishankara

fit to data measured here and ref 11 fit to supercooled H20 vapor pressures reported in ref 20 (223 K > T > 173 K) fit to H*HCIfrom ref 8 (230 K > T > 190 K) ratio of second-order rate coeff for CION02 HCl us CION02 surface reaction rate term using k;" = 5 x IO-%~'

+

1

surface reaction term ratio of first-order loss rate coeff for CION02 ClON02 H20

P

+

+ HzO

+ HCI to that of

uptake cueff due to bulk reactions a is droplet radius in cm

the overall uptake coeff for CION02 (use cy = 0.3)

that fraction of ye that involves reaction with HC1 producing C12

y( CION02+H20) 0 Note that be neglected.

k1 is

fraction of ClONO2 that reacts with H20 and produces HOCl the sum of ki and kkc,, The I parameter, however, significantly affects the uptake only when kkcl