Dominating influence of nh3 on the oxidation of aqueous SO2: The coupling of NH3 and SO2 in atmospheric water

Atmospheric Enoironment Printed in Great Britain.

Vol. 23. No. 12, pp. 2691-2707,

OW46981/89 53.00+0.00 Pergamon Press plc

1989.

DOMINATING INFLUENCE OF NH, ON THE OXIDATION OF AQUEOUS SO,: THE COUPLING OF NH, AND SO, IN ATMOSPHERIC WATER PHILIPPE BEHRA,* LAURA SIGG

and WERNER STUMM

Institute for Water Resources and Water Pollution Control (EAWAG), Swiss Federal Institute of Technology Ziirich (ETH), CH 8600 Diibendorf, Switzerland (First received 12 December

1988 and received for publication

15 May 1989)

Abstract-The oxidation of SO, in atmospheric water (cloud, rain, liquid aerosol and fog) is influenced by the presence of NH,. The enhancing effect of NH, is especially pronounced if the oxidation occurs with an oxidant such as OX for which the reaction rate increases strongly with increasing pH, because NH, (i) codetermines the pH of the water and thus in turn the solubility of SO,, and (ii) provides acid neutralizing capacity as well as buffer intensity to the heterogeneous atmosphere-water system in counteracting the acidity produced by the oxidation of SO,. At low buffer intensity, the acidity production leads to the alleviation of further SO,-oxidation. A computer model is used to assess the influence of SO,, NH, and other potential acids and bases, of aerosols and of the liquid water content on the composition and its temporal variation of closed or open atmospheric systems as a consequence of SO1 oxidation by 0,. An essential corollary to this model is a definition of atmospheric alkalinity (or acidity). Model results are compared with field data obtained in measuring the temporal variation in urban/rural fog composition. Key word index: Acid neutralizing capacity, aqueous-phase reactions, atmospheric alkalinity, modelling, simulations, NH,, O,, SO, oxidation.

INTRODUCTION

It is well known that oxides of S and N emitted into the atmosphere are converted by various oxidative pathways in the gaseous and aqueous phases into SOi- and NO;. With regards to SOz, this transformation is relatively fast in the aqueous phase and occurs there primarily (Jacob and Hoffmann, 1983; HoignC et al., 1985; Seinfeld, 1986; Lamb et al., 1987; Waldman and Hoffmann, 1987). On the other hand, the rate of oxidation of nitrogen oxide (NO,) occurring mostly in the gaseous phase is considered to be slow in’the aqueous phase (Schwartz and White, 1983). Since these oxidation reactions are accompanied by a generation of protons, i.e. the aqueous phase becomes acidic, the overall process is termed acid deposition (Hoffmann and Jacob, 1984, Finlayson-Pitts and Pitts, 1986; Schwartz, 1986; Seinfeld, 1986). Although emissions of NH, may neutralize part of the aqueous acidity, NH, has consequences by affect ing the rate of SO, oxidation and thus the H,SO, generation; furthermore, any NH, deposited on soils adds to soil acidification by becoming eventually converted into HNO, by microbially mediated nitrification. The importance of NH, in the atmosphere and its role in acid deposition and indirect effects on the oxidation of SO, have been recognized by various authors (Junge and Ryan, 1958; Scott and Hobbs, 1967; McKay, 1971; Penkett et al., 1979; Stelson et al., *To whom correspondence should be addressed.

1979; Munger et al., 1983; Hoffmann and Jacob, 1984; Durham et al., 1984; Ziircher and Gisler, 1987; Schuurkes et al., 1988). Our interest in the role of NH, in acid deposition was stimulated by measurements of the chemical composition of fog and aerosols, and the assessment of factors that regulate their composition (Johnson et al., 1987; Sigg et al., 1987; Ruprecht and Sigg, 1989). Although the source functions of NH, and SO2 showed considerable variations, remarkably constant molar proportions of NH: and SOi- have been observed in fogwater during events recorded in Diibendorf, a suburb of Ziirich, Switzerland. Of particular interest to us were the observations made during an accidental release of a few kg NH, within 500 m of our sampling station and during a similar NH,-experiment. The measurements of the fog composition showed in these studied cases that the SOiconcentration ‘followed’ the variations of the NHf concentrations (Sigg et al., 1987). These data lend support to the simple hypothesis that NH, mediates the oxidation of SO, in aqueous phase by 0, because (1) the absorption of SOz into water is enhanced by an increase in pH; (2) the oxidation of aqueous SOz increases with increasing pH; (3) the buffering intensity provided by the NH, in the water and especially in the gas phase mitigates the lowering of the pH and thus in turn assists in sustaining relatively high oxidation rates; and (4) the atmospheric acid neutralizing capacity (ANC), i.e. the alkalinity of the volume of atmospheric system including gas, aerosol and water phases, ultimately determines the extent of SO, oxida-

2691

2692

PHILIPPE BEHRA et ul.

tion and the NH: to SOi- ratio in the atmospheric water. The question, whether H,O, or 0, acts as an oxidant, may be answered differently if the fog occurs in a rural (uncontaminated) or in an urban environment during winter. In the latter situation, H,O1 would be used up relatively quickly in comparison to 0, by the contaminants present in the atmosphere. The objective of this paper is to describe a model of the NH,-mediated oxidation of SOI in aqueous phase by 0, for urban conditions and in wintertime, and to illustrate and discuss the applicability of the model to selected case studies. First, the equilibrium system is described in the absence of oxidation; the acid-base neutralization process in the atmospheric system, gas-aerosol-water phases, is considered, and the concept of alkalinity is investigated for a gas-water system. Secondly, when the oxidation of SO2 is possible, the evolution of the system with time is studied with regard to the alkalinity of the system, particularly the aqueous phase alkalinity. The role of different variables in the rate of the SO, oxidation, e.g. the addition of strong acid or aerosols into the system, as well as the variation of the liquid water content, is investigated and shown from simple cases. Finally, results from fog samples and from computations with this model are compared in order to improve the model, to establish the influence of NH, on SO, oxidation in the atmospheric aqueous phase, and to explain the fact that the NH:, SOi- and NO; concentrations are in relatively constant proportions in the fog water droplets (Sigg et al., 1987). CHEMICAL

MODEL

The oxidation of the SO2 to SO:- in water is accompanied by the release of two protons:

so I(gas)+ H,O + SO + SO:~ + 2H ’ + s

(1)

where SO is a source for oxidant, i.e. H,O,, 0, or 0, (Seinfeld, 1986). In addition to the H,SO, generated, various other strong acids (HCI, HNO,, HNO,) and bases (NH,, alkaline dust containing CaCO,, fly ash) are present in the atmosphere (Jacob et al., 1986; Johnson et al., 1987). The atmospheric liquid water is characterized by the electroneutrality equation: [NH:]+[H+]+Cn[Cat.“‘] =[OH-]+[HSO,]+2[SO;-] + [HSO;]

+ 2[SO:--]

+ Xm[An.“-]

(2)

where Cn[Cat.” + ] is the sum of the other cations, e.g. Ca’ + from alkaline dust; Zm[An.“- ] is the sum of the other anions, e.g. NO; , Cl-, or dissolved anionic CO, species; and [ ] indicates concentration in molt _ ‘. The rate of oxidation of SO, in water depends on the oxidant (SO), on solution variables, and on the presence of catalysts. Our model system is built up by mixing, within a given volume, in a closed or open system, various gases (CO,, NH,, SO,, oxidant), and liquid water (water droplets). Also, acids (e.g. HCl, HNO,) and bases (e.g. alkalinity from dust) can be added to the system (Fig. 1). Then, a typical reaction sequence is established by oxidizing SO1 ; the H + ions produced as a consequence of this oxidation ‘titrate’ the gas-water system to a lower pH.

Chemical equilibrium model

Water, aerosol, and gas phases are taken into account in the chemical equilibrium model. Equilibria in our model system are characterized by Henry’s law for exchanges between the gaseous and aqueous phases, and by acid--base equilibria in the aqueous

rate reiationshlp d[ZS(IV)f,ql]/dt

Fig. 1. Pictorial representation of processes occurring in the aqueous phase: oxidation of SO2 species by O3 in the presence of NH, (after Ziircher and Gisler, 1987; Hoignt, 1988). For every time increment, some of the H,SO, is formed. In turn, the pH and the aqueous phase alkalinity decrease.

2693

Influence of NH, on oxidation of SO, Tabie 1. The~~~~ic Equilibrium number

(RI) W (R3) (R4) (RS) (R6) (R7) (R8) (R9) (RlO) (Rll)

data considered in atmosphe~c-water

Equilibrium

log

H,OeH+ +OHco ztpa,~“H,CO: H,CO$F~H+ + HCO; H&O;F~H+ +CO;ztp,,*H,SO:(l) H,SO$+H+ +HSO; HSO;*H+ +SO$yHO; E~H+ +SO:J,~&NHQ~) NH: srt NHa(.,,) + H+ 0 3lrd=O3w

K,= -14.00 K,= - 1.41 K,= -6.37 & = - 10.33 KS = 9.34 x 1o-2 Kg= -1.89 K,= -1.22 K,= -1.92 Kg = 1.79 K,,= -9.23 K,,= -2.03 K,z= -2.72

w,~,*N%,,

:2::; (Rl4) (RI51 (Rl6) (Rl7) (Rl8) (R19f I::; (R22) 1~23)

K398,iS

NO Z~~st@NOz@l) 2NOz~p~,+HzO@H+ +HNO,,,,I+NO; NOfrps)+ NO,iy,,t + H,Oe2HNO,,, HNOxpas, FTH +NO; HNOz,p,*HNOx,q, HNO,,,@H + + NO; HQ..,,+.---,-‘H+ +Cl-

CI-W,a,,~-Ww,, CH,O,,,, + H,O~H,C(OHh

CH,O,,,,+ HSO; @HOCH,SO; HOCH,SO;F~H+ +OCH,SO$-

K,,=-1.92

K,,=S.68 K,,= -1.26 K1,=6.S1 K1,= 1.69 Kla= -3.29 K,,=6.30 Kz,=3.8S K,, = 3.26 K,,=7.00 I(,,= - 11.70

model H%s.15 (kcal mol- ‘)

Reference

13.34s -4.846 1.825 3.55 - 6.247 -4.161 - 2.23 - 5.23 -8.17 12.47 -5.04 -2.94 - 5.0 -25.6 - 18.2 - 17.3 -9.5 -2.5 - 17.89 - 12.85 -8.04 0 0

S S S S S s S SM S SM S SW J SW SW JH JH SW L JH J K J

[H$O:] is defined as for CO, according to Stumm and Morgan (1981). References: J: Jacob, 1986; JH: Jacob and Hoffmann, 1983; K: Kok et al., 1986; L: Liljestrand, 1985; S: Seinfeld, 1986; SM: Stumm and Morgan, 1981; SW: Schwartz and White, 1983.

phase. The different equ~lib~a and species are listed in Table 1. (1) In the open system mode!, constant partial pressure of each gas component, (pso2) and (pm,), is maintained. The total mol balances for the sum of S(W), neglecting the formation of other complexes as HOCH,SO; , and NH, species in water then becomes for equilibria:

E&& = ~EGJ.

m

The open and closed equilibrium systems are depicted for NH,, SO, and CO, in the presence of water in Figs Za,b. Double logarithmic diagrams, with log molar concentration of the individual species in water or log partial pressure (atm) of the gas vs pH, are used for atmospheric systems (Charlson and Rodhe, 1982). In Fig. 24 partial pressures of NH3, SOz and CO2 are constant, i.e. the system is open for all components; in W(IV),.,,1 = (PS**).(KS+KS’K~.CH+I-’ Fig. 2b, the system is open for CO, only, but closed for + K, K6.K,-[H+-j-2) (3) SO2 and NH,. In a closed system, significant amounts of the NH3 CNHWIT= IN%,,,,1+ [NH:3 and SO, are present in the gas phase (Fig. 2~). The =(PNHJ).(K~+~Y~.R;~~.CH+I) (4) distribution of NH, and SOz between the gas and where the K-values refer to those given in Table 1. water phases depends on pH and on the liquid water (2) In the closed system model, there is no exchange content. Below pH 5.5, nearly all the NH, dissolves in of matter between the considered volume of the system water to form NH:, while SO, is predominantly and the outside. Thus, the sum of component concengaseous. Above pH 8, NH, is mainly in the gas phase, tration in gas and water phases within the total system and nearly all the SO? dissolves to form SO;- (Fig. 2, remains constant. For a component C, this sum can be parts b and c). In the closed system, the maximum characterized by an initial partial pressure noted concentrations in the aqueous phase, in mole- I, are (pc)o, e.g. SO, or NH,, which is assumed for a volume given by: of the atmosphere before water droplets condense. Then, we can write in the presence of a liquid water WI: I,,. = PH3M andCSO~-l,,,= PW&. content in the atmosphere (q, in Gmm3): (7) {Cl, = (P&IRT=

W/R~f qf&,,l,

(5) Atmospheric acidity and alkalinity

where R is the gas constant (8.2057x lOIs atm m3 K-’ mol-‘). Tthe temperature, {C), the total concentration in the gas-water system (in mol me3), Ci the concentration of the species, i, and

Acid deposition results primarily from a disturbance of the redox balance (the reactions of the fuel combustion, i.e. the reactions of oxidation of C, N and S exceed reduction reactions in these elemental cycles).

2694

PHILIPPE

BEHRA

et al.

6

0

2

4

P”

6 PH

a

10

12

P”

open system for SO2

I-

ol!_-4-L 2 PI

homogenwS,&ck$d

-5 acid c

0

2

8

4

10

0

Alkalimetrlc

5

or

acldlmetrlc

1

10 (meqli) L

12

p”H

s$$rr

titration

base curve

Fig. 2. Equilibria of SO,, NH,, CO, and H,O of water and/or gas system as a function of acid and base added. @so,) = (pnn,) = 1.22 x lO_satm or {SO,} = {NH,} = 5 x 10-7molm-3; liquid water content (4): 10-4~m-3; temperature 298 K. For the calculation, (a) an open system for SO, and NH,, or (b) a closed system for SO, and NH, was assumed; (c) the percentage of the total amount of NH, and SO, present in the system is given as a function of pH; (d) the alkalimetric and acidimetric titration curves of heterogeneous (gas + water) corresponding to both cases (a) and (b) are compared with that of a homogeneous system in which the components are considered as being non-volatile in the absence of CO,. Note that the buffering capacity of the open system is much larger than that of the closed and heterogeneous one which is much larger than that of the closed and homogeneous one.

Because the transfer of electrons is coupled with the transfer of protons, this leads to a net production of H+ ions in atmospheric precipitation. In order to understand how the disturbance in e-- and H+balance is transferred through the atmosphere into atmospheric water and to the terrestrial and aquatic environment, it is essential to consider the acid-base neutralization process in the entire atmospheric system. Liljestrand (1985), and Jacob et al. (1986) introduced the concept of atmospheric acidity and alkalinity to interpret the interactions of NH, with strong acids emitted into and/or produced within the atmosphere. A net atmospheric acidity can be de&ted by summing over all potential acids and bases in the gas,

aerosol and liquid phases, using reference conditions for specific redox conditions, and assuming the presence of liquid water. Figure 3 exemplifies the concept of alkalinity (Alk), and acidity (Acy) for a representative gas-water environment and defines the relevant reference conditions. In Fig. 3a,b, it is shown how the gases NH,, SOz, NO,, HNO, , HCl and CO, (potential bases or acids), subsequent to their dissolution in water and the oxidation of SO, to H,SO, and for NO, to HNO,, become alkalinity or acidity components. Alkalinity is the acid neutralizing capacity, and is defined by a net proton balance with regard to a reference level; that is

2695

Influence of NH, on oxidation of SO, Reference:

Oxic

condition

and

presence

of

water

a) Gas phase

t 0

I I 40

I 20

b) Gas &Solved

0.0

I , SO n mol mm3

, 60 in

water and under oxic conditions

*

& = w 1W perf@)

I

8

1

I

,

0.2

0.4

0.6

0.8

I

1.0

1.2 meq

1-f

c) Aerosol Ho’ d’

MI+ so:t 0

1

t 100

Na, 1 200

1

(AW,.

- - (AcyJM= (NH&, + (Na% + 2(Cart - (NO;), - 2(S@;t~ - lx),,

x-

I 300

t 400neq

me3

d) Gas dissolved in water prior to oxidation of SO2to SO?

Reference Species:H,O.

1

[Alkh

Ii+

M4’

-

$50;.

NH:, H,COj, (7, NO;. Or0

[NH& + [HSOcjh + [S(lV)b t 2(Sqt,

- INH& + W&q

+ [HCO& + z(Ccrjt, - [H+tw

- IrJo;lw- Wly, - (Wl,

where(S(W)], representsaWehydeadditioncompwtxtset Wita

c

*

I

0.0

0.4

0.2

I

0.6

0.8 meq I-’

Fig. 3. Alkalinity/acidity in atmosphere, aerosols and atmospheric water. Alkalinity and acidity can be defined for the atmosphere using a reference state valid for oxic conditions (SOa and NO, oxidized to HaSO, and HNO,) and in the presence of water. The neutralixation of atmospheric acidity by NH, is a major driving force in atmosphe~c deposition.

the sum of the concentrations of all the species containing protons in deficiency minus the concentrations of the species containing protons in excess of the proton reference level. Thus (Alkf,,,, and (Ac~&~.~), for the gas phase, is defined by the folfowing relation:

(AlkL,

= (NH3 X,m, - 2iSO2 1~ - IHNO&,,j

- {HClL,

- 1NO,1,,,

CAlk),,j

= - (Ac~Lsj

(9)

where { ) indicate mol per m3, and HOrg is the sum of volatile organic acids. In case of aerosols, we can define the alkalinity by a charge balance of the sum of conservative cations, WCat.‘+ )(aej,of NH: , (NH: )(a5r,and of the sum of conservative anions, XCm{An.“- Jfac)(Fig. 3~): (AlkL,

= WCat.“+

)

- Zm{An.“- )(_)

(IO)

with WCat.“+

Lr = (Na+ itnet + 21Cazf jfstr + W’&C) + 2{Mg2+ )(acf+ . * (11)

Le)

+ {NO; 1~ + - . .

(12)

In addition to this, the aqueous phase alkalinity includes S(W) species (Fig. 3d). The aqueous phase alkalinity is defined as: {Alk}(,,, = q*[Alk] = q.(2[CO:-]

- IHOrgL, (8)

and

Cm{An.“- }(a*)= {Cl- )(ne)+ 21s0:-

+ 2[SO:-

I- [HCO;]

] + [HSO, J

+ CS(IV)- 3 + ENH,<,,,l + [OH-]

+ [Org-]

- [H’])

(13)

or by a charge balance on the aqueous phase (see Equation (2)): {Alk),,,, = q*(L;n[Cat.“+] +

CNHWI~- WA=“- 1

(14)

where S(W)are the aldehyde adducts of HSO; . Zn[Cat.“‘] and Cm[An.“l-] are respectively the equilvalent sum of the concentrations of conservative cations (Ca’“, IL+, Na+, Mg*+, . . . ) and anions (Cl-, SO;-, NO;, NO;, . . .). The emission of gaseous NH, into the atmosphere essentially ‘titrates’ the inputs of the potentialstrong

PHILIPPE BEHRA et al.

2696

acids. In case of a residual atmospheric alkalinity, i.e. if

all the SO, absorbed into a water droplet will be relatively readily converted into H,SO,, On the other hand, SO, oxidation by 0, becomes kinetically delayed (auto-inhibition due to the lowering of the pH as a consequence of SO, oxidation) in the case of residual atmospheric acidity. The NH, cannot ‘neutralize’ the acids present as well as the acidity generated by SO, oxidation. Thus, the NH,-by influencing pH, buffer intensity and acid neutralizing capacity+ssentially regulates the extent of SO, oxidation by 0, that occurs incipiently and within a time period of a few hours. Thus, we can expect in atmospheric water droplets (cloud, rain, liquid aerosols) relatively constant proportion of NH: to SOi(mol-ratio [NH; ]/[SO:- ] z 2). The alkalimetric or acidimetric (addition of base or acid, respectively) titration curves of the systems given in Fig. 2a,b are drawn in Fig. 2d. The gradient of the curves, -dpH/dC, or dpH/dC,, where C, is the concentration of added acid, and C, the concentration of added base, respectively, in meq e- ’ , is the inverse of the total buffer intensity of the system, fi (Stumm and Morgan, 1981). If NH, and SO, are in an open system, case(i), the buffer intensity is very high and the pH variation is not large: gaseous NH, is able to neutralize the acid additions by forming NHf , while SO, can neutralize the base additions (Fig. 2a). On the other hand, in the absence of CO,, and in a homogeneous and closed system in which NH, and SO, are considered as being non-volatile, case (iii), the pHrange is very large: the buffer intensity is very low if some acid is added. If NH, and SO, are in a closed system either with CO, or without CO,, case (ii), the buffer intensity is significant as long as all NH, is not exhausted in the gas phase, and is able to be dissolved into water to neutralize acid addition. In the latter case, the buffer intensity of CO, is shown when a base is added to the system: in the absence of CO,, pH increases, whereas the variation of pH is very low in the presence of CO,, because dissolved CO, neutralizes base addition (Fig. 2b). Chemical kinetic model

The two most important ways to oxidize SO, in water are by dissolved O,, or dissolved H,O,; the other possibilities are the oxidation by NO2 or N(III), and by O2 catalyzed by Fe (III), or Mn (II) (Seinfeld, 1986). The oxidation rate with 0, is -d[ZS(IV),,,,]/dt

= (k,.[H,SO:]

+ k, .[HSO;]

+ kz.CSO:- IPCQ,,,,]

(16)

where [H$O$] is the total analytical concentration of dissolved SO, defined like for CO, accord-

ing to Stumm and Morgan (1981); k,=(2+2) to4 L moI_‘s-‘, k,=(3.2f0.2) lo5 1*mol.‘s- ‘. and k,=(l.Of0.2) 10g~mols-’ at 295 K (Hoigne et al., 1985) with activation energies equal to 46.0 kJ mol- ’ and 43.9 kJmol- ’ for k, and k2, respectively (Erickson et al., 1977); and with H,O,:

-dC=fW,,,,l!d~ = k~CH+l~Ci-f,O,&! .[~s(Iv),,,l’~HsOl~(t+~.[H’I)

’

(17)

with k=7.45xt07Pmol~‘s-‘, K=13/mol’-‘, and xHSO; = [HSO;]/[CS(IV),,,,] at 298 K (Hoffmann and Calvert, 1985, in Seinfeld, 1986). By comparing Equations (16) and (17), it is observed that oxidation with 0, is strongly pH dependent, while oxidation with H,O, is not, if pH > 2. In this case, it can be shown that the Equation (I 7) becomes:

-dCWW,,,,lldt z k” CH,O,~,,,l . CHaWl (18) where k’, equal to 9.61 x lo5 emol-i s-i, is a constant equal to k times K,. fn an open system with regard to SO,, [HZSOz] is constant (Fig. Za), and the oxidation rate does not depend on pH. On the other hand, in a closed system with regard to SO,, [H$O:} is not always constant and the oxidation rate decreases when [H$O;] decreases (Fig. 2b). Moreover, if SO, is in excess with respect to H,O,, HzO, is not constant, and the oxidation rate is lowered as the concentration of H,O, decreases (Hoignt, 1988). The acidity of water droplets can also be due to the oxidation of NO,. According to Schwartz and White (1983), the possible reactions are from the gaseous phase to the aqueous phase (see Table 1, Reactions R12-R15 and R19): 2NO2(8asj+ H,O : H+ +NO; k J

I- HNO,,,,,(19)

and

k-4

with k3 =4.25 x lo3 mol/-is-iatm-*; k_, = 8.88 x 1O-3&2 molw2 s-‘. k., = 7.03 x lo2 mol rP_’ s-i at 298 K and’ k-,~5.57~mol~~‘s~~’ atm-*, (Schwartz and White, 1983). The rate is mainly limited by the low values of Henry’s law coefficients of NO and NO, components. Numerical methods and model assumptions

We used the chemical equilibrium program, MICROQL (Westail et at., 1976; Westall, 1979; Stumm and Morgan, 1981). This program was modified to allow the calculation when at least one of the components is in closed system. Alkalimetric and acidimetric titration curves can be calculated using the electroneutrality equation or proton condition. In our exampies, the ionic strength wiIt normafly not exceed lo- 2 moi d- I. In first approximation, the solution is

Influence of NH, on oxidation of SO, considered not to be very different from an ideal one. Thus, we did not correct for activity effects. At initial time (t = 0), water droplets are assumed to condense in a given volume containing various atmospheric gases. The extent of droplet formation is assumed to be instantaneous, and is reflected in the liquid water content. Acid-base and gas-water equilibria and dissolution of aerosols in water are assumed instantaneous. Mass transport to the droplet surface or mass transfer into the bulk phase is not considered limiting (Schwartz and Freiberg, 1981; Finlayson-Pitts and Pitts, 1986; Schwartz, 1988), and is not taken into account. As shown by Schwartz and Freiberg (1981),

0a

2691

and Schwartz (1988), the solubility equilibria of SO,, and 0, at the gas-water interface may be considered to be fast for small water droplets, except perhaps at high pH, i.e. pH > 7. Although the role of temperature in equilibria (McKay, 1971) and rates (Hoffmann, 1986) is important, we made all computations at 298 K (Figs 2-9). A sensitivity analysis showed that the effect of temperature on equilibrium constants are compensated by the effect of temperature on rate constants. The rate equation for the oxidation of SO1 is numerically integrated with the fourth-order method of Runge-Kutta (Sibony and Mardon, 1984). With

03

6.001 03 In open system 2- 03 In closed system

z l -

04

f.3

L-

@NH3)OC2 @SO~)O 1 : 03 in open system 2’ 03 in closed system

@N,,)O'2(pS0,)0 04

*Au

1 : 03 m open system 2’ 03 m closed system

z 02 [so2-1’1

[so32 -1.0

0

I 20000

I 40000

I 60000

t(s)

Fig. 4. System with residual atmospheric acidity, i.e. low NH, partial pressure (P~“,)~ c 2(pso,),. Variation in aqueous phase composition as a consequence of SO, oxidation: (a) [NH:] and [SO:-], (b) pH, (c) [NH:]*, [SO:-]*, and CO,]*, and (d) [NH: ]/[SO:- ] vs time. Initial conditions: (P~“,)~= 4.3 x 10m9atm ([NH,], = 9.6 x IO-*mold-‘k (~s~~)~= 2.3 x lo-* atm ([SO,lo = 5.1 x 10m3mold-‘); {NH,NO,},,,O,, = 1.0 x 10-smolm-3 ([NO;] = 5.4 x 10-5molL’-1); (P~,)~ = 1.5 x lo-* atm (pco,)o= 1O-3.5 atm; q = 1.84 x lo-“ Gmv3; temperature 298 K; NH, and SO, are in closed system; CO, is in open system; 0, in open or closed system. N.B.: mHf]* = [NHi]/[NH,],, [SO:-]* = [SO:-]/[SO,lO, and [O,]: = [O,],/[O,],,.

2698

PHILIPPEBEHRAet al.

0a

@

lO”r

i”

I

,_.A._

5.5

[so:-11 ,_.-.-.-

& ‘\

. AH / _________________*----------.bo4-12

NH)2

_

‘,_y)l ---_

(PNH3)O ’ 2(b02)0 1 03 in open system 2’ 03 In closed system I

loo

I

20000

40000

5

IO

I

I

45: 0

1

I

60000

t(s)

15

20 t(h)

/

‘,oi

I

20000

40000

1p

~

60000

1:

t(s

2F”i

4.0 1 03 In opansystem 2 03 In closed system

A2

0.C

ul y -

3.0

;” 3

04 I : 03 m open system 2 03 I” closed system

I

Oi

\...;--___~~~~____~_____ 0.C 20000

40000

i 0

60000 t(s)

20000

40000

60000

t(s)

.5. System with atmospheric ‘neutrality’, i.e. (pNH3)0 = 2(p,,),, . Variation in aqueous phase composition as a consequence of SO, oxidation: (a) [NH:] and [SO:-], (b)pH, (c)[NH:]*, [SO:-]*, and [O,]*, and = 4.6 x 10-s atm ([NH,], = 1.1 x lob2 mole-l); for (d) [NH: ]/[SO:- ] vs time. Initial conditions: (pNHI)O other conditions like Fig. 4.

this method, the concentration of S(W) is calculated at time (t + At), where At is the time step. The method is explicit with time. In this condition, the numerical process is consistent, convergent, and stable, if the time step accords with the following condition at time t: 0 <(At)’ < (At): = l/S; - (((At)‘)4.(S;)3/6

+((Ac)‘)~.(S;)*/~ + S((At)‘)’ . S5/6)

(21)

where (At): is the critical time step, and s; = (kJ.ab + k,.& + k~.cc:).Co,,.,,l’

(22)

with c& = (1 +K,/[H+]‘+Kg.K7/([H+~)2)-L

(23)

xi = K,/[H + 1’. a;

(24)

x; = K, .K,/([H+]‘)’

.a;.

(25)

The first condition, which states that At must be positive, is trivial. The second one is very and more restrictive than in the case of simple finite difference technique described by Hoffmann and Jacob (1984) or Seinfeld (1986), where the right term of condition (21) only equalled l/S& Condition (21) is a necessary condition for stabihty; nevertheless, it is still possible

Influence of NH, on oxidation of SO,

0a 12*r

10 I

5

I

15 I

2699

@a

7.0°

20 t(h) I 1

bJHtl1

5 I

10 I

15 I

20 t(h) 1 *

PH 6.5 -

(PNH~)oQ(Pso&o 1 o3 I” open system 2’ 03 In closed system

6.0

?

\

-24

----_--_-----

5.5 -

iW)t

1 : 03 in open system 2: 03 in closed system

1

*o

I

I

t

20000

a000

5

10

6oooo t(s)

I

L

0

0c 1.0:

20 t(h)

15

I

/+-

‘(so2-I”1

i

.

7 Iso:-)‘

‘z:

(PNH3)0’2(%02t0 1 : 03 in open system 2.03 in closed system

0

*--

’ 20000

103 19 I 4OOOO

3.0

‘ii 3

)NH;)“P

/.

2: 03 in closed system

s:

.C”._.L.-.-.-.-.L

0.0

2pt(i

4.0

(NH;]*1

\

1:

_..ASa

0.2 1

1p

’

0.8 -/ I *_CL-v-.-_..

:

2.0

I 60000

I

UL----J * o

t(s)

20000

4oooo

60000

t(s)

Fig. 6. System with residual a~osphe~c alkalinity, i.e. NH3 in relative excess (pmJo > Z(p,,),. Variation in aqueous phase composition as a consequence. of SO, oxidation: (a) [NH:] and [SO:-], (b) pH, (c) [NH:]*, [SO:- ]*, and [O,]*, and(d) [NH: ]/[SOZ,-] vs time. Initial conditions: (p&, = 8.0 x lo-* atm (fNH3]* = I.8 x 10q2 molL_‘); for other conditions like Fig. 4.

that the deviations from the true exact solution of the original differential equation are very large. Moreover, it is very important that At is not too small to avoid dispersion due to numerica error during calculation. Time step increment is calculated at each iteration, and is taken to comply with the conditions above as: (l/S;)/100

G (Ar)’ < (l,‘S#O.

(26)

Then, the new data of S(W), S(VI), and 03tasl are introduced into MICROQL. This subroutine is used on the one hand to compute the speciation at the starting of the computation (t = O), and on the other hand to calculate the new pH in the iterative part over time (at time t). Reactions progress models are, in this

case, based on the concept of partial equilibrium in the system (Stumm and Morgan, 1981), where partial equi~b~um describes a state in which the system is in equilibrium with some but not all reactions at each time step. The variables are the initial partial pressures of SO*, NH, and O,, the liquid water content, the aerosol concentrations, and the concentrations of base (alkaline dust, e.g. CaC03) or strong acid (HN03, HCl or H,SO,). In this model, 0, is selected as the oxidant for SQ2. The rate expression (see Equations (16) is characterized by a pH-dependence that is in accordance with the observations (Johnson et al., 1987; Sigg er al., 1987); for urban

2700

PHILIPPE BEHRA et al.

0a

05

P”

60

-1

0

1 [Acy]

(water,

0

meq/l)

0.8O

05 I

10 I

2000

10

/

15 20 t(h) / , , [Strong Acid 1 added

4000

6000

t(s)

1.5 20 I I I [Strong Acld] added

Fig. 7. Addition of strong acid, such as HNO, or HCI to the system; its effect on progress of SO, oxidation as depicted by the development of pH and SO:-. The addition qfacid brings the system to a lower buffef intensity and to a lower pH. The zoncentration-range of strong ands added IS lo-’ to lOmeqL_‘. Irutlal condiEonnj3 (p,.,,,,),,=l.Ox lo- atm ([NH,],=4.1 x lo-” mold-‘); (p&=6.0x low9 atm ([SO,],=2.05 mol e-l); (p,-&, = 1.5 x lo-* atm; (pcol),, = 10-3.s atm; q = low4 G rnm3; temperature 298 K; NH, and SO, are in closed system; 0, and CO, are in open system.

conditions and in wintertime, it is reasonable to assume that pHlol is lower than po, but also lower than psol (Hoigni, 1988; Olszyna et al., 1988). Therefore, H,O,, if initially present at all, is exhausted during the oxidation of SO,. Thus, even if the rate of SO2 oxidation by H,O, is faster than that by 0, (Seinfeld, 1986), the partial pressure of H,O, would decrease rapidly and become negligible (Seinfeld, 1986; Ho@, 1988). In the model, we only take into account the oxidation of SO,. The NO, oxidation in the aqueous phase is not considered since calculations on the rate of the oxidation of NO,, Equations (19) and (20), showed

of acidity due to aqueous phase oxidation is negligible within the time of interest. that the contribution

RESULTS AND DI!SCUSMON Dynamic behaviour aqueous phase

ofsolutions with SO,

oxidation in

In order to illustrate the results of our computer simulations, we will first consider the simplified system C02, NH,, SO1, 0, and H20. Three cases exemplify the concept of atmospheric acidity or alkalinity with respect to Equation (15): (i) a system with residual

2701

Influence of NH, on oxidation of SO2 atmospheric acidity, i.e. (pNt& < 2(pso,), in Fig. 4; (ii) ‘neutral’ atmosphere, i.e. (pmr,)0 = 2(pso,)a in Fig 5; (iii) a system with residual atmospheric alkalinity, i.e. (pN& > 2(p,,), in Fig. 6. In the three cases, the system is either open or closed with respect to 0s. An initial rapid decrease in pH (concordant with an increase in SOiand NH: concentrations) is followed by an asymptotic approach to a steady state value. The reaction is also characterized by a variation in the mol-ratio [NHZJ to [SO:-]. The amount of SOi- or NH: present in the aqueous phase, and consequently the amount of SO, consumed by oxidation or the amount of NH, still present in gaseous phase, can be directly determined by using the relative coordinates defined by mol-ratios [SO:-]* =[SO:-]/[SO&, and [NH~]*=[NHfl/CNH& (Figs 46), where [SO,], and [NH,], are the total and initial concentrations of SO2 and NH,, respectively, in the entire atmospheric system,

Residual

atmospheric

alkalinity

(hH3),, > 2(p,,),

(Fig. 6) Although H,SO, production is extensive, the pHdecrease is much smaller than in both previous cases. The drop in pH comes to an end as soon as SO, or O3 becomes exhausted. At the beginning of the simulation, the aqueous phase alkalinity is very large, but it decreases and tends to 0: the mol-ratio [NH:] to [SO:-] is equal to 2 and is constant with time. The excess of NH, is sufficient to neutralize H,SO* and maintain the pH at a high value. The oxidation rate is very fast. Here, the difference between systems open and closed with respect to 0, is significant. Moreover, NH3 is not in sufficient excess for the system to be considered as an open one: an important part of it is used to neutralize the H+ ions formed by the oxidation to SO:-.

Acidimetric titration curve Residual (Fig. 4)

atmospheric

acidity

(pNH,),, c 2(psoz)~

At the banning of the simulation, the aqueous phase alkalinity is positive; but, the acid neutralizing capacity is exhausted very quickly, and nearly all the NH, dissolves to form NHf , i.e. [NH: ]* tends to 1. At the conclusion of the simulation, the SO:- concentration and pH are lo.w, whether the system is open or closed with respect to 0,. Since the pH is too low (pH z 4.0), the oxidation of SO, is very slow. At the end of the simulation, SOi- is still being produced at a very low rate (see Equation (16)). The mol-ratio [NH:] to [SO:-] approaches 2.0 within a few hours. For this case, there is no difference between open and closed system with respect to O3 because SO2 and O3 are in excess with respect to NH,. If the consumption of a component, i.e. in our case 0s or SO,, is small in comparison to the total and initial concentration of O3 and SO,, respectively, the system becomes as if it was open with regard to O3 and SOZ, (po, z constant and psoZ x constant), respectively.

Armospheric ‘neutrality’ (pNH3)0= 2(p,,,),

(Fig. 5)

In this case, the production of SOi- is more extensive than in the previous case, and the decrease in the pH-range is smaller, depending on the 0, initial conditions. If O3 is in closed system, this component is not completely exhausted within the time span of calculation. The mol-ratio [NH: J to [SO:- J tends to 2. It is important to notice that, in these conditions of atmospheric ‘neutrality’, the aqueous phase is not ‘neutral’, i.e. the pH is < 5.6, the reference pH corresponding to pure water in equilibrium with the weak acid CO, only (Charlson and Rodhe, 1982). At the beginning of the simulation, the aqueous phase alkalinity is very large, but it decreases and tends to zero.

For every time increment, At, some of the HzS04 is formed and the H+ ions produced titrate the gas-water system along the ‘titration curve’ (see Fig 2d) to a slightly lower PH. The titration curves depend on the initial conditions of the calculations, i.e. (pNH3IO, and on (P~,M~(PSOJW

The slope

of the

titration curve is inversely proportional to the buffer intensity, fi. Obviously, if (pNHs),, > 2(p,,),, the buffer intensity is large, and the acid neutralizing capacity of the NH, (mostly in the gas phase) is capable of readily neutralizing all the acidity generated from SOZ oxidation. Alternatively, if (pm,&, < 2(p,,),, the buffer intensity is very weak, and the acid neutralizing capacity of NH, becomes rapidly exhausted; most of the NH, in the system will be in the water phase as NH;. The ideas presented here for simple NH,, SO,, 0,) CO2 and H,O systems are readily extended to more involved situations. The addition of other strong acids (like HNO, or HCl) or strong bases (such as CaO from fly ash) has as a consequence the shift of the gas-water system along the titration curve. It is also interesting to assess the influence of the uptake by the water of aerosols and the effect of a change in the liquid water content.

A~itian

of strong acid and aerosols

If strong acids, such as HNO, or HCl produced or advected, are introduced into our system, these acids become readily absorbed in the water phase and they titrate the system along the titration curve towards lower pH. Furthermore, they reduce the alkalinity of the total system and lead to decrease the buffer intensity because with decreasing pH the gas phase becomes more depleted in NHscW,. In general, we can quantify the ‘titration curve’ by

2702

PHILIPPE BEHRA et

the following equation: {Alk) = {NH, 1~) + q~UWb,J +[OH-]+2[CO~-]+[HCO~] +2[SO;-Ii-[HSO;]-[A-])

(27)

where [A-] is the sum of the anions of strong acid introduced into the system (Cl-, NO; ) or generated by oxidation of SO, (SO:- ). We can investigate the addition of strong acids or aerosols in the system before a fog event. Considering that in our system there are only two cations, H+ and NH:, i.e. Cn[Cat.“‘] is equal to zero, the electroneutrality Equation (2) becomes: %n[An.“-]-[H+]

= [NH:]-[A-]

= [AIL],,, (28)

where Cm [An.“’ .^] is the sum of the concentrations of the anions (OH-, HSO;. SO:-, HCO;, CO:-, S(IV)_ ). Therefore, we can consider that one part of NH, is equal to the concentrations of conservative anions coming either from strong acids or from aerosol into the aqueous phase, {NH, },; the remaining part distributed between the gaseous phase and the aqueous phase is the (NH, jr. From this assumption, the total NH, in the system can be written: {NH, jr = {NH, lr + (NH, In = (NH,}, + q.[A-1.

(29)

From the relation (5) and Table 1, we can infer: xb = CNH: MPH,

I+-’ f

=K,K;i[H+]((qRTf-‘+I& + K,K;,‘[H+])-‘.

(30)

The relation (28) becomes: [Alk],,,,=

X[An.“~~] -[H’] = {NHj},(xb(l

+ r) - r)lq

(31)

where (32) This aqueous phase alkalinity written like this is only defined with respect to NH,. It appears that the ratio r is a determining factor and determines essentially the acidity or alkalinity of the aqueous phase. If (NH, >, is very high with regard to {NH,},, i.e. the amount of acid or aerosol is tow, r tends to zero and nearly all the NH, is in the gaseous phase. In this case, the aqueous phase alkalinity tends to [NH: 1. On the contrary, if (NH, ),, is very high with regard to {NH, I,, i.e. the amount of acid or aerosol added to or generated by the system is high, r tends to be much larger than 1. The aqueous phase alkalinity tends to zero, and [H ’ ] tends to Z[An.“-1. Moreover, if the aqueous phase alkalinity is negative, i.e. there is an excess of acid with. respect to the reference point, the pH tends to

al.

tog( - [Alk]faqj) according to Equation (3 I) or log([Acy](,,,) (see Equation (9) applied to the aqueous phase) (Fig. 7a). If, for given conditions, the pH of the water phase is oniy able to follow one titration curve (relation (27), Fig. 7a). the variation of pH with time due to the generation of strong acid during the oxidation of SO, depends on the initial conditions as well, i.e. the total amount of strong acid in the system at time O(Fig. 7b). Indeed, if the strong acid concentration increases at time 0, the capability of the entire system to produce more SO$- from SOZ oxidation is reduced (Fig. 7c), since the pH tends to be too low and the rate of oxidation of SO, becomes too slow. The relation (31) is true if aerosols are present. Our simulation model considers the uptake of aerosols into the water droplets by adjusting the electroneutrahty equation. The aerosols used in Fig. 8 are (NH‘,),SO, .2NH,NO,, which are mixtures of (NH,),SO, and NH,NO, aerosols (Stelson and Seinfeld, 1982; Seinfeld, 1986). It is interesting to note that the addition of this “neutral” aerosol, e.g. [NH:] = 2[SOi-] + [NO, 1, may change the alkalinity of the water droplets because the partitioning of the alkalinity between the gas phase and the water phase (due to NH, distribution) may become changed; in other words, the uptake of ‘neutral’ aerosols may reduce the alkalinity of the water phase, and in turn decrease the rate of SO, oxidation, because some of the aerosol NH: ends up as NH, in the gas phase. The higher the ratio r, on the one hand the lower the production of SOi- from SO, oxidation---all the SO:-- present in the aqueous phase is the result of the dissolution of aerosols+Fig. 8a), and on the other hand the lower the pH. Moreover, the pH-range is smaller between starting and end of simulations (Fig. 8b). But, what about the mol-ratio [NHf],,,,/ PO: - Iw when aerosols as (NH,),SO,.2NH,NOs are present in the system? For the concentrations of aerosols 1.5 x 10-s, ISxlO~‘, and 2.0~10~~ mol mV3, the ratio decreases from 6.5, 4.0 and 4.0 to 2.2, 3.6 and 4.0, respectively. Thus, on the one hand, the oxidation of SO, tends to lower the mol-ratio to 2.0 as previously shown in CNHfl~,,,iCSO:--I,,,, Figs 4d, 5d and 6d; on the other hand, the higher the aerosol concentration, the closer to 4.0 the mot-ratio

CNH;l,,,,/CSO:.-I,,,,. Thus, the aqueous phase alkalinity and the rate of SOZ oxidation are also affected by the presence of NH,-containing ‘neutral’ components as aerosols, since NH: from aerosol is distributed between the gas and liquid phases subsequent to their dissolution in the last one. E@ct of ~~q~i~ water content

As shown in Equations (28), (30) and (31), the titration curve for a closed system and for given initial conditions in a closed system is a function of the ratio volume of water to volume of system, i.e. the liquid water content, q, and PH. From this point of view, the

2703

Influence of NH, on oxidation of SO,

0a loO0

15 20 t(h) 05 10 . . . . . . . . . . . . . .1. . . . . . . . . . . . .I . . . . . . . . . . .I. . . . . . . . . . .I [NH:]

[so2-13 __-__

___-__w___

s

3

2 loE _-----

[NHtb ----

‘;;

[NH;]

.5 b F NW

_

1

_______--1

Comparison calculations

3 {Wb)z

01

0

60°

SOd(NH4NOd)

15 150 2000

1. 2 3.

2000

0.5

I

Aerosol

5.0 -

r.

nmol m-’ nmol mq3 nmol me3 I 4000

015 1 6000

2

i t(s)

15

10

20 t(h)

I

I

I

S0~2Wd%)}

= ((NH&

When q is very large, the conditions tend to the homogeneous system, the gaseous phase being negligible (see Fig. 2d). On the contrary, if q is small, the gaseous phase can be considered as a large reservoir for the water soluble species, specially for NH,. As Fig. 9 illustrates, the pH at equilibrium increases with a decrease of the liquid water content and thus, in turn the rate of SO, oxidation is affected. This illustrates how important it is to measure reliably the water content during a fog event because all solution variables are affected. of field

data of fog

event with kinetic

The simulation of a fog event requires the knowledge of input variables such as the concentrations of the gas phase, the concentrations of aerosols, the composition of condensation nuclei, and the evolution of the liquid water content, of the gas phase content and of the temperature with time. In turn, it will not be easy to simulate and calculate in advance the evolution and the modifications of the composition of the total system. The comparison of simulation data with the experimental results observed during a fog event is a valuable test and may help to improve the model. We selected two fog events, an acid fog and a ‘neutral fog (pH = 7), for this comparison. The composition of these fogs and its variations with time were described earlier (Sigg et al., 1987). Before and during the two fog

150 nmol rn+ 0.5

6.E

I

1.0

I

1.5

I

20

I

t 1)

1,2 and3

;;J_.F._?~E!$._ 1 PH

0

2000

4000

6000

.4

t(s)

Fig. 8. Dissolution of ammonium aerosols into fog droplets: effect of the presence of aerosols, {(NH,),SO,*Z(NH,NO,)}, on SO, oxidation rate and the pH. The addition of NH; in the system affects the ratio CNHd I/CNH31T or {NH31n/W3J,, andin turn lowers the aqueous phase alkalinity as well as the pH and the rate of SO, oxidation. The range of concentration of aerosols is: 1.5 x lo-*, 1.5 x lo-‘, and 2.0 x 10e6 mol me3. Initial conditions: @NH,)0= 1.0 x lo-* atm ([NH,], = 4.1 x lo-’ molL_I); (Pso, )o = 2.2 x 10eg atm ([SO,], = 9.4 x 1o-4 molt-I); (poJo = 1.5 x 10-s atm; (P~,)~ = 10-3.5 atm; q = 10M4dm-“; temperature 298 K; NH, and SO, are in closed system; 0, and CO, are in open system.

consequence of an increase of q (isothermal condensation) is a modification of the distribution of NH, between the gas and water phases: since the ratio r increases, the aqueous phase alkalinity decreases.

5.5

45

2000

4000

6000

t(s)

Fig. 9. Effect of the liquid water content (q). The titration curve for a closed system depends on q (see Equations (28), (30) and (31)): if q decreases, the pH at equilibrium increases, and in turn the rate of oxidation of SO, is affected. (1) q=2.0x10-5Lm-3; (2) q=l.O~lO-~ dmW3; (4)=5.0x 10-41m-3; (4) q=2.5 x 10T3 dm-3. Initial conditions: (pNHJO= 1.0 x lo-* atm; (psoJo= 1.14 x 10-s atm; (PO,)0 = 1.50 x 10-s atm; (pco,)0= 10-“~5atm; temperature 298 K; NH, and SO, are in closed system; 0, and CO, are in open system.

2704

PHILIPPE BEHRA er al.

events, both gas phase SO, and 0, are measured near the sampling site (Gehrig, pers. comm., 1988; see captions of Figs 10 and 11). It was observed in our investigations on fog events that the NH: concentration is often approximately equal to the equivalent sum of the NO; and SOiconcentrations. Moreover, the mol-ratios EN% I~dWIfaqj andIN-C l~a,d?‘JOS lfaqt are constant and equal to 3.6 and 1.9, respectively. Two possible explanations were proposed (Sigg et al., 1987): (i) after its absorption in the water droplets, the gas phase SO, is oxidized to SOi- in proportion to NH,; (ii) the NH:, SOi- and NO; concentrations are in relatively constant proportions in mixtures of aerosols of approximate composition ~(NH~)~SO~.ZN~~NO~ 1, i.e. SO, oxidation and neutralization occurred in the aerosols, prior to their becoming dissolved in the fogwater droplets. The aerosols provide a substantial fraction of these components in the fogwater. The presence of mainly (NH,),SO, in continental samples was also reported by Tu and Kanapilly (I978).

5.0

4.0 u a. I, 3.0

2.0

I

I

I

I

22 09 86

-

[SO$(e) field data

----- [SO$-l(c) calculated

Acid fog

Advection of HCI, from stack gases of a refuse incineration plant located ca 3 km from the sampling point, affected the pH and the composition of the fog. The larger the pulses of HCl, the lower the pH: at low pH values, the concentration of protons is nearly equal to the Cl- concentration, i.e. HCl is regulating the pH. For our model calculations (Fig. lo), we used as input data the concentrations of the gases (NH,, HCI, HNO,) and the liquid water content, and assumed a system closed with respect to NH, and SO,, and open with respect to 0, and CO,. At initial conditions, the ratio (pNH3)0to (p,,),, is equal to 1, i.e. the same conditions as depicted in Fig. 4 ((pNHa)*< 2(pso,),). In this case, we showed that the total atmospheric alkatinity is negative (Fig. 3). From Fig. 4, we can predict that (i) if pH < 5, the rate of SO, oxidation by 0, is slow, and (ii) when pH < 4, the aqueous phase acidity is not due to H,SO, but to other strong acids. The pH-range of water droplets is 3.05-4.61; the latter value is the pH at the beginning of the fog event. Because of this low pH, relatively little SO, became oxidized subsequent to the fog formation. On the other hand, we applied to SOi- the empirical scavenging factor F defined as (Fuzzi et al., 1988): (33) 4= 2 Since [SOi-](,, = 1.7 x 10-4molC-‘, x 10-4Lm-3(Siggetal., 1987),and{SO:J,aej= 1.7 3 (Ruprecht and Sigg, 1989), F is equal x lo-‘molmto 0.2. This value is in the same order of magnitude than data reported by Fuzzi et al. (1988). According to these remarks, an incipient concentration of SO:equal to 1.7 x tom4 mol P- f corresponding with the

Fig. 10. Comparison of field data with kinetic calculations: acid fog by absorption of HCl(22.09.86, data after Sigg et al., 1987);(a) pH, pC1 vs time, and (b) SO:-, and q vs time (e) Measured field data, and (c) results of the computations. Initial conditions: (pso,fo = 4.2 [SO.$-lo = 1.7 x fOv4 mol Lp- ‘; (pNWJO x low8 atm; = 4.2 x 10e9 atm; [NO;], = 3.5 x 10e4 mole-‘; (poJO = 5.1 x 10e9 atm; (pco,)0 = 3.0 x lo-’ atm; NH, and SO, are in closed system; 0, and CO, are in open system. The concentrations of NHs, HCI, HNO, and the liquid water content used in the program varied during the simulation as they varied during the fog event. Temperature: 283 K (02~5~35~, 284 K (O635-1015).

measured [SO:-]0 The two parameters ~m~sition of the were (i) HCl that

was used in the computed

model. that had the greatest effect on the fog and of its temporal variations regulates H’ concentration, the

Influence of NH, on oxidation of SO,

latter coinciding with Cl- concentration (Fig. Ma); and (ii) the liquid water content, i.e. the density of the fog, whose variations with time essentially determined the variations in the concentrations of ions which are present in constant proportions (Fig. lob). The results of the simulation confirm that the low pH of the water droplets precluded the oxidation of the SO,; thus the conditions before the formation of the fog, especially the compositions of the aerosols, that were then taken up by the water droplets, determine the proportions of the different components in the fogwater, particularly the NH:, SO:- and NO; concentrations.

-

measuring 4 and in sampling fogwater during fog dissipation. There are some additional minor inconsistencies between field data and model predictions in the latter part of the fog event. Although the temporal devel-

opment of the observed con~ntration of SOi- agrees with the calculation, there are two deviations: (i) the SO, measured in the water phase is higher than that

\ \

pti (e) held data pH (cl ulcutated

--

\ \, __-a.--

5~02-I,

‘Neutral fog’ This ‘experiment’ is particularly interesting as it demonstrates the mediating role of NH,. Indeed, because of the continued release of gas phase NH3, the excess of NRS was su%cient to maintain the pH close to 7. We were justified in assuming that the system is open with respect to NH3, i.e. its partial pressure is nearly constant over the entire duration of the fog. The Cl- and NH3 concentrations, and the liquid water content used in the program, varied during the simulation as they varied with the density of the fog during the event. The results presented in Fig. 11 are obtained with the assumption that the system is closed with respect to SO,, and the initial concentration of SO:- is equal to 0. At initial conditions, the ratio @NM,)0to (pso,)e is much greater than 2 (see Fig. 6). Thus, the total atmospheric alkalinity is positive (Fig. 3). According to Fig. 6, we can predict that pH > 6, and the rate of SO2 oxidation by O3 is very fast. Thus, all the SO, is exhausted by oxidation by O,, and transform into SO:-, what it is shown in Fig. 1lb. The agreement between field data and computed results is good for pH, NH: and SO:- : we can simulate the variations of the concentrations with time. Contrary to the previous case, i.e. acid fog, we obtain the expected values for the mol-ratio [NH: ]~~~~/[SO~-]~*~~about (3.5) by assuming that all the SOi- is due to the oxidation of the SO, present at time 0. Two conditions have to be combinedz a sufficient amount of SO, and a quite high pH( > 6). The first assumption of Sigg et al. (1987) is in accord with the constant ratios observed in this ‘neutral’ fog. As in the previous case, the role of the liquid water content is evident. When the fog disappears at 8.00 p.m., i.e. 4 decreases, a discrepancy between field data and calculated concentrations (Figs 10 and 11) is probably due to the difficulties in

2705

7-

C-40”

I

10-4_

SO?,- (e) . . .._.. _ SOa- (c) --NHf (e) -.-_ I NH2 I (c)

i

-

”

! 10-J !

I I

L

\

q

2

fietd data calculated field data ~l~ulated

5

\

field data

1 4

1

, 6

I 8

10Wd

Fig. II. Comparison of field data with kinetic calculations: ‘neutral’ fog by absorption of NH3 (24.09.86, data after Sigg et cf., 1987);(a) pH vs time, and (b) SO: - , NH: and Qvs time.(e) Measured field data and(c) results of the computations. Initial conditions: (pso2h = 2.9 x lO-p atm; [SOi-10 = 0, (pNH&= 1.0 x 1O-s atm; [NO; Jo = 4.8 x lo-* molC-‘;(po,)c = 1.9 x 10m9atm; (pm,),, - 3.0 x 10W4atm; SO, is in closed system; NH,, 0, and CO2 are in open system. The concentrations of Cl-, NH, and the liquid water content used in the program varied during the simulation as they varied during the fog event. Temperature: 283 K (0235-0805), 285 K (0805-1005).

calculated from SOz(sss) and Henry coefficient; and (ii) the calculated total amount of SOi- produced during this fog event depends also on the assumption of an open or closed system for SO,. Assuming an open system for SO,, the pH is lower than the

2706

PHILIPPEBEHRAet ol.

measured one but the total concentration of SOiagrees then with the SOi- measured in the aerosols after the fog event. At this time, we cannot decide whether these deviations, occurring only at the end of the fog event, are due to analytical complications (liquid water content, aerosol analyses during and after the fog event, S02(gas,-measurement) or to transport limitation in SO,- or O,-absorption.

Durham J. L., Barnes H. M. and Overton J. H. Jr. (1984) Acidification of rain by oxidation of dissolved sulfur dioxide and absorption of nitric acid. In Chemistry of Particles, Fogs and Rain (edited by Durham J. L.), pp. 197-235. Butterworth, Boston. Erickson R. E., Yates L. M., Clark R. L. and McEwen D. (1977) The reaction of sulfur dioxide with ozone in water and its possible atmospheric significance. Atmospheric Enuironment 11, 813-817. Finlayson-Pitts B. J. and Pitts J. N. Jr. (1986) Atmospheric Chemistry,

CONCLUSIONS

A model on the oxidation of SO, by 0, in atmospheric aqueous phase (fog, cloud, rain water for urban conditions and in wintertime) shows that NH, plays a fundamental role in neutralizing the acidity produced by the strongly pH-dependent oxidation of SO,; the amount of available NH, and its partition between the gaseous and aqueous phases (depending on the pH of the aqueous phase, and in a closed system on the liquid water content) regulates the alkalinity of the aqueous phase, and in turn drives the production of SO:-. If in the initial air volume SO2 is present in excess with respect to NH,, the production of SOiwill be limited by acidification of the aqueous phase after exhaustion of the buffer capacity by NH,; if on the contrary NH, is present in excess, the production of SO:- is very effective and may exhaust the available S02. The presence of other acid gases (HCI, HNO, , etc), of ‘neutral’ ((NH,),SO,, NH,NO, , etc) aerosol components, and of basic (CaCO,) components affects the pH of the aqueous phase and the dissolution equilibria of NH, and SOz , and influences thus the production of SOifrom SO,. Thus, the mol-ratio [NH: ],,,/[SO:Jnqj gives informations about the source of SOi- present in fog water droplets. If the concentrations of (NH&S0,~2NH,N03 aerosols are negligible with respect to the gaseous NH, concentration, the oxidation of SO, by O3 is the main source of SOi- in the aqueous phase, and the mol-ratio [NHt],,,,/[SO:-I,,,, tends to 2. On the contrary, if the concentrations of (NH,),SO,.ZNH,NO, aerosols are larger than the gaseous NH3 concentration, the main part of aqueous SO:- is due to the dissolution of these aerosols, and the mol-ratio [NH:],,,,/ [SO:-]<,,,, tends to 4. The comparison of field measurements on the composition of radiation fogs with model computations supports the reaction sequence postulated. Acknowledgement-We would like to thank Jiirg Ho&n6 for stimulating discussions. Appreciiition is expressed to Mrs H. Bolliger and U. Huser for drawing the pictures. This research was supported by Schweizerischer Nationalfond (National Programme on Air Resources NFP-14).

REFERENCES Charlson R. J. and Rodhe H. (1982) Factors controlling the acidity of natural rainwater. Nature 295, 683685.

Fundamentals

and Experimental

Techniques.

Wiley, New York. Fuzzi S., Orsi G., Nardini G., Facchini M. C., McLaren S., McLaren E. and Mariotti M. (1988) Heterogeneous processes in the PO Valley radiation fog. J. geophys. Rex 93, 11,141-11,151. Hoffmann M. R. (1986) On the kinetics and mechanism of oxidation of aquatic sulfur dioxide by ozone. Atmospheric Enoironment 20, 1145-l 154. Hoffmann M. R. and Jacob D. J. (1984) Kinetics and mechanisms of the catalytic oxidation of dissolved sulfur dioxide in aqueous solution: and application to nighttime fogwater chemistry. In SO,, NO and NO, Oxidation Mechanisms: Atmospheric Considerations (edited by Calvert J. G.), pp. 101-172. Butterworth, Boston. Hoignt J. (1988) Bildung und chemische Eedeutung von Photooxydantien in der wlssrigen Phase. Proc. ARGE ALP Internationales Symposium Verteilung und Wirkung van Photooxydantien im Alpenruum. Garmisch-Partenkirthen, 11-15 April 1988, 166175. Hoigni J., Bader H., Haag W. R. and Staehelin J. (1985) Rate constants of reactions of ozone with organic and inorganic compounds in water-III. Inorganic compounds and radicals. Water Res. 19, 993-1004. Jacob D. J. (1986) Chemistry of OH in remote clouds and its role in the production of formic acid and peroxymonosulfate. J. geophys. Res. 91, 9807-9826. Jacob D. J. and Hoffmann M. R. (1983) A dynamical model for the production of H+, NO;, and SO:- in urban fog. J. geophys. Res. 88, 6611-6621. Jacob D. J., Munger J. W., Waldman J. M. and Hoffmann M. R. (1986) The H,SO,-HNO,-NH, system at high humidities and in fogs: 1. Spatial and temporal patterns-in the San Joaauin Vallev of California. J. aeouhvs. Res. 91, _ ._ 107j-1088. Johnson C. A., Sigg L. and Zobrist J. (1987) Case studies on the chemical composition of fogwater: the influence of local gaseous emissions. Atmospheric Environment 21, 2365-2374. Junge Chr. E. and Ryan T. G. (1958) Study of the SO, oxidation and its role in atmospheric chemistry. Q. Jf R. met. Sot. 84, 46-55.

Kok G. L., Gitlin S. N. and Lazrus A. L. (1986) Kinetics of the formation and decomposition of hydroxymethanesulfotane. J. geophys. Res. 91,2801-2804. Lamb D., Miller D. F., Robinson N. F. and Gertler A. W. (1987) The importance of liquid water concentration in the atmospheric oxidation of SO,. Atmospheric Environment 21, 2333-2344.

Liljestrand H. M. (1985) Average rainwater pH, concepts of atmospheric acidity, and buffering in open systems. Atmospheric Enoiron&nt 19,487-49$. McKav H. A. C. (1971) The atmosuheric oxidation ofsulDhur dioxide in water droplets in presence of ammonia: Atmospheric Environment 5, 7-14.

Munger J. W., Jacob D. J., Waldman J. M. and Hoffmann M. R. (1983) Fogwater chemistry in an urban atmosphere. J. geophys. Res. 88, 5109-5121. Olszyna K. J., Meagher J. F. and Bailey E. M. (1988) Gasphase, cloud and rain-water measurements of hydrogen peroxide at a high-elevation site. Atmospheric Enufronment 22, 1699-1706.

Penkett S. A., Jones B. M. R. and Eggleton A. E. J. (1979) A

Influence of NH, on oxidation of SO, study of SO, oxidation in stored rainwater samples. Environment 13, 139-147. Ruprecht H. and Sigg L. (1989) Interactions of aerosols (ammonium sulfate, ammonium nitrate and ammonium chloride) and of gases (HCl, HNO,) with fogwater. Atmospheric Enoironment (submitted). Schuurkes J. A. A. R., Maenen M. M. J. and Roelofs J. G. M. (1988) Chemical characteristics of precipitation in NH,affected areas. Atmospheric Environment 22, 1689-1698. Schwartz S. E. (1986) Mass-transport considerations pertinent to aqueous phase reactions of gases in liquid-water clouds. In Chemistry of Multiphase Atmospheric Systems (edited by Jaeschke W.), pp. 415-471. Springer-Verlag, Berlin. Schwartz S. E. (1988) Mass-transport limitation to the rate of in-cloud oxidation of SO,: re-examination in the light of new data. Atmospheric Environment 22, 2491-2499. Schwartz S. E. and Freiberg J. E. (1981) Mass-transport limitation to the rate of reaction of gases in liquid droplets: application to oxidation of ‘SO, in aqueous solutions. Atmospheric Environment 15, 1129-l 144. Schwartz S. E. and White W. H. (1983) Kinetics of reactive dissolution of nitrogen oxides into aqueous solution. In Atmospheric

Trace Atmospheric Constituents: Properties, Transformations & Fates (edited by Schwartz S. E.), pp. 1-116.

Wiley, New York. Scott W. D. and Hobbs P. V. (1967) The formation of sulfate in water drops. J. atmos. Sci. 24, 54-57. Seinfeld J. H. (1986) Atmospheric Chemistry and Physics ofAir Pollution. Wiley, New York. Sibony M. and Mardon J. Cl. (1984) M6thodes num&iques de r&solution d’bquations diff&entielles. In Analyse Numkrique II: Approximations et equations Diffirentielles, pp. V.l-V.118. Hermann, Paris.

2101

Sigg L., Stumm W., Zobrist J. and Ziircher F. (1987) The chemistry of fog: factors regulating its composition. Chimia 41, 159-165.

Stelson A. W., Friedlander S. K. and Seinfeld J. H. (1979) A note on the equilibrium relationship between ammonia and nitric acid and particulate ammonium nitrate. Atmospheric Environmenr 13, 369-371.

Stelson A. W. and Seinfeld J. H. (1982) Thermodynamic prediction of the water activity, NH&NO, dissociation constant, density and refractive index for the NH,NO,+NH,),SO,-H,O system at 25°C. Atmospheric Environment 16, 2507-2514.

Stumm W. and Morgan J. J. (1981) Aquatic Chemistry, ed. Wiley, New York. Tu K. W. and Kanapilly G. M. (1978) Generation characterization of submicron ammonium sulfate ammonium hydrogen sulfate aerosols. Atmospheric

2nd and and En-

oironment 12, 1623-1629.

Waldman J. M. and Hoffmann M. R. (1987) Depositional aspects of pollutant behavior in fog and intercepted clouds. In Sources and Fates of Aquatic Pollutants (edited by Hites R. A. and Eisenreich S. J.), pp. 7&129. Advances in Chemistry Series, Amer. Chem. Sot., Washington. Westall J. (1979) MICROQL-I: A chemical equilibrium program in BASIC. Report, EAWAG, Diibe.nd&f (CH). Westall J. C., Zachary J. L. and Morel F. M. M. (1976) MINEQL, a computer program for the calculation of chemical equilibriim co&p&ition of aqueous systems. TN-18, Parsons Laboratory, M.I.T., Cambridge, MA. Ziircher F. and Gisler B. (1987) Beitrag von Ammonium zur nasse.n Deposition von Schwefelverbindungen. Proc. 4th Europ. Sympos. on Physico-Chemical Behaviour of Atmospheric Pollutants (edited by Angeletti G., Restelli G.),

pp. 48&488. D. Reidel, Dordrecht.