Vous êtes sur la page 1sur 21

Journal of Atmospheric Chemistry 16: 1-21, I993.

1
© 1993 Kluwer Academic Publishers. Prinwd in the Netherlands.

Mass Transfer at the Air/Water Interface:


Mass Accommodation Coefficients of SO2,
HNO3, NO 2 and NH 3
J. L. P O N C H E , Ch. G E O R G E , and Ph. M I R A B E L
Chemistry Department, Universitf ~Louis Pasteu5 1 rue Blaise Pascal, 67000 Strasboulg, France

(Received: 2 December 1991; in final form: 30 April 1992)

Abstract. We present here experimental determinations of mass accommodation coefficients fi using a


low pressure tube reactor in which monodispersed droplets, generated by a vibrating orifice, are
brought into contact with known amounts of trace gases. The uptake of the gases and the accommoda-
tion coefficient are determined by chemical analysis of the aqueous phase.
" ~ report in this article measurements of/3~,~p= (6.0 -+0.8) x 10 -2 at 298 K and with a total pres-
sure of 38 Torr for SO~,, (5.0 -+ 1.0) x 10 -z at 297 K and total pressure of 52 "Ibrr for HNO3,
(1.5 -+ 0.6) x 10 --~ at 298 K and total pressure of 50 Torr for NO2, (2.4 -+ 1.0) x 10 -~ at 290 K and total
pressure of 70 Tort for NH 3.
These values are corrected for mass transport limitations in the gas phase leading to /3 =
(i.3 ± 0 . t ) x 10 -~ (298 K) for SO2, (1.1 ±0.1) x t0 -~ (298 K) for HNO3, (9.7 +_0.9) x 10 -2 (290 K)
for NH3, (1.5 ± 0.8) x I0 ~'~ (298 K) for NO 2 but this last value should not be considered as the true
value of/3 for NO a because of possible chemical interferences.
Results are discussed in temps of experimental conditions which determine the presence of limita-
tions on the mass transport rates of gaseous species into an aqueous phase, which permits the correc-
tion of the experimental values.

Key words. Mass accommodation coefficient, sticking coefficient, absorption by monodispersed drop-
lets, HNO3, NO2, SO2, NH3.

1. Introduction
Mechanisms of heterogenous processes are fundamental to the understanding of
the chemical transformations and reactions of trace gases under amlospheric con-
ditions. The incorporation of trace gases into droplets is a very efficient process in
the removal of natural and anthropogenic trace gases from the atmosphere. One of
the difficulties of measuring these coefficients is to first obtain clean and well
defined water surfaces with constant physico-chemical properties. Another diffi-
culty is due to the fact that gas to liquid mass transfer is a multistep process in
which it is difficult to isolate each step. One can consider the following subpro-
cesses:

(t) Diffusion of the molecules in the gas phase towards the liquid surface,
(2) mass transfer across the interface vapor/liquid,
(3) hydrolysis, ionization or chemical reactions in the liquid phase (if any),
(4) diffusion of the species in the liquid phase.
2 J.L. PONCHE, Ch. GEORGE, AND Ph. MIRABEL

Subprocesses 1 and 4 are governed by diffusion, in both the gas and liquid
phases. The second process, transport across the interface is characterized by the
coefficient /3 known as the mass accommodation coefficient (or sticking coeffi-
cient), the determination of which is the goal of our work. Under special condi-
tions, it is possible to kinetically separate these four processes. Direct measurement
of the coefficient/3 is possible when step 2 constitutes the limiting factor of the
overall process (Schwartz, 1986).
For droplets of the liquid phase, this coefficient/3 is defined as:
number of molecules absorbed through the interface
/3 = (1)
number of molecular collisions with the surface of droplets
/3 is thus the effective fraction of molecules of trace gases absorbed at the interface.
However, the experimentally measured mass accommodation coefficient is a lower
limit to the actual coefficient due to limitations of the uptake rate by gas phase
diffusion, surface saturation or chemical reactions.
The determination of the mass accommodation coefficient/3 permits to identify',
under atmospheric conditions, the rate limiting step for the incorporation of trace
gases into liquid aerosols, fog, or cloud droplets. When the value of fl is larger than
approximately 0.01, main limitation to the uptake is diffusion in the gas phase
(Schwartz, 1986). However, for lower values of/3, mass transfer at the gas/liquid
interface can become the limiting step.
Interest in/3 originated with studies of the evaporation and condensation coeffi-
cients of pure liquids. Ma W attempts have been made to determine this coefficient
/3 for water condensing on water surfaces. It is now generally recognized that the
magnitude of/3 for water surfaces is close to unity (Schwartz, 1984, 1986). How-
ever, many older experiments lead to values much lower than unity. For example,
Gamier et al. (1987) measured a condensation coefficient of 10 -2 for water on
liquid droplets containing NaC1. Other earlier measurements are reported by
Pruppacher (1978).
With regard to the physical chemistry of trace gas uptakes by water surfaces,
only few experimental data exist today. For example, recent experimental determi-
nations of mass accommodation coefficients have been performed for SO 2 and
H202 (Gardner et al., 1987; Worsnop et at., 1989), for HNO3, HC1, N20 s (Van
Doren et al., 1990; Kirchner et al., 1990), for NO2 (Lee and Tang, 1988), for HO2
radicals (Mozurkewitch et al., 1987), and for NH 3 (Larson and Taylor, 1983;
Richardson and Hightower, 1987; Harrison et al., 1990).
The uptake rate can be limited by several factors. In fact, for highly soluble
gases, the limitation originates in the diffusion in the gas phase. The uptake near the
droplet could be such that diffusion in the gas could be insufficient to maintain a
constant concentration around the droplets. Also, if the diffusion rate in the liquid
is smaller than the rate of uptake near the interface, the liquid surface can rapidly
saturate, so that desorption of gas can occur. In these two cases, the effect is to
decrease the value of the experimental mass accommodation coefficient fi~xp com-
pared with that in the absence of limitation (Gardner et al., 1987).
MASS TRANSFER AT THE AIR/WATER INTERFACE 3

In order to minimize these limitations, we use an experimental method that com-


bines low pressure (to increase diffusion rates in the gas phase) and short drop ex-
posure times to limit, as much as possible, saturation of the liquid surfaces.
In this paper, we report measurements of/~exp for the four gases SO 2, HNO3, NO2, at
298 K, and NH 3 at 290 K.

2. Experimental
A principal difficulty involved the preparation of a rapidly and constantly renewed
liquid surface. Lee and Tang (1988) performed this experiment for NO2 using macro-
scopic surfaces of liquids containing chemical sinks and exposure times of the order of
several seconds. Kirchner et at. (1990), use micrometric jets and exposure or inter-
action times in the range 0.03-1 ms. We preferred to produce micrometric droplet
stream (diameters in the range 80-110 gin), having short contact times with the trace
gases of the order of a few milliseconds.
The method used here follows closely that of Gardner et al. (1987). The experi-
mental apparatus (Figure 1) is a low pressure flow tube reactor which consists of five
successive chambers: (1) a droplet generating chamber, (2) a buffer chamber contain-
ing liquid water on the wall and an electromagnetic shutter which controls access to the
next chamber, (3) the interaction chamber where trace gases and droplets interact, (4)
another buffer chamber, and (5) the droplet collection chamber. The stream of drop-
lets and the flow of trace gases and carrier gas are co-axial.
The droplets are generated using the vibrating orifice method (Plateau, 1873;
Rayleigh, 1878; Str6m, 1969; Berglund and Liu, 1973). Filtered water is forced
through a calibrated orifice, formed in stainless steel sealed to a piezoelectric ceramic
(Quartz et Silice, S.A.) connected to a frequency generator. When the generator is
turned on, the vibrations control the disintegration of the water jet, and for particular
frequencies, a monodisperse stream of droplets is produced. The optimal frequency
for obtaining monodispersed droplets is given by Rayleigh (1879) as

Vi'et (2)
fopt 4.508 Diet '
where viet and Dies are, respectively, the water jet velocity and the diameter of the jet.
Since thej et passes through the orifice and disintegrates with negligible loss of ener-
gy, one can assume that viet = va and Diet = D where v a and D are, respectively, the
velocity of the droplets and the diameter of the calibrated orifice. This leads to
Vd
fopt 4.5 D " (3)

In fact, monodispersed droplets are produced for frequencies f in the range


(Schneider and Hendricks, 1964):

~)iet < f < Ujet


7Diet 3.5 Diet " (4)
4 J.L. PONCHE, Ch. GEORGE, AND Ph. MIRABEL

generation of the droplets

buffer zone Nitrogen


Nitrogen+trace gas

\IV

interaction chamber

--I
buffer zone Argon

collection chamber

Fig. 1. Schematic of experimental apparatus.

The velocity and the diameter of the droplets, in the case of a monodispersed
droplet stream are given by

4 F1 (5)
va = srD 2

and

d = k(6nfF~)v3, (6)

where G is the liquid flow rate.


For frequencies in the range of 40 to 80 kHz, and with a back pressure of 1 to 4
bars forcing the water through an orifice of 41.1 + 0.3 or 76 _+0.4 ~tm in diameter,
70-110 btm diameter droplets are generated having velocities of 480 to 2000 cm/s.
The diameters and velocities of the droplets have been measured at the Institfit
ffir Thermodynamik und Raumfahrt in Stuttgart (Germany), using laser scattering
MASS TRANSFER AT THE AIR/WATER INTERFACE 5

(Koenig et al., 1986). The measurements show an agreement between theoretical


and measured diameters to within less than 3%.
During an experiment, the stream of monodispersed droplets is visualized by a
stroboscopic lamp connected to the frequency generator. After leaving the genera-
tion chamber, the droplets enter the buffer zone where the presence of liquid water
minimizes evaporation and maintains them in equilibrium with the water vapor. A
thermocouple permits control of temperature inside. When the electromagnetic
shutter is open, the stream enters the interaction chamber, where the N 2 carrier gas
containing the trace gas flows. The mean velocity of the carrier gas in the inter-
action chamber can be varied from 20 to 60 cm/s. The interaction tubes are 0.75
cm in radius and 7.5 to 21.5 cm in length. Combining the different droplet velocities
with the different lengths provides interaction times between 3.7 and 45 ms.
The train of droplets leaves the interaction chamber through a 3 mm diameter
diaphragm and enters a second buffer zone, where argon is injected to sweep out
the droplets. In the last chamber; these droplets are collected in a Pyrex flask and
immediately frozen in liquid nitrogen to preserve them from possible contamina-
tion.
After each experiment, which lasts about 5 to 10 min, the small flask is isolated
and separated from the reactor. The fraction of trace gas incorporated in the liquid
phase is then determined by means of 'ion chromatography. In some cases (NO2
and SO2), the dissolved species had to be oxidized before analysis.
The pressures inside the different chambers are measured by three pressure sen-
sors (Eurosensor), and electronically controlled valves insure the uniformity, of
pressure across the apparatus. Our working pressures are typically in the range of
35-70 Tort, with pressure differences between the chambers never exceeding 0.3
Torr.
In the two buffer chambers, a counter flow permits the fixing of the length of
interaction between trace gas and droplets. It involves injecting a neutral gas
(nitrogen, argon) near the diaphragm of entering and exiting droplets in the inter-
action chamber. To work with this system it is necessary to have a low pressure
difference (0.5 Torr max.) between the interaction and the contiguous chambers.
The trace gases used are of two kinds. Known quantities of high purity SO2 (39
ppm) and NO 2 (36.8 ppm) were prepared by Alphagaz and premixed with nitro-
gen. These concentrations were known with an accuracy of 5% by Alphagaz (L~Xir
Liquide, S.A.) before they were used directly or mixed with known fluxes of nitro-
gen. For HNO3 and N H 3 , a carrier gas (nitrogen) carried away saturated vapor of
solution and by mixing with a second flux of nitrogen, we could obtain different
concentrations of trace gases in nitrogen. These concentrations were measured
before and after each experiment by means of ion chromatography. In the case of
HNO3, it is possible to provide a good agreement of the observed concentrations
with those predicted by the compilation of Jaecker-Voirol et al. (1990).
Carrier gas is saturated with water vapor through a thermostated water bubbler
just before entering in the two buffer chambers and in the interaction chamber. The
6 J.L. PONCHE, Ch. GEORGE, AND Ph. MIRABEL

temperature of the bubbler is the same than temperature inside the reactor. Fur-
thermore, the bottom of the first buffer zone is permanently covered with a thin
pool of liquid water. So we can assume that the droplet stream is in equilibrium
with ambient water vapor. Values of the water vapor pressure as a function of tem-
perature are given by Tabatta (1973).
As noted before, the mass accommodation coefficient is defined as the ratio of
the number of molecules absorbed by droplets to the total number of molecular
collisions with droplets.
From the kinetic theory of gases, we can express the rate of collision between
molecules and droplets as"

1 (VA} N S d hA, O, (7)

where (vA} represents the thermal mean molecular velocity of the trace gas A (cm/
s), N is the number of droplets in the interaction chamber, J~'d is the total area of
the N droplets of area Sd (cm2), and hA. 0 is the number density of A (molecules
c m -3) in the carrier gas flow before interaction with the droplets.
The total number of molecules absorbed by the droplets, n~b~, is given by the
expression:

= F. An , o, (s)

where Fg is the volume rate of flow (cm 3 s -1) of the carrier gas, AnA, 0 = hA, 0 -- rtA, 1
is the number of molecules absorbed per c m 3, and nA, 1 is the trace gas density after
an interaction length l (l represents the length of the interaction tube).
The coefficient ~exp can be expressed as

Fg A hA, 0 (9)
/~exp ~ i

This expression is valid only if absorption of the trace gas A by the droplets is
small, implying that the radial concentration gradient is small. However, a concen-
tration gradient develops along the length of the interaction tube due to the absorp-
tion of the trace gas by the droplets and by the walls (wall losses).
To express the gradient due to the absorption by the droplets, one can assume a
first-order reaction (Gardner et al., 198 7) such that
dnA = k~ ) hA, (10)
dx v
where k~ ) represents the first-order rate constant for absorption of the trace gas A
by the droplets and ~ is the average velocity of the carrier gas in the interaction
tube of radius r t 05 -- ]~/~r2).
Integrating over the length l of the interaction tube, Equation (9) yields
MASS TRANSFER AT TIIE AIR/WATER INTERFACE 7
_ = k~ )
In nA,0 - - l, (11)
HA, 1 ?-)
with the first-order rate constant k~ ) expressed by

k ~ ) _- (vA) flexp.NSd (12)


4V~ '
where 17I is the volume of the interaction chamber.
Finally, we obtain an expression for flexp as

/~exp 1 In (13)

where (VA) is given by the kinetic theory of gases


( S R T ) I/2 (14)

in which Ma is the molar mass of the gas A, T the temperature, and R the universal
gas constant.
#exp in Equation (13) can be expressed as a function of our experimental meas-
urables as

/3~p = 6.5135 RF/2i--D2 In nA, o . (15)


F/A, 1
This relation supposes two conditions:

(i) Desorption of trace gas from the droplet has to be negligible. This implies
that the contact time between the trace gas and droplet has to be short to
avoid saturation of the interface. This point will be discussed for each trace
gas in the result section.
(ii) The flow rate of the carrier gas has to be faster than diffusion in the gas
phase in order to minimize radial concentration gradients. This implies that
the uptake by the droplets must be slow compared to the rate of supply and
that wall losses are small so that the gas phase concentration changes are
negligible.

This last point was checked by changing the trace gas concentration in the
carrier gas (up to a factor 10 for HNO3, see Figure 3) or by varying the flow veloci-
ty by a factor 2 (25-50 cm s-l). No significant variation in the values of #exp has
ever been observed confirming that condition (ii) is always fulfilled.
Typical experimental conditions and droplet parameters are summarized in
Tables I and II. For SOz and HNO3 more than 50 experimental runs, in each case,
were performed while for NO2 and NH3, the number exceeded 80.
Our determination of fle×p is based on the measurement of the trace gas dis-
solved in the aqueous phase. For HNO 3 and NH 3 our analytical method allows
8 J.L. PONCHE, Ch. GEORGE, AND Ph. MIRABEL

I0 o
Mass accommodation coefficient

S02
I0-I oH= 11

10-2
:t
10-~ Interaction time (seconds) i
O. OOE+OO 3.00E-°3 6.00E'°3 9.00E-O3 1.20E'°2 1.50E-°2
Fig. 2. Plot of the observed mass accommodation coefficient fi~p versus interaction time for SO2.
Results are obtained at 298 K, with a total pressure of 38 Torr and for a gas phase concentration of 28
ppm (lower curve) and 2.8 ppm (upper curves). Theoretical curves are calculated from Equation (29).

Table I. Typical experimental conditions

Parameters Values

T internal reactor temperature 290 to 298 K


P total pressure (cartier gas + H20) 35-70 Tort
PH~o partial pressure vapor of HzO 18-32 Torr
hA, a initial concentration of gas traces 6.9 x 1011 - 2 x 10 ~4 molecules cm-3
(VA} average molecular thermal velocity SO2:31100-31300 cm s-~
HNO3:31200-31600 cm s-1
NO2:36500-37000 cm s-I
NH3:59800-60000 cm s-1
73flow velocity of carrier gas in the interaction
chamber 25-50 cm s-1
r, radius of the interaction chamber 7.5 x 10 -1 cm
t interaction lengths 7.5; 9.5; 13.5; 21.5 cm
pH0 initial pH of droplets 4; 5.6; 6; 9; 11
D orifice diameter 4.1 x 10-3 and 7.6 x 10 -3 cm
F I liquid flow rate 1.5 x 10 - a - 3 x 10 -2 cm 3 s -~
f frequency 40-80 kHz

direct m e a s u r e m e n t of N O ? a n d NH~. B u t f o r SO2 a n d N O 2 , we have to oxidize the


s o l u t i o n b e f o r e i n t r o d u c i n g it i n t o the c h r o m a t o g r a p h y c o l u m n since o u r c o l u m n
c o u l d o n l y d e t e c t SO4z- a n d NO3. T h u s , the liquid p h a s e is oxidized b y the a d d i t i o n
of small q u a n t i t i e s of c o n c e n t r a t e d H 2 0 2 to b e sure to c o n v e r t all S a n d N as S(VI)
a n d N(V).
MASS T R A N S F E R AT T H E A I R / W A T E R I N T E R F A C E

Table II. Experiments realized for the four gases

Trace gas Concentration Initial value of Interact. lengths (cm)


range (ppm) droplet p H 7.5 9.5 13.5 21.5

SO2 28 ± 2 4 x x x
2.8 ± 0.3 5.6 x x x
2.8 ± 0 . 3 11 x x x

HNO 3 9-101 9.5 x


101 ± 5 9.5 x x x
101 ± 5 11 x x x

NO 2 36.8±0.8 4 x x x
36.8±0.8 5.6 x x x x
36.8±0.8 11 x x x x

NH 3 115 ± 3 4.2 x x x
115±3 6.2 x x x
115+_3 11.5 x x x

For S (VI):

HSO 3 + H 2 O 2 -~ SO2OOH- + H2O (16)


S O 2 0 0 H - + H + -. H2SO 4 (17)

For N(V):

NO~ + H202 ~ NO~ + H 2 0 (18)


The characteristic times of oxidation are, respectively, for S(VI) and N(V): 6.9 x
10 -6 and 1.36 x 102 s at pH = 6 and for H202 concentration of 0.16 mole 1-~. In
the case of N(V), this time increases with pH, and as a result, the aqueous phase
must be treated to reduce the pH to 6 - 7 for the experiments in which the initial pH
is 11.

3. Results and Discussion


The limits on mass transport can be estimated for the gas phase, interface, and
liquid phase processes. It is possible to determine what step constitutes the limiting
factor for the incorporation of trace gases into droplets. Freiberg and Schwartz
(1980), Schwartz and Freiberg (1981), Schwartz (1986)have analysed this question
of mass transport limits using the characteristic times r associated with each
process. These times are given below and calculated using physical solubilities and
droplets diameter of 100 ~tm (see also Table III).
10 J, L, P O N C H E , Ch. G E O R G E , A N D Ph, M I R A B E L

Table Ili. Physical constants and characteristic times

Parameters 802 (298 K) H N O 3 (298 K) NO2 (298 K) N H 3 (290 K)

H(M/atm) 1.26 (2,1 _+ 0.3) x 10 s (1.2 _+ 0,4) x 10 -2 91.0


(a) (b) (b) (i)
Dg(cm 2 s -t) 0.132 0,160 0.192 0.230
(c, d) (d) (d) (d)
D~(cm 2 s "-1) 1.8 x 10 .5 = 2.6 x 10 -5 ~ 10 -5 = 1.9 x 10 -s
(e) (f) (g) (k)

Characteristic times (in seconds)


( t ) Gas Phase
r~ 3.2 x 10 .6 2.8 x 10 -6 2.3 x 10 -6 2.6 x 10 -6
(38 Torr) (50 Torr) (50 Torr) (70 Tort)
/:gs,~ 1,0 x 10 .4 1.44 x 10 ~ 6.8 x 10 -`7 3,7 x 10 -3
(38 Torr) (50 Torr) (50 Tort) (70 Torr)
(2) Interface
(fl = 0.1)
ri 3.2 x 10 -6 3.8 x I0 -6 1.1 x 10 -4 10 -6
r~a' 1.0 x 10-'* 2 x 10 L 3.2 x i 0 -s 1.4 x 10 --~

(3) Aqueous phase


%d 1.1 x 10 -1 7.9 x 10 -2 2.0 x 10 -1 3.2 x 10 <
r, 3 x 10 "-7 <5 x 10 -9 1.6 2 x 10-6
(e) (h) (b, i) (j)
reap: experimental
interaction time ( 4 - 1 5 ) x 10 -3 (4-15) x 10 .3 ( 4 - 1 5 ) x 10 -3 ( 1 0 - 4 5 ) x 10 -3

(a) Schwartz and Freiberg (1981);


(b) Schwartz and White (1981);
(c) D u r h a m etaL (1981);
(d) Reid and Sherwood (1966);
(e) ~¥brsnop et aL (1989);
(f) International Critical Tabtes of Numerical Data (1929);
(g) Kirchner etal. (1990);
(h) Van D o r e n et al. (1990);
(i) Seinfeld (1986) with NO2 concentration in the gas phase 2.42 x 10 -6 arm.
(j) Eigen and Schoen (1955).
(k) Leaist (1987).

- gas phase diffusion to a spherical droplet:

d
d2
rg = 12 Dg '

- saturation of the droplet by gas phase diffusion:

Tgs~t ~ H R T I:d

interracial mass transport:


2d
MASS TRANSFER AT THE AIR/WATER INTERFACE 11

- saturation of the droplet by the interfacial mass transport:

sat
r, = H R T ~:i,

- aqueous phase diffusion:


d2
d
ra 4~2Da '

- chemical reaction:
1
r~ = ~ (k pseudo-first-order rate constant).

The characteristic times of saturation of the droplet by gas phase diffusion T gsa~
and inteffacial mass transport Ti~atgave
• the timescale in which either diffusion in the
gas phase or transport across the interface control the establishment of the equi-
librium between the gas and liquid phases leading to uniform concentration in the
sat sat
liquid phase. For NH3, NO; and SO2, rg and Ti are smaller than the experi-
mental interaction time te×p meaning that phase equilibrium can be reached in our
timescate. However, the characteristic times r d for aqueous phase diffusion are
larger than texp assuring that the entire droplet will not be saturated by the dissolved
trace gases. For H N O 3 , a similar conclusion can be reached since "ggSat,Ti'~at,and va
are all larger than rexp.
In order to discuss and explain our experimental results, we also need to know
the following physical constants: the Henry's law constant H, the gas phase diffu-
sion coefficient Dg and the aqueous phase diffusion coefficient Da. All these values
are listed in Table III.
Considering these characteristic times, it becomes easy to determine which limi-
tation is involved in our measurements. We will examine the results of the four
gases separately, and discuss them with the aid of the theory of Schwartz (1986)
and the model proposed by Gardner et al. (1987).

3.1. Resutt f o r S 0 2
Experimental results, for three different values of pH (see Table II), are given in
Figure 2. These results show that the observed mass accommodation coefficient is
strongly dependent upon the interaction time texp and also upon the droplet's pH.
This can be explained by consideration of the characteristic times of the different
sub-processes.
The characteristic time for the chemical reaction r~ is much smaller than other
times, so it appears that aqueous phase reactions could not be a limiting factor. For
the gas phase, the characteristic time for diffusion r~ is much smaller than texp
meaning that the gas phase can be considered at steady state. Elsewhere, rdg de-
creases with pressure and with the radius of the droplet justifying our decision to
12 J.L. PONCHE, Ch. GEORGE, AND Ph. MIRABEL

work with small droplets and under reduced pressure. The characteristic times for
Sat
saturation of the droplet by diffusion in the gas phase rg and by the interracial
mass transport rTat are lower than rexp so that the droplet surface can be considered
at equilibrium with ambient SO2. However, the characteristic time of aqueous
phase diffusion r~ is much greater than texp implying that this process is not rapid
enough to evacuate adsorbed molecules from the droplets surfaces into the liquid
phase. As a consequence, the interface will become rapidly saturated by SO2 within
a penetration depth dx.
Such a phenomenon will be followed by re-evaporation of
802 from the interface introducing a time dependence in the observed mass
accommodation coefficient. In order to quantify this dependence one has first to
consider the real solubility of SO 2 at the interface.
Incorporation of sulfur dioxide molecules in the aqueous phase leads to the
acid-base equilibria (19-21), and allows one to define another Henry's law con-
stant Hs(w} (22). This pseudo-constant takes account of all the dissolved species
resulting from the dissociation of SO 2 in aqueous phase, and describes the effective
solubility of SO 2 in water. This constant appears to highly depend on the pH.

SO2(g) ~ SO2(aq) Hso 2 (19)

SO2(aq) -}- H 2 0 ~ HSO~ + H30 + K1 (20)

HSO3 + H 2 0 ¢- 802- +H3 O+ K2 (21)

, (
Hs(w, = Hso2 1 + [ ~ + ~
K1K1K2) (22)

These equilibria between the gas and aqueous phase change the local value of
the droplet surface pH and this in turn modifies the solubility of SO 2 at the inter-
face. If we suppose that a steady state exists at the interface, implying that SO2(g) is
in continual equilibrium with SO~(aq> it becomes possible to calculate the value of
the pH at the droplet's surface (Gardner et al.,
1987). By neglecting the second dis-
sociation (Equation (21)), and introducing the equilibrium involving water dis-
sociation (23), namely:

2 H20 ~ OH- + H3 O+ g w, (23)

one can obtain

[H+]i = E + (E 2 + K w + K 1 Hso 2 Ps02) 1/2, (24)

with

E=-~
1( [H+]o [~-V]° , (25)
MASS TRANSFER AT THE AIR/WATER INTERFACE 13

where the indices o and i correspond, respectively, to the initial state and the equi-
librium state at the surface, while Pso2 is the partial pressure of SO2 in the gas
phase. Using, K w = 10-14(298 K), K 1 = 1.74 x 10 2 M (Schwartz and Freiberg,
1981), Hso 2 = 1.26 M/atm (Schwartz and Freiberg, 1981), we can calculate the
pseudo Henry's law constant using the different surface's pH (Table IV). These cal-
culations show a strong dependence of the solubility upon the droplet's pH explain-
ing the differences observed for the three series of measurements depicted in
Figure 2.
The desorption of SO 2 from the droplet surfaces appears as a decrease of the
incorporation rate as a function of time (Gardner et al., 1987, Worsnop et al.,
(1989). This time dependence is expressed by Worsnop et al. (1989) by consider-
ing the real flux J of SO2, i.e. the difference between the influx and outflux from the
droplet:
j= (VSO2)/~exp
4 Hs(w~RT (Hso2 Pso2 - [S(IV)(t)]), (26)

where [S(ivfl is the average concentration of S(iv~ in the liquid phase. At the time t,
the penetration depth dx of Sav ~ is given by dx = (Dat) 1/2 and within this distance
from the surface, the average concentration is
Jt
[S(IV)] (Dal)l/2. (27)

It leads to an expression for time dependence of flexp (Worsnop et al., 1989):

1 (Vs@ ) <. (28)


/~exp(t) = /~exp(0~ -~- 4 H s ( w ) R T ( D J t ) 1/2

Using our experimental conditions, we obtain for the theoretical dependence of


flexp on interaction time Gp and the initial pH droplet, the expression:
( 1 75236t~ )i.
/~exp(l) = /~exp(0~ + Hs(rv) (29)
By fitting this expression to the experimental data, we can determine the value of
flexp(0), defined as the limit o f flexp(t) a s rexp approaches zero. This means that
fiexp(0) is independent of pH. This method makes it possible to determine the value
of ]~exp(SO2)in good approximation. At 298 K, we found: /~exp(SO2) = (6.0 + 0.8) x
10 -2 at 38 Torr.

3.2. Results for H N O 3


Experimental data are shown in Figures 3 and 4. They correspond to the experi-
mental conditions listed in Table II. We have chosen basic pH for the aqueous
phase in order to prevent saturation of the interface by the trace HNO 3. But solu-
bility of HNO 3 in water is so great that it appears that the initial values of pH have
14 J.L. PONCHE, Ch. GEORGE, AND Ph. MIRABEL

Table IV. Hem3' pseudo-constant for SO 2 and pH


values at the interface

PsQ (atm) pH o Hs(w) (M/atm) pit i

2.6 × 10 -7 4 76 3.53
2.6 x 10 -s 5.6 289 4.12
2.6 x 10 .8 11 3964 5.25

I0 o
Mass accommodation coefficient

HN03
10-1
0 QP .... ~ . . . . . . . . . . . 0 ~ n
ull " - 0 a

10-2

io- Gas phase concentration (p.p.m.)


I I . i i i i l

0 20 40 60 80 100

Fig. 3. Plot of the observed mass accommodation coefficient/3exp versus gas phase concentration for
HNO 3. Results are obtained at 298 K, with a total pressure of 50 Torr and for pH = 9.5.

no effect on the measured value of flexp. We conducted a first series of measure-


ments that allowed us to determine the dependence of flexp on the concentration of
the trace gas. No significant dependence was discovered. The second series of
experiments was conducted to provide data on the dependence of/3exv on texp. For
the two values of pH, we found the s a m e flexp within the experimental errors.
The characteristic time of the reaction of ionization r~ is very much smaller than
the other times, meaning that the reaction does not constitute a limiting step, The
gas phase steady-state hypothesis is justified because the interaction time [exp is
always greater than the characteristic time for diffusion ~g.d As for SO2, the aqueous
phase diffusion will limit the incorporation rate but because of the great solubility
of H N O 3 it was not possible to observe any dependence with the experimental
interaction time. The characteristic times for saturation of the droplet ~gsat and r~sat
MASS TRANSFER AT THE AIR/WATER INTERFACE 15

lass accommodation coefficient


I0 o

HNOs
• pH= 9.5

10-1 ° pH= 11

° !

i0-2 Interaction time (ms)


0 3 6 9 12 15
Fig. 4. Plot of the observed mass accommodation coefficien~ versus interaction time for PINO3.
Results are obtained at 298 K, with a totN pressure of 52 Torr, for gas phase concentration of 101 ppm
(pH = 9.5) and 50 ppm (pH = 11).

are m o r e important than texp implying that the droplet surface does not b e c o m e
saturated with gaseous HNO3. In such a situation, the experimental value will cor-
r e s p o n d to the mass a c c o m m o d a t i o n coefficient for a given pressure and t e m p e r a -
ture. WE found an experimental value flexp = (5.0 --+0.9) X 10 -2 at 298 K and 50
Torr total pressure.

3.3. Results for NO 2

T h e experimental data are illustrated in Figure 5. T h e y c o r r e s p o n d to the three


series of runs with three different values of p H and with the conditions described in
Table II.
T h e three series of experimental data were conducted to m a k e conspicuous the
interaction time d e p e n d e n c e of fiexp- It appears that p H does not have any signifi-
cant effect on the value of flexv-
In o r d e r to calculate the characteristic times of reaction Tr, one has to consider
the following reactions (Seinfeld, 1986):

NO2(g) ~ NO2(aq ) HNO z (34)

2 NO2(aq ) ~ 2 H + + N O 2 + N O 3 K 3 = 2.44 x 106 M 2 (35)

H ÷ + N O 2 ~ HNOz(~q ) K 4 = 1.96 x 103 M (36)


16 J.L. PONCHE, Ch. GEORGE, AND Ph. MIRABEL

~a55 accommodation coefficient


I0"I

Nil2 . pH= 11
I0-2
° pH= 5 . 6
" pH= 4
$ +p + " '* o 'm<~

I0-3 ~ +$
o
" '0+
o
11

"
'+ II

a0-+ , . . Interaction time (seconds)


O.OOE+°° 3.00E-°3 6.00E-°3 q. OOE43 1.20E42 1.50E42
Fig. 5. Plot of the observed mass accommodation coefficient fl~×p versus interaction time for NO 2.
Results are obtained at 298 K, with a total pressure of 50 Ton+ with a gas phase concentration in NO2
of 36.8 ppm.

For pH value larger than 3.5, HNOa concentration is much smaller than NO2
concentration and can be neglected. The characteristic time r r of reaction (35) is
under our experimental conditions, in the order of 1.6 s which is much larger than
any other characteristic time (see Table III). This implies that, for our experiments,
only reaction (34) takes place, i.e. only physical solubility of NO: occurs. Note that
NO2 has a very low physical solubility in water compared to SO2 and HNO 3.
Aqueous phase diffusion o f NO2(aq) is then the limiting process and the droplet
surface becomes rapidly saturated with NO2. The analysis developed for SO 2
should also apply in this case, except that the interracial pH remains the same as the
initial pH. However, it was impossible to fit our experimental values with theoreti-
cal curves such as those predicted by Equation (28). Even by using a value of
/3exp(0) = 1, the curves calculated from Equation (28) lie below our experimental
values.
The observed uptake rate in our system seems to greatly exceed the value pre-
dicted by the above described model. Such a discrepancy has already been ob-
served in previous studies. In fact, Borok (1960) and Beilke (1970) observed that,
at tow partial pressure of NO 2 (lower than 10 .7 atm), the uptake rate is much larger
than the value calculated by considering only the reactions (34) and (35) and also
Henry's taw constant as high as 0.55 M atm -~ has been reported. Schwartz and
White (1983) explained these facts by considering additional (but unknown) reac-
tions of NO2 with dissolved impurities. Such an explanation may also apply to our
system.
MASS TRANSFER AT THE AIR/WATER hNTERFACE 17

We present average values of fl~xp, which are given in Table V, determined by


extrapolation of the fitted experimental data of each serie to zero of time of inter-
action &~p.The average value fi~xp= (1.5 _+0.6) x 10 -3 but this value has to be con-
sidered carefully because of transport limitation and possible unknown chemical
reactions which may affect significantly fi (0).

3.4. R e s u l t s f o r N H 3
Experimental results are given in Figure 6 for three different values of pH (see
Table II). By considering the characteristic times for the different steps involved, it
appears that the hypothesis of the steady state is, in fact, verified for diffusion in the
gas phase. But it also appears that characteristic times for saturation of the droplet
sat sat
rg and r~ are smaller than rexp SO that the droplets surfaces are in equilibrium
with ammonia traces. Mass transport limitations also exist in this case (similar to
that of SO2) because of a slow aqueous phase diffusion.
To estimate the effective solubility of ammonia at the interface, one has to con-
sider the following equilibria (Leaist, 1987; Durrant et al., 1962; Berg et al., 1953):

NH3(g ) ~ N H 3 . H 2 0 HNH3, (37)

NH 3 • H20 ~ NH~ + O t t - K 5 = 1.7 x 10 -5 mole/1 (38)


Considering all these species, it is again possible to define a pseudo-Henry's law
constant HNH3as

HNH 3 = HNH~ 1+ ~ . (40)

The presence of these equilibria changes the pH on the droplet surface which
can be calculated (Table VI) using the same hypotheses as those for SO>
It appears that for the three values of pH chosen the interface is rapidIy satu-
rated, implying a decrease of the observed coefficient as a function of time. The
uptake is a convolution of absorption and desorption. Once again, invoking the
same analysis as that for SO> one obtains

flexp(t) =
(VNH31 )-1.
+ 4 H *NH3R T (,L)a/texp)
. . . . 1/2 (41)
/~exp(O)

By fitting the curves generated by Equation (41) to the experimental data, we


can determine the limiting value fie~p(NH3)= (2.4 _+1.0) X 10 -2 at 70 Torr.
For all the results presented here, there exists a limitation which may affect the
observed coefficient. Worsnop et al. (1989) have studied the effect of gas-phase
diffusion on the observed mass accommodation coefficient for substances such as
18 J.L. PONCHE, Ch. GEORGE, AND Ph. MIRABEL

Mass accommodation coefficient

10-t NH 3

~, pH= 4 . 2
Q pH = 6 . 2
~> pH= 1 1 . 5
t0-2

i0-3 , Interaction t!,me(<


0 10 20 30 40 50
Fig. 6. Plot of the observed mass accommodation coefficient versus interaction for NH 3. Results are
obtained at 290 K, with a total pressure of 70 Torr with a gas phase concentration in NH 3 of 115 ppm.

TabteV. Experimental mass accommoda-


tion coefficient for NOz

Initial value of pH fl~xp

pH = 4 (1.7 ± 0,6) X 10 -3
pH = 5.6 (1.3 ± 0.6) x 10 -3
pH = 11 (15 ±0.6) x 10 -3

Table VI. Henry's law pseudo-constant for NH 3 and


values for the pH at the interface

PNH~(atm) pH0 Hm~l atm -1) pH i

10 -5 4.2 106 10
10 -5 6.2 103 10.1
10 -5 11.5 91 11.5

SO 2 and H202 in different carrier gas. Their observations show that the observed
uptake is dependent upon the total pressure in the reactor, meaning that the accom-
modation coefficient is pressure dependent. A solution to this mass transport limi-
tation has been given by Schwartz (1986) and by Gardner et al. (1987). For the
case of a spherical droplet, the mass accommodation coefficient for a zero total
pressure/3 (P = 0), i.e. when no gas phase limitation exists, is given by
MASS TRANSFER AT THE AIR/WATER INTERFACE 19

1 1 (v)Pd
(42)
/3(P = o) /3exp(O) 8Dg '

where P is the total pressure in the reactor. Using this expression, we obtain the
corrected mass accommodation coefficient/3 (P = 0) listed in Table VII.
However, Equation (42) has been derived for a static single droplet and may not
be directly applicable to our fast moving train. Recent work by Worsnop el al.
(1989) suggest that the droplet diameter in Equation (42) should be replaced by a
corrected droplet diameter df = 1.9 D. But it should be noted that this relation can
be directly obtained by combination of Equations (2) to (6) meaning that, in our
case, Equation (42) already includes the correction for the diameter However,
further analysis must be done to completely resolve this issue.

In conclusion, the corrected values of/3exp for SO2, H N O 3 and N H 3 a r e consist-


ent with previous results. For SO 2 our value of/3 (1.3 + 0.1) x 10 -1 can be com-
pared with the value obtained by Gardner et al. (1987); 1.2 x 10 -1 at 295 K and
Worsnop et al. (1989); (1.1 + 0.2) x 10 -I at 273 K. For HNO3 the value of/3 ob-
tained here (1.1 + 0.1) x 10 -1 is consistent with that found by Van Doren el al.
(1990) namely (7.1 _+0.2) x 10 -2 at 293 K. Kirchner et al. (1990) have also deter-
mined/3exp for H N O > but with another method based on microjets of water. Their
results provide 0.01 as a lower limit. Also for N H 3 o u r value (9.7 + 0.9) x 10 -2 is
comparable to other results found in the literature. The reported accommodation
coefficients vary between 0.02 and 1 (Larson and Taylor, 1983; Richardson and
Hightower, 1987; Harrison et al., 1990). Note, however, that the reported values
were determined by measuring the evaporation rate of ammonium salts, in which
N H 3 and HC1 or H N O 3 were evaporing simultaneously. In such cases, the evapora-
tion coefficients were considered as adjustable parameters and no direct measure-
ments of the evaporation coefficient or mass accommodation coefficient were per-
formed. Finally, for NO2 our result (1.5 + 0.8) x 10 3 should be considered careful-
ly because of possible unknown chemical interferences. However, our average
value compares with that reported by Tang and Lee (1988);/3 = (6.3 + 0.7) x 10 4
at 273 K, with a different experimental method, using macroscopic surfaces and a
chemical scavenger to avoid solubility dependence.

Table VII. M a s s a c c o m m o d a t i o n coef-


ficient corrected for gas p h a s e dif-
fusion

Compounds fi ( P = O)

SO2 (1,3 _+ 0.1) x 10 -1


HNO 3 (1,1 _+ 0.1) x 10 -1
NO2 (1.5 + 0.8) x 10 3
NH 3 (9.7 -I- 0.9) x 10 -2
20 J.L. PONCHE, Ch. GEORGE, AND Ph. MIRABEL

Acknowledgements
It gives us great pleasure to thank Professor H. Reiss, UCLA, who has always
shown great interest in our work. Support of this work by the Minist~re de
l'Environnement and by the ATP Phase Atmosph6rique des Grands Cycles Bio-
g6ochimiques and by the european project EUROTRAC/HALIPP are gratefully
acknowledged.

References
Berg, D. and Patterson, A., Jr., 1953, The high field conductance of aqueous solutions of ammonia at
25 °C,J. Am. (2hem. Soc. 75, 5731-5733.
Berglund, R. N., and Liu, B. Y. H., 1973, Generation of monodisperse aerosol standards, Environ. Sci.
Technol. 7 (2), 147-153.
Borok, M. T., 1960, Dependence of the degree of absorption of nitrogen dioxide in water on its con-
centration in a gaseous mixture, J. Appl. Chem, USSR 33, 1761-1766.
Beilke, S., 1970, Laboratory investigations on washout of trace gases, in Precipitation Scavenging,
USAEC Symposium Series 22, December 1970~,pp. 261-269.
Durham, J. L., Overton, J. H , and Aneja, V. R, 1981, Ini'luence of gaseous nitric acid on sulfate pro-
duction and acidity in rain, Atmos. Environ. 15 (6), 10~9-1068.
Durrant, R J. and Dun:ant, B., 1962, Introduction to Advanced Inorganic Chemistry, Longmans Lon-
don.
Eigen, M. and Schoen, J., 1955a, Z. Phys. Chem. 3, 126.
Eigen, M. and Schoen, J., 1955b, Z. Elektrochem. 59,483.
Freiberg, J. E. and Schwartz, S. A., 1981, Oxidation of SO2 in aqueous droplets: mass transport limita-
tion in laboratory studies and the ambient atmosphere, Atmos. Environ. 15, 1145-1154.
Gardner, J. A., Watson, L. R., Adewuyi, Y. G., Davidovits, E, Zahniser, M. S., Worsnop, D.R., and
Kolb, C. E., 1987, Measurements of the mass accommodation coefficient of SOz(~)on water drop-
lets, J. Geophys. Re.s. 92, 10887-10895.
Gamier, J. R, Ehrhard, Ph., and Mirabel, Ph., 1987, Water droplet growth study in continuous flow
diffusion cloud chamber Atmos. Res. 21, 41-51.
Harrison, R. M., Smrges, W. T., Kitto, N., and Yuanqian, Li, 1990, Kinetics of evaporation of armno-
nium chloride and ammonium nitrate aerosols, Atmos. Environ. 24, 1883-1888.
International Critical Tables of Numerical Data, Physics, Chemistry and Technology/National
Research Council of U.S.A., Vol. 5, 1929, MacGraw Hill, New York, pp. 62-78.
Jaecker-Voirol, A., Ponche, J. L., and Mirabel, Ph., t990, Vapor pressures in the ternary system water-
nitric acid-sulfilric acid at low temperature, J. Geophys. Res. 95, 11857-11863.
Kirchner, W., Welter, E, Bongartz, A., Kames, J, Schweighoeffer, S., and Schurath, U, 1990, Trace gas
exchange at the air/water interface: measurements of mass accommodation coefficients, J. Atmos.
Chem. 10,427-449.
Koenig, G., Anders, K., and Frohn, A., 1986, New light scattering technique to measure the diameter
of periodically generated moving droplets, J. AerosolSci. 17 (2), 157-167.
Larson, T.V. and Taylor, G, S., 1983, On the evaporation of ammonium nitrate aerosol, Atmos.
Environ. 12, 2489-2495.
Leaist, D. G., 1987, Diffusion of aqueous carbon dioxyde, sulfur dioxyde, sulfuric acid and ammonia at
very low concentrations, J. Phys. Chem. 91, 4635-4638.
Lee, J.H. and Tang, I.N., 1988, Accommodation coefficient of gaseous NO2 on water surfaces,
Atmos. Environ. 22, t147-1151.
Mozurkewich, M., McMurry, R H., Gupta, A., and Calvert, J. G., 1987, Mass accommodation coeffi-
cient for HO2 radicals on aqueous particles, J. Geophys. Res. 92, 4163-4170,
Plateau, J, 1873, Statique exp~rimentale et th~orique des liquides soumis aux seules forces mol~culaires,
Gauthier Villard, Paris.
Pruppacher, H.R. and Klett, J.D., 1978, Microphysics of Clouds and Precipitations, D. Reidel,
Dordrecht, pp. 136-162,
MASS TRANSFER AT THF AIP./WATER INTERFACE 21

Rayleigh, J. W. S., 1878, Proc. London Math. Soc. 10, 4.


Rayleigh, J. W. S., 1879, Proc. Roy. Soc. 29, 71.
Reid, C. R. and Sherwood, T. K., 1986, The Properties of Gases and Liquids, McGraw Hill, New York,
pp. 520-565.
Richardson, C. B. and Hightower, R. L., 1987, Evaporation of ammonium nitrate particles, Atmos.
Environ. 21,971-975.
Schneider, J.M. and tIendricks, C.D., 1964, Source of uniform-sized liquid droplets, Rev. Sci.
Instrum. 35 (10), 1349-1352.
Schwartz, S.E., 1986, Chemistry of Multiphase Atmospheric Systems, (ed. W. Jaeschke), Springer-
Verlag, Berlin, Heidelberg.
Schwartz, S. E., 1984, Gas and aqueous chemistry of HO 2 in liquid water clouds, J. Geophys. Res. 89,
11589-11598.
Schwartz, S. E., 1990, The characteristic time to achieve interfacial phase equilibrium in cloud drops,
Atmos. Env. 24, 2892-2893.
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 2 in aqueous solutions, Atmos. Environ. 15, 1129-
1144.
Schwartz, S. E. and White, W., 1981, Solubility of the nitrogen oxides and oxyacids in dilute aqueous
solution, Adv. Environ. Sci. Eng. 4, 1-45.
Schwartz, S. E. and White, W., 1983, Kinetics of reactive dissolution of nitrogen oxides into aqueous
solution, Adv. Environ. Sci. TechnoL 12, 1-116.
Seinfetd, J. H., 1986, Atmospheric Chemistry and Physic of Air Pollution, Wiley, New York, pp. 198-
234.
Str6m, L., 1969, The generation of monodisperse aerosols by means of a disintegrated jet of liquid,
Rev. Sci. Instrum. 40 (6), 778-782.
Tabatta, S., 1973, A simple but accurate formula for the saturation vapor pressure over liquid water, J.
AppI. Meteorol. 12, 1410-1412.
Van Doren, J. M., Watson, L. R., Davidovits, R, Worsnop, D. R., Zahniser, M. S., and Kolb, C. E.,
1990, Temperature dependence of the uptake coefficients of HNO 3, HC1 and N205 by water drop-
lets, J. Phys. Chem. 94, 3265-3269.
Worsnop, D. R., Zaliniser, M. S., Kolb, C. E., Gardner, J. A., Jayne, J. T., Watson, L. R., Van Doren,
J. M., and Davidovits, E, 1989, Temperature dependence of mass accommodation of SO2 and
H202 on aqueous surfaces, J. Phys. Chem. 93, 1159-1172.

Vous aimerez peut-être aussi