JOURNAL
OF GEOPHYSICAL
RESEARCH, VOL. 91, NO. D1, PAGES 1089-1096, JANUARY 20, 1986
The H2SO4-HNO3-NH 3 Systemat High Humidities and in Fogs
2. Comparison of Field Data With Thermodynamic Calculations
DANIEL J. JACOB,
1 JED M. WALDMAN,J. WILLIAM MUNGER,AND MICHAEL R. HOFFMANN
EnvironmentalEngineeringScience,W. M. Keck EngineeringLaboratories, California Institute of Technology,Pasadena
Concentrationsof HNO3(g ) and NH3(g) determinedin the field were compared to predictionsfrom
aerosolequilibrium models.The productsof HNO3(g) and NH3(g) concentrationsmeasuredunder cool
and humid nonfoggy conditions agreed in magnitude with predictions from a comprehensivethermodynamic model for the atmosphericH2SO,•-HNO3-NH3-H20 system.Observedconcentrationsof NH3(g)
in fogs were generally consistentwith those predicted at equilibrium with fog water, but important
discrepancieswere noted in some cases.These discrepanciesmay be due to fluctuationsin fog water
compositionover the courseof samplecollectionor to the samplingof nonfoggypocketsof air present
withinthe fog.Detectableconcentrations
of HNO3(g) (up to 23 neqm-3) wereoftenfoundin fogswith
pH < 5 and were attributed to the samplingof nonfoggyair or to the slow rate of HNO3(g) diffusionto
fog droplets.Concentrationsof HNO3(g) in fogswith pH > 5 were below the detectionlimit of 4--8 neq
m-3
INTRODUCTION
The chemical speciation of the H2SO4-HNO3-NH3-H20
system between atmosphericphaseshas been the subject of
much recent interest. Sulfuric acid nucleateswith water vapor
under usual atmospheric conditions [Kiang et al., 1973], but
HNO 3 remains in the gas phase at relative humidities up to
about 98% [Nair et al., 1983]. Ammonia is scavengedby
acidic
sulfate
aerosols
until
eventual
neutralization
is
achieved; however, it has a substantial vapor pressure over
ammonium nitrate aerosol. A key question to our understanding of aerosol nitrate formation is the determination of NH 3
and HNO 3 vapor pressures over aerosol formed from
H 2SO,•-HN O 3-NH 3 atmospheric mixtures.
A number of thermodynamic models have attempted to
answer that question. Tang [1980] and Stelsonand Seinfeld
[1982a] studied the effect of relative humidity and pH on the
vapor pressuresof HNO3 and NH 3 over their aqueoussolutions. Stelsonand Seinfeld[1982b] determinedthe dependence
on temperature and relative humidity of the ammonium ni-
trate aerosol dissociationconstantK = PHNO3
X PNH3and
later reported [Stelson and Seinfeld, 1982c] that addition of
H2SO4 to the mixture does not lower K greatly unlessthe
H2SO4/HNO 3 ratio is very large. Saxena et al. [1983] and
Bassett and Seinfeld [1983, 1984] proposed comprehensive
multiphase thermodynamic models for the H•SO4HNO3-NH3-H•O system.
All of the above models are basically consistentone with
the other and representvarious degreesof sophisticationin
the treatment of the H2SO4-HNO3-NH3 system.The Bassett
and Seinfeld models offer at present the greatest level of
chemical detail. In these models the equilibrium composition
is calculatedby minimizing the Gibbs free energyof a system
composedof the gas phase speciesH2SO4(g), HNO3(g), and
Seinfeld [1984] model considersin addition the increasein
vapor pressuredue to the Kelvin effect and thus predictsthe
size distribution of NO 3 - aerosolfrom the sizedistribution of
SO42- aerosol.The BassenandSeinfeld[1983] modelignores
the Kelvin effect,but this is reported to causeonly a negligible
underestimate of K. In the special case of the
HNO3-NH3-H20 system the Bassett and Seinfeld [1983]
model reduces to the Stelson and Seinfeld [1982b] model,
which usesthe same thermodynamic data and ionic strength
correction procedures.
The hygroscopicaerosol is an aqueous solution at high
humidities, and Table 1 gives the reactions determining the
speciationof HNO 3 and NH 3 under those conditions.K decreases rapidly with increasing humidity above the deliquescencepoint [Stelsonand Seinfeld,1982b]. Above 100%
relative humidity, fog droplets form by activation of condensation nuclei, resulting in a considerableincrease of the
atmospheric liquid water content and a corresponding enhancementof HNO 3 and NH 3 heterogeneouscondensation.
Fog droplets are not at stable equilibrium with the surrounding water vapor, and thermodynamicmodels predicting
K as a function of relative humidity cannot be applied; however, HNO3(g) and NH3(g) concentrationsat equilibrium with
the fog water can be directly calculated from measuredfog
water concentrations.Fog droplets are sufficientlydilute solutions that the Debye-Hfickel expression for activity coefficients [Stummand Mor•7an, 1981] is appropriate.
In fogs the equilibrium partitioning of HNO 3 and NH 3
between the gas phase and the fog water can be determined
from the liquid water content, the fog water pH, and (R1)-
(R3). Liquid water contentsin fog rangefrom 0.01-1 g m-3,
and fog water pH valueshave been found to range from about
2 to 8 [Mun•7eret al., 1983; Jacobet al., this issue].Over 99%
of total HNO 3 at equilibrium with fog water is scavengedas
NO 3- in this range of conditions(reaction(R1)). No measur-
NH3(g); the aqueousphasespeciesH +, HSO½-, SO4:-,
NO3- , and NH½+; and the solid phasesNH,•HSO•(s), able HNO3(g) shouldthereforebe foundat equilibriumin fog,
(NH•)3H(SO,•)2(s), (NH,•)2SO,•(s),NH,•NO3(s), (NH½)2SO,•'
3NHcNO3(s), and (NH,•)2SO,•' 2NH•NO3(s). The Basserrand
even acidic fog. On tl•e other hand, the equilibrium speciation
of NH 3 is strongly dependenton droplet pH, liquid water
content, and temperature.The fraction F of total NH 3 scavenged by fog is given by:
x Now at Center for Earth and Planetary Sciences,Harvard University,Cambridge,Massachusetts.
F =
Copyright 1986 by the AmericanGeophysicalUnion.
LRTK2[1 + (K•[H+])]
1 + LRTK2[1 + (K3[H+])]
(•)
where the equilibrium constantsK 2 and K 3 are the K298
Paper number 5D0728.
0148-0227/86/005D-0728 $05.00
values of Table 1 corrected for temperature with the Van't
1089
1090
JACOB
ETAL.:H 2SO,•-HNO3-NH 3 IN FOGS--THERMODYNAMIC
CALCULATIONS
TABLE 1. Dissociationand Vapor PressureEquilibria of HNO3 and NH3
K298,
Reaction
AH298,
kcal
kcal
Reaction
mol - •
mol - •
(R1)
HNO3(g)= NO 3- + H +
3.2 X 106M e atm-•
--17.3
(R2)
NH3(g) = NH3(aq)
7.4 x 10• M atm-x
-6.8
Halesand
(R3)
NH3(aq) + H + -- NH,• +
1.7 X 109 M -1
--12.5
Smithand
Number
Reference
Schwartz
and
White [1981]
Drewes [1979]
Martell [1976]
Hoff equation, L is the liquid water content (cubic meters of
water per cubic meter of air), R is the gas constant (8.20568
The readeris referredto Jacobet al. [this issue]for a description of sampling sites, analytical methods, and a detailed
x 10-2 L atm mol-• K-•), and T is the absolutetemper- analysisof measurementerrors and biasesassociatedwith our
ature. Note that in the derivation of (1) the Debye-Hiickel
activity coefficientscancel each other. F is plotted in Figure 1
as a function of pH under various conditions.
The thermodynamic approaches outlined above provide
simple ways for treating aerosol formation and composition in
atmospheric chemistry models; therefore it is important to
determine the accuracy with which they can describereal atmospheric conditions. The model of Stelson and Seinfeld
[1982b] has been tested with field data in three studies[Stelson and Seinfeld,1982b; Harrison and Pio, 1983; Hildentann et
al., 1984-1,which all reported that the products of measured
HNO3(g) and NH3(g) concentrations agreed in magnitude
with predicted values of K. Tanner [1983] found that the
model of Tang [1980] adequately predicted concentrationsof
HNO3(g) and NH3(g) over dry aerosol, but he found large
discrepanciesat high humidities betweenhis field data and the
model of Tang [1980] for vapor pressuresover aqueousaerosol. Bassettand Seinfeld[1983] reexaminedthe data of Tanner
[1983] in light of their own model and again noted large
discrepanciesat high humidities between model and observations. No further testsof the Bassettand Seinfeld[1983] model
have been reported to date.
Measurements of HNO3(g) and NH3(g) in fogs or clouds
are few. Daunt et al. [ 1984] found concentrationsof both gases
to be below the detectionlimit of 0.4 ppb (17 neq m-3) in
acidic stratusclouds(pH 3.2-4.2). No measurementsare available in nonacidic fogs, where NH 3 should have a substantial
vapor pressure.Ayers et al. [1984] have compared measured
ground level NH3(g)concentrations to those predicted from
[H +] and [NH,• +] rainwaterconcentrations
and reported
samplingtechniques.
We will generally present HNO3(g) and NH3(g) concentrationsin units of equivalentsper cubicmeter, for consist-
ency with the units of NO 3- and NH,• + aerosol concentrations."Equivalent" in that senserefers to the proton
donor or acceptorcapacityof the gas when scavengedby the
aerosol.Both HNO3(g) and NH3(g) contributeone equivalent
per mole; 1 ppb = 43 neq m-3 at 5øC.
NONFOGGY CONDITIONS
The main site of the San Joaquin Valley samplingprogram
was located in Bakersfield, California. The Bakersfield data
are given in Table 2. For comparison with the Bassett and
Seinfeld[1983] model the aerosolcompositionswere reduced
to model mixturesof SO42-, NO3- , NH4 +, and H +. This
was done by assumingconcentrationsof H + and OH- to
satisfyelectroneutralitywith the measuredconcentrationsof
SO42-, NO3-, and NH½+. The presence
of otherionsin the
real aerosolwill perturbthe thermodynamics,
but this pertur-
bationshouldbe smallbecause
the ionsSO½
2-, NO3-, and
NH½+ contributedover90% of the total aerosolioniccontent
[Jacob et al., this issue].Aerosol concentrationsof HCO•and RCOO- werenot determined,but the balancesof NH,, +
to (NO3- + SO½2-) in Table 2 showno evidenceof excess
NH4 + attributableto HCO3- or RCOO- ammoniumsalts.
Alkaline ammonium salts appear to be volatile under nonfoggy conditions [Jacob et al., this issue]. Therefore the
(SO½
2-, NO3-, NH½+, H +)mixtureis an adequate
modelfor
the ionic content of the Bakersfield
aerosol.
We applied the Bassett and Seinfeld [1983] model to de-
large discrepanciesbetweenthe two.
scribetwo differentpartitioningmodesfor SO½
2- andNO3-:
2- and NO•- presentexclusively
in differentaerosol
Over the course of an extensivesampling program in the (1) SO½
wintertime atmosphereof the San Joaquin Valley of Califor- phases("externalmixture"),and (2) SO42- and NO•- alnia [Jacob et al., this issue], we determined HNO3(g) and
NH3(g) concentrations simultaneously with aerosol and fog
water composition. In this paper we compare the concentrations of HNO3(g) and Nlt3(g ) determined under nonfoggyconditionsto thosepredictedby the Bassettand Seinfeld
[1983] model for the HeSO,•-HNO3-NH3-H20 system.Our
data is well suited for such a comparison because the ions
SO,•2-, NO3-, and NH,• + contributedover90% of the ionic
content of the aerosol. In fogs we compare measured concentrationsof HNO3(g) and NH3(g) to thoseexpectedat equilibrium with the fog water concentrations; fog water pH
values ranged from 2.9 to 7.6, thus spanninga broad range of
predictedNH3(g)-scavengingefficiencies(Figure 1).
Gaseous HNO3 and NH 3 were determined by dual-filter
methods [Russell and Cass, 1984]. Aerosol was collected on an
open-faced Teflon filter surmounted by a cover to prevent
collection of large particlesby sedimentation.Fog water was
collected with a rotating arm collector [Jacob et al., 1984].
g
/
/
\
\ ,, ',,
\
',. ',,
O/ I I I%'"'
I• •
3
4
5
6
7
B
9
pH
Fig. 1. Fraction of NH3 scavengedby fog water at equilibrium.
All fog droplets are assumedto be at the same pH. Liquid water
contentsof 0.05 g m -3 (solidline) and 0.5 g m-3 (dashedline) are
considered.
JACOB
ETAL.'H2SO•.-HNO3-NH3
INFOGS--THERMODYNAMIC
CALCULATIONS
1091
TABLE 2. Bakersfield1983-1984Data for Aerosoland GaseousH2SO,•-HNO3-NH 3 Species
Relative
Temperature, Humidity,
Date
Dec. 31
Time
NH4 +
NO 3-
SO•,2-
NH3(g)
HNO3(g)
0000-0345
1200-1610
0215-0815
1200-1615
(X)00-0415
351
467
170
161
168
201
370
445
427
442
362
388
142
85
117
152
351
297
89
161
109
124
226
445
660
731
596
855
118
219
570
417
19
158
212
68
67
107
483
113
55
63
60
47
52
139
60
48
17
4
<4
35
14
øC
1200-1615
573
669
242
275
273
344
545
769
1057
1141
923
1149
272
260
621
533
Jan. 11
Jan. 12
1200-1615
0000-0415
1200-1615
144
345
331
105
204
244
78
149
169
65
157
334
46
<4
26
9.0 _ 0.7
3.0 _ 0.9
11.2 _ 1.0
Jan. 13
0000-0400
0700-1115
1200-1615
1800-2210
0050-0410
0735-1130
1200-1615
486
729
855
692
567
641
318
283
349
437
402
423
423
263
407
475
507
306
229
214
129
208
122
240
161
180
470
118
<4
6
14
10
5
6
43
4.6 _
5.9 _
9.9 _
6.4 _
4.9 _
6.4 _
10.7 _
Jan. 1
Jan. 2
1200-1620
Jan. 3
Jan. 4
Jan. 5
Jan. 6
Jan. 8
Jan. 9
Jan. 10
Jan. 14
0000-0415
1200-1615
1200-1700
0100-0510
1200-1615
0000-0415
0000-0405
1420-1825
0000-0415
25
<4
11
24
6
19
7
6
8
46
14
7.8 _
10.8 _
10.4 _
13.3 _
7.2 _
11.8 _
9.1 _
9.2 _
9.9 _
6.9 _
8.8 _
7.4 _
6.8 _
7.3 _
6.2 _
6.9 _
0.9
0.6
0.6
0.7
1.1
0.7
0.8
0.6
1.3
0.8
0.6
0.6
0.6
0.6
0.6
0.8
0.7
0.8
1.1
0.8
1.2
1.0
1.5
%
96
93
96
76
89
74
87
81
80
95
85
95
95
86
89
85
71
91
66
96
91
73
85
89
85
61
Completeaerosolanalyses
arereportedby Jacob[1985a].
Valuesare in nanoequivalents
per cubicmeterunlessotherwise
indicated;timeis localtime(PST);
andrelativehumidityis calculatedfromtemperature
anddewpoint.
lowed in the same aerosol phases("internal mixture"). The
actual partitioningwill lie somewherebetweenthesetwo extremes,dependingon the aerosolformationprocesses,
the kineticsof coagulation,and the Kelvin effect on HNO 3 vapor
pressure
I-Bassett
andSeinfeld,
1984].SinceNH,•+ wasalways
HNO3, and NH3, the temperature,
and the relativehumidity.
To calculate the errors on the predictionsfrom the errors on
the measurements,100 setsof input variableswere generated
for each data point from a random schemebasedon the observed values and standard errors for each of the variables;
in excessof SO,•2-, the externalmixturecasewastreatedby
assuming
that all SO,•2- waspresentas(NH,02SO,•andsolving the equilibriumproblemfor the remainingHNO 3 and
NH 3 with no SO,•:- present.The internalmixturecasewas
treatedby directlysolvingthe equilibriumproblemfor the
H:SO,•-HNO3-NH3-H:O atmospheric
system.The presence
of SO4:- and NO3- in the sameaqueousphaseincreases
the
solubilityof HNO3 and NH3, comparedto an externalmix-
standarderrors on the experimentaldeterminationsof concentrationswere taken from Jacob et al. [this issue].The ex-
ture [Stelsonand Seinfeld,1982c].
minationsof temperatureand relativehumidity; they were not
pectedvaluesand standarderrorsfor the output variables
were determinedfrom the outputsfor the 100 setsof input
variables.Resultsare shownin Figures2a and 2b. The aerosol
was predictedto be entirely aqueousexceptin a few cases
(indicatedon Figures2a and 2b).The standarderrorson the
predictedvaluesof K weremostlydue to errorson the deter-
Sampleswere collectedover 4-hour samplingperiods, affectedby errorson H2SO•, HNO3, and NH3 concentrations
duringwhichtemperatures
andrelative
humidities
remainedin
the external mixture assumption and were affected only
stable becauseof the prevalent low-overcastconditions(Table
very weaklyin the internalmixtureassumption.
On the other
2). Temperatureand dew point were measuredhourly at the
National Weather Service(NWS) station 8 km north of the
Bakersfieldsamplingsite,and standarddeviationsfor temperature over the 4-hour period were calculatedfrom the hourly
temperaturerecord; 0.5øC was added to the standard deviation to account for the sensitivityof the readings.The dew
point did not changesignificantlyover the 4-hour sampling
periods.A few temperaturemeasurementstaken at our sampling site were within IøC of those taken by NWS. Measurements by the NWS confirm that temperaturesand relative
humidities at the NWS station are representative of those
found at our Bakersfield site (D. Gudgel, Bakersfield NWS,
private communication,1984).
The Basserr and Seinfeld [1983] model was run with the
computercodeprovidedby Basserr[1984]. This coderequired
as input the total (gasplus aerosol)concentrationsof H2SO,•,
hand, errors on the concentrationsof the individual gases
werestronglyaffectedby errorson the determinations
of concentrations,especiallywhen a closebalanceexistedbetween
total NH, and total acids(HNO3 + H2SO,0. In thosecases
error bars were very large.
The productsof partial pressuresobservedin the field
agreedin magnitudewith the valuesof K predictedby the
model. However, the model predictionswere consistentlytoo
low. Predictions with the external mixture assumption were
closer to observations than with the internal mixture assump-
tion. The modelappearedto givebetterpredictionsfor individualgasesthan for K, but this is deceiving.First, the errors
on the predictedgasconcentrations
werelarge.Second,K was
often so small that one of the gaseswas almostentirelydepleted, and the remaininggas was then simply presentat its
concentration in excess of the neutralized aerosol.
1092
JACOBET AL.' H :SO,,-HNO 3-NH 3 IN FOGS--THERMODYNAMIC
CALCULATIONS
6O0
,
ß
!
-
!
,
!
ß
!
'
!
ß
i
ß
i
I
ß
'
I
I
I
'
I
ß
i
KEY:
r]OBSERVED
NH$(g)
XPREDICTED
400
EXTERNAL
MIXTURE
neq m-3
200
,tl
ß
i
100
' • •' ' '
D.... t
,
.I
i
ß
i
ß
• ' ' ' ' ' ' ' '1
i
ß
,
HNO$(g)
2.0
75
1.0
25
o ! I ..,
105
'
I
'
I
•
ß
0.5
o
!
I
I
'
I
'
!
'
I
'
I
'
!
'
1
i
ß
I
'
IOt
o
I0'•
t
0
10o
ppb
z
( neqm-S,)
z
i0 -I
lo z
101
,
$1
1
i
,
2
,
3
I
,
4
,
5
,
6
DECEMBER
I
,
7
f
8
I
,
9
I
i
10
I
....
II
12
13
14
JANUARY
Fig. 2. Comparisonof Bakersfieldfield data to the Bassettand Seinfeld[1983] aerosolequilibriummodel: concentrationproducts(HNO3(g))x (NH3(g)), and individualconcentrations
of HNO3(g) and NH3(g). Resultsare givenas
error bars extendingbetweenthe limits (mean -a) and (mean +a). The aerosolwas predictedby the model to be an
aqueoussolutionexceptwhen indicatedby mixed aqueous-solid(opencircle),or solid (dagger).Two limiting caseswere
considered:
(a) SO,•2- and NO 3- presentexclusively
in differentaerosolphases(externalmixtureassumption)
and (b)
SO½
2- andNO 3- allowedin thesameaerosolphases
(internalmixtureassumption).
Stelsonand Seinfeld[1982b] also found that the products of
vapor pressuresover aqueousaerosol were systematicallyunderpredictedby their model, by factors similar to the ones we
observed.Harrison and Pio [1983] did not find sucha systematic trend, although their determinationsmay have been subject to large errors becauseof the considerablefluctuationsin
temperatures and relative humidities over the course of the
sampling periods. Becauseof the many complicated processes
occurring in the atmosphereit is difficult to assessthe significanceof the model underpredictions.Considerationof droplet
curvature would increase the predicted K, but Bassett and
Seinfeld [1984] found this increaseto be negligible.Laboratory data of vapor pressures over concentrated ammonium
nitrate solutions should provide a check on the accuracy of
the thermodynamic calculations,but they are unavailable at
this time.
The dual-filter method is known to be subject to positive
interferences,that is, concentrationsof HNO3(g ) and NH3(g)
may be overestimated[Appel et al., 1980]. Volatilization of
NH,•NO 3 from the Teflon prefilter leadsto artifact HNO3(g)
and NH3(g), but in our case this problem is minimized because temperaturesand relative humidities remained stable
over the course of the sampling periods (Table 2). Another
sourceof positive interferenceis the displacementof NO 3 - or
NH,• + by nonvolatilematerial collectedon the filter, for example, acid sulfatesor alkaline carbonates.However, if such
dispacementreactionsoccurredon the filter, this would mean
that the atmosphereitselfwas not at equilibrium; in that case,
HNO3(g) and NH3(g) atmosphericconcentrationscould have
differed substantiallyfrom their equilibrium values.
Indeed, a possible explanation for the discrepanciesobservedis that the atmospherewas not at chemicalequilibrium.
The NH 3 observedat Bakersfieldoriginatedmostlyfrom local
ground level sources[Jacob et al., this issue]; HNO 3 and
NH 3 may not have had time to mix sufficientlyfor equilibrium to be achieved.The incorporationof HNO3(g) and NH3(g)
JACOB
ETAL.'H2SO4-HNO
3-NH
3INFOGs--THERMODYNAMIC
CALCULATIONS
1093
b
6OO
4OO
nlq m'$
lOs
ß
i
ß
I
'
I
ß
i
-'
i
ß
i
ß
i
,
I
ß
!
ß
!
'
i
'
I
ß
!
ß
I
ß
K, (NHI (g)) x(HNO•(g))
10'
o
10 4
o o
0
t
ioo
ppb'
(.eqre's)• I0 )
i0 't
I0 z
i0 t
31
I
2
3
4
5
ß
,
DECEMBER
7
.
•
.
ß
1•)' II
12 13 14
JANUARY
Fig. 2. (continued)
(equivalents
per liter of air). The standarderrors
into the aerosolat highhumidities
proceeds
mostlyby hetero- concentration
on
predicted
(NH3(g))
were
determinedfrom standarderrors
geneous
condensation,
whichcanbea slowprocess.
Basedon
of
the calculationsof Fuchsand Sutugin[1971] over a rangeof of + IøC on temperatureand 15% on the determinations
representative
conditions,one finds that the diffusion- [NH4 +] and [H +] [Jacobet al., thisissue].Severalfog water
controlledequilibrationtimeof a gaswith low vaporpressure sampleswere generallycollectedover the cours.eof one
period;equilibriumNH3(g) concentrations
mayrangefromminutesto hours(depending,
amongother NH3(g)-sampling
things,onthe"sticking
coefficient"
of thegasmolecule
on the
aerosol).
FOGGY CONDITIONS
Nine concurrentsamplesof fog water, HNO3(g), and
NH3(g) werecollectedduringthe winter 1983-1984San Joaquin Valley samplingprogram at Bakersfieldand several
other sites,and 19 more sampleswere collectedin a similar
programconductedthe followingwinter. All sampleswere
collectedduringintervalswith densefog throughout.The data
are givenin Table 3. Concentrations
of NH3(g)at equilibrium
with the fogwaterwerecalculatedfrom(R2) and (R3):
were calculatedfrom the fog water concentrationsin each
individualsampleand then time averaged.Resultsare shown
in Table 3 and Figure 3.
The NH3(g) concentrations
calculated
from(2) wereconsistent with observedNH3(g) concentrations
in 22 of the 28
comparisons'
theyagreedto withina factorof 2 for NH3(g)
concentrationsabove the detectionlimit (nine cases)and cor-
rectly predictedNH3(g) concentrations
below the detection
limit (13 cases).In the remainingsix comparisons,
however,
the agreementwaspoor.Nonattainmentof equilibriumis an
unlikelyexplanationfor the observed
discrepancies;
equilibrium of NH3(g) with fog water •,• ?H > 5 is achievedrapidly
becauseof the relativelylow solubilityof NH 3 [Chameides,
1
(NH3(g))= •
[NH,•+]/[H +]
(2) 1984].Also,seediscussion
by Jacobs[1985b]and Chameides
RTK2K3
[1985]).However,a majorsourceof errorin our calculations
where[NH4 +] and I-H+] are the fog water concentrations is that the fog dropletscollectedin a fog water samplewere
(equivalents
per liter of water),and (NH3(g))is the NH3(g) not of uniform composition.Fog droplets supporting a
1094
JACOB
ETAL.'H2SO,•-HNO3-NH
3INFOGS--THERMODYNAMIC
CALCULATIONS
TABLE3. Concentrations
of HNO3(g)andNH3(g
) in Fogs
Temperature,
Date
øC
HNOa(g)/NH3(g
) FogWater
Sampling
Sampling
Period,PST
Periods
pHø
NH3(g)
t'
NH3(g)
NH,•+,ø Predicted, Observed,
#eqL- x neqm- 3
neq m- 3
HNO3(g)
Observed,
neqm- 3
Bakersfield
Jan. 13, 1984
0(0X)-0400
0130-0200
6.38
3190
0200-0300
Dec. 28, 1984
4
Jan. 2, 1985
3.5
Jan. 3, 1985
4
Jan. 3, 1985
2
Jan. 3, 1985
1
Jan. 3-4, 1985
-1
5.92
2800
0300-0330 6.72
5920
0330-0400
3300
6.74
0130-0215
6.20
(X)00-0225
0035-0115
6.80 2780
2680
} 309ñ 73
2300-0000
5.60
2115--0000
2100-2300
6.20 1990
1100}38ñ 15
0200-0305 5.72
5.66
1290
0(X)0-0300
(XX)0-0200
1340
} 19ñ 7
118
<7
143 ñ 42
91
<6
22 ñ 7
40
<6
Jan. 14, 1985
379
268
2100-2130
5.16
1340
2200-2300
5.46
1440
0000-0100
5.20
1770
0100-0200
5.10
1850
0500-0600
5.92
0600-0700 4.80
0700-0800
4.5
4.52
11 ñ 3
<25
<6
<17
<4
165 ñ 45
<25
N/A
14 ñ 7
<25
9
1080
1660
7ñ 3
720
0130-0430
0130-0230
6.17
2430
}
2015-2315
2005-2105
5.85
2400}
0(X)0-0300
0000-0100
3.99
1450
0720-0940
0715-O800
3.06
3130
Jan. 18, 1985
Jan. 19, 1985
Jan. 20,1985
<6
<4
0400-0805
0400-0500
4.96
1500}
3.5
7.57
6.76
<25
262
Jan. 5, 1985
Jan. 4-5, 1985
0000-0100
363
<6
0530-0700
6.83
928
0300-0700
0305-0500
4.30
781
} 64ñ 30
0830-0930
6.51
0700-0930
0700-0830
6.82 2900
1070
} 161ñ 47
0124-0155
4c
39
2300-0155
2300-0000
7.08592
2000-2255
2010-2100
6.26946
2310-0205
230(00000
5.55
1860
Jan. 4, 1985
208
370ñ 86
0230-0330
5.56
2390
0330-0430
6.72
2170
2105-2200
4.85
1710
2200-2300
4.18
1610
0100-0200
0200-0300
3.86
3.67
1450
1590
< 1
<25
9
< 1
<29
23
0800-0845
3.06
3360
0845-0930
2.92
3500
McKittrick
Jan. 5, 1984
0045-0415
0005-0100
4.02
399
0100-0200
Jan. 6, 1984
(X)00-0410
Jan. 7, 1984
(X)(X)-0415
4.00
347
0200-0300 4.02
344
0300-0435
357
4.21
0035-0100
4.23
241
0100-0200
0200-0300
0300-0400
0400-0500
4.03
4.22
4.20
4.14
294
499
333
345
0(0)0-0100
3.96
136
0115-0150
0200-0300
030(00400
0400-0500
4.41
4.26
4.40
4.44
186
162
204
166
<1
<19
< 1
<16
7
< 1
<16
9
<16
11
<16
5
<17
10
<5
Jan. 7, 1984
Jan. 8, 1984
1200-1615
1030-1230
4.23
683
t
Jan. 10, 1984
0000-0410
2035-0035
4.18599
1230-1440
4.24
870
1440-1540 4.28
740
1540-1655
522
4.12
0035-0205
4.50
555
0205-0405
5.01
658
0120-0525
3.85
0045-0445
0010-0120
3.71 351
845
}
<1
1ñ 1
<1
Buttonwillow
Jan. 7, 1984
Jan. 3, 1985
Jan. 4, 1985
Jan. 4, 1985
6
4
3
1.5
0100-0500
(X)00-0220
0020-0220
0415-0810
0040-0140
5.18
1270
0410-0620
5.33
1070
0140-0410
5.40969
t 10
ñ4
(X)00-0215 5.25
0015-0215 5.73
0430-0630 6.14
2080
553
519
11ñ 4
8ñ 3
16 ñ 6
69
159
42
<17
<7
<8
<4
JACOB
ETAL.' H2SO•.-HNO3-NH3 IN FOGS--THERMODYNAMIC
CALCULATIONS
TABLE 3.
Date
(continued)
HNO3(g)/q'qH
3(g)
FogWater
Sampling
Period,PST
Sampling
Periods
Temperature,
øC
1095
NH3(g)t'
pHø
NH3(g)
HNO3(g)
NH,• +,ø Predicted, Observed, Observed,
#eq L- • neqm- 3
neqm- 3
neqm- 3
Buttonwillow (continued)
Jan.
4,1985
Jan.
5,1985
4
2.5
2030-2135
6.76
858}
1930-2135
1930-2030
6.55 968
0300-0400
5.87
0220-0410
0205-0300
5.70 483
529
}
124+ 33
79
N/A
8+ 2
102
<9
Jan.
5,
1985 2 0415-0810
0400-0500
6.16
456
t
145
<8
0500-0615
6.78
0615-0700
6.89
683
582
0700-0800
542
6.87
75+
21
-
Visalia
Jan.
7,1984
6
0145-0335
0120--0250
6.97 860
1 355
+90
0250--0430
7.23
678
-
592 <8
Measurement
errorsfor NH3(g) andHNO3(g) wereabout20%. Detectionlimitsdepended
on filterrun time [Jacobet al., thisissue].
N/A = not analyzed.
aFogwaterconcentrations.
Completefogwateranalyses
arereportedby Jacob[1985a]for theJanuary1984data andby Waldman[1986]
for the December1984to January 1985data.
hAverage
overtimeof NH3(g) concentrations
predicted
fromfogwaterconcentrations
andequation
(2).
CDetection
limit.
measurablepartial pressureof NH 3 contain alkalinity, and
than the bulk foggyatmosphereand would correspondingly
[H +] will not be conserved
uponthe mixingof alkalinity- affectthe measured
NH3(g) concentration.
Further,sampling
containingdropletsof differentcompositions
in the sample.
Further,(NH3(g))is proportionalto 1/[H +], a quantitywhich
is not properly averagedby measurementof pH in the collectedsample.The resultingerrorswill dependon the extent
of nonuniformityin the fog water compositionand are difficult to estimate,but theycouldclearlyaccountfor mostof the
discrepancies.
The large variationsin pH often observedbetweentwo consecutivefog water samples(Table 3) are indicative of the variabilityin fog water composition.
The largestdiscrepancies
were observedfor four sampling
periodsat Buttonwillowwhen the fog water pH was in the
range 5-6. In those four cases,equilibrium with fog water
considerably underpredictedthe observed NH3(g) concentrations.Fog water at pH 5-6 scavenges
most of the atmosphericNH 3 (Figure 1) but returnsit to the gasphaseupon
evaporation[Jacobet al., this issue].An explanationfor the
large discrepancies
that were sometimes
observedin this pH
rangeis the existencewithin the fog of pocketsof air undersaturated with respectto water vapor [Gerber,1981]. Theseair
pocketswouldsupportmuchlargerconcentrations
of NH3(g)
.,[
I
!
[
!
! •],
,
!
,
[
of undersaturated
air may evaporatefog dropletspreviously
collectedon theaerosolprefilterandcauseartifactNH3(g).
Concentrationsof HNO3(g) in fogs were at or below the
detectionlimit in fogsWithpH > 5, but detectableHNO3(g)
concentrations
(5-23neqm-3) werefoundin fogswithlower
pH. The presenceof HNO3(g) in acidic fog cannot be explainedon the basisof equilibriumwith fog water.However,
the presenceof HNO3(g) can be explainedby the substantial
vapor pressureof HNO3(g) over the acid precursoraerosol;
undersaturatedair pocketswithin the fog would support
HNO3(g) at equilibrium.An additional explanationis that
detectableHNO3(g) concentrations
may subsistfor sometime
in thefogbecause
of theslowrateof HNO3(g)diffusionto the
droplets. Chameides[1984] reports that scavengingof
HNO3(g) by diffusionto the fog dropletsproceedson a time
scaleof a few minutes,but this time may be longerif organic
filmsformatthesurface
ofthedroplets
[Gilletal.,1983].
CONCLUSION
Concentrationsof HNO3(g) and NH3(g) weredeterminedin
the field under both foggyand nonfoggyconditions,simultaneously with aerosol and fog water composition.Observed
, ,!
concentrations
werecomparedto predictions
from thermodynamic models.
Measurementsunder nonfoggy,cool, and humid conditions
(temperatures3ø-13øC, relative humidities 60-100%) were
comparedto predictionsfrom the Basserrand Seinfeld[1983]
model for the H2SO½-HNO3-NH3-H20 system.The observed
productsof HNO3 and NH3 vapor pressureswereof the same
magnitude as those predictedby the model; howevei•,the
o, 10a
z
modelpredictions
wereconsistently
too low. The agreement
.o_
betweenobservationsand model was slightly improved by
assumingthat SO½
2- and NO 3- were presentin different
101
102
103
Observed
NH3(g) (neqm-3)
Fig. 3. NH3(g) concentrationsat equilibrium with œogwater
versusobservedNI-[3(g) concentrations.
Both were below the detection limit for 13 of the points (shaded area). Line represents1:1
agreement between predictions and observations.
phases(externalmixture).
Concentrationsof NH3(g) were below the detectionlimit of
17-30 neq m-3 in fogswith pH < 5, and this is consistent
with thermodynamicpredictions.SubstantialNH3(g) concentrationswere observedin fogs with higher pH; the observed concentrationswere usually within a factor of 2 of
those predictedat equilibrium with the fog water, but some
1096
JACOB
ETAL.:H2SO,•-HNO3-NH3 IN FOGS•THERMODYNAMIC
CALCULATIONS
importantdiscrepancies
werenoted.Thesediscrepancies
may
be due to fluctuationsin fog water compositionover the
courseof samplecollectionor to the presenceof pocketsof
Jacob, D. J., Comment on "The photochemistryof a remote stratiform cloud" by William L. Chameides,J. Geophys.Res.,90, 5864,
1985b.
Jacob,D. J., R.-F. T. Wang, and R. C. Flagan, Fogwater collector
design and characterization,Environ. Sci. Technol., 18, 827-833,
undersaturatedair within the fog.
Concentrationsof HNO3(g) were at or below the detection
1984.
limit of 4-8 neq m -3 in fogs with pH > 5, but detectable Jacob, D. J., J. W. Munger, J. M. Waldman, and M. R. Hoffmann,
concentrations
(5-23 neq m-3) wereoftenfoundin fogswith
lower pH. BecauseHNO3(g) shouldnot be presentat equilibrium with fog water, we attribute our observationsto the
presenceof undersaturatedair within the fog or to the slow
rate of HNO 3(g)diffusionto the fog droplets.
The H2SO,,-HNO3-NH 3 systemat high humiditiesand in fogs,1,
Spatial and temporalpatternsin the San JoaquinValley of California, J. Geophys.Res.,this issue.
Kiang, C. S., V. A. Mohnen,J. Bricard,and D. Vigla, Heteromolecular nucleationtheory applied to gas-to-particleconversion,Atmos.
Environ., 7, i279-1283, 1973.
Munger, J. W., D. J. Jacob, J. M. Waldman, and M. R. Hoffmann,
Fogwater chemistryin an urban atmosphere,J. Geophys.Res.,88,
Acknowledgments.
A.G. Russelland G. R. Cass(CaliforniaInsti5109-5121, 1983.
tute of Technology)are thanked for many helpful discussions.
This
work was fundedby the CaliforniaAir Resources
Board (A2-048-32). Nair, P. V. N., P. V. Joshi,U. C. Mishra, and K. G. Vohra, Growth of
aqueous
solution
droplets
of HNO3 andHC1in theatmosphere,
J.
Correspondenceshouldbe addressedto Michael R. Hoffmann.
Atmos. Sci., 40, 107-115, 1983.
REFERENCES
Russell,A. G., and G. R. Cass,Acquisitionof regionalair quality
model validationfor nitrate, sulfate,ammoniumion and their pre-
cursors, Atmos. Environ., 18, 1815•-1827, 1984.
Appel, B. R., S. M. Wall, Y. Tokiwa, and M. Haik, Simultaneous
nitric acid,particulatenitrate,and aciditymeasurements
in ambient Saxena,P., C. Seigneur,and T. W. Peterson,Modelingof multiphase
atmosphericaerosols,Atmos.Environ.,17, 1315-1329,1983.
air, Atmos.Environ.,14, 549-554, 1980.
Ayers,G. P., J. L. Gras,A. Adriaansen,
and R. W. Gillett, Solubility Schwartz,S. E., and W. H. White, Solubilityequilibriaof the nitrogen
oxidesand oxyacidsin dilute aqueoussolution,Adv. Environ.Eng.
of ammoniain rainwater,Tellus,36B, 85-91, 1984.
Bassett,M. E., Mathematicalmodelingof atmospheric
aerosoldynamicsand equilibria,Ph.D. thesis,Calif. Inst. of Technol.,Pasadena, 1984.
Bassett,M. E., and J. H. Seinfeld,Atmospheric
equilibriummodelof
sulfateand nitrateaerosols,
Atmos.Environ.,17, 2237-2252,1983.
Bassett,M. E., and J. H. Seinfeld,Atmospheric
equilibriummodelfor
sulfateand nitrate aerosols,2, Particlesizeanalysis,Atmos.Environ., 18, 1163-1170, 1984.
Chameides,W. L., The photochemistry
of a remotemarinestratiform
cloud,J. Geophys.Res.,89, 4739-4755, 1984.
Chameides,
W. L., Reply,J. Geophys.
Res.,90, 5865-5866,1985.
Daum, P. H., S. E. Schwartz,and L. Newman,Acidic and related
constituents
in liquidwaterstratiformclouds,J. Geophys.
Res.,89,
1447-1458, 1984.
Fuchs,N. A., and A. G. Sutugin,High-dispersed
aerosols,in InternationalReviews
of AerosolPhysics
andChemistry,
vol. 2, editedby G.
M. Hidy and J. R. Brock,pp. 1-60, Pergamon,New York, 1971.
Gerber,H. E., Microstructureof a radiationfog, J. Atmos.Sci.,38,
454-458, 1981.
Gill, P.S., T. E. Graedel,and C. J. Wechsler,Organicfilmson atmosphericparticles,fogdroplets,clouddroplets,raindrops,and snowflakes,Rev. Geophys.,
21, 903-920, 1983.
Sci., 4, 1-45, 1981.
Smith, R. M., and A. E. Martell, Critical Stability Constants,vol. 4,
Plenum, New York, 1976.
_
Stelson,A. W., and J. H. Seinfeld,Relativehumidityand pH dependenceof the vapor pressureof ammonium nitrate•nitric acid solutions at 25øC,Atmos.Environ.,16, 993-1000, 1982a.
Stelson,A. W., and J. H. Seinfeld,Relativehumidityand temperature
dependenceof the ammoniumnitrate dissociationconstant,Atmos.
Environ., 16, 983-992, 1982b.
Stelson,A. W., and J. H. Seinfeld,Thermodynamicpredictionof the
water activity,NH,•NO3 dissociation
constant,densityand refractive index for the systemNH,•NO3-(NH,02SO,•-H20at 25øC,
Atmos. Environ., 16, 2507-2514, 1982c.
Stumm,W., and J. J. Morgan, AquaticChemistry,2nd ed., p. 134,
Wiley-Interscience,New York, 1981.
Tang, I. N., On the equilibriumpartial pressures
of nitric acid and
ammonia in the atmosphere,Atmos.Environ.,14, 819-828, 1980.
Tanner,R. L., An ambientexperimentalstudyof phaseequilibriumin
the atmosphericsystem:AerosolH +, NH,• +, SO,••--, NO3-,
NH3(g), HNO3(g), Atmos.Environ.,16, 2935-2942, 1983.
Waldman,J. M., Depositionalaspectsof pollutant behaviorin fog,
Ph.D. thesis,Calif. Inst. of Technol.Pasadena,1986.
Hales,J. M., andD. R. Drewes,Solubilityof ammoniain waterat low
concentrations,Atmos.Environ.,13, 1133-1147, 1979.
D. J. Jacob, Center for Earth and Planetary Sciences,Harvard
Harrison,R. M., and C. A. Pio, An investigation
of the atmospheric University, Cambridge, MA 02138.
M. R. Hoffmann, J. W. Munger and J. M. Waldman, EnvironmenHNO3-NH3-NH,,NO 3 equilibriumrelationshipin a cool, humid
climate, Tellus, 35B, 155-159, 1983.
Hildemann,L. M., A. G. Russell,and G. R. Cass,Ammonia and nitric
acid concentrations
in equilibriumwith atmospheric
aerosol:Experimentvs.theory,Atmos.Environ.,18, 1737-1750,1984.
Jacob,D. J., The originsof inorganicacidityin fogs,Ph.D. thesis,
Calif. Inst. of Technol., Pasadena,1985a.
tal EngineeringScience,W. M. Keck EngineeringLaboratories,California Instituteof Technology,Pasadena,CA 91125.
(ReceivedFebruary 15, 1985;
revisedJuly 25, 1985;
acceptedJuly 31, 1985.)