A theoretical evaluation on the HNO
3
artifact of the annular
denuder system due to evaporation and diffusional deposition
of NH
4
NO
3
-containing aerosols
Kuanfoo Chang
a, Chungsying Lu
a,*, Hsunling Bai
b, Guor-Cheng Fang
ca
Department of Environmental Engineering, National Chung-Hsing University, Taichung 402, Taiwan
b
Institute of Environmental Engineering, National Chiao-Tung University, 75, Pa-Ai Street, Hsinchu 300, Taiwan
c
Air Toxic and Environmental Analysis Laboratory, Hungkuang Institute of Technology, Sha-Lu, Taichung 433, Taiwan Received 7 July 2001; accepted 17 May 2002
Abstract
A mathematical model was developed to evaluate HNO3artifact of the annular denuder system due to evaporation
and diffusional deposition of nitrate-containing aerosols. The model performance was validated by comparing its numerical solutions with laboratory and numerical data available in the literature for evaporation and diffusional deposition of monodisperse and polydisperse NH4NO3aerosols. Measurement artifacts were evaluated by varying
typical sampling ranges of ambient temperature, HNO3gas concentration, aerosol number concentration, aerosol mass
median diameter, and nitrate mass fraction ofo2.5 mm aerosols to see their respective effects. Potential application of the present model on estimating HNO3artifacts was demonstrated using literature data sampled in USA, Taiwan,
Netherlands, Korea and Japan. Significant measurement artifact could be found in Taiwan and Netherlands due either to low HNO3gas concentration and high nitrate concentration ino2.5 mm aerosols or to high ambient temperature.
r2002 Elsevier Science Ltd. All rights reserved.
Keywords: Annular denuder system; Measurement artifact; HNO3; NH4NO3; Evaporation; Mass accommodation coefficient
1. Introduction
The annular denuder system is an instrument being widely employed for sampling and collecting reactive gases such as SO2, HNO3, NH3and organic vapors in
the atmosphere (Koutrakis et al., 1988; Benner et al., 1991; Lee et al., 1993; Matsumoto and Okita, 1998). The sampling mechanism of the denuder is the diffusion of gaseous pollutants onto the denuder walls coated with reactive substances. Despite the fact that the denuder has advantages of allowing high sampling velocities and possessing large sampling capacities, measurement artifacts are possible and may be significant.
As the gaseous pollutants are absorbed onto the denuder wall, the state of equilibrium between the aerosol and gas-phases is disturbed, which may result in evaporation of aerosols and release of additional gaseous pollutants. Therefore, an excessive amount of gaseous pollutants is being sampled. This phenomenon may particularly be true for volatile aerosols such as NH4NO3 and NH4Cl and their associated gases.
Furthermore, the diffusional deposition of aerosols containing compounds similar to the sampling gases may also introduce additional gaseous pollutants.
Durham et al. (1987) conducted a study on the nitric acid–nitrate aerosol measurements by a tubular diffu-sion denuder and compared their experiment data with those predicted by the Gormley–Kennedy equation (Gormley and Kennedy, 1949). The relative errors, which are defined as the difference between measured
*Corresponding author. Fax: +886-4-22862587. E-mail address:cslu@enve.ev.nchu.edu.tw (C. Lu).
1352-2310/02/$ - see front matterr 2002 Elsevier Science Ltd. All rights reserved. PII: S 1 3 5 2 - 2 3 1 0 ( 0 2 ) 0 0 3 5 2 - 7
and theoretical results divided by the theoretical result, were in the range of 10–90%. They attributed the deviations to the deposition of gaseous N-compound that is extracted and analyzed as NO3
or the released of HNO3gas from NH4NO3aerosols during transit in the
denuder. Measurement artifacts were found for sam-pling HNO3gas in the study of Bai and Wen (2000). The
relative errors strongly depend on the atmospheric HNO3concentration and may be as high as over 80%.
Sampling artifacts of the denuder due to evaporation of NH4NO3aerosols have been theoretically studied by
Pratsinis et al. (1989) and Biswas et al. (1990). Over 200% excessive absorbed NH3gases may occur at 351C.
However, in their studies, the aerosol stream was assumed to contain a monodisperse aerosol. A moment model approximating the aerosol-size distribution by a lognormal function was developed by Bai et al. (1995) to evaluate the evaporation rate of dry NH4NO3 and
NH4Cl aerosols. The model was then employed to derive
the denuder performance equation for sampling atmo-spheric HNO3 gas (Lu et al., 1995). The aerosol
polydispersity leads to a significant reduction in the evaporation rate of dry NH4NO3and NH4Cl aerosols as
compared to the monodisperse aerosol. Therefore, the actual measurement artifact may not be as significant as predicted by Pratsinis et al. (1989) and Biswas et al. (1990).
On the other hand, Forrest et al. (1982) and Larson and Taylor (1983) conducted an experiment on the evaporation of NH4NO3aerosols in a tubular diffusion
denuder and a parallel-plate stripper, respectively. They
concluded that the HNO3 gas evaporated from
NH4NO3 aerosols could be neglected as compared to
measured HNO3 gas concentration. Koutrakis et al.
(1988) demonstrated the high performance of the Harvard-EPA denuder for sampling atmospheric SO2,
HNO3 and HNO2 gases. The differences between
measured and calculated collection efficiencies were small in their study. Tsai et al. (2000) found that positive HNO3artifact due to volatilization of NH4NO3
aerosols is generally negligible compared to measured nitrate-containing particles for an annular denuder and a honeycomb denuder system. According to these studies, the measurement artifacts seem unimportant in a denuder system. However, most of their emphases are on the evaporation of aerosols. Many factors such as ambient temperature, HNO3 concentration, NH4NO3
concentration as well as aerosol size distribution may simultaneously influence the accuracy of HNO3
measurement.
The mass accommodation coefficient, a; is a kinetics-originating factor describing the interface limitations to the mass transfer between the aerosol and gas-phases. The coefficient is yet not well understood; the process is believed to be inhibited by a resistance in transport of molecules across the particle/air interfaces (Dassios and
Pandis, 1999). Bai et al. (1995) used a of 0.1–1 to fit the change of radius of dry NH4NO3aerosol in the denuder.
Recently, Dassios and Pandis (1999) found that the a decreases from 0.8 to 0.5 as the temperature increases from 211C to 271C. In order to evaluate the effect of organic compounds on the NH4NO3evaporation, Cruz
et al. (2000) coated the NH4NO3particles of 100–200 nm
diameter with dioctyl phthalate (DOP) and allowed them to evaporate in a constant temperature laminar flow reactor. A decrease, up to 50%, in the evaporation rate of NH4NO3 aerosol due to the presence of DOP
film was observed which corresponds to the decrease of a from 0.4 to 0.25.
The goals of this study were to develop a mathema-tical model for identifying atmospheric conditions that may lead to significant HNO3artifacts. Major sampling
artifacts under consideration are from evaporation and diffusional deposition of NH4NO3-containing aerosols.
Effects of the two mechanisms on the HNO3artifacts for
a typical range of atmospheric aerosol and gas proper-ties are evaluated and quantitatively determined.
2. Theory
When NH4NO3 aerosols in equilibrium with HNO3
and NH3gases enter the denuder, HNO3gas is removed
by diffusional absorption at the denuder walls with which the equilibrium is distorted. This results in evaporation of NH4NO3 aerosols and release of
addi-tional HNO3and NH3gases.
2.1. Basic assumptions
The following major assumptions are made in developing a mathematical model to study the HNO3
artifacts in the denuder:
1. The system is in a steady-state operating condition. 2. The flow field in the denuder is a fully developed
laminar flow.
3. Only dry atmospheric aerosols are considered, that is, the relative humidity of the atmosphere is below the deliquescent humidity of investigated aerosols. 4. Effect of diffusion in the direction of flow is neglected
as compared to convection. This is a valid assump-tion since the Peclet number of the denuder is much greater than unity.
5. No production or reaction of the gas or aerosol occurs in the denuder.
6. Gas-particle equilibrium, with a modification by introducing a of less than unity, is assumed at the interfaces of solid particles and their surrounding gases and is independent of the denuder residence time.
7. The collection efficiencies of HNO3 gas and
NH4NO3-containing aerosols are 100% on the
collector wall surface. 2.2. Model development
For NH4NO3 aerosols undergoing simultaneous
diffusion and evaporation in the denuder, the steady-state general dynamic equation is expressed as (Fried-lander, 1977) Uavf ðrÞ qn qz¼ 1 r q qr r qðDpnÞ qr þqðGnÞ qv ; ð1Þ
where Uavis the average gas velocity in the denuder, f ðrÞ is the flow profile, r is the radial distance, n is the volume distribution function, z is the axial distance, Dp is the particle diffusivity, G is the volume evaporation/ condensation rate (Go0 when evaporation is faster than condensation), and v is the volume of an aerosol. The left-hand side (LHS) term accounts for convection of aerosol stream. The first term on the right-hand side (RHS) accounts for radial diffusion of particles. The second RHS term accounts for the loss or gain of particles by evaporation/condensation at rate G: The fully developed flow profile in the denuder is written as (Bird et al., 1960)
f ðrÞ ¼2½ð1 k 2Þlnðr=r
0Þ þ lnð1=kÞð1 r2=r20Þ
ð1 þ k2Þlnð1=kÞ ð1 k2Þ ; ð2Þ where k is the inner-to-outer radius ratio, r0is the outer radius of the denuder. The G can be expressed as (Friedlander, 1977) G ¼dv dt¼ 2pDmMdpðp1 pdÞ RrpT 1 þ Kn 1 þ 0:3773Kn þ 1:33Knð1 þ KnÞ=a ð3Þ where Dmis the diffusivity of NH4NO3monomer, M is
the molecule weight, dp is the particle diameter, p1and pdare the vapor pressures of NH4NO3monomer in the
bulk phase and in equilibrium with the aerosol surface, R is the gas constant, rpis the particle density, T is the absolute temperature in K, Kn is the Knudsen number of the particle (Kn is defined as 2l=dp; with l the mean free path of gas molecules).
The evaporation of NH4NO3 aerosol leads to the
release of HNO3 and NH3 gases. The mass balance
equation of the gaseous species j is expressed as (Lu et al., 1995) Uavf ðrÞ qCj qz ¼ Dj r q qr r qCj qr 1 vm Z N 0 Gn dv; ð4Þ
where Cj is the concentration of species j (j¼ HNO3or NH3), Dj is the diffusivity of species j; vmis the molar volume of NH4NO3 aerosol. The LHS term considers
convection of species j: The first RHS term considers
radial diffusion of species j while the second RHS term considers the gain of species j by aerosol evaporation.
The sectional approach was employed for simulating the evolution of aerosol size distribution along the denuder (Gelbard et al., 1980). The studies of Landgrebe and Pratsinis (1990) showed that the volume-square-based model provided a better prediction on the evolution of aerosol-size distribution than the number-based and number-based models. Therefore, the volume-square-based model was employed in this study. By dividing the entire aerosol-size domain into L arbitrary sections, the volume-squared size distribution is approximated as follows:
QI¼ Z vI
vI 1
nðvÞv2dv; ð5Þ
where QIis the aerosol volume-squared concentration in section I ð1pIpLÞ; vI 1 and vI represent the volumes of the smallest and largest aerosols, respectively, in section I.
The governing equations for the sectional volume-squared distribution can be deduced from Eqs. (1) and (4) to give the following partial differential equations (PDEs): qQIðr; zÞ qz ¼ 1 er q qr r q qrðoQIðr; zÞÞ þ 1 Uavf ðrÞ ðwIQI HIþ HI 1Þ; ð6Þ Uavf ðrÞ qCNH3 qz ¼ DNH3 r q qr r qCNH3 qr ðS 1ÞzcX I ¼L I ¼1 QI ðvI vI 1Þ ; ð7Þ Uavf ðrÞ qCHNO3 qz ¼ DHNO3 r q qr r qCHNO3 qr ðS 1ÞzcX I ¼L I ¼1 QI ðvI vI 1Þ ; ð8Þ
where S is the system saturation ratio (S is defined as CHNO3CNH3=KP; with KP the equilibrium constant of NH4NO3particles with the two associated gases, HNO3
and NH3). When the S value is unity, the system is in an
external state of equilibrium. The driving force of the evaporation process of NH4NO3particles is determined
by ð1 SÞ: The definitions of e; z; o; w; HI; HI1; z and c are listed in Table 1.
It is assumed that the NH4NO3 aerosols and
associated gases, HNO3and NH3,reach equilibrium at
the denuder inlet (S ¼ 1 at z ¼ 0). So the inlet NH3
concentration can be assumed to be CNH3;in¼
Kp=CHNO3;in: The Kp value is taken from Mozurkewich (1993) and can be expressed as
ln Kp¼ 118:87 24084
The initial condition for the inlet aerosols is QIðr; 0Þ ¼
Z I I 1
v2nðv; 0Þ dv: ð10Þ
The denuder wall is designed as perfect sink for
NH4NO3 aerosol ðQI¼ 0Þ and HNO3 gas
(CHNO3 ¼ 0), but inert for NH3 gas (CNH3=qr ¼ 0). The physical quantity of QI¼ 0 means the aerosol volume-squared concentration is equal to zero at the denuder wall.
Eq. (6) describes aerosol diffusion and evaporation while Eqs. (7) and (8) consider the diffusion of NH3and
HNO3gases and the source term from evaporation of
NH4HNO3 aerosol. These equations along with the
appropriate boundary conditions constitute a set of coupled PDEs. Using an explicit-finite difference scheme at P radial points across the denuder, the three PDEs are transformed to ordinary differential equations (ODEs). These ODEs are then solved by a stiff ODE solver, DIVPAG (IMSL, 1987). The integrations are solved using the Gauss–Lagendre quadrature method (Horn-beck, 1982). The cup-mixing average is used to evaluate effects of process parameter on the evolution of the aerosol and the gaseous concentrations.
3. Results and discussion 3.1. Numerical stability analysis
Prior to analysis of the denuder performance, the numerical stability of the model is checked. The
stability of numerical solutions as a function of grid size ðDr; DzÞ and number of sections ðLÞ employed in the model was investigated by a sensitivity analysis. Trials with varying P radial points from 10 to 30, Dz from 106 to 104cm, and L from 5 to 20 were conducted and the results showed that nearly identical solutions are achieved for P radial points X20, Dzp104 and LX10: Therefore, radial points of 20, Dz of 104cm, and L of 10 were used in the following studies.
3.2. Model verification for diffusional deposition of HNO3gas and NH4NO3aerosol
The present model for diffusional deposition of HNO3 gas and NH4NO3 aerosol was first examined
by comparing its numerical solutions with those predicted by the Gormley–Kennedy equation (Gormley and Kennedy, 1949) in a tubular pipe. The inner-to-outer radius ratio, k; was set to zero and the flow profile is set to f ðrÞ ¼ 2ð1 r2=r2
0Þ: The boundary conditions were set to axis of symmetry and zero wall concentration. The aerosol diffusivity was estimated by the Stokes–Einstein expression (Friedlander, 1977) while the diffusivities of NH3, HNO3 and NH4NO3
monomer were chosen as 0.227, 0.118 (Winiwarter, 1989) and 6.03 102cm2/s (Bai et al., 1995), respec-tively. The comparison results showed that the numer-ical solutions predicted by the present model are in excellent agreement with the Gormley–Kennedy solution.
Table 1
Definitions of parameters used in the governing equations
w 1 ðvI vI 1Þ 12DmMðp1 pdÞ RrpT RvI vI 1 1 d2 p 1 þ Kn 1 þ 0:3773Kn þ 1:33Knð1 þ KnÞ=a ! dv e Uavf ðrÞRvvI 1I 1=v2dv ¼ Uavf ðrÞð1=vI 1 1=vIÞ o RvI vI 1kbTCc=v 23pmd pdv j RvI þ1 vI ðdp=v 2Þð1 þ KnÞ=ð1 þ 0:3773Kn þ 1:33Knð1 þ KnÞ=aÞ dv z ð1=vmÞð2pDmpdM=rpRTÞ HI mI=ðmI þ1 mIÞwQI HI 1 mI 1=ðmI mI 1ÞwI 1QI 1
Note: v; volume of an aerosol; Dm; diffusivity of NH4NO3monomer; M; molecular weight; dp; particle diameter; p1; bulk phase vapor
pressure of NH4NO3 monomer; pd; equilibrium vapor pressure of NH4NO3 monomer; R; gas constant; rp; particle density; Kn;
Knudsen number; a; accommodation coefficient; Uav; average gas velocity; f ðrÞ; flow profile; r; radial distance; kb; Boltzmann’s
constant; T; ambient temperature; Cc; Cunningham correction factor; m; dynamic viscosity; vm; molar volume of NH4NO3aerosol;
mI ¼ VI=NI; VI; aerosol volume in section I; NI; aerosol number in section I; and QI; aerosol volume-squared concentration in
3.3. Model verification for evaporation of a monodisperse NH4NO3aerosol
The present model for evaporation of a monodisperse NH4NO3 aerosol was then evaluated by the measured
data of Dassios and Pandis (1999). The NH4NO3
aerosols of 80–220 nm diameter were evaporated in the reactor under relative humidity of around 10% and temperature of 20–271C. An average residence time of 30 s was employed in most experiments. The a value is unknown and needs to be justified.
Fig. 1 shows the change in size of the NH4NO3
aerosols at 211C predicted by the present model and measured by Dassios and Pandis (1999). The employed a values were equal to 0.5, 0.8 and 1. As can be seen, close agreement was obtained between the predicted and measured results in the a range of 0.5–1.0. The a ¼ 0:8 case provides a best fit to the measured data and was used in the following studies.
3.4. Model verification for evaporation of a polydisperse NH4NO3aerosol
The present model for evaporation of a polydisperse NH4NO3aerosol in the denuder was also tested by the
measured data of Harrison et al. (1990). The inlet NH4NO3aerosols were lognormally distributed with a
geometric mean radius of 0.46 mm and a standard deviation of 1.6. The temperature and relative humidity were 20721C and 30–60%, respectively. The inlet number concentration of NH4NO3 aerosol was set to
105/cm3(Bai et al., 1995).
Fig. 2 shows the measured and predicted results on the change of particle radius of dry NH4NO3aerosols in
the denuder. It is seen that the predicted results agree well with the measured data.
3.5. Model application in the field study
The present model was then applied to predict the measured data of Bai and Wen (2000) for the relative sampling errors of HNO3 gas by the annular denuder
system. Bai and Wen (2000) used three or four denuders in series coated with NaCl for HNO3 sampling.
Assuming that the absorption efficiency of HNO3 gas
ðZÞ is high and the absorption efficiency of the interfering species ðZintÞ is low, the first and second denuders collected both HNO3gas and interfering species (CHNO3 and Cint) and the third and fourth denuders collected interfering species only; the collected gas concentrations by each denuder are
first denuder ¼ CHNO3Z þ CintZint; ð14Þ
second denuder ¼ CHNO3ð1 ZÞ þ CintZint; ð15Þ third denuder ¼ fourth denuder ¼ CintZint: ð16Þ The HNO3gas concentration can thus be calculated
by subtracting double the concentration collected by the third from the sum of those collected by the first and second denuders. The percentage excess of absorbed HNO3 gas is calculated as CintZint=CHNO3 100%: Detailed information concerning HNO3artifact can be
seen in Bai and Wen, 2000. Because the size distribution of the atmospheric aerosols was not measured in their study, the atmospheric aerosol and gas properties of the Hsinchu area are adopted from average values of a typical range available in the literature (Lee and Hsu, 1996; Bai and Wen, 1998) and are listed in Table 2.
Fig. 3 shows the predicted and measured results on the percentage excess of absorbed HNO3 gas in the
denuder. As can be seen, the percentage excess of absorbed HNO3gas was underpredicted by the present
model. This can be attributed to the oxidation of other Inlet particle diameter (nm)
50 100 150 200 250 Final particle diameter (nm) 0 50 100 150 200 250 Measured data This study ( α α=0.5) This study ( α α=0.8) This study ( α α=1.0)
Fig. 1. Comparisons of the change in size for a monodisperse NH4NO3aerosol at 211C predicted by the present model and
measured by Dassios and Pandis (1999).
0 6 12 18 24 30 36 Particle geometric mean radius ( µ m) 0.0 0.1 0.2 0.3 0.4 0.5 Measured data This study
Residence time of aerosol in the denuder (min)
Fig. 2. Comparisons of the change in size for a polydisperse NH4NO3aerosol predicted by the present model and measured
gaseous N-compounds such as HNO2 and NOX to HNO3 gas, which was not considered in this study.
These reactions are still unclear in the literature and further studies are needed to determine the real mechanisms of oxidation and whether they can be controlled. Significant deviations were observed for inlet HNO3concentrationso0.2 mg/m3and could be as high
as 40%. However, as the inlet HNO3 concentrations
increased to 1.0 mg/m3, the deviations between the measured and predicted results became small. This is because for the same amount of HNO3 artifact the
percentage error is high for low inlet HNO3
concentra-tion since the denominator in the calculaconcentra-tion is low.
3.6. Predicting the excess of absorbed HNO3gas
Having established the accuracy and consistency of the present model on the evaporation and diffusional deposition of NH4NO3 aerosol in the denuder, a
sensitivity analysis of the important parameters on the HNO3artifact was carried out. The denuder described in
the study of Bai and Wen (2000) was employed. The
investigated parameters are: inlet aerosol number con-centration (Ni), inlet mass median diameter (MMDi),
nitrate mass fraction of o2.5 mm aerosols ðMfÞ; and inlet HNO3gas concentration (Ci). The employed values
of Table 2 are selected as the base case. Trials are conducted for each parameter in the high and low levels, keeping all other variables constants as the base case.
Figs. 4–7 show the predicted percentage excess of absorbed HNO3 as a function of ambient temperature
for varying Ni; MMDi, Mf and Ci; respectively. The residence time of aerosols in the denuder system was set to 1.0 s. The employed ambient temperature was in the range of 15–351C which is the typical temperature range
Table 2
Typical range of the atmospheric aerosol and gas properties Parameters Typical range Employed values Ni(#/cm3) 104–106 105
MMDi(mm) 0.1–1.0 0.6
Mf(%) 4–12 8
Ci(mg/m3) 0.2–1.0 0.6
Note: Ni; inlet aerosol number concentration; MMDi, inlet
aerosol mass medium diameter; Mf; nitrate mass fraction of less
than 2.5 mm aerosols; and Ci, initial HNO3gas concentration.
Initial HNO3concentration (µg/m3)
0 2 4 6 8 Percentage excess o f absorbed HNO 3 gas (% ) 0 20 40 60 80 100 Measured data This study
Fig. 3. Comparisons of the percentage excess of absorbed HNO3gas predicted by the present model and measured by Bai
and Wen (2000). Ambient temperature (oC) 0 10 20 30 40 Percentage excess of absorbed HNO 3 (% ) 0 20 40 60 80 100 120 140 Ni= 104/cm3 Ni= 105/cm3 Ni= 106/cm3
Fig. 4. Predicted percentage excess of absorbed HNO3gas as a
function of ambient temperature for varying inlet aerosol number concentration (Ni) with inlet conditions of aerosol mass
median diameter=0.6 mm, nitrate mass fraction of o2.5 mm aerosols=0.08, and inlet HNO3gas concentration=0.6 mg/m3.
Ambient temperature (oC) 0 10 20 30 40 Percentage excess of absorbed HNO 3 (% ) 0 20 40 60 80 100 120 140 MMD = 1.0µm MMD = 0.6µm MMD = 0.1µm
Fig. 5. Predicted percentage excess of absorbed HNO3gas as a
function of ambient temperature for varying inlet aerosol mass median diameter (MMDi) with inlet conditions of aerosol
number concentration=105/cm3, nitrate mass fraction of
o2.5 mm aerosols=0.08, and inlet HNO3 gas
in Taiwan. It is seen that the percentage excess of absorbed HNO3 increased as the ambient temperature
increased. This can be attributed to the fact that the NH4NO3 evaporation rate is dependent on the partial
pressure driving force, p1 pd as indicated in Eq. (3). The pdis a function of ambient temperature. An increase of ambient temperature causes an increase of pd; and thus results in an increase of partial pressure driving force and NH4NO3evaporation rate.
As can be noted in Fig. 4, the percentage excess of absorbed HNO3 gas increased as Ni increased. This is due to the fact that an increase of Niprovides a higher NH4NO3mass available for evaporation and release of
more HNO3 gases. Furthermore, more NH4NO3
-con-taining aerosols were deposited on the denuder wall for a higher Ni; which may also enhance the HNO3artifact.
The Ni=106 case has a very significant effect on the HNO3 artifact under ambient temperature >301C. At
an ambient temperature of 301C, the percentage excess (artifact concentration) of absorbed HNO3 gas was
equal to 33% (0.2 mg/m3), 5% (0.03 mg/m3) and 1% (0.006 mg/m3) for aerosols with Ni of 104, 105 and 106cm3, respectively.
It is seen in Fig. 5 that a higher sampling artifact is introduced for aerosols with a larger MMDi. This is due
to the fact that for NH4NO3-containing aerosols with
same Mf and Ni; a larger MMDiaerosol has a higher
NH4NO3 mass available for the evaporation and
diffusional deposition processes. At an ambient tem-perature of 301C, the percentage excess (artifact concentration) of absorbed HNO3 gas was equal to
45% (0.27 mg/m3), 20% (0.12 mg/m3) and 2% (0.012 mg/ m3) for aerosols with MMD
i of 1.0, 0.6 and 0.1 mm,
respectively.
It is noted in Fig. 6 that the percentage excess of absorbed HNO3gas increased as Mf increased. This is again due to a high NH4NO3 mass available for
evaporation and diffusional deposition processes. However, the effect of Mf on the HNO3 artifact is
relatively small as compared to other parameters. This is because the typical value of Mf in Taiwan was usually o0.1 (Chang et al. 2001). Hence, the Mf was changed only by a factor of three. At an ambient temperature of 301C the percentage excess (artifact concentration) of absorbed HNO3 gas increased from
2% (0.012 mg/m3) to 4% (0.024 mg/m3) as Mf increased from 0.04 to 0.12.
As can be seen from Fig. 7, an increase of Ciresults in a reduction of the percentage excess of absorbed HNO3
gas increased as Cidecreased, but an opposite trend was observed for artifact concentration. For example, at an ambient temperature of 401C the percentage excess of absorbed HNO3gas decreased from 55% to 18% as Ci increased from 0.2 to 1.0 mg/m3, respectively, but the artifact concentration increased from 0.11 to 0.18 mg/m3. This is because the NH3–HNO3–NH4NO3 system was
assumed to be in equilibrium at the entrance of the denuder. Therefore, the inlet NH3 gas concentration
depends on the inlet HNO3 gas concentration. For a
high inlet HNO3gas concentration, the inlet NH3 gas
concentration is low. As the HNO3 gas is being
absorbed rapidly at the denuder wall upon entering the denuder, the evaporation driving force, S 1; would be high, which caused a high NH4NO3 evaporation rate
and thus led to the release of more HNO3gas. However,
when expressed as the percentage excess of absorbed HNO3gas, the artifact becomes relatively low for a high
inlet HNO3 gas concentration since the denominator
is high. Ambient temperature (oC) 0 10 20 30 40 Percentage excess of absorbed HNO 3 (% ) 0 2 4 6 8 10 12 14 16 18 Mf= 0.12 Mf= 0.08 Mf= 0.04
Fig. 6. Predicted percentage excess of absorbed HNO3gas as a
function of ambient temperature for varying nitrate mass fraction of o2.5 mm aerosols ðMfÞ with inlet conditions of
aerosol mass median diameter=0.6 mm, aerosol number con-centration=105/cm3, and HNO3gas concentration=0.6 mg/m
3 . Ambient temperature (oC) 0 10 20 30 40 Percentage excess o f absorbed HNO 3 gas (% ) 0 10 20 30 40 50 60 Ci= 1.0µg/m3 Ci= 0.6µg/m3 Ci= 0.2µg/m3
Fig. 7. Predicted percentage excess of absorbed HNO3gas as a
function of ambient temperature for varying inlet HNO3gas
concentration (Ci) with inlet conditions of aerosol mass median
diameter=0.6 mm, aerosol number concentration=105/cm3, and nitrate mass fraction ofo2.5 mm aerosols=0.08.
Fig. 8 shows the effects of the residence time of aerosols in the denuder from 0 to 5 s on the percentage excess of absorbed HNO3 gas. With the exception
of Ci; the parameter values were adopted from the base
case of sensitivity analysis. The Ciwas set to 0.1 mg/m3. As can be seen, the magnitude of HNO3 artifact
was proportional to the residence time of aerosols in the denuder. This is because more HNO3 gases
were evaporated from nitrate-containing aerosols and diffused to the denuder wall at a longer residence time.
3.7. Practical application of the present model
Potential application of the present model was demonstrated on estimating the HNO3 artifact using
literature data sampled in USA (Keeler and Spengler, 1991), Taiwan (Perng, 1995), Netherlands (Hoek et al., 1996), Korea (Lee et al., 1997) and Japan (Matsumoto and Okita, 1998) and the results were listed in Table 3. These studies provided the measured data of HNO3gas
concentration and nitrate concentration in o2.5 mm aerosols. However, the information of aerosol size distribution and ambient temperatures was not avail-able. Therefore, it is assumed that the size distribution of o2.5 mm aerosols was lognormally distributed with Ni of 105/cm3 and MMDi of 0.6 mm. The HNO3 artifact
was estimated using the highest monthly average temperature in 1 yr. As can be seen in Table 3, 0 1 2 3 4 5 Percentage excess of absorbed HNO 3 gas (% ) 0 20 40 60 80
Residence time of aerosol in the denuder (sec)
Fig. 8. Predicated percentage excess of absorbed HNO3gas as
functions of residence time of aerosols in the denuder with inlet conditions of HNO3 gas concentration=0.1 mg/m
3
, aerosol mass median diameter=0.6 mm, aerosol number concentra-tion=105/cm3, nitrate mass fraction of o2.5 mm aero-sols=0.08, and ambient temperature=301C.
Table 3
Estimated measurement artifacts of HNO3gas by one annular denuder for typical gas and aerosol properties available in the literature
Country Taiwan USA Netherlands Korea Japan
Reference Perng (1995) Keeler and Spengler (1991) Hoke et al. (1996) Lee et al. (1997) Matsumoto and Okita (1998) Average HNO3gas concentration
(mg/m3) 0.30 1.10 0.30 0.80 1.60 [NO3 ] in less than 2.5 mm aerosols (mg/m3) Avg. 10.0 0.4 25.0 2.50 2.24 Max. 21.6 0.79 51.4 4.53 3.20
The highest monthly average temperature in 1 ha(1C) 35 26 27 24 25 Estimated HNO3 artifactb(mg/m3) Avg. 0.11 0.0071 0.093 0.017 0.017 Max. 0.23 0.03 0.22 0.05 0.04 Estimated percentage excess of absorbed HNO3(%) Avg. 35 1.8 31 2.1 1.1 Max. 77 3.6 73 6.2 2.2
aObtained fromhttp://www.wunderground.com/global/ko.html. bThe HNO
3 artifact was estimated using the highest monthly average temperature in 1 yr with the assumption that the size
distribution ofo2.5 mm aerosols was lognormally distributed with inlet aerosol number concentration of 105/cm3and aerosol mass
significant measurement artifact could be found in Taiwan and Netherlands. This is due either to low HNO3 gas concentration and high nitrate
concentra-tions ino2.5 mm aerosols or to high ambient tempera-ture. The estimated HNO3 artifact concentration and
the percentage excess of absorbed HNO3gas could be as
high as 0.23 mg/m3and 77%, respectively. The estimated HNO3 artifacts for the field data sampled in USA,
Korea and Japan were very low as compared to those sampled in Taiwan and Netherlands.
4. Conclusions
A mathematical model was developed to determine the HNO3 artifact in the denuder due to evaporation
and diffusional deposition of NH4NO3-containing
aerosols. The model performance was validated by comparing its predictions with the measured data for evaporation and diffusional deposition of mono-disperse and polymono-disperse NH4NO3aerosols available in
the literature. The neglect of the oxidation of other
gaseous N-compounds such as HNO2 and NOX to
HNO3 gas may lead to the underprediction of the
percentage excess of absorbed HNO3 gas measured in
the field.
The model was then applied to evaluate effects of the important parameters on the HNO3 artifact. The
investigated parameters are: inlet aerosol number con-centration (Ni), inlet mass median diameter (MMDi),
nitrate mass fraction of o2.5 mm aerosols ðMfÞ; and inlet HNO3 gas concentration (Ci). These parameters were usually neglected and unreported in the literature for evaluation of the sampling artifact in the denuder system. For the atmospheric conditions with high values of ambient temperature, Ni; MMDiand a
low value of Ci; the percentage excess of absorbed HNO3gas may be significant and therefore should be
avoided.
Potential application of the present model on estimating HNO3 artifact was demonstrated using
literature data sampled in USA, Taiwan, Netherlands, Korea and Japan. Significant measurement artifact could be found in Taiwan and Netherlands due either to low HNO3 gas concentration and high nitrate
concentrations ino2.5 mm aerosols or to high ambient temperature.
Ultimately, there are other artifact sources neglected in the present model. Therefore, further experimental and theoretical studies are needed to determine the real mechanisms of the oxidation of other N-compounds to HNO3 gas in the denuder and whether they can be
controlled. The field conditions leading to a significant measurement artifact in the denuder can then be avoided.
Acknowledgements
Support from the National Science Council, Taiwan (NSC89-2211-E-009-058) is gratefully acknowledged.
References
Bai, H., Wen, H.Y., 1998. Applicability of an annular denuder to measure acidic and basic gases in the Hsinchu Area. Journal of the Chinese Institute of Environmental En-gineering 8, 73–81.
Bai, H., Wen, H.Y., 2000. Performance of the annular denuder system with different arrangements for HNO3and HNO2
measurements in Taiwan. Journal of the Air and Waste Management Association 50, 125–130.
Bai, H., Lu, C., Ling, Y.M., 1995. A theoretical study on the evaporation of dry ammonium chloride and ammonium nitrate aerosols. Atmospheric Environment 29A, 313–321. Benner, C.L., Eatough, D.J., Eatough, N.L., Bhardwaja, P.,
1991. Comparison of annular denuder system and filter pack collection of HNO3(g), HNO2(g)and particulate-phase
nitrate, nitrite and sulfate in the south-west desert. Atmo-spheric Environment 25A, 1537–1545.
Bird, R.B., Stewart, W.E., Lightfood, E.N., 1960. Transport Phenomena. Wiley, New York.
Biswas, P., Lu, C., Xu, M., Pratsinis, S.E., 1990. Design equations for gas sampling in diffusion denuders and the effect of particle diffusion. In: Mathai, C.V. (Ed.), Transactions on Visibility and Fine Particles, AWMA, Publication No. TR-17, Estes Park, Colorado, USA, October 1989 pp. 131–145.
Chang, K.F., Fang, G.C., Lu, C., Bai, H., 2001. Particle size distribution and anion content at a traffic site in Sha-Lu, Taiwan. Chemosphere 45, 791–799.
Cruz, C.N., Dassios, K.G., Pandis, S.N., 2000. The effect of dioctyl phthalate films on the ammonium nitrate aerosol evaporation rate. Atmospheric Environment 34, 3897–3905. Dassios, K.G., Pandis, S.N., 1999. The mass accommodation coefficient of ammonium nitrate aerosol. Atmospheric Environment 33, 2993–3003.
Durham, J.L., Spiller, L.L., Ellestad, T.G., 1987. Nitric acid– nitrate aerosol measurements by a diffusion denuder: a performance evaluation. Atmospheric Environment 21, 589–598.
Forrest, J., Spandau, D.J., Tanner, R.L., Newman, L., 1982. Determination of atmospheric nitrate and nitric acid employing a diffusion denuder with a filter pack. Atmo-spheric Environment 16, 1473–1485.
Friedlander, S.K., 1977. Smoke, Dust and Haze. Wiley, New York.
Gelbard, F., Tambour, Y.R., Seinfeld, J.H., 1980. Sectional representation for simulating aerosols dynamics. Journal of Colloid and Interface Science 76, 541–556.
Gormley, P.G., Kennedy, M., 1949. Diffusion from a stream flowing through a cylindrical tube. Proceedings of the Royal Irish Academic A 52, 163–169.
Harrison, R.M., Sturges, W.T., Kitto, A.N., Li, Y., 1990. Kinetics of evaporation of ammonium chloride and ammonium nitrate aerosols. Atmospheric Environment 24A, 1883–1888.
Hoek, G., Mennen, M.G., Allen, G.A., Hofschreuder, P., Meulen, T.V.D., 1996. Concentrations of acidic air pollu-tants in the Netherlands. Atmospheric Environment 30 (18), 3141–3150.
Hornbeck, R.W., 1982. Numerical Methods. Prentice-Hall, Englewood Cliffs, NJ.
IMSL (International Mathematics and Statistics Libraries), 1987. Contents Document, Vol. 2. Version 1.0, IMSL, Houston, Texas.
Keeler, G.J., Spengler, J.D., 1991. Acid aerosol measurements at a suburban Connecticut site. Atmospheric Environment 25, 681–690.
Koutrakis, P., Wolfson, J.M., Slater, J.L., Brauer, M., Spengler, J.D., Stevens, R.K., Stone, C.L., 1988. Evaluation of an annular denuder/filter pack system to collect acidic aerosols and glass. Environmental Science and Technology 22, 1463–1468.
Landgrebe, J.D., Pratsinis, S.E., 1990. A discrete-sectional model for particulate production by gas-phase chemical reaction and aerosol coagulation in the free-molecular regime. Journal of Colloid and Interface Science 139, 63–85.
Larson, T.V., Taylor, R.L., 1983. On the evaporation of the ammonium nitrate aerosol. Atmospheric Environment 17, 2489–2495.
Lee, C.T., Hsu, W.C., 1996. The source apportionment of urban aerosols from chemical properties of aerosol spectra near atmospheric sources. Journal of the Chinese Institute of Engineer 19, 1–13.
Lee, H.S., Wadden, R.A., Scheff, P.A., 1993. Measurement and evaluation of acid air pollutants in Chicago using an
annular denuder system. Atmospheric Environment 27A, 543–553.
Lee, H.S., Kang, B.W., Cheong, J.P., Lee, S.K., 1997. Relationship between indoor and outdoor air quality during the summer season in Korea. Atmospheric Environment 31 (11), 1689–1693.
Lu, C., Bai, H., Lin, Y.M., 1995. A model for predicting performance of an annular denuder system. Journal of Aerosol Science 26, 1117–1129.
Matsumoto, M., Okita, T., 1998. Long term measurements of atmospheric gaseous and aerosol species using an annular denuder system in Nara, Japan. Atmospheric Environment 32, 1419–1425.
Mozurkewich, M., 1993. The dissociation constant of ammo-nium nitrate and its dependence on temperature, relative humidity and particle size. Atmospheric Environment 27A, 261–270.
Perng, S.N., 1995. A study of ambient ionic species using diffusion denuder, dichotomous and Hi-Vol PM10samplers.
MS Thesis, Institute of Environmental Engineering, National Chiao Tung University, Taiwan.
Pratsinis, S.E., Xu, M., Biswas, P., Willeke, K., 1989. Theory for aerosol sampling through annular diffusion denuders. Journal of Aerosol Science 20, 1597–1600.
Tsai, C.J., Perng, S.B., Chiou, S.F., 2000. Use of two different acidic aerosol samplers to measure acidic aerosols in Hsinchu, Taiwan. Journal of the Air and Waste Manage-ment Association 50, 2120–2128.
Winiwarter, W., 1989. A calculation procedure for the determination of the collection efficiency in annular denuder. Atmospheric Environment 23, 1997–2002.