Post on 03-Feb-2021
Hindawi Publishing CorporationInternational Journal of GeophysicsVolume 2013, Article ID 612375, 13 pageshttp://dx.doi.org/10.1155/2013/612375
Research ArticleEvaluation of Vapor Pressure Estimation Methods for Use inSimulating the Dynamic of Atmospheric Organic Aerosols
A. J. Komkoua Mbienda,1 C. Tchawoua,2 D. A. Vondou,1 and F. Mkankam Kamga1
1 Laboratory of Environmental Modeling and Atmospheric Physics, Department of Physics, Faculty of Science, University of Yaounde 1,P.O. Box 812, Yaounde, Cameroon
2 Laboratory of Mechanics, Department of Physics, Faculty of Science, University of Yaounde 1, P.O. Box 812, Yaounde, Cameroon
Correspondence should be addressed to A. J. Komkoua Mbienda; kombiend@gmail.com
Received 24 March 2013; Accepted 5 June 2013
Academic Editor: Robert Tenzer
Copyright Β© 2013 A. J. Komkoua Mbienda et al.This is an open access article distributed under the Creative Commons AttributionLicense, which permits unrestricted use, distribution, and reproduction in anymedium, provided the originalwork is properly cited.
The modified Mackay (mM), the Grain-Watson (GW), Myrdal and Yalkovsky (MY), Lee and Kesler (LK), and Ambrose-Walton(AW) methods for estimating vapor pressures (πvap) are tested against experimental data for a set of volatile organic compounds(VOC). πvap required to determine gas-particle partitioning of such organic compounds is used as a parameter for simulating thedynamic of atmospheric aerosols. Here, we use the structure-property relationships of VOC to estimate πvap. The accuracy of eachof the aforementionedmethods is also assessed for each class of compounds (hydrocarbons, monofunctionalized, difunctionalized,and tri- andmore functionalized volatile organic species). It is found that the bestmethod for eachVOCdepends on its functionality.
1. Introduction
Atmospheric aerosols (AA) have a strong influence on theearthβs energy balance [1] and a great importance in theunderstanding of climate change and human health (respira-tory and cardiac diseases, cancer).They are complexmixturesof inorganic and organic compounds, with compositionvarying over the size range from a few nanometers to severalmicrometers. Given this complexity and the desire to con-trol AA concentration, models that accurately describe theimportant processes that affect size distribution are crucial.Therefore, the representation of particle size distributionis of interest in aerosol dynamics modeling. However, inspite of the impressive advances in the recent years, ourknowledge of AA and physical and chemical processes inwhich they participate is still very limited, compared tothe gas phase [2]. Several models have been developed thatinclude a very thorough treatment of AA processes such asin Adams and Seinfeld [3], Gons et al. [4], and Whitby andMcMurry [5]. Indeed, the evolution of size distribution of AAis made by a mathematical formulation of processes calledthe general dynamic equation (GDE). It is well known thatthe first step in developing a numerical aerosol model is toassemble expressions for the relevant physical processes. The
second step is to approximate the particle size distributionwith a mathematical size distribution function. Thus, thetime evolution of the particle size distribution of aerosolsundergoing coagulation, deposition, nucleation, and conden-sation/evaporation phenomena is finally governed by GDE[1]. This latter phenomenon is characterized by the mass fluxπΌπfor volatile species π between gas phase and particle which
is computed using the following expression [6]:
πΌπ=
πππ
ππ‘= 2ππ·
π
ππππFS (πΎππ , πΌπ) (π
π
πβ ππ
π) . (1)
πFS describes the noncontinuous effects [7]. When πΌπ β₯0 (πΌπ
β€ 0), there is condensation (evaporation). ππ πis
assumed to be at local thermodynamic equilibrium with theparticle composition [8] and can be obtained from the vapourpressure (πvap
π) of each volatile compound π with average
molar mass of the atmospheric aerosol, mole fraction, andactivity coefficient, through the following equation:
ππ
π= π₯
πΎπvapπ
π106
π π, (2)
where πvapπ
is given by the Clausius-Clapeyron law:
πvapπ
(π) = πvapπ
(298) exp[β(π»vap/π )((1/π)β(1/298))]. (3)
2 International Journal of Geophysics
Parameters in (1)β(3) are named in Table 1.In (3), π»vap is equal to 156 kJ/mol as stated by Derby
et al. [8]. Consequently, the vapour pressures of all aerosolcompounds are needed to calculate the mass flux of con-densing and evaporating compounds. The knowledge of πvap
π
for organic compound π at the atmospheric temperature πis required whenever phase equilibrium between gas phaseand particle is of interest. Often, most of the compounds ableto condense have experimental vapor pressures unavailable,and because of that, their estimation becomes necessary. Tosolve this problem of estimation, many methods have beendeveloped. For example, in Tong et al. [9], a method basedon atomic simulation is applied only for compounds bearingacid moieties. Quantum-mechanical calculations are makingsteady effort in vapor pressure prediction (Banerjee et al.[10]; Diedenhofen et al. [11]). Furthermore, current mod-els describing gas-particle partitioning use semiempiricalmethods for vapor pressure estimation based on molecularstructure, often in the form of a group contribution approach.Therefore, these methods require in most cases molecularstructures (e.g., boiling point π
π, critical temperature π
π,
and critical pressure ππ), which usually have themselves to
be estimated. For example, The MY method [12] was usedby Griffin et al. [13] and Pun et al. [14] for modeling theformation of secondary organic aerosol. Jenkin [15], in a gasparticle partitioning model, used the modified form of theMackay method [16]. Some methods (Pankow and Asher[17], Capouet and MuΜller [18]) assume a linear logarithmicdependence (lnπvap
π) on several functional groups, but this
consideration fails when multiple hydrogen bonding groupsare present. Furthermore, more investigations are needed toclarify which method can give values closer to the exper-imental data. Camredon and Aumont [19], Compernolleet al. [20], and Barley and McFiggans [21] have made anassessment of different vapor pressure estimation methodswith experimental data for compounds of relatively highervolatility. For this latter reason, large differences in theestimated vapor pressure have been reported.
In this paper, our focus will be on the (i) evaluationof a number of vapor pressure estimation methods againstexperimental data using all volatile organic compoundspresent in our database and (ii) assessment of the accuracy ofeach of thesemethods on the base of each class of compounds.
The rest of the paper is organized as follows. In Section 2,we describe experimental data and methods. The results arepresented in Section 3. Finally, we summarize our finding inSection 4.
2. Data and Vapor PressureEstimation Methods
2.1. Experimental Data. The molecules selected in thisstudy have been identified during in situ campaigns [22]and during chamber experiments [23]. In fact, they arehydrocarbons, monofunctionalized and multifunctionalizedspecies, and bearing alcohol, aldehyde, ketone, carboxylicacid, ester, ether, and alkyl nitrate functions. The exper-imental vapor pressures are taken from NIST chemistry
Table 1: Nomenclature.
Parameters NamesπΌπ
Mass flux for volatile species πππ
The particle wet diameterππ
Mass of species ππ·π
πMolar diffusivity in the air of species π
ππ
πGas-phase concentration of species π
πFS Fuchs-Sutugin functionππ
πConcentration at the surface of species π
π Gas constantπ Temperature in Kelvinπ»vap Vaporisation enthalpyπ₯ Mole fractionπΎ Activity coefficientπvapπ
Vapour pressure of each volatile compound π
website (http://www.nist.gov/chemestry) and from Myrdaland Yalkowsky [12], Asher et al. [24], Lide [25], Yams [26],and Boulik et al. [27], and they have a range from 10β8 atmto 1 atm. Molecular properties (boiling point, critical tem-perature, and critical pressure) are also taken from the NISTchemistry website. All vapor pressure estimation methodsused in this study take into account these properties. In mostcases, these properties have also to be estimated.
2.2. Estimation of Molecular Properties
2.2.1. Boiling Temperature ππ. Using the boiling temperature
of Joback [28] and its extension, the group contributiontechnique denoted by πJob
πis written as
πJobπ
= 198 +β
π
πππ‘ππ, (4)
where π‘ππ
is the contribution of group π, and ππis the
occurrence of this group in the molecule. As in CamredonandAumont [19], the extension ismade by adding some othergroup contribution to take into account molecules bearinghydroperoxidemoiety (βOOH), alkyl nitratemoiety (ONO
2),
and peroxyacyl nitrate moiety (βC(=O)OONO2). Thus, the
first group is divided into the existing Joback groups βOβand βOH. The second group value is provided by the NISTchemistry website, and the last group value is provided byCamredon and Aumont [19] using boiling point value fromBruckmann and Willner [29].
2.2.2. Critical Temperature ππ. We have used Joback [28]
and Lydersenβs [30] techniques to estimate ππ. Denoting by
πJobπΆ
and πLydπ
the Joback and Lydersen critical temperatures,respectively, we have
πJobπ
=ππ
0, 584 + 0, 965βππππ‘πβ (βππππ‘π)2,
πLydπ
=ππ
0, 567 + βππππ‘πβ (βππππ‘π)2.
(5)
International Journal of Geophysics 3
New group contributions have been added by Camredonand Aumont [19] for hydroperoxide, alkyl nitrate, and perox-yacyl nitrate moieties:
π‘βOOHπ
= π‘βOβπ
+ π‘βOHπ
,
π‘βONO2π
= π‘βOβπ
+ π‘βNO2π
,
π‘βC(=O)OONO2π
= π‘βC(=O)Oπ
+ π‘βOβπ
+ π‘βNO2π
.
(6)
2.2.3. Critical Pressure ππ. The two techniques listed in the
previous section have been used to estimate critical pressureππ. Here, it is also assumed that π
πis the sum of group
contributions. Therefore, denoting by πJobπ
and πLydπ
theJoback and Lydersen critical pressures, respectively, we canwrite
πJobπ
=1
(0, 113 + 0, 0032π β βπππππ)2,
πLydπ
=π
(0, 34 β βπππππ)2,
(7)
where π is the molar mass, π is the number of atoms in themolecule, and π
πis the critical pressure contribution of group
π. The new group contributions described in the previoussubsection are taken into account here.
2.3. Vapor Pressure EstimationMethods. As said earlier, manymethods for vapor pressure estimation have been developedand are based on the Antoine or on the extended form of theClausius-Clapeyron equation. Let us present in what followseach of the five methods used.
2.3.1. The Myrdal and Yalkowsky (MY) Method. The MYmethod [12] starts from the extended form of the Clausius-Clapeyron equation obtained by using Eulerβs cyclic relation.Here, the expression of πvap is given by
lnπvap =Ξππ(ππβ π)
π π+
ΞπΆππ
π (ππβ π
πβ ln
ππ
π) , (8)
where Ξππis the vaporization entropy at the boiling temper-
ature, πΆππ
is the gas-liquid heat capacity, and π is the gasconstant. Ξπ
πused in this method is an empirical expression
given by Myrdal et al. [31]:
Ξππ= 86 + 0.4π + 1421HBN. (9)
In (9), the parameters π and HBN which characterize themolecular structure represent the torsional bond (see Vidal[32]) and the hydrogen bonding number (see [19, Section3.1.2]), respectively. ΞπΆ
ππis a linear dependence of π [32]:
ΞπΆππ
= β (90 + 2.5π) . (10)
Thus, in the MY method, vapor pressure is estimated bythe relatively simplified formula
lnπvap = β (21, 2 + 0, 3π + 177HBN) (ππβ π
π)
+ (10, 8 + 0, 25π) lnππ
π.
(11)
2.3.2. The Modified Mackay (mM) Method. Often calledsimplified expression of Baum [33], this method is also basedon the extended form of the Clausius-Clapeyron equation.The simplifying assumption here is to consider the ratioΞπΆππ/Ξππto be constant [34]:
ΞπΆππ
Ξππ
= β0.8. (12)
In (12), the vaporization entropy, Ξππ, takes into account
the van der Waals interactions and is based upon theTroutonβs rule. For its calculation, Lyman [35] has suggestedthe following expression:
Ξππ(ππ) = πΎπ(36, 6 + 8, 31 lnπ
π) , (13)
whereπΎπis a structural factor of Fishtine [36] which corrects
many polar interactions. It has different values as follows:
(i) πΎπ= 1 for nonpolar and monopolar compounds;
(ii) πΎπ= 1.04 for compounds with a weak bipolar char-
acter;(iii) πΎ
π= 1.1 for primary amines;
(iv) πΎπ= 1.3 for aliphatic alcohols.
Finally, themMmethod is reduced to the following equation:
lnπvap = πΎπ(4, 4 + lnπ
π) (1, 8
ππβ π
πβ 0, 8 ln
ππ
π) . (14)
2.3.3. The Grain-Watson (GW) Method. The GW method isbased on the following equation [37]:
lnπvap = Ξππ
[
[
1 β
(3 β 2ππ)π
ππ
β 2π(3 β 2ππ)πβ1
β lnππ]
]
,
(15)
whereπ = 0.4133β0.2575ππandππis inversely proportional
to ππ(ππ= π/π
π). Ξπ has the same form like that used in the
mMmethod.
2.3.4. The Lee and Kesler (LK) Method. Like the methodsdescribed previously, the LK method required critical tem-perature, critical pressure, and boiling temperature. Here, thevapor pressure is estimated on the base of Pitzer expansion[38]:
lnπvapπ
= π0(ππ) + π€π
1(ππ) , (16)
4 International Journal of Geophysics
where πvapπ
is reduced vapor pressure, ππis reduced temper-
ature, and π€ is the Pitzerβs acentric factor which accounts forthe nonsphericity of molecules:
π€ =β lnππβ π0(π)
π1 (π)(17)
with π = ππ/ππ. In (16) and (17), π0 and π1 are the Pitzerβs
functions which are polynomials in ππ. Lee and Kesler have
suggested the following equations [39]:
π0(ππ) = 5, 92714 β
6.09648
ππ
β 1.28862 lnππ+ 0.169347 lnπ6
π,
π1(ππ) = 15.2518 β
15.6875
ππ
β 13.4721 lnππ+ 0.43577 lnπ6
π.
(18)
2.3.5.TheAmbrose-Walton (AW)Method. AWmethod [40] isalso based on the Pitzer expansion. They have reported theiranalytical expressions of Pitzerβs functions in the form of aWagner type of vapor pressure equation:
π0(ππ)
=β5.9761π + 1.29874π
1.5β 0.60394π
2.5β 1.06841π
5
ππ
,
π1(ππ)
=β5.03365π + 1.11505π
1.5β 5.41217π
2.5β 7.46628π
5
ππ
.
(19)
In (19), π = 1 β ππ.
3. Results
3.1. Molecular Properties. We present in this section theresults obtained for the Joback and Lydersen techniquesdescribed previously. The accuracy of each of the five vaporpressure estimation methods used in this study is assessedtaking into account the reliability of pure substance propertyestimates. The reliability of the two techniques presented inSection 2.2 is therefore crucial.
In Figure 1, where the results of Joback technique aredisplayed, πJob
πis plotted against experimental values for a
set of 253 volatile organic compounds. The scatter tends tobe larger for boiling temperature higher than 500K. Thecorrelation coefficient (π 2 = 0.97) shows that estimatedvalues match very well experimental π
π. The root mean
square error (RMSE) and the mean absolute error (MAE)are, respectively, 17.60K and 12.65 K. This MAE agrees with12.9 K and 12.1 K calculated in Reid et al. [39] and Camredonand Aumont [19] for a set of 252 and 438 volatile organiccompounds, respectively. Hence, these results show that πJob
π
300
350
400
450
500
550
600
650
300 350 400 450 500 550 600 650
TJob
b(K
)
Texpb
(K)
N = 253
R2= 0.97
Figure 1: Estimated boiling point for a set of 253 species versusexperimental values for the Joback technique. The black line is the1 : 1 diagonal.
can be used in vapor pressure estimationmethods.Moreover,Joback reevaluated Lydersenβs group contribution scheme.He added several new functional groups and deducted newcontribution values.
The two techniques for the estimation of ππare compared
to experimental data of 138 compounds in Figure 2.Figures 2(a) and 2(b) show that the Joback and Lydersen
techniques give similar results forππvalues lower than 700K.
Joback technique shows a negative bias for ππhigher than
700K. This technique gives for the overall compounds anRMSE of 24.98K. This value is higher than the 19.81 K pro-vided by the Lydersen technique.Themean bias error (MBE)for πJobπ
and πLydπ
is β4.9 and β1.9, respectively. These resultsand Figures 2(a) and 2(b) show clearly that Joback techniqueunderpredicts critical temperature, mostly for compoundswhich can be condensed onto particle phase, with highboiling temperature. The experimental group contributionsprovided by Lydersen are therefore more accurate than thoseprovided by Joback. Thus, the Lydersen technique is morereliable than the Joback technique to estimate π
π.
Figure 3 shows πJobπ
and πLydπ
versus experimental valuesfor a set of 117 compounds. The RMSE is 4.5 atm and 2.6 atmfor Joback and Lydersen, respectively. According to Figures3(a) and 3(b), π
πestimated by Lydersen matches fairly better
(π 2 = 0.95) experimental data than ππestimated by Joback
(π 2 = 0.85). Joback technique considerably overpredictscritical pressure with MBE of 1,95 K higher than 0,39Kobtained with the Lydersen technique. In fact, besides groupcontribution, Lydersen technique takes into account molec-ular weight. Therefore, the Lydersen technique is retainedto the critical pressure estimation in this paper. This is inagreement with Poling et al. [38] who found that the Lydersentechnique is one of the best techniques for estimating criticalproperties. For the five vapor pressure estimation methodsdescribed previously, we will use estimated π
π, ππ, and π
π
because there is in general a lack of experimental data.
International Journal of Geophysics 5
400 500 600 700 800 900400
500
600
700
800
900
Texpc (K)
TJob
c(K
)
N = 138
R2= 0.93
(a)
400 500 600 700 800 900400
500
600
700
800
900
Texpc (K)
TLyd
c(K
)
N = 138
R2= 0.95
(b)
Figure 2: Estimated critical temperature for a set of 138 species versus experimental values for the (a) Joback technique and (b) Lydersentechnique. The black line is the 1 : 1 diagonal.
0 20 40 60 80 1000
20
40
60
80
100
Pexpc (atm)
PJob
c(atm
)
N = 117
R2= 0.85
(a)
0 20 40 60 80 1000
20
40
60
80
100
Pexpc (atm)
PLyd
c(atm
)
N = 117
R2= 0.95
(b)
Figure 3: Estimated critical pressure for a set of 117 species versus experimental values for the (a) Joback technique and (b) Lydersen technique.The black line is the 1 : 1 diagonal.
Some of the five methods described in Section 2.3 needππ, ππ, and π
π, while others need only π
πand π
π. This last
pure substance property is estimated by the Joback technique,and the two critical properties are estimated by the Lydersentechnique.
3.2. Evaluation of Vapor Pressure Estimation Methods. Theaccuracy of each method is assessed in terms of the meanabsolute error (MAE), the main bias error (MBE), andthe root mean square error (RMSE) (Table 3). The MBEmeasures the average difference between the estimated andexperimental values, while the MAE measures the averagemagnitude of the error. The RMSE also measures the error
magnitude, but gives some greater weight to the larger errors.Their expressions are given by
MBE = 1π
π
β
π=1
(logπvapest,π β logπvapexp,π) ,
MAE = 1π
π
β
π=1
logπvapest,π β logπ
vapexp,π
,
RMSE = β 1π
π
β
π=1
(logπvapest,π β logπvapexp,π)2
.
(20)
6 International Journal of Geophysics
Table 2: MAE, MBE, and RMSE of vapor pressure computed based on various methods.
GW LK mM MY AWMAE 0.3235 0.3240 0.3186 0.2652 0.3013MBE 0.0736 β0.1513 0.0263 0.0270 β0.0980RMSE 0.4496 0.4909 0.4426 0.3811 0.4525
Table 3: MAE, MBE, and RMSE of vapor pressure computed based on various methods for each class of compounds.
GW LK mM MY AWHydrocarbonsMAE 0.1455 0.1303 0.1424 0.1251 0.1384MBE 0.0547 0.0427 0.0486 0.0054 0.0603RMSE 0.1983 0.1750 0.1930 0.1711 0.1848Monofunctionalized speciesMAE 0.3692 0.3168 0.1424 0.2699 0.2900MBE 0.0355 β0.1712 β0.0189 β0.0402 β0.1141RMSE 0.4992 0.4471 0.4955 0.3891 0.4130Difunctionalized speciesMAE 0.4494 0.5641 0.4109 0.4291 0.5063MBE 0.2875 β0.3322 0.2084 0.2378 β0.2456RMSE 0.5506 0.6978 0.5120 0.5398 0.6291Tri- and more functionalized speciesMAE 0.4407 0.5946 0.4596 0.4261 0.5489MBE 0.0536 β0.3661 β0.0339 0.1632 β0.2739RMSE 0.5494 0.8386 0.5592 0.5065 0.7704
In (20),πvapest,π andπvapexp,π are the estimated and experimental
values of VOC π, respectively, and π is the total numberof VOC. We have also used linear correlation coefficientwhich measures the degree of correspondence between theestimated and experimental distributions.
The logarithms of vapor pressures estimated at π =298K for different methods are compared in the scatter plotsshown in Figure 4 for a set of 262VOC. CorrespondingMAE,MBE, and RMSE are given in Table 1. In this figure, it isclear that all the five methods give similar scatter for vaporpressures higher than 10β5 atm. The species concerned arehydrocarbons (Figure 5). For the set of 28 tri- and morefunctionalized species used in this study, vapor pressures arelower than 10β2 atm (Figure 8).
Figure 4(d) shows thatMYmethod is well correlated withexperimental values. For this method, we have one of the bestcorrelation coefficients π 2 = 0.968. This method shows nosystematic bias for vapor pressure lower than 10β5 atm, whileit is not the case for other methods. The MBE found here is0.027.This value is one of the lowest ones of the total VOC (seeTable 1). Thus, the MYmethod does not show any systematicbias. This method also provides the smallest values of MAEand RMSE (0.265 and 0.381, resp.). These results are inagreement with those found by Camredon and Aumont [19]using Ambrose technique to estimate critical properties. It isalso found that this method provides the smallest values ofthese errors for a set of 74 hydrocarbons (see Table 2). Thoseare compounds with vapor pressure higher than 10β2 atm.
This result is not the same for mono- and difunctionalizedspecies whose errors are some of the largest ones. The vaporpressures higher than 10β4 atm are fairly well estimated.Using a set of 45 multifunctional compounds, Barley andMcFiggans [21] found that MY method tends to overpredictvapor pressure of lower volatility compounds. Furthermore,it is important to note that MY method provides one ofthe poor results for difunctionalized VOC (see Table 2).Thus, for a set of 32 difunctionalized VOC, estimation valuesfit the experimental ones with a coefficient π 2 = 0.957(Figure 7) which is the smallest value obtained for this classof species. Vapor pressures are overpredicted with a biasof 0.24, while the RMSE = 0.54 is of the same order ofmagnitude as those obtained by other methods. In contrast,Figure 8 shows that MY method has the best correlation fortri- and more functionalized species and is therefore the bestmethod to estimate vapor pressure for this class of species.The systematic errors reported in Table 2 allow us to concludethat assumption. Indeed, the peculiarity of theMYmethod isthat it takes into account the molecular structure.
Figure 4(c) displays the results for the mM method. Thismethod gives one of the lowest scatterings with a coefficientπ 2= 0.957 and agrees with other methods for the highest
vapor pressures. It shows a positive bias (MBE = 0.026)for the total set of 262VOC. As the GW method, the mMmethod tends to overpredict vapor pressures lower than10β6 atm. Furthermore, these methods describe vaporisation
entropy by taking into account van der Waals interactions.
International Journal of Geophysics 7
β8
β6
β4
β2
0
β8 β6 β4 β2 0logPvapexp (atm)
logP
vap
GW
(atm
)
R2= 0.957
(a)
β8
β6
β4
β2
0
β8 β6 β4 β2 0logPvapexp (atm)
R2= 0.968
logP
vap
LK(a
tm)
(b)
β8
β6
β4
β2
0
β8 β6 β4 β2 0logPvapexp (atm)
R2= 0.957
logP
vap
mM
(atm
)
(c)
β8
β6
β4
β2
0
β8 β6 β4 β2 0logPvapexp (atm)
R2= 0.968
logP
vap
MY
(atm
)
(d)
β8
β6
β4
β2
0
β8 β6 β4 β2 0logPvapexp (atm)
logP
vap
AW
(atm
)
R2= 0.968
(e)
Figure 4: Logarithm of estimated vapor pressure of all VOCs used in this study versus experimental values for the (a) GW, (b) LK, (c) mM,(d) MY, and (e) AWmethods. The black line is the 1 : 1 diagonal.
The root mean square error (RMSE = 0.442) is close to thoseprovided by GW and AW methods. The mM method isthen less appropriate than the four others for all classes ofVOC.
Figure 5(c) shows that the predicted values match theexperimental values with a coefficient π 2 = 0.96 for 32
difunctionalized species. For this class of species, estimatesare provided with a positive bias (MBE = 0.208) and anRMSE of 0.512.These are the best values obtained from all thefive methods (see Table 2) for difunctionalized species. Thus,mM method is more accurate than others to estimate vaporpressure for difunctionalized species, but does not provide
8 International Journal of Geophysics
β8
β6
β4
β2
0
β8 β6 β4 β2 0
R2= 0.959
log Pvapexp (atm)
log P
vap
GW
(atm
)
(a)
β8
β6
β4
β2
0
β8 β6 β4 β2 0
R2= 0.967
log P
vap
LK(a
tm)
log Pvapexp (atm)
(b)
β8
β6
β4
β2
0
β8 β6 β4 β2 0
R2= 0.960
log P
vap
mM
(atm
)
log Pvapexp (atm)
(c)
β8
β6
β4
β2
0
β8 β6 β4 β2 0
R2= 0.962
log P
vap
MY
(atm
)
log Pvapexp (atm)
(d)
β8
β6
β4
β2
0
β8 β6 β4 β2 0
R2= 0.966
log P
vap
AW
(atm
)
log Pvapexp (atm)
(e)
Figure 5: Logarithm of estimated vapor pressure for a set of 74 hydrocarbons versus experimental values for the (a) GW, (b) LK, (c) mM, (d)MY, and (e) AWmethods. The black line is the 1 : 1 diagonal.
good results for monofunctionalized (Figure 3) and tri- andmore functionalized species (Figure 5).
It can be seen in Figure 7 that estimated values providedby GW method are strongly correlated with experimentalvalues (π 2 = 0.960) for difunctionalized species.Thismethodtends to overpredict vapor pressure for this class of species
(MBE = 0.288), but does not show any bias for other classes ofspecies (Table 2). Figure 8 shows that we have very acceptableresults for tri- and more functionalized species with RMSEand MAE equal to 0.549 and 0.440, respectively.
Except for tri- and more functionalized species (seeFigure 8), it is clear from Figures 4 to 7 that the LK method
International Journal of Geophysics 9
β8
β6
β4
β2
0
β8 β6 β4 β2 0
R2= 0.930
log P
vap
GW
(atm
)
log Pvapexp (atm)
(a)
β8
β6
β4
β2
0
β8 β6 β4 β2 0
R2= 0.964
log P
vap
LK(a
tm)
log Pvapexp (atm)
(b)
β8
β6
β4
β2
0
β8 β6 β4 β2 0
R2= 0.932
log P
vap
mM
(atm
)
log Pvapexp (atm)
(c)
β8
β6
β4
β2
0
β8 β6 β4 β2 0
R2= 0.960
log P
vap
MY
(atm
)
log Pvapexp (atm)
(d)
β8
β6
β4
β2
0
β8 β6 β4 β2 0
R2= 0.964
log P
vap
AW
(atm
)
log Pvapexp (atm)
(e)
Figure 6: Logarithm of estimated vapor pressure for a set of 128 monofunctionalized species versus experimental values for the (a) GW, (b)LK, (c) mM, (d) MY, and (e) AWmethods. The black line is the 1 : 1 diagonal.
gives accurate values, based upon best correlation coefficientvalues. This method is the second best one of the fivemethods, but it has the greatest systematic bias (RMSE =0.490, MAE = 0.32) for the total set of VOC and fordifunctionalized species (RMSE = 0.70, MAE = 0.56). TheMAE of hydrocarbons and monofunctionalized species are
0.142 and 0.36, respectively. For these species, Figures 5 and6 give the best correlations.
The AW and LK methods are both based on Antoineβsequation. According to all figures plotted, it is clear thatthese two methods give very similar results. The peculiarityof AW is that, for the monofunctionalized compounds,
10 International Journal of Geophysics
β8
β6
β4
β2
0
β8 β6 β4 β2 0
R2= 0.960
log P
vap
GW
(atm
)
log Pvapexp (atm)
(a)
β8
β6
β4
β2
0
β8 β6 β4 β2 0
R2= 0.961
log P
vap
LK(a
tm)
log Pvapexp (atm)
(b)
β8
β6
β4
β2
0
β8 β6 β4 β2 0
R2= 0.960
log P
vap
mM
(atm
)
log Pvapexp (atm)
(c)
β8
β6
β4
β2
0
β8 β6 β4 β2 0
R2= 0.957
log P
vap
MY
(atm
)
log Pvapexp (atm)
(d)
β8
β6
β4
β2
0
β8 β6 β4 β2 0
R2= 0.961
log P
vap
AW
(atm
)
log Pvapexp (atm)
(e)
Figure 7: Logarithm of estimated vapor pressure for a set of 32 difunctionalized species versus experimental values for the (a) GW, (b) LK,(c) mM, (d) MY, and (e) AWmethods. The black line is the 1 : 1 diagonal.
the predicted and experimental values are strongly correlatedwith a coefficient π 2 = 0.9638.
Vapor pressures for a set of 74 hydrocarbons are higherthan 10β3 atm. It is found for the five methods that estimatedvalues for this class of species are well correlated (Figure 5).
Furthermore, vapor pressures of tri- andmore functionalizedspecies are below 102 atm (Figure 8). For this class of species,LK method yields a weak correlation and has the largest pos-itive bias. Therefore, this method is the least reliable to esti-mate vapor pressure for tri- and more functionalized species.
International Journal of Geophysics 11
β8
β6
β4
β2
0
β8 β6 β4 β2 0
R2= 0.900
log P
vap
GW
(atm
)
log Pvapexp (atm)
(a)
β8
β6
β4
β2
0
β8 β6 β4 β2 0
R2= 0.898
log P
vap
LK(a
tm)
log Pvapexp (atm)
(b)
β8
β6
β4
β2
0
β8 β6 β4 β2 0
R2= 0.901
log P
vap
mM
(atm
)
log Pvapexp (atm)
(c)
β8
β6
β4
β2
0
β8 β6 β4 β2 0
R2= 0.923
log P
vap
MY
(atm
)
log Pvapexp (atm)
(d)
β8
β6
β4
β2
0
β8 β6 β4 β2 0
R2= 0.899
log P
vap
AW
(atm
)
log Pvapexp (atm)
(e)
Figure 8: Logarithm of estimated vapor pressure for a set of 28 tri- and more functionalized species versus experimental values for the (a)GW, (b) LK, (c) mM, (d) MY, and (e) AWmethods. The black line is the 1 : 1 diagonal.
4. Conclusion
Wehave evaluated in this study five vapor pressure estimationmethods useful for simulating the dynamics of atmosphericorganic aerosols. These are the Myrdal and Yalkovsky (MY),
the Lee and Kesler (LK), the Grain-Watson (GW), themodified Mackay (mM), and the Ambrose-Walton (AW)methods. Some of them are based on the Antoine equation,while others are based on the extended form of the Clausius-Clapeyron equation. But all of them take into account boiling
12 International Journal of Geophysics
temperature ππand (or) critical temperature π
π. Therefore,
Joback technique has been used to estimate ππ, while the
Lydersen technique was found to be better for ππestimation.
When using Joback to provide the ππvalues, LK, AW,
and MY are the best three methods for all classes of species.Moreover, for a set of 262 volatile organic compounds andas illustrated in the scatter plots and errors computed, theMY method which appears to be the best one fails fordifunctionalized species. For these latter species, the mMmethod provides good results, according to the correlationcoefficientπ 2 = 0.960 and the least errors reported in Table 2.GW method is the least reliable, which provides the lowestresults for all VOC and also for each class of species. Pre-dictions made with the AWmethod for monofunctionalizedspecies are more reliable than those made with the other fourmethods employing the Joback technique to provide the π
π.
For vapor pressure higher than 10β2 atm, all the five methodsgive similar results.
This work highlights that the choice of a method to pre-dict vapor pressure of volatile organic compounds dependson the number of functional groups existing in the species.
References
[1] J. Seinfeld and S. Pandis, Amospheric Chemestry and Physics:From a Pollution to Climate Change, John Wiley & Sons, NewYork, NY, USA, 1998.
[2] M. Tombette, ModeΜlisation des aeΜrosols et de leurs propriaΜlteΜsoptiques sur lβEurope et lβΜΔ±le de France: validation, sensibiliteΜ etassimilation des donneΜes [TheΜse], Ecole nationale des ponts etchausseΜes, 2007.
[3] P. Adams and J. Seinfeld, βPredicting global aerosol size distri-butions in general circulation models,β Journal of GeophysicalResearch, vol. 107, no. 19, pp. AAC 4-1βAAC 4-23, 2002.
[4] S. Gons, L. A. Barrie, J. P. Blanchet et al., βCanadian aerosolsmodule: a size-segregated simulation of atmospheric aerosolprocesses for climate to and air quality models. I Moduledevelopment,β Journal of Geophysical Research, vol. 108, no. 1,pp. AAC 3-1βAAC 3-16, 2003.
[5] E. R. Whitby and P. H. McMurry, βModal aerosol dynamicsmodeling,β Aerosol Science and Technology, vol. 27, no. 6, pp.673β688, 1997.
[6] E. Debry, ModeΜlisation et simulation de la dynamique desaeΜrosols atmospheΜriques [TheΜse], Ecole nationale des ponts etchausseΜes, 2004.
[7] B. Dahneke, Theory of Dispersed Multiphase Flow, Academicpress, New York, NY, USA, 1983.
[8] E. Debry, K. Fahey, K. Sartelet, B. Sportisse, and M. Tombette,βTechnical note: a new SIze REsolved Aerosol Model(SIREAM),β Atmospheric Chemistry and Physics, vol. 7, no.6, pp. 1537β1547, 2007.
[9] C. Tong, M. Blanco, W. A. Goddard III, and J. H. Seinfeld,βThermodynamic properties of multifunctional oxygenates inatmospheric aerosols from quantum mechanics and moleculardynamics: dicarboxylic acids,β Environmental Science and Tech-nology, vol. 38, no. 14, pp. 3941β3949, 2004.
[10] T. Banerjee, M. K. Singh, and A. Khanna, βPrediction ofbinary VLE for imidazolium based ionic liquid systems usingCOSMO-RS,β Industrial and Engineering Chemistry Research,vol. 45, no. 9, pp. 3207β3219, 2006.
[11] M. Diedenhofen, A. Klamt, K. Marsh, and A. SchaΜfer, βPredic-tion of the vapor pressure and vaporization enthalpy of 1-n-alkyl-3-methylimidazolium-bis-(trifluoromethanesulfonyl)amide ionic liquids,β Physical Chemistry Chemical Physics, vol.9, no. 33, pp. 4653β4656, 2007.
[12] P. B. Myrdal and S. H. Yalkowsky, βEstimating pure componentvapor pressures of complex organic molecules,β Industrial andEngineering Chemistry Research, vol. 36, no. 6, pp. 2494β2499,1997.
[13] R. J. Griffin, K. Nguyen, D. Dabdub, and J. H. Seinfeld, βA cou-pled hydrophobic-hydrophilic model for predicting secondaryorganic aerosol formation,β Journal of Atmospheric Chemistry,vol. 44, no. 2, pp. 171β190, 2003.
[14] B. K. Pun, R. J. Griffin, C. Seigneur, and J. H. Seinfeld, βSec-ondary organic aerosol 2. Thermodynamic model for gas/particle partitioning of molecular constituents,β Journal ofGeophysical Research D, vol. 107, no. 17, pp. AAC 4-1βAAC 4-15,2002.
[15] M. E. Jenkin, βModelling the formation and composition ofsecondary organic aerosol from πΌ- and π½-pinene ozonolysisusing MCM v3,β Atmospheric Chemistry and Physics, vol. 4, no.7, pp. 1741β1757, 2004.
[16] D. Mackay, A. Bobra, D. W. Chan, and W. Y. Shiu, βVapor pres-sure correlations for low-volatility environmental chemicals,βEnvironmental Science and Technology, vol. 16, no. 10, pp. 645β649, 1982.
[17] J. F. Pankow and W. E. Asher, βSIMPOL.1: a simple group con-tribution method for predicting vapor pressures and enthalpiesof vaporization of multifunctional organic compounds,β Atmo-spheric Chemistry and Physics, vol. 8, no. 10, pp. 2773β2796,2008.
[18] M. Capouet and J.-F. MuΜller, βA group contribution methodfor estimating the vapour pressures of πΌ-pinene oxidationproducts,β Atmospheric Chemistry and Physics, vol. 6, no. 6, pp.1455β1467, 2006.
[19] M. Camredon and B. Aumont, βAssessment of vapor pressureestimation methods for secondary organic aerosol modeling,βAtmospheric Environment, vol. 40, no. 12, pp. 2105β2116, 2006.
[20] S. Compernolle, K. Ceulemans, and J.-F. MuΜller, βTechnicalNote: vapor pressure estimation methods applied to secondaryorganic aerosol constituents from πΌ-pinene oxidation: an inter-comparison study,β Atmospheric Chemistry and Physics, vol. 10,no. 13, pp. 6271β6282, 2010.
[21] M. H. Barley and G. McFiggans, βThe critical assessmentof vapour pressure estimation methods for use in modellingthe formation of atmospheric organic aerosol,β AtmosphericChemistry and Physics, vol. 10, no. 2, pp. 749β767, 2010.
[22] L. E. Yu, M. L. Shulman, R. Kopperud, and L. M. Hilde-mann, βCharacterization of organic compounds collected dur-ing Southeastern Aerosol and Visibility Study: water-solubleorganic species,β Environmental Science and Technology, vol. 39,no. 3, pp. 707β715, 2005.
[23] M. Jang and R. M. Kamens, βCharacterization of secondaryaerosol from the photooxidation of toluene in the presence ofNOx and 1-propene,β Environmental Science and Technology,vol. 35, no. 18, pp. 3626β3639, 2001.
[24] W. E. Asher, J. F. Pankow, G. B. Erdakos, and J. H. Seinfeld,βEstimating the vapor pressures of multi-functional oxygen-containing organic compounds using group contributionmeth-ods,β Atmospheric Environment, vol. 36, no. 9, pp. 1483β1498,2002.
International Journal of Geophysics 13
[25] D. R. Lide, Handbook of Chemistry and Physics, 86th edition,1997.
[26] C. L. Yams, Hanbook of Vapour Pressure, Vols. 2 and 3, GuefPublishing Compagny, Houston, Tex, USA, 1994.
[27] T. Boulik, V. Freid, and E. Hala, The Vapour Pressure of PureSubstances, Elsevier, Amsterdam, The Netherlands, 1984.
[28] K. J. Joback and R. C. Reid, βEstimation of pure componentproperties from group contribution,β Chemical EngineeringCommunications, vol. 57, pp. 233β243, 1987.
[29] P.W. Bruckmann and H.Willner, βInfrared spectroscopic studyof peroxyacetyl nitrate (PAN) and its decomposition products,βEnvironmental Science andTechnology, vol. 17, no. 6, pp. 352β357,1983.
[30] Y. Nannoolal, Development and critical evaluation of groupcontribution methods for the estimation of critical properties,liquid vapour pressure and liquid viscosity of organic compounds[Ph.D. thesis], University of Kwazulu Natal, 2006.
[31] P. B. Myrdal, J. F. Krzyzaniak, and S. H. Yalkowsky, βModifiedTroutonβs rule for predicting the entropy of boiling,β Industrialand Engineering Chemistry Research, vol. 35, no. 5, pp. 1788β1792, 1996.
[32] J. Vidal, Thermodynamique: Application au GeΜnie Chimique etAΜa lβindustrie du PeΜtrolieΜre, Technip, Paris, Farnce, 1997.
[33] E. J. Baum, Chemical Property Estimation-Theory and Applica-tion, section 6.3, Lewis, Boca Raton, Fla, USA, 1998.
[34] R. P. Schwarzenbasch, P. M. Gschwend, and D. M. Imboden,Environmental Organic Chemistry, John Wiley & Sons, NewYork, NY, USA, 2nd edition, 2003.
[35] W. J. Lyman, Environmental Exposure from Chemicals, vol. 1,chapter 2, CRC Press, Boca Raton, Fla, USA, 1985.
[36] S. H. Fishtine, βReliable latent heats of vaporization,β Environ-mental Science & Technology, vol. 55, pp. 47β56, 1963.
[37] C. F. Grain, βVapor pressure,β inHandbook of Chemical PropertyEstimation Methods, chapter 14, McGraw-Hill, New York, NY,USA, 1982.
[38] B. E. Poling, J. M. Prausnitz, and J. P. OβConnell,The Propertiesof Gases and Iquids, McGraw-Hill, New York, NY, USA, 5thedition, 2001.
[39] R. C. Reid, J. M. Prausnitz, and B. E. Poling, The Propertiesof Gases and Liquids, McGraw-Hill, New York, NY, USA, 4thedition, 1986.
[40] D. Ambrose and J. Walton, βVapor pressures up to their criticaltemperatures of normal alkanes and 1-alkohols,β Pure andApplied Chemistry, vol. 61, pp. 1395β1403, 1989.
Submit your manuscripts athttp://www.hindawi.com
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
ClimatologyJournal of
EcologyInternational Journal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
EarthquakesJournal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Hindawi Publishing Corporationhttp://www.hindawi.com
Applied &EnvironmentalSoil Science
Volume 2014
Mining
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Journal of
Hindawi Publishing Corporation http://www.hindawi.com Volume 2014
International Journal of
Geophysics
OceanographyInternational Journal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Journal of Computational Environmental SciencesHindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Journal ofPetroleum Engineering
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
GeochemistryHindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Journal of
Atmospheric SciencesInternational Journal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
OceanographyHindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Advances in
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
MineralogyInternational Journal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
MeteorologyAdvances in
The Scientific World JournalHindawi Publishing Corporation http://www.hindawi.com Volume 2014
Paleontology JournalHindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
ScientificaHindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Geological ResearchJournal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Geology Advances in