Você está na página 1de 13

A ~ p k e r ~ F~i:,enrasntVol. 20, No. 5, pp.

1025-1037, 1986

(3004--.69111/116S3.00 -t- 0.00


C~ 19116~r~,mon pre*s Ltd.

P~nted in Cheat! llritain.

EFFECT OF COMPOSITION ON AEROSOL LIGHT


SCATTERING EFFICIENCIES
CHRISTINE S. SLOANE
Environmental Science Department, G. M. Research Laboratories, Warren, MI 48090-9055, U.S.A.
(First received 1 April 1985 and in final form 30 August 1985)

Abstrsct--Forecasts of the impact of emissions changes on visibility use light scattering efficienc~es---the
change in the amount of light scattered with a change in mass of an aerosol constituent. This paper
demonstrates how fight scattering ef~ciencies depend on interactions between the aerosol constituents.
Calculations are presented for two model aerosols: NH,HSO4 droplets and a 'typical' urban aerosol. These
calculations demonstrate that the traditional means of predicting visibility impairment for increases in
atmospheric emissions is not generally appropriate. Different light scattering ef~ciencies are needed for
distinct meteorological conditions and aerosol compositions thus for typical high and low visibility events.
Key word index: Aerosols, light scattering, atmospheric optics, sulfate, visibility.

!. I N T R O D U C T I O N

Forecasts of the impact of emissions changes on


visibility are based on a presumed relationship between visual air quality and pollutant emissions. That
relationship has two parts: (1) the relationship between
emission source strengths and particulate composition
and loading, and (2) the amount of light scattered (and
absorbed) by particles, characterized by chemical composition, mass loading and size distribution. This work
focuses on the second issue, the relationship between
visual air quality and airborne particles.
Visual air quality can be evaluated in terms of visual
range. For a homogeneous sight path, visual range is
inversely related to Bext, the optical extinction coefficient, by: Visual Range = 3.9/Bext (Middleton, 1952).
In the absence of substantial multiple scattering, the
extinction can be separated into scattering and absorption attributed to gases and airborne particles (lkxt
= Bsp + Bap + Bsg + Bag; s and a refer to scattering
and absorption; g and p, to gases and particles,
respectively). During the past decade, numerous
studies have shown that, for the most part, the light
scattering attributed to fine aerosol particles (aerodynamic diameter < 2.0/Jm) usually dominates the light
extinction. Hence, the particulate fight scattering coefficient, Bsp, is the focus of this work.
Light scattering efficiencies ;)rovide the connection
between the amount of light scattered by airborne
aerosols and their airborne mass concentration. They
are of key importance in forecasting visibility changes
and in establishing extinction budgets.
The light scattering efficiency of an aerosol is
commonly defined as the change in Bsp per unit
aerosol mass: ABsp/AM (M = mass). It is often taken
as the ratio Bsp/FPM, which represents the change in
the light scattering (ABsp = B s p - 0 ) which would
result from the complete removal of the fine particle

mass (AM = FPM - 0). A light scattering efficiency of


3-4 m 2 g- t has generally been observed for dry airborne fine particles.
By analogy, the light scattering efficiency per chemical species is the change in Bsp caused by a change in
the mass of one chemical constituent of the airborne
particles. Commonly, Bsp is assumed to be proportional to the mass concentrations of the aerosol
constituents. In that case, the fight scattering efficiencies are the proportionality constants which are
usually estimated by multivariate regression analysis
of field measurements. The equation:
Bsp = ~'~a i m i

(1)

is the usual starting point for statistical fitting procedures. Here, M i represents the fine particle mass (per
volume of air) of chemical species i. aj is the fitting
constant which is usually interpreted as the light
scattering efficiency of chemical species i. Indeed, if the
a i were true constants then a change in the mass of
species i, AM i, would produce a Change in Bsp given by
ABsp ffi aiAM i, so that a i ffi ABsp/AM i would correspond to a unique light scattering efficiency.
Recently, White (1986) has noted that regression
estimates of fight scattering efficiencies could be confounded by interactions between aerosol constituents.
The calculations to be presented in this paper are
intended to provide a realistic evaluation of the
sensitivity of light scattering efficienciu to interactions
between aerosol chemical coMtituents. The light scattered by model aerosols is calculated exactly. The light
scattering eff|ciencies of the individual chemical species
are evaluated by calculating Bsp before and after a
change in the mass of one species. The difference
between these calculations, ABsp, is divided by the
mass change, AMt, to give the light acattering effic-

1025

1026

CHIUb~HNES. SLOANE

iency. The calculations have been performed for two


aerosols, NH4HSO4 droplets and a 'typical' urban
aerosol. In this paper, Sections 2 and 3 develop
background information; the results are presented and
discussed in Sections 4 and 5.

x (vJWi)Mi(D)/M i dD]

(3)

i = chemical species (including H=O).


2. BACKGROUND: VISIBILITY FORECASTS AND

EXTINCTION BUDGETS

2.1. Theoretical basis for empirical light scattering


eticiencies
The theoretical, or physical, basis for Equation (1) is
found in the optical equation for the fight scattering
coefficient. That equation is given by

Here vi is the partial molar volume of chemical species i


and Wi is its molecular weight. The partial molar
volume of some species[such as (NH4)zS04 dissolved
in an aqueous particlecoating] will depend on the
concentration of all dissolved species. More conveniently,Bsp can be written in terms of the dry mass
size distributionas:

Bsp = ~ M,[ f (3Q(m,~,D)/2D)f Y/ Yo)

x Mi(D)/Mip ~ d D ] --- ~ M,[a,]

Bsp = J o (m, A, D) n (D) dD

f [3q

i H20

D)/2pa M(D)dD

= f[3Q(m,~,D)/2DPo'I(y/Yo)Mo(D)dD

(2)

for the illustrative case of fight scattered by homogencous, spherical particles of similar composition.
Here, (m,,t, D) is the scattering cross-section, ,l is the
wavelength of the fight, m is the index of refraction, and
D is the particle diameter. Q(m, lt, D) equals
4(m,~,D)/~D 2, the ratio of the scattering crosssection to the geometrical cross-section; that ratio is
the optical scattering efficiency (Kerker, 1969).
(3Q/2D) is the mattering cross-section per unit volume
of particles with diameter D. n(D)dD is the particle
number density, which is related to the mass size
distribution by n(D)dD = (6/p(D)xD3)M(D)dD.
M(D)dD is the mass of aerosol particles (per unit
volume of air) with diameters in the size interval dD. p
is the ambient aerosol density.
The total aerosol mass (including water) is generally
not measured. Therefore, Bsp is more usefully expressed in terms of the size distributions of the 'dry'
species (i.e. species other than water): n(D)dD
-, (6/po (D) nD~)Mo (D)dD. Here the subscript 'o' signifies 'dry' particulate mass and Vo (= xD~/6) refers to
the volume an ambient particle of diameter D would
have without a water component.
Equation (2) is easily generalized to describe aerosols in which all particlfs are not homogeneous or of
similar composition (Sloane, 1984). For example, if
aerosol particles of like size were of like chemical
composition, but particles differing in size possibly had
different chemical compositions, then Equation (2)
would be changed by making the density, refractive
index and growth factor (V/Vo) into functions of D,
the particle diameter.
An analogous expression can be obtained for the
mass size distribution differentiated by chemical
SpeCies"

(4)

where pa ~- Wi/vi in the absence of water. Equations


(3) and (4) are equivalent; in each, 3Q/2D, the scattering
per unit aerosol volume, is multiplied by the ambient
volume, which in Equation (4) is Y/Vo times the dry
particle volume.
The 'traditional' theoretical expression for the light
scattering efficiency,aj, evolved from its definition as
the change in Bsp upon the removal of speciesj. Taken
from Equation (3), it has been expressed (Ouimette
and Flagan, 1982) as:

,_-

f[3 (m.

(,;

M,(D)/Mj dD

+ ~j ~3/2D(v3wl)UQ(m,~-D)
-Q(m',A,D)] Mi(D)/Mj dD
where m' is the refractive index evaluated without
species j. In the likely event that the refractive index of
species j is not appreciably different from that of the
other major species, then the fight scattering efficiency
for species j is simply:
dD

oj-and at low relative h u ~ d i t y (where


I'/Vo ~- 1),

Poj "" Wj/vj and

fE30.(.o)/2DpMj(D)/Mj
] dD.

(5)

This is the standard theoretical formulation which will


be referred to as 'traditional.' Figures 1A and IB
illustrate the 'traditional' response of Bsp to a change
in constituent loading. Shown is the overall mass size
distribution beCore and after the removal of one
chemical constituent from a hypothetical aerosol assuming no response in the mass size distributions,
Mi(D)dD, of the remaining species. Shown also is the

Effect of composition on fight scattering efficiencies


function 3Q/2D. Bsp is largely dependent on the
overlap of 3Q/2D with the size distribution. The
change in Bsp predicted 'traditionally' is the result of
the illustrated change in the mass size distribution.

1027

+ i#jy f[3Q(m,W)/2Op3[(v/Vo)M(O)

HzO

- (V/V'o)M~(O)]/Mj dO

= aj ('traditional')

2.2. Traditional extinction badgers and visibility


forecasts
The 'traditional' and statistical approaches are consistent in that in both, the incremental change in lisp
from a change in the mass of one aerosol constituent
(species j) is simply ajAMj. Therefore, after the
removal of species j, Bsp = ~ aim i (i ~ j); the physical
i

and chemical properties of the remaining constituents


are assumed to be unaltered. Hence, the incremental
change in Bsp from the removal of any one aerosol
constituent, a iMj, is independent of the order in which
that constituent is removed (i.e. before or after the
removal of other constituents). The incremental changes in Bsp predicted for the sole removal of each
constituent from the aerosol therefore satisfy a simple
additive constraint: ~ aiM i = Bsp.
i

From Equation (5), the 'traditional' light scattering


efficiency of species j is seen to be independent of the
remaining aerosol constituents if the refractive index
does not provide a significant interaction. Therefore,
the 'traditional' formulation supports a separable light
scattering budget in which each chemical species is
'responsible' for a specific fraction of the average light
scattering. For example, 40 % of the average light
scattering at a measurement site could be attributed to
sulfates if ai<Mi)/<Bsp)=0.40 for i = s u l f a t e
species, a i being the statistically fit coefficient of sulfate
mass in Equation (1).

2.3. Interactive light scattering e~ciency


The 'traditional' light scattering efficiency has a
major shortcoming; it disregards changes in the size
distributions of the remaining species caused by the
removal of one constituent. It describes a static aerosol
in which responsive inter-reactions between the aerosol constituents are disregarded. The light scattering
efficiency for the removal of species j is more completely given by:

aj = f[3Q (m, 4, D)/2D P0j] (V/Vo) Mj (D)/Mj dD


+

i)
4. H 2 0

- (V/V'o)Q(m',A,D)M~(D)]/Mj dD

f[3Q

(,,,, ~, D)/2Dpoj] ( F / r o)Mj (D)/Mj OD

(6)

+ ~ f[3Q(m,~,o)/20p~}
ij

H~O

[(F/Fo)ni(D ) - (V/V'o)M~(D)]/Mj dD.


Equation (6) refers to the light scattering e~ciency of
species j for its total removal. In this case, M'(D)
represents the mass size distribution of the remaining
aerosol species i after the mass change (in this case,
removal) of species j. For an increase or smaller (than
total) decrease in the mass of species j, the light
scattering efficiency would be given by:

aj = ~,o f(z/2op~)[(V/Vo)Q(m,).,D
) Mi(D )
- (V/V'o)Q(m',2,D)M~(D)]/Mj dO.
Interactions between the aerosol constituents contribute to the inaccuracy of the 'traditional' prediction.
The calculations presented in this paper will illustrate
the effect of two such interactions. A simple example of
the first is the decrease in particle volume after the
removal of one species for the case where the number
of aerosol particles and the ma_~sesof the other species
are not affected. In that case, the mass size distributions
of the remaining constituents shift to smaller sizes as
illustrated in Figure 1C. The overlap of the final mass
size distributions with 3Q/2D is smaller. The size
distribution could also be distorted if the species
eliminated had not been evenly apportioned within the
overall mass size distribution. This simple effect can
most easily be visualized for internally mixed (i.e.
multicomponent) aerosol particles, but it would also
occur in externally mixed aerosols (i.e. mixtures of
chemically pure particles) if the species mass change
were less than total and the size (as opposed to the
number) of chemically pure particles was affected.
A second type of interaction occurs when the
concentration of one constituent influences the concentration of another in the aerosol. An example is
presented by the affinity of water for water-soluble
salts. A change in the mass of a salt constituent could
affect the aerosol water content. Therefore, if the
formation processes for the other species were not
affected, the mass size distribution of those remaining
constituents would shift in response to the loss of the
salt species and associated water from each particle.
Some distortion of the size distributions would result if
the salt were not evenly apportioned within the overall
mass size distribution. The average index of refraction
of the aerosol could also shift noticeably.
This last type of interaction between aerosol constituents will be referred to as i/chemical interaction in
contrast to the more physical interaction described

CHILSTINE ~. SLOANE

1028

,.,~..,..,.--......

~.

3 Q(m,A,D) I 2D

AttedLognorma, Mass Size


Dh t rlbut/on

Distl il)ullon

~,J'~;?:~111!) ii:;?71~?~i?:

,,,

.2

.4

.6

.8

1.2

Diameter (},m)

3Q(m,k,D) / 2D

fitted Lognor~l Mass Size


Distribution

DIstrlbutl0n

il . . . . . . . . .
I

.2

.I

.4
.6
Diameter ~m}

.8

,1.2

3 Q(m,k,D) /2D

Remainin9 Mass Size

/_

.1

.2

.4

.6

.8

L2

Diameter (~mb

rlg.

Compon ,,s or light

~,D)/2DPi]

Mi(D)dD , 30(m,,~D)/2D,=scsttering per unit volume o f aerosol (m,=1.5, ,~


=, 0.55/nn), Pi " density of species i, Mi(D)dD = ~
size distribution of aerosol
constituent i (i ,= I, 2) plotted as normalized mass per log diameter inter~l, species I
,,,s/utded arm, species 2 -- clear area. (A) Mass size distribution including species ] and
2. (B)"rr~ditional' mass size distribution after removal of spies 2. (C) In tcractivc
mass size distribution after removal of species 2.

Effect of composition on light scattering efliciencies


first. In both, a change in the mass of one aerosol
constituent affects the size distribution of each remaining species.

3. DESCRIPTION

OF

SEMI-EMPIRICAL

PHYSICAL

MODEL

Bsp will be evaluated for a multicomponent aerosol


using a semi-empirical model which provides for the
aerosol water content. This physical model was developed (Sloane, 1984) to predict light extinction from
ambient measurements of aerosol composition, size
distribution and loading in both humid and dry urban
settings. It has been tested successfully against field
measurements made in Denver and Detroit (Sloane,
1983, 1984; Sloane and Wolff, 1985), and therefore
appears to realistically provide for the chemical and
physical processes most important in describing light
scattering. Since a full description of the model has
been given previously (Sloane, 1984), only its salient
features will be outlined here. Readers not concerned
with details of the model formulation might proceed to
Section 4.
3.1. Water absorption
From a semi-empirical application of the thermodynamic properties of solution droplets, the particle
growth factor, D/Do (Do = dried aerosol diameter, D
= ambient aerosol diameter), is related to the relative
humidity by:
rh
V / V o = (D/Do) ~ = 1 + - l-rh

<i) <t>
- M,Ep o.
(M,)

(7)
Here, rh - relative humidity/100, E = water soluble
fraction of dry aerosol mass, Mw = molecular weight
of water, and P0 - dry aerosol density. The ambient
aerosol diameter, D, is taken as the Stokes diameter,
which is related to the aerodynamic diameter through
Dstok. = D,erody,~ ~/(S!/$2 p) where the Ss are the
usual slip correction factors (Friedlander, 1977) and p
is the ambient aerosol density. The Kelvin Effect is not
included in Equation (7) becauseit is not important for
particles in the size range which dominates visible light
scattering. The remaining parameters form the composite function, < i ) ( ~ ) / ( M , ) .
They are derived
from the thermodynamic relationship: rh = water activity [1 +in(s)/n(w)]-' where: n(s) = moles of dissolved solute, n(w) = moles of water and i = van't
Hoff factor, n(s) equals E~ Mo/W, with ~ = dissolved
fraction of soluble aerosol mass Mo is the dry aerosol
mass and W, is the average molecular weight of
dissolved aerosol constituents, n(w) equals (V
- Vo)p(w)/qMw; p(w) is the density of water, q is the
ratio of the partial molar volume of water in the
aerosol particle to the molar volume of water, q = 1 if
volume is conserved when water is absorbed by an
aerosol l~rticle. M , - , W,]q, so M, is sensitive to
deviations from the conservation of volume upon

1029

dissolution. The composite function ( i ) ( ~ ) / ( M , )


is a function of relative humidity.
It has been observed (Sloane, 1983,1984; Sloane and
Wolff, 1985) that the 4- and 8-h average ambient light
scattering coefficients, Bsp, can be predicted well for a
4-8-week field study if a single function for i t / M , is
used for a site. That function (of relative humidity) is
designated ( i ) ( ~ ) / ( M, ). It is treated as 'typical' of
the average aerosol at the site. The calculations
presented here use this approximation: deviations of
i~/M, from ( i ) ( ~ ) / < M , ) are neglected. This does
not mean that the amount of water absorbed by the
ambient aerosol is always the same at a given relative
humidity, quite the contrary. As Equation (7) shows,
the swelling due to water absorption varies with
changes in the water soluble mass fraction, E, and the
dry density, Po. The amount of water absorbed is
determined separately for each size interval for which
mass measurements, Mi(D)dD, are recorded.
The composite function < i ) < r . ) / < M , ) has not
been evaluated rigorously for chemically complex
ambient aerosols. It has been measured for some
chemically pure aerosols and for some binary salt
mixtures. It can be estimated empirically from measurements of the water content of ambient aerosols with
varying composition (Hanel, 1976; Hanel and
Lehmann, 1981). That procedure has proved useful in
interpreting ambient light scattering measurements
(Sloane, 1984). The relative humidity dependence of
this composite function is shown in Table 1 for several
chemically pure aerosols, binary mixtures and aerosols
of typical urban composition. Shown are values appropriate to the hysteresis pathway which includes supersaturated droplets (Sloane, 1984).
3.2. Size distribution
3.2.a. Reference distribution for model aerosol.
Lognormal mass size distributions used for the model
calculations of this work were taken from measurements made of urban aerosols in Detroit (Sloane and
Wolff, 1985) which are typical of many urban aerosol
measurements. The values assigned to the geometric
mean aerodynamic diameter, Ds(gm), and log standard geometric deviation, In ~s, for each of the principal chemical species are: sulfate (D s = 0.49, in o s
= 0.73), organic carbon (0.29, 0.84) and elemental
carbon (0.31, 0.73) at 59% relative humidity. The
aerosol dry mass, FPM, is taken to be 30.6 #gin-3
apportioned as 12.7/~g m - 3 to sulfate compounds, and
11.8/Jgm-3 total carbonaceous material (comprised
of 9.5 #g m - 3 organic material and 2.3 #g m - 3 elemental carbon). Residual material (i.e. constituents of
the urban aerosols other than these principal species)
comprises 6.1/zgm -3 of which 80% is assigned to the
largest size interval (the coarse mode tail) and 20 % is
apportioned as (0.49, 0.73).
The soluble aerosol mass is assumed to consist of the
total sulfate mass, the nitrate mass, 60 % of the organic
mass (hence, 48 % of total carbonaceous material) and
10 % of the residue mass (Sloane and Wolff, 1985), The

1030

CHRmTn~ S. StOANS

Table 1. Thermodynamk functionfor l~rticle growth: (1) ( e ) M w / ( M .)


Relative humidity (~/o)
50
60
70

A~rosol type

30

40

H2SO,
NH,HSO4
(NH4)aSO4 '
NH,NO~
Binary salts*:
(0.5 (NH,),SO, and
0.5 NH,NO3)t
(0.5 (NH4hSO4 and
0.5 H=SO,)
Mixed aerosols :
Sea salt (E = 0.89)1"

1.60

1.40
0.61
, 1.10
0.29

1.20
0.48
0.71
0.23

1.00
0.45
0.53
0.37

~:

:~

0.61

0.04

80

90

0.80
0.46
0.39
0.31

0.66
0.40
0.32
0.32

0.52
0.35
0.27
0.33

0.32

0.33

0.35

0.48

0.45

0.46

0.40

0.35

0.06

0.09

0.13

0.83

0.71

0.57

0.24

0.24

0.30

0.54

0.67

0.58

0.46

0.18

0.18.

0.20

0.34

0.31

0.29

0.24

(0.9 (NH,,)=SO,, and


0.1 H2SO,; E = 0.63)

0.25

0.48

0.62

0.65

0.69

0.56

0.37

'Typical' urban aerosol


(0.4 NH4HSO4 and
0.4 carbonaceous and
0.2 residual; E = 0.63)

0.20

0.40

0.60

0.70

0.75

0.65

0.60

Continental urban aerosoist


(0.5 NH4NO3 and
0.1 (NH,)2SO4
+ 0.4 sea salt; E = 0.50)
(0.6 (NH,)2SO, and
0.2 NH4NO;
0.2 H2SO,; E = 0.40)

*Adapted from Tang (1980), and Tang and Munkelwitz (1977).


tAssuming partial molar volume of water = molar volume of water (0 = I).
:[:Measurements not available.
Adapted from Hanel and lehman (1981).

consequent average soluble ma_ss fraction, E, and the


average dry density, Po, for each size interval used in
the calculation is shown in Table 2. The growth
factors, D/Do, predicted for this 'typical' aerosol and
for pure NH4HSO4 are shown in Table 3. From the
growth factors, the refractive indices and aerosol
densities are evaluated as a function of particle size and
relative humidity. The densities are used to compute
Stokes particle diameters for the optical computations.
3.2.b. Response of size distribution to mass change.
The response of the mass size distributions to a change
in the mass of one or more chemical constituents of the
aerosol was discussed in Section 2.3. The simple physical interaction cited would produce a change in panicle
diameter from D = (D/Do)Do to:

D'

(D/Do)XDo

where
L

J/L

.I

The response of the water content to a change in mass


of a soluble constituent is provided by (D/Do) which
responds (in each size range) to a change in the soluble
mass fraction, E, and in the dry particle density, Po, as
prescribed by Equation (7). In these calculations, the

mass of a chemical species will be altered proportionally throughout that species' reference mass size distribution, so that in all size ranges AMi(D)/Mi(D)

= AMilM i.
3.3. lnhomooeneity
In this model, the ambient aerosol is taken to have
an interior core of the water-insoluble material and an
exterior water-soluble layer. The exterior layer is
presumed to take up/release water with changes in the
relative humidity. The refractive index and density of
each layer are represented by volume averages of the
refractive indices and densities, respectively, of the
chemical constituents of the layer (SIoane, 1983). They
are re-evaluated to be consistent with aerosol composition for calculations involving changes in the
masses of individual species. Values of the constituent
refractive indices and densities used in these calculations are given in Table 4.
The chemical composition of all particles in each size
interval is proportioned according to the mass size
distribution for each chemical constituent as given in
Section 3.2. Thus, the composition and refractive index
of the core layer vary with particle diameter. The
composition and refractive index of the exterior layer

Effect of c0mp~ition on lif,ht scattering efli~ien~ies

1031

'

~ .

g.
e~

~" --"

..=
m
m

" ~
o

.=
~s

e4

.G

.o

"~'--

e~
t~s

o
.q

o
. ~

~k

,- ~,o..

~ ,,.~

1032

CnmsrmE S. St.oxw:

Table 4. Physical properties of chemical constituents


Chemical spzcies

NH,,HSO,
Organic carbon
Elemental carbon
Residue

Index of refraction

Density

1.47"
1.55t
1.90-0.60i~
,1.53-0.005i H

1.78"
1.401"
2.00
2.3~

"Handbook of Chemistryand Physics.


tWithin range of many organic compounds.
:[:Lee and Tien (1981).
Reessler and Faxvog (1980).
IIPatterson (1977).
Hanel (1976).

depend, in addition, on the water content, and hence,


the relative humidity. The relative thickness of the core
and shell layers also varies with particle diameter and
relative humidity. The scattering cross section for each
size interval is evaluated using an extension of Mie
theory for concentric spheres (SIoane, 1984) and
0.48 ~m wavelength light.
4. ItESULTS

The light scattering efficiencies of individual chemical species have been evaluated by taking the difference
between Bsp calculated for a reference aerosol composition (Section 3.2.a) and Bsp calculated for an
aerosol which differs from the reference aerosol by the
removal (or addition) of mass for one or more chemical
constituents (Section 3.2.b). That difference in Bsp is
divided by the mass change.
Results of the calculations based on the 'typical'
urban aerosol model are presented in Table 5. Section
A ofTable 5 shows the calculated scattering efficiencies
for the sulfate, carbonaceous and residual aerosol
constituents given as functions of relative humidity
and percentage change in mass. For example, the light
scattering efficiency (change in light scattering per unit
mass change) for a 50 ~ increase in mass of sulfate
compound at 60 ~ relative humidity is 8.7 m a g- ~ for
this aerosol. The sulfate compound represented by the
dens/ty and index of refraction in Table4 is
NH4HSO,,, the sulfate species likely to have dominated the Detroit aerosol (Sloane and Wolff, 1985).
For comparison, the calculations were repeated for
the hypothetical case where the three constituents
(referred to as $1, $2 and $3 in Table 5B) differed only
in affinity for water and mass fraction. All three
constituents were assigned the same size distribution
(D e -- 0.49, In# s = 0.73), density (1.75 gec- 1) and refractive index (1.50). The solubilities of the three
constituents were taken to be 100, 50 and 0 ~ ,
respectively. And, as in the 'typical' urban aerosol the
average aerosol mass loading was apportioned as S1,
12.7 Sa, 11.8 and $3,6.1/~g m - 3 for a total fine particle
mass (FPM) of 30.6 pgm -3. The results of these
calculations are given in Table 5B. Values shown in
parentheses correspond to calculations in which the

size distribution of S2 was shifted to (D s -- 0.30, in a,


,= 0.80), to emulate the size distribution of total
carbonaceous material (TC) in the modelled 'typical'
urban aerosol. Another hypothetical case in which all
three constituents have the same chemical affinity for
water (but still differ in average mass fraction) is also
presented. These results suggest that a chemical interaction (affinity for water) could have a greater impact
on light scattering efficiency than a constituent's size
distribution.
The sensitivity oftbe light scattering efficiency to the
direction and magnitude of a constituent mass change
is demonstrated in Tables 5C and 6. Clearly, these
nonlinearities are small. The scattering efficiency is
much more sensitive to relative humidity than to the
size of mass change.
The calculated light scattering efficiencies for
NH,HSO4 droplets are given in Table 6. There the
thermodynamic growth of NH4HSO4 droplets is
simulated as specified in Table 1 for 'pure species'
NH4HSO4. The mass of residue, R, and carbonaceous
material TC, are set to zero. These Bsp calculations
and the resulting light scattering efficiencies agree with
experimental measurements on NH4HSO4 droplets
(Tang, 1980;, Charlson et aL, 1978).
Figure 2 shows the light scattering efficiency of the
NH,HSO4 given as a function of relative humidity.
The solid curve represents a 5 0 ~ decrease in
NH4HSO4 mass with the related loss of water. The
mass reduction is associated with a decrease in particle
size as described in Section 2.3; the number of aerosol
particles is unchanged.
For comparison, the dashed curve in Figure 2 gives
the light scattering efficiency for the same reduction in
NH4HSO4 mass and associated water, but the number
of aerosol particles is reduced to accommodate the
mass change without shifting the mass distribution to
smaller sizes. Hence, the simple physical effect described in Section 2.3 is disregarded. Since the dry
aerosol is composed of only one species, V/Vo is not
affected by the NH4HSO, mass loss. As a result, the
lower curve in Fig 2 gives the 'traditional' NH4HSO4
light scattering efficiency for an external mixture.
The light scattering efficiency for a 50 ~ decrease in
sulfate mass in the modelled urban aerosol is plotted as
a function of relative humidity in Fig. 3. Also shown is
the light scattering efficiency evaluated without including the simple physical interaction (Section 2.3) between the dry aerosol species' size distributions. The
latter calculation differs slightly from the 'traditional'
result since the growth factor, V/Vo, can be affected by
the NH4HSO4 reduction.

5. DISCUSSION

These calculations show that interactions between


aerosol species can make important contributions to
the light scattering efficiencies. For the simple illustrative cases modelled here, the chemical interaction

1033

Effect of composition on light scattering effieiencies


Table 5. 'Typical' urban aerosol
A. Light scattering efficiencies by chemical species* (m2 g-1). AM i ., X ~o Mi
Re,five
humidity

Mass
chan~

Aerosol con~ituent
S
TC
R
FPM

+ 50 ~
-50%

3.7
3.7
3.9
3.9
4.8
4.7
6.4
6.3
8.7
8.6
11.5
11.9
14.6
15.8

30

50

70
80

3.5
3.3
3.5
3.4
3.9
3.7
4.5
4.5
5.6
5.4
6.8
6.6
8.6
8.5

1.7
1.0
1.6
I.I
1.5
1.1
1.8
0.9
2.6
1.2
2.1
1.9
2.7
2.3

3.2
3~
3.3
3.1
3.8
3.6
4.9
4.6
6.2
5.9
7.9
7.9
9.9
1~4

~ aiMi/Bsp
i

1.1
1.1
1.1
1.1
1.1
1.1
1.1

B. Light scattering efficiencies for constituents differing in water affinity (m2 g-i).
AM~ = + 50% M i
Relative
humidity

Aerosol constituent**
St
S2
$3

FPM

30

4.4
(4.4)
4.4

4.2
(3.5)
4.4

4.0
(4.0)
4.5

4.3
(4.1)
4.5

50

6.8
(6.9)
6.8

5.4
(4.7)
6.9

4.2
(4.2)
6.9

5.8
(5.6)
6.9

70

12.4
(12.3)
12.3

8.5
(7.4)
12.2

4.3
{4.2)
12.1

9.1
(8.7)
12.1

Solubilities (~/o)
Sl 100
St 100
S t 100

$2 50
S~ 50
Sz 100

$3 0
$3 0
$3 100

C. Nonlinearities in light scattering efficiencies ABscatt/AM. NH4HSO4 component


of modelled urban aerosol (m 2 g-t). AM ~ X ~ M
Relative
humidity

+ 2 0 ~o

-20~

30
50
70

3.9
6.2
11.6

4,0
6.4
11.5

Percentage mass change


+50~/o -50~/o
+99~
3.9
6.4
11.5

3.9
6.3
11.9

3.9
6.5
11.5

-99~
3.8
6.1
I 1.8

* + 50 % mass change from: S (sulfate compound: 1.2 SOt mass) of 12.72 pg m - 3 TC


(carbonaceous material; 1.5 volatile C + nonvolatile C) of I 1.75/Agm- 3 R (residual fine
mass; FPM-S-TC) of 6.13pgm -3 FPM (fine particle __n~s)of 30.60/agm -3.
Wavelength 0.48
*0 All constituents have aerodynamic size distribution (Dt I
0.49 ~, In er8 ,- 0.73) at
5 9 ~ relative humidity. Po ffi I and refractive index ,~ 1.50.'Values in parentheses are
for S, (Ds ,- 0.30 p, lno s = 0.80) at 59~o relative humidity. 5 0 ~ mass change from
12.72 pgm -3 St, ! 1.75 ~gm -3 S: and 6.13 p.gm -3 S 3.
represented by the affinity of hygroscopic species for
water can lead to two- and three-fold changes in the
light scattering efficiency of the hygroscopic species.
On the other hand, the physical interaction is seen
from Figs2 and 3 to contribute 8-13~o to the light
scattering efficiency. The significance of the chemical
interaction in these examples is, therefore, far more
important.
5.1. Interactive light scattering effzciency
5.l.a. Variability. F r o m Table 5A it is evident that
the scattering efficiency of the aerosol constituent with

the g r m t ~ ' t affinity for water (in this case, the most
soluble) is the strongest function o f relative humidity.
The N H , H S O 4 efficiency is calculated to change from
3.9 m 2 g - t at 3 0 ~ rh to 8.6 m a g - l at 60~0 rh within
the composite urban aerosol, over a two-fold increase.
The effgiency o f the carbonaggous m a t , ~ t l ( 4 8 ~
water soluble) changed from 3.4 m a g - x at 30 ~ rh to
5.4 mZg - l at 6 0 ~ rh, a 5 9 ~ increase. The residual
material (10 ~o water soluble and with 2 0 ~ of its mass
in the optically efficient size ranse ) showed an,increase
in light scattering efficiency from 1.1 m a g - 1 at 30 ~ to
1.2 m s g - I at 60 ~ rh. Table 5B further establishes that

1034

CHtlSTINZ S. SLO^NE

Table 6. NH,HSO, droplets (ms g- J~ Light scattering effiCiency, a (m2g "t)


Relative
humidity

-50%

-100%

a(-50,%) M/Bsp

20
30
40
50
60
70
80

3.7
3.7
5.6
6.0
6.9
8.8
11.3

3.3
3.3
5.1
5.4
6.4
8.2
10.6

1.1
1.1
1.1
1.1
1.1
1.1
1.1

16
15
14
13
12
11
10
9
8
765-

AM

*Thermodynamic
(Table 1) assumed.

growth function

MIlS

for NH,HSO,

4-

16-

/'

is14-

/I

121i -

/ 6
//

'-6-

3 I
20

,
40

RELATIVE

,
50

50

60

70

80

s
60

70

HUMIDITY

mluble man

/4/'

,
30

40

Fig. 3. Light scattering efficiency of NH,HSO, within


the modelled 'typical' urban aerosol shown as a function
ofrelative humidity.
Chemical and physical interactions included,----physical interaction neglected.

//

30

RELATIVE

/~

.:bg 13 -

l0

20

I
80

su.

IL

IH

HUMIDITY

Fig. 2. Light scattering efficiency of NH4HSO, droplets as a function of relative humidity.


Chemical and
physical interactions included, - - - - physical interaction neglected.

Reduced

soh,d)le mrs

RL

It,l~ve Humiolty

the difference in affinity for water (not density, size


distribution, or refractive index) between the three
constituents controlled the difference in sensitivity to
relative humidity.
The calculations presented in Tables 5A and 5B also
suggest that the overall, or FPM, light scattering
efficiency (that for a simultaneous and equivalent
percentage change in the mass of each constituent) is
nearly a mass average of the light scattering efficiencies
of the aerosol constituents. It is not equivalent to
Bsp/M (=ABsp/AM, AM = M). The results in
Table 5A also indicate that X,iatM | > Bsp by approximately 10~o.
The relative humidity dependence of the light
scattering efficiencies produced in these calculations
results from the simulation of the aerosol behavior
depicted in Fig. 4. The 'start' position in Fi& 4 shows

RH

High RIIsUwHumidity

Fig. 4. Response ofaerosol size and composition to a mass


change in a soluble constituent. Black = insoluble core
volume; stripes ,- soluble salt volume; white -, volume
added by water absorption. Water and soluble salt mixture
occupies acombined, homogeneous volume in the modelled
aerosol.
an airborne aerosol at low relative humidity composed
of three species: water, a hygroscopic species and a
hydrophobic species. At low relative humidity the
water content is not appreciable. The light scattering
efficiency at low rh would result from the difference in
light scattering per unit mass between particles modelled at 'start' and those same particles transformed by
an increase/reduction in mass as shown. The difference
in light scattered by particles at 'IL' (increased mass/
low relative humidity) and particles at 'start' is not the
same as the difference in light scattered by particles at

Effect of oompesifion on light scattering e~ficiencies


'RL' (reduced mass/low relative humidity) and at
'start'. Th/s is the basis for the nonlinearity in the light
scattering e/ficiency shown in Table 5A.
At low relative b~midity, only the physical interaction is important for these calculations. With a change
in mass, the particles shift to a different size range and,
as a consequence, the number of part!cles within any
size range changes, but the total number of particles is
unchanged. These calculations may best simulate
changes in secondary aerosol material where the
number of condensation nuclei is unchanged.
Increases in mass correspond to particle growth, not
the production of additional particles. There is considerable evidence for such volume growth for sulfate
species in atmospheric aerosols (Hering and
Friedlander, 1982).
5.l.b. Comparison: "traditional' and statistical approaches. Three means of evaluating the light mattering efficiency have been discussed. In the statistical, or
empirical, approach the proportionality constants of
Equation (I) are estimated by statistically optimizing
the fit of field measurements. The fitting constants (e.g.
regression coefficients) are interpreted as the light
scattering efficiencies. Hence, the light scattering efficiencies are presumed to be constants, and thus,
independent of the relative concentration of constituents. The calculations summarized in Tables 5 and
6 and in Figs 2 and 3 demonstrate that the statistical
evaluation of the light scattering efficiencies as physically meaningful constants is not well founded.
In the 'traditional' theoretical formulation given in
Equation (5), the light scattering efficiencies of the
individual aerosol constituents are also independent of
one another. A minor interdependence would be
realized only if the removal of one species significantly
shifted the average refractive index or the species' molar
volumes in the remaining aerosol. The 'traditional' light
scattering efficiency is sensitive to variations in the
normalized mass size distribution, and therefore, need
not be a constant. However, it does not fully provide for
the relative humidity dependence. The relative humidity dependence must include any change in the
hygroscopicity with a change in aerosol composition. A
change in hygroscopicity alters the aerosol water
content, and hence, the size distribution of all remaining species. The average refractive index is also affected
by a change in water content.
In these more 'traditional' calculations, the total
number of airborne particles changes with a change in
mass. The size of individual particles does not shift
according to Do -~ XDo in Equation (8). The result is
illustrated by comparison of Figs IA and lB. Hence,
the 'traditional' approach fails to account for the shift
in the size distribution of the remaining species caused
by the loss of volume associated with the mass of the
species removed. In effect, that volume is replaced by
additional mass for the remaining species in each
particle. (Hence, the number of particles decreases for
constant total mass.) This does not correspond to a
realistic aerosol response process.

1035

As a result of these shortcomings, the "traditional'


light scattering efficiency is inaccurate by approximately 1 0 ~ in these calculations. Furthermore, it is
insensitive to the magnitude and direction of the mass
change in a chemical constituent. However, it may
serve to approximate the impact of changes in the size
distribution produced by the simple ventilation that is
sometimes responsible for large changes in light scattering observed in field studies.
It has been observed (White, 1986) that since light
scattering efficiencies calculated from the "traditional'
approach systematically underestimate the interactive
light scattering efficiencies of secondary species (such
as sulfate), they do not provide firm evidence that
statistical estimates are sometimes unphysically high.
However, these calculations indicate that this systematic error in the 'traditional' evaluations is too
slight (approx. 10 %) to undermine the conclusion (e.g.
Sloane and Wolff, 1984) that statistical estimates of
ambient sulfate fight scattering efficiencies have been
higher than would be expected from the physical and
chemical aerosol properties as modelled here.
5.2. Modelling limitations
The 'typical' urban aerosol calculations represent a
simplification of ambient aerosol behavior. The affinity of the hygroscopic species for water is modelled
as shown in Equation (7). In this model, the affinity of
water for the aerosol is responsive to the relative
concentration of soluble and insoluble species. The
model does not include any dependence of the semiempirical thermodynamic function ( i ) ( e ) / ( M s ~,
and hence the water content, on the relative concentrations of the soluble species. As noted earlier, light
scattering measurements made over a wide range of
relative humidity in two field studies were predicted
well with the physical modelling approach used in
these calculations. The relative concentrations of the
aerosol constituents did vary substantially in those
tests. However, if the variability of ie/M, with the
chemical identity of the soluble species were significant, that interaction would affect the relative humidity dependence of the light scattering efliciencies.
Unfortunately, the coarseness of the field measurements available (Sloane and Wolff, 1985) has not
permitted a more precise investigation of the aerosol
growth factor.
The calculations based on the modelled 'typical'
urban aerosol illustrate that the variability of the light
scattering efficiencies can be substantial even if the
ambient aerosol is not composed primarily of hygroscopic species. The effect illustrated is intended to
be realistic since it is based on a modelling approach
which has simulated measurements of light scattered
from an urban aerosol. However, these calculations are
meant to serve primarily as model calculations designed to demonstrate the significance of chemical and
physical interactions rather than simulations of a
specific ambient situation.
As stated in the introduction, this work addresses

1036

CtlaWnsE S. SLO^NE

the impact of aerosol loading and composition on


atmospheric optics. It does not address the relationship between emissions and the resultant aerosol
loading and chemical character; i.e. aerosol formation
processes. Therefore, the calculations represent the
sensitivity of the amount of light scattered by aerosols
to specific changes in aerosol composition (for example, a 50 ~o change in sulfate ma~). Emissions changes
realistically anticipated to affect a particular locale
would be likely to produce changes in aerosol composition different from the specific cases modelled
herein.
5.3. Test of statisticalmodels for relative humidity
dependence of Bscatt
The calculations for pure NH4HSOs are based on
experimentally verified thermodynamic growth factors, DIDo, and light scattering measurements.
Therefore, these calculations can be used as standards
against which to test the appropriateness of the
common empirical (i.e. statistical) approaches to establishing a relationship between the aerosol species'
masses, relative humidity and the light scattered by the
pure N H , H S O , aerosol.
The aerosol water content is generally not measured.
So, the light scattering attributed to water is sometimes
taken as the difference between heated and unheated
nephelometer measurements. This requires the somewhat risky assumption that the heated nephelometer
measures the light scattered when all water is eliminated from the the aerosol with no other change. That
difference is usually related to the relative humidity by:
Bsp(ambient) - Bsp(heated) = ~ aiM if~(rh),
I

i = soluble species, fi(rh) is a function of the relative


humidity, rh, often taken to be (l - r h ) - 1 . Then the
scattering efficiencies of the primary aerosol constituents other than water are given by a i - a ~ '
+a~fj(rh). The a~ are the constants statistically estimated from Bsp(heated)= ~. a~M i. The light scati

tering efficiencies, a~, derived from the heated


nephelometer measurements would be interpreted
from Equation(4) as the average value of
[3~(mo. Z, Do)/2Dop~] M~ (Do)/ ( M, ~ dDo with
M~(Do) =Mi(XDo) and X = DIDo.
We can test the appropriateness of the basic relationship Bsp = a'M + o'(l - rh)- ~M = aM for
droplets composed of the pure species NH4HS04.

Using the results of the calculations given in Table 6


and the usual empirical expectation that Bsp(heated)
= Bsp(40~ rh), a" = 5.1 mZg -!. Setting a(00~o rh )
equal to 6.4 mZg - ~ yields a' = 0.44 m 2 g - t. But these
values do not provide an accurate extrapolation even
from 60 to 8 0 ~ rh. as the predicted light scattering
efficiency at 80 ~/o rh would be 7.3 m z g- t which is 30
too low. Of course, the expectation that the difference
Bsp(ambient)-Bsp(heated) is proportional to (1

- rh)- t is flawed since the difference should approach


zero as rh decreases. Unfortunately, the simple relationship Bsp = a'(1 - r h ) - l M = aM works no better at simulating the relative humidity dependence of
Bsp or of the light scattering efficiency. Here a' (60 ~o
rh) = 2.6m2g - t leads to a(80~,~,rh) = 12.8m2g -1
which is 2 0 ~ too high. The simple function
rh(1 - r h ) - 1 might be suggested by Equations (4)and
(7), but it, too, proves to be inaccurate. It does not
effectively provide for the relative humidity dependence of (i ~ ~e ~/( Ms ~ nor of the refractive index.
These are not stringent tests. The empirical light
scattering efficiency estimates were assured of being
correct at 60 ~ relative humidity and then extrapolated
to 80~. A statistical fit to empirical data would not
necessarily have produced the correct 'starting' value at
60 ~. Thus, the simple relationships commonly used as
the starting point for estimation of the light scattering
efficiencies by statistical means were tested only for
accuracy in extrapolating from 60 to 8 0 ~ relative
humidity; a minimal requirement. In addition, neither
(1 - r h ) - t nor rh(l - rh)- 1 satisfies the extreme relative humidity criteria: f ( r h --, 1) =~ co.

6. S U M M A R Y

Forecasts of the impact of a change in aerosol


composition on visibility are based on light scattering
efficiencies, defined as a i = ABsp/M i, i = aerosol
constituent. This work explores the importance of
responsive interactions between aerosol constituents
in the evaluation of light scattering efficiencies. The
interaction of hygroscopic species with water is used as
one example of a chemical interaction between two
species that influences effective light scattering elficiencies. Many different types of chemical interactions can be envisaged which are not considered here;
for example, the influence of pH on nitrate
concentration.
The calculations presented here illustrate the extent
to which light scattering efficiencies are not constants.
Their variability should be considered in forecasting.
For example, if the lowest visibilities at a site occur
under conditions of higher than average relative
humidity, then the light scattering efficiency used to
forecast the change in visibility at that site would not
be the same for adverse and median visibility events.
The calculations presented in this work demonstrate
that:
(I) Chemical interactions between aerosol constitucnts can causea two-fold variation in the response

of the light scattering to a change in the mass of a


hygroscopic aerosol constituent.
(2) The 'traditional' light scattering efficiencies are
underestimates in error by 10 % for these calculations.
(3) The individual light scattering efficiencies for
each aerosol constituent need not satisfy the additive
constraint:

~aiMt = Bsp. In these calculations,


i

Effect of composition on fight scattering ei~ciencies


aim i was greater than Bsp by approximately 10%.
i

(4) The response of the light scattered by atmospheric aerosols to a change in the mass of an aerosol
constituent is not linear. Nonlinearities are generally
less than 10% of the light scattering efficiencies.
(5) Simple relationships (such as Bsp = ~ a i M i ( 1
i

- r h ) - l ) , which are commonly used to emulate the


relative humidity dependence of the total light scat,
tered or the light scattering efficiencies,are shown to be
inaccurate by at least 20 %.
(6) The statistical evaluation of ambient light scat.
toting efficiencies is not well founded because the light
scattering efficiencies are variables which deviate
markedly from constant values under normal atmospheric conditions. Field measurements should at least
be stratified by relative humidity before using statistical estimation procedures to evaluate ambient light
scattering efficiencies.
(7) The light scattering efficiency of an aerosol
constituent, such as N H , H S O , , is not of constant
value, but varies with the overall aerosol composition.
REFERENCES

Charlson R. J., Covert D. S., Larson T. V. and Waggoncr A. P.


(1978) Chemical properties of tropospheric sulfur aerosols.
Atmospheric Environment 12, 39-53.
Friedlander S. K. (1977) Smoke, Dust and Haze. Wiley, New
York.
Handbook of Chemistr.v and Physics (1967) The Chemical
Rubber Publishing Co., Cleveland, Ohio.
Hanoi G. (1976) The properties of atmospheric aerosol
particles as functions of the relative humidity at thermodynamic equilibrium with the surrounding moist air. Adv.
Geophys. 19, 73-188.

AE 20:5-0

1037

Hanel G. and Lehmann M. (1981) Equilibrium size ofaerosol


particles and relative humidity: new experimental data
from various aerosol types and their treatment for cloud
physics application. Comrib. Atmos. Phys, .~4, 57-71.
Hering S. V. and Friedlander S. IL (1982) Origins of the
aerosol sulfur size distributions in the Los Angeles Basin.
Atmospheric Environment 16, 2647-2656.
Kerker M. (1969) The Scattering of Light. Academic Press,
New York.
Lee S. C. and Tien C. L. (1981) Optical constants of soot in
hydrocarbon flames, l gth Syrup. (Int.) on Carbon Combust.,
Combust. Inst., 1159--1166.
Middleton W. E. K. (1952) Vision Through the Atmosphere.
University of Toronto Press, Toronto.
Ouimette J. R. and Flagan R. C. (1982) The extinction
coefficient of muiticomponent aerosols. Atmospheric
Environment 16, 2405--2419.
Patterson E. M. (1977) Atmospheric extinction between
0.55 t~mand 10.6/~mdue to soil derived aerosols. Appl. Opt.
16, 2414--2418.
Roessler D. and Faxvog F. (1980) Optical properties of
agglomerated acetylene smoke particles at 0.514/ira and
10.6/zm wavelengths. J. Opt. Soc. Am. 70, 230-235.
Sloane C. S. (1983) Optical properties of aerosols-comparisons of measurements with model calculations.
Atmospheric Environment 17, 409-416.
Sloane C. S. (1984) Optical properties of aerosols of mixed
composition. Atmospheric Environment lg, 871-878.
SIoane C. S. and Wolff G. T. (1985) Prediction of ambient
light scattering using a physical model responsive to
relative humidity: validation with measurements from
Detroit. Atmospheric Environment 19, 669-680.
Tang I. N. (1980) Deliquescence properties and particle size
change of hygroscopic aerosols. In Generation of Aerosols
and Facilities for Exposure Experiments (edited by Willek
K.), pp. 153-167. Ann Arbor Science, Ann Arbor.
Tang L N. and Mukelwita H. R. (1977) Aerosol growth
studies--III: ammonium bisulfate aerosols in a moist
atmosphere. J. Aerosol Sci. 8, 321-329.
White W. H. (1986)On the theoretical and empirical basis for
apprortioning extinction by aerosols: a critical review.
Atmospheric Environment (in press).