JOURNALOF APPLIED PHYSIOLOGY

Vol. 38, No. l,January 1975. Printed in U.S. A.

A theory of aerosol deposition

in the human respiratory tract

D. B. TAULBEE AND C. P. YU
Department of Engineering Science, Aerospace Engineering and Nuclear Engineering,
State UniversitvJ ofd New York at Buffalo.
d.J J
Buff’alo.
yu J
New York 14214

TAULBEE, D, B., AND Cm P. Yu. A theory of ~tmxml defiosition in the depths in the lung va ries with a) the physical properties of
human resjiratory tract. J. Appl. Physiol. 38(l): 77-85. 1975.-- aerosol particles such as size, shape, mass, and charge, b)
The deposition of inhaled aerosol particles in the human respira- the breithing conditions including tidal volume, frequency,
tory tract is due to the mechanisms of inertia impaction, Brownian and breathing pattern, and c) the physiological conditions
diffusion, and gravitational settling. A theory is developed to pre-
such as residual capacity and airway structure which varies
dict the particle deposition and its distribution in human respira-
from subject to subject. Rigorous determination of deposi-
tory tract for any breathing condition. A convection-diffusion
equation for the particle concentration with a loss term is used to tion requires a complete knowledge of the airflow pattern
describe the transport and deposition of particles. In this equation, in the lung, which is very complex and yet not fully under-
an apparent diffusion coefficient due to the velocity dispersion in stood.
the lung is present and found to be the dominant diffusion mecha- In spite of these difficulties, several theoretical attempts
nism for the cases considered here. Expressions for deposition by have been made in recent years by Findeisen (5), Landahl
various mechanisms are also derived. The governing equation is (9), and Beeckmans (2) based upon a compartmental model.
solved numerically with Weibel’s lung model A. The particle con- The important characteristics of Findeisen-Landahl-
centration at the mouth is calculated during washin and washout Beeckmans theory are a) the anatomical structure of the
and compared favorably with experimental recordings for 0.5-
lung was represented by a series of well-mixed compart-
pm diameter di(Z-ethylhexyl) sebacate particles. The total deposi-
ments; b) particles which did not deposit in a compartment
tion in the lung for particle size ranging from 0.05 to 5 pm is also
computed for a 500-cm3 tidal volume and 15 breaths/min. The were transferred to the following component and an empiri-
results in general agree with recent measurements of Heyder et al. cal function was used to account for the mixing process of
However, a particle size of minimum deposition is found to exist intrapulmonary gas flow; c) the breathing cycle was ap-
theoretically near 0.3 pm. proximated by four steady-flow processes : inhalation with
constant velocity, a pause, exhalation with constant ve-
transport equation; inhaled aerosols in airways; industrial hygiene; locity and followed by another pause; d) the particle deposi-
drug inhalation; diagnosis of pulmonary disease tion by combined mechanisms of impaction, Brownian
diffusion, and sedimentation was assumed to be the super-
position of depositions due to three independent mecha-
nisms. Since exact expressions for deposition in a realistic
THE DETERMINATION of aerosol deposition and its distribution geometry and flow pattern in the lung airways were not
in the lung during breathing is important in the field of in- available and difficult to determine, semiempirical expres-
dustrial hygiene, in the administration of therapeutic drugs sions were used. The impaction deposition was estimated by
by inhalation, and in the diagnosis of obstructive pulmonary the results in a 90” bend in a smooth tube. The diffusion
diseases. A large number of studies, both experimentally deposition was obtained by the analytic solution for a con-
and theoretically, have been devoted to this subject over vection-diffusion flow in an infinitely long horizontal circu-
many years. General reviews can be found in several recent lar tube under fully developed flow conditions. An expres-
papers and monographs (I, 6, 13). sion, based upon the particle deposition by sedimentation
While the total deposition of aerosols in the lung can be during flow in a horizontal tube was also used to calculate
determined by an inhale-exhale experiment, it is not pos- the sedimentation deposition. In spite of these many obvious
sible however to achieve a detailed measurement on its simplifications, the Findeisen-Landahl-Bee&mans theory
distribution at various depths in the lung at the present is successful in comparison with the total deposition of
time. This information can only be obtained by theoretical earlier measurements and has been formally used in pre-
methods. Unfortunately, a theoretical study of this problem dicting the regional deposition of aerosol particles in the
is also very complex because of the complicated structure respiratory system ( 13).
of the lung. The experimental determination of particle deposition
The deposition of aerosol particles in the lung occurs is usually done by an inhale-exhale experiment with con-
primarily by three mechanisms-inertia impaction, gravi- tinuous recordings of the lung volume and particle con-
tational settling, and Brownian diffusion. The relative centration at the mouth (4). The breathing pattern is con-
amount of the deposition due to each mechanism at different trolled by the subject with the use of a spirometer. The total
77

Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (134.129.182.074) on November 16, 2018.

Copyright © 1975 the American Physiological Society. All rights reserved.
78 D. B. TAULBEE AND C. P. YU

fraction of particles deposit in the lung is then calculated corresponds to the following relation
from these volume and concentration curves. The Findeisen-
La ndahl-Beeckmans theory described above is steady state c(A + v> ]t+dx - c(A + v> ItAx
(0
nature; it can On ly pred ict the par title d eposi tion for a = AJl,At - AJI,+a,At - LAXA~
steady-state breathing cycle consisting of constant-velocity
inhalation and exhalation processes. For arbitrary breath- where c is the particle concentration, A is the total airway
ing, a time-dependent computational theory is needed. This cross-sectional area, v is the alveolar volume per unit
paper is intended to develop such a theory, based upon a length of the airway, L is the combined particle Ioss due to
physical transport theory approach. It will be seen that this impaction, diffusion, and sedimentation deposition per unit
theory, to be developed in the foIlowing section, not only length of the airway per unit time, and J is the particle flux
gives more physical insight about air mixing and particle per unit airway cross-sectional area. If the particle inertia
deposition in the lung, but also provide results which may is neglected, the expression for J may be written as
directly be compared with the experimenta recording of dC
particle concentration at the mouth for any arbitrary J =uc-D-
dX
(2)
breathing condition. It is thought that only through such a
fine level of comparison between theory and experiment, where u is the average air velocity at the cross section and D
we would have confidence on the theoretical prediction of is an apparent diffusion coefficient that we shall discuss later.
the distribution of particle deposition in the lung. Dividing Eq. I by AxAt and let Ax + 0 and At --+ 0, we
obtain
THEORY
; [(A + v)c] = -& (AJ) - L (3)
Weibel (16) assumes that the lung is a symmetric, dichot-
omously branching system, consisting of 23 generations. The air velocity in the airways is small compared with
The total airway cross section A is a function of generation the velocity of sound and any compressibility eflect of the
number and it increases drastically in the last few genera- air is negligible. The conservation of air volume leads to the
tions. Starting from the 17th generation, the airways become relation
alveolated. The alveolar volume expands and contracts aQ
-= -- dv (0
during breathing+ The geometrical model of the lung that dX dt
we shall use in this study is the airway model proposed by
Scherer et al. (10) in a recent study of gaseous diffusion wh ere Q = uA is the total flow rate of air passing through
problem. an airway cross set tlon.
Consider that the lung airway is approximated by a We integrate Eq. 4 to give
variable cross-section channel (Fig. 1) with cross-sectional
area being a function of generation, given by Weibel. In Q = Qo - ; 1‘ v dx = Q. -
0
F (5)
the last seven generations, additional volume due to alveoli
encircles the channel. This alveolar volume again is a func- where Qo(t) = Q(0, t) is the airflow rate at the mouth
tion of the generation number. The breathing process is which is chosen to be located at x = 0, and V, is the total
then pictured as the movement of air in which particles are alveolar volume between x = 0 and X.
susDended into and out of this channel as the alveolar Let F(x) be the fraction of total alveolar volume located
volume expands and contracts uniformly. The airflow is between x = 0 and x. Then
considered to be along the airway path x while being uni-
form over the cross section of the duct. Q(x, t) = [l - F(x>]Q-o (6)
The equation of mass balance for the particles can easily We integrate Eq. 5 with respect to t and make use of
be established by considering an element of the channel Eq. 6 to give
with length Ax as shown in Fig. 2. The increase of particles
in the element during the time At must be equal to the net vab, 0 = v”,(x)+ F(x)St Qodt (7)
particle flow into the element minus the particle loss due to 0

deposition in the element during At. Mathematically, it where V.O(x) is the alveolar volume between x = 0 and x
at rest.
Differentiating Eq. 7 with x, we finally obtain the follow-
ing expression for v(x, t)

v(x, t) = g(v, + [Qodl) (8)

LENGTH x where Vi, is the total alveolar volume in the airways at

rest.
With the use of Eq. 4, Eq. 3 may be further cast in the
form

FIG. 1. Airway model .

Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (134.129.182.074) on November 16, 2018.

Copyright © 1975 the American Physiological Society. All rights reserved.
AEROSOL DEPOSITION IN HUMAN RESPIRATORY TRACT 79

ALVEOLAR tubes simultaneously at an initial time t = 0. Let u’ be the

VOLUME
velocity dispersion, i.e., difference between the air velocity
in an airway tube with the average air velocity of all tubes
in a generation. The velocity u’ remains constant in a given
airway tube. For convenience, we assume that u’ follows a
Ix+Ax normal distribution in the form
1

f(
u’

>
= e
exp ( - u’~/%~u~) W)
42 ?rfl
L-Ax-1
where f(u’)du’ is the fraction of airway tubes which have
FIG 2. Differential element of the airway. velocity dispersion between u’ and u’ + du’ and G is the
standard deviation. Then, after a time period T, the par-
ticle concentration at a distance x’ = u’T from the average
Equations 8 and 9 are the two equations which form. the
air front will be
basis of our study. For the convenience of performing
00
numerical analysis of these equations, it is desirable to
introduce a stretched coordinate z based upon the genera- c(x, t) = co f(u’) du’ ua
s x’/T
tion number of the airways. Thus, z takes on values of U-23
corresponding to locations at the end of the trachea to the This equation states that concentration for the entire flow at
end of generation 23. The value z = - 1 corresponds to the a location x’ ahead of the average airfront is equal to the
mouth. Let Z(z) be the length of the airways per generation. sum over all tubes in which the particle material has moved
Then I = dx/dz and d/dx = (l/ Z) (d/dz). Upon substitu- past the location x’.
tion, Eg. 8 and 9 become, respectively The substitution of Eq 12 into E+ 13 and the use of a
new variable 4 = u’/u gives
v(z,t) = -; pT+ 4’Qodt) c(x’, T) = & r00 49
-32
dE (14)

An apparent diffusion coefficient can be found by com-

/(A + v) $ + Q 2 = ; - IL (II) paring the development of diffusion profile of a stationary
concentration front located at x’ = 0, where c = co for
Equation 11 as it stands contains two unknown functions x’ < 0 and c = 0 for x’ > 0 initially. In this case, the solu-
2) and L, representing, respectively, the effects of air mixing
and particle loss by deposi tion on the airway surfaces. The
quantitative success of the proposed theory depends largely
tion for the concentration

c(x’, T)
is (8)

= L-
00
s d( e-t2/2
W)
on the choice of these two functions. We shall discuss them 42 r r’/Z/2DT

separately in the sections to follow. The comparison of Eq# 14 and 15 leads to the relation

AIR MIXING AND APPARENT DIFFUSION COEFFICIENT

The longitudinal diffusion of particles in the airways The characteristic time scale for T is Z/u, Eq* 16 thus gives
arises from several different mechanisms. Brownian diffu-
D = D, = 42 WI
sion is unimportant over the entire airways except for
extremely small particles. If the airflow is laminar, Taylor It turns out from calculation that this apparent diflusion
diflusion ( 14) which results from the combined action of coefficient D, is the dominant diflusion coefficient in the
radial diffusion and air velocity profile in each airway tube lung airways, which accounts for the pulmonary air mixing
is present. In the upper airways, particularly in the trachea, process. The value of standard deviation G in Eg. 17 is not
the Reynolds number for the airflow may be greater than exactly known. Presumably, it has a value similar to the
critical and the flow is turbulent. Turbulent diffusion thus standard deviation of the length and diameter distribution
may take place (15). of the airways given by Weibel. In our computation, G is
Starting from the end of the trachea where the branching given a value of 0.6.
of the airways begins, additional diffusion arises from the
fact that the air in each airway tube at the same generation
DEPOSITION FUNCTIONS
does not move with the same velocity but follows a velocity
distribution. This velocity dispersion may result from the The function L in Eguution II represents the deposition
randomness in geometrica sizes (diameter and length) and loss of particles to the airway surfaces per unit time and unit
dynamical properties (resistance and compliance) of the length of airways. The mechanisms of deposition are due
airway tubes at the same generation. From Weibel’s data to impaction, Brownian diffusion, and sedimentation. The
the value of the standard deviation of the airway length and determination of L due to the combined mechanisms under
diameter distribution is found to be between 0.5 and 0.8. realistic conditions of airflow pattern and airway geometry
Consider the air, which carries a uniform concentration is extremely difficult. In the Findeisen-Landahl-Beeckmans
of particles CO , moving into a new generation of airway theory, the total deposition by combined mechanisms was

Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (134.129.182.074) on November 16, 2018.

Copyright © 1975 the American Physiological Society. All rights reserved.
80 D. B. TAULBEE AND C. P. YU

60 cated problem. One may be asked as to whether these

0 TRACHEA
calculations are necessary. If the time scale for deposition
MAIN BRONCHI
LABOR BRONCHI taking place is very short compared with the time duration
SEGMENTAL BRONCHI of a breathing cycle, To , the deposition can be regarded as a
quasi-steady process. The three time scales for the particle
deposition are
f=1.3(Stk -
Ti = X2/v for impaction deposition
T d = X2/DB for diffusion deposition
T, = X/u, for sedimentation deposition W)
where Y is the kinematic viscosity of the air, Ds is the
Brownian diffusion coefficient, u, is the settling velocity of a
l
particle, and X is the cha racteristi .c length, ta ken to be
otlr 4+4+&b
either the diameter of an airway tube or the boundary
U3 IO+ lo-’ IO0
STOKES’ NUMBER Stk layer thickness if the particle concentration is of a boundary
layer type+
FIG. 3. Deposition efficiency of the airway generations of human
lung cast measured by Schlesinger and Lippmann (11). It may be shown that Ti < To everywhere in the lung
and the impaction deposition may be treated as a quasi-
assumed to be the superposition of deposition due to three steady process. An estimation of Td and Ts for micron or
independent mechanisms, and they were calculated by submicron particles shows that Td and T, in the airways
assuming a Poiseuille velocity profile in a tube geometry. may be greater than To. Thus, for an exact calculation,
This simplification in geometry and in flow field will lead time-dependent expressions for deposition function bY
to obvious errors in the calculation, particularly for upper diffusion and sedimentation may be required.
airways where the effect of air motion on the deposition is Since our primary concern in this paper is to establish a
important. theory of computation described above (THEORY), for the
To improve the geometrical considerations, Bell and moment, we shall only try to make appropriate improve-
Friedlander (3) recently considered ment of the deposition functions used in the Findeisen-
the deposition near a
bifurcation point in the upper respiratory airways. Based Landahl-Beeckmans theory in our calculation. Similar to
upon a two-dimensional wedge model, they calculated the their theory, we assume that the deposition loss due to
particle deposition by impaction and diffusion. The theo- combined mechanisms is the sum of that due to three inde-
retical results were compared with their experimental data pendent mechanisms, i.e.
and showed good agreement. Experimental measurements L = Li + Ld + 1;s (20)
were also made in a three-dimensional bifurcation, but only
limited data were available and the results cannot be put where Li, Ld, and Ls are, respectively, the deposition
in use. functions (particle loss to the wall per unit length per unit
When there is more than one airway generation, the air- time) by impaction, diffusion, and sedimentation. We
flow problem also becomes complicated. Schroter and further assume that L i , Ld , and L, take the same form
Sudlow (12) have demonstrated that the velocity profiles during both inhalation and exhalation processes.
depart drastically from that of the Poiseuille flow and For impaction deposition, the results of Schlesinger and
usually exhibit more than one peak. Such velocity profiles Lippmann given by Eg. 18 is used and we have
would promote transverse air mixing in the airways and cAu
strongly affect the particle deposition behavior. Schlesinger Li = 1.3 7 (Stk - 0.0001) for Stk -> 0.0001 (21)
and Lippmann (11) recently measured the particle deposi-
tion for inhalation in the upper airways by means of hol- Both diffusion and sedimentation are important in the deep
low Silastic casts extending from the trachea through the lung where the airways are alveolated. The geometry for the
segmental bronchi. The deposition mechanism in their range alveolated airways is considerably different from a circular
of experiment is primarily due to inertia impaction. Their tube model in the Findeisen-Lahdahl-Beeckmans theory.
results on the fraction f of aerosol deposition in each genera- We therefore assume that these results only apply to those
tion to total aerosol entering is plotted versus the Stokes airway surfaces which are not alveolated. For the alveolated
number Stk = TU/d in Fig. 3, where T = pd$/ 18~ (p being surfaces, different formula are derived in the APPENDIX for
the mass density of the particle, d, the particle diameter, and diffusion and sedimentation. Let the number of alveoli per
1-1the air viscosity) is the relaxation time of a particle and d generation be n, and the total number of airways per
is the diameter of the airway. We find a simple algebraic generation be n, then the fraction of airways surfaces which
expression to correlate their results for Stk < 0.00 1 as folIows are alveolated is
N,s dF
f= 1.3 (Stk - 0.001) u8) a =-- (22)
nndl dz
and f = 0 for Stk < 0.001.
where s is the entrance area to each alveolus. Denoting va
Another complication arises from the fact that breathing
as the volume of each alveolus, Eq. 22 may be written as
is a time-dependent process. The determination of the
deposition function in various anatomical units under time- &d s dF
dependent airflow conditions poses an even more compli- a ~
= 4Al v, dz
(23)

Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (134.129.182.074) on November 16, 2018.

Copyright © 1975 the American Physiological Society. All rights reserved.
AEROSOL DEPOSITION IN HUMAN RESPIRATORY TRACT 81

From the assumed geometry of an alveolus described in the easily be solved by the method of characteristics. This sim-
APPENDIX,~~ have plification significantly reduces the difficulty in carrying
out the numerical solution. Without this simplification the
s 1
-= 0.963 d (24) full Eq. II would have to be solved in the mouth and trachea
Va a region, where upon inhalation, a slowly diEusing concen-
where d, is the diameter of an alveolus. a then takes the tration front moves down a relatively long passage (22 cm)
form before apparent diffusion attenuates the large concentra-
tion gradients in the bronchioles. A very fine grid would be
dV, dF required to resolve the flow-wise concentration profiIe so
a = 0.241 - - (25)
Aid, dz as to prevent significant numerical type errors.
In addition the loss function in Eq. 30 will be neglected in
From the expressions used by Beeckmans, we may obtain
the mouth and trachea. Since the surface area of this region
the following deposition functions for diffusion and sedi-
is relatively small the amount of sedimentation and diffu-
mentation for the airway surfaces which are not alveolated
sional deposition, as compared to the total lung, would be
to be
small. Furthermore, for the upright position, sedimentation
deposition would be minimized. Only inertial deposition of
Ld=: (1 - a)?[1 - exp(-3F)] (26) large particles in the mouth and throat could be of signifi-
cance. This deposition is neglected in this study. Hence,
the concentration distribution in the mouth and trachea
L,- (1 - dF[l - exp(-&T)] (27) can be obtained from

where a = (r/4)d2 is th e cross-sectional area of each airway c(x,t) = c[x -&ltQo(l)dt] (31)
tube and g is the gravitational acceleration.
For the alveolated airway surfaces, the deposition func- where A0 is the area of the mouth,
tions for diffusion and sedimentation are determined in the Equation 11 is to be solved in the region from the end of
APPENDTX. The results are, respectively the trachea (z = 0) to the end of generation 23 (z = 23).
The generation number is taken to correspond to the loca-
Ld = 4.23 ; (28) tion at the end of that generation. The geometrica quan-
tities were determined as follows. From tabulated values of
length, diameter, and the three-fourths inflated alveoli
Ls = 1.13 7 N;t3(VaT)2’3 &f W-9 volume for each generation, the accumulated airway length,
x, airway volume, V, and alveoli volume, Va”g are computed
where N, is the total number of alveoli in the lung. The total for integer values of the generation number z. The fraction
deposition functions are then the sum of Eg. 26 and 28 for F(z) of accumulated alveoli volume in the interval 0, z is
diffusion and Eq. 27 and 29 for sedimentation. simply Voa/VaoT . The airway length per generation 2 is
obtained as a tabulated function of z by performing the
PROCEDURE OF NUMERICAL COMPUTATION differentiation dx/dz numerically at each integer point
Z = 0 to z = 23. The volume per generation at each integer
From the transport model described above (THEORY) a
point 2 = 0 to z = 23 is obtained by numerically perform-
numerical method was developed to compute the particle
ing the differentiation dV/dz. The airway cross section
concentration throughout the mouth, trachea and 23 A(z) is then obtained by A = (dV/dz)/Z. It is noted that
generations of the lung as given by Weibel’s three-fourths
these procedures consistently define the geometric quantities
inflated model. In the mouth and trachea the transport of
for integer values of z which corresponds to the location at
particles is assumed solely by convection since the effects the end of the respective generation. The tabulated func-
of Brownian and Taylor particle diffusion are small, there
tions at integer values of z are interpolated to a finer mesh of
is no apparent diffusion in this region and there is no
interval AZ for use in the numerical solution.
turbulent diffusion if the flow in the trachea is laminar. The
Due to the complexity of the coefficients Eq- Xl was
latter restriction is satisfied, if for a constant flow rate and a
solved numerically using finite differences. The simple
critical Reynolds number of 2,000, we have tidal volume/
implicit method for solving parabolic partial differential
period < 208 in the trachea. Quiet breathing satisfies this
equations was adopted here. Scherer et al. (10) also used
restriction; however, turbulence can occur for rapid breath-
this meth .od an .d found it to be stable. Other more accurate
ing with large tidal volumes. We intend to investigate the
methods, such as the Crank-N ‘ichelson method were tried,
effects of turbulence on the particle transport; however, for however, overshoot errors were present in the computed
the present work we will restrict ourselves to breathing
results.
conditions with subcritical Reynolds numbers. Hence, all
Eauation II can be rewritten
axial diffusion in the mouth and trachea will be neglected.
For the mouth and trachea Eg. II simplifies to d2C
E+pc+q$rs
A$+Qg= -L W) where
which is a hyperbolic partial differential equation and can P = L/IA -I- V)C

Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (134.129.182.074) on November 16, 2018.

Copyright © 1975 the American Physiological Society. All rights reserved.
82 D. B. TAULBEE AND C. P. YU

and

q C [ Q-Fz d WEJO
r = AD./P(A
I/ w-I-v>
+ v)
Given the geometry of the model and the flow rate these
coefficients can be computed as functions of z and f. Finite
differencing E+ 32 using a backwards diflerence in time
and rearranging gives
w 5 IO 15 20 23
Az2 GENERATION NUMBER z
-(ri ct> + *AZ qi “‘)c~?~ + at + 2$’ + pl!t)A~’ cFE’
( )
(33)
- ( $1 - +AZ q;“‘)c/~; = AZ2
At &At1

where the subscript i denotes a grid point in the interval

0 -< z -< 23 such that
0

Zi = 23(i - l)/(imax - 1) 1 I< i _< i,,,

Given the boundary conditions at z = 0 and z = 23 and
EXPIRATION
the values of the quantities pi , qi , and ri Es, 33 with its
tridiagonal coefficient matrix can be solved for ci (1 -< i <-
.
lrn ax ) by a straightforward elimination process. The quan-
. . GENERATION NUMBER z
tlties pi , q i , and ri can be reevaluated and the solution
found for successive time steps. FIG. 6. Theoretical predictions of particle concentration along the
airway during the eighth breath of washin. Particle diameter = 0.5
The boundary conditions are as follows. During inhala-
pm, tidal volume = 600 cm3, and breathing period To = 3.75 s. Par-
ticle concentration of the inhaled aerosol is normalized.
PARTICLE CONCENTRATION (WASHIN) I

tion the concentration is specified at the mouth and with

Eq. 31 the concentration at the end of the trachea (z = 0)
is determined. During exhalation dc/dz = 0 at z = 0
since there is only convective transport in the trachea. At
z = 23 we put &z/&z = 0 since we do not want a diffusional
PARTICLE CONCENTRATION (WASHOUT) deposition flux at the end of the model.

RESULTS AND DISCUSSION

To test our theory, calculations were carried out for those

t (SEC!
cases where recent experimental measurements were made
FIG. 4. Theoretical predictions (dotted lines) of particle concentra- (4, 7). Davies et al. (4) conducted measurements using
tion at the mouth compared with the experimental recordings (solid 0.5-rJ,rn diameter particles produced from di (Z-ethylhexyl)
lines) by Davies et al. (4) for a given breathing pattern indicated by
sebacate (density 0.91 g/ cm3) by a modified Sinclair-LaMer
lung volurm- curves. Particle diameter = 0.5 pm, tidal volume = 600
cm3, and breathing period = 3.75 s. Particle concentration of the in- type aerosol generator. The particle size was nearly uni-
haled aerosol is normalized. form and had a standard deviation of 1 15. When the tidal
l

volume is 600 cm3 and frequency is 16 breaths/min, their

recordings of lung volume and particle concentration at the
mouth for the washin and washout are compared with our
theoretical calculations as shown in Fig. 4. In fact, the
theoretical volume curve is a given time function which can
be chosen to fit exactly the experimental values. Here we
have approximated the experimental curve by a sawtooth
function for the convenience of computation. With this
WASHIN WASHOUT choice of volume function, the particle concentration at the
mouth is calculated as a function of time. It is seen in Fig. 4
012345012345
that the theoretical prediction of particle concentration is
NUMBER OF NUMBER OF
favorably compared with the experimental values of both
BREATHS n BREATHS N washin and washout In the first two breaths, there is a
slight overestimation of the minimum concentration during
FIG. 5. Theoretical fractional recovery of particles (solid lines)
during washin and washout compared with the experimental. data of washin and an underestimation of the maximum concen-
Davies et al. (4). tration during washout. The comparison between theoreti-

Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (134.129.182.074) on November 16, 2018.

Copyright © 1975 the American Physiological Society. All rights reserved.
AEROSOL DEPOSITION IN HUMAN RESPIRATORY TRACT 83

cal and experimental results becomes very good in later 1

,NORMAL ALVEOLUS
breaths.
Davies et al. also made measurements on the fractional
recovery during washin and washout. Following their
definitions, let I, be the amount of particles inhaled in the
nth breath of washin, E, be the amount of particles exhaled
in this breath, then the fractional recovery is

Similarly, let 11 be the amount of particles which was in-

INFLATED ALVEOLUS
haled in the last breath of steady inhalation of particles, E,
be the amount of particles exhaled in the Nth breath of air, FIG. 8. Geometry of an alveolus.
the fractional recovery during washout is then
particle, size with minimum deposition, which is generally
RN = EN/II. (38 believed to be at between 0.3 and 0.5 pm, depending upon
If R is the fractional recovery during steady breathing, then particle density air temperature, etc. Further experimental
R,/R increases from zero to unity during washin while data in this particle range are needed to support their find-
1 - RN/R increases during washout. A comparison of ings. In Fig. 7, the theoretical results of total deposition are
R,/Rd and 1 - RN/R between theoretical prediction and compared with the experimental data of the subject 3 in the
experiment is given in Fig. 5 for the subject JEZ in their experiment. The tidal volume used is 500 cm3 and frequency
experiment. The comparison is again very good. is 15 breaths/min. The comparison is extremely good for
It is worth mentioning that the experiment of Davies particle size between 0.2 and 0.6 pm* For particle larger
et al. seems to show that the steady-state breathing can be than 0.6 pm, the theoretical prediction gives a higher value
reached after four breaths of washin. This is not exactly of deposition. Furthermore, our theory predicts that a
correct from our calculations. Our results actually show particle size of minimum deposition does exist and it occurs
that the number of breaths of washin needed to reach the near 0.3 pm, Deposition increases for particle smaller than
steady state depends upon the particle size. For 0.5~pm 0.3 pm due to Brownian diffusion, it increases for larger
particles, it requires as much as 10 breaths of washin to particles due to sedimentation+ For the particle size smaller
reach the stage where the total deposition of two successive than 2 pm, the present tidal volume and frequency of
cycles differs by less than 1 %. At the eighth breath of breathing do not appear to give any appreciable impaction
washin, the particle concentration profiles at various deposition. The calculation on the distribution of deposited
generations of the lung are plotted in Fig. 6 during inhala- particles at various airway generations and its contribution
tion and exhalation. The movement of particle front due to from various mechanisms is in progress and will be given in a
air convection is clearly displayed in this figure. future publication.
Di (Z-ethylhexyl) sebacate particles with diRerent size We have also made comparison between our theoretical
ranging from 0.2 to 1 pm in diameter were also produced by values of total deposition with those predicted based upon
Heyder et al, (7) to study the particle deposition in the lung, the Findeisen-Landahl-Beeckmans theory. Our results for
Due to an improved aerosol generator they built, the par- all particle sizes and breathing conditions are considerably
ticle size was more uniform than Davies’ group with stand- smaller and thus closer to the measurements. It appears that
ard deviation between 1.04 and 1.06. It is interesting to the theory that we proposed here is an attractive one.
note that their results failed to show the existence of a Better agreement between theoretical prediction and
experiment can be achieved by further improvements of the
deposition function, diffusion coefficient and lung geometry.

APPENDIX

Determination of Deposition Function in Alvaolated

Airways lhe to IX&ion and Sedimentation

Model geometry of an alveolus. According to Weibel, the shape of an

alveolus is similar to a portion of a sphere with diameter d, and height
h (see Fig. 8). The total volume of an alveolus is then

v, = E h2(3d;, - 2h) WI
6

and the inner surface area is

aa - ILULI \AAk /
0.05 0.1 0.2 0.5 1.0 2.0 5.0
PARTICLEDIAMETER(pm) Define 81 be the angle between the axis of symmetry and a line joining
the center of the sphere and any point on the circular boundary of the
FIG. 7. Theoretical prediction of total fraction of particle deposi- alveolar opening. Then
tion (solid line) for different particIe sizes compared with the experi-
mental data of Heyder et al. (7). Tidal volume = 500 cm3 and breath- d, - 2h
cos 01 = ~ (A3
ing period = 4 s. 6%

Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (134.129.182.074) on November 16, 2018.

Copyright © 1975 the American Physiological Society. All rights reserved.
84 D. B. TAULBEE AND C. P, YU

and the area of the alveolar opening is particle concentration

diffusion coefficient
2 apparent diffusion coefficient, EQ. 17
S = = ; d,z( 1 - cos2 &) Brownian diffusion coefficient
diameter of an airway tube
We assume that during the process of breathing 81 remains constant, diameter of an alveolus
i.e., the alveolus remains the same shape all the time and only its particle diameter
diameter varies. From Weibel’s data for three-fourths maximum inha- amount of particles exhaled in the Nth or nth breath
lation, h = 0.23 mm, d, = 0.28 mm, which gives cos 81 = -9/14 or fraction of total alveolar volume betxeen x = 0
81 = 2.268 radians. Thus and x
f fraction of aerosol deposition in each generation to the
Va = 0.478 d$ (As> total aerosol entering
% = 2.58 da2 W) function of U’
fb’>
s = 0.561 da2 WI gravitational acceleration
s
Di$usion deposition.The Peclet number, ud,/DB , in the alveolar h height of an alveolus, Fig. 8
region is usually larger than unity. For 0.5-pm particles under a 11 amount of particles inhaled in the last breath of steady
breathing condition of 500 cm3 tidal volume and frequency 16 breaths/ inhalation
min, Peclet number is about 100. Thus the concentration gradient L amount of particles inhaled in the nth breath of
occurs mainly in a thin boundary layer near the alveolar wall. For a washin
flat surface suddenly exposed to an aerosol concentration c, the diffu- i grid point in the interval 0 5 z 5 23
sion deposition to the wall per unit area per unit time is J particle flux per unit airway cross-sectional area
L total deposition function, combined particle deposi-
tion to the airway surfaces per unit length of airway
f>O (~4~7) per unit time
Ld deposition function due to Brownian diffusion
To each alveolar inner surface, the deposition rate is s&d . Thus, Li deposition function due to impaction
the deposition rate per unit length of the airways in the alveoli is L deposition function due to sedimentation
2 length of airway per generation
Na total number of alveoli in the lung
n total number of airways per generation
. 1 dz
na number of alveoli per generation
where N, is the total number of alveoli in the lung. Using Q. A6, we P parameter defined in Eq. 32
obtain Q total Aowrate of air passing through an airway cross
section, UA
VaT a3 q parameter defined in Eq. 32
s, = (Am R fractional recovery during steady breathing
4.23 ( N a ) fractional recovery in the Nth or nth breath
RN, Rn
and parameter defined in Eq. 32
itk Stokes’ number, Tu/d
SaN* = 4.23 Nal13 (VaT) 2/3 WQ S area of an alveolar opening
Sa inner surface area of an alveolus
Finally, we may rewrite Eq. A9 as T time scale, Z/u
Td,Ti,Ts time scales for diffusion, impaction and sedimentation
C
Ld = 4.23 - deposition
2 (AM period of a breathing cycle
TO
t time
Sedimentat ion deposition. The deposition rate due to sedimentation in U average air velocity in a generation
an alveolus depends on the relative orientation of the alveolus with velocity dispersion,
U’ difference between the air ve-
respect to gravity. So far as calculation of sedimentation deposition is
locity in an airway tube with the average velocity u
concerned, we assume that an average alveolus has its axis of symmetry
uf7 settling velocity of a particle
perpendicular to the gravity g. Then, in each alveolus, the deposition
VW airway volume between x = 0 and x
rate rate is alveolar volume between x = 0 and x
V,(X)
V aT total alveolar volume in the lung
V alveolar volume per unit length of airway
3-a volume of an alveolus
X airway length
z airway generation number
fraction of airway surfaces which are alveolated
il angle defined in Fig. 8
= 0.69 c gd& 8 spherical coordinate
x characteristic length
The sedimentation deposition rate in the alveoli per unit length of the P air viscosity
airways is found to be kinematic viscosity of the air
i variable, u’/au
Na dF
L, = Qi, - - P air density
2 dz d standard deviation
7 relaxation time of a particle, pdp2/1 8~
dl?
= 1.13 7 Na113(V#3 dz a’s deposition rate in an alveolus due to sedimentation
4 spherical coordinate
#d deposition rate in an alveolus per unit area due to
LIST OF SYMBOLS Brownian diffusion
subscript 0 referring to the condition at the mouth
A total airway cross-sectional area superscript 0 rest condition
a cross-sectional area of an airway tube, ?rd2/4

Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (134.129.182.074) on November 16, 2018.

Copyright © 1975 the American Physiological Society. All rights reserved.
AEROSOL DEPOSITION IN HUMAN RESPIRATORY TRACT 85

The initial thought of this paper was developed when one of us This research was supported by the National Science Foundation
(C. P. Yu) was visiting the Medical Research Council Aerosol Labora- under Grant K.040794.
tory, University of Essex, England, in the Spring of 1972. He is grateful
to Dr. C. N. Davies for a number of conversations regarding this
problem. Received for publication 27 March 1974.

REFERENCES

1. ALTSHWLER, 8. Behavior of airborne particles in the respiratory 8. HOLMAN, J. P. Heat Transfer. New York: McGraw, 1968.
tract. In : Circulatory and Respiratory Mass Transport, edited by G. F. 9. Landahl, H. D. On the removal of airborne droplets by the human
Walstenholme and J. Knight. London : Churchill, 1969, p. 21.5- respiratory tract. I. The lung. Bull. Math. Biophys. 12 : 43-56, 1950.
235. 10. SCHERER, P. W,,L. H. SHENDALMAN, AND N. M. GREEN. Simul-
2. BEECKMANS, J. M. The deposition of aerosols in the respiratory taneous diffusion and convection in single breath lung washout.
tract. Can. J. Physiol. Pharmacol. 43 : 157-l 72, 1965, Bull. Math. Biophys. 34: 393412, 1972.
3. Bell, K. A., and S. K. FrieIander. Aerosol deposition in models 11. SCHLESINGER, R. B., AND M. LIPPMANN. Particle deposition in
of a human lung bifurcation. Paper presented at: International casts of the human upper tracheobronchial tree+ Am. Id ITyg.
Conference on Dust Hacurds and Dust Control, Bonn, Germany, September, Assoc. J. 33 : 237-261, 1972.
26-28, 1972. 12. SCHROTER, R. C., AND Me F* SUDLOW. Flow patterns in models of
4. DAVIES, C. N.,J+ HEYDER, AND M. C. SUBBA RAMU. Breathing if the human airways. Respiration Physiol. 7 : 341-355, 1969.
half-micron aerosols. I. Experimental. J. &$Z. Physiol. 32 : 591- 13. TASK GROUP ON LUNG DYNAMICS. Deposition and retention models
600, 1972. for internal dosime of human respiratory tract. Health Phys. 12 :
5. FXNDEISEN, W. Uber das Absetzen Kleiner in der Luft suspendier- 173-207, 1966.
ten Teilchen in der Menschlichen Lunge bei der Atmung. 14. TAYLOR, G. I. Dispersion of soluble matter in solvent flowing
Pjueger Arch. Ges. Phvsiol. 236 : 367-379, 1935. slowly through a tube. Proc. Roy. SW, London, Ser. A 2 19 : 186-203,
6. HATCH, T. F., AND D. GRUSS. Pulmonary Deposition and Retenfion in 1953.
Inhabd Aerosols. New York : Academic, 1964.
15. TAYLOR, G. L The dispersion of matter in turbulent flow through
7. HEYDER, J.,J. GEBHART, G. HEIGWER, CROTH, AND W. STAHL-
a pipe. Proc. Roy. Sot., London, Ser. A 223 : 446-268, 1954.
HOPEN. Experimental studies of the total deposition of aerosol --_ - .- ---
particles in the human respiratory tract. J. Aerosol Sci. 4 : 19 l-208, 16. W&BYE. R. Morbhometry of Human Lti-KBerlin : Springer-
1973. Verlag, V’ 1963 1

Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (134.129.182.074) on November 16, 2018.

Copyright © 1975 the American Physiological Society. All rights reserved.