Sunteți pe pagina 1din 12

Abstract: This paper provides a rigorous analysis in the accuracy of Stokes for calculating particle-size

distributions of nonspherical equation for calculating particle-size to the equivalent Stokes diameter are
presented for disk-shaped particles. Disks and rods are chosen to present the high aspect ratios typizally
exhibited by real clay particles. For Particle sizes ranging from 0.1. m to 100 . m and for aspect ratios
ranging from 10 to 500, it is shown than Stokes equation underestimates the maximum particle dimension by
up to two orders of magnitude. Consequently, particle-size distributions calculated from conventional
hydrometer analysis are shown to be misleading. To confirm the conception that Stokes equation may not be
appropriate for calculating particle sizes and particle-size distributions for fine-grained clays, Kaolinite and
pulverized mica are characterized by comparing results from hydrometer analysis, sieve analysis, laser
diffraction analysis, and scanning electron microscopy (SEM). The experimental results confirm the error
estimates from the theoretical evaluation.

KEYWORDS: Sedimentation test, hydrometer analysis, fine-grained clays, scanning


electron microscopy, laser diffraction, particle sizing.

The Standards Test Method for Particle-Size Analysis (ASTM D 422-63) relies on the well-
known Stokes equation for estimating particle diameters (ASTM 1990a; AASHTO 1983).
The validity of Stokes equation for fine-grained particle size analysis is based on
assumptions including:
1. Particle-to-particle interference and boundary effects from the walls of the
sedimentation column are negligible
2. Particle sizes are small enough to ensure that the induced fluid flow is well within
the laminar flow regime.
3. Actual particle shapes can be approximated by smooth spheres.
For all practical purposes, the first two assumptions are satisfied. The first is validated by
limiting the maximum concentration of soil in the suspension. It has been shown that if no
more than 50g of dry soil are used in 1000 cc of suspension, the effects of interference are
negligible (Weartherly 1929).
The second assumption is validated by prescribing an upper limit to the measurable particle
size range. For particle sizes in the micrometer range, it has been shown that the laminar
flow assumption is mostly true (e.g., Mysels 959; Clift et al. 1978). The upper limit of
Reynols number for the valid application of Stokes equation is shown in the literature to be
between 0.1 and 1.0 (Allen 1990; Bernhardt 1994; Bardet 1997). These values correspond
to free-falling spherical particles up to 2 mm in diameter (Mysels 1959). For fine particles,
the terminal settling velocities are very slow and the Reynolds number is much smaller than
0.1. To ensure that the upper particle size range is controlled, samples are presieved using
mechanical methods prior to the sedimentation analysis. Generally, material passing the
No. 200 (75 . m) sieve is analyzed by the hydrometer method.
The validity of the spherical shape assumption and the consequent impact on the accuracy
of standard hydrometer analysis for nonspherical fine-grained soils are the subjects of this
paper. The sphericity assumption has been the subject of much research in the past. Kunkel
(1948) examined the magnitude of errors produced by shape factors in Stokes Equation
and found that size estimation errors resulting from nonspherical shapes could be up to 50%
for particles 2 mm in size with an aspect ratio of 2.0. Parslow and Jennings (1986)
conducted simultaneous particle-sizing tests using electric birefringence and X-ray
absorption to investigate nonspherical of micrometer scale clay particles. They found that
difference in calculations of major dimensions for spherical and nonspherical particles
could reach 150%. Recently, Nettleship et al. (1997) showed that various measurement
methods yield significant differences in particle-size distributions for identical clay
minerals. More importantly, they concluded that the standard hydrometer analysis should
not be recommended for submicron minerals. Vitton and Dadler (1997) compared particle
size result obtained from hydrometer and laser-diffraction based measurements for eleven
soils. They found that, in general, the laser measurement indicated a lower percentage of
fines than the hydrometer and partially attribute the discrepancies to the nonespherical
shape of the particles. Similarly, Jonert and Vandenberghe (1997) compared results
obtained from sedimentation based (StokesLaw) pipette analysis and laser-diffraction were
comparable for blocky quartz particles but were strikingly different for platy clay particles.
Other indirect studies using light scattering techniques including optical light (Stamm and
Svedberg 1925) and lasers (Singer et al. 1988; Hildebrand and Row 1995) also support the
contention that sedimentation test based on Stokes equation may seriously underestimate
particles sizes. While previous studies have identified the underestimation of particles sizes
and size distributions, a quantitative magnitude of the error caused by using Stokes
equation for fine-grained clays has yet to be determined. This study attempts to provide a
rigorous theoretical treatment with supporting evidence from particle size experiments to
answer this question.

General consideration

When a particle moves in a fluid whit a velocity U parallel to the principal axis i, the
viscous drag is given by

= = , , (1)
Where stands for fluid viscosity; and the three components of 1 are termed the principal
translational drag resistances (Clift et al. 1978). Usually, the particle reaches a constant
terminal velocity after a short period of turbulent free fall. At steady state, the particle
movement is best characterized by the dimensionless Reynodls number, , defined as

= (2)
Where denotes the density of the fluid and L is a characteristic length. Since clay
particles are typically at micro and sub micrometer scales, settling velocities are usually
very slow so that the Reynolds number is much smaller than 0.1. This condition secures
that the particle motion is in the laminar flow regime and a constant settling velocity is
achieved shortly after a free fall. At steady state, the drag force induced by the terminal
settling velocity is balanced by the gravitational and the buoyancy forces:

=
+
(3)

Terminal Settling Velocity for Spherical Particles

In 1851, Stokes published a seminal paper describing the settling behavior of a spherical
particle within the laminar flow regime. The translational drag resistance for spherical
particles was expressed by Stokes in an amazingly simple analytical from as:
= 6 (4)
Where stands for the radius of the sphere. By balancing the viscous drag, the
gravitational force, and the buoyancy force (i.e. Eqs 1,3 and 4, Strokes arrived at the
following analytical equation relating the particles settling velocity U, diameter (= 2 ),
fluid density , particle densidy , and fluid viscosity as:
2 ( ) ( )
= 9 ( )2 = 2 (5)
18
This is the well-known Stokes equations used in the Standard Test Method for Particle-
Size Analysis of Soils (ASTM D 422-63 1990a) as the theoretical basis for calculating fine-
grained particle size distributions. Because it is simple, economical, and reliable,
hydrometer analysis is the only standard ASTM geotechnical laboratory test method for
sizing fine-grained soils. For ideal soil particles of spherical shape, the method is valid for
particle diameters ranging from 0.2 . m to 200 . m. Beyond this limits, Stokes equation
is invalid either because turbulent flow may occur for larger particles or because Brownian
motion may be significant for smaller particles.

Terminal Settling Velocity for Ellipsoidal Particles

For nonspherical particles, the settling behavior is far from that for an idealized sphere.
Because the viscous drag force may not be in the same direction of gravitational and
buoyancy forces and many not act through the centroid of the particle, complicated motions
such as rotation and nonvertical translation will occur, analyzing the motions of generally
irregular shapes of particles is far from straightforward. However, insight can be gained by
approximating these irregular particles as spheroids, or ellipsoids. For example the platelet
shape of Kaolinite may be approximated by a disk shaped ellipsoid. The tubular shape of
halloysite may de approximated by a rod-Shaped ellipsoid. The mathematical formula for
the shape of an ellipsoid is:

2 2 2
( ) + ( ) + ( ) = 1 (6)

Where a, b and c are the principal axes. Disk shapes and rod shapes are chosen to present
extreme cases of an ellipsoid and are illustrated in Fig. 1 For a disk-shaped particle (Fig.
1a), the two longer axes a and c are equal and the particle is rotated about the short axis b to
give a full three-dimensional configuration. For a rod-shaped particle (Fig. 1.b), the two
shorter axes b and c are equal and the particle is rotated about the long axis a to yield a full
three-dimension configuration. Both shapes can be characterized by an aspect ratio E
defined as b/a for disk-shaped particles and a/b for rod-shaped particles. Therefore, for
fine-grained soils approximated by disk-shaped ellipsoid, the aspedt ratio E is less than 1.
For fine-grained soils approximated by rod-shaped ellipsoid, the aspect ratio E is greater
than 1.

The hydrometer of ellipsoidal particles is a classical subject in fluid mechanics (Oberbeck


1876; Gans 1911). Characteristics of particle motion and mathematical solution of the
earlier work have been abstracted in some handy forms (Happel and Brenner 1973; Clift et
al. 1978). Some relevant conclusions concerning the terminal velocity for small sllipsoidal
particles are summarized as follows:

(1) Due to the symmetry of an ellipsoid, there is no torque component of drag, thus
no particle rotation.
(2) At terminal velocity, there is no preferred orientation for particle settling and
particles will fall at a constant local inclination angle.
(3) In general, the terminal velocity vector will not be parallel to gravity but will act a
certain angle from vertical.

Based on the above conclusions, the total drag force acting on a free-falling ellipsoid can be
obtained by adding individual drag components due to the velocity acting parallel to each
of the three orthogonal axes (unit vector ei ) of the ellipsoid:

FD = [c1 U1
e1 + c2 U2 e3 ]
e2 + c3 U3 (7)

If we are to consider ellipsoidal particles as approximations for fine-grained soils free


falling in the hydrometer test, three fundamental questions must be addressed:

(1) For an individual ellipsoid, how different are the settling velocities between
vertical and horizontal orientations?
(2) For and assemblage of ellipsoids with identical dimensions and densities, what is
the mean settling velocity if the particles are randomly oriented in a suspension?
(3) For an ellipsoidal particle and spherical particle having the same settling velocity,
how different are the major particle dimension?

Insight into the first question can be gained by investigating the literature. It was illustrated
quantitatively by Bernhardt (1994) that the classical solutions (Oberbeck 1876; Gans 1911)
lead to some interesting and useful conclusion on the terminal velocity for individual
ellipsoidal particles. In general, a particles terminal velocity depends on the particles
settling orientation. For extremely thin disk-shaped particles, the velocity ratio between
horizontal settling (shown in Fig 1a) and vertical settling in 2:3. In other words, a particle
settling vertically will fall 50% faster than one that is settling horizontally. For extremely
thin rod-shaped particles, the velocity ratio between horizontal settling and vertical settling
(Shown in Fig. 1b) is 1:2. The velocity ratios for ellipsoidal particles of finite sizes will be
smaller than these limiting cases, implying less difference in velocity among different
particle orientations. The answers to the second and third fundamental questions are far less
apparent and are addressed as follows.

Because particles in the sedimentation cylinder are well dispersed and mixed at the start of
the hydrometer analysis, a random particle orientation can be assumed. Although particles
with different orientations will fall atdifferent velocities, the mean drag resistance for a
random assemblage of particles with identical dimensions can be expressed as (Barr 1931:
Rouse 1946; Payne and Pell 1960):
3
= 1 2 3 (8)
1 +2 +3

The mean direction of drag is parallel but apposite to the gravitation field. Given a mean
drag resistance, terminal velocities for disk-shaped and rod-shaped particles can be derived.

Mean Terminal Velocity for Disk-Shaped Particles

In the case of disk-shaped particles, the two longer axes are equal, and c2 = c3 in the
direction perpendicular to the vertical axis shown in Fig. 1a. Then mean drag resistance af
disk-shaped particles is given as (Barr 1931; Rouse 1946; Payne and Pell 1960);

3 6a1E2
ca = c 1 +2c 2 = (9)
1 2 cos1 E

Where E was previously defined as the aspect b/a. A drag ratio may be defined as the ratio
of the drag resistance f an allipsoid to that of a sphere with the same equatotial radius a:

ca

a = 6a (10)

By balancing the drag force with the gravitational and buoyancy forces, we obtain:
2 ab (s f )g
U = 9 (11)
a

To compare the Strokes dameter with the ellipsoids dimensions, we define the concept of
a hydraulic equivalent sphere. The equivalent sphere and the ellipsoid will have identical
settling velocities and particle dessity. By comparing Eqs 5 and 11 and applying this
hydraulic equivalent condition, the radius of the hydraulic equivalent sphere is given by

E
a = a (12)
a
The above equation is used to analyze the influence of the aspect ratio on the accuracy of
the Strokes diameter calculation for disk-shapes particles. Consider a measured terminal
settling velocity for an assemblage of nonspherical clay particles with the same shapes ans
dimensions but randomly oriented in a suspension. They equivalent spherical radius a of
these particles can be calculated from the hydrometer test using Stokes equation. If the
actual particles are disk-shaped with an aspect ratio E, the relationship between the
normalized radius (Defined as the ratio of the Strokes diameter to the disk diameter,a /a)
and the aspect ratio E can be described analytically in question have an aspect ratio 0.1, the
radius ratio (a /a) will be 0.38 and the major axis will be 2.6 times longer than the radius
of the corresponding hydraulic equivalent spheres. For a well crystallized kaolinite with a
typical aspect ratio of 0.001 the major axis will be eight times longer than that predicted by
hydrometer analysis.

The impact of nonspherical on the terminal velocity can be directly appreciated by


rearranging Eq 11 usingb = aE:

2 2 ( )
= 9
(13)

The relationship among the terminal velocity, aspect ratio, and the disk radius is illustrated
in Fig. 3a. Here, the terminal velocity for particles in water at 20C is considered as a
function of aspect ratio for a series of fixed values of the major axis radius. For example,
for a spherical particle (E =1) with a diameter of 100m (a = 50m), the terminal velocity
is about 1cm/s. however, for a disk-shaped particle with a disk diameter of 100m (a =
50m) and an aspect ratio of 0.01, the terminal velocity drops to about 0.02 cm/s, a
reduction of nearly two orders of magnitude. The same reduction trend applies to particles
with different dimensions. It can be concluded that a particles settling velocity is quite
sensitive to the major dimension (Diameter of disks) and aspect ratio.

Mean Terminal Velocity for Rod-shaped Particles


For rod-shaped particles (See Fig. 1b), such as certain forms of the clay mineral halloysite,
the two shorter axes a and c are equal and the aspect ratio is defined as E = a/b > 1. The
axial and normal drag resistance, denoted by c1 and c2 , are given in Clift et al. (1978) and
the mean resistance for randomly oriented particles is expressed as:

3 6bE2 1
cr = c 1 1 = (14)
1 +2c2 ln(E+E2 1)

Since the volume of the Rod-shaped ellipsoid particles is given by Vn = 4a2 b/3, the
terminal velocity n the laminar flow regine is:
2 ab E(s f )g
U= (15)
9 b

The drag ratio is defined as:

b =

cr
(16)
6a

By comparing Eq 15, the Stokes Radius a , of the hydraulic equivalent sphere rod-shaped
particles is given by:

E a
a = b = (17)
b b
E

The above equation defines the relationship among


the Strokes radiusa , half-length of rod-shaped
particles, aspect ratio, and drag ratio, as shown in
Fig. 4. By comparing Fig. 4 with Fig. 2, it is
observed that the dependence of the normalized
radius ratio on the aspect ratio for rod-shaped clay
particles is much higher than that for disk-shaped
particles. For example, at the same terminal velocity,
a particle with an aspect ratio of 10 (Atypical value
for Halloysite) is four times longer than predicted
spherical diameter.

As shown in Fig. 3b, a rod-shaped particles terminal


velocity is very sensitive to the aspect ratio. A ten-
fold change n the aspect ratio (i.e., from E=1 to 10)
will result in a two-orders of magnitude change in the
terminal velocity.
Inaccuracy of Stokes Equation for Nonspherical Clay Particle Size Distribution

So far we have demonstrated that, for identical sizes of disk-shaped and rod-shaped
particles, the Strokes diameter gives ambiguous size estimations. It is now imperative to
determine how thw nonspherycity affects the interpretation of particle size distributions.
This can be illustrated by examining idealized particle size distributions for disk and rod
shapes as follows.

Size Distribution Transformation Function for Disk-shaped Particles

Suppose we obtain a poorly graded particle size distribution of a fine-drained soil from a
conventional hydrometer analysis as shown by the solid line in Fig. 5a. The mean particle
size, calculated from Strokes equation, is D = 2a = 10m. If the soil consists of well-
crystallized disk shaped kaolinite with a constant thickness b of about 2 m, we can
determine the disk diameter distribution by deriving a simple analytical relationship
between the Stokes diameter D and the disks diameter 2a. For disk-shaped particles, Eq
12 is rewritten as

cos1 E
D = 2a = 2b
E1E2

(18)

The above equation implies that once we know the half-minor axis f of disk-Shaped clay
particles, we can directly relate the aspect ratio E to the Stokes diameter D using the
following approximate relation:
D2
2a = b
(19)

It can be that the error created by using Eq 19 to approximate Eq 18 is less than 5% for
aspect ratios ranging from 0.001 to 0.1 for a disk thickness b = 1 m and. Therefore, the
transformation of the spherical particle size distribution (Solid line in Fig 5a) to the disk-
Shaped particle size distribution can be accomplished using Eq 19 (Dashed line for b =
1 m and dotted line for b = 2 m in Fig. 5a)

Several observations on the effect of nonsphericity on the particle size distribution can be
made. The mean diameter distribution changes from 10 m for spherical particles to
16 m for disk-shaped particles with thickness of 2 m, and to 32 m for disk-shaped
particles with a thickness of 1 m (Fig. 5a). For thinner particles such as kaolinite where
the thickness can be small as 0.01 m, the increase in mean disk diameter will be much
greater.

Size Distribution Transformation Function for Rod-Shaped Particles

The size distribution transformation function for rod-shaped particles can be arrived by
rearranging Eq 17:

E ln(+ 2 1)
= 2a = 2b = 2b (20)
b 2 1

By conducing series expansions, the above equation can be approximated as:

3 1.1664
2 = 2b exp [2 ( 2) ] (21)

For particle aspect ratios 1 to 1000, the above equation approximates the particle diameter
(Eq 20) with a maximum error less than 5%. Transformation of a measured particle size
distribution from hydrometer analysis (Shown as solid line in Fig. 5b) to a known rod
diameter can be readily accomplished using Eq 21(Shown dashed and dotted lines in Fig
5b). To illustrate the effect of aspect ratio on particle size distribution, rod diameter 2b is
given as a constant for 3 m(Dash lines) and 5 m (Dot lines), respectively.

Whereas the changing patterns in particle distributions are similar to that of disk-shaped
particles, the aspect ratio of rod shapes has a more radical impact on particle size
distributions. For example, particle 3 m in diameter(compared to 10m forfor shperes)
will increase the mean size by two orders of magnitude, whereas particles 5 min diameter
remain about the same mena size (Fig. 5b).

Experimental Evidence

Two series of experiments were conducted to demonstrate, qualitatively and quantitatively,


the errors introduces into hydrometer analysis by the spherical shape assumption. Materials
that exhibit disk-shaped structures (Kaolinite and pulverized mica) were chosen for analysis
by various methods for comparison. These methods include standard hydrometer analysis,
sieve analysis, laser-diffraction analysis, and scanning electron microscopy (SEM)

Experimental Confirmation for Georgia Kaolinite

To indirectly demonstrate discrepancies that may be attributable to the nonspherical shape


or natural clay minerals, particle size distributions for kaolinite obtained from conventional
hydrometer analyses were compared to those obtained from laser diffraction analysis. For
the hydrometer analysis, standard testing procedures according to ASTM d 422 were
followed. For the laser diffraction based measurement, a Malvern Mastersizer with a
detection limit of 0.5 m and an analytical range from 0.5 m to 2000 m was used. The
Malvern device employs a flow-through cell where suspended particles are continuously
streamed in front of a focused laser beam. Based on the forward-scattered diffraction
patterns of the laser light, direct estimates of individual cross-sectional particle diameters
are obtained (Singer et al. 1988). The use of laser diffraction for fine-particle size
measurement is reported extensively in the literature (Vitton and Sadler 1997; Singer et al.
1998; McCave et al. 1986). Detailed principles controlling the laser-based measurements
are described elsewhere by Swithenbank et al. (1977).

Figure 6 shows a particle size comparison of the laser and hydrometer methods for a
powdered sample of standard Georgia Kaolinite. Sample preparation for both methods was
identical. The sample was air-dried and dispersed prior to testing by soaking for 24h and
mixing for 1 min in a 4% solution of solution of sodium hexametaphosphate. Hygroscopic
moisture contents of the air-dried samples were on the order of 0.6%. On the figure, shape
effects arising from the platy kaolinite particles are evidenced by an underestimation of
particle size from the hydrometer analysis. The median values (50% finer) for the
hydrometer and laser measurement are 0.55 m and 2.5 m, respectively. If we accept that
the laser measurement reflects are average major dimension of the kaolinite (e.g., Singer et
al. 1988), the hydrometer analysis underestimates the median particle size by a factor of
4.5. Although the particle size range is widely distributed, concurrent SEM analysis form a
limited representative elemental volume confirms that the kaolinite particles are highly
nonspherical and that the actual average major dimension ranges from one three m.

Experimental Confirmation for Pulverized Mica

To quantitatively demonstrate discrepancies that may be attributable to the nonspherical


shape of natural materials, particle size distribution for finely ground muscovite were
obtained. Conventional hydrometer analyses were compared to particle size distributions
obtained from mechanical sieve analysis and to the theoretical evaluation for disk-shaped
particles developed earlier in this paper.

Muscovite, or mica, is a widespread and commonly occurring rick-forming mineral. It is


generally present as foliated sheets owing to its perfect cleavage along the {011}
crystallographic axis. It has a specific gravity ranging from 2.76 to 2.88 (Klein and Hurlbut
a977), Mica was chosen for this experimental effort because, unlike kaolinite, it allows
much stricter control over particle size, shape, and degree of dispersion.

When pulverized by mortar and pestle, mica sheets break down to distinct platy flakes of
relatively uniform thickness. Figure 7a shows a SEM micrograph of pulverized mica which
was mechanically sieved and wet-washed to retain the portion finer than a No. 200 sieve
(75 m) but coaster than a No 325 sieve (43 m). Note from the figure that the flakes are of
uniform thickness and are generally within the prescribed size range. Some exceptions to
the upper limit are evident where long narrow flakes passed through the No. 200 sieve
opening along their long axis.

Hydrometer analyses were performed on the me mechanically sieved (43 m to 75 m)


portion of ground mica following standard procedures (ASTM D 42). Figure 7b shows the
particle size distribution from the hydrometer analysis, sieve analysis, and theoretical
calculation. The actual size range controlled by the sieving effort is shown as the shaded
area from 43 mto 75 m. Note that the hydrometer analysis clearly underestimates the
particle size. According to the hydrometer, 100% of the material (by extrapolation) lies
below the actual lower limit controlled by the No. 325 sieve (43 m). The median size
approximated by the hydrometer is 26 m. If we assume that the median size from the
mechanical analysis occurs between the upper and lower limits at 59 m, the hydrometer
underestimate the median particle size by a factor of 2.3. Recall also that the actual median
size is probably skewed towards the upper limit of the range owing to the elongated shape
of a certain portion of particles. As such, the underestimation by the hydrometer is likely to
be even greater.

Figure 7b includes theoretical estimations of size distributions using the Stokes diameter
from the hydrometer data and the transfer function for disk-shaped particles developed in
the theoretical portion of this paper (Eq). Theoretical distributions are shown for particle
thickness (b) values of 2 mand 5 m. It can be drawn from comparison that the range of
the theoretical estimate overlaps with the sieve analysis quite well, except in the area of the
upper limit. This discrepancy may be attributable to the previously mentioned error in the
mechanical control to the upper limit and to the fact a constant thickness is assumed in the
theoretical estimation, In actuality, the thickness may vary with the major dimension. The
difference between the hydrometer analysis and the sieve analysis clearly demonstrates the
error in the hydrometer analysis.

Summary and conclusions

Understanding the engineering properties of soil often demands detailed information on


particle sizes and shapes, particularly for fine-grained clays. Studies on clays have shown
that Atterberg limits, microscopic physico-chemical forces, shear strength, interparticle
friction angle, and permeability all depend on clay particle sizes and shapes. Studies also
show that clay particles are far from spherical; aspect rations of tens to hundreds are
common. However, hydrometer analysis (ASTM D 422) is the only standardized test for
the sizing of fine-grained soils. The basis for obtaining particle size distributions is the
assumption that all particles are spheres.

This study contemplates this assumption by conducting a rigorous theoretical treatment and
comparing it with experimental data obtained for natural materials. Analysis expressions
relating the nonspherical particle dimension and the Stokes diameter are established for
hydrometer analysis test. Analytical functions for transferring the Stokes diameter
distribution to size distribution for two nonspherical shapes (Disk and rods) are presented.
Our analysis suggest that for the particle sizes ranging from 0.1 m to 100 mand for
aspect ratios ranging from 10 to 500, Stokes equation underestimates the maximum
particle dimension by up to two orders of magnitude. Experimental results using various
techniques confirmed the theoretical estimates of the errors inherent in the hydrometer
analysis. Should hydrometer analysis be used to quantify the particle size distribution if
fine-grained soils, other techniques such as SEM should be used as a basis for using the
theoretical work presented in this paper in order to better estimate particle size distribution.

Acknowledgments

The work described in this paper is supported by National Science Foundation grant
number CSM- 9713442.

S-ar putea să vă placă și