1.

The term powder technology refers to the process of mixing and pressing metal particles into a
prearranged shape, in order to set the shape the mould is then sintered. The method of
sintering refers to heating the particles just below the melting point of mixture allowing inter-
diffusion between the particles (Thmmler, Oberacker, and Jenkins 1993, 15). In order to
produce a metal powder an initial stage of size reduction needs to be carried out, this can be
done through several different processes, including: mechanical processes, chemical processes
and physical methods. In order to control quality of a powder sample particle size analysis is
carried out. The importance of powder sampling is to gather information on the characteristics
of a large population from a small fraction sample of the given population. To ensure
worthwhile results it is vital that the sample is a correct representation of the size distribution
of the components within the bulk material. Due to the random nature of sample selection the
many errors associated need to be considered, these can be due to primary methods, sample
division, and analysis errors. In order to reduce the possible errors and to gain a representative
sample make up, it is important that the powder is sampled while in motion and the whole
stream of powder should be taken for short increments of time (Allen, 2003, 23). The purpose
of the following report is to compare and contrast common sampling devices and strategies
from both stores materials and flowing streams.
Stored material sampling is taken from a powder at rest that has assumed to be previously
mixed and therefore exists as a homogeneous mixture, this type of sampling is however
considered to be not as accurate or efficient compared to flow stream sampling. Stored
material can be sampled in several different ways, depending on the storage container and
previous history of the material.
The first method of stored powder sampling is referred to surface sampling. Surface sampling
involves scooping into the powder to collect a sample. Surface sampling does however have
downfalls; since the sample is taken predominantly from one section it cannot be considered a
representative of the bulk material. The average standard deviation for the reliability of sample
taken via the surface sampling method is 5.14% (Particle Science 2013). In order to increase the
accuracy of surface sampling several samples can be taken and analysed independently, the
container can be shaken prior to sample being taken and using a flat-blade sample scoop to
ensure no loss of coarse or fine powder.
A second stored method involved a long tubular device that will collect the sample in one or
more open chambers. The method is referred to as thieving and the device used a sampling
thief. Thieving gives more valuable results compared to surface sampling; however it can result
in the disadvantages of fracturing particles, a tendency to jam, lack of flow into the tube and
incorrect distribution of fines.
Cone and quartering is an operator dependent process, the powder must be poured onto a flat
surface, flattened and divided into one quarter of the size with a cross-shaped cutter device.
Coning and quartering can be repeated until the desired sample size it reached (Particle
Sciences 2013). The process of coning and quartering will only give reliable results if the powder
heap was symmetrical, this however will rarely occur in practice and each quarter will represent
a different section of the sample. As a sampling method the relative reliability of the coning and
quartering method will involve a 6.81% standard deviation from the mean.
A second type of sample collection can be taken while the material is in motion. According to
the rules of sampling, flowing stream sampling is the recommended method and will produce a
more valuable representation of the bulk population of material.
The method of conveyor sampling can be carried out by a large scoop apparatus with the
opening a similar size to belt width. Samples may also be collected from flowing streams
embedding a tube within the sample stream with an open cavity that will gradually fill. This
type of sampling can only be applied to known homogeneous streams as it does not collect the
whole of the sample (Allen 2003, 32).
The process of table sampling can be carried out by feeding material onto an inclined plane,
some of the powder will then fall through holes in the plane over several removal stages. The
powder that reaches the end of the inclined plane is collected as the sample. Table sampling
can result in several errors due to initial feed not displaying uniform characteristics and no
mixing existing at stages. As a sampling method the reliability deviation of table sampling is
approximately 2.09% the highest error of all flowing sample methods (Particle Science 2013).
A chute splitter is a V-shaped channel in which one chute will feed to two trays at the bottom.
Powder can be fed into the V-slot and it will be repeatedly halved until a desired quantity is
reached (Allen 2003, 33). Chute splitting is a relatively good method for sampling; however it is
prone to operator error and has a reliability deviation of 1.01% (Particle Science, 2013) from the
mean.
A spinning riffler is the most efficient of all sampling devices; a constant flow of powder is fed
into a divider rotating at a constant speed. This allows an even divide of powder between each
chamber and the desired quantity can be combined to use as the sample specimen (Particle
Sciences 2013). As the most efficient method it will only encounter a reliability deviation of
0.125%.
By taking into account the situation of an in-situ or flow process that needs to be sampled and
by evaluating the types of error involved, correct selection of a sampling method with
associated devices can be achieved.
2.
In order to gain insight into the properties of the bulk population sampling of a small fraction
must be carried out. From the sample a particle size distribution can be calculated, this
distribution is necessary as particle size can influence several properties of the material and
indicates the operating performance (Horiba Scientific 2013, 1). A particle size distribution can
be calculated through several methods and correlations. Due to the number of correlations
available for particle size distributions they may only be applicable to specific situations, a
factor that needs to be considered before calculation. The following report will compare
common particle size distribution functions, including the applicability and corresponding
parameters.




The Gates, Gaudin and Schuhmann (GGS) correlation can be used if Y versus X graph shows a
straight-line trend, it will only be applicable if the X value is the X-intercept at Y = 1 and the
slope m is equal to the mean log value for Y and X. The GGS function will usually fit best for
the middle section of a size distribution from 2 50%. The cumulative plot tends to smooth
data variations, while the Gaudin plot will emphasize the variations (Fuerstenau and Han 2003,
24), this is useful to detect sources of error within a given distribution.






Gates, Gaudin, Schuhmann =

K is related to coarsest size
m is the measures spread in distribution
Y is cumulative weight fraction finer than X








The Rosin-Rammler correlation can be used if the value of: ln(1/[1-Y]) is linear. The value of n
will be the slope of the line according to (Fuerstenau and Han 2003, 123):
(ln(ln[1/(1 Y
i
)] ln(ln[1/(1 Y
j
)])/ln(X
i
/X
j
).
The value of R will be calculated by:
ln[1/(1 Y)] at X = 1.
The Rosin-Rammler particle size distribution is applicable for a range of powder types and sizes;
it is generally used to estimate processes involving grinding, milling and crushing. The RR
particle size distribution will usually be based on 0.5 80% of the particle size distribution. The
distribution can be graphed on a log-log scale by plotting the cumulative percentage finer than
the particle size, against particle size (Allen 1996, 53). The Rosin-Rammler generally fits
distributions better for coarse particle sizes










Rosin-Rammler = 1 exp

P refers to coarsest size
n measures spread in distribution

Figure 1: Gates, Gaudin, Schuamann distribution (Fuerstenau and Han 2003, 23)









The Gaudin and Meloy equation can be used for particle distribution if (Fuerstenau and Han
2003, 123):
(1 Y) versus (1 X/X
0
) is a straight line.
The value of r will be the slope and it must fit:
[ln(1 Y
i
)/(1 Y
j
)]/[ln(1 X
i
/X
0
)/(1 X
j
/X
0
)]







The Log Normal correlation created by Herbst and Sepulveda, the plot of X versus Y will be
symmetrical around the geometrical mean on a logarithmic scale (Fuerstenau and Han 2003,
123). Log normal will often hold for very board particle size distributions where small deviations
do not occur. Common industrial applications are for natural powders or powders produced
from grinding and milling.



Gaudin and Meloy = 1 1

X
0
is coarsest size
r is degree of fragmentation

Log Normal (Herbst and Sepulveda) =
exp[log2
2
]

2log

u is geometric mean size
t is standard deviation


Figure 2: Rosin-Rammler distribution (Fuerstenau and Han 2003, 25)
Figure 3: Log-normal distribution (Fuerstenau and Han, 26)
3.
Due to the random nature of crushing and grinding often particle size will not be produced in a
perfect spherical shaped. Due to the irregular shape, correlations for non-spherical particle
distributions need to be considered. These correlations are produced experimentally and used
to input data in order to fit to the trends. An equivalent diameter will be defined by the relation
to a specific sizing method that was developed in order to find equivalency.
Martins diameter (M) refers to the average length of a constant line drawn in any direction
which bisects the particle image (Henk 2009, 15)
Ferets diameter (F) is the distance between two parallel tangential lines that are drawn in a
fixed direction on opposite sides of a particle (Henk 2009, 15).
The projected area diameter (d
a
or d
p
) is found but drawing a circle around the edges of a
particle at rest plane to the surface, the diameter of the circle is then measured (Yang 2003, 2).

2









The equivalent sieve diameter refers to the diameter of the minimum square opening through
which particles will pass (Allen 1996, 5).



Figure 4: Equivalent Diameters (Allen 1996, 5)

The surface diameter (d
s
) refers to the diameter of a sphere that has an equivalent surface area
of the specific particle (Allen 1996, 6). The formula for surface diameter is expressed by:

(Allen 1996, 6)
A volume diameter (d
v
) is calculated by measuring the diameter of a sphere that has been
produced to have the same volume of the specific particle (Allen 1996, 6). The formula for
volume diameter is expressed by:

= (

(Allen 1996,6)
Sauters diameter is known as the surface-volume diameter (Allen 1996, 6), it can be found by
creating a sphere with the same surface area to volume ratio of the particle and measuring the
relating diameter. The formula to express Sauters diameter is expressed by:

2
=

2

(Allen 1996, 6)
The Stokes diameter (D
s
) refers to the equivalent diameter corresponding to a sphere with the
same settling velocity as the particle (Yang 2003, 6) when under the influence of laminar flow in
a fluid with constant physical properties. This equivalent diameter is formulated by the Stokes
relation: where = viscosity of the liquid medium,
s
is the density of the solid,
L
is the density
of the pure liquid, g is gravity and v is final settling velocity.

=
1

By employing an equivalent diameter for a non-spherical shaped particle it allows the analysis
of particle size distributions for powders that may not have been analysed before.

References:
Upadhyaya, G. S. 1996. Powder Metal Technology. Cambridge: Cambridge International Science
Publishing
Thmmler, F. Oberacker, R. Jenkins, I. 1993. Introduction to powder Metallurgy. London: The
Institute of Material, Minerals and Mining.
Allen, T. 2003. Powder Sampling and Particle Size Determination. Elsevier.
Particle Sciences: Drug Development Services. 2013. Sampling of Powders. Accessed March
17, http://www.particlesciences.com/news/technical-briefs/2011/sampling-of-powders.html
Horiba Scientific. 2012. A Guide to Particle Size and Analysis. California: HORBIA Instruments,
Inc.
Fuerstenau, M. Han, K. 2003. Principles of Mineral Processing. Society for Mining, Metallurgy,
and Exploration (SME)
Henk, M. 2009. Particle Size Measurement: Fundamentals, Size, Quality. Springer.
Allen, T. 1996. Particle Size Measurement: Volume 1: Powder Sampling and Particle Size
Measurement. Springer.
Yang, W.C. 2003. Handbook of Fluidization and Fluid-Particle Systems. New York: Marcel
Dekker, Inc.