Sie sind auf Seite 1von 84

1

1.POWDER METALLURGY
1.1. POWDER FABRICATION
1.1.1 MECHANICAL FABRICATION TECHNIQUE
There are four mechanisms for reducing a material into powder mechanical
combinations: impact, attrition, shear and compression. Impaction involves the rapid.
Instantaneous delivery of a blow to a material, causing cracks and resulting in size
reduction. Attritioning applies to the reduction in particle size by a rubbing motion.
Shear is a cleavage type of fracture associated with operations like crushing. Powders
formed by shearing are coarse and not often found in powder metallurgy unless the
material is extremely hard. Finally, comminution can be by compressive forces: it the
material is sufficiently brittle it will not deform. But break into a coarse powder. The
formation of metal powders by mechanical techniques generally relies on various
combinations of these four basic mechanisms. The following subdivisions demonstrate
how these fundamental comminution techniques are manifested with respect to metal
powders.
1.1.1.1 Machining
Coarse powder with irregular shape results from the shear associated with the
machining of wrought metal. Because of the large amount of machining scrap
produced in metalworking operations, machining chips are an abundant source of
powder. This scrap can be further reduced in size by grinding. Machining is not a first
choice approach to powder fabrication, and by itself proves inefficient and slow.
1.1.1.2. Milling
Milling by mechanical impaction using hard balls is a classic approach to
fabricating powders from brittle materials. A jar mill such as diagramed in Figure1.1.1
uses a ceramic lined cylindrical jar filled with balls and the material to be milled. As
the jar rolls on its side, the balls continuously impact on the material, crushing it into
2
powder. Milling is not useful for most metals because of their ductility, cold welding
and low process efficiency. Brittle materials are more responsive.
Figure1.1.1 A view of the action in a jar mill.
The jar is rotated on its side and the impact of
the falling balls leads to grinding of the
material into a powder.
1.1.2. ELECTROLYTIC FABRICATION TECHNIQUES
A powder can be precipitated at the cathode of an electrolytic cell under certain
operating conditions. Common examples of metals formed into high purity powders by
such an approach include titanium, palladium, copper, iron, and beryllium. The anode
and cathode reactions corresponding to copper and iron are shown in this Figure1.1.2.
The cathode deposit is removed and cleaned by washing and drying. Subsequently the
cathode cake is ground into fine powder and drying. Subsequently the cathode
cake is ground into fine powder and annealed to remove any strain hardening.

Figure1.1.2 The formation of metal powder
from an electrolytic cell. Material is dissolved
at the anode and deposited at the cattode
(examples of reactions are shown for copper
and iron)
1.1.3. CHEMICAL FABRICATION TECHNIQUES
Almost all metals can be fabricated into powder by a chemical technique.
Typically the particle size and shape can be adjusted over a wide range by control of
the reaction variables. There are several variants to the chemical synthesis approach;
powders can be formed by gas-solid, liquid, or vapor phase reactions.
1.1.31. Decomposition Of A Solid By A Gas
3
The classic form of metal powder fabrication is oxide reduction. The process
starts with a purified oxide such as magnetically separated iron oxide (magnetite).
Such oxides are easily milled into fine powders. Oxide reduction is achieved by
thermo chemical reactions involving reducing gases such as carbon monoxide or
hydrogen.
Feo(s) + H2(g) Fe(s) + H2O(g)
Thus, for FeO reduction by hydrogen, as long as the moisture is removed from
the reaction, the reaction can go to completion.
1.1.3.2. Thermal Decomposition
Powder particles can be fabricated by the combination of vapor decomposition
and condensation.
1.1.3.2. Precipitation from a liquid
A dissolved metal salt such as a nitrate, chloride or sulfate can be treated to
produce either a metallic precipitate or a metal containing precipitate. Precipitates
involving metallic salts are an easy means of producing powder. A soluble salt is
dissolved in water and precipitated by a second compound.
The precipitation techniques are well suited to forming composite powders. In
this case, one phase is used to nucleate the precipitation reaction. Example nuclei are
thoria, Titania, and tungsten carbide.
The precipitated powders have some characteristics in common. Generally, the
crystallite size is quite small and agglomeration is a natural tendency. The powder
purity is usually over 99.5 % with the dominant impurities coming from the reaction
bath. The particle shape is irregular or cubic, or in some instances sponge-like.
Consequently, the flow properties are poor and the packing densities are low.
1.1.4. ATOMIZATION FABRICATION TECHNIQUES
In the last twenty years, P/M has turned to several advanced powder fabrication
techniques, which fall under the general heading of atomization. Prior to the
development of atomization, powder chemistry and shape characteristics could not be
fully controlled. The flexibility of the approach coupled to its applicability to several
alloys and easy process control, make it an attractive alternative. A main feature of
atomization is the general reliance on fusion-based technology.
4
1.1.4.1.Gas Atomization
The use of air, nitrogen, helium or argon as a fluid for breaking up a molten
metal stream provides a versatile powder fabrication technique. The liquid metal
stream is disintegrated by rapid gas expansion out of a nozzle. The approach has
proven ideal for super alloys and other highly alloyed materials. The designs may vary
with respect to the metal feed mechanism and the sophistication of the melting and
collection chambers: however the main idea is to deliver energy (from a rapidly
expanding gas) to the metal stream to form droplets.
. Figure1.1.3 shows a schematic diagram of vertical inert gas atomizer. The
melt must be superheated over the melting ( liquids ) temperature.
Figure 1.1.3. A vertical gas atomizer. The main
features are a vacuum induction furnace, gas
expansion nozzle, gas recirculation/supply
system free-flight chamber and powder
collection chamber
Because of the volume of gas used in atomization, it is important to exhaust the
gas to avoid a backpressure. It is necessary to incorporate a cyclone separator.
Gas atomization can be performed totally under inert conditions. Thereby
maintaining the integrity of high alloy feedstock. The particle shape is spherical with a
fairly wide size distribution. The list includes gas type, residual atmosphere, melt
temperature and viscosity as it enters the nozzle, alloy type, metal federate, gas
pressure, gas federate and velocity, nozzle geometry, and gas temperature.
5
The atomization physics can be described by the drawing shown as
Figure1.1.4. The expanding gas around the molten metal stream causes disturbances in
the melt surface, giving a cone shape after exit from the nozzle. From the top of the
cone, expansion causes the metal stream to form into a thin sheet. The sheet is unstable
because of a high surface area to volume ratio. The liquid continues to respond to the
shear and acceleration forces, giving first ligaments and subsequently finer spherical
particles. The size reduction is limited by the melt viscosity and temperature, and by
the response to the acceleration forces. The effect of superheating the melt above the
liquids is to decrease its viscosity and to prolong the post-atomization solidification
time. The particle shape sequence with distance from the nozzle is cylinder-cone-
sheet-ligament-sphere. Depending on the amount of superheat and other variables, any
one of these shapes may be produced.
Figure 1.1.5 The formation of a metal powder
by gas atomization involves the break-up the
liquid stream by rapidly expanding gas. The
stream first form into a thin sheet, and
subsequently forms ligaments, ellipsoids, and
spheres.
Shorter distances between the gas exit and melt stream favor better energy
transfer, aiding the formation of finer powders. The gas velocity on exit from the
atomizer is the dominant factor in determining the resulting particle size. In terms of
metal characteristics a low- density metal favors coarser particle sizes because of more
rapid acceleration out of the expansion zone.
D
1 = 3
2 / 1
3
1
]
1

V
W

6
Where
m
is the melt density and is the surface energy.
There is obvious interest in linking the mean particle size to the atomization
conditions. A Droplet formation during atomization is enhanced by a large difference
between the gases and melt velocities. This occurs with high gas pressures and gas
flow rates, giving an empirical form as follows
D =
57 . 0 22 . 0
1
]
1

1
]
1

m
m
m
U
V
C

Where C is a nozzle geometry constant, and Um is the melt viscosity. A


particle size dependence on the inverse of the gas velocity has proven applicable to the
atomization of several metals, including tin, iron, led, steel and copper.
1.1.4.2. Water Atomization
Water atomization is the most common technique for producing elemental and
alloy powders from metals, which melt below approximately l600C. In Figure 1.1.5.
an example of water atomizer geometry is shown. The water can be directed by a
single jet, multiple jets or an annular ring. The process is similar to gas atomization,
except for the rapid quenching and differing fluid properties. High- pressure water jets
are directed against the melt stream, forcing disintegration and rapid solidification.
Consequently, the powder shape is more irregular than with gas as the fluid. Also the
powder surface texture is rough, with some oxidation. Because of the rapid heat
extraction, shape control requires superheats far above the liquids. Because of the high
cooling rate, the water -atomized particle takes less time to solidify. Chemical
segregation within an alloy particle tends to be quite limited.
7
Figure 1.1.5. The water atomization process, where
multiple water jets disintegrate a molten metal
stream
Mathematical models for particle size from water atomization have similarities to
those for gas atomization. A high pressure, or high water velocity, causes a decrease in
the mean particle size. In a simle form the relation can be expressed as follows.
D =
( ) a V
C
sin .
Obviously, the water velocity is a major factor in controlling particle size.
1.1.4.3. Centrifugal Atomization
The desire to control particle size and the difficulties in fabricating powders from
reactive metals have led to the development of centrifugal atomization. The centrifugal
force throws off the molten metal as fine spray which solidifies into a powder. The
rotating electrode concept is shown in Figure1.1.6.
Figure 1.1.6 Centrifugal atomization by the rotating
electrodes process is shown in this diagram. A
rapidly rotating spindle is melted by an arc using a
tungsten cathode. The powder is formed by the melt
thrown from the anode, and can be solidified in
either a vacuum or inert gas environment
8
.
The mean particle size is increased by higher melt rates, slower rotations, and
smaller anodes.
D =
43 . 0
64 . 0
2 . 0

,
_

m Wd
M

Where M is the melt rate, d is the anode diameter, W is the angular velocity, is the
surface tension of the melt and m is the density.
Typically, the melt rate is on the order of 10
-7
m /s the rotation velocity between
1000 and 50.000 rpm and the anode diameter between 2 and 5 cm
1.1.4.4. Other Atomization Approaches
The melt explosion (vacuum chamber ) technique in Figure1.1.8. uses a
hydrogen saturated liquid metal and rapid desaturation in vacuum to form a fine
powder spray. The melt is pressurized with 1 to 3 Mpa of hydrogen. A siphon tube
then exhausts the saturated melt into a large vacuum chamber. Both the high velocity
and hydrogen desaturation cause the melt to literally explode into the vacuum
chamber.
Figure 1.1.8 The melt explosion technique for
forming spherical powders. Molten metal is
pressurized with hydrogen and exhausted via a
siphon tube into a low pressure chamber. The rapid
pressure change and hydrogen desaturation from
the melt cause the liquid stream to explode into a
fountain of droplets. The droplets solidify during
free-flight are collected at the bottom of the chamber
1.2 POWDER CHARACTERIZATION
9
This chapter is designed to introduce the techniques and measures available for
describing powders. The influence of the powder characteristics on processing and
properties will be demonstrated in subsequent chapters.
1.2.1 POWDER SAMPLING
Atypical production lot may be several tons in size; a sample of this lot will
probably be on the order of a kilogram. Many of the modern analytical instruments
require sample sizes of a gram or less for particle size analysis. Assuming a spherical
shape, the particle population in one gram depends on the size and material density
( theoretical density).
1.2.2. PARTICLE SIZE
The size of a particle depends on the measurement technique, specific
parameter being measured, and particle shape. Particle size analysis can be achieved
by several techniques, which usually do not give equivalent determinations due to
differences in the measured parameters. The basis for analysis can be any of the
obvious geometric values, such as surface area, projected area, maximum dimension,
minimum cross sectional area, or volume. Particle size is probably one of the most
important powder characteristics to the powder metallurgist. Size data are most useful
when presented within the context of the measurement basis and the assumed particle
shape. A particle size analysis should convey information on the particle size
distribution, particle shape, and state the basis for measuring particle size. The desire is
to use a particle dimension most characteristic of the powder.
1.2.3. MEASUREMENT TECHNIQUES
1.2.3.l. Microscopy
The procedure is covered in American Society for Testing Materials
specification E20 (ASTM E20). Although the technique is reasonably accurate, the
tedium of sizing statistically significant quantities of particles has led to use of
automatic image analyzers. The image for analysis is generated by optical, scanning
electron or transmission electron microscopes. The instrument choice depends on the
particle size. By microscopic counting of diameter, length, height or area, a frequency
10
distribution can be generated. Counting two or more small particles as a large particle
will cause a skewing of the distribution towards the coarse sizes.
1.2.3.2 Screening
The most common technique for rapidly analyzing particle size is based on
screening. A square grid of evenly spaced wires creates a mesh. Mesh sizes can not go
to very small opening sizes; thus, the screening technique is usually applied only to
particles larger than 38 m. There are electroformed meshes available down to 5 m,
but agglomeration and particle adhesion to the mesh generally make the electroformed
screens of little practical use. A listing of mesh sizes and opening sizes for the U. S.
Standard series of screens appears in Table l.2.3 (ASTM E11).
TABLE 1.2.1 Standard Sieve Sizes (U. S. Standard. ASTM E119 )
Mesh
Size
Opening in m
Permissible Variation
+,- m
Maximum Individual Opening
in m
18 1000 40 1135
20 850 35 970
25 710 30 815
30 600 25 695
35 500 20 585
40 425 19 502
45 355 16 425
50 300 14 363
60 250 12 306
70 212 10 263
80 180 9 227
100 150 8 192
120 125 7 163
140 106 6 141
170 90 5 122
200 75 5 103
230 63 4 89
270 53 4 76
325 45 3 66
400 38 3 57

11
The powder is loaded onto the top screen and the screen stack is vibrated for a
period of 20 to 30 minutes. For particle size analysis, a sample size of 200 g is usually
sufficient when using 20 cm diameter screens. After vibration, the amount of powder
in each size interval is weighed and the interval percent calculated for each size
fraction (ASTM B214)

1.2.3.3. Sedimentation
Particle size analysis by sedimentation is most applicable to the finer particle
sizes. Particles settling in a fluid (liquid or gas), reach a terminal velocity dependent on
both the particle size and the fluid viscosity. On this basis, particle size can be
estimated from the settling velocity. Depending on the particle density and shape,
sedimentation techniques are most applicable to particles in the 0.05 to 60 m range.
Assuming a spherical particle shape, settling at the terminal velocity in a
viscous medium is represented by a balance of forces. The buovancy and viscous drag
forces act to retard particle settling as diagramed in Figure. Alternatively the
gravitational force, at the terminal velocity the forces are balanced. The settling force
equals mass times acceleration.
F
G
= ( ) g
D
m

6
3
Where D is the particle diameter, g is the acceleration (gravity) and
m
is the
particle density. The buoyancy force is determined by the volume of fluid displaced by
the particle,
F
B
= ( ) g
D
t

6
3
Finally, the viscous drag force Fv is given as.
F
V
= DVU 3
U is the fluid viscosity. For a sedimentation experiment, the velocity is
calculated from the height and time. Combining equations gives.
V =
( )
( ) U
gD
t m
18
2
For the terminal velocity, which is known as Stokes law. It is experimentally
most convenient to work with a known settling height H while measuring the time for
settling.
12
D =
( ) ( )
2 / 1
18
1
]
1

t m
gt
HU

Figure 1.2.1 The force balance leading to a constant settling velocity for a spherical particle in a
viscous fluid.
The technique of particle size analysis by sedimentation uses a predetermined
settling height and places a dispersed powder at the top of a tube. Measurements of the
amount of powder setting at the bottom of the tube (weight or volume) versus settling
time then allows calculation of the particle size distribution. Obviously, the fastest
settling particles are the largest while the smallest can take considerable time to settle.
Automatic instrumentation for performing sedimentation based analyses use light
blocking, x-ray attenuation, weight or settled cake height to determine the size
distribution.
Internal porosity in the powder decreases the mass, thereby causing slower
particle settling.
There are mathematical limits to Stokes law. The derivation assumes that
viscosity controls settling. Accordingly, at Reynolds numbers R in the range of 0.2 to
l.2 the assumption of viscosity controlled settling break down. At high settling
velocities the sedimentation model is invalid if the calculated Reynolds number is
large, where gives the Reynolds number in terms of the settling parameters. Finally,
the fluid and powder can not react chemically. In spite of these several difficulties,
sedimentation techniques are in use for several powder systems such as the refractory
metals.
13
1.2.3.4. Light Scattering
1.2.3.5. Electrical Conductivity
1.2.3..6. Light Blocking
1.2.3.7. X-Ray Techniques
1.2.4. PROBLEMS IN PARTICLE SIZE ANALYSIS
For sieves, this is generally above 38 m. optical microscopy is restricted to
particles above l m. In contrast, techniques such as sedimentation are only applicable
to a narrow size range because of limitations in the applicable physics.
1.2.5. PARTICLE SHAPE
The shape of a particle is a distributed parameter, which can influence packing,
flow, and compressibility of a powder. Particle shape provides information on the
powder fabrication route and helps explain many processing characterictics. Because
of the difficulty in quantifying particle shape, qualitative descriptors are used. Figure
1.2.2 gives a collection of particle shapes and shows the appropriate qualitative
descriptors.
The most straightforward such descriptor is the aspect ratio. The aspect ratio is
defined as the maximum particle dimension divided by the minimum particle
dimension. For a sphere, the aspect ratio is unity, while for a ligament type particle a
value near 3 to 5 is more likely. A flake particle can have an aspect ratio in excess of
ten and in some instances can be as high as 200.
14
Figure 1.2.2 A collection of possible particle shapes and qualitative descriptors.
1.2.6. INTERPARTICLE FRICTION
Under the general heading of interparticle friction come two main concerns:
powder flow and packing. As the surface area increases, the amount of friction in a
powder mass increases. Consequently, the friction between particles increases, giving
less efficient flow and packing. These concerns are important in automatic die filling
during powder compaction, as well as packaging, transportation, blending and mixing
of powders.
The main feature of friction is a resistance to flow. Also, the density or
packing properties decrease because of poor flow past neighboring particles. The
apparent density of a powder is the density (mass/volume) when the powder is in the
15
loose state without agitation. This is also known as the bulk density. The tap density is
the highest density that can be achieved by vibration of a powder without the
application of external pressure. The theoretical density corresponds to the handbook
density for a powder material; the density when there is no porosity present. The angle
of repose is another friction index. It is the angle formed by pouring a powder into a
pile as shown in Figure 1.2.3, where the tangent of a equals the height divided by the
radius of the loose powder pile. Finally, the flow rate is a measure of the rate a powder
will feed under gravity through a small opening. Most fine powders will not flow
because of their high interparticle friction. Such powders are termed non-free flowing,
and present particularly difficult problems to engineers looking for high productivity in
forming operations.
Figure 1.2.3 The angle of repose is a measure
of the interparticle friction. It is determined
from the height and radius of the powder alter
passing through a funnel.
There are two common devices for measuring the apparent density. Examples
of both the Hall flowmeter and the Scott volumeter are given in Figure 1.2.4. The Hall
flowmeter is used for the coarser particles; both the flowrate and the apparent density
are measured by this device. Alternatively, the Scott device is applied to the fine
refractory powders which have higher interparticle friction. Both devices are covered
by ASTM specifications (ASTM B212 and B213 for the Hall and ASTM B329 for the
Scott).
The flow rate for a powder is usually expressed as the time for 50 g of powder
to flow through the Hall flowmeter , Smal flow times indicate free flowing powders
while long times are an indicator of high interparticle friction. The apparent density
16
and flow times are easily obtained with the Hall flowmeter by combining a precision
volume cup with the funnel. In this case the apparent density is the weigh of powder
divided by the cup volume.
Another simple test for interparticle friction is the tap density. Powder is
Vibrated in a cylindrical volume for l000 cycles at 284 cycles per minute using a 3.2
mm throw from an eccentric cam (ASTM B527). Usually the initial powder volume is
250 ml. The tap density is the weight divided by the final volume. Both the tap and
apparent densities can be expressed as fractions of theoretical density.
Figure 1.2.4 The basic compenents of the Hall flowmeter and Scott volumeter for measuring the flow
and packing of powders.
1.2.7. CHEMICAL CHARACTERIZATION
The elemental powders are relatively high-purity materials where chemical
analysis focuses on the impurity concentration. The prealloyed
Powders constitute micro-castings with multiple elements in a predetermined
ratio. For the prealloyed powders, attention is given to the alloy composition
as well as the impurity concentrations.
Beyond the bulk chemical information, there is often a need to know the
surface condition of the powder. Hence, there is concern with oxides, adsorbed organic
films and the presence of surface coatings like silica.
17
Beyond such tests, metallography can be applied to assess the inclusion
concentration. Also, for gas atomized powders, metallographic examination will show
the presence of internal gas pockets. In a few instances, techniques such as Auger
electron spectroscopy have been applied to characterize the surface chemistry. Most
recently, transmission electron microscopy has been applied to thinned particles to
determine the micro segregations and phases.
Bulk chemical characterization of a powder can be obtained from emission
spectroscopy, colorimetry, x-ray fluorescence and neutron activation analysis.
The prealloyed powders should be checked to verify adherence to chemical
tolerances. These powders tend to have higher oxide surface concentrations. In the
cases involving premixed powder blends (such as copper and tin to form bronze during
sintering), the blend chemistry should be checked and blend uniformity should be
assured by reblending and deagglomeration.
1.3. PRECOMPACTION POWDER HANDLING
This part discuses the powder handling steps before compaction, including
blending, mixing, classification and lubrication. The characteristics of common
lubricants are discussed as they affect powder properties.
1.3.1. PRECOMPACTION
It is necessary to tailor specific properties into a powder for easier compaction
and sintering. Examples of operations that occur in the pre-compaction stage include
classification, blending, mixing, agglomeration, de-agglomeration, and lubrication.
Classification is used to obtain a specific size fraction from a powder..
1.3.2. MIXING and BLENDING APPROACHES
Blending and mixing both combine powders into a homogeneous mass.
Blending refers to the combination of different sized powders of the same chemistry,
while mixing implies different powder chemistries. In spite of the recognized necessity
of such precompaction steps, the processes are poorly understood. These operations
can be a source of problems in fabricating components. Some simple rules reduce the
likelihood of problems:
18
1. Do not use a powder after transport without reblending.
2. Do not vibrate a powder.
3. Do not feed a powder through a free-fall where fine and coarse sizes can
segregate due to different settling rates.
The mechanisms of powder mixing are diffusion, convection and shear. These
three types of mixing are illustrated in Figure 1.3.1 as diffusional mixing in a rotating
drum, convective mixing in a screw mixer, and shear mixing in a blade mixer. A
diffusional mix occurs by the motion of individual particles into the powder lot. An
inclined plane of the powder bed breaks down at the outer edge, allowing flow over the
surface.
Figure 1.3.1 The three modes of powder mixing are diffusion, convection and shear. These processes
are given schematically, although in powder mixing all three contribute to the homogenization sequence
1.3.2.2. Powder Lubrication
Interparticle friction reduces the powder flow and packing properties. A more
fundamental problem is the friction between the die wall and the powder during
pressing. As the compaction pressure is increased, ejection of the powder mass from
the die becomes more difficult. Consequently, lubricants are used to minimize die wear
ease ejection from the die body.
There are two means of lubricating a pressing; die wall and powder lubrication.
Die wall lubrication is preferred in theory, but is not easy to incorporate into automatic
compaction equipment. The lubricants are usually mixed with the metal powder as a
final step before pressing. For metal powders, stearates based on Al. Zn. Li. Mg. Or
Ca are in common use. The stearate is added to the metal powder as a fine (typically
19
atomized) spherical form. A mean size of 30 m is common. Concentrations of the
lubricant range up to 2.0 wt.%.
1.4. COMPACTION
Pressure is used to form powders into engineered shapes with close
dimensional control. This compaction process involves both rearrangement and
deformation of the particles, leading to the development of inter-particle bonds.
1.4.1 PHENOMENOLOGY of COMPACTION
An external pressure is needed to both shape the powder and promote higher
packing densities. The schematic of powder compaction shown in Figure 1.4.1
provides a basis for defining the stages of compaction. The initial transition with
pressurization is from a loose array of particles to a closer packing. Subsequently, the
point contacts deform as the pressure increases. Finally, the particles undergo
extensive plastic deformation. At the beginning of a compaction cycle, the powder has
a density approximately equal to the apparent density. Voids exist between the
particles, and even with vibration, the highest obtainable density is only the tap
density. For a loose powder there is an excess of void space, no strength and a low
coordination number (number of touching neighbor particles). As pressure is applied,
the first response is rearrangement of the particles, giving a higher packing
coordination. The initial pressurization is therefore analogous to vibrating the powder,
because the density increases by powder restacking. Large pores caused by particle
bridging are initially filled by rearrangement. The rearrangement portion of
compaction is aided by hard particle surfaces (such as with oxides).
20
Figure 1.4.1 A simplified view of the stages of metal powder compaction. Initially, repacking occurs
with the elimination of particle bridges. With higher compaction pressures, particle deformation is the
dominant mode of densification.
. High pressures increase density by contact enlargement through plastic
deformation. The interparticle contact zones take on a flattened appearance. During
deformation, cold welding at the interparticle contacts contributes to the development
of strength in the compact. The strength after pressing, but before sintering, is termed
the green strength.
At low pressures, plastic flow is localized to particle contacts. As the pressure
increases, homogeneous plastic flow occurs throughout the compact. With sufficient
pressurization, the entire particle becomes work (strain) hardened as the amount of
porosity decreases.
1.4.2. CONVENTIONAL COMPACTION
Conventional powder compaction is performed in hard tooling of the type
shown in Figure 1.4.2.
When pressure is transmitted from both the bottom and top punches, the
process is termed double action pressing. Alternatively, when pressure is transmitted
from only one punch, the process is termed single action pressing.
21
Figure 1.4.2 A conventional punch and die set
for powder compaction; the punches provide
compression and the die gives lateral support
to the powder.
A general system of part classification exists for declaring the shape
complexity. As the number of part levels and the complexity of the pressing directions
increase, the part classification also increases as noted in table 1.4.1.
There are several modes of pressing and accordingly there are several types of
presses, including hydraulic, mechanical, rotary, isostatic, and anvil.
TABLE 1.4.1 The Classifications of P/M Patrs
Class Part Levels Pressing Directions
1 One One
2 One Two
3 Two Two
4 Sveral Two
1.4.3. THEORETICAL BASIS
The main problem in powder compaction is the die wall friction with the
powder. This friction causes the applied pressure to decrease with depth in the powder
bed. There are many important intrinsic characteristics of a powder that affect the
pressure-density-strength relations in a powder compact. These include the material
properties like hardness, work (strain) hardening rate, surface friction, and chemical
bonding between particles. Equally important are the extrinsic factors associated with
the powder size, shape, lubrication and the mode of compaction.
1.4.3.l. Fundamentals of Compaction
22
Consider a cylindrical compact of diameter D and height H such as drawn in
Figure 1.4.4. Analyzing a thin section of height d H when theres is an external pressing
force, shows that the pressure on top of the element P and that transmitted through the
element bottom Pb will differ by the normal force acting against friction.
Mathematically, the balance of forces can be expressed as follows:
( )
n b
uF P P A F +

0
Figure 1.4.4 The balance of forces during die
compaction, where the difference in the
applied and transmitted pressures results from
the frictional force at the die wall. A small
element from the compact serves as the basis
for calculating the pressure distribution.
Where F
n
is the normal force, u is the coofficient of friction between the
powder and the die wall, and A is the cross sectional area. The normal force can be
given in terms of the applied pressure with a proportionality constant z. the factor z
represents the ratio of the radial stress to the axial stress, thus
F
n
= zPDdH
The friction force F
f
is calculated directly from the normal force and the
coefficient of friction as,
F
r
= zPDdH u
Combining terms gives the pressure difference between the top and bottom of
the powder element d P as,
dP = P-P
b
= -F
t
/a = -4uzPdh / D
Integration of the pressure term with respect to compact height shows that the
pressure at any position x below the punch is given by the following expression:
23
[ ]
D
uzx
P
P
x 4
exp

This equation is applicable to a single action pressing. It shows that the


pressure decreases with depth in the powder bed. Examples of plots of this expression
are given in Figure. Note the effect of an increase in the term uz H/D; attention will be
given to the significance of this term in part 3 of this section. The wall friction
contributes to a decreased pressure with depth. Hence, for homogeneous compaction,
small height to diameter ratios are desirable.
For a single ended pressing, the average compaction stress is estimated as,
( )
D
uzH
P
2 1

and for a double ended pressing the average stress is approximately.
( )
D
uzH
P

The average stress is dependent on both the geometry (H/D), the interparticle
friction (z), and the die wall friction (U). High average streses are attained in short
compacts, with large diameters and lubricated die walls. The most important parameter
is the height to diameter ratio of the compact.
1.4.3.2. Particle Bonding in the Green State
A high initial packing density aids the formation of interparticle bonds.
Additionally, a clean powder surface aids bond strength. When the compaction force is
sufficiently high, shear forces will act to discrupt surface films.
1.4.3.3. Goals in Compaction
The predominant goal in powder compaction is to achieve compact properties
with minimal wall friction. Thus, efforts are made to decrease the axial to radial forces
to minimize die wear and improve pressing efficiency. The height to diameter ratio is
important to uniform compact properties. Generally, when the height to diameter ratio
exceeds five, die compaction is unsuccessful. A low compact height allows for
successful single action pressing: however, double action pressing is the predominant
approach.
The ratio zu H/D is a sensitive gauge of the pressing operation. Powder
lubrication raises z while lowering u. First, the die wall friction u depends on the
24
amount of lubricant, the more lubricant, the lower the friction. Second, the die wall
friction decreases as the pressure increases. Third, the pressure ratio z increases with
approximately the square root of the applied pressure.
In die pressed powder compacts, density gradients result from the pressure
gradients. For a copper powder, the density gradients with both single and double
action pressing are shown in figure. In both compacts, the height to diameter ratio is
unity, the coefficient of friction is 0.3 and the pressure ratio is 0.5.In the single action
pressing, the lowest density occurs at the compact bottom. Alternatively, the double
action pressing has the lowest density in the very center of the compact.
The other important factor is the height to diameter ratio. As this ratio is
increased, density gradients in a compact will increase and the overall compact density
will decrease. For a single action pressing of copper using a constant compaction
pressure of 700 M Pa. Plots of the approximate pressure distribution in the compacts
are given for height to diameter ratios of 0.42, 0.79 and l.66. An increase in the height
to diameter ratio results in greater density gradients and a lower bulk density.
1.5 SINTERING
Sintering is the process whereby particles bond together at temperatures typically
below the melting point by atomic transport events. A characteristic feature of
sintering is that the rate is very sensitive to temperature. The driving force for sintering
is a reduction in the system free energy, manifested by decreased surface curvatures
and an elimination of surface area.
1.5.1. SINTERING THEORY
1.5.1.l. Characteristic Stages
Consider two spherical particles in contact such as shown in Figure 1.5.1. As the
bond between the particles grows, the microstructure changes as shown in Figure
1.5.2. The initial stage of sintering is defined as occurring while the neck size ratio
X/R is less than 0.3 for uncompacted powders.
25
Figure 1.5.2 The development of the interparticle bond during sintering, starting with a point contact.
The pore volume shrinks and the pores become smoother. As pore spheroidization occurs, the pores are
replaced by grain boundaries.
During initial stage sintering shows that it is the curvature gradient at the neck
which guides the mass flow.
In the intermediate stage, the pore structure is much smoother. The pores have
an interconnected, cylindrical structure. At this point attention shifts from the
interparticle neck growth to the grain-pore structure. The predominant development of
compact properties occurs in the intermediate stage. The driving force is the interfacial
energy, including both the surface and grain boundary energy. It is common for grain
growth to occur in the latter portion of the intermediate stage. As a consequence, either
pore motion or pore isolation can occur.
With shrinkage of the pore structure, the cylinders become unstable at
approximately 8% porosity. At this point, the cylindrical pores collapse into spherical
pores which are not effective in slowing grain growth. In many cases, the
microstructure exhibits pores separated from the grain boundaries. The isolation of the
pores at grain interiors results in a drastic decrease in the densification rate. In the final
stage, the kinetics are very slow. The driving force is strictly the elimination of the
pore-solid interfacial area. The presence of a gas in the pore will limit the amount of
final stage densification.
1.5.1.2.Transport Mechanisms
The transport mechanisms are the ways in which mass flow occurs in response
to the driving forces. There are two classes of transport mechanisms surface transport
26
and bulk transport. Surface transport, as shown in Figure 1.5.2 involves neck growth
without a change in particle spacing (without densification).these paths are shown in
Figure 1.5.2. In contrast to surface transport, bulk transport controlled sintering results
in net dimensional changes. The mass originate at the particle interior with deposition
at the neck region, as shown in Figure 1.5.2. The bulk transport mechanisms include
volume diffusion, grain boundary diffusion, plastic flow, and viscous flow (for the
amorphous solids)
Figure 1.5.3 The two classes of sintering
mechanisms as applied to sphere-sphere
sintering. Surface transport mechanisms
provide for neck growth by moving mass from
surface sources (E C= evaporation-
condensation, SD = surface diffusion, VD =
volume diffusion). Bulk transport processes
provide for neck growth using internal mass
sources (PF = plastic flow, GB = grain
boundary diffusion. VD = volume diffusion).
Only bulk transport mechanisms give
shrinkage.
1.5.1.3.Initial Stage Sintering
A point contact between particles leads to the growth of a neck at a rate which
depends on the mechanism of mass transport. The sintering rate depends on the rate of
material arriving from the various transport paths. Although viscous flow is a possible
transport mechanism, for crystalline materials it is not applicable.
The model for neck growth during initial stage sintering represents the
contribution of several investigators as noted by Thummler and Thomma, and Exner.
Assuming monosized spheres initially in point contact, the neck growth by a single
mechanism can be represented by a generalized equation,
( )
m
n
R
Bt
R
X

27
Where X is the neck radius, R is the particle radius, t is the isothermal sintering time,
and B is a collection of material and geometric constants.
It is important to consider the distributed nature of powders when discussing
the particle size effect. The models for sintering assume a homogeneous geometry.
However, in real powder systems there is a distribution in particle size, number of
contacts per particle, and contact flattening due to compaction.
1.5.1.4. Intermediate Stage Sintering
The intermediate stage is the most important in determining the properties of
the sintered compact. This sccond stage is characterized by densification coupled to
grain growth. The pore structure becomes smooth but remains interconnected until the
final stage. In many instances, dimensional change is not acceptable during sintering.
For such cases, short sintering times are typically combined with lower sintering
temperatures and high compaction pressures to minimize densification. Alternatively,
with the refractory metals, emphasis is on achieving densification. Consequently, the
intermediate stage is viewed quite differently. For this discussion, the focus will be on
the physics of intermediate stage sintering without a distinction between the possible
merits and demerits of the observed densification
1.5.1.5.Final Stage Sintering
Final stage sintering is a slow process wherein isolated, spherical pores shrink
by a bulk diffusion mechanism. The isolation of a pore in the final stage of sintering is
illustrated in Figure 1.5.3. For the pore sitting on a grain boundary, a small dihedral
angle causes a large pinning force. Spherical pores are expected after grain boundary
breakaway. After boundary breakaway, the pore must diffuse vacancies to distant grain
boundaries to continue shrinking, which is a slow process. Also, with prolonged
heating, pore coarsening will cause the mean pore size to increase while the number of
pores will decrease. Differences in the pore curvature will lead to growth of the larger
pores at the expense of the smaller, less stable pores.
28
Figure 1.5.3 The sequence of steps leading to pore isolation and spheroidization in the final stage of
sintering; a) pore on the grain boundary exhibiting an equilibrium dihedral angle, b)and c)correspond
to grain growth with pore drag, and d)represents pore isolation because of boundary breakaway.
1.5.2 ENHANCED SINTERING
There are four common approaches to sintering enhancement in metal
powders: hot pressing, phase stabilization (or mixed phase sintering). Activated
sintering, and liquid phase sintering.
1.5.2.l. Hot Pressing
Uniaxial hot pressing resembles die compaction with both an upper and lower
punch. The rate of densification due to an external stress can be estimated in terms of a
surface energy enhanced driving force. The action of an external stress is to promote
grain and particle sliding by diffusional processes, to generate excess vacancies, and to
cause pore collapse. The overt effect is a more rapid densification. More extensive
discussion of hot pressing is given in the next chapter on full density processing.
1.5.2.2. Phase Stabilization
The volume diffusivity of a material is determined by several factors including
the temperature, crystal structure and the defect configuration. For a material like iron,
the volume diffusivity at 9l0C is 330 times higher in the body- centered cubic (BCC)
phase, ferrite, than the face-centered cubic (FCC) phase, austenite. Stabilization of the
29
BCC phase provides an avenue to more rapid sintering of iron. Elements like
molybdenum, phosphorus and silicon stabilize ferrite above the polymorphic
transformation temperature.
1.5.2.3.Activated Sintering
The term activated sintering refers to any of several techniques which lower the
activation energy for sintering. The term implies enhanced densification or improved
properties in the sintered product. Activated sintering allows for a lower sintering
temperature, shorter sintering time, or better properties. Several techniques have been
invented to achieve this goal, ranging from chemical additions to the powder, to the
application of external electrical fields. In this respect, mixed phase sintering
treatments can be categorized as activated sintering.
In activated sintering, the amount of activator and the particle size are very
important parameters. First, the activator must be either a metal or compound which
fores a low melting temperature phase during sintering. Secondly, the activator must
have a large solubility for the base metal, while the base metal shown have a low
solubility for the activator. The operation of the activator is to slay segregated to the
interparticle interfaces during sintering. Such a segregated layer provides a high
diffusivity path for rapid sintering. The lower melting point ensures a lower activation
energy for diffusion, while the solubility ensures that the activator is not dissolved into
the base metal during sintering.
Chemical additions are the most successful means of activating sintering.
1.5.2.4. Liquid Phase Sintering
In two phase systems involving mixed powders, liquid formation is possible
because of differing melting ranges for the two or more components. In such a system,
the liquid may provide for rapid transport and therefore rapid sintering if certain
criteria are met. The liquid must form a film surrounding the solid phase, thus wetting
is the first requirement. Secondly, the liquid must have a solubility for the solid. The
formation of a liquid film has the benefit of a surface tension force acting to aid
densification and pore elimination. In this sense, the liquid phase acts like a low
30
pressure external stress. Common systems involving liquid phase formation during
sintering include Cu- Co, W-Cu, W-Ni-Fe, W-Ag, Cu-Sn, Fe-Cu, WC-Co, and Cu-P.
The combination of wetting, liquid flow and particle rearrangement all
contribute to a rapid change in volume of the compact. With continued heating in the
presence of a liquid phase, the solid phase begins dissolving into the liquid. If the solid
has a high solubility in the liquid, then it is possible for the liquid composition to
recross a solidus boundary and solidify.
1.5.3. SINTERING ATMOSPHERES
Due to their porous structure, pressed powder components react more readly
with the surrounding atmosphere than fully dense materials. For this reason the
sintering atmosphere is very important. The protective atmospheres used are mainly
gases, in specials cases vacuum. The choice of gas must taken into account possible
reaction between the gas. These reactions depend on temperature and pressure and are
numerous. Because gases used in commercial production often contain trace gases
such as H
2
, H2O, CO, CO2 or N
2
. Additional gases may be envolved during
aannealing due to interactions with the sintering components.
1.5.3.1.Pure gases
Hydrogen (H
2
) is the most common of the commercially pure gases. Although
pure dry hydrogen is a relativly expensesive protective gas, it is widely used as a
sintering atmosphere because it provides the most effective reducing atmosphere.
Because of the high damger of explosion resulting from air ingress, special safety
precautions must be taken. Hydrogen-air mixture4s are explosive in the range of 4 to
74% H
2
; their minimum ignition temperature is 574C.
1.5.3.2.Dissociated (cracked) ammonia
Sintering plants often use dissociated ammonia ewhen a furnace gas is required
which will have a reducing effect. Dissociated ammonia has a high hydrogen content is
free from CO, CO
2
and water, and does not contain any other oxygen or sulhur
31
compounds. Its use is suitable for many materials including those, which contain
alloying components, which oxidise readly (Cr-Ni steels). Due to its high H
2
content
the gas is decarburising and is therefore not suitable for sintering steels containing C.
1.5.3.3.Protective gases produced by burning hydrocarbons
1.5.3.4.Vacuum
Various developments in vacuum technology have widened the use of vacuum
in sintering. these developments include the development of high performance pumps
and other additional equipment as well as the availability of furnaces capable of
continues operation. Vacuum has the advantages that metals are protected from
oxidation (a high vacuum of 10
-2
Pa, for instance yields a dew point of about -90C)
And no impurities enter the furnace zone during sintering. Furthermore, evaporation
from the surfaces of the parts causes a certain self purification of the sintering stocks
vacuum deoxidation is used to remove oxides, which is adhere to most mass produced
sintering materials after pressing. . However it is a slow process which is a
disadvantages of this processing method for this reason vacuum (high vacuum) tends
to be used as a sintering atmosphere only when the materials to be sintered have a very
high sintering reactivity especially with oxygen, moisture and the hydrogen.
1.5.4. SINTERING FURNACES
The sintering furnace provides the time- temperature control to the sintering
cycle. Additionally, it contains the atmosphere, provides for removal of the lubricants
and binders, and controls the heat treatment. The furnace performs these various
functions in either batch or continuous mode. A batch furnace is loaded with the
material to be sintered and then is raised to temperature. A continuous furnace
provides for the compact treatment by controlling the position in a preheated furnace.
Figure 1.5.4 shows the type of time-temperature cycle needed in commercial sintering
treatments. The difference between furnace types depends on control of either the
furnace temperature or compact position versus time.
32
Figure 1.5.10 The sequence of operations occurring in a sintering furnace. The lower diagram shows
the time-temperature profile typical to metal powder sintering.
In continuous furnaces, parts are moved through a multiple zone furnace by a
conveyor, such as a belt, pusher or other mechanical device. Usually, the conveyor
proves to be a major limitation in the furnace operating temperature. Several different
heating elements are available for generating the temperature. Elements which require
a reducing atmosphere are kept under hydrogen while at temperature. Otherwise, the
furnace heating elements can be located external to the atmosphere, with radiant
heating through a furnace muffle.
33
2.PARTICLE PACKING CHARACTERISTICS
2.1. STRUCTURES IN ONE AND TWO DIMENSIONS
There are two very different types of packing structures, random and ordered.
A random packing is constructed by a sequence of events that are not correlated with
one another. The result of such a random assembly procedure is a structures without
long-range repetition. Random structures have a lower packing density than attainable
with the high packing density ordered structures. An ordered structure occurs when
objects are placed systematically into periodic positions, as are the bricks in a wall or
atoms in a crystal structure.
The spherical particle shape has received greatest attention in three-
dimensional packing. This is because only one size parameter, the diameter, is needed
to specify the dimension of spheres. Other than regular polyhedral shapes, (for
example, a cube), most other shapes require multiple parameters to specify their size
and shape.
The analogous packing problem in one dimension is termed parking and is
analyzed in terms of segments placed on a line. For an ordered structure, the parking
density of line segments will equal 100%, because of perfect end-to-end alignment of
the segments. This is similar to a string of pearls without any gaps. Without a random
placement of segments the coverage is less than complete. In three dimensions the
problem is termed packing, while in one and two dimensions the problem can also be
referred to as parking and covering, respectively.
2.1.1. PACKING OF MONOSIZED SEGMENTS IN ONE DIMENSION
In the ordered structure, the assembly of the packing is systematic, and the
fractional packing density is 1,00. As illustrated in figure 2.1.12 The coordination
number is two with one contact at each end. Conceptually the first segment is placed at
one end of the string, the second is placed just in contact with the first, and the process
is repeated to fill the string. Each added segment creates one new contact, which is
34
shared by two segments, giving a net coordination increase of two contacts per
segments. Total coverage occurs with this end-to-end ordered structure.
figure 2.1.1 one dimensional packing of segments along a line a) ordered packing packing with full
coverage b) random packing with incomplete coverage
2.1.2. PACKING OF MONOSIZED SEGMENTS IN TWO DIMENSION
2.1.2.1 Ordered Packing
Because of the shape, the packing structure is not totally covering for the
underlying structure. Indeed, significant overlap of the disks is required to attain total
coverage, with in an estimated excess in disk area of 21% over the covered area
necessary to produce a complete covering. For this reason, adisk (or sphere in three
dimension) is termed a low packing efficiency.
There are three ordered packing of disks that can be repeated to fill space, as
shown in figure 2.1.2. these are the best characterized by the number of contact points
for each disk, which is the coordination number N
c
, and the fraction density . The
fraction density is the density expressed as a fraction of the theoretical density for the
material. The lowest density structure has a coordination number of three and the
highest density structure has a coordination number of six. The packing density
increases with the coordination number. There is no repetive unit with five-fold
symmetry. However a statistical model of two-dimensional packing gives an estimated
density of 0.78 for the five-fold geometry.
35
Figure 2.1.2. a sketch of the ordered structures of disks packed in two dimensions with
coordination numbers of three, four, and six
The six-fold geometry is termed closed-packed. Packing disks on every other
lattice site in a square grid, where the occupied sites are those having either both
coordinates odd and both even, can construct it. This structure gives the theoretical
maximum packing density for disks. This planar structure is the same as that which
makes up the most dense ordered packing in three dimensions. The actual minimum
unit for creating this geometry is cluster of three disks. This is the unit cell in two-
dimensional space that is space filling.
2.1.2.2. Random Packing
The random packing density of disks in two-dimension is largely dependent on the
procedure used in two assembling the packing. Indeed, two random conditions are
possible, loose and dense. Most studies use some force in maintaining contact between
the disks, giving a maximum packing density for the random structure, the dense
random packing. This can be unidirectional force, such as gravity, or a central
attraction force. Random packing without such a force will allow low density regions
with zero or one contact per disk, the random loose packing. Alternatively, packing
formed with a force will have at least two contacts per disks and a correspondingly
higher density. Vibration has a slight influence on the packing density, but this
influence is less than observed in three-dimensional packings. With small packing
cluster, cluster size also effect. Generally at least 1000 disks are needed to ensure
behavior representative of true random packing without disruption from container.
36
The actual packing density for a random mixture can vary over a considerable
range, yet the transition between random and ordered packing occurs when the
fractional density exceeds 0.82.
The coordination number in a random packing is variable from point to point.
Some regions with a coordination of six will exist while other regions will have values
as low as two. The structure is better characterized as regions of order separated by
random or defective regions. Consequently, a random packing with a density over
approximately 0,82 can be treated as a mixture of random and ordered regions.
2.1.3. PACKING OF MIXED SEGMENTS IN ONE DIMENSION
The random placement is important in packing problems, since random events
are more typical in natural process. Although randomly placed segments produce a
packing density below that for an order structure, if two segments sizes are used to
randomly pack a line, then it is possible that the smaller segments can preferentially
fill the voids left in packing of larger segments. For a bimodal one-dimensional
packing density depends on the size ratio between the segments and the compositional
ratio. The size ratio denoted by D
L
/D
S
, which is the ratio of the large to small sizes.
The composition is given in terms of the fraction of small segments X
S
and the packing
density ig given as .f the segments are very different in size then the smaller
segment will increase the packing efficiency by filling many of voids between the
larger segments. However, the additional factor of composition becomes important: if
there are too few of the small segments then there is little packing benefit.
Alternatively if there are too many of the small segments, then the packing density is
controlled by their inherent void space.
It is possible that a distribution in segment sizes would be beneficial in
attaining a high packing density from randomly placed segments. The character of the
optimal distribution presents a difficult problem. In computer simulations of random
one-dimensional packings, it has been determined that wide variations in the
distribution width are indeed beneficial. The packing density increases as the
normalized coefficient of variation increases. The coefficient of variation for a
distribution is defined as the standard deviation divided by the mean size. Goldman et
al. examined several types of distribution and found that the packing density not only
37
increased with large coefficient of variation, but a high population of large segments
was most favorable. For a given packing density the small segments break up the line
length more rapidly than the large segments. That is the small segments preempt space
disproportionately to their packing contribution.
2.1.4. PACKING OF MIXED DISKS IN TWO DIMENSION
It is possible to improve the packing or covering of disks by placing small
disks in the voids between large disks. Both ordered and random structures are
possible. The size ratio of disks determines the packing coordination at the highest
density
For the most efficient packing improvement as the gain in density per disk, the
added disk should just touch the three neighboring disks. Using the generalized
geometry shown in figure 2.1.4.1 The three existing disks have radiuses of R1 R
2
and
R3
.
The radius of the disks that just touches these disks is noted as R
4
and can be
calculated as follows:
1 / R
4
=(1 / R
1)
+ (1 / R
2)
+ (1 / R
3)
+2[(1/ R
1
R
2
) + (1/ R
1
R
3
) +(1/ R
2
R
3
)]
1/2
If the three large disks are equal in size, this equation predicts R
4
will have a
radius of 1/6.464 or 0.1547 of the large disk radius. The procedure for filling with
smaller disks can be continued with still smaller disks fitted into each of the voids
remaining after the first level of filling. Accordingly, the number of disks increases by
a factor of three as each new size class added.
38
A general case is that of the disks sizes packed into the voids of an existing
structures where the disks sizes are very different. In this case more than one disks is
allowed to fit into each void. If each disks size has associated with it a fractional
density , then by filling the voids with smaller disks the packing density of the
mixture
mix
is improved to give,

mix
=1 (1- )
k

Where K is the number of levels of sizes. This says that the voids existing after
positioning the first level of diks, equal to 1 are filled to increase the packing
density by an additional amount . For an ordered bimodal mixture with an infinite
ratio of large to small disks sizes, the peak packing density will converge to 0.9913 at
the optimal composition. This differs from the efficient bimodal case where the large
to small size ratio is 6.464, which gives a maximum packing density of 0.9503. The
fractional density approaches unity as the number of size levels increase. How ever in
practice putting the small disks into the proper sites presents a mechanical problem
that limits the actual coverage.
The structure of binary disks mixtures has important implications with respect
to amorphous materials and random packing of particles. Consequently several studies
have been conducted to view thee two- dimensional structures using disks or spheres
with a size ratio close to unity. The two-dimensional factors are the size ratio and
composition of the mixture. The fractional packing density of a random assembly is
increased by mixing large and small disks. The degree of improvement depends on the
size ratio: if the sizes of the disks in a mixtures are not very different, then the packing
density shows little improvement over the random case.
In a mixture of large and small, three structures can form, depending on the
density. A random packing exists at low densities. At high densities, the structure is
ordered or at least exhibits regions of order surrounded by regions of disorder.
Depending on the composition and size ratio, segregation by size can also be seen,
especially when the mixtures is vibrated. At intermediate densities with small size
ratios the structure is termed hexatic since there are local regions of order.
39
2.2. FACTORS AFFECTING PARTICLE PACKINGS
Generally the density of a material has no significant influence on its fractional
packing density. Particles of an equal size and shape will pack to the same
fractional density in spite of differing theoretical densities. However, several
other factors do cause differences in fractional packing densities. The factors are
particle size, particle shape, agitation, particle size distribution, surface texture,
agglomeration, container size, segregation, bridging, surface-active agents, internal
powder structure, and cohesion.
2.2.1.STABLE POSITION
For gravitational stability a particle must have contact with at least three other
particles. Figure 2.2.1 illustrates the addition of a new particle to an existing
layer of particles. The initial point contact is unstable in comparison with the
lower energy double contact. In turn, a lower height and energy can be achieved
by further rolling the sphere into the valley between the three stationary
spheres. Within a powder mass, an individual particle requires at least four contacts
within a powder mass to ensure stability. These four contacts cannot lie on a
single equator or single hemisphere of the particle
Figure 2.2.1: an added particle attains a stable position by first impacting on one existing particle,
rolling to contact two particles and finally rolling into a valley between existing particle
As the packing density of a powder decreases, conceptually it reaches a point
where the compact is no longer stable. Under perfect conditions stability might
persist at very low packing densities. However, a random structure becomes
unstable if there are fewer than 4.75 average contacts per particle. Based on
40
various correlations between packing density and coordination number, a fractional
density of 0.35 for monosized spheres corresponds to approximately 3.5 to 5.0
contacts per sphere. This range agrees with the mean value of 4.75 contacts per sphere
mentioned above. Packed particles with fewer contacts will be unstable
PARTICLE SIZE
For packings composed of large particles, the particles size is not important
to the density. However, when the mean particle size is below approximately 100
m there is more interparticle friction, and particle bridging is more likely to occur.
The decreasing packing density with smaller particles is due to an increase in the
surface area, a lower particle mass, and a greater significance of the short-range,
weak forces such as electrostatic fields, moisture, and surface adsorption. Since
interparticle cohesion increases with a smaller particle size, there is more
agglomeration and inhibited packing. The smaller particles give a lower
packing density. The packing densities can be very low for particle sizes
significantly below 1 m.
2.2.3.PARTICLE SHAPE AND SURFACE TEXTURE
Another form of interparticle friction arises from irregularities on the particle
surface. The greater the surface roughness or the more irregular the particle shape,
then the lower the packing density. The data for the particle shape effect on
packing density are scattered, yet some general patterns are apparent. Figure 2.3.1
provides a schematic of the general particle shape and surface roughness
effects on the fractional packing density. On the left, the packing density is shown
as a function of the relative sphericity, which is defined as the surface area of a
sphere of equivalent volume divided by the actual surface area of the particle. The
closer the particle shape is to being spherical, the larger the relative sphericity.
On this basis both particle shape and surface texture are included in the relative
sphericity. The right half of this figure shows the effect of the relative surface
roughness. The relative roughness is a measure of the texture on the powder
surface for an otherwise spherical shape. In this regard, the effect of surface
roughness is similar to the particle shape effect. This is due to bridging of the
41
particles. The use of vibration or lubricants can help attain a high packing density,
but problems may arise with agglomeration or size segregation. Water or various
oils can reduce the interparticle friction; however, treatments that increase the
surface stickiness of a powder will degrade the packing density.
Figure
2.3.1 schematic plots of the effects of particle shape and surface texture on the dense random fractional
packing density. The highest density is associated with smooth spherical particles.
For powders of the same size but different shapes, the packing density
will decrease as the shape departs from equiaxed (spherical). This is easily seen in
the packing of fibers compared to spheres. The length to diameter ratio provides a
measure of the departure from an equiaxed shape. As the shape becomes more
fibrous, with a larger ratio of length to diameter, the packing density is reduced. In
powder mixing, an irregular particle shape will interfere with mixing, but will also
help maintain a homogeneous mixture by interfering with demixing. Density can
be improved by mixing different sizes of particles. This packing benefit is
independent of shape, but the starting densities are lower with irregular particle
shapes. However, with certain shapes under vibration, a high packing density
may be achieved by orienting the irregular particles. Such a high packing
density occurs most typically with equiaxed particles.
2.2.4. AGGLOMERATION
Small particles cause difficulty in attaining a high packing density
42
because of agglomeration due to cohesion. The attractive forces between
particles become larger as the surface area increases and the
particle mass decreases. In addition, small particles have a greater
tendency for vapor condensation at the particle contacts. Agglomeration
can be induced in a powder by inhomogeneously distributing a wetting
liquid. Most commonly, a condensed vapor will form pendular bonds
between particles. These bonds strengthen the particle cluster, but
inhibit dense packing. This is a particular problem with submicron-sized
powders exposed to humid air.
Agglomeration makes mixing more difficult. An alternative is to create
conditions that give rise to repulsive forces between particles by using thin coatings
of polar molecules. Surface repulsive forces contribute to high packing densities by
reducing the interparticle friction found with powders possessing cohesive forces.
Particles agglomerate into clusters of high coordination number separated by high
porosity regions, as sketched in Figure 2.2.3. Although this is an idealized situation,
many examples of agglomeration are evident in submicron powders.
Agglomeration occurs mostly with smaller particles, because of a high
surface area and the action of one of the weak forces. The common weak forces
are van der Waals attraction, electrostatic charges, chemical bonding, capillary
liquid forces or magnetic force. The van der Waals force is significant for
particles below 0.05 m in size, Agglomeration can also occur during mixing
due to cold welding at the particle contacts.
43
Figure 2.4.1 A sketch showing the large pores associated with the interagglomerate in an agglomerated
spherical powder the overall packing density is reduced by agglomeration.
Most typically, particles agglomerate due to adsorbed surface films,
especially during periods of agitation. Atmospheric vapors condense and establish
a surface concentration dependent on their partial pressure. Water is the most
typical vapor to condense on a powder. The amount of water adsorption depends on
the relative humidity and the surface curvature. Four cases of adsorption are
possible. First, at low vapor pressures the powder surface IB uniformly coated
with a thin layer of adsorbed vapor. Second, as the partial pressure increases, a
critical level is reached where the vapor condenses to form capillary bridges
localized at the particle contacts. These bridges are termed pendular bonds and are
shown schematically in Figure 2.4.2a.Third, as the vapor pressure increases, the
funicular state occurs. In the funicular state the pendular bonds merge, but the
pores are less than totally filled by liquid. In this state the pores are smooth and
surrounded by liquid. A connected path exists in the both the vapor and liquid
phases as shown in Figure 2.4.2.b, with the vapor phase existing as a cylindrical
shape. Finally, when the pore structure is saturated, the pores are filled with
liquid. This case is shown in Figure 2.4.2.c. For water vapor and spherical powders,
the-pendular bond state is anticipated at relative humidity levels between
approximately 65 and 80%.
44
Figure 2.4.2 The states associated with agglomeration of powders due to wetting liquid: (a) pendular, (b) funicular, and (c)
saturated
A negative consequence of agglomeration is a decrease in, the packing
density. This is a particular problem with small particles. Clustering of particles,
especially those of similar size, is an unexplained characteristic of many structures. As
the particle size decreases and the number of particles in each agglomerate
increases, the agglomerates become stronger and exhibit a lower maximum packing
density. Besides decreasing packing density, agglomeration also creates problems
with mixing, settling from suspensions, compaction, and sintering, especially for
those systems with wide particle size distributions. The sensitivity to the particle size
distribution arises because the small particles are the primary cause of
agglomeration. In a wide particle size distribution, the smallest particles can exert a
strong effect on the larger particles. Alternatively, agglomeration can be selectively
used to minimize size segregation in powder handling.
2.2.5. SURFACE ACTIVE AGENTS
Small quantities of surface-active agents are often added to particles to alter
packing or mixing characteristics. Some common additives are polyvinyl alcohol,
stearic acid, sodium oleate, glycerine, and oleic acid. These additives reduce
interparticle friction by lubricating surfaces via short-range repulsive forces.
Generally, the flow and packing of particles are improved by the presence of' the
appropriate surface-active agent. The level of improvement is dependent on the
molecular size of the additive, its polar character, the layers of coverage, the particle
surface condition, the particle size, and the temperature. Polar molecular
45
coatings with short-range interactions aid in keeping particles from agglomerating.
This is most important with the submicrometer particles that are dispersed in a
fluid for processing, such as in slip casting. The particle dispersion is
maintained by use of short range repulsive forces, possibly induced by surface
charges as measured by the zeta potential. The viscosity and surface energy of
the additive are not as important as the polarity and wetting ability. Agglomeration
is less of a problem with a narrow particle size distribution and larger particle sizes.
In contrast, sticky particle surfaces will have a high level of agglomeration, leading to
a lower particle packing density.
Intermediate chain-length polymeric molecules are used to lubricate powder
compaction, where they are used in high concentration. Alternatively, at low
concentrations small organic molecules are optimal because of their polar
character. Generally, a backbone chain length of approximately ten carbon
atoms is optimal as a surface additive for inorganic particles. Even so, for large
particles the relative packing benefit is low.
2.2.6 . INTERNAL POWDER POROSITY
Many powders contain internal porosity, which is sometimes totally
isolated from the powder surface. Shows sketches of cross-sectioned
particles with differing internal pore structures: a) fully dense, b) entrapped
pore, and c) open pore. Then closed porosity will not. Interact with a penetrating
f l ui d. In contrast the sponge-like structure has vapor phase communication
from the external surface to the inner pores. Figure 2.6.1 contains two optical
photographs of cross-sectioned and polished iron' powders.
46

Figure.2.6.1 sketches of three forms of the particles with varying internal pore structures (a) dense (no
pores) (b) closed internal pore and (c) open internal pore.
The packing laws for particles ignore the internal porosity within the particles;
the focus is on the interparticle pores with the assumption of dense particles. If a
particle contains internal pores, then the intraparticle porosity will degrade the
packing density. Thus, in mixed particle size systems, the predicted mixture density
must be corrected for the internal porosity. Let
s
be the fractional solids content
for the powder, giving the fractional intraparticle porosity as 1
s
Then, with f as
the predicted external fractional packing density (interparticle porosity is (1
e
),
the overall packing fractional density is,
=
e
.
s
This equation says that the overall packing density is the product of the
fractional densities of the particles times the particle packing density. Generally in
this treatment, the focus will be on the interparticle porosity. However, when porous
particles are involved in the packing, the actual solids content will be reduced by
the intraparticle porosity. For very porous particles, this correction can be
substantial.
Beyond lowering packing density, internal porosity can influence several other
attributes. For example, the open pores will increase the measured specific surface
area. These are typically much smaller pores than the interparticle voids;
consequently adsorption and trapping of fluids will be favored in these
intraparticle pores. These will preferentially collect and retain wetting fluids. In a
sense, agglomerated particles represent one case of powders with internal pores.
47
2.2.7 CONTAINER WALL
The container used to hold a powder will induce a local area of order at the
container wall in an otherwise random packing. The effect is more pronounced for
flat, smooth containers, giving local regions of oscillating high and low porosity in
the first few particle layers near the wall. Even in flexible containers there is
ordering near the wall. Besides packing density, the container wall influences
measurements because of the low-density regions near the wall. The packing
coordination is higher in the first few layers of particles near a container wall.
Figures 2.7.1 and 2.7.2 illustrate the magnitude of the container wall effect in terms
of the fractional packing density. Figure 2.2.6plots the local packing density versus
distance from the wall in sphere diameters. This decaying oscillation in packing
density with distance has been confirmed in several studies. The wall effect persists
for several particle diameters into the packing. This same influence is evident at the
interface between a crystallized solid and its liquid; damped density oscillations can
be seen in the liquid bordering on the solidification interface. Figure 2.2.7 shows the
integral packing density versus distance for spheres packed in a cylinder. The inte-
gral density decreases from a low value at the wall to a constant value of
approximately 0.64 within the first two particle diameters.
Figure 2. 7. 1 The l ocal fractional density versus the distance from the container wall for monosized
spheres packed in a cylindrical container.
48
Figure 2.7.2 The integral fractional packing-density for spheres in a cylindrical container.
There is less sensitivity of the integral density in comparison with the
local density shown in Figure 2.71.
Depending on the particle size, particle shape, and container shape, it takes
from one to ten particle diameters from the wall to establish truly random
packing. The effect is larger at higher packing fractions. In each case, as the
randomness increases, the packing density approaches an asymptotic value as the
ratio of container diameter to particle diameter increases. The
effect of the container wall on the packing density has been expressed in various
mathematical forms. For a given container, the container size influence on
density increases with the particle size and with the surface area of the
container. This same behavior can be seen when a close-packed structure
is separated. The resulting unfilled depressions on the surface of the
close-packed layer result in a slight density decrease. For spherical particles
the results by Ayer and Soppet fit the following equation:
f = 0.635 0.216 exp(-0,313 D
c
/D) (2.2)
Where D is the sphere diameter, D is the container diameter, and f is
the factional packing density. They found the overall packing den sity to be
within 1.5% of the calculated value.
For nonspherical powders there is inherently more randomness to the
packing; thus, the wall effect decays over a shorter distance from the container
wall. Equations for rough spheres and cylinders are analogous to those shown
above for hard, monosized spheres. Mixed particle sizes aid in minimizing the
49
porosity variations at the container wall. Accordingly, as the width of the particle
size distribution increases, there is less influence from the container wall

Figure 2.7.3 Fractional packing density for random loose and random dense packings
of steel spheres, showing the extrapolated limiting densities with no container wall effects
2.2.8 SEGREGATION
Another concern in studying powder-packing characteristics is segregation in
mixed powders. This can lead to uneven packing densities and to distortion in
compaction and sintering. There are three causes of segregation: differences in
particle size, density, and shape. Of these three, size segregation is dominant. For
example, a powder poured into a container will segregate by size if the small
particles can pass through the holes between the large particles. This leads to
banding, especially at low filling rates.
A common cause of size segregation is agitation. In a randomly mixed
powder with differing particle sizes (or with the presence of agglomerates), the
larger particles Segregate to the top of the container when the particles are agitated.
Such segregation is most pronounced when there is a vertical component to the
agitation. The agitation opens local interstices between particles, making it more
probable for the small particles to fall by gravitational action into these interstices.
This process enriches the lower gravity positions with smaller particles over time,
leading to a rise in the large particles as illustrated in the figure 2.8.1
50
Figure 2.8.1 Size segregation of mixed powders occurs by percolation of the smaller
particles to lower gravity positions during agitation. The net effect is a rise of the larger
particles to the container top.
2.2.9. BRIDGING AND VIBRATION
The low packing density found in many powders is attributed to bridging, which is the
creation of large pores by arching, such as shown in the two-dimensional schematic of
Figure 2.9.1. A nonspherical particle shape, especially an angular shape, will increase
the tendency of a powder to bridge. Likewise, as the particle size decreases there is less
mass to an individual particle and more surface friction. Both factors contribute to a
lower packing density. Bridging is most apparent near container walls, where it gives
voids larger than the particle size.

Figure 2.,9.1 A schematic diagram of particle bridging and the creation of large
pores because of high interparticle friction, irregular particle shapes, or
particle cohesion
To eliminate the large pores associated with bridging, agitation is used to
bring a powder to a high packing density. Vibrating the container without a direct
51
external pressure on the powder performs this. This density requires the particles
to attain a close proximity to one another. The tap density is generally accepted as
the maximum density attainable under vibration or agitation. The tap or vibrated
density depends on the material, vibration amplitude, vibration direction, applied
pressure, vibration frequency, particle density, shear, and test apparatus. Powders
will reach the dense random packing limit more rapidly as the particle size
increases. Likewise, a dense packing is aided by a spherical shape and a lack of
interparticle cohesion.
For any powder there is an optimal condition for attaining rapid packing.
The angle, type, amplitude, and frequency of vibration must all be considered;
their effects vary between powders. The optimal procedure is to vibrate the
powder with both a lateral and an axial component. Generally, the most rapid
packing occurs with acceleration near 6 to 7 times gravity, and with an amplitude
near 1 mm. Many powder systems maximize their rate of densification with a
specific frequency. Best packing is attained with frequencies over 300 Hz for
submicron powders, 200 to 300 Hz for intermediate particle sizes, and 100 to 200
Hz for large particles over 100 m. Because of this dependence on vibration
frequency, variable vibration frequencies are sometimes employed, especially with
particle mixtures. A three-component mixture of particles has been vibrated to a
fractional density of 0.94
2.3.PACKING OF MONSIZED SPHERE
2.3.1 COORDINATION NUMBER
The determination of the packing coordination of monosized spheres is one of
the classic problems encountered in science. Monosized spheres can be packed in an
orderly close array with a coordination number of twelve and a fractional density of
0.7405. Generally, for geometric stability the packing coordination can not go
below four. A random packing assembled under the action of gravity requires each
sphere that joins the packing rest on at least three contact points. Thus, each sphere
adds three new contacts to the packing. Since two spheres share each contact point,
the net increase in packing coordination number is six contacts per sphere.
52
Accordingly, an average coordination number of six is expected for a random yet
stable packing of monosized spheres. This value is generally substantiated by
experiments using large spheres, including packings of mixed sphere sizes.
2.3.2 ORDERED PACKING
In analyzing the packing of monosized spheres, first attention will be given
to ordered packings. These are well known because of their similarities to hard
sphere models of crystalline atomic structure For example, Figure 3.2.1 shows the
geometries of regular ordered packings with coordination numbers of six and
twelve.
Figure 3.2.1 The structure of two ordered packings of monosized spheres; a) simple cubic packing
with a coordination number of si x and b) face-centered cubic packing with a coordination
number of twelve.
Several simple unit cells characterize packing and pore structures. Figure
3.2.2 shows their six basic packings, corresponding to cubic, orthorhombic (two
forms), rhombohedral (two forms), and tetragonal geometries.
Figure3.2.2 the unit cells of the six packing geometries introduced r for construction
of random packings; cubic, orthorhombic (two forms), rhombohedral (two forms) and tetragonal
geometries.
53
When the spheres are compressed they will form new contacts only if
particle rearrangement can occur. Ordered packings of monosized spheres at a high
coordination number do not show an increase in coordination with compression; the
final geometric shape is usually a dodecahedron (12 sided polyhedra).
Various models linking the coordination number and packing density are
possible for ordered packings at integer coordination numbers. Table 3.2.1 collects
the recognized solutions for standard even and odd coordinations.
Table 3.2.1 The Densities and Coordination Numbers for Ordered Packings
Packing type Coordination
number
Packing
density
Porosity
Face-centered cubic
12 0.7405 0.2595
11 0.7120 0.2880
Tetragonal-sphenoidal
10 0.7081 0.3019
9 0.6134 0.3866
Body-centered cubic 8 0.6802 0.3198
Orthorhombic
8 0.6046 0.3954
7 0.5612 0.4388
Simple cubic
6 0.5236 0.4764
5 0.4031 0.5969
Diamond 4 0.3401 0.6599

Figure 3.2.3 the fractional packing density versus the coordination number for
ordered packings of monosized spheres.
As the coordination number increases, the spacing between layers of spheres
decreases. From Figure 3.2.3 it is clear that the coordination number varies with
54
the fractional solids content. Although the proof is difficult, it is mathematically
accepted that the maximum possible packing density for equal size spheres is
given by a coordination number of 12. This packing structure has special
characteristics because of the 48 symmetry groups. Furthermore, the close-
packed structure is a mixture of tetrahedral and octahedral units with two
tetrahedral units for every octahedral unit.
Several mathematical approximations have been proposed to link the packing
density and the coordination number for mono sized spheres. Most of these are
based on the correlations using data for ordered packings. There si no exact
relation between packing density and the corrdination number. Various simple
models have been applied to data such as shown in Table 3.2.2 to approximate
the relation of coordination number N and fractional density f. Some of these
expressions are collected in Table 3.2.2. The relations range from the simple to
complex. Not included in Table 3.2.2 are several polynomials that involve
undefined coefficients or infinite series.
Table 3.2.2.mathematical expressions for the packing density and the average
coordinaions number for monosized shperes
Equation1 f= -0.072 -f 0.1193 N
c
- 0.00431 N
c
2
Equation2 N
c
= /(1-f)
Equation3 N
c
= 26.49 - 10.73/f
Equation 4 N
c
= 2 exp(2.4 f)
Equation5 f=((N
c
-l)/N/N
c
)
3
Equation6 f= 0.4696 + 0.02539 N
c
Equation7 N
c
= 1 6 f - 2
Equation 8: N
c
= 19. 7 f- 2 . 5
55
When the spheres are compressed they will form new contacts only if
particle rearrangement can occur. Ordered packings of monosized spheres at a high
coordination number do not show an increase in coordination with compression; the
final geometric shape is usually a dodecahedron (12 sided polyhedra). In contrast,
compressed random spheres deform to polyhedra with 14 sides
2.3.3 RANDOM PACKING
There are three recognized random packing densities for monosized spheres.
Loose and dense random packing concepts have already been introduced. The third is
defined by the density at the onset of fluidization, which ranges from fractional
densities of 0.53 to 0.54. Of the three densities, the loose and dense random packings
are important to most uses of particulate materials. Various reports have placed the
fractional densities at 0.610 to 0.667 for the random dense packing, while values
between 0.560 and 0.625 are cited for the random loose density. One difficulty in
dealing with random structures is that several different particle arrangements give
similar densities and other attribute. As more energy is put into assembling a
particle packing by cascading or impacting particles, the fractional density
increases to the random dense packing limit.
In scientific studies on the packing of spheres there is considerable interest in
the radial distribution function. The character of this function can be related to the
structure of liquids and glasses. The radial distribution function is the average
number of sphere centers per unit area in a spherical shell about a central sphere.
Figure 3.3.1 plots the radial distribution function for both random loose and
random dense packings of monosized spheres.
56
Figure3.3.1 The radial distribution of monosized spheres randomly packed in both the loose and
dense structures. The number of particles perunit spherical area is shown versus the
normalized radius.
As expected for a random packing, there is a uniform probability distribution as the
distance increases. As noted in Figure 3.3.1 the radial distribution functions for the
random dense and loose packings of monosized spheres are similar.
Random packings are characterized by clusters of high packing coordination
and small pores embedded in a matrix of low packing density and low packing
coordination. Such clustering is noted in soap bubbles and polycrys-tailine solids,
although it is generally assumed that random packing has no correlation in particle
placement beyond the first nearest neighbor region. The cumulative radial
distribution function provides another measure of the packing coordination number
behavior in a random packing.
Many observations of particle packing are made on random cross sections
through the packing. It is reassuring that the size distribution of the cross-sectioned
circles in the simulations agree with theory for the size distribution for monosized
spheres in random cross sections. Figure 3.3.2 plots the cumulative distribution of
circle sizes seen in two dimensions for random cross sections through monosized
spheres. In such two-dimensional cross sections there is an apparent slight
clustering of the particles as can be seen in Figure 3.3.3. The local clusters indicate
a distribution in pore sizes. In a two-dimensional section both the pore size and
57
particle size will appear smaller than actual. Even so, the presence of large pores is
apparent in Figure 3.3.3.
Figure 3.3.2 The cumulative frequency distribution of circle sizes expected from a
random cross section through a packing of monosized spheres
Figure 3.3.3 An example of the particle
clustering and circle size distribution
apparent in a random cross section of
monosized spheres
2.3.3.2 Random Loose Packing
In practice the apparent density best represents the random loose density.
Several investigations have reported the fractional density for random loose packings
of monosized spheres. This density is very sensitive to experimental
techniques. As aconsequence, there is considerable variation between
measurements. The values range from 0.660 to 0.625, with two-thirds of the
reported values between 0.680 and 0.606 .The average of the reported fractional
densities is 0.593 with a standard deviation of 0.014. Most of these measurements
have been corrected for wall effects on packing.
58
A problem with the random loose packing is that the observed density has
a high dependence on the experimental procedure. The measurement is influenced by
the kinetic energy of each particle as it enters the bed; thus, higher densities are
observed with particles of higher mass or higher velocity. The theoretical predictions
are as widely scattered as the experimental determinations.
The most probable value for the fractional packing density on a random loose
array of monosized spheres is between 0.59 and 0.60, with approximately six
contacts per sphere. There is an inherent variation point-to-point in the packing
density and coordination number. The mean and mode coordination numbers are
near six and the distribution is approximately binomial.
2.3.3.3 Random Dense Packing
The fractional density for a random dense packing is better defined than that
for the loose packing, since it is a limiting condition where nearest neighbors are in
contact. A random dense packing for monosized spheres corresponds to their
maximum density without ordering or deformation. The fractional packing density
for the random dense case is 0.637. Early experimental measurements of the
packing density did not always correct for wall effects or other sources of error
such as surface roughness on the spheres. Accordingly, reports of fractional
density vary from 0.610 to 0.665, but largely cluster around 0.637; two-thirds of
the reported values are between 0.636 and 0.644.
The coordination number for thq dense random packing is estimated to range
frorp six to nine. The most accurate measurements give a mean and mode value
near six with a standard deviation near 1.1 contacts. As noted earlier, there is an
experimental problem with determining the existence of an actual contact versus a
near contact. The computer simulated coordination numbers range between six
and seven contacts per sphere, and tend to be lower than some of the experimental
determinations. However, the recent computer simulations which converge to a
fractional density of 0.637 give a coordination number near six.
59
When a random dense packing is subjected to shear, the density increases
slightly although the coordination number appears to remain constant. The
fractional density at the onset of shear deformation is between 0.66 and 0.67
The pore structure for a random dense packing is often modeled as a
collection of various sized tetrahedral units. This is valid since the radial position of
neighboring spheres proves to be highly correlated. There is a high probability that
contacting spheres will have separation angles of 2/3, or /3. The geometric
model based on dilated tetrahedral units has proven successful in reducing the
dense random packing into an easily analyzed structure.
A random dense packing occurs when the particles attain a close proximity to
one another through agitation. Accordingly, the tap density is a good first
approximation of the maximum loading attainable without application of an external
stress. The tap or vibrated density depends on the material, vibration amplitude,
vibration direction, frequency, shear, and test apparatus. During vibration the
density varies with the number of vibrations. The random dense packing and
random loose packing are fairly similar for large spherical particles.
Consequently, spherical particles of sizes greater than approximately 100 m
undergo little densiflcation during vibration. In contrast, smaller spheres and
nonspherical particles exhibit a greater difference between the apparent and tap
densities. These particles undergo a large packing increase with vibration. Indeed,
the more irregular the particle shape, the greater the increase in packing density with
vibration, but a longer vibration period is needed to reach the tap density.
2.4. PACKING OF MONOSIZED NONSPHERICAL PARTICLES
Particles with a rough surface texture or shape are more susceptible to
adsorption of surface films and to agglomeration. Along with an increase in surface
60
texture goes an increase in the interparticle friction. Accordingly, the packing density
and coordination number decrease. The surface friction between particles can be
characterized by an angle of friction, one form of which is the angle of repose
2.4.1 PARTICLE SHAPE EFFECT
Because of the higher interparticle friction, nonspherical particles give a
lower random packing density and coordination number than spherical particles.
Figure 4.1.1 shows the fractional packing density for various monosized irregular
particle shapes. As the particle shape becomes more rounded (spherical) the
observed packing density increases. Consequently, powders with a highly irregular
particle shape do riot match the packing density for spheres. Indeed, the range
between random and ordered packing densities increases as the particle shape
becomes nonspherical. Spherical particles are more desirable in applications that
require a high packing density or low interparticle friction.
Figure 4.1.1 A plot of the
fractional density versus the relative
roundness for randomly packed monosized
particles
If the particle has a regular geometric shape, then a decrement in packing
density is not always observed even though the shape is not spherical.
Nonisotropic particles can be packed to high densities if they are ordered. For
example, cubic or rectangular particles can be packed to a fractional density of
1.00 when placed in perfect packing. Other polygonal shapes also exhibit high
packing density structures.
The random packing of fibers provides an illustration of a decreasing packing
density as the particles have a larger length-to-diameter ratio L/D. and 4.1.3 plot the
fractional packing density versus the length-to-diameter ratio for fibers . The first
61
figure is scaled linearly to better emphasize the behavior of short fibers, while the
second figure extends to large values using a logarithmic scaling. These figures show
that fractional densities as low as 0.10 result from the random packing of fibers with
length-to-diameter ratios over approximately 50.The highest packing densities
and the most isotropic structures are observed with equiaxed particles. In the
extreme example of very long fibers with a length-to-diameter ratio over 100, the
packing density decreases to 0.005 or less.
Figures 4.1.2 The fractional packing density
of fibers decreases as the length-to-
diameter ratio increases
Figures 4.1.2 A logarithmic plot of the
fractional density versus the length-to-
diameter ratio for fibers in the random
dense packing condition
Generally, the wall effect on packing density is less pronounced with nonspherical
particles. A comparison with the wall effect for spherical particles shows less of an
effect. The packing density f depends on the relative container size D as follows:
f =A- (K/D) - B exp(-C D
c
/D) (4.1)
where D is the particle size and the constants A, B> C and K depend on the particle
shape. This equation involves both the particle size and container size effects.
Obviously, there is an influence of particle size on packing density for angular
62
particles. For angular particles in an infinite container, Equation 4.1 can be modified
to give
f= A-(K/D) (4.2)
with A approximately equal to 0.58. The parameter K varies with particle shape. For
angular, equiaxed particles it is approximately equal to 7 m, and for tetragonal
particles it is near 32 m. Equation 4.2 clearly demonstrates the particle size
effect on packing density. Larger particles and spherical particle shapes give
higher packing densities. For fibers, the density decreases as the length-to-
diameter ratio {L/D) increases Figure 4.1.2 shows the systematic decrease in
packing density with fiber length-to-diameter ratio.
Clearly, changes in the particle size and particle shape both affect packing ,
but particle shape is usually most important. For the typical case of a nonisotropic
irregular solid, the nonspherical particle shape gives a lower packing density compared
to a shperical shape. Figure 4.1.3 sketches six simple shapes and tabulates the
corresponding sphericity index.
The packing density follows a pattern similar to that observed with the
interparticle friction; a lower density correlates with a lower coordination number
and a higher angle of repose. The random dense packing of spheres is contrasted with
various other geometries in Table 4.1.1. The.simple and equiaxed geometric shapes,
such as cylinders and cubes, have random dense packings that rival those of spheres.
Alternatively, the more irregular particle shapes have much lower
%
packing densities.
As with all packing structures, there are numerous factors, which influence the actual
density; however, the role of particle shape is dominant, as is evident from table
4.1.1
63
Figure 4.1.3 Sketches of six simple particle shapes with the sphericity index
indicated for each.
Table 4.1.1. Particle Shape and Random Dense Packing
Particle shape
fractional density
Spheres 0.64
Cubes 0.72 to 0.79
Rectangles (2x5x10) 0.51
Plates (4x4x1) 0.67
Plates (8x8x1) 0.59
Tetrahedron 0.41 to 0.64
Cylinders (L/D= 167) 0.03
Cylinders (L/D = 60) 0.09
Cylinders (L/D = 25) 0.21
Cylinders (L/D=U5) 0.28
Cylinders (L/D-10) 0.48
Cylinders (L/D = 5) 0.52
Cylinders (L/D=l) 0.60 to 0.67
Disks (L/D = 0.5) 0.62 to 0.64
Dagger shaped fibers 0.33
Irregular filaments 0.06 to 0.35
Irregular, isotropic 0.55 to 0.63
Irregular, nonisotropic 0.30 to 0.50
2.5. BIMODAL MIXTURES OF SHPERICAL PARTICLES
64
Such simple experiments illustrate that bimodal particle combinations can
pack to higher densities than can monosized particles. The key to improved
packing is the particle size ratio. Small particles are selected such that they fit
into the interstices between larger particles without forcing the large particles
apart. In turn, even smaller particles can be selected to fit into the next level of
pores, giving a corresponding improvement in packing density. Figure 5.1 urther
illustrates the basic concept using an ordered packing of two-dimensional disks to
repreaent three-dimensional spherical packing. In the first case, shown in Figure
2.5.1a onosized disks are packed into an orderly two-dimensional array with pores
between them. In the binary packing case of Figure 6.2b, the secondary disks are
of a smaller size selected to fill the pores between the larger disks without
distorting the original packing. Accordingly, a ternary packing, such as shown in
Figure 5.1 c can be created using even smaller disks that fill the pores in the
binary case. However, as noted in the lower right drawing, Figure 5.1 d , the
selection of too large a disk will not improve the packing density since the added
disks force the primary structure apart.
Figure2. 5.1. Two-dimensional
representations of the effects from
combinations of differing sizes; a)
monosized, b) bimodal with a large
difference, c) trimodal with a large
difference, and d) bimodal with a
small difference.
2.5.1 QUALITATIVE DESCRIPTION
65
As illustrated in figure 2.5.1. packings of monosized spheres leave holes
between the particles. In most cases, the density of the random packing can be
improved by the addition of smaller spheres. The requirement is that the void spaces
be filled without dilating the overall volume.
For an ordered packing, the diameter of the sphere that will fill the
interstitial pore decreases as the packing coordination increases. Table 5.1
summarizes the effect of adding selective small spheres to three ordered packings.
In the case of a coordination number of twelve, two different interstitials exist
(octahedral and tetrahedral); thus, two sphere sizes and contents are needed for
minimal porosity. For optimal packing the number of small particles increases
as their relative size decreases. The rules for constructing ordered packings from
various size ratios of particles follow exactly from those deduced for ionic crystals.
In both cases, the porosity is minimized by adding a specific amount and.size of
interstitial spheres. However, if the packing is hot ordered, then the actual packing
density tends to be less than that predicted by Table 5.1. Indeed, this provides a
basis for constructing models of ionic glasses using randomly oriented unit cells
containing the appropriate stoichiometry.
Table 5.1Interstitial Sphere Packing Density Improvements for Ordered
Packings
Base structure Initial
poros
ity
Diameter
added
Volume
added
Mixtures
porosity
Cubic N
c
=6 0.476 0.723
0.391 0.271
orthorhombic
N
c
=8
0.395 0.528 0.147 0.307
rhombohedral
N
c
=12
0.260 0.225
0.414
0.019
0.070
0.190
The basic behavior for random dense packing is sketched in Figure 5.1.1. The
packing volume, termed the specific volume (volume to mass ratio), is shown as
a function of the composition for a mixture composed of large and small spheres.
There is a, composition of maximum packing density. The relative improvement in
66
packing density depends on the particle size ratio of the large and small
particles. Within a limited range, the greater the size ratio of large to small
particles, the higher the maximum packing density. This is true up to a limiting
size ratio of approximately 20:1, but requires at least a 20% difference in particle
sizes to occur , Using Figure 5.1.1, the specific volume versus composition can
be analyzed to determine the optimal composition. Assuming dense particles, the
specific volume of a powder V is defined as the inverse of the apparent density.
V = l/(pf) (5.1)
where p is the theoretical density For the powder and f is the Fractional density For
the powder.
Figure 5.1.1 The reduction in specific volume for mixed large and small spheres, showing the
condition of optimal packing where the small spheres fill all voids in the large sphere packing.
Beginning with the large particles, the specific volume initially decreases as small
particles are added because the smaller particles fill the unoccupied voids between the
large particles. A plot of the specific volume versus composition for this process is
given in the right-hand portion of Figure 5.1.1. Eventually, beyond the optimal
packing,, the quantity of small particles becomes too great since all of the voids are
67
filled. Accordingly, further additions force the large particles apart and no longer
improve the packing density.
In contrast, starting with a container filled with the small particles,
clusters of small particles and their associated voids canbe removed and replaced
by large particles. Since the large particles are fully dense, this process leads to an
increase in density" because of the simultaneous elimination of voids between small
particles. Hence, a porous region is replaced with a fully dense region everywhere
a large particle is added. The packing benefit from replacing small particles with
large particles continues until a concentration where the large particles contact one
another. Figure 5.1.1 shows this process as the left-hand plot. The point of minimum
specific volume for the mixture corresponds to the intersection of these two curves.
2.5.2 QUANTITAVTIVE MODEL FOR OPTIMAL PACKING
The maximum packing density for a bimodal mixture of spheres
corresponds to the minimum specific volume. This is termed the saturation
point. At saturation the large particles are in point contact with one another and
all of the interstitial voids are filled with small particles. A mathematical
description of bimodal packing starts by designating the large particles with the
subscript "L" and the small particles with the subscript "S". Calculation of the
saturation composition in terms of the weight fraction of large particles X is an
obvious goal of the mathematical treatments. The weight fraction of large particles
at any composition depends on the following equation:
X
L
=

W
L
/(W
L +
W
S
)
(5.2)
with W indicating the weight. The weight of large particles is calculated from the
theoretical density p and fractional packing density f
L
a s
W
L
= f
L

L
. (5,3)
For maximum packing density the desire is to add sufficient small particles to
just fill the void space between the large particles without forcing the large particles
68
apart. The amount of void space equals I - f and the fractional packing density for
the small parti-cles times the small particle theoretical density gives its weight frac-
tion as
W
S
=(1 f) f
S
.
S
(5.4)
Thus, for saturation X
L
is given as follows
X
L
= f
L

L /
(f
L

L +
(1
-
f
L
) f
S

S
) (5.5)
As illustrated in Figure 5.2.1 the fractional density is dependent on the composition
and particle size ratio. The larger the particle size ratio, the higher the packing density
at all compositions.

FIGURE 5.1.1 A plot of the fractional density versus composition for six particle size ratios
2.5.3 POROSITY AND PARTICLE SIZE RATIO EFFECTS
For a binary mixture of spheres, the amount of small particles at the
saturation point varies with the porosity of the pure constituents. As a first case,
consider two powders with a large difference in particle size that exhibit random
dense packing with the usual fractional density of 0.637. The corresponding weight
69
fraction of large particles for maximum packing is 0.734, or 26.6 wt.% of the smaller
particles.
The fractional packing density variation with the particle size ratio and
composition has been curve fit as follows for the compositions rich in larger
particles.
f = 0.64/[I - (0.362 - 0.315 (D
S
/D
L
)
0.7
X
L
+0.955 (D
s
/D
L
)
4
(X
2
L
/(1 - X
L
))] . (5.6)
In this equation, D is the particle diameter with the subscript denoting
either the large or small sphere, and X is the fraction of large spheres. For the case
of an infinite particle size ratio, then the relation becomes much simplified:
f= 0.64/(1 - 0.362 X
L
). (5.6)
This assumes that the particles are spherical with a monosize random dense
packing fraction of 0.64.
There is no gain in packing density if the particles have the same
size.Alternatively, the packing benefit is maximized by a large difference in
particle sizes. ). In many instances the inverse ratio is used in analyzing packing
effects since this compresses the size scale at large size differences. Since the highest
binary packing density occurs when the larger particles provide a fixed bed for the
smaller particles to fill, the smaller the relative size of the small particles the better
the packing.
In randomly mixed particles there is a complication in predicting the mixture
density because of localized distortions. The small spheres preferentially collect on
the surface of the large spheres. As a consequence, the small sphere packing
structure is distorted at the interface in the otherwise random packing. The large
particle surface acts as a curved boundary that induces nonrandom structure in
the small spheres near the interface. Associated with the nonrandom packing
structure at the interface is a decrease in the local density similar to that
encountered at a container wall. Since the small spheres do not efficiently cover the
large sphere surface, the localized fractional density is approximately 67% of the
inherent density for the small particles. The actual distortion in the small sphere
70
packing depends on the particle size ratio. Because of this localized surface effect,
the composition for maximum packing density in a random mixture tends to be
slightly shifted to a lower content of large particles.
2.5.4 COORDINATION NUMBERS
Under optimal conditions, the maximum fractional packing density for a
binary mixture of spheres has been estimated to be between 0.84 and 0.88. The
coordination of the two constituents depends on the amount and size ratio of the
particles. As the particle size ratio of large to small particles D
L
/D
S
goes up, the
packing coordination of the large particles increases. For example, with a particle size
ratio of 1000, the coordination number of a large particle surrounded by small
particles exceeds 3 million contacts. At a particle size ratio of 100, the large particle
will experience over 31,000 contacts with small particles. Based on computer
simulations, the coordination number N of the large sphere at optimal packing
can be estimated as follows:
N
c
=3.0 + 3.16(D
L
/D
S
)
2
(5.7)
This equation gives a coordination number of 5.8 when the particles are of
equal size
N
c
=3,5 + 1.25(D
L
/(n
S
D
L
+ n
L
D
L
))(1 + D
L
/(n
S
D
L
+ n
L
D
L
)) (5.8)
where n
s
is the number fraction of small particles and n
L
is the number
fraction of large particles. Across the binary mixture the coordination number for
the small particles increases from 3 to 6, and the coordination number for the
large particles increases from 6 to 25.
2.5.5 HOMOGENEOUS EFFECTS
The homogeneity of a mixture has a major influence on the packing density It is
clearly established that the density varies with the composition for mixtures of
differing particle sizes. Consider that in an inhomogeneously mixed packing some
regions;will have more large spheres than others. Consequently, the fractional
71
density will vary from point to point in proportion to the localcomposition. The
typical structure of mixed partides will have random packing and the resulting
segregation orfluctuations in local composition will result in fluctuations in
density .Because of this influence, models of packing density for bimodal mixtures
overestimate that attainable in random mixtures.
There is a particular problem in mixtures formed without vibration since the
small spheres may actually cause a dilation of the mixture, as sketched in Figure
5.5.1. The bimodal mixtures introduced for high packing densities assume the small
spheres are sited in the interstitial voids. This becomes a special problem for those
mixtures in which the particles are similar in size since the small particles can not
fall through the constricted openings between the large particles.
Figure 5.5.1. A sketch showing how small particles can inhibit optimal packing due to
improper vibration into interstitial sites
As the fractional packing density of the large particles increases, the size of the
opening decreases; for a simple cubic lattice the small sphere size m vis t be less
than 1/2.4 of the large sphere size to pass through the interparticle throat
constriction. If the large sphere packing is close-packed cubic, then the small sphere
must be less than 1/6.5 of the large sphere to pass through the interparticle
throat constriction. In both cases, the sphere size that can fit into the pores is
larger than the size of the sphere that can pass through the constrictions between
the large spheres. The density
improves with the degree of mixing. The upper limit is based on perfect
mixing, while the lower limit assumes a totally segregated structure. In the
72
unmixed or segregated condition, the fractional packing density is linearly dependent
on the composition and weight fractions of the two powders.

S S L L u
f X f X f / /
1
+

(5.9)
where f
U
is the fractional density for the unmixed system, composed of X
L
weight fraction of large particles with a fractional density of f
L
and X
S
as the
volume fraction of small particles with f
S
as the fractional density. If the
fractional packing densities for the large and small particles are the same, then in
an unmixed system the packing density will be unaffected by composition. Randomly
mixed systems will range between the unmixed and fully mixed limits, depending
on the mixture homogeneity.
2.6.BIMODAL MIXTURES INVOLVING NONSPHERICAL
PARTICLE
A major difference between spheres and nonspherical particle shapes is that
the initial monosize packing is generally higher for the spheres. The greater is the
surface roughness; shape irregularity, or particle aspect ratio, then the lower the
inherent packing density. Thus, although the relative gain in den-. sity is similar
for spheres and nonspherical particles, the starting density for nonspherical particles
is lower. Accordingly, at all compositions the nonspherical mixture will be lower in
density. The relative gain in density for binary mixture (of either spheres or
nonspheres) increases as the difference in particles sizes between the two
powders increases. with a mixture of spherical and nonspherical particles, the
behavior has a sensitivity to shape that influences the size ratio for optimal packing.
This behavior is especially important for the fabrication of composite materials.
Hence, the first case considered in this chapter is the packing of spheres and fibers.
2.6.1 MIXTURES OF SPHERES AND FIBERS
73
Mixtures of fibers and spherical particles provide the basis for fabricating
many high performance composite materials, especially those with ceramic or
metallic matrices. In those cases involving aligned fibers, the packing behavior can
be treated as a two-dimensional problem.It is beneficial to the consolidation process
to start with a high packing density. The maximum packing for sphere-fiber
mixtures requires attention to the particle size, fiber length, fiber diameter, and
amount of fiber. Of particular importance is the fiber aspect ratio, defined as
the length divided by the diameter. The peak in the packing density generally occurs
at small concentrations of fiber, near 25 volume percent for fibers with aspect ratios
over 10. Generally, the packing of high concentrations of long fibers requires a
particle size much smaller than the fiber diameter. Furthermore, the packing
density is increased as the fiber aspect ratio is reduced; this is often contrary to
the requirements for optimal strength. If we let R be the ratio of sphere diameter
D to fiber diameter D
f

R = D/ D
f
( 6.1)
then two combinations exist for high packing densities. The first case
corresponds to spheres filling the voids between large fibers for a size ratio R less
than unity. The second case is for the fibers filling the voids between large spheres
for R greater than unity. The packing density for a sphere-fiber combination
increases with very large or very small R values. The cases of R near unity give
the lowest packing densities. Figure 6.1.1 shows an example of the packing density
variation with composition and size ratio R for as constant fiber length-to-diameter
ratio of ten {L/ D
f
= 10). As the size ratio R increases, the packing density likewise
increases at all compositions. There is a composition of maximum packing at high
volume fractions of spheres, typically near 75% spheres.
74
Figure 6.1.1 The packing density variation with composition for fiber-sphere mixtures with
a fiber length-to-diameter ratio of 10 and various ratios R of sphere diameter to fiber
diameter
The interaction between the sphere-to-fiber diameter ratio, and the fiber
length-to-diameter ratio is illustrated in Figure 6.1.2 This figure plots the relative
gain in fractional packing density versus the logarithm of the sphere-to-fiber
diameter ratio (R) for a length to diameter ratio L/D of 15.5, The data are applicable
to the whole range of possible compositions. As illustrated in Figure 6.1.1 , for a
given ratio of diameters there is a condition for minimum packing density.

Figure 6.1.2 An example of the interaction between the fiber and particle characteristics with
respect to the packing density for a range of sphere-to-fiber diameter ratios
There is a relation between the fiber loading, the diameter ratio, and the
fiber length-to-diameter ratio. Figure 6.1.3 shows the key interaction, where the
choice of small particles improves the packing density with high fiber loadings and
long fibers. Alternatively, large spheres are favored with short fibers, especially at
75
low fiber concentrations. An approximate formula for the optimal fiber content X
based on the fractional packing density for the spheres f and fractional packing
density for fibers f is given as follows:
X
f
=f
f
(2 f
s
)/(2 f
f
+ f
s
- f
s
f
f
)) (6.2)
This model generally predicts optimal packing at relatively low fiber contents.

Fiigure 6.1.3 The effect of the fiber content
and fiber length-to-
diameter ratio on the choice of either
relatively large or small
Spheres for optimal packing
2.6.2 NONSPHERICAL PARTICLE MIXTURES
Irregular, small particles can be mixed with spheres at low concentrations
without significantly degrading the packing density. Mixtures of spheres with low
concentrations (less than 10%) of smaller irregular particles do not exhibit
significant packing density degradation , though at high concentrations of irregular
particles the packing structure is degraded by the irregular particles. This is
especially true if the irregular particles approach the size of the spheres.
Consequently, the density decreases in proportion with the concentration of irregular
particles when the particles are of similar sizes.
The packing of monosized nonspherical particles typically is inferior to that for
spheres. Accordingly, there is interest in increasing the packing density o.f
nonspherical particles by mixing differing particle sizes. Unfortunately, volume
76
expansion is sometimes observed in mixtures of irregular particles. Thus, a signifi-
cant problem arises in determining the particle shape effect on packing. In
general, the packing density increases with large differences in particle sizes. At
small particle size ratios there is no benefit from mixtures of the two component
powders. Under optimal conditions, the improvement in packing density depends
on the size ratio as follows:
f=A-B exp(-CD
L
/D
S
) (6.3)
where A, B, and C are constants determined by the particle shapes and
inherent packing characteristics. The particle size ratio D
L
/D
S
is the size of the large
particle divided by the size of the small particle. Figure 6.1.4 plots the observed
fractional density versus particle size ratio for irregular feldspar and quartz
particles This plot is at a composition of 77% large particles, and it shows that
packing density improves as the particle size ratio becomes larger. However,
increasing the particle size ratio over approximately 8 does not further. Improve the
packing density. Optimal packing conditions occur over a narrow composition range
near 25 to 30% of the small particles. Other models for the bimodal nonspherical case
include a particle size effect in the A and B constants.
Figure 6.2.1 The maximum packing density
(at the 77% large composition) versus the
particle size ratio for bimodal mixtures of
irregular mineral particles
Measurements on coke particles further demonstrate the packing behavior of
irregular particles. Depending on the particle shape and size ratio, the peak in
packing density can occur over a broad range of compositions. This range is
broader than observed with spherical particles. Such behavior is illustrated in
Figure 6.2.3 for two particle size ratios. In contrast with Figure 6.2.2 the relative
packing density improvement with a mixture of particle sizes is less. The most
likely cause for the difference between Figures 6.2.2 and 6.2.2 is the particle size
77
ratio. This behavior illustrates a basic problem in dealing with irregular particles.
Because of the wide range of possible shape combinations, it is difficult to predict the
specific behavior for mixtures involving non-spherical particles. Typically, the
principles associated with bimodal sphere mixtures are observed, but the combination
of two different shapes is untreated by theory.
Figures 6.2.2 The dependence of the
fractional packing density on
composition for mixtures of
irregular (crushed) and spherical
glass with particle size ratios of
3.4 and 3.
Figures 6.2.2 Fractional packing
densities of mixtures of irregular coke
particles with particle size ratios of
1.4 and 1,9
2.7. COMPACTION OF PARTICLES
Many Uses of particles involve the application of pressure to increase density
and strength. The external pressure repacks and deforms the particles into a higher
density mass. Simultaneously compaction imparts shape to the powder mass, which is
78
usefull in manufacturing products ranging from pills to automotive gears. Thus interest
in particle compaction spans fields including powder metallurgy, ceramics soils
pharmaceuticals and geology.
2.7.1 STAGES OF COMPACTION
Figure 7.1.1 is a simplified diagram of the change in density versus applied pressure
during particle compaction. The diagram is nominally divided into four zones,
which are not distinctly separate. The initial density without an applied pressure
corresponds to the loose packing condition. The densification rate (first derivative)
continuously declines as the compact density increases. As illustrated in Figure
7.1.2, the porosity, coordination number (number of contacts), and contact area
all vary with the pressure . These observations were made on spherical bronze
particles during compaction. Based on such observations a generalized concept of
particle compaction has emerged.
Figure 7.1.1 An idealized plot of fractional density versus pressure for particle
compaction showing the four overlapping stages.
Various measurements and tools have been used to monitor the changes in
packing structure during compaction. These have included scanning and
transmission electron microscopy, optical metallography, x-ray diffraction, hardness,
microhardness, resistivity, surface area, mercury porosimetry, permeability, and
small-angle neutron scattering. The results of these studies allow a nominal
79
division of compaction into four stages. Initially, elastic deformation occurs at
point contacts. With further pressurization the number of contacts grows as
particle rearrangement and sliding occur. Concomitantly, the contact areas expand
and the zone of plastic deformation around each contact spreads. At high compac-
tion pressures, rearrangement ceases and densification is by contact enlargement.
The concentrated pressure at the contact points causes mass flow into the neighboring
pores,- reducing the pore size and porosity. Along with contact enlargement,
strain hardening occurs; both factors increase the stress needed for continued densi-
fication. At very high densities the, actual particle character is lost as massive
deformation takes place.
figure 7.1.2 The compaction behavior of
spherical bronze particles as measured by
the compact porosity, number of contacts
(coordination number), and area of each
contact
In the case of spherical particles, the contacts deform and the spheres are
transformed into polyhedra. During such deformation, shear occurs along the
particle contacts by sliding. A consequence of this shear is that the contacts bond
together by cold welding, giving an increase in compact strength. Additionally,
the initially random structure takes on a more ordered character as density
increases. There is danger in generalizing compaction behavior because of the several
80
variables that can influence the process: the details of the compaction stages depend
on particle and material characteristics. Although this makes the specific response of
particles to an applied pressure dependent on the situation, there are still some
general behavior patterns.
The initial transition with pressurization is from a loose array of particles
to a closer packing. This is termed rearrangement, where both the packing density
and coordination number increase with small levels of pressurization. Generally,
rearrangement occurs at pressures less than 0.05 MPa. The particles start with a
packing density characteristic of random packing. From this starting condition,
the amount of rearrangement is inversely dependent on the initial packing
density. At low loads, the point contacts between particles undergo elastic
deformation. Additionally, large packing voids due to particle bridging are collapsed
by particle motion under small stresses. Increasing pressure enlarges the contacts
and provides closer packing with lower porosity and smaller pores. The more
irregular the particle shape, the lower the initial packing density. Accordingly, the
density change by rearrangement is larger for powders of lower packing density.
The random dense packing is often cited as the upper limit to the
rearrangement stage of compaction. However, other densifi-cation processes
become active prior to this limit. For irregular particles, fracture is observed at
pressures as low as 30 kPa. Nonspherical particles undergo a substantial
reduction in volume by rearrangement (up to a 55% decrease in volume for
flakes). In contrast, spherical particles undergo as little as a 1% density
increase by rearrangement.
The second stage of compaction involves localized deformation starting at
the point contacts between particles. As pressure is increased,, the localized
deformation flattens the asperities and the deformation spreads throughout
the contact region. At the contact, the stress distribution is not uniform. The
maximurp stress occurs at the center of the contact and the minimum
compressive stress is at the edge of the contact. For a compressive force F
applied over a circular contact of diameter X, the maximum stress
M
is given
as
81

M
= 6 F/( X
2
) . (7.1)
In contrast, the maximum shear stress occurs away from the contact zone with a
magnitude of approximately 0.3
M
. Along the contact face, the shear stress is
dependent on the amount of slip between the particles.
For brittle particles, the onset of plastic deformation can lead to fracture,
giving fragmentation of the original particles and densifi-cation by repacking of
the fragments. In experiments with glass spheres of varying sizes, fracture was
observed at a fractional density of 0.63, independent of the particle size.
Fracture initiates at the point of maximum tensile stress, which occurs on the
contact boundary as a radial stress. The onset of fracture can be detected by an
increase in the surface area and a decrease in the mean particle size. Figure 7.1.3
illustrates the difference in surface area versus density for coal and salt.
Figure 7.1.3 The change in specific surface area versus compaction pressure for
coal and salt; the different behaviors reflect the difference in compaction events
For ductile particles, the concentrated pressure at the particle contacts will
eventually exceed the yield strength of the material. This causes localized plastic
deformation at the contact points. Material flows into the unoccupied regions
neighboring the contacts, leading to further densification. During compaction, there is
a preferential elimination of the larger pores and a shift of the pore size distribution to
82
smaller pores. Since the particle centers move closer together, new contacts are
formed continuously as densification takes place. Indeed, sliding along the contact
faces is observed well into the compaction process. The zone of plastic deformation
spreads as the pressure is increased until the entire particle deforms.
For a brittle substance, the deformation process is partially replaced by
fragmentation. In contrast with pore filling by plastic flow, the fragments move into
the voids, decreasing the porosity and mean pore size. When fragmentation begins
there is a decrease in the compaction rate. The degree of fragmentation versus
deformation during compaction is dependent on the material hardness and plasticity.
Fragmentation reduces the particle size, increases the surface area, alters the
particle shape, and increases the number of contacts per unit volume. This latter
event leads to a stress reduction at each contact point as the number of contact
points increases. As a generalization, plastic deformation gives faster densiflcation
than fragmentation.
For spheres, the onset of the third stage of compaction occurs at a
fractional packing density of approximately 0.85. In this stage, independent of the
densification process, further elimination of pores requires ever increasing
expenditures of energy. The densi-fication rate decreases as the density increases,
which means that the energy input is inversely related to the remaining porosity.
At very high compaction pressures, massive deformation occurs, leaving small pores
at the junction of three or more particles. This typically starts at fractional densities
over 0.95 and compaction pressures in excess of 1 GPa. The fourth stage of
compaction corresponds to a fully hardened compact. Plastic deformation is
exhausted and the compact behaves as a full density elastic solid undergoing
hydrostatic bulk deformation. An increase in pressure leads to a corresponding
decrease in volume, yet no significant permanent volume change is obtained upon
relaxation of the pressure. The fourth stage of compaction is rarely observed in
practice.
After compaction, the external stress is removed and the compact
undergoes elastic relaxation; this is referred to as springback. The elastic
relaxation is very evident from the failure of a compact to fit back into rigid tooling
after ejection. At low compaction pressures the degree of springback is low. For high
83
compaction pressures, springback increases with approximately the square of the
compaction pressure. As the particle hardness increases, the degree of springback
likewise increases. One consequence of springback is that the shear stresses
associated with compact ejection can cause compact failure. This is typically
evident as delamination cracks perpendicular to the direction of pressurization.
2.7.2 DENSITY- PRESSURE RELATIONS
The relation between density and the applied pressure is an important aspect
of compaction. The response of a particle mass to pressure is termed the
compressibility, and is often defined as the density attained at a preset
compaction pressure. Various theories of densification during compaction note that
the change in fractional density f with pressure P can be expressed as follows;
df/dP = K(l-f) (7.2)
where If is a proportionality constant and 1 f is the porosity. Rearranging
and integrating Equation 7.2 gives
ln((1-f)/(1-f))=- KP (7.3)
where f
0
equals the fractional density without an applied pressure (the apparent
density)
Numerous models have been presented for the dependence of density on
compaction pressure that incorporate terms for .rearrangement, plastic deformation,
fracture, hardness, and strain hardening. Table 7.1 summarizes several models but
most of these are unsatisfactory for several reasons, not the least of which is their
empirical origin.
TABLE 7.1 Models of Compact Density as a Function of the Applied Pressure
f=A P
2/3
-B
A, B, K, K , K , and m are constants
f=A P
m
-B
84
P = applied pressure
f = fractional density
f
0
= initial fractional density
f=(1-f
0
) exp(-K P)
f=1-(1- f
0
)exp(-K P- B)
f= f
0
(1+ A P)/(1+ BP)
f=(A+B logP)
-1
1/(1-f)=A
1
exp(K
1
P)+ A
2
exp(K
2
P)
f= f
0
+ A P
1/3
f= f
0
+ A P
3/2
f=( 1 - f
0
) exp(-K P + A)
Although all models adequately predict a decrease in porosity with an
increase in pressure, they often lack a physical basis. Generally, the models
oversimplify the complexity of compaction and do not adequately include the stages
outlined above. One problem is the number of adjustable parameters. The transition
between stages is often poorly defined. For example, most models fail to include
rearrangement as a low-pressure component of densi-fication. Additionally, very high-
pressure compaction appears to be poorly treated by the existing models. Some of the
variations have attempted to incorporate packing density, yield strength, strain
hardening, fracture strength, or rearrangement corrections. Various comparisons of
the models to experimental measurements do not clarify the matter. There are
differing levels of success depending on the powder and material characteristics. In
spite of the shortfallings, empirical equations that predict the density as a function of
the compaction pressure are important in generalizing behavior and comparing
different particles. Most probably the largest problem is that compaction is not a
simple combination of stages and mechanisms; thus, the mathematical treatments
need to better reflect the complexity of the actual process. Accordingly, simple
models involving only the pressure and a few adjustable constants are nominal
acknowledgments of the underlying complexity..

Das könnte Ihnen auch gefallen