LIEHARY
Michigan State
Unlverslty

 

 

 

PLACE IN RETURN BOX to roman this Mom hum your record.
TO AVOID FINES Mum on or baton data duo.

DATE DUE DATE DUE DATE DUE
I t m I

.- -I

:I—7
ll ' ll 1
I L__J

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

    

 

 

 

 

 

 

 

 

 

 

 

 

MSU IQMMMMWOWIW
W1

FIBER ELECI'ROPHORETIC DEPOSITION PROCESSING OF CONTINUOUSLY
REINFORCED Fe-4OA1 / A1203 INTERMETALLIC MATRD( COMPOSITES

By

Christopher J. Suydam

A THESIS
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
MASTER OF SCIENCE
Department of Materials Science and Mechanics

1993

Melissa Crimp, Advisor

ABSTRACT
FIBER ELECTROPHORETIC DEPOSITION PROCESSING OF CONTINUOUSLY
REINFORCED Fc-40Al / A1203 INTERMETALLIC MATRD( COMPOSITES

By

Christopher J. Suydam

A technique for producing a continuously reinforced interrnetallic matrix
composite (IMC) was developed by applying theoretical and experimental examination
of the system components in order to determine the optimum processing conditions
and parameters. The mechanics of this process involves the adhesion and infiltration
of Fe-40Al powder onto A1203 multistrand fiber bundles without the use of a binder.
By evaluating and determining the optimum processing conditions, maximum adhesion
between the matrix powder and the fiber reinforcement (and minimum adhesion)
between similar system components is achieved. Optimum processing conditions were
generated by applying a previously developed theoretical model which describes the
multi-component system governed by traditional colloidal theory. By evaluating such
parameters as particle size, surface potential, and the component stability ratio, W,
verification of uniform adhesion of the particles to the fibers was obtained. A
prototype filament winding system was also developed for implementation into the

processing scheme.

To my wife, Tasha, and my entire family for all of their love and support.

ACKNOWLEDGEMENTS

First of all I would like to thank my family and all of my fi-iends for their
moral support throughout the course of my graduate studies. I would also like to
thank my advisor, Dr. MJ. Crimp, for funding me and for all of her support during
this research project. In addition, I would like to thank Dr. M.A. Crimp for his
support and help with the microscopy and photographic techniques. I would especially
like to thank S.C. Tonn and J. Ambriz for all of there help and encouragement.
Special thanks go out to B.A. Wilson for all of his guidance in colloidal chemistry and
extensive computer knowledge. Finally, I would like to extend thanks to Dr. E.D.

Case for all of his intellectual help, within and outside of this research project.

iv

TABLE OF CONTENTS

LIST OF TABLES
LIST OF FIGURES
INTRODUCTION
REVIEW OF LITERATURE
HISTORY OF PROCESSING
COLLOIDAL THEORY AND PRINCIPLES
ELECTROKINETIC PHENOMENA
MATERIAL SELECITON
FILAMENT WINDING
EXPERIMENTAL PROCEDURE
PARTICLE SIZE ANALYSIS AND REDUCTION
PROCESSING PARAMETERS
ESA MEASUREMENT OF SYSTEM COMPONENTS
FIBER COATING
FED WINDING SYSTEM DESIGN
RESULTS AND DISCUSSION
PARTICLE SIZE REDUCTION

PROCESSING PARAMETERS

vii

10

20

38
42
42
47
53
56
59
66
66

74

TABLE OF CONTENTS (CONTINUED)

RESULTS AND DISCUSSION
ESA MEASURMENT AND STABILTTY PREDICTION
FIBER COATING
FILAMENT WINDING SYSTEM

CONCLUSIONS

APPENDIX

REFERENCES

67

113

124

136

141

158

Table 1.
Table 2.
Table 3.
Table 4.

Table 5a.

Table 5b.

Table 6.

Table 7a.

Table 7b.

Table 7c.
Table 8.

Table 9.

Table 10.
Table 11.
Table 12.
Table 13.
Table 14.

Table 15.

LIST OF TABLES

Component Properties 35

Thermodynamic Calculations and Compatibility Results 37

Selected Materials 38
Electrolyte Concentrations 49
Measurement Variables for Fe-40AlSED 52
Measurement Variables for Fe-40Al325 53
Potentiomeuic Titration Data 54
Potentiometric Identification for Fe-40Al325 55
Potentiometric Identification for Fe-40AISED 55
Potentiomeu'ic Identification for FF AI¢O3 56
Acid and Base Normalities 58
Wet Powder Furnace Treatment Data 73
Particle Size Data for 0.1N KNO3 Fe-40AISED 84
Particle Size Data for 0.01N KNO3 Fe-40AlSED 84
Particle Size Data for 0.001N KNO, Fe-40AlSED 85
Particle Size Data for 0.0001N KNO3 Fe-40AlSED 85

Particle Size Data for 0.01N KNO3 pH 3.0 Fe-40AISED 87

Particle Size Data for 0.001N KNO3 pH 5.5 Fe—40AlS- 87

vii

LIST OF TABLES (CONTINUED)

Table 16. Particle Size Data for 0.0001N KNO3 pH 8.0
Fe-40A1SED

Table 17. Particle Size Data for 0.1N KN 03 Fe-40Al325

Table 18. Particle Size Data for 0.01N KNO3 Fe-40A1325

Table 19. Particle Size Data for 0.001N KNO3 Fe—40A1325

Table 20. Particle Size Data for 0.0001N KNO3 Fe-40Al325

Table 21. Particle Size Data for 0.01N KNO3 pH 3.0 Fc-40Al325

Table 22. Particle Size Data for 0.001N KNO3 pH 5.5 Fe-40A1325

Table 23. Particle Size Data for 0.0001N KNO, pH 8.0
Fe-40Al325

viii

Figure 1.

Figure 2.

Figure 3.

Figure 4.

Figure 5.

Figure 6.

Figure 7.

Figure 8.

Km 9.
Figure 10.
Figme lla.

Figure llb.

Figure 12.

LIST OF FIGURES

Illustration of possible fiber alignment configurations
using binder-assisted molding with either an expanding
contracting flow through the die.

Illustration of the steps involved with the powder cloth
processing technique.

Illustration of pressure casting technique.

Illusu'ation representing the diffuse elecu'ical
double layer.

Illustration representing the electrical double layer
with repect to Stern’s Theory.

Graphical representation of the total energy of
interaction, V1, obtained by the summation of the
energy of repulsion and attraction curves, V3 and VA.
Basic filament winding system.

Schematic illustrating the particle size sedimentation

technique.

Hypothetical roller setup.

Filament carraige assembly design.
Fiber cradle assembly.

SEM micrograph of Fe—40AISED as-received
powder(8OOX).

SEM micrograph of sedimented Fe-4OSED powder
sample.

13

18

30

8

61

62

67

68

LIST OF FIGURES (CONTINUED)

Figure 13. Illustration of grid analysis used to determine 70
maximum particle size.
Figure 14. SEM micrograph of Fe-40ALSED (850K). 71
Figure 15. SEM micrograph of Fe-40A1SED (3500)(). 73
Figure 16a. SEM micrograph of Fe—40Al325 (200K). 75
Figure 16b. SEM micrograph of Fe-40A1325 (llOOX). 76
Figure 17. Time lapse photograph for c.c.c. determination (t = 0s) 79
Figure 18. Time lapse photograph for c.c.c. determination (t = 458) 80
Figure 19. Time lapse photograph for c.c.c. determination (t = 90s) 81
Figure 20a. SEM micrograph of FF alumina fiber bundle. (500X). 98
Figure 20b. SEM micrograph of single FP alumina fiber. (4000K). 99
Figure 20c. SEM micrograph of ball-milled FP alumina. (800X). 100
Figure 21. Zeta potential data for Fe-40AISED vs. FP alumina. 101

Tiu'ation from pH 4.0 to pH 10.0.

Figure 22. Zeta potential data for Fe-40A1325 vs. FP alumina. 104
Titration from pH 3.0 to pH 10.0.

Figure 23. Stability prediction for Fe-40AlSED vs. FP alumina 106
multicomponent system.

Figure 24. Stability prediction for Fe—40A1325 vs. FP alumina 107
multicomponent system.

Figure 25. Stability prediction for Fe-40A1325 using p.s. = 1/2 110
simulation.

Figure 26.
Figure 27.

Figure 28.

Figure 29.

Figure 30.

Figure 31.

Figure 32.

Figure 33.

Figure 34.

Figure 35.
Figm'e 36.
Figure 37.
Figure 38.
Figure 39.

Figure 40.

LIST OF FIGURES (CONTINUED)

Stability prediction for Fe-40A1325 using p.s. < 10pm
simulation.

Zeta potential vs pH data for Fe-40Al 325 simulation
(titration from pH 4.0 to pH 10.0).

Stability prediction for Fe-40Al325 simulation.

SEM micrograph of coated FP alumina w/ Fe-40AISED
matrix particles at pH = 5.5 to 6.0.

SEM micrograph showing the degree of coating at
pH 3.0 for the Fc-40Al325/FP A1203 system (100X).

SEM micrograph showing the degree of coating at
pH 6.0 for the Fe-40A1325/FP 1511203 system (80X).

SEM micrograph showing the degree of coating at
pH 8.0 for the Fe-40Al325/FP A1203 system (100X).

SEM micrograph showing the uniform su'ucture of the
Fe—40A1325/FP A1203 system (800X).

35mm photograph showing the various components of
the FED filament winding system.130

Multi-strand fiber delivery assembly.

Cradle roller guide and suspension bath assembly.
Rotational motion of cradle roller assembly.

Fiber guide roller.

Rotational and transverse motor assembly.

Mandrel assembly.

113

112

114

118

120

121

122

123

127

128

130

131

132

133

134

LIST OF FIGURES (CONTINUED)

Figure 41. Side-view of mandrel assembly. 135

INTRODUCTION

Due to the increasing technological demands by the aeronautical/aerospace
industry there has become a need to produce materials meeting the strict requirements
of today’s high performance systems, such as reduced weight savings and increased
stcngth and toughness properties. The recurrent theme in the majority of expectations
is that the materials need to accommodate significant weight to volume reduction
requirements while retaining other desirable properties such as strength and toughness,
over a wide range of temperatures.

Intermetallic alloys, and specifically the B2 aluminides, have received
considerable attention as possible candidates to accommodate the indusuial demand for
high performance materials. In addition to their relatively low densities, intermetallic
alloys have excellent environmental and oxidation resistance. Unfortunately, there are
limitations to the use of intermetallics. Intermetallic alloys suffer from poor ductility
and toughness at low temperature ranges and lack high strengths at elevated
temperature ranges. It has become apparent that intermetallic alloys have very limited
uses as monolithic materials in high performance applications. However, it may be
possible to correct the undesirable properties of intermetallic alloys by incorporating
them into multiocomponent systems. By employing the interrnetallic alloy as a mauix

material for a composite system, varying types of reinforcements can be applied which

enhance the properties of the matrix. This provides a composite material which
defines the now common classification of intermetallic matrix composites or IMCs.

A The major objective in designing and producing IMCs is to provide a near-fully
dense composite containing a high volume fraction of fiber while taking advantage of
the desirable properties of the intermetallic matrix. Since the advent of composite
materials processing, there has become a need to not only produce materials that meet
these general design requirements, but also to establish a means of processing that is

effective, simplistic, and economicalll,2,3].

REVIEW OF LITERATURE

Histog of Processing

Since the introduction of composite materials and composite materials
manufacturing, there has been a constant need to produce a material which is cost
effective without degrading the original design parameters. Therefore, it is necessary
to establish a means to produce a composite having optimum material properties. One
of the most evident problems encountered in the processing of IMCs is the inability to
produce a material which is fully or near-fully dense with a uniform fiber (or
reinforcement) distribution.

There exists a variety of processes which have been used to produce IMCs.
Although, many of these processes have been successful in providing near-fully dense
composite systems, there are limitations. Several successful IMC processes exist
which are generally referred to as powder metallurgy processing techniques. Some of
these techniques include: reactive processing [4], powder injection molding [4,5], hot
exu'usion [4,6], XD" synthesis [4,6,7,8], and mechanical alloying [7].

Reactive processing utilizes the reaction between the elemental powders. One
variation of reaction processing is known as reactive sintering, RS. This process
consists of mixing elemental powders and producing a reaction by heating the mixture

to the lowest liquidus temperature of the components [4]. Once the reaction has been

4

accomplished, the product is then separated via grinding, mixed with the desired
reinforcement, and then sintered to form the composite. Several intermetallic alloys
and their composites have been successfully formed in this manner, including N iAl,
Ni3Al, Ti3Al, Nb3Al and TiAl. Modifications of the reactive sintering process include
the use of uniaxial pressure during sintering (reactive hot pressing, RI-IP) and isostatic
pressure during sintering (reactive hot isostatic pressing, RHIP).

A similar technique to that of reactive processing was developed by Martin
Marietta. This process, known as XD" synthesis[4,6,7,8], employs the use of an
exothermic reaction created between the elemental components of the powder mixture.
This reaction provides enough heat to form a liquid phase which help reduce sintering
times and ultimately increases the densification of the powder compact. Once reacted,
the product is pulverized, mixed and consolidated via hot pressing or HlPing.

Another powder processing technique is mechanical alloying [7]. This
typically consists of high energy milling of a dry, ductile matrix with ceramic
particles. The mixture is then compacted and consolidated via traditional techniques.
Although this type of technique has found limited application in the production of
IMCs, there has been some success. Strothers and Vedula successfully produced an
FeAl/Y,,O3 composite system via mechanical alloying. This technique is explained in
detail in references discussed elsewhere [7,9,10].

Of the previously described techniques, there is one technique which has not
only achieved near-fully dense composite systems, but has been able to control the
alignment of the reinforcement within the matrix. This technique is known as powder
injection molding [4,6,11]. This process utilizes a heated polymeric binder which is

mixed with a powder matrix and chopped, discontinuous reinforcement. This mixture

5

is then extruded through a die. Alrnan and Stoloff were successful in forming as-
CIPed green bodies of Ni+AI/15v/o FP ALO, with 35v/o binder using this technique
[11]. As can be seen in Figure 1, this technique has the possibility of providing
alignment of the fiber reinforcement. In order for this process to be successful, a fine
powder and a proper binder must be selected. Pr0per binder selection is crucial in that
it must be completely removed during the post-sintering / consolidation steps thereby
not becoming a source of contamination [4,6].

Each powder metallurgy technique previously described has been successful in
providing near fully dense IMCs. One limitation of all the powder metallurgy
techniques is that the processes are limited to applications requiring short,
discontinuous reinforcement. There exist a variety of processes, though, that have
extended their range to include continuous reinforcement. These include: the powder
cloth technique [4,6], the foil-fiber-foil technique [6], various melt processing
techniques [6,7], and filament coating techniques [7]. The powder cloth method has
achieved considerable success at the NASA Lewis Research Center in producing
continuously reinforced IMCs. This technique is utilized to produce alternating
lamellae of matrix and reinforcing material. The matrix lamellae is formed by
combining a powder matrix with an organic binder and wetting agent and then rolling
to form a thin sheet. The reinforcing lamellae are formed by winding the continuous
reinforcing fiber around a mandrel or drum and then coating the drum with a
polymeric binder. Lamellae are then stacked in the desired arrangement, and
consolidated [6]. This process is illustrated in Figure 2. The major problem
encountered is the entrapment of the binder within lamellae which provides for a

source of contamination. The foil-fiber-foil uses the same design and consolidation

 

 

WWW -_ /

I/

 

 

 

 

 

 

—+ $98 'g-gfimcmc
RANDOM INPUT ORIENTED OUTPUT
___.. ::EEEEEEEEEE __.

Figure 1. Illustration of possible fiber aligrnent configurations using binder-assisted
molding with either an expanding or contracting flow through the die.

<@> -*m-*

INTERMETALLIC

POWDER+TEFLON
POWDER+STODDARD t W
SOLUTION "
MIX TOGETHER

 

 

 

 

 

 

 

 

 

 

 

 

STACK AND
CONSOLIDATE

FIBER POLYMERIC , w /

WINDING BINDER

FIBER
MAT

Figure 2. Illustration of the steps involved with the powder cloth processing
technique.

8

principles except that the matrix lamellae are in the form of a rolled sheet instead of a
rolled powder slurry. Therefore, no binder-contamination is contributed by the
formation of the matrix ply but contamination is produced by entrapment of binder
between the plys.

Another successful IMC processing technique for producing continuously
reinforced composites is pressure casting. This technique consists of melting the
mauix material under vacuum and then forcing the molten material through a
continuous fiber preform. This process is schematically illustrated in Figure 3.
Nourbakhsh, et al have achieved success in forming IMCs of TiAl, Fe,Al, Ni,Al
reinforced with up to 60v/o of ALO, [6,7,9]. The inability to provide adequate wetting
at the matrix/fiber interface for many materials and the resulting poor bond structure
poses a problem as applied to IMC systems. Other processes which have been applied
to continuously reinforced composites include thermal/arc spray [4,7], CVD [7],
plasma spray [4,7] and sol-gel processes [7]. The groundwork for all of these
processes involves the deposition of matrix material onto a mandrel containing
continuously wound fiber using varying methods for deposition. For the thermal, are
and plasma spray techniques, the matrix is deposited in a molten form in order to
provide an impregnated tape which can be cut/stacked and consolidated. Since the
matrix is molten, this type of processing encounters the same problems associated with
pressure casting [6,7,9].

All of the processes described above are capable of producing near-fully dense
IMCs. The limitations inherent to each of these processes hinders the type and range
of applications where an IMC can be implemented. The most apparent problems

encountered during current processing techniques include: inability to implement

APPLIED LOAD

I

    
   
 

CASTING PISTON

TEN
INTERMEI'ALLIC
MATRIX

FIBER
PREFORM

Figure 3. Illustration of the pressure casting technique.

10

continuous fiber reinforcement, contamination from binders, and wetting problems
associated with melt processing. In order to address these problems, an adhesion
process, coupled with standard continuous fiber winding techniques, must be designed
which does not introduce the complications of polymeric binders and melt processing.
One solution would be to implement established colloidal theory to standard processes

inherent to powder metallurgy and continuously reinforced composite systems.

Colloid Theog and Principles
The basic fundamentals and principles of colloidal theory describe the behavior

and interactions of particles suspended in a fluid medium. The basis for the design of
the Fiber Electrophoretic Deposition (FED) process is to primarily take advantage of
the existing electrostatic characteristics inherent to particulate suspensions and
secondly to control these parameters in order to produce attraction between the powder
suspension (intermetallic rmuix) and the continuous reinforcement (ceramic fiber). In
order to provide a complete explanation of this adhesion phenomena, several colloidal
principles inherent to particle suspensions must be examined. Upon immersing some
particles into a fluid medium, a charge potential is created via a variety of charge
generation mechanisms. These mechanisms include ionization, ion adsorption and ion
dissolution [12]. The ionization charge mechanism is dependent on the pH of the
particle suspension. Deviation from the suspension’s iso-electric point (i.e.p), the pH
where the net potential on the particle surface is equal to zero, dictates what the net
charge of the particle will be and can be related to the particle’s surface charge. A net
surface charge may also be acquired by the unequal or preferential adsorption of one

ionic species over another, which is known as ion adsorption. Ion dissolution creates

11

a net surface charge for ionic species by the preferential adsorption of one of the
different ionic groups which determines the net surface charge. These individual ionic
species, which are responsible for creating a net surface charge, or potential, are also
know as potential determining ions, or p.d.i.s. Shaw gives examples pertaining to both
the classic AgI system, where both Ag+ and I' are the potential detemrining ions, and
the hydrous metal oxide system, where both the H+ and OH' ions are the p.d.i.s [12].

Unique to the adsorption of ions onto a particle surface is the creation of an
electrical double layer. The electric double layer is modelled as two layers; the
diffuse layer of ions and a layer of rigidly held ions immediately adjacent to the
particle surface. The diffuse layer of ions has been described in a quantitative manner
by Gouy and Chapman [12,13,14]. The layer containing rigidly held ions has also
been described quantitatively by Stern [12,13,14]. These two descriptions will be
discussed in detail to provide a basis for understanding and interpreting the charge
potential phenomena associated with spherical particles immersed in an aqueous
solution.

The Guoy and Chapman treatment of the diffuse layer of the electrical double
layer is based on the following assumptions [12]: (l) the charge density of the
particle surface is assumed to be uniform across the particle surface, (2) the particle
surface is assumed to be infinite and flat in geometry, (3) the charge distribution
surrounding the particle is assumed to be governed by a Boltzmann distribution, (4)
the only influence the solution has on the diffuse region is through the characteristic
dielectric constant, which is also assumed to remain constant through the bulk of the
medium, and (5) the electrolyte in the bulk medium is assumed to have a fixed single

charge number, z. Figure 4 illustrates the assumed surface geometry with a fixed

12

surface charge, in this case, positive, having the characteristic Boltzmann distribution
of ions. Given the concentration of positive and negative ionic species at a potential,

‘I’, is expressed as [12,14]:

 

 

n<+> =n,exp[ 'Z‘Vl (I)
and
n(-) =n,exp[ *2va (2)

where n, = concentration of ionic species in the bulk, zev is the electric potential
energy created by each ionic specie, and k and T are Boltzmann’s constant and

temperature respectively. The net volume charge density, p, is given by:

p =ze[n(+) -n( -)] . (3)

The volume charge density can be related to the surface potential by considering

Poisson’s equation for a geometry of a flat plate, given by:

 

dz‘l’ =-.£_ (4)
dz:2 8

where e is the permittivity of the particle. By combining equations (3) and (4), the

13

 

 
   
  

 

 

 

 

 

g "’o
z
9 S
1.. .
g COUNTER-IONS g
2 no """""" 2
8 CO-IONS
l A l m
1m 1II<
DISTANCE (x) DISTANCE ()0

Figure 4. Illustration representing the diffuse electrical double layer.

14

following expression for relating the surface potential is found:

d2“, = 2zeno cinh zgw . (5)
dx2 8 kT

From this equation, the appropriate boundary conditions can be applied. Using the
assumptions of the Gouy and Chapman treatment of the electrical double layer, the
appropriate boundary conditions for Equation (5) are as follows: I]! = w, as x —9 0 and
\y = 0 as x -> oo. Applying these conditions to the Laplacian form of equation (5)
yields the following relationship:

(my) (29%)
M +1][exp ZkT

(zev) (2%)
M ZkT

[cxra -l]

 

 

ln[ (6)

[exp +1]

 

 

+l][exp

This equation can then be simplified to the expression for the Gouy-Chapman
expression for the variation of the particle potential with respect to distance within the
double layer[12]:

zevo
[PXP(7k-T—)-l]
7 =10expl-tcx] where: 7 = . (7)

taxmgvpm

 

15
For small potentials, i.e. if zetyJZkT << 1, or kT/e = 25.7mV at 25°C, the Debye-

Hiickel approximation:

2N0] = 1 41E] (8)

e
xp[ 2kT 2kT

 

can be used to obtain the following relationship

 

ty=wacxp( -m) (9)
where:
K=[2e2n022]; . (10)
ekT

Equation (8) describes the Debye-Hiickel approximation for the Gouy-Chapman
treatment of the diffuse double layer. This Arrhenius-type relationship describes the
variation in particle potential with respect to distance from the srn'face. Using the
Debye—Htickel approximation, the relationship given in equation (9) only holds for
small potentials, \y S 25.7mV [12].

The potential \y., can be related to the surface charge density by considering the
condition of electroneuu'ality at an interface between the particle and the fluid
medium. The total charge contained in a volume element of unit cross sectional area
of the diffuse layer must contain the same magnitude of charge, (opposite sign) of that
of a unit element of the surface [14]. In simpler terms, the net charge of the diffuse

layer is equated with that of the net surface charge of the particle surface. Therefore:

obfpd. . an

Applying the Poisson-Boltzmann distribution function and assuming low potentials,

Equation (11) reduces to the form:

c =ercwa . (12)

This treatment provides an adequate approximation for a distribution.of an
"ion-cloud" around a specific particle immersed in a fluid medium, but it does not
account for a problem that arises due to the stated assumptions. Primarily, at
increased magnitudes of surface potential, the net charge density calculations become
so unrealistic that it can no longer be assumed that the ions may be effectively treated
as point charges. As the charge density increases at the surface, it is then assumed
that the ions have some characteristic ionic volume [12,14]. Stern [12,13,14]
examined the problems associated with the assumpan of point-charge ions and
devised a new geometric interpretation of the diffuse layer model. He proposed a
model which divided the aqueous fluid medium into two layers: the diffuse layer and
the Stern layer.

The interface between the Stern layer and the diffuse layer arises from the
stipulation that there are two basic types of ionic species present in the electrical
double layer. These include physisorbed ions and chemisorbed ions. Physically
adsorbed ions, or physisorbed ions, are classified as any ionic species which are held

to the surface by relatively weaker, physical bonds. Chemically adsorbed ions, or

17

chemisorbed ions, are any ionic specie which is attached to the particle surface by
stronger chemical bonds. Chemisorbed ions are also commonly known as specifically
absorbed ions. Both ionic species as either potential determining ions or indifferent
electrolyte ions. The Stern model holds that particles do not remain still while
suspended in a fluid medium. Particle motion results from a variety of forces
including: (1) gravitational forces obeying Stoke’s law, (2) convective forces due to
thermal gradients, and (3) particle-particle interactions caused by Brownian motion.
Therefore, Stern was able to assume that due to the existence of two types of ionic
species and given the unrealistic charge density values obtainable through the Gouy-
Chapman treatment, there must exist a layer which separates those ions which are
attached, (or which determine the potential) and those that remain behind as the
particle moves through an aqueous medium. Stern also assumed that there existed a
Langmuir adsorption isotherm [14] that adequately described the mono-layer (Stern
layer) adsorption of specifically potential determining ions.

The major difference between the Gouy-Chapman and Stern treatments for the
elecuic double layer is that the potential, W0, is measured at the interface between the
Stern layer and the diffuse double layer, whereas for the Gouy-Chapman treatment, the
potential, v, is evaluated directly on the particle surface. This Stem/diffuse layer
interface is located at a characteristic distance, 5, as shown in Figure 6. The
characteristic potential at that interface is denoted as We Assuming that only the
specific adsorption of counter-ions exists, the following equation can be generated

which describes the sm'face charge density within the Stern layer [14]:

 

 

a
o = "
. NA WM. (13)
1+_exp[ ]
n v kT

18

  

 

5:23: 9/ i Storgupégg: of shear
9 Ir) 6) e
G) I 9 Q
I

e '6)

e: 69 9
e l e

\_ . >
'\ Stem I232” layer

 

L
7

POTENTIAL
d€~ 5$

 

 

0 E‘— .
"K DISTANCE

‘7

Figure 5. Illustration representing the electrical double layer with respect to Stern’s
Theory.

19

where o; is the surface charge density at the Stem layer, 0.. is the surface charge
density associated with the Langmuir-type adsorption of counter-ions within a mono-
layer, VIII is the molar volume of the solvent, and 9 is the specific chemical interaction
energy [12,13,14]. Therefore, assuming that electroneutrality exists throughout the
entire double layer, the sum of each surface charge density, charge density for the
particle surface, charge density for the Stern layer and the charge density for the
diffuse layer will equal zero. Thus the following mathematical expression for the

Stern model of the double layer can be obtained:

 

 

3 _ out _ é . 1]] ze¢d=
3(4)" ¢)+ NA ze¢d+¢o (Swim s 2kT 0 . (14)
1+ apt—l
:1on kT

Heimenz [14] states that although the model proposed by Stern does
accommodate the specific adsorption of ions, it also generates a number of difficulties.
Most importantly, the complexity of the description of the variation of potential with
respect to distance from the particle surface is dramatically increased when applying
Stern’s treatment to that of Gouy and Chapman for the electrical double layer. This is
brought about by the added number of variables that must be considered during the
treatment. In addition, it becomes unclear as to what exactly is physically defined by
the characteristic distance, 8. Furthermore, it is uncertain as to where the potential, V
is actually evaluated. In the design of the FED system, the main parameter of concern

is this evaluated potential located somewhere within the Stern layer, known as the zeta

patential (C).

20

As was previously illustrated, there exists a characteristic potential which is
inherent to the electrical double layer surrounding a particle suspended in a fluid
medium. The location and definition of this "potential" varies with respect to location
from the particle sm'face. w, is defined as the potential located at the particle surface.
As was shown in Figure 5, there also exists a potential We which is defined as the
potential at the plane separating the specifically adsorbed ions with those that are not.
This plane of specifically adsorbed ions is also known as the Stern plane. Thus w, is
denoted as the Stern potential. In the same manner as the Stern plane was defined, the
sm'face of shear is defined as the imaginary plane which separates the ions which
remain attached to the particle surface and ions that are released as the particle moves
through the fluid medium. The potential inherent to this "surface of shear" is defined
as C, the zeta potential. As will be shown in the next section, it is the zeta potential,
1;, with which we are particularly interested in defining in order to ultimately
characterize the degree of heterogeneous attraction between the Fe40Al intermetallic

powder and the ALO, continuous fiber.

Electrgkinetic Phenomena

In the previous section, it was shown that when a charged particle is immersed
into a fluid medium, it acquires a characteristic "ionic cloud" which is defined as the
electrical double layer. This double layer arises from the observation that the ionic
distribution around the surface of the particle consists of differing types of ions.

These ions are either specifically adsorbed to the surface and comprise the adsorbed
layer or Stern layer surrounding the particle, or, they can reside in the a region beyond

the Stern layer defined as the diffuse layer. The existence of particle motion within

21

the fluid medium, causes important changes to this entire ionic region surrounding the
particle. As the particle moves through the fluid, attempts are made by the generated
fluid forces to remove the ions fi'orn the specifically adsorbed layer surrounding the
particle surface. As a result of this observation, four phenomena result which
comprise what is generally referred to within colloidal theory as electrokinetic
phenomena [14,15].

Elecuokinetic phenomena is a result of the combination of particle motion and
particle electrical effects. As applied to the previous discussion, this phenomena is
more accurately defined as the combination of the theory describing the elecuical
double layer and that of fluid flow. There exists four separate situations which are
adequately used to describe electrokinetic phenomena. These include:

(1) electrophoresis, (2) electro-osrnosis, (3) streaming potential and (4) sedimentation
potential [14]. Electro- phoresis is the phenomenon with which most of the attention
has been given with respect to the results contained herein. Specifically, research was
performed using electroacoustics which is very similar to electrophoresis.

Electrophoresis describes the movement of a charged particle and it attached
ionic layer as is moves through a fluid medium. The movement associated with the
particle is caused by the application of an electric field. The main stipulation of this
phenomenon is that the velocity caused by the applied electric field is directly
proportional to the magnitude of the electric field. The constant of proportionality is
commonly defined as [1,, the electrophoretic mobility. By measuring the
electrophoretic mobility, it becomes possible to determine the potential characteristic
of the adsorbed ionic layer, C, the zeta-potential. The calculation of the zeta potential

depends on the shape of the double layer and how the charged particle and its

22

associated electrical double layer are treated [14].

The electrical double layer is defined by the parameter, UK or Ic’ 1. This is
known as the Debye length, or the distance over which the potential decreases by an
exponential factor. Shaw showed that for low potentials the double layer has the same
capacity as a parallel plate capacitor with a distance of K" between the plates
[12,13,14]. This parameter, K‘, is common referred to as the "thickness" of the double
layer as shown in Figure 5. The shape of the double layer can be described by
considering the radius of curvature of the particle surface, a, and the characteristic
thickness of the double layer, It 1. With respect to examining the phenomena of
electrophoresis, there exist two treatments which account for the different shape that
the double layer can have and how this effects the calculation of the zeta potential, C.
The first case deals with a experimental scenario where K3, the ratio of the radius of
crn'vature to the double layer thickness is small and the second case deals with the
inverse of this. For tca small, i.e. a much smaller than UK. the particle may be treated
as a point charge. For Ica large, i.e. a much larger than UK, the particle surface may
be treated as a flat plate. For the case where K3 is small, it is considered that Stoke’s
Law is still applicable even though the particle size is very small.

For Ira small, the particle size is much smaller than the length of the double
layer. In order to treat this case and generate an expression for the zeta potential as a
function of electrophoretic mobility, the physical nature of the motion of the particle
through a fluid medium must be considered. By equating the electrical forces with the

particleffluid fiictional forces, the following expression is obtained [12,15]:

(QNH)E = 61t11vca (15)

23
where QNET is the net charge on the particle, E is the magnitude of the applied electric

field, 1] = viscosity, v° is the electrophoretic velocity and "a" is the radius of curvature.

Simplifying eq. (15) gives the following result for the electrophoretic mobility [12,15]:

 

The zeta potential, C, is the calculated resultant between the charge on the particle,
+Qmm and the charge on the adsorbed layer, ONE... Since K8 is assumed to be << 1,
the following equation is obtained [12,15]:

1.5
g... ”E“ (17)
808'

 

where e, and e, are the permittivity in vacuum and relative permittivity respectively.
Eq. (17) is known as the Huckel equation and is applicable to elecu'ophoresis of non-
aqueous media which has a low conductivity. As shown by Shaw, this inapplicability
is due to the fact that extremely low electrolytic concenu'ations are required when
using small particle sizes.

For ka large, i.e. for particle size much larger than the thickness of the double-
layer, the same basic principles that were previously applied can be used to obtain the
Helrnholtz-Smoluchowski equation for zeta-potential as a function of electrophoretic
mobility. This is achieved by equating the electrical and viscous forces generated by

the motion of the liquid through the diffuse layer relative to the particle surface.

24
Thus, as shown by Shaw [12]. the expression:

EEC =1] v. (18)
or simplified:
=Vc=C8 as)
”e 72' ‘n—

C- M . (20)

This equation is valid for aqueous suspensions. It should be pointed out that since
there is some discrepancy as to the exact location where the "potential" is measured
and to which potential is actually used to characterize the particle, Lu, is assumed
equal to C at the Stern/diffuse layer interface.

As applied to the research project specifically, the phenomena examined is
known as electroacoustics. The mechanisms by which the zeta-potential is determined
are basically concrn'rent with those which define electrophoresis. The only major
difference is that the motion the particle experiences due to the application of an
electric field generates a measurable sound wave with a characteristic frequency. The
technique commonly used to evaluate the elecu'oacoustical phenomenon is known as
Electrokinetic Sonic Amplitude or BSA [17.18.19]. The ESA technique consists of the

application of an electric field of some known magnitude. E. to a fluid suspension.

25

An acoustic wave is generated by the back-and-forth motion of the particles though
the electric field. This motion is caused by the charge and density differences between
the particles and the fluid medium. The ESA then measures the amplitude of the
pressure generated per unit electric field. This measured quantity is given in terms of
the calculated dynamic mobility, 310(0)), such that:

(co=
It.) We

 

where ESA((I)) is the measured angular frequency. It is the volume fraction of
particles. Ap is the density diflerence and c is the speed of sound within the

suspension. The zeta potential is then determined using It, and an added correlation

 

factor:
=fllc(a-l)| (22)
808'
where:
or = a) a’p . (23)
v

From this technique. it is possible to characterize the zeta potential in terms of
dynamic mobility associated with the sound wave generation. Once an understanding
has been achieved regrading the determination of the zeta potential. this parameter. C.

which is inherent to the particular solution and conditions surrounding it, can be used

26

to further describe the solution. This description extends to the determination of the
attractive and repulsive forces which dictate the stability and lack-there-of for all
colloidal suspensions. By ultimately understanding the forces present within the
suspension. it becomes possible to control these parameters in order achieve the goal
of producing a binder-less composite material. Before proceeding into a discussion
regarding the forces present within a colloidal suspension. attention must be given to
the understanding of suspension stability.

The most important property associated with colloidal processing is the stability
of the suspension. The stability of the suspension is characterized by the amount of
coagulation that takes place within the system. Coagulation can be defined as the
formation of agglomerates. or bundles, comprised of the individual particles. If these
agglomerates form, the system is considered to be unstable. Colloidal coagulation is
considered a rate process. therefore it is important to analyze and understand the rate
of coagulation of a colloidal system, rather than being concerned as to whether the
agglomerates fornr or not.

With respect to suspension stability, there exists two classifications of
suspension solutions. or. sols. These are lyophilic and lyophobic sols [12]. Lyophilic
sols include all sols which are stabilized by and have a strong affinity for the liquid
medium within which it is dispersed. On the other hand. lyophobic solutions refer to
sols which have little affinity for and are not stabilized by the liquid medium within
which they are dispersed. As pertaining to the processing of IMCs. the solutions or
sols used are lophobic. Therefore. the stabilizing of the suspension must be achieved
by modeling the forces inherent to the dispersed particles. The exists two basic force

regimes which affect the stability of a colloidal system. These forces include an

II

27

attractive forces. which results fiom the existence of universal long-range forces
known as van der Waals and London forces. and a repulsive force. generated by the
interaction and overlapping of electrical double layers. The theories based on the
existence of these two force regimes is commonly grouped in a cumulative theory
known as the DLVO theory. The DLVO theory. named for the persons responsible
for its development. Deraguin. Landau, Verwey, and Overbeek, is the culmination of
individual research combined to describe to effects of interaction between two
particles. The basis for the theory is the evaluation of the total interaction energy in
terms of interparticle distance. In order to apply this theory to the situation of a
multicomponent systerrr, a model for representing two unequal spheres will be
presented at the conclusion of the discussion regrading the DLVO theory
[l2.13,l4,21.33].

The first principle of the DLVO theory is that the particles have a characteristic
double layer. This double layer is the combination of a rigidly held inner layer,
known as the Stern layer, and a diffuse "outer" layer both of which obey a Boltzmann
distribution of point-like charges. The second principle of the DLVO theory is that

the total particle interaction energy is described by following relation:

V, = VR + VA (24)

where: VT = the total energy associated with the particle interaction. V A = energy of
attraction. and VR = energy of repulsion. Therefore. an evaluation of each of these
terms is required in order to produce a relation between the total particle energy and

the relative distance between them.

28

As pertaining to attraction between particles. there exist three basic forces
which are encountered in colloidal dreary. These include Debye, Keesom. and London
interaction [14]. It is the relative contribution of these forces which comprise the
energy of attraction VA. An expression for V A was developed by Hamaker in 1937.
Hamaker was able to assume that at very close ranges the London attractive forces
would predominate. This is due to the fact that at close range the attractive force
varies inversely to a power of 6 with respect to interparticle separation. Therefore the

following equation was generated by Hamaker:

 

= ’4“ 25
VA (12”) ( )

where A = Hamaker constant. a = particle radius. and H = interparticle distance.
The DLVO theory generates an expression for the repulsive energy of interaction for
two identical spherical particles by using the poisson equation for the flow of an

electric field in a dielectric medium, given by [12]:

V24) = -_E_ (26)

where p is the net charge density and a, and a. are the permittivities in a free space
and a vacuum respectively. By applying the previously defined Boltzmann equation

for the distribution of charges within the diffuse double layer. given by:

"I = "racxPI-gl - (27)

29
With to s new. the net charge density,p, becomes [12]:

1 . (28)

 

. -z.ew
P-zngcxPI [CT

Combining equations (26) and (28). and simplifying using the previously defined

Debye-Hilckel approximation, (WSZSmV @ 25°C), gives [12]:

 

- l 2 nab—Z "2‘9”” . (29)
l 8 k7!

Assuming electroneutrality in a bulk medium:

w=_[£"2"2"‘ 0‘") . (30)

Assuming boundary conditions which describe a one dimensional, linear situation:

W4¢.asx->OAv—90asx-)oo . (31)

An expression for the potential. w. is obtained:

‘l’ = \I',cxr)( -ch) (32)

30

 

Potential energy of interaction (V)
O

 

 

Figure 6. Graphical representation of the total energy of interaction, VT. obtained
by the summation of the energy of repulsion and attraction curves. VR and VA.

31

where:

 

K _J 92: niaziz (33)
- —_kT .

Deraguin used this result to obtain the following expression for the repulsive energy of

interaction. VR:

V, = 26083]?an + exp( -rcHo)] (34)

where H, is the shortest interparticle distance. By combining the previously derived
expression for VA and VR, and expression for the total energy of interaction is obtained
for a single component system. Stability of the system is then indicated by graphing
the potential energy, V1, versus the particle separation. H. The condition of stability is
indicated by the existence of a potential barrier. V,. of at least 20—25kT.

Expanding the results of the DLVO theory, Hogg, Healy, and Furstenau
developed a quantitative kinetic theory for stability for multicomponent systems. This
theory is commonly referred to as the HHF theory for stability [20]. The IMF theory
assumes that only small separations exist and that the particle potentials remain

constant. The expanded equation for the attractive energy is given as:

= -Aalaz

. (35)
50:02”

 

where A = Hamaker constant and al and a2 are the particle radii of species 1 and 2

32
respectively. Therefore. evaluation of the Hamaker constant provides the energy

associated with interparticle attraction. I-II-IF showed that the w. values are actually
considered valid up to 50-60 mV as opposed to $2va as indicated by the DLVO
theory. In the same manner as the DLVO showed. Hogg. Healy, and Furstenau
generated a relation for the energy of interparticle repulsion. V... by solving the

Laplacian given in equation (30) and applying the following boundary conditions:

W=Walasx_)0 AW=WaZasx42d (36)

By solving the Laplacian. the following relation for potential is obtained:

(“’02 - VoICOSh(2Kd)) (37)
31°11’10“!)

 

\|l =\|Io,cosh(rcx) +

Therefore using the equation given by Deraguin [21] for the interparticle repulsion

energy for a single component system:

V.=e.e,n are. (wor’wmai 2"°""°’In “WW“ +ln(1-cxr>(-2KH))I 08’

(al-I-az) we, 022 l-cxp(-KH)

 

The combination of the two components. VA and VR, gives and expression for the total
energy of the system. i

The stability for a system of non-identical particles is indicated by the stability
ratio of the system. The stability ratio. W1, is defined as the ratio of particle collisions

to collisions resulting in coagulation. For a single component system:

’3
r

33

.. V dr
W=2a .3; _ (39)
um”) r

20 2

Expanding to include a multicomponent system:

.. VT 0"? 40
W0. = (a5 + aj)uexp(fi)72_ ( )

Hogg. Healy, and Furstenau [20] integrated and applied this to show:

2 l-n)2 2n(l -n) -
w =["_._( +_]I . (41)
T W11 W22 W12

I-IHF theory provides the best quantitative prediction for the stability of a
multicomponent system. This is primarily attributed to the ability to make adjustments
to improve the stability prediction. As applied to the analyzed multi-component
system contained herein, it is possible to use the zeta-potential, C, to characterize the
the component interaction. The only requirements for determining the C-potential are
prior knowledge of the particle size and the surface charge density. Therefore, with
respect to the research contained within, the H“ and OH' ions are considered the
potential determining ions and the C-potential is nreasured as a function of the relative
Change in pH. For a change in concentration, the surface charge changes and

therefore the C-potential changes. By calculating the C-potential and then assuming

34
that §=\|r. the stability of the system can be characterized. A system is considered

stable for WT values 2 1010 or for log(W) = 10 [13].

Material Selecdon

The demand for high-perforrnance materials which can accommodate
the requirements for high temperature aerospace applications has steadily increased
over the last decade. The irnmergence of high performance materials represents a new
age of materials applications. where weight savings and increased structural integrity
are primary design criteria. With this increasing demand, intermetallic alloys have
become possible candidates. The majority of ongoing research has focused on the
internretallic aluminides. specifically the Ll2 ordered aluminides. titanium alurrrinides,
and the B2 ordered aluminides. Of particular interest to high performance applications
is the BZ ordered intermetallics. due to their low densities. high melting temperatures.
and good environmental resistance.

A few alloys which are included within the B2 intermetallic system are FeAl.
MM and COM. Each of these alloys exhibits the B2 crystal structure or ordered
BCC. where the Al atom occupies the body-centered position and the Fe. Ni and Co
atoms occupy the comer atoms for each of their respective systems. The most
advantageous characteristic of the 132 aluminides is that they have relatively low
densities and high melting temperatures as shown in Table 1.

In addition to these characteristics. the B2 aluminides also exhibit excellent
oxidation resistance. Although these materials exhibit very desirable properties. as
monolithic materials. they do not represent the complete solution to high performance

aI’Plications. In addition to lacking high strength characteristics at elevated

35
Table 1. Component Properties

 

 

Material Dcnsig [glcc] Melting Temp. (K)
I FeAl 5.56 1724
NiAl 6.07 1921

 

 

temperatures, the B2 aluminides also lack ductility. toughness and creep suength at
low temperatures. Due to the intrinsic brittle behavior of these materials. it has been
of major research interest to investigate the effects of alloying additions such as Zr
and B to improve the ductility [22.23.24]. Aoki and Izunri showed that by adding up
to 1000ppm of boron to Ni3Al provides up to 50% elongation [25]. Therefore. the
main focus in intermetallic research is the improvement of the materials strength and
toughness over the wide range of possible operating temperatures. In order to
accomplish this. a reinforcement material is required to alleviate the problems
associated with the monolithic intermetallics.

There exists two major criteria for selecting a suitable reinforcing material to
be incorporated to form a composite system. These criteria include the need for both
chemical compatibility and a suitable coefficient of thermal expansion, CIE, between
the fiber and the matrix. The need for chemical compatibility arises due to the
expected high temperature environments that can provide for extreme changes in the
reaction rate kinetics subsequently degrading the reinforcing material and altering the
composite properties [8]. The need for an acceptable match between the CTE’s for

limb the fiber and the matrix is apparent since the associated operating range and

36
thermal gradients are very large. Improperly matched components can generate

thermal stress cracking and cause system degradation.

At elevated temperature conditions. the reinforcement can exhibit three basic
types of behavior. These include: highly reactive. partially reactive and non-reactive
behavior [26]. A highly reactive reinforcement will allow a large amount of reaction
product to form. Subsequently, a partially reactive product will also allow a reaction
product to form, but the extent to which this happens is significantly less. This is due
to the reduction in the reaction rate kinetics. Both of these types of reinforcement
behavior produce detrimental effects due to the degradation and reduced effectiveness
of the reinforcement. The non-reactive behavior prohibits any interdiffusion of
constituents or the formation of reaction products [26]. The reaction rates and
chemical compatibility are not only a concern with respect to the operating
temperatures of the material but they are also a concern during processing where
elevated temperatures are characteristic of many consolidation processes.

Shah, Anton and Musson [22] performed a compatibility study which examined
several possible intermetallic matrix materials coupled with several reinforcing
candidates. Some of the interrnetallics examined included NiAl from the B2 aluminide
group and (Nb, Ti)2Al and (Nb, T'i)A12 fi'om the ternary titanium aluminide group.
The several reinforcing candidates used included A1203, SiC. TiC, Si3N, and Y203.
The primary objective of the compatibility study was to vacuum hot press several
fiber/mauix samples and characterize each microstructure in terms of porosity. fiber
damage. uniformity and the matrix/reinforcement reaction zone. From their work,
Shah et al concluded that in terms of chemically stable reinforcing material, Dupont

FP Ale, and Saphikon fibers represented the best possible candidates for use with

37

most intermetallic systems [22].

Draper. Gaydosh and Nathal performed a similar study which examined the
chemical and thermodynamic stability of various fibers with an Fe-40at%Al matrix
material. Using thermodynamic calculations to determine the reaction mode of the
system, Draper et al were able to determine whether the selected reinforcement was
compatible with the Fe-40Al mauix. A summary of results from the thermodynamic

calculations can be found in Table 2 [23].

Table 2. Thermodmamic Calculations and Compatibility Results

I Fiber Reaction Mode Comatible? I

 

 

 

 

 

SiC Forms FeSi and A140, No
B,C-B Forms PCB and fi'ee Carbon No
A1203 No reaction Yes
W Solubility of W in Fe-40Al = 2% at 1500K Yes

Solubility of Al, Fe in W = 2.7%Al. 0.5% Fe

 

MonC Forms M03Al No

 

 

 

 

Upon performing a heat treatment and microstructural analysis of each system, the
A1203 fibers produced no reaction product with the Fe-40Al matrix even at a heat treat
of 1500K at 25 hours, thus agreeing with the corresponding thermodynamic
calculations. Of the possible reinforcements examined. only the W and A1203 fibers
provided chemical compatibility. The existence of Kirkendal porosity and high density

of W limits the usefulness of W in the W/Fe-40Al composite system. Therefore. the

38

most suitable reinforcing fiber for use with Fe-40AI is A1203 due to its chemical
compatibility and low density. A1103 also exhibits a high CTE (9.4 x 10" K“) as
compared to W (5.3 x 1045 K"). The CTE for A1103 also more closely matches that of
Fe-40Al (21.8 x 10“ K“) [23]. Although the expected operating range of most IMC
applications will be ~1273k, the mismatch in CTE between Fc-40Al and A120, may
still be large enough to limit its cyclic thermal fatigue properties. It should be noted
here that it is beyond the scope of this research project to provide a solution to the
CTE mismatch problem. The focus of this work is to select the most qualified
reinforcing candidate to be incorporated into a composite system which utilizes
electrokinetic phenomena to provide a binder-less. continuously reinforced. multi-

component system. A summary of the selected materials is given in Table 3.

Table 3. Selected Materials

 

Material Melting Temp. (C) CTE (1000 C) II

 

 

 

 

 

 

A1203 2050 9.4 x 10‘ K" II
II Fe-40Al 1451 21.8 x 10‘ K" "
Filament Winding

As was previously discussed, there exists a variety of methods for the
manufacture and consolidation of intermetallic matrix composites. The selection and
implementation of a particular method depends on several variables associated with
each components of the composite system Since the application presented here
involves the design and manufacture of a continuously reinforced IMC, the selection

process becomes limited. A widely accepted method for the manufacture of

39

composites which incorporate the use of continuous reinforcement is that of filament
winding [7,27]. Filament winding basically consists of a fiber source which is wound
continuously around a mandrel. whose motion is controlled by a movable cross-head.
The basic instrumentation of a typical winding system is illustrated in Figure 7. The
filament winding process consists of two different method in which composites are
formed. These include dry and wet filament winding. Dry filament winding consists
of winding a fiber or tow which has previously been processed in order to attach the
mauix material onto the fiber. This process is commonly referred to as pre-
irnpregnation and consequently the winding medium is referred to as a preirnpregnated
tow. or "prepreg-tow." The general purpose for using this type of method is to
provide specific geometries and shapes of parts by adjusting the variables associated
with traditional winding principles. Wet filament winding consists of

incorporating the impregnation process and the winding process into one system in
order to produce a composite green body for manufacture using a variety of
consolidation process mentioned previously. The system which has been designed for
the manufacture of IMCs using fiber electrophoretic deposition will implement the
principles associated with wet filament winding since it is the purpose of this project
to incorporate a novel technique for generating a pre-impregnated tow with traditional
winding practices.

The most important variables associated with the wet filament winding process
are the rotational speed of the mandrel and the relative speed of the cross-head. The
degree to which the fiber is impregnated depends on the length of time it is allowed to
remain in the impregnation chamber. It is necessary to control the winding speed in

order to produce sufficient impregnation of the fiber prior to being wound onto the

    

Fiber 8 ools
/ P
Motor Driven F'I F d /
Mandrel lament ee ’l/IIII ,
7-

 
   
 
 

I
\m Resin Bath /

 

 

 

 

 

 

 

\\\\\\\

 

 

 

 

I
E W
//

W/ // W Y/// /////

 

 

 

 

 

 

 

 

 

Figure 7. Basic filament winding system.

41
mandrel. The geometry and properties of the composite are controlled by the relative

size and shape of the mandrel onto which the tow is wound. The typical size range
for composite parts produced via filament winding can be as small as 1" (25mm) or as
large as 20’(6m) [27]. The specific details associated with these controlling
parameters will be given more detailed attention when the design of the winding

system is presented later.

EXPERIMENTAL PROCEDURE

In order to design and develop an acceptable process for the production of IMC
specimens utilizing the FED process previously described. it is necessary to consider _
the parameters involved. The outline for developing the FED technique includes five
experimental topics. These include: (1) particle size analysis, (2) evaluation of the
processing parameters involved. (3) comparison of system component behavior using
the electrophoretic Sonic Amplitude (BSA) measurement technique, (4) experimental
examination of the FP A1,.O3 continuous fiber with the FeAl intermetallic matrix. and

(5) design and development of the FED continuous winding system.

Particlp Size Analysis and Reduction

During the course of this research project two batches of Fe-40Al intermetallic
particles were utilized. The preliminary results for this project were obtained using a
powder that was produced by Alloy Metals, Inc. of Troy. MI and supplied by
Professor K.M. Vedula at Iowa State University. The powder provided was a gas
atomized raw powder in an 80-mesh (150].Lm > diameter > 0.1},Im) size range. Upon
initial examination. it was readily apparent that due to the fact that the FP fiber strands
consisted of individual fibers with an average diameter of 20-30 microns, adequate

coating could not be achieved using the FeAl powder in its raw form. Previously. an

42

attempt had been made by Glime to reduce the particle size via a dry sieving
technique which implemented the use of a ro-tap shaker and standard 3 inch ASTM
screens supplied by Endecotts, Ltd., London. England, in sizes of 63, 38. 15 and 5pm.
This method proved inadequate due to the large time commitment and relatively low
yield percentage. A new technique was then developed for reducing the particle size
in a more efficient manner.

With the use of the raw Fe-40Al powder a suitable size range had to be
chosen. The range chosen for this powder was lOwn > d > 0.1].trn. This particular
range was chosen for a variety of reasons. First an upper limit had to be established
which would guarantee that the gravitational forces involved would not offset the
colloidal forces which are required for adequate coating of the fiber. Second. a size
distribution had to be chosen that would provide a narrow distribution so that not only
would the particle/fiber interactions be maximized, but the stability problems
associated with having a wide size range avoided. In addition to this. a more uniform
coating of the particles around the fiber could be achieved. Therefore. since the
average fiber diameter for the FP fiber is 20-30|.I.m, it was assumed that a ceiling limit
of 10m for the matrix particle size range would be suitable for this particular raw
powder. This limit met both the demands for a small enough particle for coating and
a size range narrow enough for maximizing interactions. The specifics associated with
the selection of this size range will be expanded during the discussion of the
composite processing parameters.

The particle size reduction technique developed for this project incorporated the

use of natural sedimentation coupled with high frequency sonication and vacuum

43

44

assisted filtering. The following equation was used which describes the motion of a
particle in a fluid medium, as given by Stoke’s Law [28]:

V: SIP “P2152 (42)
181]

where V = particle velocity. g = gravitational constant. p = particle density,
pf = fluid density, 1] = viscosity of fluid medium, and x = particle diameter
The following values were then used to determine the velocity of a 10|.tm particle in a

fluid medium of doubly-deionized water:

p... at 25°C = 0.99707 g/mL
p... = 5.56 g/cc

1].... = 0.8904cp (cp = centipoise = gm/[(9cm)(sec)(10’)

From Equation 42, V = 2.79 x 10“ m/s. Once the particle velocity is known for the
particular fluid medium, an average falling depth. y,WE and sedimentation time. t. can
be calculated relative to the size of the container used. This was accomplished by

incorporating the result from Equation 42 and performing the following calculation:

yAVE=40XIOJm

V = (yAvQ/(t)
t = 143.373

45
It should be noted that yAVE is defined as half of the total falling height for a particle

that begin sedimentation at the surface of the container. The value for vae is used
instead of y (total falling height) because the settling time. t. must not only account for
particles which begin their decent somewhere between the surface and the bottom of
the medium. but also those particles which begin their decent at the surface. This is a
safety measure to insure that once the sedimentation time. t, for a 10m has elapsed,
there will exist no particles larger than those specified within the given falling length,
Yawn-

Once the values for yAVE and t are known. a suspension with a desired amount
of powder is then mixed and sonicated using a high frequency ultrasonic probe. The
ultrasonic probe is used in conjunction with any mixing steps performed within this
research project. The sonication provides an initial even dispersion by eliminating any
flocculation which may have been present prior to the mixing operation. Once the
sonication has been performed. the sedimentation time, t. is allowed to elapse. The
amount of suspension corresponding to van is then drained into a cellulose membrane
filter container. The de-ionized water is then draw off by applying a vacuum of
approximately 3.3 - 4.0kPa. The remaining solution is then re-suspended and the
process is repeated until it becomes apparent that the re-suspended solution has
become supernatant and the solution no longer contains a significant amount of fines.
i.e. the amount of solution corresponding to y,“ is essentially clear in appearance.
The collected fines are then dried using a tube furnace at a temperature of 110°C for 1
hour and then stored in a vacuum desicCator. Figure 8 gives a schematic illusuation
of the particle sizing sedimentation process. The powder provided by Alloy Metals,

Inc. was previously referred to as raw Fe-40Al powder. Since two powder samples

 

Pyrex Container
\. Raw Mixed Suspension

 

Sedimentation Process

Recovered Fines

.15. . W -.;:- 1.31.5.5,“ .6... - -.-..-.-- /

# Filtered Water

Vacuum *—

 

 

 

 

Filtration Process

 

 

 

Figure 8. Schematic illustrating the particle size sedimentation technique.

47
were used during the project, the sedimented powder will be referred to herein as Fe-

40AlSED.

For reasons which will be discussed later, it became necessary to implement
another Fe-40Al powder for primary use in the composite processing. This specific
Fe-40Al powder was donated by Mr. R. Noebe at NASA Lewis Research Center,
Cleveland, Ohio. The initial particle size of the as-received was -325 mesh. This
corresponds to a screen Opening of 44.5um. Although this particle size is larger than
that of the desired size previously stated. it allowed for the continuation of
experimental work to include other aspects which had not been possible with the use
of the Fe-40AISBD. The reasons surrounding this fact will be discussed in a later
section. In order to avoid confusion between the two different powder samples. the

Fe-40Al received in the -325 mesh form will be referred herein as Fe-40A1325.

Processing Parameters

Before being able to accurately examine the colloidal forces which are inherent
to an aqueous particle suspension using the BSA measurement technique, two main
parameters associated with processing of the IMC composite system had to be
evaluated. Two major processing parameters are related directly to the colloidal
application within this system. The first parameter which was evaluated was the
critical coagulation concenu'ation or c.c.c. of the suspended Fe-40Al matrix particles.
The second parameter which was analyzed was the "efl’ective" particle size of the Fe-
40Al matrix particles. The term "effective" refers to the size of the particles while
they are suspended in the aqueous solution. By examining these two parameters. the

conditions under which the experimental data is taken are optinrized.

III;

“7“]:
a.»
I

set

1161

w...

IMF

Pal

CXI

the

the

43
The procedure used for determining the c.c.c. for the Fe-40A1 aqueous

suspension was derived from the commonly known Shultz-Hardy Rule [13]. The
critical coagulation concenu'ation refers to the concentration of the electrolyte in the
aqueous solution that will give way to rapid coagulation. As was discussed in the
section on colloidal dreary. a pair or more of particles will flocculate or attract when
the force of attraction between them is greater than the force of repulsion.
Fmthermore, it was stated that the force of repulsion that prevents the flocculation of
particles is directly related to the length, UK, of the double layer inherent to the
individual particles. If the length of the double layer is less extensive than the
curvature of the surface. the particle will have a small repulsive force and be
susceptible to rapid coagulation or rapid flocculation. Rapid coagulation refers to the '
elimination of the electrical barrier that previously existed between the particles.
Therefore, without this barrier. every collision which is a result of the Brownian
motion of the particle. will form a combination of particles, or "floc". By adding a
common ion source such as an electrolyte to a solution. the length of the double layer
surrounding the particles can be controlled. The higher the concentration of
electrolyte. the more compressed the characteristic double layer becomes. thus the
probability for coagulation increases. If a solution experiences rapid flocculation. the
particles will adhere to one another and "fall-out" of the suspension at a rate far
exceeding that given by Stoke’s law as related to a single particle. The exact value of
the c.c.c. depends on the nature of the sol. the temperature of the environment in and
around the sol, the nature of the electrolyte and the electrical charge on the surface of

the particles.

49
The procedure for the determination of the c.c.c. for the Fc-40Al aqueous

system was performed by selecting 4 different elecu'olyte concentrations. These
included 0.0001N. 0.001N. 0.01N. and 0.1N. The electrolyte chosen was potassium
nitrite or KNO3 with an atomic weight of 101.1032 g/mole. 200mL solutions of de-
ionized water and the corresponding electrolyte concentrations were mixed. The
following table lists the concentration and relative amounts of KNO3 added to the

solution:

Table 4. Electrolm Concentrations

 

 

 

 

 

Once the four different electrolyte concentrations are mixed the same concentration of
powder is added to each 200mL electrolyte solution. For this experiment a 2"/o Fe-
40Al325 (22.24g/196mL DI H,O) was added to each solution. After the powder is
added, the solution is sonicated. poured into 80 mL glass test tubes. and then allowed
to stand in a fixed support for a period of 2 hours. Once the solution has remained
undisturbed for the specified time period. it is then re-shaken and returned to its fixed
support. Time lapse photographs of the four test tubes are then taken at 5 second
intervals. The c.c.c. is then estimated to be between the test tubes for which one
attains a supernatant or "clear" state while the other contains particles which remain

suspended.

50

By evaluating the c.c.c., it becomes possible to determine the amount of
allowable time in which the processing of the composite can take place. For example.
if a solution with a given amount of electrolyte remain suspended then the flocculation
kinetics are at a minimum and therefore the mauix suspension will remain dispersed
and the settling ldnetics can then be estimated by simply considering the upper limit of
the matrix particle size distribution and combining with Stoke’s law [28]. In addition.
the value of the c.c.c. is extremely important when calculating the potentials associated
with the system components using the ESA measurement technique. By knowing
what concentration of electrolyte to use for a specific component which will produce a
stable and dispersed suspension. a greater accuracy in the values obtained from the
ESA measurement is achieved due to the minimization of flocculation kinetics. This
is primary due to the fact that a major variable involved in the measurement of zeta-
potentials using ESA is particle size and if flocculation kinetics are not kept to a
minimum, the particle size will change due to the formation of flocs within the
system. This particle size change that can occur once a dry powder is dispersed into
solution results in a different particle size from that of the dry powder. The particle
size of the suspended powder is known as the "effective" particle size. The effective
particle size of the suspension is not only a function of c.c.c. but it is also a function
of the size range of the particles.

As was stated in the section on colloidal theory. the stability of a system is
very dependent on the size range of the particles within the system. For a wide size
range, any collisions between very large and very small particles will result in
coagulation. thus producing stability ratio with respect to the two particles does not

apply. With respect to the systems examined. Fc-40AISBD and Fe-40Al325, the

51

possibility for even a small amount of flocculation exists between the upper and lower
limits of the size ranges. It was also previously stated that since the zeta-potential of
a suspension also varies with respect to pH of the suspension the stability of the
system. i.e. the number of collision which result in flocculation, is indicated by the
change in the effective particle size. Therefore. by being able to evaluate the size of
the particles in suspension. an accurate representation of the particle size can be
obtained which will subsequently increase the accuracy of the zeta-potential data
which will ultimately detemrine the ideal processing range.

The Malvern Mastersizer provided by the Hydrocyclone Development
Consortium was selected to generate the particle size data for both the Fe-40AlSED
and the Fe-40AL325. The Malvern obtains its measurement by analyzing the
diffracted light from an emitted laser beam through a sample cell which contains an
aqueous or non-aqueous suspension. The purpose of this experiment is to not only
attempt to examine the flocculation behavior and kinetics associated with changing
electrolyte concentration and changing pH values, but also to get an accurate
representation of the particle size for use in the BSA measurement of the solution for
which the c.c.c. is used. An experiment was set-up in which both powder batches
would be analyzed at the same electrolyte concentrations as the c.c.c. calculation was
performed at. 3 different pH values (3. 5.5, 8.0) were also selected in an attempt to
gain additional understanding of the flocculation kinetics over a range of pH values.
The following tables list the different samples measured and give a sample number for
each for comparison with the results later. The experimental runs were performed by
adding a small amount of Fe-40Al to premixed 200mL solutions containing each

different variable mixed with de-ionizcd water. Prior to the measurement each sample

52

was sonicated for l min.

Before any BSA testing can be performed an accurate particle size has to be
generated for each of the components which are to be analyzed. As was stated
previously the Fe-40AISED and Fe-40A1325 samples were analyzed using the Malvern
Mastersizer. The particle size for the FP A1203 was obtained by implementing a
simple line intercept technique. The fiber was initially ball-milled for 4 hours in a
mixture of Borudum grinding media, provided by Stoneware. Inc.. and propanol. At
the completion of the milling operation. a 6mm x 6mm piece of carbon tape was
inserted into the milling container and a sample of the sized FP A1203 was taken. This
sample was then attached to a metal stage and microscopic examination was
performed using the Hitachi S-2500C Scanning Electron Microscope. A grid analysis
was performed which produced an array of micrographs which were then analyzed
using a line intercept method for determining particle size. The reasons for using a

different method for measuring the fiber will be discussed in a later section.

Table 5a. Measurement Variables for Fe—40AlSBD

 

 

 

 

 

 

S -. ' n Number Vii-131213
FBALlA 0.0001N KNO,
FBAL2A 0.001N KNO,
FBAL3A 0.01N KNO3
FEAL4A 0.1N KNO.
FBALSA pH = 3.0
FEAL6A pH = 5.5

 

53
Ta 1e . M nt Variables for Fe—40Al325

0.0001N KNC.
0.001N KNC,
0.01N KNO.

 

 

 

0.1N KNO3
pH = 3.0
pH = 5.5
H = 8.0

 

 

 

 

 

BSA Measmement of System Commnents
The ideal pH range for processing of the IMC was determined by applying a

potentiometric titration using the BSA measurement system from Matec Applied
Sciences. Dining a potentiometric titration the zeta-potential, C. is measured with
respect to specific pH value. The pH of the system is varied by adding small amounts
of nitric acid. HN03. to make the suspension more acidic or by adding small amounts
of potassium hydroxide. KOH’. This allows for the measurement to take place over a
range of pH values. Once C-potential versus pH curves have been generated for both
components of the composite systenr, the stability is calculated using a computer
simulation so that an estimation as to the proper pH range for processing can be made.
The potentiometric tiuation is performed by mixing a specified volume percent
of suspension and then selecting a phase reference material. This preliminary
operation is performed so that the sign (+.-) of the zeta-potential values obtained

during the experiment can be correctly interpreted. For the BSA potentiometric

54
titration the following criteria regarding the sign interpretation with respect to phase
difference of single point zeta-potential values is as follows:

If phase is between: --15 to 15 then material is same sign as reference
material.

-l65 to 180 or -l65 to -180 then the material is of
opposite sign.

If the phase measured meets neither of the requirements listed above. then a different
reference material must be chosen and the single point nreasurernent repeated until a
proper interpretation can be made. Once a proper reference has been chosen. the
parameters required for the BSA potentiometric titration can be entered into the
system. The following table summarizes the required information and the
corresponding values used to perform the potentiomeuic titration for both the Fe-

40AISED, Fe-40AL325. and FF A120,.

Table 6. Potentiometric Tim’ on Data

Particle Densig Particle Size
Fe-40AlS- 5.56g/cc 1.90pm

 

 

Fe-40A1325 5.56g/cc 11.90pm
3.965g/cc _ 0.53pm

 

 

 

 

 

Once these material parameters are entered. the BSA potentiometric titration can take
begin. For the components analyzed here. the following tables list the sets of data that
were obtained in the terms of C—potential versus pH. Bach table list the sample

numbers of each of the data set(s) and the corresponding titration(s) performed.

55

The data compiled from the potentiometric titration was then entered into a stability
prediction program developed by Wilson and Crimp [15]. The basis for the program
is to predict the state of flocculation of the system components. The resulting data-
from the computer run is compiled in the form of the stability ratios. W11. W22. and
Wu. where the subscripts l and 2 correspond to the different system components. The
values for the stability ratios are then plotted graphically with respect to their
corresponding pH values. At the completion of both the potentiometric titration and
the stability prediction data runs. an estimation can be made of the pH range within
which the greatest instability exists between dissimilar materials while at the same

time exhibiting stability with respect to like components.

Table 7a. Potentiometric Identification for Fe-40AL325

S f- le Numbers Titration Performed (pfl) Comments
510. 511. 512 30:10.0. 10.0:3.0, 3.0:10.0 Mixed at pH = 3.0
513, 514 10.0:3.0. 3.0:10.0 * MIxed at pH = 10.0
515. 516 10.0:3.0. 3.0:10.0 Redispersed S.N. 523. 514
I 500. 501, 502 3.0:8.0. 8.0:3.0, 3.0:8.0 Mixed at pH = 7.0
_ 504. 505 3.0=>.121>1.0=>111..2. 3012.11 1M 3 SN.

    
 

 

 

  

 

  

 

  

 

 

 

 

   

 

Table 7b. Potentiometric Identification for Fe-40AISBD

Titration Performed (pH)

40:11.0, 11.0=>4.0 Mixed at pH = 7.0

 

 

4.0:8.3. 8.3:4.0 Mixed at pH = 7.0

 

 

 

 

4.0.0=>7. 7-°=>40 , -- ,- , _M1md_'H= 70

 

56
Table 7c. Pothtiometric Identification for FF g0.

 

 

 

 

 

 

 

 

1 ,, :7 , , ~ g , if 7

’ Sam 1e _- Titration Performed (pH) Comments A _ p

._ 530, 5314.0:110, 11.0=>4.0 Mixed at pH = 7. _ §
«1.0-1.4 M:_-.....o 3

 

 

___- ,,,,. __,____ “___—___ ...-__.. _ . _ _ ___—___--. __.__ __._._...___.,,v ___ ,_ .1

Fiber Coating

By using the data obtained from the BSA measurements. an approximate pH
range was established for which maximum coating of the FP AI¢O3 fiber by the Fe-
40Al matrix could be achieved. Upon dcterrrrining this ideal pH range. it was
necessary to experimentally verify the results. This was accomplished by performing a
series of experiments in which attempts were made to physically coat the fiber strands
with the powder matrix. These experiments were initially performed using the Fe-
40AISBD powder. The usage of the Fe—40AlSED had to be superceeded. though, due
to an extreme shortage of sedimented powder. Preliminary results include the coating
experiments using the Fe-40AISBD powder. but the majority of the data includes
results from experiments performed using the Fe-40A1325. This was primarily due to
the availability and the adequacy of the powder with respect to the experimental work.

The coating experiments were performed by initially cutting 70mm strands of
the FP A1203 fiber. Each individual fiber was approximately 25—30um in diameter and
each bundle contained approximately 70 fibers. A long fiber was laid out on a
plexiglass sheet where adhesive was placed over the fiber at 70mm increments. Upon
drying the fibers were cut and stored in a vacuum desiccator. A 25v/o suspension was
then prepared in a 150mL Pyrex beaker using 166.80g of Fe-40Al in 90mL of de-

ionized HzO. The beaker was then placed on a Corning stirring plate and allowed to

57

mix using a l" magnetic stir bar. At this time the pH of the system was monitored
using a Fisher Scientific Model Accumet 915 pH meter.

For alteration of the pH value of the suspension, different normalities of acid
and base were used. The specific acid and base used for these experiments was nitric
acid. (HNO3), and potassium hydroxide. (KOH'). For determining the relative weights

needed for each of the different normalities. the following conversion was used:

. _‘ moles of solute

molarr . . . x atomic weight x liters of solvent = grams of solute needed
liters of solution

 

To obtain the normality of the solution multiply the molarity by the number of
equivalents of the substance or the mass that exactly reacts with the equivalent of
another reactant. For each of the reactions for HNO, and KOH'. the number of
equivalents is equal to 1 so the number of grams required for a specific normality is
equivalent to the weight required by the same molarity. The following table lists the
different normalities used of acid and base and the grams needed per 200mL solution.
Once the pH of the suspension had stabilized. small quantities of acid were added to
bring the pH down to a starting point between pH3.0-pH3.5. Specific pH values were
not targeted due to slight variations in the pH values. Prior to the coating of the
fibers. the solution was sonicated using a Branson Model 950 sonifier to break any
agglomerates that may have formed at the point of mixing. An individually cut fiber
bundle was then taken and bent into a semicircular shape and immersed into the
suspension for a time period of 20 seconds. The fiber was then placed on a glass slide
and labeled. This process was repeated for the entire range of pH = 3.0 to pH = 11.0.

Each of the fibers were placed on glass slides and photographed using a 35mm lens to

58

examine the amount of powder matrix which was canied out of solution by the fiber
bundle.

The next experiment performed implemented the use of the Hitachi Model S-
2500C Scanning Electron Microscope to characterize the coated fibers at different pH
ranges. The procedure for coating the fibers is exactly the same as stated above for
the residual coating experiment. Instead of placing the previously immersed fibers
onto glass slides. the fibers were clipped at one end and allowed to dry before being
placed on a 2" diameter SEM stage. The "dipped" fiber were secured to the stage by
first adhering a glass slide to stage using Plyabond brand adhesive. SPI 5006 brand
carbon paint was then used to cover the corners of the slide. The fiber(s) were then
placed on the glass slide and secured to the stage using Plyabond and carbon paint on
the ends of each strand. The stage was then given a 40nm thick gold coating using a

gold coating device provided by the Composite Materials and Structures Center,

Table 8. Acid and Base Normalities

 

 

  
   

    
    
      
   
     

Normalig Grams of HNO3 Grams of KOH‘

 

0.0013
0.0126
0.1260
_1'2350 ,

0.0011
0.0112
0.1122
11.12

 

 

   

 

 

 

Michigan State University. Pictures characterizing both sufficiently and poorly coated
fiber specimens were then taken using the Hitachi Model S-2500C Scanning Electron

Microscope.

59
Fiber Blectrpphoretic geposition Winding System Desigp

In order to successftu produce intermetallic composite green bodies for
consolidation. a laboratory-scale winding system was developed. The design and
development of this system was dependent on many factors. These factors stemmed
from both traditional composite winding practices and those specifically inherent to the
electrodeposition procedure. The following procedure will outline those factors which
were identified as having the greatest impact in the design and development of the
FED winding system.

The behavior of the multi-strand FP filament used in this process is
synonymous with most winding process which employ the use of a multiple fiber
bundles. The key element in attaching, or impregnating, the desired component to the
fiber strand is to allow each of the fiber to act individual. or more simply. to allow the
fiber strand to become temporarily separated from the bulk of the filament bundle.
The easiest way to insure that this will happen is to provide a means by which the
fiber bundle can effectively "lay-down" or spread out dining the impregnation process.
The most common method for achieving this goal is to allow the fiber bundle to
traverse across a series of rolling surface. The more rollers that are included in the
system, the more effective the system will be in spreading the fibers out. Thus, the
first step in the design of the FED winding system was to produce a roller system
which exhibited the configuration shown in Figure 9. The rollers provided were cut
from extruded UHDP (Ultra-High Density Polyethylene). This material was chosen
due to its availibility. cost and very low coefficient of friction. Stainless steel sealed
bearings were countersunk into each end of the rollers and a 5/16" drill rod was used

to provide the support at each end. Aluminum supports were then machined and three

fired

posit
that

Clif-

asse
sup;
cho:
to 1’

roll

the
Th.

SUP

10:

\\

60
fixed supports and 1 adjustable support were implemented as illustrated in Figure 9.

Another factor which is required for proper fiber lay-down is the use of
positive tension on the fiber tow during operation. It is important to consider the fact
that the individual FP fiber strands are relatively fragile and that caution must be
exercised when planning to apply a positive tension. A simple method for providing
this important variable is accomplished within the design of the filament cartridge
assembly shown in Figure 10. 19mm stainless steel hearings were inserted into the
support arms which house the 1/2" threaded rod assembly. These bearings were
chosen in a larger configuration than normal so as to provide a slight amount of drag
to the system thus enhancing the ability for the fiber-tow to spread out across the
rollers.

The third factor considered in developing the FED winding system is the
design of a rolling system which can be integrated within the suspension bath. The
purpose for integrated roller system is to not only keep the tow constantly flowing
through the powder bath but also to aid in guiding the impregnated tow out of the bath
and onto a take-up device. The roller cradle designed is shown in Figure 11. Each of
the integral rollers are full adjustable with complete angular rotational capabilities.
The material used to construct each of the rolling assemblies was virgin Teflon
supplied by Almac Plastics Inc.. Grand Rapids. Michigan. and was selected due to its
ease of cleaning. non-reactivity, and low coefficient of fiiction. Springs were attached
to the comers 0" ‘1 in order to provid a convenient means to remove the
roller carriage leaning, etc.

on taken in the design is the take-up device unto which

3 wound. The main objective in the take-up device or

61

 

 

Tow Direction

 

 

Roller

 

 

 

 

 

 

 

 

 

 

TOW DIISCIIOD

 

 

Figure 9. Hypothetical roller setup.

 

62

\

Fiber Spool Support Brackets
Side Wow

I

   

Bearing Assisted
Adjustable Support Rod

Front View

Figure 10. Filament carriage assembly design.

63

mandrel design is to provide transverse as well as rotational freedom. This insures
that the fiber strand can be wound completely around the mandrel in desired interfiber
spacings. Virgin Teflon was employed as the mandrel material. for the same reasons
as stated above for the other applications, i.e. non-reactivity, low coefficient of
friction, etc. The mandrel is supported by a 1/2" bearing supported tool-steel shaft.
The rotational movement of the mandrel is provided by a Dayton 0.35A, 12V DC.
motor. The entire rotational assembly is mounted to a plate which is supported by a
threaded base. The threaded base, or cross-head provides the transverse motion. which
coupled with the rotational motion of the mandrel, generates the necessary winding
action. The threaded cross-head is also powered by a 0.35A, 12V D.C.motor
providedby Dayton. Inc. To make the production of multi-layer composites, a
switching track was installed which takes advantage of the movement of the cross-
head by employing the use of two polarized switches. By correctly positioning the
switches along the track, the movement of the cross-head will activate each switch
every time a transverse run is completed. This allows for a reversal in the polarity of
the motors thus changing the direction of the cross-head and provides another
complete wind in the opposite direction. A schematic of the proposed mandrel design
is shown in Figure 11.

The power for the system is provided by an HPB3612A Dual Range Digital
Power Supply. Located on the face plate of the system are the system controls.
These include two variable resistors to provide fine adjustment of the rotational and
cross-head speed. Another switch is supplied which, when activated, will provide a

reversal in the transverse motion. identical to that provided by the two polarized

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

* “V
/ Top View \

Cradle Housing Suspension Rollers

\.

 

 

 

 

 

 

   

 

 

6

Side View

 

Figure 11b. Fiber cradle assembly.

65

switches previously discussed. The only remaining switch on the face-plate is an
intermediate power switch which allows for the adjustment of the HP power supply

prior to the engagement of the motors.

RESULTS AND DISCUSSION

Particle Size Reduction

During the course of this research project, two intermetallic powders were
employed. The composition for both intermetallic powders is Fe40Al with 40 atomic
percent of aluminum present. The differences between the two powders include the as
received condition and small additions of Zr and B used to influence the mechanical
deformation behavior. At the outset of the research project the only available
intermetallic powder was the raw powder supplied in the by Alloy Metals. Inc. of
Troy. M]. This gas atomized pre-alloyed powder was received with a characteristic
particle size of 150m > diameter > 0.11mi. As was previously discussed. the need
arose for the reduction of this nominal size to a range more suitable for the
elecu'ophoretic deposition of the particle mauix on to the continuous fiber
reinforcement. The range initially chosen was lOurn > diameter > 0.1 um. A smaller
particle size allows colloidal forces to dominate over the gravitational forces which
become a significant factor with the use of large particle sizes. In order to effectively
reduce the particle size of the raw Fe-40Al powder. a sedimentation technique was
developed and is discussed in detail in the previous section. The SEM rnicrographs in

Figures 11 and 12 show the as-received Fe-40Al powder and the sedimented Fe40Al

66

 

67

 

Figure 11b. SEM micrograph
of Fe-40AlSED as-received
powder.(800X)

68

 

Figure 12. SEM micrograph
of sedimented Fe-40AISBD
powder sample.

69
powder. The specimens used in the SEM analysis were produced by applying 6mm x

6mm double-sided carbon tape to the powder sample and then removing the excess via
compressed air. This sample preparation technique provided for a well grounded
system which would prevent charging during examination. In order to determine if a
proper particle size range was achieved, a grid analysis was performed. The
verification of the correctparticle size was obtained by locating a large sample area in
the SEM. This large area was then sectioned into a smaller area in a grid fashion.
Each of the smaller areas were then visually examined using the micron marker on the
SEM video monitor to verify that the largest particles encountered were within the
specified size limit of lOum. The grid method is illustrated in Figure 13 and two of
the eight micrographs which comprise the grid are shown in Figures 14 and 15. As is
observed in Figures 14 and 15, the particle size achieved was below the intended size
range limit of lOurn.

Although success was achieved and an accurate distribution of Fe-40Al
particles was obtained. the developed particle sizing method encountered a number of
potential problems. Initially. the sedimented powder was allowed to remain hydrated
for extended periods of time. e.g. t > 24hours. and this resulted in a significant color
change in the powder. The fact that the powder surface had been altered in some way
became readily apparent when immersing the sedimented powder in doubly deionized
water and attempting to stabilize the pH. Upon mixing, the suspension would exhibit
a consistent tendency to stabilize in the more basic range of pH 10 to pH 11. Any
attempts to lower this value to the acidic range would result in the pH of the
suspension increasing again towards values of leO and le 1. An identical

experiment was performed on the raw powder and immediate stabilization was

70

 

 

800x Micrograph

 

 

 

3500X 3500X
Micrograph Micrograph

 

 

 

 

3500X 3500X
Micrograph Micrograph

 

 

 

 

3500X 3500X
Micrograph Micrograph

 

 

 

 

3500X 3500X
Micrograph Micrograph

 

 

 

 

 

 

 

 

 

 

 

 

Figure 13. Illustration of grid analysis used to determine maximum particle size.

 

71

 

Figure 14. SEM micrograph
of Fe-40AISED. (850K).

 

72

 

Figure 15. SEM micrograph
of Fe-40AISED. (3500X).

Ion

1011.

31h

P161

Obgf

73
obtained for a wide range of pH values. It was then concluded that possible surface

reactions had taken place by allowing the surface to remain hydrated in an
nonprotective atmosphere.

An attempt was made to remedy the surface change problem previously
encountered. A furnace treatment was employed which would attempt to dehydrate
the powder. The variables selected for the furnace operation included a 2 hour furnace
u'eatment at a temperature of 110°C. Table 9 provides the relative weight of the
powder sample with respect to operation time.

Table 9. Wet Powder Furnace Treatment Data

Average Weight (g) Furnace Time Elapsed (min) I

 

54.2096 30

 

53.9378

 

53.9242

 

53.9372

 

 

As can be seen in the previous table. the weight of the sedimented powder sample no
longer decreases after a 2 hour fin-nace operation. thus showing that the sample is no
longer hydrated. At the conclusion of the furnace operation, the specimen still
exhibited some degree of instability when suspended and the pH monitored. as seen
previously. It was concluded that the specimen may have undergone frnther sm'face
damage during the furnace operation, possibly oxidation. In light of the previous

observations. the frunace operation was negated as a possible solution to the instability

74

problem encountered. The solution to the instability problem was ultimately produced
by developing a vacuum assisted desiccating system in order to completely dry the
powder immediately after it was sedimented. Upon performing this type of
dehydration process. the Fe-40Al powder exhibited a vast decrease in the observed
instability. The remaining instability resulted in a much slower increase of the pH
level with respect to time. The small degree of instability may have resulted in
adsorbed ions present on the particle surface which will become active when re-wettcd
by an aqueous solution.

The other intermetallic Fe-40Al powder was supplied by the NASA/Lewis
research Center, Cleveland. Ohio. This powder was obtained in a -325 mesh fornr
which corresponds to the largest screen opening of 441m. The same grid analysis
procedure for verifying the expected particle size range was used. Since it was
initially assumed that the as-received particle size was acceptable. no addition
reduction methods were employed. Therefore an instability problems observed while
using the previous reduction technique could be avoided. The SEM micrographs are

shown in Figures 16a and 16b.

Processing Parameters

The next major area of experimentation involved the determination and
examination of two processing parameters. These included the calculation of the
critical coagulation concentration (c.c.c.) and the determination and examination of the

effective particle sizing. Both of these parameters are extremely important with

 

75

 

Figure 16a. SEM micrograph
of Fe-40Al325. (200X).

 

III.- Il... {I ( ;‘§§‘(\(’TI/

l‘v

76

 

Figure 16b. SEM micrograph
of Fc-40A1325. (llOOX).

77

. respect to the generation of the characteristic potentiomeuic titration data and the
subsequent stability prediction. Therefore the accuracy of the BSA and stability data
is dependent on the accuracy of the values obtained for the two parameters listed
above.

The c.c.c. value is defined as the specific concentration for which the
suspension coagulates and forms flocs which fall out of the system in a rapid manner,
and the concenuation at which the suspension exhibits stable behavior. During
thecourse of the experiment it was originally intended to provide c.c.c. data for the Fe-
40AlSBD powder. Instead. the Fe-40AlSBD was not included as acceptable data for
the calculation of the c.c.c.. This was primarily due to the previously explained fact
that when powder is hydrated and then dried. the adsorbed ions, which were present
when the particles were wet, become active upon rehydration of the particle surface.
During the attempt to calculate the c.c.c. using 2"/o suspensions containing Fe-
40AlSBD powder, the time-lapse sedimentation utilizing the different concentrations
were initially successful. Due to the limited supply of the Fe-40AISBD powder. the
powder used for the initial c.c.c. calculation was wet-filtered through a cellulose
membrane. Since a proper washing technique was not employed. the electrolyte ions
which were present in solution during the first run of the c.c.c. experiment were
reactivated during the second time-lapse sedimentation. This resulted in equal settling
in each test tube so no differentiation can be made as to which should settle out and
which should not. The result of this initial c.c.c. experiment verified the existence of
adsorbed ions since each test tube suspension exhibited the same behavior.

At the conclusion of the initial attempt to determine the c.c.c. for the Fe-40Al

system it was decided that further endeavors in its determination should employ the

78
use of the Fc40AB25 powder instead of the Fc-40AlSED powder. This is primarily

due to the abundance of the Fe-40Al325 and the realization that this particular powder
would most likely be employed in the processing of the IMC. The c.c.c. was
determined in the same manner as previous attempts. The procedure for the
calculation of the c.c.c. is given in detail in the previous section. The time lapse
photos for the calculation of the c.c.c. for the Fe-40Al325 system are shown in Figures
l7, l8, and 19. The end result of the experiment is dependent on each stage of the
experiment. Initially. the solutions are mixed to the specifications present earlier and
sonicated with an ultrasonic probe. This sonication procedure ensures that any flocs
which form as a result of initial mixing are broken up and act as individual particles.
The solutions are then shaken and allowed to stand for 2 hours. This stage of the
experiment allows for the formation of the flocs. The suspensions are then
resuspended and allowed to stand. This is the most important step within the
experiment. This provides an actual representation of the effect of a particle floc as it
falls through a suspension. The floc will not only cause the particles comprising it to
fall out of the suspension. but also those particles which get pushed to the bottom by
the moving floc. As can be seen from the following figures, the suspensions
containing the 0.1N and 0.01N KNO3 electrolyte concentrations virtually settle out
completely. The suspensions with values of 0.0001N and 0.001N KNO3, negating the
effects that Stokc’s law has on the larger particles in the distribution. exhibit stable
behavior. Therefore. the values of 0.0001N and 0.001N KNO3 electrolyte
concentration. are considered acceptable for use in the processing of the IMC.

The next parameter analyzed was the effective particle size of the powder in

suspension. The scope of this experiment included the analysis of several variables.

rv r . _/
w w v v v L

:- mvmrw w v-rw

l V “v. _z 77

_

y

F “A“ nae—4......—

a T‘TI‘.TIITIIII9IIIII Inglis“. their. “ti-hue. 1.3.1.. 2%.. ‘ Nit-pl xiv .h. V .tIt.II‘r-‘.II‘I'II III.‘ I II 2|.I:I.I. IIIlIIII It. I II I III. .Itl) IL. 2‘1'. *Ifiv‘lmr‘.!_.;- ‘—

79

 

Figure 17. Time lapse photograph for c.c.c.
determination. (time = 0 seconds).

 

80

 

Figure 18. Time lapse photograph for c.c.c.
determination. (time = 45 seconds).

 

81

 

Figure 19. Time lapse photograph for c.c.c.
determination. (time = 90 seconds).

82
The initial purpose of this experiment was to verify the previous experimental work

performed which concluded that the particle size limits which were specified were
accurate. Upon beginning the effective particle size measurements using the Malvern
Mastersizer. it became possible to measure several other variables. Additional
experimental variables examined included: (1) measuring the effective particle size
with respect to different electrolytic concentrations and to compare those values with
the experimental work performed to calculate the c.c.c. for the systerrr, (2) examine the
flocculation behavior of the suspension with respect to pH and electrolyte
concentration. and (3) explore the possibility of examining flocculation kinetics.

The first experiment performed utilizing the determination of effective particle
sizing included comparing the data gathered from performing the calculation of the
c.c.c. with values obtained for the effective particle size at the same electrolyte
concentrations. After the data for the effective particle size was corrrpleted using the
Malvern Mastersizer, several results are given which were then used to interpolate
different types of information. With respect to comparing the values from the
Mastersizer and the c.c.c. time-lapse photo experiment. the results used to characterize
the data include the mean particle size. (measured in um), of the entire distribution
and the specific surface area, (measured in mzlcc) of the suspension sample. The
results for the four particle sizing data runs are presented in Tables 10 and 11 on the
following pages. By examining the data, the following values were obtained for the
Fe-40AIS- powder with respect to elecu‘olyte concentrations of 0.0001N, 0.001N.

0.01N. 0.1N KNO3:

83
D[v,0.5] = mean p.s. = 5.32pm for 0.1N KNO...

= 4.89pm for 0.01N KNO3.
= 5.3011111 for 0.001N KNO3. '

= 5.53pm for 0.0001N KNC,

Specific S.A. = 1.2746m2/cc for 0.1N KNO3.

= 1.3846m’lcc for 0.01N KNO.

= 12867me for 0.001N KNO3.

= 1.2292m2/cc for 0.0001N KNO3.
By examining the values listed above and on page 84, no differentiation or correlation
can be generated from the patterns exhibited. The most reasonable explanation for the
inconclusiveness of the data taken for the Fe-40AlSED is that the particle size range is
so narrow that there exists minimal flocculation since there is not an over abundance
of very large and very small particles. This eludes to the explanation presented by
Hogg. Healy and Furstenau [20]. where it was stated that the prediction for stability
within a multicomponent system does not apply for particles in which there is such a
large size difference. Therefore for a system where very large and very small particles
exist. these particles will tend to attract more readily than those which are more equal
in size. It is difficult to evaluate any flocculation that may be present because the
interaction of large and small particles do not adhere to the governing stability and
may flocculate. Therefore. the amount the Mastersizer laser beam is defracted is so
small that the results obtained do not outweigh those values associated with the
standard deviation. Flocculation does exist in some capacity if the previous grid

analysis for particle size limits is used as a benchmark for comparing the data obtained

84

Table 10. Particle Size Data for 0.1N KNO3 Fe-40AISED

 

Table 11. Particle Size Data for 0.01N KNOa Fe-40AISED

Length-14.3mm
112.2545

I 0.1348
Dim‘

 

85
Taple 12. Particle Size Data for 0.001N KNO3 Fe-40AISED

 

Table 13. Particle Size Data for 0.0001N KNO3 Fe-40AISED

 

86

for the effective particle size. As can be seen fi'om the histogram data in Tables 10
and 11. between 7.5% and 8.7% of the samples tested were above the upper limit of
1011m. Therefore. these particles which are measured as being greater than 10pm can
be considered as the result of flocculated particles. The pick-up lens of the
mastersizer. which produces the diffracted data for the measurement, is unable to tell
the difference between particles which are a single entity or an entity which is made
up of a collection of particles. Therefore. the computer pick-up lens will report a floc
as a larger particle rather than a collection of snull particles.

The flocculation behavior for the Fe-40AISBD was more apparent when
comparing the effective particle size values over a range of pH values. This is shown
by examining Tables 12. 13, and 14. Each of the data acquisition runs was performed
at a pH 3.0, pH. 5.5 and pH 8.0 respectively. Again, the variable which provided the
most revealing information was the amount of fines which were below the project

lOttm limit. These values are listed as follows:

pH 3.0 : for a high size = 9.94pm : 96.0% of sample below
pH 5.5 : for a high size = 9.94pm : 96.4% of sample below

pH 8.0 : for a high size = 9.94tun : 89.3% of sample below.

This result indicates that the flocculation behavior is excited at the higher levels of pH
due to the amount of fines below the project lOum limit decreases as the pH of the

suspension increases.

87

Table 14. Particle Size Data for 0.0001N KNO3 pH 3.0 Fe-40AlSED

Name My
Infill-lulu:
”331de

 

Table 15. Particle Size Data for 0.0001N KNOa pH 5.5 Fe-40AlSBD

 

88

Table 16. Particle Size Data for 0.0001N KNO3 pH 8.0 Fe-40AISED

143m "nip

Sample
lads-labia

Dhtn‘ht'xn IIJS’" Ifl'loc

 

The initial experiment performed on the Fe-40AlSED for comparing the time-
lapse c.c.c. results with the effective particle size data was also performed on the Fe-
40A1325 powder. Similar suspensions were mixed and sonicated prior to measuring
with the Mastersizer. The results for each of the data runs are given in Tables 15, 16,
17, and 18. By examining the data, the following values were obtained for the Fe-
40A1325 powder with respect to electrolyte concentrations of 0.0001N, 0.001N, 0.01N,

0.1N 1010,:

D[v,0.5] = mean p.s. = 27.67pm for 0.1N KNO3.
= 23.32pm for 0.01N KNO3.
= 23.18pm for 0.001N KN03.

= 23.71pm for 0.0001N KNO3.

89

Table 1?. Particle Size Data for 0.1N KNO3 Fe-40A1325

up

MOW”
SA-Omm’lx

 

Table 18. Particle Size Data for 0.01N KNO3 FefiQAlB2§

up

Can-W35
SA-OJ’TBSII’IOC

 

90

Table 19. Particle Size Data for 0.001N KN03 Fe-4OA1325

. any» mun-143m up
Inga-soon mm
.u non-I9 elm-comm
W rut-ammo:

 

 

Table 20. Particle Size Data for 0.0001N KNO Fe-40A1325

 

"A 4
- ‘ _
- ‘
‘8
,.
\

 

 

 

 

 

Tarrrrrrfi
'-TFT?TFTT?T?:

 

 

 

fl

FT.
——TTTTTTTTT_

 

 

u
u
“i?“
Jan
I»
~-J~
Jae
In
Jan
u
in

91
Specific S.A. = 0.3269m2/cc for 0.1N KNO3.

= 03785me for 0.01N KNO
= 04004me for 0.001N KNO3_

= 0.3834m2/cc for 0.0001N KNO3_

By initial examination of the mean and specific sm'face area values it is apparent that
flocculation is present at an electrolyte concentration of 0.1N KNO3 and not very
apparent at the other levels of electrolyte concentration. Even with respect to the
results generated for the mean and specific surface area, the values presented could be
suspect and not necessarily represent flocculation since there exits no understanding as
to the standard deviation of the measure. A qualitative estimate can be made though
regarding the flocculation behavior with respect to pH if we consider the average
particle size at the 90th percentile. This particular percentile is chosen since we are
concerned with particles coagulating and forming flocs which will be represented at
the upper end of the sizing spectrum The values for the 90 percentile are given as
follows:
D[v,0.5] = .90 particle size = 66.06pm for 0.1N I-INO,.
= 55.42pm for 0.01N HNO,.
= 51.43um for 0.001N HNO,

= 56.23pm for 0.0001N moi...

From examination of these results it is a much more apparent comparison than that
generated by the mean and specific sm'face area results. A further comparison can be

made with the c.c.c. data by using the data representing the amount of the sample

92
which is less than the expected value of 44m for the Fe—40Al325. This data is

represented as follows:

0.1N KNO3 = for a high size = 46.2].tm :9 78.4% of sample below
0.01N KNO3 = for a high size = 46.2pm = 84.6% of sample below
0.001N KNO3 :9 for a high size = 46.2um = 86.1% of sample below

0.0001N KN03 :9 for a high size = 46.2}.tm = 84.1% of sample below

This data also confirms the fact that the flocculation behavior has increased for an
electrolyte value of 0.1N KN 03. Although this data show a correlation with the
observed experimental results used in determining the c.c.c., it does not account for
the fact that the c.c.c. visually determined is between 0.001N KNO3 and 0.01N KN03.
This observation can be attributed to the fact that the driving force for the
determination of the critical coagulation concentration is the evolution of the
flocculation kinetics inherent to the particular system. It became apparent during the
conclusion of the effective particle sizing experiment that it was not feasible to
examine the flocculation kinetics of either the Fe-40AlSED or Fe-40A1325 systems.
Because the sample sizes required for the particle size measurement in the Malvern
Mastersizer are so small, (sample size range 0.0011% to 0.0073), that the interactions
of the individual particles is dramatically decreased, therefore the probability for the
formation of floc will also decrease. Furthermore, the sample container and the area
within which thelaserbeamactsisalsosolimitedthatthetimerequiredfor
measurement, (approximately 5 min. for 1000 sweeps = single data acquisition), far

exceeds the settling time of a particle in an upper size region. Therefore, a majority

93
of the particles which actually do flocculate, have already settled out before the

measurement takes place, so they are simply ignored when the final data is processed
and the results are received. Therefore it is highly probable that the results presented
here only indicate a fraction of the total flocs that are formed during the experiment.
Even though the data acquired does not give the desired representation that

would allow for the examination of flocculation kinetics, it does present some insight
as to the flocculation behavior with respect to pH. Fe-40Al325 samples were again
prepared in a similar fashion as the Fe-40AlSED and an effective particle sizing run
was performed for the same pH range. The complete particle size data with respect to
pH is given in Tables 19, 20, and 21. A similar set of data is presented here which,
again, represents the amount of powder which is below the specified of the upper size
limit:

pH 3.0 = for a high size = 46.2tlm = 79.4% of sample below

pH 5.5 = for a high size = 46.an = 80.6% of sample below

pH 8.0 = for a high size = 46.2um = 89.2% of sample below

Therefore for a pH of 8.0 the amount of powder within specification exceed the other
pH values by approximately 10%,thus indicating that flocculation is much more
prevalent at the lower pH values. This point is can be enrphasized even further by
once again examining the 90 percentile particle size value for each of the pH ranges.
This data is given as follows:
D[v,.90] = .90 particle size = 524.58prn for pH 3.0 .
= 111.05ttm for pH 5.5 .

= 47.32pm for pH 8.0 .

94

Table 21. Particle Size Data for 0.0001 KNO3 pH 3.0 Fe-40Al325

 

Table 22. Patricle Size Data for 0.0001N KN03 2H 5.5 Fe-40A1325

 

95

Table 23. Particle Size Data for 0.0001 KNOa pH 8.0 Fe-40A1325

 

This data clearly shows that with respect to flocculation initially present at a pH 3.0 is
far greater than that experienced during the measurements taken at pH 5.5 and pH 8.0.
This also gives a better understanding as to the differences in flocculation at the
various pH values. The specific importance of this data will be expanded upon when
the data for the BSA measurement and subsequent stability prediction is presented.

By being able to measure the particle size of the powder as it "appears" in
solution, an accurate determination of the effective particle size can be obtained. As
was discussed previously, one of the most important input variables involved in both
the ESA measurement and the stability prediction is the particle size. Even though an
accurate particle size estimation can be made on a dry powder sample using a variety
of techniques, the BSA measurement takes into account that the particle is in
suspension and not dry. By using a particle size which is inaccurate, the zeta potential

values obtained by the BSA measurement will change in magnitude. This was

96
experienced during previous work performed on the Fe-4OAlSBD [15,29]. Generally,

particle sizes, entered into the computer prior to BSA measurement, which are greater
than the actual in-situ value observed by the BSA will tend to shift the zeta-potential
curves down in magnitude. Therefore, accurate curves can be obtained but they are

limited to the accuracy of the measurement used for the input parameters.

BSA Measureflnt and Stabilig Prediction

The next step in the development of a technique for processing IMCs via
electrophoretic deposition is the analysis of the zeta potential of each component of
the system and the subsequent stability prediction associated with this. As was
discussed in the section on colloidal theory the BSA technique can measure
characteristic zeta potential values for a range of pH values. The specifics of this
measurement include the micro-titration of acid or base in order to achieve different
pH values. The BSA then takes individual measurements of potential at each of the
pH values. Once this data is compiled, a stability prediction program, developed by
Wilson, is used to give a theoretical prediction of the degree of homo- and
heterocoagulation between the system components.

The first potentiometric titration performed include the Fe-40AlSED and the
ball milled FP A503. An SEM micrograph showing the ball milled A1203 is shown in
Figm'e 20c. Both titrations were performed independently. Each titration consisted of
mixing with a neutral solution of deionized water at a pH of 7. The 1.0N HNO3
solution was then micro-titrated to lower the pH to a level of pH 4.0. Once this level
was established the titration began and the zeta potential measurements were taken

over the range from pH 4.0 to pH 10.0. The plot corresponding to the potentiometric

97
titration for the Fe-40AlSED and FF A1203 is shown in Figure 21. Upon examining

the curves it is apparent that the A1203 exhibits expected behavior compared to data
seen elsewhere [15]. The A1203 zeta potential curve shows that both positive and
negative values are present and that the A1203 has an iso—electric point or i.e.p. at pH
8.5, where homocoagulation will be at a maximum. Recalling the previously
established criteria that homocoagulation is minimized for higher values of pH, the
curve indicates that the flocculation of FP Al¢03 with respect to itself is minimized for
pH values less than approximately pH 8.0. Otherwise, theory dictates that the
homocoagulation of A1203 will be maximized for those pH values which are within the
range pH 7.5 surrounding the i.e.p.

The Fe-40Al curve presented in Figure 21 illustrates markedly different
behavior than that showed by the potentiometric titration for A1203. Assuming again
that theory is correct and that a like component will flocculate for zeta potential values
which approach zero, one would expect flocculation to worn at all pH levels as show
by the characteristic potentiometric titration. Although, theoretically, flocculation is
known to exist over pH ranges which are near zero, (i.e. near the i.e.p.)and since no
i.e.p. is shown for the Fe-40AlSBD curve, it can also be stipulated that for positive
zeta potential values, flocculation will be at a minimum for pH values where the zeta
potential is maximized. Using this realization, a range for which the zeta potential is
consistently maximized is between pH 5.5 and pH 8.5. In comparing the
potentiometric titration data with the previous effective particle sizing experiment, the
range identified as that of minimum attraction does not agree entirely with that shown
by the Mastersizer data. The most apparent correlation between the two sets of data is

that for a pH 5.5, the flocculation as shown by the particle sizing is at a minimum,

’1
r.
I
l
l
l
l
l
r

98

 

Figure 20a. SEM micrograph of PP
A1203 Fiber bundle. (500X).

b”‘*"‘"\.f\~’\/\r\'\’\MMGAMAAAA «exflx

Elillvlllllllllll'vlillllnlfiq

 

Figure 20b. SEM micrograph of FF
A1203 single fiber. (4000X).

‘V -' , J
—r‘w—'vd'\’ rr‘

W w ‘9- "'~.-!’" 'q,-." -

 

 

Figure 20c. SEM micrograph of ball-
milled alumina. (800K).

101

 

 

   
   

 

 

30
1 Component
2 * Alumina
25:” O FeAOAISED
20:-

A

0'

Jill
1

Zeta Potential (mV)
<75

 

 

 

 

0:
-103...+.+...:...
3 4 5 6 7 8 910

 

 

 

Figure 21. Zeta potential data for Fe-40AlSED and FF A1103. Titration
from pH 4.0 to pH 10.0.

102

and the zeta potential is at a maximum, therefore also illustrating minimum
flocculation.

The second potentiometric titration performed involved the Fe-40A1325
powder. The titration was performed over a range of pH 3.0 to pH 11.0. The data
presented in Figure 22 includes the data for the FP A1203 that was previously obtained.
This data is superimposed on the same plot as before for comparative purposes. As
can be seen from the plot presented, the data points for the Fe—40A1325 are scattered
in a much more exaggerated fashion than that exhibited by the Fe-40AlSBD. This can
be primarily attributed to the wider range in particle size for the -325 mesh sample.
This identical phenomena was also exhibited in previous work [15,29]. Again, using
the assumption that the flocculation is minimized for maximum values of zeta
potential, the curve for the Fe-40A1325 sample in Figure 22 shows minimized
flocculation to occur at pH levels above 5.5, because the zeta potential at these pH
values is consistently maximized.

Once the BSA measurements were completed, the zeta potential vs pH values
could then be imported into a computer program developed and written by Wilson and
Crimp [15,30] which theoretically predicts the colloidal stability of multicomponent
composite systems. As was shown in Figure 6, the there exists a total energy of
interaction which varies with increasing interparticle distance. The interparticle
distance, H, at which the total potential energy, V1, is maximum is know as the
distance at which the potential barrier exists. As was stated previously, the stability of
a system is indicated by graphing the total potential energy of a system versus the
interparticle distance. The condition of stability is accepted where the potential barrier

is at least 20-25kT [15,20,31]. It was also previously introduced that the stability ratio

103

W of a system is equivalent to the ratio of the number of particles interactions and the
number of particle interactions which result in flocculation. A stable or unflocculated
dispersion is approximated by log(W) > 10 which corresponds to a potential barrier
being at least as large as 20-25kT. Therefore, the ultimate goal of the BSA
measurement and stability prediction is to generate a plot which illustrates the stability
behavior of a multicomponent system by plotting the log(W) values versus the pH at
which the corresponding zeta potential values were taken. From this plot, a prediction
can be made as to the IMC processing pH range for the Fe-40Al / A1203 system.

The zeta potential data given in both Figures 21 and 22 was inputed into the
stability prediction program previously described. The main goal for stability
prediction program is to gather predicted data values for each component which makes
up the total stability ratio, W1. These include the stability ratio for components with
respect to themselves, given by Wu and Wu, and with respect to each other, given by
Wu. The value of most concern here is the stability ratio which predicts
heterocoagulation, Wu. The purpose for using the experimentally determined data
generated by the BSA, is to give a prediction as to the range of pH which may be
suitable for the process and to compare the predicted processing range with that found
by physically attempting to coat the A1203 fibers with Fe-40Al particles at various pH
levels. This comparison not only reveals the actual desired coating range but also
examines the accuracy of the stability prediction as applied to this particular system.

The zeta potential data for both the Fe-40AlSBD/FP A1203 and the Fe-
40A1325/FP Al¢O3 systems was used to develop the stability prediction plots shown in
Figures 23 and 24. Both plots show almost identical behavior for the

heterocoagulation stability ratio, Wu. The prediction given by both plots shows that

104

 

 

Component

. . + Fe40Al325
25 _- I. g * Alumina

 

 

-L
0'!

LLLILJ

l l L I

Zeta Potential (mV)
8

 

 

 

 

 

 

 

Figure 22. Zeta potential data for Fe-40A1325 and PP Al¢03. Titration
from pH 3.0 to pH 10.0.

105
stability exists for pH values less than approximately pH 8.0. From this data, it is

predicted that the desired processing range for which maximum heterocaogulation
would occur is located at the higher pH values of pH 8.0, pH 9.0, and pH 10.0 from
the stipulation that stability is assumed for log(W) values >10. The most promising
aspect of these plots is shown in the stability prediction for the Fe-40A1325 / FP A1203
plot which shows that the Fe-40Al powder is stable with respect to itself

for the range of pH 4.5 to pH 10.0 while the components exhibit instability for pH
ranges in excess of pH 8.0, thus resulting in heterocoagulation.

By comparing Figures 21 and 22, the curves for both Fe-40Al appear to be
different in shape and in magnitude. This discrepancy is magnified upon incorporating
the zeta potential data into the stability prediction program. This can be seen by
examining Figures 23 and 24. Given that both powder samples are approximately
identical in terms of atomic weight, it was initially assumed that both samples would
generate approximately the same prediction. Since the same prediction was not
achieved, some possible explanations for the difference in the acquired data for both
samples is given. First, as was stated before, there existed a fundamental difference
between the sedimented powder and the as-received -325 mesh powder, in that the
surface charge of the sedimented powder was affected by the adsorbed ions which
were present as a result of the sedimentation procedure. Second, even though the
powder samples were almost identical in terms of atomic percent, the -325 mesh
sample contained small additions of B and Zr to enhance mechanical behavior of the
intermetallic. The constituents of the Fe-40Al325 sample are (in atomic percent)
59.70% Fe, 39.80% Al, 0.10% Zr, and 0.41% B. This is compared to the Fe-

40AlSBD sample which contained (in atomic percent) 60.00% Fe and 40.00%Al.

106

 

 

350

 

 

 

 

 

 

 

 

 

 

 

Figure 23. Stability prediction for Fe-40Al325 and A1203 multicomponent
system.

107

 

 

 

 

 

 

 

 

 

 

 

 

 

350
300.
250{
d l
. |
200:- g
‘ StablltyRatloe
E 3 ---w11
0,150:-
3 3 +W12
1oo-:L ,. _ *W22
l ..
50‘” I .
i
: .,‘ . . ’1’
or 744-1.. .’
I i. 'I..'.. ‘1‘.
'50 I '1 l i l : I i r
4 5 6 7 9 10
pH

 

Figure 24. Stability prediction for Fe-40AlSBD and A1203 multicomponent

system.

 

108
Therefore, even thought the additions of Zr and B are so minimal, they should not be

disregarded as having no effect on the chemical make-up of the powder surface. The
third possible reason for the discrepancy in the two data sets is that the particle size
range for each of the samples was markedly different, (1 < 10ttm for the
Fe-40AlSED and d < 44pm for the Fe-40A1325. This difference not only changes the
zeta potential behavior but also significantly alters the stability ratio data describing
homocoagulation of the Fe-40Al powder. This difference is primarily due to the
settling out that occurs during the BSA measurement. For example, if a powder of
relatively small size, (1 < 10pm, were suspended in a dilute concentration, 2"/o, the
settling time for this suspension is relatively adequate with respect to the measurement
time. Therefore the ESA system, is "tol " that the average particle size is
commensurate with the value obtained during the particle sizing. If, during the
measurement, the powder remains suspended then the BSA will "see" an average
particle size concurrent with the value that was initially used. On the contrary, if the
system is given an average particle size value for a larger powder range, i.e. d <
44pm, and the powder sample experiences some significant settling effects during the
measurement, then what the BSA "sees" as the average particle size and what that
actual particle size is are two different values. The value that is inputed into the
system will subsequently be larger than that which is actually "seen" by the BSA
measmement system Therefore with regard to the difference in data, the fact that
there is some error experienced when using a larger particle size range explains the
discrepancy in the data between the two powder samples.

This fact was illustrated by performing two additional stability prediction runs

which would subsequently use reduce average particle size values. The first additional

109
stability prediction used an average particle size of d = 11.86pm, which corresponds to

half the value used for the previous run. The second additional stability prediction
used an average particle size of d = 2.46pm, which corresponds to the value used for
the run using Fe-4OSED. This allow for the generation of a series curves which can
be examined to see if a trend exists for decreasing particle size. The stability
predictions generated are shown in Figures 25 and 26. As can be seen from these
plots the curve corresponding to the homocoagulation of Fe-40A1325 decreases
considerably in magnitude and approaches the behavior exhibited by the curve for the
Fe-4OS- shown in Figure 23, as the particle size decreases. Therefore, this
illustrated the importance in having an accurate correlation between the actual particle
size of the material and the particle size that is used by the ESA to provide the zeta
potential data.

Even in light of the differences which exist between the two powder samples,
the predicted stability ratio curves for the heterocoagulation of Fe-40Al and A1203, i.e.
Wu, change very little. Therefore, since it is the stability ratio describing
heterocoagulation, W12, that is of most concern, the discrepancy between the Fe-40Al
powder samples can be disregarded. An additional experiment was performed which
used the previous data for the Fe-40A1325 plot with exagerrated zeta potential values.
This data is shown in Figure 27 . Once the hypothetical potential values were
compiled, a stability prediction was again performed to see if any changes would
occur in the W12 stability ratio which governs the heterocoagulation. It should be

noted here that the possibility exist for taking single point measurements using the

110

 

300
250
l
200

150

L09 (W)

100

50

 

350*

 

Stability Mice

---W11
+w12
+W22

 

 

 

 

 

Figure 25. Stability prediction for the Fe-40A1325/A1203 system using
p.s. = 1/2 simulation.

lll

 

 

   
   

 

 

 

 

 

 

350
Stabilityfiatloe
1 "’W11
300 -.--.- ""O\ .
\\ *W12
+W22
250-
2001
8,150:
—' a
100{
50{
q
I
'50‘I'l'i'l'i'l
4 5 6 7 8 9 10
pH

 

 

 

Figure 26. Stability Prediction for Fe-40Al325 / A1103 system using
p.s. < 10|rm simulation.

112

 

 

   
 

 

 

 

ao .
. Component
: * Alumina

25 T + Fe40AISIM
1

209-

A

0"
1111
l

I l l

Zeta Potential (mV)
2;

 

 

 

 

 

 

 

Figure 27 . Zeta potential vs. pH data for the Fe-40A1325 Simulation
(titration from pH 4.0 to pH 10.0).

1 13
BSA. One possible method for taking the flocculation kinetics into account is by

performing point measurements using data from the Mastersizer for the same pH
values. This would allow the zeta potential to be measured with repect to the particle
size at that particular pH instead of assuming that the particle size remians constant
over the entire pH range. The hypothetical stability prediction plot for the Fe-40A1325
/ A1203 system is shown in Figure 28.

By examining the data it is apparent that the exaggerated zeta potential values
have a significant affect on the WH stability ratio behavior, as expected. These
exaggerated values, however, do not provide any change in the stability ratio
governing heterocoagulation, W“. This result indicates that the heterocoagulation data
is controlled by other factors such as particle size and flocculation kinetics. The
significance of this result will be explained in the next section. Experimental
verification utilizing the both of the Fe-40Al powders and the FP A1203 fiber is

presented in the next section.

Fiber Coating
As was stated previously, the stability prediction program revealed that the

ideal processing range for the Fe-40Al/Ale3 system existed at pH vales greater than
pH 8.0. The net set of experiments performed were designed to not only verify the
accuracy of the prediction relative to the Fe-40Al/Ale3 system but also to determine
the actual ideal processing range by physical coating the FP A1203 fibers with the Fe-

40Al matrix powder. Two experiments were set up for the characterization of particle—

114

 

 

   
   

 

 

 

 

 

 

 

350 .
.. Stabllty Ratio.
2 ---W11
3001-0-
; +w12
j +W22
250:-
200{-
E 1 s
m150':’ :
O 4 j
"’ . :
wo-i- .2
so{L i
1 I 5'
0. -. 5
: a;
'50 r i I II # i l I I ll r
4 5 6 7 8 9 10

 

 

 

Figure 28. Stability prediction for Fe-40Al325 simulation.

1 15
coated fibers. The first experiment was designed to qualitatively evaluate the

magnitude and distribution of coating with respect to varying pH values. The results
were obtained by visually examining the residual coating of the fibers by the matrix
particles at different pH values. The term "residual" is used to define the amount of
matrix which is electrophoretically deposited onto the fiber when the fiber is immersed
into the powder suspension. The residual coating experiment was primarily focused
on using the Fe-40Al325 powder as the matrix instead of the Fe-40AlSED. For
reasons that will be expanded upon later, in the best interest of the research project, it
only became feasible to use the -325 mesh powder for the matrix. The second
experiment performed expanded upon the results of the first experiment utilizing the
scanning electron microscope to characterize the coated specimens.

The residual coating experiment was performed by mixing up a 25"/o Fe-
40A1325 powder suspension. Once the desired pH of the suspension was obtained and
stabilized, a previously cut 70mm fiber strand was immersed into suspension and then
immediately placed on a glass slide and labeled. The basis for this experiment was to
allow the surface tension of the water to remove the powder form the fiber surface by
allowing the wet fiber strand to come into contact with the glass microscope slide.
This allowed for the amount of "residual" powder to be visually characterized and
subsequently the ideal processing value(s) identified. As was observed, the pH range
which generates the highest magnitude of coating is between the ranges of 5.5 and
6.0. This differs slightly from the predicted values of pH 8.0 and higher. The values
that each coating was performed at is actually represented by a small range. For
example, the value of pH 5.5 was actually between the range of pH 5.5 and pH 6.0.

this approach had to be taken as a result of the minor problems associated with

l 16
achieving a pH value and holding that value to an accuracy of two significant decimal

places by titrating different normalities of acid and base.

Once the overall pH range was macroscOpically characterized and the actual
ideal processing range established, the second experiment could then be performed
utilizing the SEM. Initial coating experiments were performed which utilized the Fe-
40AlSED powder. An experiment similar to the residual coating experiment was
performed by A. Martin revealed that the ideal processing range would also be
between pH 5.0 and pH 6.0. Therefore, a 10"/o suspension was prepared using the
Fe-40ALSED and specimens were prepared over a range of pH’s from pH 4.0 to pH
7.0. The major reason that the number of specimens prepared for SEM
characterization was substantially less than those prepared for residual coating was due
to the immense difficulty in preparing the SEM samples. Due to the relative
sensitivity of individually coated fiber samples being handled. In addition to this,
difficulty was consistently encountered in providing a conductive sanmle within the
SEM. This was remedied by providing a 40nm thick gold coating using a sputter
coating device. The work involving the Fe-40AlSED was eventually replaced by the
Fe-40A1325 powder because of several problems encountered using the Fe-40AlSBD.

The major problem encountered while using the Fe-40AISBD was the
extremely limited supply of usable fines (i.e. d <10um). In order to generate enough
powder for a 5'lo suspension of powder extensive preparation time was expended for
the sedimentation. In addition to this, it was later noticed that in order for the coating
to be effective, larger volume percentage suspension were required. Therefore, due to
the shortage of Fe-40AlSBD, this requirement became impossible to fulfill. Another

additional problem encountered when coating the FP A1103 fibers with the

1 l7
Fe-40ALSED powder was that due to the reactivation of the adsorbed ions once the

suspension was formed. This made it extremely difficult to stabilize the pH in order
to coat the fibers and record the conditions at which the process took place in an
accurate manner. As pertaining to the formation of the SEM specimens, the major
problem encountered here involved excess damage of the coated fibers via the electron
beam. In light of many of these initial problems encountered, a successful coating of
an FP A1203 using the Fe-40AlSBD powder was achieved. The ideal coating was
achieved within the range of pH 5.5 to pH 6.0. As can be observed from Figure 29,
uniform coating of the FP fiber is achieved. Also evident from this picture is the
existence of some homocoagulation between the very large and small particles which
are active in the coating process. This attraction between like components can be
eliminated by narrowing the working size range of the powder used. This will not
only provide less homocoagulation but also aid in providing a more uniform
distribution of matrix particles around the fiber. The photograph in Figure 29 was
initially representative of the specimen itself prior to the electron beam damage which
eventually destroyed the specimen.

Once the Fe—40A1325 was obtained, the Fe-40AlSED was replaced and the
residual coating experiments were performed to provide an approximate processing
range were the SEM characterization could take place, as explained previously. As
shown in Figure 29, the range which produces the most residual powder is located at a
pH of 5.5 to pH of 6.5. Therefore, an SEM characterization experiment was
performed by coating three 70mm fiber strands at three different pH values. The
values chosen, as dictated by the residual coatingexperiment, were pH 3.0, pH 6.0, pH

8.0. Contrary to the previous experiment, the specimens produced for the SEM

I l '- I‘ I ' Kr .

’ M" ed a/ v/ -....

118

 

Figure 29. SEM micrograph of coated FP
A1203 fiber with Fe-40AlSED matrix
particles at pH 5.5 to 6.0.

119

characterization were submersed for 20 seconds in a 25"/o suspension in a "bow-like"
manner. This geometry was provided by securing each end and then producing a
bendin the fiber that produces a horse-shoe shape. This geometry is

preferred over the initial configuration used which included simply dipping the fiber
longitudinally into the suspension. This is because the geometry produced by forming
a bend and securing the fiber at both ends more adequately approximates the
interactions that the fibers and powder will experience in a continuous winding system.
Once the fiber were immersed for 20 seconds, they were then hung vertically to allow
for the water to run off the end of the fiber. This drying procedure was performed in
this manner so as to imitate the scenario that the coated fiber will go through as it is
pulled out of the suspension and wound onto the mandrel of the winding system. As
will be shown in the following micrographs, the surface tension of the water was not
great enough to remove the adhered powder along with it as it transversed the length
of the fiber when hung vertically. Figures 30, 31, and 32 show the variability
achieved in coating the fibers at pH 3.0, pH 6.0, and pH 8.0. As can be seen from
both Figures 30 and 32, unacceptable coating of the fiber is exhibited. This is
characterized by the absence of uniform coating and the small volume fraction of the
matrix particles present. An example of an acceptable degree of coating is exhibited
in Figure 31. From these micrographs it is apparent that the not only is uniform
coating of the fiber achieved but there exists an adequate volume fraction of powder
present. Also, infiltration is achieved between the fibers thus indicating that there is
no homocoagulation between the fibers themselves which would prevent this. A
higher magnification micrograph of the same specimen coated at pH 6.0 is shown in

Figure 33. Upon examination it is apparent that homocoagulation between the powder

 

VIII: ( ..I x f .. I z

,llll ll

120

 

Figure 30. SEM micrograph
showing the degree of coating
at pH 3.0 for the Fe-40A1325
/ FP A1103 system. (100X).

\1\/\;)))),.\z.i)\r\r )33.)\!\r\rllilxr)i ) ), \: \-).\.,\; \2 \2 \\

 

 

‘

I... [It .(l rill III. .II Ill .((i(({ (.(l .. f (.4. _ ---':‘¢““f ( [It lift

121

'i
l
r
i,

'9
’.
4..
i

 

Figure 31. SEM micrograph showing the
degree of coating at pH 6.0 for the Fe-
40A1325 / FP A1203 system. (80X).

PH. ...

lkfl "On‘c J

 

122

 

Figure 32. SEM micrograph showing the
degree of coating at a pH 8.0 for the Fe-
40A1325 / FP A1203 system. (100X).

,.

A _‘\'-‘\ 'fi\ 'K\

123

 

Figure 33. SEM micrograph showing the
uniform structure of the Fe-40Al325 / FP
A1203 system. (800X).

124
particles is limited and that an even distribution of powder around the reinforcing

fibers is achieved.

From the previous two experiments performed, the actual range within which
the IMCs should be processed at using fiber electrophoretic deposition method is
located between pH 5.0 and pH 6.5. This is verified by the production of individually
coated fiber specimens which show a uniform distribution of matrix particles and
exhibit limited homocoagulation with respect to both components. Now attention will
be shifted to present and discuss the Filament winding system developed for future

implementation into the [MC production process.

Filament Winding System

The fundamental intent for the design of the Fiber Electrophoretic Deposition
(FED) filament winding system was to develop a prototype winding system which
would enable future processing of continuously reinforced IMCs. The basis for the
design of the winding system followed the fundamental requirements pertaining to
conventional winding systems in addition to those requirements generated by the use
of electrophoretic deposition.

The overall design requirement was to provide a laboratory scale winding
system which would generate IMC composite green bodies for consolidation into near-
fully dense composites. Figure 39 shows the completed prototype in its current state.
From this photograph, the several components necessary for a successful winding
system are shown. These include: (1)the fiber delivery assembly, (2)successive
multiple rollers, (3)suspension bath and roller/guide cradle, (4)take-up mandrel, and

(5)system control panel. In addition to the global requirements already discussed,

125
there exists specific design requirements which pertain to each component of the

system. The design of each of these components will be discussed in detail.

The main design requirement of the fiber delivery system was to force the
multistrand fiber to spread-out or "lay down" prior to entry into the impregnating
suspension bath. This is initially accomplished by providing a consistent drag force
upon the fiber while it is being pulled through the system. This is accomplished by
implementing a bearing assisted shaft though the multistrand filament spool. The
specific bearings were selected such that the diameter was large enough to provide
enough positive tension without the fiber brealn'ng. Figure 35 shows the finished
filament delivery assembly. The bearing rod is threaded along with the end caps so
that the spool is secure during operation and can be easily changed when needed.

In addition to providing positive tension to the multistrand fiber, there exists
additional requirement necessary for adequate fiber lay-down. This is accomplished
by allowing the fiber to transverse over a succession of bearing assisted rollers which
provide the curved surface over which the multistrand fiber will spread out. The
specific multiple rolling assembly developed for this system included the use of 3
bearing assisted rollers made from ultra-high density polyethylene which have 2
dimensional rotational motion and 1 bearing assisted roller with both rotational and
transverse motion. The purpose for the rotational and transfers motion in the last
roller is to provide an multi-adjustable roller which will give prOper alignment of the
fiber into the suspension bath. The multiple rolling assembly is depicted in Figure 34.

Once the fiber has properly spread out, it is ready to be incorporated into the
suspension bath were the matrix powder will infiltrate and electrophoretically deposit

onto the fiber strands. Several design requirements were adhered to. The first

126

requirement involved providing enough space to accommodate an acid/base titration
system, metering pumps and ultrasonic probe if necessary. Also, a cradle was
designed to further the spreading out of the multistrand fiber which was previously
done by the multiple rolling assembly. 3 rollers were fashioned from virgin Teflon
and given rotational motion with the structure of the cradle. The cradle itself was
provided with eyelets at each comer so that the entire assembly could be attached to
the winding system base via tension springs. Figures 36 and 37 illustrate these design
requirements. The remaining design requirement pertaining to the cradle assembly
involved providing the ability to guide the impregnated fiber out of the suspension and
onto the take-up mandrel. This was accomplished by cutting a 1mm wide square
groove 3mm deep into the roller which resides closest to the mandrel assembly. This
would groove controls the impregnated fiber onto the mandrel and allows the motion
of the mandrel to dictate the geometry of the wound fiber. The major difference here,
as compared to previously introduced design methods [27], is that the guide roller
must remain within the suspension. Whereas with conventional winding techniques,
there exists a guide roller directly atop of or even connected to the mandrel assembly
[27]. This is illustrated in Figure 38.

The final step in the winding process involves collecting the impregnated fiber
onto a mandrel for future consolidation using a variety of techniques [48]. The
design anddevelopment for the mandrel in this winding system was adapted from
previously done work by D. Maglaya [32]. The modifications performed on the
existing assembly included providing independent rotational and transverse motion so
that precise control can be achieved in producing an successfully wound IMC. This

independent motion was provided by attaching a one Dayton D.C. motor to the

5.4.1.

127

 

Figure 34. 35mm photograph showing the
various components of the laboratory-scale
FED filament winding system.

 

128

 

Figure 35. Multi-strand fiber delivery
assembly.

129
mandrel assembly and another Dayton D.C. motor to threaded cross-head assembly.

This is shown in Figure 39. Figures 40 and 41 show the complete mandrel assembly.
Several components of the assembly include: (1)teflon mandrel, (2)reversible
polarizing switches, (3)transverse motion cross-head assembly motor, (4)and mandrel
rotation motor.

Once the fiber winding system was completed, physical runs were made in
which a the A1203 was drawn through dry so that all imperfections and parameters
could be adjusted to provide a mechanically sound system. From the runs performed
the following judgements were made regarding the modifications for the FED winding
system. First, the fiber delivery assembly needed to be relocated to a position much
farther away than that originally anticipated due to the fact that the 11" fiber coil
provided from the manufacturer contained a helical wind and this caused excess
transverse motion of the fiber along the rolling assembly. In addition to this, the
suspension bath itself had to be elevated to lessen the radius of curvature that the fiber
experienced while going into and coming out of the suspension bath. The last
modification is currently involving the re-shaping of the teflon mandrel due to the

change in the foreseen consolidation requirements.

\

\

In )“u‘ddu‘JWJvfi/ I

/ _

j A

r a -. t. ’ " - .. l'
-—r - " ‘——- V—r — ~— — r“. v; ._. v" “ya-7W)? 2 ’y v3~ ‘v Wu F ‘I‘ wr ‘7' "' V v V J —

. .,_ -
_“ m-

 

130

 

Figure 36. Cradle-roller guide and
suspension bath assembly.

.0, .11...“ ... 4... 1.... s I a gqflaflhisfi .Uv ‘flflw

 

 

 

 

(

1“
(It (i {l ..
((
((( ‘.
(r ‘i
(1‘
‘
A. f
- “V‘
‘v-“ _

131

 

Figure 37. Rotational motion of
cradle-roller assembly.

 

4 .i . . . ‘.‘I r . 41".“. u .
.l 1! 1...! l'illllll .1 I‘n‘.‘ a «F.-. I I.

\./..\\I\I\./\))\}\I\l)\Ii)\.}\l\l))\\l\l.\lil\l\l)\.l;”””\l\,\i\Kl

.' -. . y, rk.’ v v w ‘— w w"-

1.1

ill", ii,

( (l {i (ll (1 i ,(i (l /(. (( [it I I I (Iii ( l l . i l.‘ ( (t ( ((I‘ ’ ’ I I ( ( (I. ( //{ ,.../l

132

 

Figure 38. Fiber guide roller.

 

133

 

Figure 39. Rotational and transverse motor
assembly.

 

 

 

/ ..{t ( ,(,/I\( Ill {it (((((((I I““‘:-‘(“(( {/1

134

 

Figure 40. Mandrel assembly.

 

 

 

 

 

‘ .( (I...

r(/

135

 

Figure 41. Side-view of
mandrel assembly

CONCLUSIONS

Over the course of this research project several aspects relating to the
development of a technique for the processing of intermetallic matrix composites were
analyzed. These several aspects included determining a method for reducing the
particle size, developing a means for determining both the actual and efl’ective particle
size, and correlating both theoretical and experimental results which describe the
colloidal behavior of the multi component composite system. In addition to this a
suitable mechanical winding system was also designed and developed for future
implementation in to the processing scheme.

From the experimental work performed it was determined that an acceptable
means for reducing the initial particle size of the as-received powder was developed.
This technique involved combining sedimentation behavior and vacuum assisted
dehydration. As was in the data presented, surface changes due to consistent
hydration of the powder surface, which may include oxidation, contributed to the
observed instability of the single component. Upon realizing which steps within the
reduction method were unnecessary, the relative instability of the system improved.
Nevertheless, since the powder was suspended in an aqueous solution during the
process, the existence of adsorbed ions was apparent as revealed by the experiments to

determine the relative stability of the system. Therefore, it became evident that the

136

137
only necessary step within the reduction process was a washing step which would

remove the adsorbed ions and produce a reduced particle size with the same
characteristics as the as-received supply.

It was also determined that the ideal particle size range for the matrix powder
is relatively small (i.e. a few um) and the size range as narrow as possible. This type
of processing necessitates a small particle size in order that the colloidal forces can
dominate over gravitational forces. This size range stipulation was made so that the
previously observed large particle/small particle flocculation which indicated instability
would be alleviated.

During the examination of the processing parameters which are important to
the colloidal processing of IMC’s it was determined that the ideal electrolyte
concentration for the powder suspension was between 0.001N and 0.0001N KNO,.
The c.c.c. experiment that was performed using the Fe-40AlSED powder generated
experimental verification that adsorbed ions existed as a result of the particles being
suspended in an aqueous solution. The second parameter examined was the effective
particle size. The purpose of this experiment was not only to provide particle size
data which could be cross-referenced with particle size expectations from the grid
analysis but also to generate an array of data which could aid in determing the
flocculation behavior which is extremely important with respect to the determination
of srn'face potential and stability. It was discovered over the duration of the particle
size experiments that the possibility for obtaining a complete examination of the
flocculation kinetics was not possible with the equipment that was utilized. Data was
produced, though, which gave an excellent description as to the flocculation behavior

for each matrix component analyzed. Although, the data provided inconclusive

138
flocculation evidence for the Fe-40AlS. sample over a wide range of pHs, it did

provide information which revealed that maximum flocculation occurs at a pH 3.0 and
pH 5.5 for the Fe-40A1325 sample. As was discovered during the ESA measurement
and stability predictions, accuracy and knowledge of particle size is crucial in
providing accurate results for the theoretical model.

During the ESA measurement for the Fe-40AlSBD, it was determined that the
maximum potential, which corresponds to minimum flocculation was achieved for a
range of pH 5.0 - 6.5. A similar but more generalized result was obtained for the Fe-
40Al325$ample which showed maximum potential for pH greater than 5.5. The data
scatter shown in the potentiometric titration plot can be primarily attributed to the
large particle size range characteristic to the -325 mesh sample. This again reinforces
the initial stipulation that not only is a small particle size desirable but also a narrow
particle size range. Upon comparison with the effective particle size data, the
behavior exhibited during the potentiometric titration was very consistent with the data
conrplied from the particle size analysis. The discrepancies which existed between the
Fe-40AlSED and the Fe-40A1325 zeta potential data were the resultant of several
factors. These included: (1) Fundamental difference in the surfaces due to the
sedimentation procedure, (2) Possible difference in the chemical make-up of the -325
mesh surface due to the minor additions of Zr and B, (3) Particle size range and
accuracy of reported values, and (4) Inability to determine the true particle size due to
flocculation kinetics.

By using the acquired zeta potential data, stability plots were generated
utilizing a computer program which attempted to predict the most desirable processing

range by maximizing the stability of like components and minimizing the stability of

139
dissimilar components, thus providing heterocoagulation of the powder to the fiber

surface. Both stability plots which were provided yielded the same conclusion for
both powder samples. The ideal range indicated from the stability data showed that
stability was minimized (in both cases) for a pH value approximately greater that pH
8.0. Once this predicted range was established, physical coating experiments were
then performed to generate actual pH processing range for which the results are most
desirable. From the fiber coating experiments it was determined that the ideal range is
located between pH 5.0 to pH 6.5. Since this result did not agree with the theoretical
predictions from before, the validity of the computer model, as applied to this system,
was in question. In order to determine whether the program itself did not apply to this
system or whether inaccurate particle sizing was the cause for the conflicting results, a
simulation was performed which utilized exaggerated zeta potential data so that any
inaccuracies related to particle size changes caused by flocculation kinetics could be
eliminated. This leads us to conclude that in order to generate more accurate zeta
potential and stability data, the current standing of the BSA measurement, which
assumes one constant particle size over the entire pH range, must be changed. This
would allow for one particle size value to be specified per a single measurement, thus
accounting for the flocculation ln'netics present.

An adequate mechanical fiber winding system was also developed to eventually
facilitate the processing of the IMCs. During the test runs which were performed,
several modifications were implemented which were not previously introduced. The
minor required modifications which were identified include: (1) the development of a
different mandrel which eliminate high stress areas around its circumference and

prevent fiber damage, (2) relocation of fiber delivery assembly to account for existing

140
helical wind on the provided fiber spool, and (3) implementation of a second power

supply to make the D.C. motors independent of each other so that accurate and
repeatable results can be obtained.

Upon completion of the experimental examination of the several aspects of
FED processing, an excellent understanding as to the behavior of the system was
gained. Several variables which were important to the validity of the testing procedure
were identified and examined. From the gathered results such parameters as optimum
pH processing range, critical coagulation concentration, and effective particle size were
determined. Information from these experiments were then used to contrast the
theoretical and experimental results. It was further shown that it is physically possible
and feasible to produce a powder coawd fiber tow without the use of expensive and
contaminating binders. By understanding and controlling the parameters inherent to
the FED processing scheme, it has become possible to produce uniformly coated

continuously reinforced composites.

APPENDIX

141

CCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCC
Rev. 06-22-92

MASTERS THESIS RESEARCH

BRETT A. WILSON

MICHIGAN STATE UNIVERSITY

COLLEGE OF ENGINEERING

DEPARTMENT OF MET ALLURGY, MECHANICS.
AND MATERIAL SCIENCE ‘

000
00000000

000000

CCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCC

CCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCC

THIS PROGRAM USES MEASURABLE MATERIAL DATA TO
CREATE INFORMATION WHICH SHOULD PREDICT THE
FLOCCULATION STATE OF TWO COMPONENT COLLOIDAL
SUSPENSIONS.

THE PROGRAM RELIES ON A METHOD WHICH IS A MOD-
IFICATION OF THE H...HF METHOD, WHICH IS BASED ON
THE DEBYE-HUCKEL APPROXIMATION FOR THE REPULSION
BETWEEN TWO PLATES OF CONSTANT POTENTIAL.

MODIFICATIONS INCLUDE:
-USING AN EFFECTIVE HAMAKER CONSTANT FOR TWO
PARTICLES IN A DISPESING MEDIUM.

-USING ZETA POTENTIAL DATA INSTEAD OF CALCULAT-
ING THE SURFACE POTNTIAL FROM POINT-OF—ZERO-
CHARGE DATA.

000000000000000000
000000000000000000

0

CCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCC

142

FILB NAME:FE40ALF\BFE40ALF

FILE DESCRIPTION:

RUN WITH Constant Potential for Fe40Al/Alumina fiber

THIS DATA WAS AQUIRED ON: 06/02/1993

AQUTSITION BY THE PROGRAM STARTED AT... 12:07 :52:90
AND FINISHED AT... 12:13:58z38

OVERALL PROPORTION OF COMPONENT 1 IN SYSTEM =.500

CONCENTRATION OF l-l ELBCTROLYTB IN SYSTEM = .00010 N

ATOMIC PARTICLE RADIUS OF COMPONENT l = 11860.0 nm

ATOMIC PARTICLE RADIUS OF COMPONENT 2 = 15000.0 nm

TEMPERATURE OF SYSTEM (IN DEG. C.) = 25.0

HAMAKER CONSTANT OF MEDIUM = .45B-l9 J

HAMAKER CONSTANT OF COMPONENT l = .22E-18 J

HAMAKER CONSTANT OF COMPONENT 2 = .11E-18 J

ZETA POTENTIAL DATA WAS USED FOR CALCULATIONS.

ZETA POTENTIAL DATA FOR PARTICLE 1:

ZET A POTENTIAL VALUES: CORESPONDING PH VALUES:
8.2 4.0
8.4 4.1

10.8 4.4
10.7 4.5
12.7 4.9
12.8 5.0
14.4 5.5
15.3 5.5
15.1 6.0
16.1 6.1
14.6 6.4
15.6 6.6
16.3 6.8
15.3 7.1
16.1 7.3
16.9 7.5
15.4 7.9
16.3 8.0
15.6 8.5
15.6 8.9
16.4 9.1
16.3 9.5
15.4 9.5
15.8 9.9

15.5 10.0

143
ZET A POTENTIAL DATA FOR PARTICLE 2:

ZET A POTENTIAL VALUES: CORESPONDING PH VALUES:

27.2 4.0
27.1 4.1
27.1 4.5
27.2 4.9
27.1 5.0
26.0 5.4
25.5 5.6
23.5 5.9
22.9 6.0
19.9 6.4
18.4 6.6
15.4 6.9
14.5 7.0
9.2 7.5
8.5 7.6
4.5 7.9
3.3 8.1
1.9 8.4
-1.8 8.5
-3.8 8.9
-4.2 9.0
-5.7 9.5
-5.8 9.6
-6.3 9.9
-6.4 10.0

CALCULATED OVERALL STABILI'I'Y RATIO DATA:

OVERALL STABILITY RATIO: CORESPONDING PH
(W 1 1,W12,W22,WT) VALUES:

4.00 , .3320E+00 , .9314+259 , .2596+304 , .7414E+00
4.50 , .2620E-15 , .2325+304 , .2596+304 , .5850E-15
5.00 . .1662E+66 , .2325+304 , .2596+304 , .3711E+66
5.50 , .3070-l-181 , .2325+304 , .2596+304 , .6854+181
6.00 , .1240+216 , .2325+304 , .2596+304 , .2770-l-216
6.50 , .5857+197 , .2325+304 , .2596+304 , .l308+l98
7.00 , .1429+234 , .2325+304 , .2596+304 , .3191+234
7.50 , .2053+304 , .2516+205 , .8491+146 , .7760+147
8.00 , 35454-271 , .1130E+00 , .6953E+00 , .2454E+00
8.50 , .6530+231 , .5728B-I-00 , .6953E-l-00 , .1075E+01
9.00 , .3804+249 , .5728E-l-00 , .6953E+00 , .1075E+01
9.50 , .2602+230 , .5728E+00 , .6162E-l-09 , .1294E+01
10.00 , .2156+226 , .5728E-l-00 , .5465E+28 , .1294E-l-01

144

FILE NAME:FB40ALG\BFE40ALG
FILE DESCRIPTION:
RUN WITH Constant Potential for Fe40A1/Alumina fiber, reduced pwdr size
THIS DATA WAS AQUIRED ON: 06/02/1993
AQUTSI'ITON BY THE PROGRAM STARTED AT... 12:16:26:57

AND FINISHED AT... 12:20:46:20
OVERALL PROPORTION OF COMPONENT 1 IN SYSTEM =.500
CONCENTRATION OF l-l ELECTROLYTE IN SYSTEM = .00010 N
ATOMIC PARTICLE RADIUS OF COMPONENT l = 6000.0 nm.
ATOMIC PARTICLE RADIUS OF COMPONENT 2 = 15000.0 nm
TEMPERATURE OF SYSTEM (IN DEG. C.) = 25.0
HAMAKER CONSTANT OF MEDIUM = .45E-19 J
HAMAKER CONSTANT OF COMPONENT l = .22E-18 J
HAMAKER CONSTANT OF COMPONENT 2 = .llB-18 J
ZETA POTENTIAL DATA WAS USED FOR CALCULATIONS.
ZETA POTENTIAL DATA FOR PARTICLE 1:

ZETA POTENTIAL VALUES: CORESPONDING PH VALUES:
8.2 4.0
8.4 4.1

10.8 4.4
10.7 4.5
12.7 4.9
12.8 5.0
14.4 5.5
15.3 5.5
15.1 6.0
16.1 6.1
14.6 6.4
15.6 6.6
16.3 6.8
15.3 7.1
16.1 7.3
16.9 7.5
15.4 7.9
16.3 8.0
15.6 8.5
15.6 8.9
16.4 9.1
16.3 9.5
15.4 9.5
15.8 9.9

15.5 10.0

145

ZETA POTENTIAL DATA FOR PARTICLE 2:
ZETA POTENTIAL VALUES:

27.2
27.1
27.1
27.2
27. 1
26.0
25.5
23.5
22.9
19.9
18.4
15.4
14.5

9.2

8.5

4.5

3.3

1.9
-1.8
-3.8
-4.2
-5.7
-5.8
-6.3
-6.4

4.0
4.1
4.5
4.9
5.0
5.4
5.6
5.9
6.0
6.4
6.6
6.9
7.0
7.5
7.6
7.9
8.1
8.4
8.5
8.9
9.0
9.5
9.6
9.9
10.0

CORESPONDING PH VALUES:

CALCULATED OVERALL STABILITY RATIO DATA:
OVERALL STABILITY RATIO:
(W 1 1,W12,W22,WT)

4.00 ,
4.50 ,
5.00 a
5.50 ,
6.00 .
6.50 .
7.00 9
7.50 ,
8.00 ,
8.50 ,
9.00 .
9.50 ,
10.00 ,

.3346E+00 ,
.9661B-07 ,
.8223E-l-34 ,
.1876E+93 ,
.6178+110 ,
.3269+101 .
.8571+l l9 ,
33824-165 .
.72S4+l38 ,
.5605+ll8 ,
.5505+127 ,
.1097+118 ,
.9420-I-115 ,

.1800+l91 ,
.1818+304 ,

.1818+304 ,
.1818+304 ,
.1818+304 ,

.1818+304 ,
.1818+304 ,

.1247+162 ,
.1547B+15 ,
.2007E+01 ,
.2007E+01 ,
.2007E+01 ,
.2007E+01 ,

.2596+304 .
.2596+304 ,
.2596+304 .
.2596+304 ,
.2596+304 .
.2596+304 ,
.2596+304 .
.8491+146 ,
.6953E+00 .
.6953E+00 ,
.6953E+OO ,
.6162E+09 ,
.5465E+28 ,

CORESPONDING PH

VALUES:

.3788E+00
.lO94E-06

.9309E+34
.2123E+93
.6994+l 10
.3701+101

.9704+l 19
23474-149

. 1922E+03
. 1625E+02
. 1625E+02
. 1775E+02
.1775E+02

146

FILE NAME:FB40ALH\BFE40ALH
FILE DESCRIPTION:
RUN WITH Constant Potential for Fe40Al/Alumina fiber, p.s.<10 microns
THIS DATA WAS AQUIRED ON: 06/08/ 1993
AQUISITION BY THE PROGRAM STARTED AT... 02:35:18:05

AND FINISHED AT... 02:46:51:26
OVERALL PROPORTION OF COMPONENT 1 IN SYSTEM =.500
CONCENTRATION OF l-l ELECTROLYTE IN SYSTEM = .00010 N
ATOMIC PARTICLE RADIUS OF COMPONENT 1 = 3000.0 nm
ATOMIC PARTICLE RADIUS OF COMPONENT 2 = 15000.0 nm
TEMPERATURE OF SYSTEM (IN DEG. C.) = 25.0
HAMAKER CONSTANT OF MEDIUM = .45B-19 J
HAMAKER CONSTANT OF COMPONENT 1 = .22E-18 J
HAMAKER CONSTANT OF COMPONENT 2 = .llE-18 J
ZETA POTENTIAL DATA WAS USED FOR CALCULATIONS.
ZETA POTENTIAL DATA FOR PARTICLE l:

ZETA POTENTIAL VALUES: CORESPONDING PH VALUES:
8.2 4.0
8.4 4.1
10.8 4.4

10.7 4.5
12.7 4.9
12.8 5.0
14.4 5.5
15.3 5.5
15.1 6.0
16.1 6.1
14.6 6.4
15.6 6.6
16.3 6.8
15.3 7.1
16.1 7.3
16.9 7.5
15.4 7.9
16.3 8.0
15.6 8.5
15.6 8.9
16.4 9.1
16.3 9.5
15.4 9.5
15.8 9.9

15.5 10.0

147

ZET A POTENTIAL DATA FOR PARTICLE 2:

ZET A POTENTIAL VALUES: CORESPONDING PH VALUES:
27.2 4.0
27.1 4.1
27.1 4.5
27.2 4.9
27.1 5.0
26.0 5.4
25.5 5.6
23.5 5.9
22.9 6.0
19.9 6.4
18.4 6.6
15.4 6.9
14.5 7.0
9.2 7.5
8.5 7.6
4.5 7.9
3.3 8.1
1.9 8.4
-1.8 8.5
-3.8 8.9
-4.2 9.0
-5.7 9.5
-5.8 9.6
-6.3 9.9
-6.4 10.0
CALCULATED OVERALL STABILITY RATIO DATA:
OVERALL STABILITY RATIO: CORESPONDING PH
(W1 1,W12,W22,WT) VALUES:
.3365E+00, .2178+128, 2596+304, .3419E-l-00 4.00
.1504E-02, .4062+211, .2596+304, .1528E-02 4.50
.4991E-i-18, .3367+293, .2596+304, .5071E+18 5.00
.8402E+47, .1558+304, .2596+304, .8537E+47 5.50
.4943E+56, .1558-l-304. .2596+304, .5022E+56 6.00
.1123E+52, .1558-i-304, .2596+304, .1141E+52 6.50
.1863E+61, .2422+232, .2596+304, .1893E+61 7.00
.1233E+84, .1884+114, .8491+146, .1253E+84 7.50
.5547E+70, .4843E+22, .6953E+00, .1104E+05 8.00
.4758E+60, .6231E+05, .6953E-i-00, .1101E+05 8.50
.1508E+65, .5734E+05, .6953E+00, .1100E+05 9.00
.2103E+60, .5540E-t-05, .6162E-l-09, .3518E+07 9.50
.1943E+59, .5472E+05, .5465E-l-28, .3475E-l-07 10.00

148

FILE NAME:FB40AL1\BFB40ALI
FlLB DESCRIPTION:
RUN WITH Const. Potn. for Fe40Al/Alumina fiber, mod. AlO zp, ps<10 microns
THIS DATA WAS AQUIRED ON: 06/08/ 1993
AQUISITION BY THE PROGRAM STARTED AT... 02:46:51:76

AND FINISHED AT... 02:54:2lz87
OVERALL PROPORTION OF COMPONENT 1 IN SYSTEM =.500
CONCENTRATION OF 1-1 ELECTROLYTE IN SYSTEM = .00010 N
ATOMIC PARTICLE RADIUS OF COMPONENT l = 3000.0 nm
ATOMIC PARTICLE RADIUS OF COMPONENT 2 = 15000.0 nm
TEMPERATURE OF SYSTEM (IN DEG. C.) = 25.0
HAMAKER CONSTANT OF MEDIUM = .45E-l9 J
HAMAKER CONSTANT OF COMPONENT 1 = .22E-18 J
HAMAKER CONSTANT OF COMPONENT 2 = .llE-l8 J
ZETA POTENTIAL DATA WAS USED FOR CALCULATIONS.
ZETA POTENTIAL DATA FOR PARTICLE l:

ZET A POTENTIAL VALUES: CORESPONDING PH VALUES:
8.2 4.0
8.4 4.1

10.8 4.4
10.7 4.5
12.7 4.9
12.8 5.0
14.4 5.5
15.3 5.5
15.1 6.0
16.1 6.1
14.6 6.4
15.6 6.6
16.3 6.8
15.3 7.1
16.1 7.3
16.9 7.5
15.4 7.9
16.3 8.0
15.6 8.5
15.6 8.9
16.4 9.1
16.3 9.5
15.4 9.5
15.8 9.9

15.5 10.0

149

ZET A POTENTIAL DATA FOR PARTICLE 2:

ZET A POTENTIAL VALUES: CORESPONDING PH VALUES:
45.3 4.0
45.2 4.5
45.0 5.0
42.9 5.5
38.2 6.0
32.1 6.5
24.2 7.0
15.3 7.5
6.8 8.0
-3.0 8.5
-7.0 9.0
-9.5 9.5
-10.0 10.0
CALCULATED OVERALL STABILITY RATIO DATA:
OVERALL STABILITY RATIO: CORESPONDING PH
(W 1 1,W12,W22,WT) VALUES:
.3365E-i-00, .3709+144, 2596+304, .3419E+00 4.00
.1504E-02. .3253+242, .2596+304. .1528E-02 4.50
.4991E+18, .1558+304, .2596+304, .5071E+18 5.00
.8402E+47, .1558+304, .2596+304, .8537E+47 5.50
.4943E+56, .1558+304, .2596+304, .5022E+56 6.00
.1123E+52, .1558+304, .2596+304, .1141E+52 6.50
.1863E+61, .1558+304, .2596+304, .1893E+61 7.00
.1233E+84, .2225+264, .2596+304, .1253E+84 7.50
.5547E+70, .1169E+66, .1688E+43, .2680E-I-47 8.00
.4758E+60, .5921E-I-05, .6953E-l-00, .1101E+05 8.50
.1508E+65, .5395E-I-05, .5402E+50, .3426E+07 9.00
.2103E+60, .5193E-I-05, .1762+178, .3298E+07 9.50
.1943E-I-59, .5158E+05, .1531+211, .3276E-i-07 10.00

150

FILE NAME:FB40ALT\BFB40ALT
FILE DESCRIPTION:
RUN WITH Constant Potential for Fe40Al/Alumina fiber; modified AlO zp
THIS DATA WAS AQUIRED ON: 06/03/ 1993
AQUISITION BY THE PROGRAM STARTED AT... 8:45:48:87

AND FINISHED AT... 8:57:35z21
OVERALL PROPORTION OF COMPONENT 1 IN SYSTEM =.500
CONCENTRATION OF l-l ELECTROLYTE IN SYSTEM = .00010 N
ATOMIC PARTICLE RADIUS OF COMPONENT l = 11860.0 nm
ATOMIC PARTICLE RADIUS OF COMPONENT 2 = 15000.0 nm
TEMPERATURE OF SYSTEM (IN DEG. C.) = 25.0
HAMAKER CONSTANT OF MEDIUM = .45B-19 J
HAMAKER CONSTANT OF COMPONENT 1 = .22B—l8 J
HAMAKER CONSTANT OF COMPONENT 2 = .llE-18 J
ZETA POTENTIAL DATA WAS USED FOR CALCULATIONS.
ZETA POTENTIAL DATA FOR PARTICLE 1:

ZETA POTENTIAL VALUES: CORESPONDING PH VALUES:
8.2 4.0
8.4 4.1
10.8 4.4

10.7 4.5
12.7 4.9
12.8 5.0
14.4 5.5
15.3 5.5
15.1 6.0
16.1 6.1
14.6 6.4
15.6 6.6
16.3 6.8
15.3 7.1
16.1 7.3
16.9 7.5
15.4 7.9
16.3 8.0
15.6 8.5
15.6 8.9
16.4 9.1
16.3 9.5
15.4 9.5
15.8 9.9

15.5 10.0

151
ZETA POTENTIAL DATA FOR PARTICLE 2:

ZETA POTENTIAL VALUES: CORESPONDING PH VALUES:
45.3 4.0
45.2 4.5
45.0 5.0
42.9 5.5
38.2 6.0
32.1 6.5
24.2 7.0
15.3 7.5

6.8 8.0
-3.0 8.5
-7.0 9.0
-9.5 9.5

-10.0 10.0

CALCULATED OVERALL STABILITY RATIO DATA:

OVERALL STABILITY RATIO: CORESPONDING PH

(W 1 1,W12,W22,WT) VALUES:
.3320E400, .2325+304, .2596+304, .7414E+00 4.00
.262OE- 15, .2325+304, .2596+304, .5850E- 15 4.50
.1662E+66, .23254-304, .2596+304, .3711E+66 5.00
.3070+181, .23254-304, .2596+304. .6854+181 5.50
.1240+216, .23254-304, .2596+304, .2770+216 6.00
.5857+197, .2325+304, .2596+304, .1308+198 6.50
.1429+234, .2325+304, .2596+304, .3191+234 7.00
.20534-304, .23254-304, .2596+304, .22194-304 7.50
.3545+271, .5409E4-93, .1688E4-43, .1543E4-44 8.00
.6530+23l, .5728E+00, .6953E4-00, .107 5E+01 8.50
.3804+249, .5728E+00, .5402E+50, .1294E4-01 9.00
.2602+230, .5728E4-00, . 17624-178, .1294E+01 9.50
.21564-226, .5728E+00, .15314-211, .1294E+01 10.00

152

FILE NAME:FB40ALI\BFB40ALI
FILE DESCRIPTION:
RUN WITH Const. Potn. for Fe40Al/Alumina fiber, mod. AlO zp, ps<10 mic
THIS DATA WAS AQUIRED ON: 06/08/1993
AQUISITION BY THE PROGRAM STARTED AT... 02:46:51:76

AND FINISHED AT... 02:54:21:87
OVERALL PROPORTION OF COMPONENT 1 IN SYSTEM =.500
CONCENTRATION OF l-l ELECTROLYTE IN SYSTEM = .00010 N
ATOMIC PARTICLE RADIUS OF COMPONENT 1 = 3000.0 um
ATOMIC PARTICLE RADIUS OF COMPONENT 2 = 15000.0 nm
TEMPERATURE OF SYSTEM (IN DEG. C.) = 25.0
HAMAKER CONSTANT OF MEDIUM = .45E-19 J
HAMAKER CONSTANT OF COMPONENT 1 = .22B-18 J
HAMAKER CONSTANT OF COMPONENT 2 = .llE-18 J
ZETA POTENTIAL DATA WAS USED FOR CALCULATIONS.
ZETA POTENTIAL DATA FOR PARTICLE l:

ZET A POTENTIAL VALUES: CORESPONDING PH VALUES:
8.2 4.0
8.4 4.1

10.8 4.4
10.7 4.5
12.7 4.9
12.8 5.0
14.4 5.5
15.3 5.5
15.1 6.0
16.1 6.1
14.6 6.4
15.6 6.6
16.3 6.8
15.3 7.1
16.1 7.3
16.9 7.5
15.4 7.9
16.3 8.0
15.6 8.5
15.6 8.9
16.4 9.1
16.3 9.5
15.4 9.5
15.8 9.9

15.5 10.0

153

ZET A POTENTIAL DATA FOR PARTICLE 2:

ZET A POTENTIAL VALUES:
45.3
45.2
45.0
42.9
38.2
32.1
24.2
15.3

6.8
-3.0
-7.0
-9.5

-10.0

CORESPONDING PH VALUES:

4.0
4.5
5.0
5.5
6.0
6.5
7.0
7.5
8.0
8.5
9.0
9.5
10.0

CALCULATED OVERALL STABILITY RATIO DATA:

OVERALL STABILITY RATIO:

(W 1 1,W12,W22,WT)
.3365E4-00, .37094-144, 2596+304,
.1504E-02, .3253+242, .2596+304.
.4991E+18, .1558+304, .2596+304,
.8402E+47, .1558+304, .2596+304,
.4943E+56, .1558+304, .2596+304.
.1123E+52, .1558+304, .2596+304,
.1863E+61, .1558+304, .2596+304,
.1233E+84, .2225+264, .2596+304,
.5547E4-70, .1169E4-66, .1688E+43,
.4758E+60, .5921E+05, .6953E+00,
.1508E+65, .5395E+05, .5402E+50,
.2103E+60, .5193E+05, .1762+178,
.1943E+59, .5158E+05, .15314-211,

CORESPONDING PH

VALUES:
.3419E+00 4.00
.1528E-02 4.50
.5071B4-18 5.00
.8537E+47 5.50
.5022E+56 6.00
.1141E+52 6.50
.1893B+61 7.00
.1253B+84 7.50
.2680E+47 8.00
.1101E+05 8.50
.3426B+07 9.00
.3298E+07 9.50
.3276B+07 10.00

154

FILE NAME:FEAL550\BFEAL550

FILE DESCRIPTION:

RUN WITH Constant Potential for Fe40Al/Alumina fiber file 550

THIS DATA WAS AQUIRED ON: 06/10/1993

AQUISITION BY THE PROGRAM STARTED AT... 19:46:32z83
AND FINISHED AT... 20:14:22z56

OVERALL PROPORTION OF COMPONENT 1 IN SYSTEM =.500

CONCENTRATION OF l-l ELECTROLYTE IN SYSTEM = .00010 N

ATOMIC PARTICLE RADIUS OF COMPONENT l = 11860.0 nm

ATOMIC PARTICLE RADIUS OF COMPONENT 2 = 15000.0 nm

TEMPERATURE OF SYSTEM (IN DEG. C.) = 25.0

HAMAKER CONSTANT OF MEDIUM = .45E-l9 J

HAMAKER CONSTANT OF COMPONENT l = .22B-18 J

HAMAKER CONSTANT OF COMPONENT 2 = .llB-18 J

ZETA POTENTIAL DATA WAS USED FOR CALCULATIONS.

ZETA POTENTIAL DATA FOR PARTICLE 1:

ZETA POTENTIAL VALUES: CORESPONDING PH VALUES:
9.5 4.0
9.7 4.0

10.1 4.5
10.0 4.6
10.2 5.0
10.4 5.0
12.2 5.4
12.7 5.6
13.4 6.0
13.3 6.0
13.0 6.4
13.4 6.5
12.7 7.0
12.6 7.1
12.2 7.4
12.1 7.6
1 1.5 7.9
1 1.7 8.1
1 1.5 8.4
1 1.2 8.5
10.4 9.0
9.5 9.5
9.4 9.6
9.4 9.9

9.4 10.1

155
ZBT A POTENTIAL DATA FOR PARTICLE 2:

ZET A POTENTIAL VALUES: CORESPONDING PH VALUES:
27.2 4.0
27.1 4.1
27.1 4.5
27.2 4.9
27.1 5.0
26.0 5.4
25.5 5.6
23.5 5.9
22.9 6.0
19.9 6.4
18.4 6.6
15.4 6.9
14.5 7.0
9.2 7.5
8.5 7.6
4.5 7.9
3.3 8.1
1.9 8.4
-l.8 8.5
-3.8 8.9
-4.2 9.0
-5.7 9.5
-5.8 9.6
-6.3 9.9
-6.4 10.0
CALCULATED OVERALL STABILITY RATIO DATA:
OVERALL STABILITY RATIO: CORESPONDING PH
(W 1 1,W12,W22,WT) VALUES:
.3320E4-00, .2325+304, .2596+304, .7414E+00 4.00
.6701E-33, .2325+304, .2596+304, .1496E-32 4.50
.3965E-26, .2325+304, .2596+304, .8853E-26 5.00
.2260E4-47, .2325+304, .2596+304, .5047E4-47 5.50
.2505E4-93, .2325+304, .2596+304, .5594E+93 6.00
.5083E+92, .2325+304, .2596+304, .1135E4-93 6.50
.1120E+61, .23254-304, .2596+304, .2501E+61 7.00
.1053E4-37, .1246+122, .8491+146, .2352E+37 7.50
.7698E4-14, .8123E—16, .6953E+00, .1835E-15 8.00
.1321E4-02, .5728E+00, .6953B4-00, .1037E+01 8.50
.2489E-24, .5728E+00, .6953E4-00, .5557E-24 9.00
.3320E4-00, .5728E4-00, .6162E+09, .4713E4-00 9.50
.3320E4-00, 572813400, .5465E4-28, .4713E+00 10.00

156

FILE NAME:FBAL32A\BFEA32A
FILE DESCRIPTION:
RUN WITH Constant Potential for Fe40Al 325/Alumina fiber Simulation
THIS DATA WAS AQUIRED ON: 06/ 16/ 1993
AQUISITION BY THE PROGRAM STARTED AT... 20:42:12:13

AND FINISHED AT... 21:09:16z28
OVERALL PROPORTION OF COMPONENT 1 IN SYSTEM =.500
CONCENTRATION OF l-l ELECTROLYTE IN SYSTEM = .00010 N
ATOMIC PARTICLE RADIUS OF COMPONENT l = 11860.0 nm
ATOMIC PARTICLE RADIUS OF COMPONENT 2 = 15000.0 nm
TEMPERATURE OF SYSTEM (IN DEG. C.) = 25.0
HAMAKER CONSTANT OF MEDIUM = .45E-19 J
HAMAKER CONSTANT OF COMPONENT l = .22E-l8 J
HAMAKER CONSTANT OF COMPONENT 2 = .llB-l8 J
ZETA POTENTIAL DATA WAS USED FOR CALCULATIONS.
ZETA POTENTIAL DATA FOR PARTICLE l:

ZETA POTENTIAL VALUES: CORESPONDING PH VALUES:
-2.0 4.0
3.0 4.5
10.0 5.0
15.0 5.5
15.1 6.0
15.2 6.5
15.3 7.0
15.4 7.5
15.5 8.0
15.4 8.5
15.3 9.0
15.2 9.5

15.1 10.0

157
ZETA POTENTIAL DATA FOR PARTICLE 2:

ZETA POTENTIAL VALUES: CORESPONDING PH VALUES:
27.2 4.0
27.1 4.1
27.1 4.5
27.2 4.9
27.1 5.0
26.0 5.4
25.5 5.6
23.5 5.9
22.9 6.0
19.9 6.4
18.4 6.6
15.4 6.9
14.5 7.0
9.2 7.5
8.5 7.6
4.5 7.9
3.3 8.1
1.9 8.4
-l.8 8.5
-3.8 8.9
-4.2 9.0
-5.7 9.5
-5.8 9.6
-6.3 9.9
-6.4 10.0
CALCULATED OVERALL STABILI'I'Y RATIO DATA:
OVERALL STABILITY RATIO: CORESPONDING PH
(W 11,W12,W22,W'I') VALUES:
.3320E+00, .5728E400, .2596+304, .4713E+00 4.00
.3320E+00, .7579E-05, .2596+304, .1712E-04 4.50
.1740E-34, .2325+304, .2596+304, .3885E-34 5.00
.1570+l91, .2325+304, .2596+304, .3505+l9l 5.50
.5857+197, .2325+304, .2596+304, .1308+l98 6.00
.2622+204, .2325+304, .2596+304. .5855+204 6.50
.l409+211, .2325+304, .2596+304, 3147+le 7.00
.9086+217, .l736+l82. .8491+146, .7760+147 7.50
.7032+224, .1 l77E-02, .6953E+00, .2658E-02 8.00
.9086+217, .5728E+00, .6953E+00, .lO75E+Ol 8.50
.l409+211, .5728E+00, .6953E400, .1075E+01 9.00
.2622+204, .5728E+00, .6162E+09, .1294E+01 9.50
.5857+197, .5728E4-00, .5465E+28, .1294E+01 10.00

10.

11.

12.

13.

REFERENCES

Technical Proposal 1990-1991.
R.E.F. Technical Proposal.
N.S.F. Technical Proposal.

D.E. Alman and N .S. Stoloff, "Processing and Properties of Intermetallic
Matrix Composites," Mat Res. Soc. Symp. Proc., 213, 989-1000, 1991.

RM. German and A. Bose, "Fabrication of Intermetallic Matrix Composites,"
Materials Science and Engineering, A107, 107-116, 1989.

R. Bowman and R. Noebe, "Up and Coming IMCs," Advanced Materials and
Processes, 8, 35-39, 1989.

N .S. Stoloff and DE. Alman, "Innovative Processing Techniques for Intermetallic
Matrix Composites," Mat. Res. Soc. Symp. Proc., 194, 31-43, 1990.

K8. Kumar, M.S. Dipietro, S.A. Brown, JD. Wittenberger, "Composites of Low-
Density Trialuminides: Particulate and Long Fiber Reinforcements," NASA
Technical Memorandum, 1992.

S. Mourbakhsh, F.L. Liang, H. Margolin, "Pressure Casting of NLAl/ALO,
Composites", Mat. Res. Soc. Symp. Proc., 133, 459-464, 1989.

RM. German, A. Bose, N.S. Stoloff, "Powder Processing of High Temperature
Aluminides," MRS Proceedings, 133, 403-414, 1989.

DE. Alman and N.S. Stoloff, "Powder Fabrication of Monolithic and Composite
NiAl," The International Journal of Powder Metallurgy, 27, 29-41, 1991.

DJ. Shaw, Introduction _t_r_) Colloid and Surface Chemim, Butterworths, Boston,
4‘“ Edition, 1992.

RJ. Hunter, Foundations of Colloid Science, Volume 1, Clarendon Press, Oxford,
1987.

158

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

159

RC. Heimenz, Principles of Colloid and Surface Chemisgy, Marcel Dekker, New
York, 2“"l Edition, 1986.25. K. Aoki and O. Izumi, J. Jpn. Inst. Met, 43, 1190,
1979.

B.A. Wilson, Master of Science Thesis, Michigan State Unversity, 1992.

M.J. Crimp, Master of Science Thesis, Case Western Reserve University, 1985.

R.W. O’Brien, "Electro-acoustic effects in a Dilute Suspension of Spherical
Particles," J. Fluid Mech., 190, 71-86, 1988.

R.W. O’Brien, "The Electroacoustic Equations for a Colloidal Suspension," J.
Fluid Mech., 212, 81-93, 1990.

Tony Oja, Matec Applied Sciences Division, Personal Communications.

R. Hogg. T.W. Healy, D.W. Fuerstenau, "Mutual Coagulation of Colloidal
Dispersions," Trans. Faraday Soc., 62, 1638-1651, 1966.

Deryaguin, B.V. and Landau, L., Acta Phys. Chirn. URSS, 14, 633 (1941).

D.M. Shah, D.L. Anton, C.W. Musson,"Feasibility Study of Intermetallic
Composites," MRS Symp. Proc., 194, 333-340, 1990.

S.L. Draper, D.J. Gaydosh, M.V. Nathal, A.K. Misra, "Compatibility of Fe-40Al
with Various Fibers," J. Mat. Res., 5(9), 1976-1984, 1990.

D.M. Dimiduk and DB. Miracle, "Directions in High Temperature Intermetallics
research," MRS Symp. Proc., 133, 349-359, 1989.

K. Aoki and o. Izumi, Nippon Kinzoku Gakkaishi, 43, 1190, (1979).

J .A. Moser, M. Aindow, W.A.T. Clark, S. Draper, H.L. Fraser,"Compatibility of
Potential Renforcing Ceramics with Ni and Fe Aluminides," MRS Proc., 194,
379-384, 1990.

AB. Strong, Fundamentals of Comppsite Manufacturing: Materials, Methods,
and Applications, 4th Edition, Soc. of Manufacturing Engineers, Dearbom, 1989.

F .V. Lenel, Powder Metallurgy: Principles and Applications, Metal Powder
Inductries Federation, Princeton, New Jersey, 1980.

B.A. Wilson, C.J. Suydam, M.J. Crimp, M.A. Crimp, "Processing of FeAl / A1203
IMCs Using Advanced Electrodeposition Methods," Mat. Res. Soc. Symp. Proc.,
288, 891-896, 1992.

31.
32.

33.

34.

35.

36.

37.

38.

39.

41.

42.

43.

160

B.A. Wilson and M.J. Crimp, "Modelling for Composite Colloidal Suspension
Stability Based Upon the HHF Interpretation," Langmuir, submitted for
publication, 1993.

M.J. Gimp, Personal Communications.

D. Maglaya, Senior Thesis, Michigan State University, 1991.

Verwey, E.J.W. and Overbeek, J. Th. G., Theog of the Stability of Lyophobic
Colloids Elsevier (1948).

 

D.L. Anton and D.M. Shah,"High Temperature Ordered Compounds for Advanced
Aero-Propulsion Applications," MRS Symp. Proc., 133, 361-371, 1989.

RD. Noebe, R.R. Bowman, J .J . Eldridge, "Initial Evaluation of Continuous Fiber
Reinforced NiAl Composites," MRS Symp. Proc., 194, 323-331, 1990.

D.C. Ganmer, "Fiber Coating and Characterization," Ceramic Bulletin, 68[2],
415-419, 1989.

J. Doychak, "Metal- and Intermetallic-Matrix Composites for Aerospace
Propulsion and Power Systems," J.O.M., 44[6], 46-51, 1992.

B. Schmidt, P. Nagpal, 1. Baker, "Annealing Studies of B2 FeAl," MRS Symp.
Proc., 133, 755-760. 1989.

E. Barouch, B. Matijevic, T.A. Ring, J .M. Finlan, "Heterocoagulation II.
Interaction energy of Two Unequal Spheres, J. of Colloid and Interface Science,
67(1). 197 8.

P.J. Scales and Eileen Jones, "Effect of Particle Size Distribution on the Accuracy
of Electroacoustic Mobilities," Langmuir, 8, 385-389, 1992.

M. James, R.J. Hunter, R.W. O’Brien, "Effect of Particle Size Distribution and
Aggregation on Electroacoustic Measurements of Zeta-Potential," Langmuir, 8,
420—423, 1992.

M.J. Gimp and RF. Preston, "Evaluation of Sedimentation as a Technique for
the Separation of Sub-rnicron Diameter Silicon Carbide Whiskers,"

M.J. Gimp and R.C. Piller, "Dispersion of SiC Whiskers in a Si:,N4 Matrix Using
pH Control,"

M.J. Gimp, RB. Johnson, Jr., J.W. Halloran, D.L. Feke, "Colloidal Behavior of
Silicon Carbide and Silicon Nitride,"

45.

47.

48.

49.

50.

51.

52.

53.

55.

56.

57.

161

J.T.G. Overbeek, "Recent Developments in the Understanding of Colloid
Stability"; pp. 431-45 in Colloid and Interface Science Volume 1 Edited by M.
Kerker, K.L. Rowell, A.C. Zettlmoyer, Academic Press, 1976.

 

R.O. James, G.A. Parks, "Characterization of Aqueous Colloids by their electricd
Double-Layer and Intrinsic Surface Chemical Properties"; pp. 119-216 in Surface
and Colloid Science Volume 12 Edited by E. Matijevic, Plenum Press, New
York, 1982.

 

 

E. Barouch, E. Matijevic, T.A. Ring, J.M. Finlan, "Heterocoagulation II.
Interaction Energy of Two Unequal Spheres," J. Col. Int. Sci., 67[l], 1-9, 1978.

A. Bierman, "Electrostatic Forces Between N onidentical Colloidal Particles,"
J. Col. Int. Sci., 10, 231-245, 1955.

A. Bleier and B. Matijevic, "Heterocoagulation I. Interactions of Monodispersed
Chromium Hydroxide with Polyvinyl Chloride Latex," J. Col. Int. Sci., 55[3],
510-524, 197 6.

A. Bleier and E. Matijevic, "Heterocoagulation III. Interactions of Polyvinyl
Chloride Latex with Ludox HS Silica," J. Chem. Soc., Faraday Trans. 1, 74[6],
1346-1359, 1978.

J .T.G. Overbeek, "Double-layer Interaction between Spheres with Unequal
Surface Potentials," J. Chem. Soc., Faraday Trans., 84[9], 3079-3091, 1988.

E. Barouch, "Double-layer Interaction between Spheres with Unequal Smface
Potential," J. Chem. Soc., Faraday Trans. 1, 84[9], 3093-3095, 1988.

RE. Johson, Jr., "A Thermodynamic Description of the Double Layer
Sm'rounding Hydrous Oxides," J. Col. Int. Sci., 100[2], 540-554, 1984.

R.O. James, "Surface Ionization and Compexation at the Colloid/Aqueous
Electrolyte Interface"; pp. 219-61 in Adsopption of Inorganics at Solid-Liquid
Interfaces, Edited by M.A. Anderson, A.J. Rubin, Ann Arbor Science, Ann Arbor,
1981.

T.W. Healy and L.R. White, "Ionizable Surface Group Models of Aqueous
Interfaces," Advances in Colloid and Interface Science, 9, 303-345, 1978.

G. Kar, S.Chander, T.S. Mike, "The Potential Energy of Interaction Between
Dissimilar Electrical Double Layers," J. Col. Int. Sci., 44[2], 347-355, 1973.

E. Barouch and E. Matijevic, "Double-layer Interactions of Unequal Spheres Part
I. The Effect of Electrostatic Attraction with Particles of Like Sign of Potential,"
J. Chem. Soc., Faraday trans. l, 81, 1797-1817, 1985.

58.

59.

60.

61.

62.

63.

162

E. Barouch and E. Matijevic, "Double Layer Interactions of Unlike Spheres III.
Nonlinear and Two-Dimensional Effects," J. Col. Int. Sci., 105[2], 552-559, 1985.

B. Barouch, E. Matijevic, T.H. Wright, "Double-layer Interactions of Unlike
Spheres Part 11. Numerical Analysis of Electrostatic Interaction Energy," Chem.
Soc., Faraday Trans. 1, 81, 1819-1832, 1985.

A. Bleier, "Fundamentals of Preparing Suspensions of Silicon and Related
Ceramic Powders," J. Am. Cer. Soc., 66, 7981, 1983.

J. Gregory, "The Calculation of Hamaker Constants," Advances in Colloid and
Interface Science, 2, 3W4”, 1968.

K. Vendula, R.J. Lauf, C.A. Wells, ECUT Quarterly Report, 1985, July 1 - Sept.
30, Oak Ridge National Lab, p.56.

D.R. Gaskell, Introduction to Metallurgical Thermodmamics, 2ND Bd.,
Hemishpere Publishing Corp., New York, New York, 1981.

64. CRC Handbook of Chemist_ry and Physics, 73RD Ed. CRC Press, 1993.

65

. B. Glime, Undergraduate Senior Thesis, Michigan State University, 1991.

"IWIIIIIIIIIIIIIII

 

I
J