(
flag?” . -
E?» a. .
Wat... 5 . in»
.. . $9 Thaw..."
. . mfltnwfin . 5th
.. .kui 2:». 5.10 .
"whimhnnmuwb .45..
.: 32:4???

2. v)- .591

 

O

 

 

A o y 7.
)1 .1: rim. 3.!
. 1 1 ~ Q i ..
f ”1...... 2.)! at .
.1, Ann: 1 u!

.
zyttzcs.&flum
v0 .50! .1.
arm-”.2. Purity. t .3
. .

.d’uw‘ . I.
:1 .z . _ (at
.1!- aa 3.. ). Q! . '31:!
. 2.3.). ”7.. a uh : .1. n , 1. 19.153369le 5
nflfitxfi )soi 31-2....- 232!I\ l‘l‘n ilh’&))3u . I. n
I. ) Va z.- fifnhfnv Inrrwflufind. {1.1). 5.63. 12!... q? I «2 lung. 13
£23.. .4 v .95 ”101:..1 ...L¥o\v aunt?! Brian}... :1?!"
(WW)? 5.0.5:. .. h... 3
t..hul.fitu. hmlywu‘
(at. tit \Plh .
.3 t ,
i=3...
i .3...
21%. .

)

 

 

4.}!

 

 

1.... ,
m. .wflfil. LIMQJI
h p‘ n,“ ‘3'... a]
fish.
93...... 1.
5.3.5:.
5.... .l 5.... s
12 .
.3...“
.1 I

 

u. :2:
505....

. .

25....“ M
534%..“ Jvmnfifl.‘
. S? t.)

5:. 9.3.1:...) 1C.)

.5 :«l I In. x
g I .5 1:31:51.
fun-#711:

lat:

.. .

t .. V 5...... ii.

.. Eirlta , .. Kiuuitliknfiup

. u n. ,5! 3.3m}, 51

a...».....

at ...2. 3.5.3.1 -

£92.... .6 .n. 9.5.42.3;
,....n»

:1: .

ii

 

3.533:

1.1 35 1
"Huh!” 1 ..
Elba:

w :1 it £53.... 3‘."

P:‘.-; 1 1.0 . JvalLW-flhfihsnflhah.htfiiflu ) 1

anoti. 3. (.23.... \ 5151.33... .34?

$55!): A . nav. t: 31 1?} .

. g 3.1

 

.J\‘!«Au‘. I‘lWV
5 ii...)
)3. 3

 

 

 

 

’7
2:33

. but. awn.
2 .

1
2

' m
t“!

.2 A.
.. 51‘ x»
1;
5:222..- 3.:
.ufiaieou: )1:
. {Runs .12, . .

. .

Jpn?!

 

 

i Ln! {ti
. . {xvii}:
, , 32-1
v ; .3
x unfit.
E1... .

 

 

fl.

 

‘.
A

 

6t l»«‘bs5\ .
y}; 012.. I 11.1.... ...).\3\!
. 3)....6343. 2:52: ..I..
. i .....| a .1. .r. 3;... s: -
.srlsx:v.. £1 . (if!)
Zimnrfl .

7 .
vl

 

 
 

 

 

IIlllllllllnhmwill

1044 5033

This is to certify that the

thesis entitled
Design of an Experimental Chamber for
a New Powder Metallurgy Technique to

Make Continuous Fiber Metal-Matrix Composites
presented by

Adel Saoudi

has been accepted towards fulfillment
of the requirements for

Master's degree in Materials Science

 

Major professor

0-7639 MS U is an Affirmative Action/Equal Opportunity Institution

 

a

LIBRARY
Mlchlgan State
Unlverslty

 

 

 

PLACE II RETURN BOXtoromavothbchockomfmmywrocom.
TO AVOID FINES Mum on or Mon duo duo.

DATE DUE DATE DUE DATE DUE

        
    

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

l

MSUI MAM div. 0 » Infiltwon
a an Action/Ema! womanly

 

m1

 

DESIGN OF AN EXPERIMENTAL CHAMBER FOR

A NEW POWDER METALLURGY TECHNIQUE TO

MAKE CONTINUOUS FIBER METAL-MATRIX
COMPOSITES

By
Adel Saoudi

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

v MASTER OF SCIENCE

Department of Materials Science and Mechanics

1993

ABSTRACT
DESIGN OF AN EXPERIMENTAL CHAMBER FOR
A NEW POWDER METALLURGY TECHNIQUE TO
MAKE CONTINUOUS FIBER METAL-MATRIX
COMPOSITES
By

Adel Saoudi

The existing fabrication methods Of continuous fiber metal-matrix
composites face many problems. A new powder technique was developed by
the Composite Materials and Structures Center at Michigan State University
to make continuous fiber reinforced polymer composites. This thesis
research describes the design and the fabrication of an experimental chamber
to test this process with aluminum powders. A thermodynamics approach
was used to insure the safe handling of fine aluminum powders. The
experimental chamber was used to fiuidize aluminum powders and make
aluminum prepreg tapes. Nylon coated carbon fibers were heated inside the
chamber and coated with fluidized aluminum powders. Scanning Electron
Microscope (SEM) was used to characterize these prepregs. The
experimental results Showed a Significant powder pick up in some regions of
the tape. This proved that the process can be used to make aluminum prepreg

tapes as a first Step to make metal-matrix composites.

DEDICATION

To my parents.

iii

ACKNOWLEDGMENTS

I would like to acknowledge the Research Excellence Funds from the
State of Michigan for funding this project through the Composite Center. I
am very thankful for the materials and advice that came from Dr. Lawrence
Drzal and his staff. I, also, would like to thank everybody who helped with
the experimental work especially Wang and Sanjay. I would like to thank
Margot for helping with the typing. Finally, my biggest thanks goes to my
advisor, Dr. Thomas Bieler, for his continuous support and valuable advice

through this entire research period.

iv

TABLE OF CONTENTS

List of Tables
List of Figures
Chapter 1 1. Introduction
1.1 Historic Development

Chapter 2 2. Fabrication Methods of Continuous Fiber
Metal-Matrix Composites

2.1 Liquid-Phase Processing
2.1 .1 Pressure-Infiltration Method
2.1.2 Wetting
2.1.3 Carbon/Aluminum System

2.2 Solid-Phase Processing
2.2.1 Preforming Methods
2.2.2 Consolidation By Diffusion
Bonding

Chapter 3 3. New Process

3.1 Description of The Original Process
3.1.1 Operation of The Process

3.1.2 Spreader
3.1.3 Aerosolizer

viii

ix

11
12

22
22

27
38
38
38

41
41

3.2. Advantages of The New Process
Chapter 4 4. Safe Handling of Aluminum Powders

4.1 Risk Factor: Explosions
4.2 Sources of Ignition
4.3 Basic safety Precautions

Chapter 5 5. Experimental Chamber

5.1 Theory of The Design
5.1 .l Thermodynamics Approach
5.1.2 Materials and Dimensions

5.2 Description of The Chamber
5.2.1 Outside Chamber
5.2.2 Inside Chamber
5.2.3 Speaker
5.2.4 Heating System
5.2.5 Vacuum System
5.2.6 Grounding System

5.3 Operation of The Experimental Chamber
Chapter 6 6. Procedures
6.1 Materials
6.2 Calibration of The Heater
6.3 Making of Aluminum Prepreg
6.4 Scanning Electron Microscope (SEM)
Examination

Chapter 7 7. Results and Discussions

7.1 Experimental Results

vi

44

47

47
50
51

54

54
54
62

65
65
67
7O
7O
7 l
74

74
78
78
78
81
83
84

84

7.1.1 Vacuum Level

7.1.2 Fluidization of Aluminum Powders

7.1.3 Characterization of The Aluminum
Prepreg Tape

7.2 Discussions
7.2.1 Safety Factor
7.2.2 Heating of The Prepreg
7.2.3 Fluidization of Aluminum Powders
7.2.4 Powder Pick up
Chapter 8 8. Conclusions

Bibliography

vii

84
86

88
103
103
104
107
111
116

120

LIST OF TABLES

Table 1.1 Mechanical properties of some metal-matrix composites.
Table 2.1 Surface tension values for some metals at the melting point.

Table 5 .1 The number of argon moles to the number of oxygen moles ratio
for various vacuum levels inside a chamber.

Table 5 .2 Summary of the results from the thermodynamics approach.
Table 6.1 Properties of materials used during the experiment.

Table 6.2 Results from calibration of the heater.

Table 7.1 Powder size distribution from each sample.

Table 7.2 Average particle Size for each sample.

Table 7.3 Summary of the quantitative analysis.

viii

LIST OF FIGURES

Figure 1.1 Comparison of high temperature materials in terms of operating
temperatures and specific strength.

Figure 1.2 Specific strength and stiffness of aluminum and magnesium
matrix composites as compared to the unreinforced alloys.

Figure 2.1 Casting fabrication methods for metal-matrix composites.
Figure 2.2 Condition for good and poor wetting. -

Figure 2.3 Reduction of ultimate tensile strength for temperatures above
450 C because of crack formation.

Figure 2.4 Arc-Spray process for the fabrication of composite monotape.
Figure 2.5 The sodium process for the infiltration of carbon fiber tow.
Figure 2.6 Variation of flexural strength with hot press'pressure.

Figure 2.7 Variation of flexural strength with hot press temperature.
Figure 2.8 Vacuum hot press.

Figure 3.1 Powder prepregging process.

Figure 3.2 Spreader.

Figure 3.3 Spreading operation.

ix

Figure 3.4 Aerosolizer.

Figure 4.1 Increase of pressure with time for aluminum powders during
oxidation reaction.

Figure 5.1 Addition of the experimental chamber to the powder prepregging
system.

Figure 5 .2 The experimental chamber.

Figure 5.3 Outside chamber.

Figure 5.4 Inside chamber.

Figure 5.5 Heater inside chamber.

Figure 5.6 Vacuum system.

Figure 5.7 Operation of the chamber diagram.

Figure 5.8 Picture of the actual experimental chamber.

Figure 7.1 Various types of powders.

Figure 7.2 Location of samples on the aluminum prepreg tape.
Figure 7.3(a) SEM image of one particle on a carbon fiber.

Figure 7.3(b) X-ray dispersive energy analysis identifying the powder
in figure (a) as aluminum.

Figure 7.4(a) SEM image of powder pick up in sample A.
Figure 7.4(b) SEM image of powder pick up in sample B.

Figure 7.4(c) SEM image Showing powder pick up in sample C.

Figure 7.5(a) SEM image from sample A at high magnification.
Figure 7.5(b) SEM image from sample B at high magnification.
Figure 7.5(c) SEM image from sample C at high magnification.

Figure 7.6 SEM image showing the appearance of the heated nylon
coated fibers.

Figure 7.7 SEM image showing powder loss.

xi

Chapter 1

1. Introduction

In an age of constant strive for high temperature and high performance
materials, metal-matrix composite (MMC) stands as one of the most attrac-
tive candidates. Some of the advantages of metal-based composites[l-8]
may include high specific strength, high specific modulus, high service tem-
perature, good wear and seizure resistance, low thermal expansion, low fric-
tion coefficient, resistance to moisture, good thermal and electrical
conductivity, and dimensional stability. Figure 1.1 and figure 1.2 compare
MMC to other materials in terms of specific strength, Specific modulus and
high temperature capabilities. Table 1.1 shows the mechanical properties of

some of metal-matrix composite materials.

1.1 Historical Development

Metal-matrix composites are a class of materials made of two or more

phases of different properties, the combination of which results in a new

OPERATING
TEMPEfiéTUHE.

4000

3000

2000

1000

 
   
   
 

CARBON/CARBON
COMPOSITES

        
    
 
     

 
  
  
 
 

 

/ \
I_ l \
I .
METAL - '
Mm" I coctfiggg'rces I
comosrrts - V
INTERMETALUCS
AND
INTERMETALLIC

COMPOSITES

 
   

CONVENTIONAL
/ MATERIALS
(TITANIUM & SUPERALLOYS)

¢/////////

STRENGTH/WEIGHT RATIO

 

1.0x 106

Figure 1.1 Comparison of high temperature materials in terms

of operating temperatures and Specific strength [3].

SPECIFIC PROPERTIES of Al
I— and Mg MATRIX COMPOSITES __

 

 

 

 

SPECIFIC STIFFNESS- .84! (in)

 

 

 

ZN-

. , I
QM - P21004113

. t (.4 we)

P-tOt'yN
5. d
m (.45 We)
‘ ‘ s P-SSMg (.4 we)
4.00»; - '
. ‘ Noe/AI

Scene 4 Sue ('3 m) P-SSW s'ci”.

4 (M09010) (.45 We) . "

BIN‘

2M . FPIAI‘

. Disc. SIC/Al .
1,000.3 - I I c: SIC/N (Ntcalon)

Al Mg Shel, Tl (.35 We) °.S vol Traction
o.oo«o A ,
. , ,
0-000+0 Locus 2.00.»: ' 3 no»; I

SPECIFICSTRENGTH- a/p (in)

Figure 1.2 Specific strenght and stiffness of aluminum and magnesium
matrix composites as compared to. the unreinforced alloys [18].

I
I

 

 

 

 

 

 

 

 

Temsile Modulus Strain to Transverse
Composites Strength (GPa) failure Strength
(MPa) % (MPa)
=W
Alumina .
FP/ 1240 207 0.70 138
Al
SiC/AL 862 103 1.0 -
Textron SiC/
A1 1586 207 ‘ 0.90 103
Gr/Al 620-689 207-345 050-1 .0 34-48
Gr/Mg 620 345 _ 1.0. 34
SiC/Mg 1380 214 1.0 . -

 

 

 

 

 

Table 1.1 Mechanical properties of some metal-matrix composites.

material of superior properties to either constituents alone. The idea behind

this class of materials is to meet specific applications where one single

material Or alloy is not sufficient.

The development of fiber reinforced-metal matrix composites dates back
to the late 19505 when NASA first reported its results of tungsten wires rein-
forced-copper alloy matrix composite [9]. During the 19603 the progress in
MMC was carried out mostly by the aerospace industry because performance
was more important than the cost of the material. For a period of twenty
years, only few commercial applications involved MMC. Boron monofila-
ments of 140 microns in diameter were made to reinforce aluminum alloys
and were used in the space shuttle as a tubing for cargo bay stiffeners[ 10].
Borsic fibers (Sic-coated boron fiberS)-reinforced aluminum alloys found
applications as structural components in Spacecrafts and airplanes[ l 1]. Sili-
con carbide and alumina whiskers were used as reinforcements for alumi-
num, magnesium, titanium, etc., but the high cost of the whiskers limited
their use to research only. Carbon fibers were also considered as potential
reinforcement for aluminum alloys due to their much lower cost as compared
to the boron fibers. Early work, however, showed that the wetting problem of
carbon by molten aluminum and the excessive reaction between the fiber and

the matrix are a major problem for the development of this material.

It was not until the 1980s that the future of MMC got a remarkable

commercial lift when Toyota Motor Corp. introduced the new alumina fiber-
alurninum matrix piston for automotive diesel engines[12]. This new piston
had better wear resistance and thermal stability than the previously used
materials, and was offered at a competitive price. This indicated that the
direction of MMC in the 19803 was toward cheaper reinforcements and Sim-
ple processing methods to find commercial applications in areas other than
the aerospace and the defence industries. Short fiber/particulate- reinforced
metal-matrix composites manufactured by casting techniques have emerged
as the leading materials. These composites found applications in many areas
such as the automotive, electronic and sports fields. The 19808 also regener-
ated the interest in continuous graphite fiber-aluminum matrix composites as
engineers in Lochheed Missiles and Space Corporation have developed an
aluminum-based composite with almost zero thermal expansion and very
high stiffness. The known commercial application of this composite is the
big antenna booms used by NASA for its space telescope [l 3] where thermal
stability and high stiffness are major concerns. Despite of the unique strength
and stiffness properties in the fiber direction, other applications of the conti-
nous fiber-reinforced metal-matrix composites (CFMMC) are hard to justify
becauce of the high cost of fabrication. Cheaper processing methods and

more reliable techniques are key elements for the future of CFMMC.

Currently, a big portion of the on-going research on MMC is toward

minimizing the cost of fabrication and optimizing the processing parameters
to control the mechanical properties and the interfacial reactions. This thesis
follows this trend by building an experimental chamber to test a new powder

metallurgy process to make continuous carbon fiber-aluminum powder com-

posites .

Chapter 2

2. Fabrication Methods of Continuous Fiber Metal-Matrix Composites

The fabrication methods of continuous fiber metal-matrix composites
(CFMMC) can be classified into two groups : liquid-phase infiltration tech-

niques and solid-phase processing methods.
2.1 Liquid-Phase Processing

There are many techniques that use molten metal to make MMC. Figure
2.1 lists most of these methods with all the steps involved in the fabrication
of the composite [14]. Pressure-infiltration is the one method that is used
quite frequently to make CFMMC. Therefore, only this technique will be

reviewed.
2.1.1 Pressure-Infiltration Method

This technique is the most widely used at the present time because of its
simplicity and the pre-existing technology base from the metal casting indus-

try. It consists of pouring liquid metal into a closed mold that contains

 

 

9

 

 

LIQUID METAL .

 

 

 

 

V

 

 

 

V V

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Mixing with particle tion OI Pm 0 or I Infiltration oi Iiher prelorm
or discontinuous “b" "10"" and solidification under
fiber (Theo-slurry) high pressure

High pressure
Infiltration costing
(HIP!

Remetted above Veeuum lntlltretloij—w ‘

V Inert gas '

over molten metal
C3“ into “I" °' 3" Castlnto shapes: by Squeeze Infiltration
(“MW““IWI soildiiieetion under . costing
pressure (Pressure costln

(y (Squeeze casting)
r Billet Reheeted I ->[—lnvestmeht cesungJ—1

II —>] Hot moidng |.—-V

 

 

 

 

 

(233sz r WWW I

 

 

 

 

 

 

 

Near not shapes

 

 

 

Figure 2.1 Casting fabrication methods for metal-matrix composites [14].

10

bundle of fibers arranged in certain Shapes. Pressure is then applied to the
system for complete infiltration and rapid solidification. This process allows
the fabrication of near net-shape products (where secondary processing meth-
ods are not necessary) at relatively fast rate of production. The use of very
high pressures (above 100 MPa) during solidification results in a fine grain

microstructure of the matrix and an increase in the interfacial bond strength.

Problems With The Process: This process faces some serious and unre-
solved problems[15-l8] that stem from the nature of the casting technique

itself. Some of these problems are listed below:

1) The presence of voids after solidification due to poor infiltration, shrinkage
of the matrix and entrapped gases evolved from the fiber preform. These

pores can be eliminated when very high pressure is applied.

2) The use of very high pressures degrades the strength of the composite
because of fiber breakage (from fiber contact with other fibers) and interfacial
reactions between the liquid metal and the fibers. Most fiber/matrix systems
have a critical pressure value above which a decrease in the fiber Strength
occurs [14]. This value is 45 MPa for SiC/Al composite. This value however

does not assure the elimination of porosity.

11

3) Uneven distribution of the fibers as a result of the formation of metal chan-
nels during solidification which push the fibers to concentrate in certain
regions rather than being uniformly distributed throught the matrix. This
problem is “an important and unresolved issue in MMC solidification pro-

cessing” [18].

4) Segregation of the solute atoms at the interface during solidification.
These atoms may react with the fibers and add to the complexity of the inter-
facial reaction products. Moreover, this effect results in uneven matrix con-
centration with a solute-rich microsructure near the interface and purer metal
in the rest of the marix. This reduces the strength of the matrix as well as the

composite.

5) Only small parts can be processed with this method because of the high

pressures involved.
2.1.2 Wetting:

Liquid metals have low viscosity compared to polymer matrices (five to
seven orders of magnitude less). This makes the infiltration technique appear
to be suitable for making MMC without the use of high pressures. However,

molten metals do not wet most commercial reinforcements and engage in

12

interfacial reactions that are detrimental to the Strength of the composite [19-
20]. The ability of a liquid metal to penetrate a fibrous body and cover all the
surfacearea of the fiber is dependent on the contact angle O as Shown in fig-
ure 2.2 and is given by Young’s equation:
Yrv 0089 = Tsv - 731 (2.1')
where 'st = surface energy of the solid

73, = solid/liquid interfacial energy

71,, = surface tension of the liquid
If the contact angle O is less than 90°, then liquid metal wets the solid fiber
spontaneously. If it is higher than 90°, which is the case for most liquid met-
als and inorganic reinforcements, then pressure has to be applied to the sys-
tem to spread the molten metal over the entire fiber surface. The aluminum-
carbon system causes most problems during infiltration and is used ito illus-

trate the difficulties facing liquid-phase processing of MMC.
2.1.3. Carbon/Aluminum system

Liquid Aluminum does not wet carbon fibers at its melting point tempera-
ture (660°C). The measured surface tension 81v value for aluminum is
1050mJ/m2 [19]. Table 2.1 gives 7", values for other metals at their melting
point. The contact angle 9 between most metals and graphite fibers is greater

than 90° and may exceed 150° in some cases. Therefore, spontaneous

13

 

 

 

9<90° _ 8>90°

GOOD WETTING POOR WET-TING

Figure 2.2 Condition for good and poor wetting.

 

 

 

 

 

 

 

 

 

Metal Surface Tension
(mJ/mZ)
=7‘T—7=='———=l=‘—‘:7T=T—'
Mg 560 '
Zn 780
A1 1050
Cu 1300
Ti 1650
Ni 1780
Fe 1880
Mo 2250

 

 

 

 

Table 2.1 Surface tension values for some metal at the melting point [19]

15

wetting is completely out of reach at the melting point. The wetting problem
in aluminum/carbon composites is caused by the aluminum oxide layer that
forms on the aluminum surface. This oxide layer prevents direct contact
between carbon and molten aluminum. Therefore, wetting will not occur
until the aluminum oxide layer is broken. At temperature higher than 950° C,
molten aluminum can penetrate the oxide layer and wet the carbon fibers by
engaging in chemical reactions. The main element of the reaction product
layer at the interface is the aluminum cabide phase (A14C3). The formation
of these brittle interrnetallic components is detrimental to the mechanical
properties of the composite. Therefore, the interfacial reaction must be con-
trolled by optimizing the temperature of the aluminum and the time of con-

tact between the carbon fiber and the matrix material.

a) Interface ReaCtign; Mechanism and kinetics

The reaction between carbon fibers and aluminum matrix at the interface
occurs in two stages [21]. The first stage is an interface-controlled reaction
and involves three steps. First, carbon atoms dissociate from the surface of ‘
the fibers. Second, the “free” carbon atoms diffuse through the aluminum
oxide layer and other interphases. Third, the actual reaction between the car-
bon and aluminum atoms occurs to form A14C3. The dissociation of the car-
bon occurs when the bonds between the surface carbon atoms and the

underlying graphite are weakened by the presence of oxygen at the interface

16

(oxygen is always present in form of A1203). This result is based on the oxi-
dation of graphite in air studied by Long and Sykes[22]. Furthermore, the
microstucture of the interface shows that A14C3 forms on the aluminum side
of the interfacial oxide. This implies that carbon atoms diffuse into alumi-
num to form aluminum carbide. This is expected because of the availability
of “free” carbon atoms at the interface, and the smaller atomic size of carbon
as compared to aluminum. Maryama [23] studied the effect of water vapor,
that is adsorbed on the alunirna surface, on the interfacial reaction. He con-
cluded that the presence of water increases the dissociation rate of carbon and

thereby catalyzes the formation of the aluminum carbibe.

The second stage of the reaction is the growth of the interfacial layer
(A14C3). At this point the reaction is diffusion-controlled process because
the mass transport across the interface becomes the limiting step in the alumi-
num carbide formation. As the interfacial layer grows, carbon atoms have to
diffuse through longer distance , and thus require more time to reach alumi-
num. The growth rate, therefore, shows a square root dependence with time
and is given by the following equation:

x = Aoexp(- Q/RT)t ”2 (2.2)
X = The thicness of the reaction layer
A0 = Constant

Q = Activation energy

17

R = Boltzman constant
T = Temperature
t = Time
The measured activation energy value, that is published in the literature,
ranges from l3Kcal/mol [24] to 35.17 Kcal/mol [25]. More work is needed

to study the kinetics of this reaction.

b) Microsmrcmre of The Interface

Many methods have been used to characterize the interface in graphite/
aluminum composites. TEM was used by a group of researchers at MIT [26-
27] to determine the size, the shape and the amount of the reaction products.
In the case of pure aluminum matrix, they found that the A14C3 formation is
random and highly dependent on the time of contact between the liquid alu-
minum and the graphite fibers. When the contact time was minimized, only
few isolated A14C3 precipitates formed on the surface of the fibers. These
precipitates have needle shape and range from 0.15mm to 0.40 mm in size.
As the contact time increased the amount of the A14C3 phase increased drasti-
cally. Tangles of acicular A14C3 precipitates were formed throughout the
interface. These partially connected precipitates would grow to fom a contin-
uous layer of Al4C3 upon further exposure of the graphite fibers to molten

aluminum [28].

18

X-ray photoelectron spectroscopy (XPS) and Auger electron spectros-
copy (AES) are two powerful] techniques for the characterization of inter-
faces in composites. XPS and AES have high sensitivity to light elements,
such as carbon and oxygen, and good depth of resolution (2nm). They are
being used in carbon/aluminum composites to determine the chemical state
of the interfacial products (carbide, oxide or metal), and identify the local
bonds of an atom. AES is used to analyse the fracture path in metal-matrix
composites by studying the matrix Side and the fiber side of a fractured sur-
face. It also gives the distribution of elements normal to the interface. XPS
is limited to the study of deposited aluminum on a carbOn substrate because

of its large probe size.

c) Effect ef The Interface Reaetion on The Mechanical Propem'es:

The interface in continuous fiber composites plays two important roles.
First, the interface transfers the load to the fibers through good interfacial
bonds. Second, it provides toughness to the composite by allowing crack
propagation along the interface before fracture. This last phenomenon is
called the interface fuse mechanism[30] and requires “ not too strong” inter-
facial bondS. In carbon-aluminum composites, the presence of the A14C3
phase causes interface embrittlement and changes the mode of fracture which

results in premature failure. Crack propagation is no longer along the

l9

interface, but from one fiber to the next one by passing through the A14C3
phase. This is accomplished by the presence of very strong bond between the
matrix and the fiber formed by the A14C3. Under this condition, the interface
fuse mechanism can not operate and low fracture toughness is to be expected.
Furthermore, the brittle carbide phase can serve as sites for crack initiation.
Baker [29] estimated the intrinsic crack length in the reaction layer to be
between 20nm and 30nm. The reduction in the ultimate tensile strength

’ (UTS) from the presence of these cracks is evident as shown in figure 2.3.

Moreover, the formation of the aluminum carbide damages the fiber
suface and causes Significant decrease in the fiber strength. The interfacial
reaction consumes the fiber and reduces its cross section unevenly. The edge
atoms in the basal-plane of graphite are more active sites for dissociation
than the basal-plane face atoms [23]. This inhomogeneous reactivity causes
the irregular aluminum carbide formation, which leads to stress concentration
on the rough fiber surface or on the sharp A14C3 crystals.

Finally, The formation of aluninum carbide increases the interfacial bond
strength and therefore can be beneficial to the overall strength of the compos-
ite when the interfacial thickness is controlled. There is a strong possibility
that there exists a threshold level of carbide formation before the degredation
of the composite strength starts [27]. However, more work is still required to

make the aluminum carbide formation have a positive effect on the properties

20

 

 

 

__ 24* x
N -
'e
I
§ 2-O« ‘-
5 4
0'
E’ I614.
‘ui ..
g .—
9‘
§ ‘.24
3
O
5 OBI
S
04.. " : -
l
O

 

 

O 160 260 360 460 560 666 760
Annealing temperature l°CI

Figure 2.3 Reduction of ultimate tensile strength for temperatures

above 450 C because of crack formation [29].

21

of carbon/aluminum composite.

(1) Potential Selutions:

Much work has been done to control the interfacial reaction [31-35].
Coating of the fibers and addition of alloying elements have shown some suc-
cess. Nickel coating was used in aluminum composites to protect carbon
fibers from the liquid metal. Brittle interrnetallic compounds, such as NiAl3
and Ni2A13, still formed at the interface and caused lower mechanical proper-
ties than expected. Silver, which has high solubility in aluminum, was coated
on graphite fibers. Good wetting was achieved and no brittle compounds
were formed. Titanium boride was considered as a suiable coating for car- ,
bon fibers, but the wettability problem was not fully resolved. Other possible

coatings include K2ZrF6, Ta, TiC, TiN, SiC, B4C, etc.

Addition of specific elements to the metal matrix to reduce the A14C3 for-
mation was used in carbon-aluminum composites. Ti and Si are added to alu-
minum to reduce the activity of carbon in molten aluminum. This slows the .
rate of the aluminum-carbon reaction. TEM examination of graphite fiber-
reinforced Al-7%Si matrix Shows Si segregation at the interface which inhib-
its carbon diffusion into aluminum. Chemyshova [28] showed that 58 mg of

A14C3 was formed for lg of cabon fiber in pure aluminum and only 13.3 mg

22

of A13C4 for lg of carbon in Al-7%Si alloy. SiC layer may form at the inter-
face and serve as a diffusion barrier coating. Lithium, which is more known
for promoting wetting in A1203 fiber-Al matrix composites, can also be used
with carbon fibers. Li helps weaken the aluminum oxide layer and therefore
promotes wetting at lower temperatures [19]. Cu and Si are added to reduce
the melting point of aluminum ( melting occurs by eutectic transformation)
[27]. Reducing the temperature of the aluminum matrix decreases the
amount of aluminum carbide that may form at the interface. As a general
conclusion, most of the work that is being done in this area is toward decreas-

ing the kinetics of the aluminum carbide formation.
2.2. Solid-Phase Processing

There are two steps in making CMMC using solid-phase fabrication
methods: making of preformed composites and consolidation by diffusion
bonding.

2.2.1. Preforming Methods

Preforming in metal-matrix composite consists of coating tows of fibers
with the metal matrix or with an adherent material that would assist in the
later steps of fabrication. Layers of coated fibers are then layed-up in the

desired form and are consolidated by diffusion bonding. Preforming is

23

usually chosen when good control over the interfacial properties is critical.
Low processing temperatures, short contact time and/or diffusion barrier

coatings are the means to achieve this goal.

Fiber coating must satisfy some requirements. First, the coating should pro-
mote wetting between the fiber and the matrix. Second, it should serve as a

barrier against excessive interfacial reactions spacially at high temperatures.
Third, the coating must support the expected load transfer from the matrix to
the fibers. Finally, it should have Similar density to the matrix in order to

keep the weight advantage of the composite.

The overall quality of the preform material depends on the efficiency, the
thickness and the uniformity of the coating. Each individual fiber must be
coated to avoid fiber breakage from fiber-to-fiber contact. The applied coat-
ing should be uniform throughout the fiber length. Finally, the preform
Should be easy to handle during lay-up to minimize fiber breakage and

matrix/coating loss.

There are many methods for making precursor materials. Some of these

methods are described next.

24

a) Spraying:

It consists of spraying molten metal onto bundles of prewound fibers.
The matrix material is melted by gas explosions, arc (figure 2.4), flame, or
plasma. A compressed hot air pushes small droplets of the molten matrix
toward the fibers at very high Speed. The liquid droplets freeze instantly
when they contact the cold fibers. The rapid rate of solidification causes a
fine matrix microstructure to form around the fibers. On the other hand, the
instant freezing of the droplets decreases the mobility of the liquid metal and
therefore reduces the fiber surface area that could have been wetted in the
case of a moderate cooling rate [36]. This generates voids and causes lower
mechanical properties. Microcracks are also formed during the rapid solidifi-

cation because of thermal stresses [37].

b) Electreplating:

It starts by winding fibers on almandral (or bobbin) and placing them in
an electrolic solution that contains the metal matrix. The fibers represent the
cathode while the matrix forms the anode. Positively charged metal ions
move toward the cathode and deposit onto the fibers. A uniform coating of
the matrix is usually obtained with no voids. Nickel, copper, aluminum and

lead were successfully deposited by the process [38-39]. The aluminum

Arc spray head

Power J

Spray gas-l

 

Matrix spool , Vacuum chamber

' .0 4..
_ .'. 3.. .23.)“: .
”at £5“ -‘.. _ ‘ ~
,1 3'. .vd, m...
\ . .. .
e’-I..:e)e & ‘u‘
l . .5 1 .. "S '
. . .'
. . - v . .' u.
” v . .‘ ...
I . I I I
. . . . . .

 

Figure 2.4 Arc-spray process for the fabrication of composite

monotape [52].

26

alloys use non-aquesous solutions or liquid salt for safety precautions. The
strength of the composite made by this process is above 80% of the role of

mixture (ROM) value. This method has the potential for a low cost process.

0) Chemical Vapoer Deposition (CVD [:

This process is used more frequently to coat fibers with wetting agents or
with diffusion barrier coatings againt fiber/matrix interactions. It begins by
heating chemical components until vaporization. The desired Species are
deposited on the fibers while the rest of chemicals escape the system as gases.
CVD has been used to coat carbon fibers for Al-based composites [40-41].
Some of these coatings include TiB, Cu, Ni, Ta, TiC, TIN and SiC. CVD is
highly dependent on the temperature of the reaction chamber (usually much
lower than the melting point of the matrix). Full Contol of all the process
parameters is required to make a high quality coating. This process is Slow.
A uniform coating is not always easy to achieve. The strength of the com-

posite is only 80% of ROM.
d) Physical Vapor Deppsitjen (PVD I:

This process is Similar to CVD. The matrix material is deposited on the
fibers in the vapor phase but without the help of chemical components. Alu-

minum-carbon composites were made by this process [42]. The tensile

27

strength was close to ROM value. This process offers a good control over the
deposited matrix in a clean atmosphere (no chemicals). However, PVD is

time consuming and can be expensive.

e) Dipping:

It consists of passing a tow of fibers through a molten matrix to make
tapes of preforrns. Sodium process is a well known method for the fabrica- .
tion of aluminum-carbon composites. Figure 2.5 Shows all the steps involved
in this process [43]. Graphite fibers pass through a bath of sodium (sodium
is known to wet graphite), then through tin and last through liquid aluminum.
The strength of the composite made by this method was reported to be greater
than 90% of ROM. Some of the problems in this process include uneven dis-
tribution of the fibers throughout the composite, and fiber damage from the

possible exposure to molten metals.

2.2.2. Consolidation By Diffusion Bonding

a) Meehapism:

Diffusion bonding is a solid state process that involves the joining of two

surfaces by applying heat and pressure. It requires the forrnaiton of strong

28

ARGON ATMOSPHERE

    

GRAPHITE
/_ FIBER

20 -
ABOVE M.P.
OR LlOUlDUS

Figure 2.5 The sodium process for the infiltration of carbon fiber

tow [43].

29

adhesive bonds between asperities. A bond iS formed when new grains
appear at the interface region and only few small pores might remain at the
joint line. The rate-controlling step for diffusion is the removal of interfacial
voids that exist from surface roughness. The temperature used ranges from
0.5 to 0.8 of the melting point of the metal matrix. The applied pressure
should be well below the yield stress of the matrix to avoid excessive plastic
deformation of the bulk material. The best case senario for the consolidation
of powders by diffusion bonding is to carefully choose the processing param-
eters (temperature, pressure and time) to give low flow stress and high ductil-
ity of the matrix material and at the same time retain the fine microstructure

obtained from the use of powders.

Modelling of diffusion bonding has been studied by many researchers
[44-49]. Derby and Wallach [44] were among the first authors to suggest that
diffusion bonding consists of five mechanisms. First, plastic deformation of
the powder surfaces. Second, power-law creep of the material at the contact-
ing surfaces. Third, diffusion of matter from interfacial voids at the surfaces
to the growing necks. Fourth, diffusion of matter from bonded regions to the
necks. Fifth, bond fOrmation by vapor mass transfer. This last mechanism is
the least important especially for metal diffusion bonding because of the low
vapor pressure of metals during typical bonding conditions. All these mech-

anisms are carried out by creep ( for the plastic deformation) and/or diffusion

30

(by volume, grain boundary, bonded interface and cavity surface diffusion).
Power-law creep and diffusional flow may be active simultaneously or at dif-
ferent times depending on the interface geometry and the thermodynamics of

the system (driving force) at that time.

A diffusion bonding cycle goes through three stages: first, intermediate
and final stage. During the initial stage, plastic deformation (by instantenous
yielding at the beginning and by creep most of the way) and fracture of sur-
face oxide layers are the dominant mechanisms. The desired goal is to
increase the contact area between surfaces and produce an intimate contact
between oxide-free and freshly formed surfaces. Breaking the oxide films is
very irnortant especially for powder consolidation where surface area is the
largest. Calvo et al [45] has Shown that the formation of metal oxide surfaces
depends on the standard energy of oxide formation (AG) and on the solubility
of oxygen in metals. Large AG and low solubility give probable conditions
for oxide formation. Aluminum, for example, has a AG value of -365 call
mol and oxygen solubility of 0.005 atm% for T/Tm :05. AS a result, diffu-
sion bonding of Aluminum is highly dependent on the removal of the alumi-
num oxide layers that form on the surface. During this stage, most of the
oxide films is broken by shear displacement between two asperities in contact
as a result of the applied pressure, and by dissolution or spheroidization of

the oxide layers as a result of the diffusional flow at high temperatures. At

31

the end of the first stage, the bonded area contains high void volune fraction.
The size, the Shape and the amount of these interfacial voids depend on the
initial roughness of the surfaces in contact. These voids need to be elirni-

nated in the remaining stages.

The intermediate stage is dominated by diffusion of matter across the
bonding interface. This causes changes in the Size and the Shape of the inter-
facial voids, and the nucleation of interrnetallic compounds [45]. Long pris-
matic or lenticular voids are converted to smaller spherical voids. The driving
force for the Shrinkage and the spheroidization of these pores is the difference
in the chemical potentials between the growing neck and the rest of the sys-

’ tem. Finally, the third stage starts when most of the voids have become
spherical. This stage marks the beginning of vOids collapse and disapperance
mainly by interface and volume diffusion, and the growth of any intermetal-
lics formed during the previous stage. As the Shrinkage and the elimination
of these pores may take a long time, the growth of the interrnetallic com-
pounds become very important phenomenon that must be considered during

parameter optimization.

One of the main requirements for successful diffusion bonding of MMC
is the use of optimum processing parameters. Unfortunately, specific fabrica-

tion conditions for MMC are hard to find in the open literature. This is

32

mainly due to two factors. First, most of MMC research is curried out by the
defence industry where the government considers this information as vital to
the national security interests. Second, most companies that make MMC
classify the procesSing parameters under proprietary information in order to
keep their competitive edge. Therefore, finding the optimum processing
parameters has to be achieved through experimental trials. One popular
method for parameter optimization is to vary one parameter (temperature)
and keep the other ones (pressure, time, vacuum level, fiber volume fraction,
etc.) constant. To evaluate the chosen parameter, one response (flexural
strength, ultimate tensile strength, void volume fraction, etc.) is measured.
Figure 2.6 Shows variation of the flexural strength with hot press temperature
and figure 2.7 Shows the flexural strength in terms of hot press pressure for
Al-C composite [49]. Depending on the desired properties of the composite
obtained from these tests, one can choose the “optimum” conditions for
MMC consolidation. This type of approach was used by Islam [50] to evalu-
ate Some of the processing conditions used during the consolidation of Al-C

composites.
b) Me ds:

Solid state processing of metal-matrix composites is carried out by many

methods. Since the focus in this thesis is on the fabrication of continuous

33

 

(D
O
I

Flexurol Strength( kglmm2 )

 

O

 

i—"

/ .

. '0’---‘- --
/ o 8
e f g e

 

 

70" «TI ‘ J
8” 3
o"’
50: -°- HT-r .1
:r -"--HT-9
‘ 3 6' 9

Figure 2.6 Variation of flexural Strength with hot press pressure [50].

Hot Press Pressure I Itglmmz)

b
o 8

N
O

Flemml Strength I. kglmm2 )

Figure 2.7 Variation of flexural strength with hot press temperature [50].

 

I

 

 

”0 o B“‘~O\s~0
l I III-9 I
500 540 560

Hot Press Temp. (°C I

 

35

fiber-reinforced metal-matrix composites, only the techniques that use thin
Sheets of precursor material are considered. The precursor can be made of
bundles of fibers drum-wound onto a thin foil of the metal matrix and held in
position by fugitive binders, tapes of plasma-sprayed fibers, or preformed
composite monolayers. The precursor tape is cut into pieces of certain
length, laid-up in the desired form and hot pressed by one of these methods:
Vacuum hot press [51], step pressing [52], hot isostatic pressing [53], hot-roll
bonding [54], hot-die molding [55] or hot-drawing [56]. Vacuum hot press is

reviewed next because of its popularity among researchers.

Vacatpp Hot Press: The process consists of hot pressing several layers of pre-
cursor in a vacuum environment to make dense composite materials. Pre
formed composite layers are cleaned from residual organic materials, greases
or thick oxide layers. They are cut and placed in a vacuum bag. A vacuum is
drawn to remove reactive gases such as oxygen, hydrogen and water vapor.
Oxidation of the metal matrix and the fibers (boron fibers for example) and
the formation of undesirable products are the main reasons for a vacuum
atmosphere. If organic binders are used, the vacuum-bagged preform must
be heated to burn off the binder and de-gassed to achieve the required vac-
uum level before consolidation. Once this preliminary work is completed,
the material is placed in a preheated mold and a constant pressure is applied

for a given time. Figure 2.8 Shows all the steps involved in the process.

36

Vac'uum hot press is Simple, inexpensive and gives hi gh-quality material.
However, large or complex forms are difficult to produce and may take long

time because of all the steps involved.

37

 

I

 

 

e
e . e
e . e e e

'0 l.e.I..o .

’e e e 0 U I I e
u e a e
ee‘.e.e. e e
-

 

 

 

 

 

 

 

 

 

 

 

       

 

 

   

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

[ 0000600 [ 0000006
PHECURSOR CLEANEDPRECUHSOH
1
6090006 9990093
F." '“"'.'.'-,':CCL.‘C.”’.. ...',‘. f g , .j;
(EGOGDQHOQD I quugqpqng
Baawflia “fififififif
GHOGDGDGHO GHQ GHSQD
GH3¢3€96H3 GDSWDGDGHO
OOOOOGOI “3366656‘
J
LAY-UP BAGGED LAY-UP
OOOOOOO
Iiiiilllill
(IGDCDCMCH'
3: s s:
000000 00
. o ‘ . . C C C C . .
.00....
FINISHED COMPOSITE

CONSOUDAHON

Figure 2.8 Vacuum hot press.

Chapter 3

3. New Process

This new technique was originally developed by the Composite Materials
and Structures Center at Michigan State University. It is designed to make
continuous fiber-reinforced polymer matrix composites [57]. Modification of
the process had to be made to] make it suitable and safe for the fabrication of

MMCS. All changes are described in the experimental part of this thesis.
3.] Description of The Original Process:

A general review of the process is presented first to Show how the process
works, then more detailed descriptions of the spreader and the aerosolizer are
given next because of their significant role in this research.

3.1.1 Operation of The Process:

This technique causes a fiber tow to go through different chambers to

make a prepreg tape of a polymer composite. This process is shown in

38

39

 

 

 

 

 

 

 

 

 

 

- 7—1 ‘ W -
O __ L____, W. 0
Fiber PIP-WWW .. .' . ' Heater Tm“
spool Spreader I be: drum
Imprcgnation
chamber

Figure 3.1 Powder prepregging process [57]

40

figure 3.1. A fiber tow is driven by a dc. motor from a fiber spool to pass
above a Speaker. The sound waves coming off the speaker Spread the fibers
apart. The spread fibers are held in position by ten steel shafts Spaced one
inch apart and placed on top of the speaker. After spreading, the fibers pass
through an optional pre-treatrnent chamber to modify the fiber surface or to
apply a thin coating of binder material to improve adhesion with the matrix.
After that, the fibers enter an impregnation chamber, called aerosolizer,
where small polyamide particles (about 10 microns) are suspended by the
effect of a vibrating rubber membrane placed on top of a speaker to create a
bed of polymer powders. The powders are attached to the fibers by the effect
of electrostatic forces generated from the static charges held by the fine poly-
mer particles. Coated fibers leave the fluidized bed and enter an oven cham-
ber for about 15 seconds. The particles are heated by convection and
radiation until sintering between adjacent particles occurs to form a thin film.
Surface tension forces cause the film to break into small droplets that sur-
round the entire fiber cross section. The impregnated fibers are then wound
on a take up drum. Sequences of the same events make up a Single run of this
continuous process. After one run, the resulting prepreg tape is cut into
pieces to a specific length and are laid-up in a rectangular stainless Steel mold
for consolidation. Thirty to forty pieces of the prepreg tape are hot pressed
according to a pressure-temperature-time profile. A thin Sheet of composite

material is thus formed and can be evaluated.

41

3.1.2. Spreader

The Spreader is made of a 25.5 cm (10 in ) speaker mounted on a ply-
wood box as Shown in figure 3.2. The speaker operates at various frequen-
cies and amplidutes with the help of a frequency generator and power
amplifier. The acoustic energy coming off the speaker forces the collirnated
fiber tow to spread to its individualfilaments. The efficiency of the spreader
is maximum when the speaker is operated at its natural frequency. A fiber
tow passes through a guide ring and is pulled by a pair of nip rollers to pass
above the speaker. The purpose of the nip rollers is keep a constant level of
fiber tension in the tow as it passes above the speaker. The fiber tow goes
above and under a set of polished steel Shafts mounted on top of the Speaker
to hold the spread fibers in position. These Shafts can rotate whenever too
much friction between the fibers and the Shafts is encountered. Figure 3.3

Shows the spreading operation.

3.1.3. Aerosolizer

The aerosolizer is made of a plexiglas column mounted on top of a ply-
wood box that contains a Speaker. The speaker is operated by a power
amplifier connected to a frequency generator. The colunm has two rubber
membranes that are placed on the top and the bottom to create a standing

wave inside the colunm. There are two slits on both Sides of the column near

42

 

 

POIISth Shafts Bcaring blOCk
Narrow fiber row / Spread fiber low
I
l¥fi \ I fl
_ "' . _, =— '--T =—-_T-_=
’ \\ __-.: 3‘ , _. _j"‘ :37} __=_

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

O _ I
lElIDIEIJII o O I—
Power amplifier

Frequency generator

 

Figure 3.2 Spreader [57].

 

43

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 3.3 Spreading operation [57].

the top to allow for the fibers to enter and leave the aerosolizer coated with
polymer powders. A gas inlet is designed above the packed powder bed level
near the bottom to help with the fuidization mechanism. Figure 3.4 Shows

the aerosolizer.
3.2. Advantages of The Proposed Process

This process has all the benefits of powder metallurgy techniques, such as
fine matrix microstructure, minimizing undesired interface reactions and pos-
sibility of superior mechanical properties. It also offers some additional

unique advantages:

1) Fibers are evenly distributed throughout the composite. The acoustic
energy forces the fibers to spread apart. This reduces fiber damage and

improves fiber alignment.

2) Uniform distribution of the matrix around each fiber. This is achieved
from the use of particles with smaller size (5 .5 microns) than the diameter of
the fibers (8 microns). Using the acoustic energy coming off the speaker as a
means of fluidization assures the use of fine particles Since large powders
Sink to the bottom of the bed. This method eliminates uneven matrix coating

and/or matrix-rich composites.

Vibrating rubber diaphragm

O-ring
Unirnpregnatcd WM. . ‘ Powder-Unpregnated
spread tow £ch?? . swim“; spread tow
Tire-:15" Ir“ ' ’. F-~"~"~:4"'5 ;

 

.4——-—--—- Entraincd particles

 
 
 
 
   

 

 

Nitrogen gas into ,1: Packed bed of

aerosolizer , —£ ' - p ' powder particles
Vibrating rubber
diaphragm

 

Plywood cashtK / Audio speaker

7"]

0

To power amplifier
and frequency generator

 

 

 

 

 

 

 

Figure 3.4 Aerosolizer [57].

46

3) Precise control over the fiber/matrix volume fractions is often achieved by
this technique. This reduces the scatter of the mechanical properties through-

out the composite.

4) High fiber volume fraction (more than 60%) is obtained by this process

mainly because of the effective use of the spreader.

5) Small diameter-fibers (7 microns) can be used without a Significant fiber

damage usually caused by fiber-to-fiber contact.

6) High quality composites can be made using this powder process. Homo-
geneous fibers and matrix distribution, high fiber volume fraction, small
diameter-fibers and reduced fiber and interface damage can result in the fabri-

cation of composites with superb mechanical properties.

7) This process is far less expensive than most of the existing MMC fabrica-

tion techniques since it is amenable for mass production.

8) The process is automated, though lay-up of prepregs still requires manual

labor.

Chapter 4

4. Safe Handling of Aluminum Powders
4.1 Risk Factor: Explosions

Aluminum reacts instanteneously with oxygen to form a thin film of alu-
minum oxide on the surface of the aluminum powders. The oxide layer is
stable in air and prevents further oxidation of the underlying aluminum.
However, if fine aluminum powders; usually less than 44 microns (325
mesh), are suspended in air and are heated by one source of ignition to reach
the ignition temperature, then each particle will burn entirely and release an
enormous amount of heat. This heat will increase the pressure of the system
(the volume of air that contains the suspended powders ) at a very fast rate of
138 MPa ( 20,000 psi/sec) [58] so that serious explosions may occur. .Figure

4.1 shows the increase of pressure with time for aluminum and other compo-

nents .

An explosion is often referred to as a gas-dynamic phenomenon [59]
caused by rapid increase of the system pressure. In dust explosions, the rapid

oxidation of fine particles dispersed in air results in a rapid energy release

47

PRESSURE. em

48

 

    

 

 

 

A
9 __ / \ _
\

I \
a - \ ‘

L—Aluminum
7 P I '4
6 h .4

Polyethylene
5 —l
4 _
"/7..." —
3 /../ "
2 '/--'/<—Iron sullide ore "‘
, l
'0 0.1 0.2 0.3
TIME. s

Figure 4.1 Increase of pressure with time for aluminum powders

during oxidation reaction [60].

49

which increases the temperature of the system so rapidly that a pressure
build-up follows. Destructive pressure forces are generated and can destroy

structures and hurt nearby personnel.

Dust explosions are described by the ideal gas law:
PV=nRT (4.1)
where P = Pressure
V = Volume
II = Number of gas moles
T = Temperature

R = Gas constant

If the reaction occurs in a fixed volume and the temperature of the system
increases from room temperature to a temperature T1, then the pressure will
increaselfrom its initial value Po to a maximum value Pmax according to :

Pmax o = Tl/To (4-2)
If the system volume is not fixed by confining structure, explosions may still
occur because the reaction rate is so fast that gas motions may be too Slow to
relieve the developing pressure. The system in this case, is confined by its

own dynamic state.

As a conclusion, handling explosive powders is very challenging. It

50

requires carefull planning and good understanding of the dynamics of the

system.
4.2. Sources of Ignition

Aluminum powders have very thin film of aluminum oxide on the sur-
face. This oxide layer prevents the instantenous ignition of the powders
when they are exposed to air. Aluminum powders will ignite only if a suffi-
cient amount of oxygen, a favorable concentration of powders and one source

of ignition are all present.

The oxygen volume that is necessary to ignite aluminum powders
depends on the other gases that fill up the system volume. It has been
reported [60] that 9% oxygen with the rest being nitrogen, or 10% oxygen
with the balance being helium, or 3% oxygen Willi the rest being carbon diox-

ide are some examples of gas mixtures that can cause ignition.

The minimum explosible concentration is an important parameter for the ‘
powder industry. Tests were made to determine the minimum amount of alu-
minum powders that can be dispersed in air without causing fires or explo-

sions. This value was found to be 45g/m3 [58].

51

Sources of ignition can be classified into three groups. The first group
consists of open flames and hot surfaces that can raise the the temperature of
the aluminum powder to its ignition point. As little as 25 milijouleS can
cause ignition. The second group includes electric sparks that are generated
from the electric power or from the discharge of electrostatic charges. Electri
power from the mains or batteries is a powerful source of ignition because the
energies available are usually greatly in excess of those needed to ignite alu-
minum dust. The generation of static charges can not be avoided during pow-
der handling, therefore attention Should be paid toward the safe removal of
the static charge to prevent their accumulation. The transfer of these static
charges to a conductive material that is not grounded causes a spark that
could ignite the aluminum powder. The safe measure to take in this case is to
ground all the conductive materials that may come in contact with the pow-

ders.
4.3. Basic Safety Precautions:
In this section, some basic rules that are recommended by the Aluminum

Association to minimize the danger associated with handling aluminum pow-

ders will be presented :

52

a) Static charge:

- All equipment that is used to handle the aluminum powders must be
grounded.

- Use conductive, non-Sparking metal scoops to work with powders.
Plastics and ferrous metals are not recommended.

- Avoid pouring powders on non-conductive surfaces.

- Avoid having isolated metal objects that are not grounded in the pro-

cessing room.

b) Eleetric equipment:

- Keep all the unnecessary electrical equipment out ot the working room.
- Cover all electrical wires that are connected or at nearby distance to the
processing area.
- Avoid having broken electric devices such as broken light bulbs or

loose electric power connectors.

c) Flar_r_1es a_nd Hot surfaces:

- No smoking is allowed while working with powders
- Do not allow open fire or sparks in the processing area.

- Avoid having matches, lighters or any objects that may cause a Spark or

53

a fire.

- Choose the working area to be away from steam pipes or radiators to

avoid heating the powders

(I) General Rules:

-Aluminum powders should be handled gently and carefully to avoid
heating from agitation, or suspending the powders in air to create dust.

- Keep water away from aluminum powders. Aluminum reacts with
water to generate heat and hydrogen gas which can create pressure build up.

- Keep track of any loose powders to avoid accummulation of powders
on floors or on nearby objects.

- Keep rubbish, such as dirty maintenance equipment, and combustibles
out of the precessing room.

- Wear smooth fire-resistant clothes since clothes can be a place for static

charge accumulation.

Chapter 5

5. Experimental Chamber

An experimental chamber is designed to meet two important criteria.
First, the chamber should assure safe handling of fine metal powders. These
powders are known to be explosive and present a great risk if they are
exposed toair. Second, the chamber should not interfere with the continuity

of process if future commercial scale-up is warranted.
5.1. Theory of The Design
5.1.1. Thermodynamics Approach

This approach sets safety conditions for working with aluminum pow-
ders. Aluminum reacts spontaneously with oxygen to form aluminum oxide

according to :

72A1 + 3/2 02 —> A1203 (5.1)
The heat released during this reaction AH (cal) is given in terms of the tem-
perature T (K) at which the reaction Occurs:

AH = -400,810 - 3.98 T ( in cal) (5.2)

54

55

At room temperature, T=298°K, AH = -401,996 cal/mol.
This enormous amout of heat generated during the formation of one mole of
A1203 can raise the temperature of the system to above 4900°C. A big and
sudden increase in the temperature of a system in air or in a confined chamber
results in pressure build-up which can cause explosions (see chapter 4). Pre-
venting the oxidation reaction from occuring, or more practically, being able
to control the conditions of the reaction if small amount of aluminum is oxi-
dized is the safest approach in metal powder fabrication methods. Calcula-
tions are made to determine the temperature of the system and the allowed
oxygen volume percentage for safe handling of aluminum powders. These
calculations are based on three assumptions:
1) The reaction occurs in dry air.
2) Dry air is made of 80% nitrogen and 20% oxygen.
3) The heat generated during the reaction is all used to raise the temperature
of the remaining components of the system (80% N2 and one mole of A1203).
These assumptions lead to the following equation:

2 Al + 3/2 02 + 6 N2 = A1203 + 6 N2 , AH = -401,996 cal (5.3)
The amount of heat needed to raise the temperature of a component (x) from

room temperature to a temperature T1 is given by AH:
Tl

AH = I Cp(x)dT (5.4)

298

where Cp (x) is the heat capacity. of a component (x) at constant pressure p.

56

Using equation (5.4), the amount of heat that is used to increase the tempera-
ture of one mole of A1203 (AHI) and 6 moles of N2 (AHZ) can be determined
as a function of the temperature T 13

7'

AH] = [Cp(A1203)dT (5.5.a)

298

AH2 = 6 ]Cp (N2)dT (5.5.b)
293

Applying the conservation of energy law to the whole system gives:

AH+AH1+AH2=0 (5.6)
Solving equation (5.6) for the temperature of the system after the reaction
occurs gives T1. Plugging in numerical values for Cp(A1203) and Cp (N2)
gives T1 = 4918°C. These calculations indicate that oxygen has to be
removed from the experimental chamber in order to run safe experiments.
Consequently, a vacuum system and an argon source seem to be necessary
components of the desired chamber. A vacuum would be pulled on the
chamber to reduce the oxygen level, and argon is introduced afterward to

bring the system to atmospheric pressure.

Using the same thermodynamics approach, one can determine the opti-
mum oxygen/argon ratio to keep the temperature of the system low in case all
the aluminum powders were oxidized. These calculations are based on the

following equation:

57

2A1 + 3/202 + XAr + 6N2 —> A1203 + XAr +6N2 (5.7)
The amount of the heat that is needed to raise the temperature of X moles of
argon to a given temperature T lis given by AH 3 :

AH3 = X] Cp (Ar) (17‘ (5.8)

298

Similarly, one can write:

AH+AH1+AH2+AH3=0 (5.9)
Solving equation (5.9) for X gives the number of Ar molesthat is needed to
keep the temperature of the system at T1. For example, if T1 = 43°C, then X
= 4486 and the Ar/02 = 2991. This Shows that if the reaction proceeds as
Shown in equation (5.7) the experiment becomes unpractical. Therefore, the

amount of oxygen that may be available in the system must be reduced.

The amount of oxygen left inside a chamber can be determined in the fol-
lowing way: First, assume that after pulling a vacuum on a chamber of vol-
ume V to decrease the pressure from one atmosphere(l atrn.) to a pressure P0,
only no moles of 02 and 4n0 of N2 are left. Assuming also that the equation
for an ideal gas is applicable to the system, one can write:

SnO = P0 (V/RT) (5.10)
Second, assume that 111 moles of Ar are added to the chamber to go back to
atmospheric pressure (1 atm). The total number of gas moles n is thus given

by n = 5n0+n1. Applying the equation of state gives :

58

n = (1 atm.) (V/RT) (5.11)
Substituting 11 by (5n0 + ml), and the ratio of 02 moles to the total number of
moles gives:
5n0/(5n0+n1)=P0 » (5.12)
Rearranging the above equation gives the ratio Ar/Oz moles as:
n1/n0=5[(l/P0)-1] (5.13)
If the volume of the chamber V is fixed at 37,000 cm3; this value corresponds
to a cylindrical chamber with 26.7 cm (10.5 in) wide and 66 cm (26 in) long,
then the total number of gas moles n at room temperature T and atmospheric
pressure P is given by:
n = PV/RT (5.14)
Plugging in numerical values for P, V, R and T gives 11 = 1.513 moles. There-
fore, the number of 02 moles 111 can be detemimed if the ratio Ar/02 is
known. For example, if nl/no = 99 and 5n0 + n, = 1.513 moles, then n, =

14.55 x10"3 moles. Table 5.1 gives the Ar/02 ratio and the number of 02

moles for different vacuum levels.

Once the oxygen content is known, then the problem reduces to deter-
mining the temperature of the system in the case where all the oxygen has
been consumed during the oxidation of some of the aluminum powders.

Now, equation (5.7) becomes: ,

59

 

 

 

 

 

 

Vacuum Ar/02 Number of 0:5, gen
level (torr) Ratio 0 moles I” ume
2 percentage
36.5 99 14.55 x10“3 0.96%
24.0 150 9.76 x103 0.65%
11.5 328 4.54 x103 0.30%
0.76 4995 0.30 x103 0.02%

 

 

 

 

 

Table 5.1 The number of argon moles to the number of oxygen

moles ratio for various vacuum levels inside a chamber.

'60

aAl + b02 + cAr + 4bN2—> dAle3 + eAl +CAr + 4bN2 (5.15)
The parameter a is to be chosen by the experimenter, the value of b and c can
be determined when the vacuum level is known and the parameters d and e
depend on the value of a and b. For example, if the vacuum level is 0.048
atrn. (36.5 torr), then the Ar/02 = 99 and the number of oxygen moles n1 =
14.55 x103. If the experimenter chooses to use 1 g of aluminum powder, then
the number of Al moles
a = 1/27 = 37. 04 x103. Equation (5.15) becomes:

37.04 x103 Al + 14.55 x103 02 + 1.44045 Ar + 58.0 x10-3 N2 —9

9.7 x10'3 A1203 + 17.64 x10’3 Al + 1.44045Ar + 58.0 x10-3 N2 (5.16)
Using the same thermodynamics approach used earlier, the temperature of
the system T1: 542 C.

Since the temperature of the system after the reaction occurs can be
found, then the resulting pressure inside the chamber P and the pressure dif-
ferencial P can also be determined using eqation (5.14). Table 5 .3 summer-
izes the results of this approach. An argon to oxygen ratio of 99 or above

gives a safe experiment.

As a conclusion, the above thermodynamics approach can give a good
approximation of the danger that may be involved during the aluminum oxide
formation reaction. Knowing the vacuum level reached by the experimental

chamber before introducing argon and the amount of the aluminum powders

61

 

 

 

 

 

 

 

 

 

 

Tempe- Pressure Pressure
Ar/ 02 . Change
. rature 1n MPa .
Ratio (C) (psi) In MPa
(Psi)
99 542 0.275 0.175
(40) (25)
150 360 0.220 0.115
(32) (17)
328 183 0.160 0.055
(23) (8)
4995 36 0.105 neglige-
(15) able

 

Table 5.2 Summary of the resirlts from the thermodynamics approach.

 

62

used during the experiment is the key to a safe procedure.

5.1.2 Materials and dimensions:

With the results from the thermodynamics approach, the experimental
chamber can be built.

The chamber will be made of plexi-glas material because fluidization of
the powders requires visual adjustments to determine the appropriate fre-
quency for the operation of the Speaker. The chamber will have similar
dimensions to those of the original aerosolizer. The reason for this design is
to be able to add the experimental chamber to the whole system once good
results are obtained. The process will appear as Shown in figure 5.1. This
way, the coated fibers leave the oven chamber with a Sticky polymer coating

to enter the new chamber where they get coated with aluminum powders.

Since the chamber requires a vacuum system, one cocem will be whether
the plexi-glas cylinder can withstand an external pressure P of 1 atrn.
(14.7psi). Solving the problem in cylindrical coordinates system gives the

following stresses:

on = - b2/(b2-a2) [1-32/12] P

63

 

 

 

 

 

 

 

 

 

 

 

 

Heater New Takeup
Chamber Drum

 

fiber Speaker Pre-Treatment

spool Chamber Coating

with Nylon

Figure 5 .1 Addition of the experimental chamber to the pOwder

prepregging system.

64

099 = -b2/(b2-a2) [1+a2/r2] P

Where a and b represent the inside and the outside raduis of the cylinder
respectively. For a cylinder with 26. 6 cm (10.5 in) inside diameter and 0.64
cm (1 /4 in) thickness, the maxirnun compressive stress is at the inner surface
of the cylinder 0'99 = -2.30 MPa (33lpsi) and the Stresses in the direction of
the thickness 0'rr are constant (Cirr = -P). The compressive strength of plexi-
glas is lOOMPa (14.5Ksi). This gives a safety factor of more than 40 and
Shows that a plexiglas tube can be used as a vacuum chamber. The bottom lid
will have holes for vacuum feedthroughs. A plexiglas lid may not support
the stress concentration generated around the holes. Therefore a stronger

material will be used for the top and the bottom cover of the cylinder.

Another anticipated problem with the use of a vacuum system is the rate
at which gas is removed and reintroduced to the chamber. Fine aluminum
powders may get carried by the air motion to the pump or outside the cham-
ber where major problems can occur. To deal with this problem a needle
valve will be used to Slowly evacuate the chamber and fill it back with argon.
Air filters will be placed at the beginning of each gas path to stop any pow-
ders from leaving the chamber. Moreover, the air inside the glass chamber

will be removed from the top of the chamber while the powders rest on the

65

rubber menbrane at the bottom ofthe chamber.
5.2. Description of The Chamber

The designed chamber is made of a closed plexiglas tube that contains a
speaker, a speaker wood box, a glass chamber, and a small aluminum tube
that connects between the speaker and the glass chamber. Figure (5.2) gives

an overall view of this chamber.
5.2.1 Outside Chamber

This chamber is made of plexiglas material for the purpose of observing
the behavior of the aluminum powders during fluidization. The tube has
these dimensions: 26.7 cm (10.5 in) diameter, 66 cm (26.0 in) length, and
0.64 cm (1/4 in) thickness. The chamber has two lids for the top and bottom.
The lids are made of aluminum and have o-rings around the inside to assure
sealing for vacuum purpose. Both lids are 1.27 cm (1/2 in) thick. The top lid
has Six 1.27cm (1/2 in) wide holes for electrical and vacuum feedthroughs.
During air evacuation, bending forces caused by atmospheric pressure are
applied to the top and the bottom of the tube. Having holes on top will add to
the effect of these forces Since stress‘concentration arises around the holes.

To deal with this problem, the Six holes are made around

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

; A
B
T“—
fl C
11* D
E
H—— F
A: Plexiglas Tube B & D: Membranes
C: Glass Thbe E: Speaker

F: Wood Box

Figure 5.2 The experimental chamber.

67

the edge to avoid the middle section where maximum bending forces are
applied, and the lids are made of aluminum instead of plexiglas to improve

the strength. Figure (5.3) Shows the described chamber.

5.2.2 Inside Chamber

This chamber is a hollow glass cylinder where the actual experiment
occurs. It has these dimensions: 10 cm (4.0 in) in diameter, 38 cm (15 in) in
length, and 0.32 cm (1/8 in) in thickness. Half an inch from the top, a small
indentation in the outside is for a 9 cm (3.5 in) O-ring to hold the top mem-
brane. At about 7.5 cm (3.0 in) from the top, Six tungsten pins are mounted
around the circumference to serve as electical feedthroughs for the heater and
as a means of grounding in the case of static charge build up. The glass
chamber has two 0.64 cm (1/4 in) gas ports. One near the top for pulling a
vacuum on the inside chamber, and the other one at the bottom for introduc-
ing argon after outgassing.The choice of glass over plexiglas was made on
the fact that heating may occur from the electic current going through the
tungsten pins and also from the heater itself. Figure (5.4) shows the glass

chamber.

68

 

 

 

 

 

 

 

Figure 5 .3 Outside chamber.
A: Plexiglas tube.
B: Design of the top and bottom lid.

C: Bottom view of the top lid.

69

 

 

> é O-Ring Location

:— Gas Inlet

 

15" \

Tungsten Pins

 

 

 

 

Gas Inlet

_ Figure 5.4 Inside Chamber

70

5.2.3 Speaker:

An audio speaker capable of up to 100 watts and has 4 ohms resistance is
used in this chamber. The speaker is mounted inside a wood box that has a
circular opening on top to allow the upward propagation of the sound waves.
The wood box is painted with an epoxy glue to avoid the release of volatiles
that could interfere with the desired vacuum level. The speaker box is con-
nected to the glass chamber through an aluminum lip; 10 cm (4.0 in) wide,
7.6 cm (3.0 in) long with a 12.7 cm (5.0 in) circular base that covers the open-
ing in the wood box. The aluminum lip iS attached to the speaker box by
screws. It also has an outside indentation for an o-ring to hold the bottom
rubber membrane where the glass chamber is fitted. The speaker is con-
trolled by a frequency generator and a power amplifier located near the

experimental chamber.
5.2.4 Heating System

The heating system includes a heater and electrical feedthroughs. The
heater is made of a 9.0 cm (3.5 in) soldering iron placed inside 1.3 cm (1/2 in)
wide, 7.6 cm (3.0 in) long copper tube. The role of the copper tube is to
equilibrate the temperature beween both ends of the soldering iron since the

temperature distribution is far from being uniform. The heater is placed

7]

inside the glass chamber with the help of the tungsten pins. The heater hangs
down from the pins to the middle of the glass chamber by its electric wires
from one end, and by a nicrome wire that connects the tip of the soldering
iron to a different pin from the other end. Figure (5.5) Shows the heater
inside the glass chamber. Feedthroughs are needed to pass a Signal from the
inside of the chamber to the outside without interferring with the vacuum
level. The feedthrough used in this chamber are made of bulk head unions
that fit the holes of the top lid. O-rings are needed for a complete sealing.
Electric and thermocouple wires pass through the bulk head union, are sealed
by using vacuum epoxy.

5.2.5. Vacuum System

The vacuum system is best understood by referring to figure (5.6). Half
inch copper tube pieces are put together by compression fittings and ball
valves to control the gas flow in and out of the experimental chamber. One
needle valve is used to have a good control over the gas flow rate. During
evacuation of the chamber, the air inside the glass chamber should be
removed from the top gas port to minimize powders movement. The vacuum
system is connected to the chamber by red vacuum hoses. Vacuum

feedthroughs are made in a similar way to the electric feedthroughs. Copper

tube pieces are attached using compression fitting.

 

O-Ring Location

4.. ’
L.:l.‘_ Gas Inlet

[<—
7%

 

15" -

q Tungsren Pins

 

 

 

 

L

 

Gas Inlet

Figure 5 .5 Heater inside glass chamber.

73

 

 

 

 

 

 

 

 

' ' 133: I _ .
l E ‘7 . «211’ . .. .. _-.-- .’
[\sg;;~.;.._.. . -33», .
.
I .
.-.~(~ -::->*-7-:-u .\
3. . - . ‘; »‘;:: . ‘
‘w: .~:~:-:»:»x- «. 43.3...
. :x‘ .. I .‘ _ :.;.;. ..
‘3‘ . ..,.-.:.\~.¢:~..‘.:..-,.~.,,~ .
'42,; V5; 21.! .2.“
“My“: .

 

 

 

 

Figure 5 .6 Vacuum system.

74

Vacuum hoses are used to connect the top lid to the vacuum system. Figure
(5 .7) shows how the experimental chamber and the vacuum system are put

together.
5.2.4. Grounding System:

The idea of this system is to ground all conductive objects that may come
in contact with the aluminum powders. This can be accomplished by con-
necting the metal objects ( aluminum lip, tungsten pins, heater, etc. ) by a
long nicrome wire. The nicrome wire passes through one feedthrough to the

ground.
5.3. Operation of the Chamber

Once the chamber is assembled as Shown in figure 5.8 , it should be oper-

ated in the following way:

1) Have all valves closed.

2) Start the pump and open valve 1.

3) Open the needle valve.

4) Open valve 3 very Slowly until the vacuum gauge Starts reading the vac-

uum level Slowly.

75

 

 

 

 

 

 

 

 

 

 

 

 

 

r—I—r
L-J

 

 

 

 

 

 

 

 

© 0

 

 

Figure 5 .7 Operation of the chamber diagram.

 

 

76

I... .
Prat/we;

-215.

$.11 gm 93.3.“!

 

9...... ..

Figure 5.8 Picture of the actual experimental chamber.

77

5) When the vacuum gauge shows 20 in Hg, open valve 3 completly.

6) After 10 to 15 minutes, The vacuum gauge will Show satisfactory vacum
level (below 27 in Hg). At that point, close valve 3 and valve 1.

7) Open valve 2 to let argon in the chamber.

8) Wait few minutes and open valve 3 very slowly to fill the chamber with
argon.

9) After the vacuum gauge Shows atmospheric pressure, close all valves and
start the prepregging experiment.

10) During fluidization of the powders, argon can be introduced (if needed) to

the glass chamber by opening valve 4 very slowly.

Chapter 6

6. Procedure
6.1 Materials

The materials used for this experiment were prepreg tapes of nylon-
coated carbon fibers produced by the powder prepregging system at the Com-
posite Materials and Structutres Center (CMSC). The tapes are 30.5 cm
(12in) long and 2.5 cm (1 in) wide and have fiber volume fraction of 70 %.
The nylon powders that were used to coat the carbon fibers had an average
particle Size of 5.0 microns. The aluminum powders have spherical shape
and 5.5 microns in diameter. Table 6.1 gives more information about the

properties of each material.
6.2 Calibration of The Heater

The heater was calibrated by varying the voltage and measuring the max-
imum temperature and the time needed to reach that temperature. The volt-
age is varied from 80 to 110 volts. Table 6.2 shows theSe results. During the

prepregging experiment with aluminum powders, The nylon-coated fibers

78

79

 

 

 

 

 

 

 

Material / Preperty Value
=

Hercules as-4 Carbon Fibers

Diameter (microns) 8.0

Specific gravity (g/cm3) 1.80

Thermal Conducrivity (W/rn/C) 5.73

Orgasol Polyamide

Average particle size (microns) 5.0

Specific gravity (g / cm3) 1.02

Melting point (C) 175

Thermal conductivity (W/m/C) 0.22

Suface tension (ml/m2) 30.0

Aluminum powders

Average particle size 5 .5 .

Density (g/cm3) 2.69

Apparent density (g/cm3) 0.6

Chemical composition : Aluminum 99.7%
Iron 0.18%
Silicon 0.12%

 

Table 6.1 Properties of materials used during the experiment.

 

 

 

 

 

 

 

 

 

 

Voltage Tmax Time
(Volts) (C) (min)
80 169 18
90 191 16
100 220 1 5
1 10 243 12

 

 

 

Table 6.2 Results from calibration of the heater.

 

81

were heated from room temperature until 220°C for 15 minutes and are held
at that temperature (220°C) for 15 minutes.

6.3 Making of Aluminum Prepreg Tapes

An outline of all the steps that are involved in this experiment was pre-

pared and followed in this order:

1) Cut polymer prepreg tape into 30.5 cm (12 in) pieces

2) Slightly bend some prepreg pieces from the middle and lay them on top of
the heater inside the glass chamber

3) Put an aluminum cap on top of the prepreg section that is in contact with
the heater. This will help hold the prepreg in position. The prepreg tapes will
hang down from the heater to about 10 cm (4 in) from the bottom of the glass
chamber.

4) Place the bottom membrane in position on the aluminum lip with the help
of an 0-ring.

5) Put the speaker box on top of the bottom aluminum lid.

6) Take one Spoon of aluminum powders and deposit it gently on the bottom
membrane

7) Fit the glass chamber on top of the aluminum lip. Make sure that it is held
tight.

8) Place the plexiglas tube in position on the bottom aluminum lid.

82

9) Put all the vacuum hoses in position.

10) Connect the inside electric wires and hoses to the their repective vacuum
feedthroughs on the top aluminum lid and close chamber.

1 1) Start outgassing following the instructions given in the “operation of
chamber” section.

12) When the pressure inside the chamber is reduced to below 25 in Hg, open
the argon valve slowly to bring pressure to one atrn.

l3) Redo 11 and 12

14) Turn on the heater at the desired voltage for a period of time.

15) Turn off the heater.

16) Turn on the speaker to start fluidizing the aluminum powders.

17) Turn off the speaker once good powder pick up is achieved.

18) Let the powders settle down inside the chamber for about 45 minutes.
19) Open the chamber from the top and disconnect the electric wires and the
vacuum hoses from the top lid.
20) Take off the top membrane.
21) Disconnect the heater from the tungsten pins and take it out with the
prepreg tapes attached to it.
22) Put back the membrane and close the chamber.

23) Keep the chamber full of argon and leave it in a safe place.

83

6.4 Scanning Electron Microscope (SEM) Examination

Four small pieces of the aluminum prepreg tape were cut from different
places along the length of the tape. A gold coating was applied to improve
the electric conductivity for the SEM. A Hitachi S-2500C Scanning Electron
Microscope was used to examine the prepreg. Samples of the aluminum pow-

ders and the prepreg tape were observed in a Hitachi H-400 SEM.

Chapter 7

7. Results and Discussions
7.1 Experimental Results
7.1.] Vacuum Level

The main reason for designing the experimental chamber is to be able to
test the polymer powder prepreg system with fine aluminum powders in a
safe way. Assuring the safety of the experimenter and the working area can

be accomplished by obtaining an air-free atmosphere inside the chamber.

Before rtmning the experiment with aluminum powders, the chamber was
tested three times to determine the allowable vacuum level obtained by the
chamber. The ball valves and the compression fitting copper joints were
tested first separately so that any leakage can be easily detected and corrected
before connecting the vacuum system to the big plexiglas tube. The designed
fixture was pumped down from 0 to 30 inch Hg reading on a mechanical vac-
uum pressure gauge in about one second. That proved that the vacuum sys- -

tem had no significant leakage.

85

When the big chamber was connected to the vacuum system, the vacuum
gauge Showed 28.5 in Hg after about 15 to 20 minutes. This implied that the
plexiglas tube was not totally sealed. However, the vacuum level obtained by

the chamber was still sufficient to run a safe experiment.

During the prepreging experiment with aluminum powders the vacuum
level obtained was only 27 in Hg. This can be explained in relation to 3
effects:

1) A smaller Sized o-ring was used on the top aluminum cover to
facilitate opening the chamber after the end of the experiment.
2) Shorter pumping time.

3) A smaller pump was used with a slower pumping speed.

The experiment was done without any obvious powder oxidation sug-

gesting that this procedure can be safely used.

86

7.1.2 Fluidization of Aluminum Powders

The average particle Size of the aluminum powders used in this experi-
ment is between 6.5 and 4.5 microns. According to Geldart [65], these pow-
ders belong to type C as shown in figure 7.1. Type C powders are cohesive

and very hard to fluidize.

The first step in this experiment was to detemime the natural frequency
of the speaker. This was determined visually Since the membrane vibrates
with maximum amplitude at that frequency level. During the experiment,
which was run in an argon medium, the natural frequency was found to be
equal to 87 Hz. At this frequency the bed reached its maximum height (about
4 inches) and a cloud of dispersed fine aluminum powders was seen. It was
noticed that a fluidized bed with a constant particle concentration was hard to
maintain at the same frequency for more than few seconds. AS time pro-
gressed the concentration of the fluidized powders and the height of the bed
was decreasing until no fluidized powders could be seen. Caking of powders
on the surface of the glass chamber and some evidence of large clustters of

packed powders sticking to the glass wall were seen.

87

 

D-powders

C-powders

 

 

 

10

 

100 . 1000 5000 um

Figure 7.1 Various types of powders [80].

88

7.1.3 Characterization of The Aluminum Prepreg Tape

This part looks into answering some of the questions concerning the pos-
Isibility of using this polymer powder prepreg system to make metal powder
prepregs. The first step is to see if aluminum powders can adhere to the
nylon-COated carbon fibers. If so, how Significant the particle pickup is and

whether there exists a concentration profile along the bed height ?

The pregreg tape that was examined on the SEM was laid up on top of
the heater with one Side hanging down to about 3 inches above the packed
powders bed while the other side was 6 inches above the bed. Four samples
from the aluminum prepreg tape were cut. Figure 7.2 shows the location of
each sample from the prepreg tape. Two samples, called samples Aland A2,
were from closest to the packed powders, one sample B from the middle of
the same prepreg tape section, and the fourth sample C was from the bottom
of the shorter Side of the prepreg tape. The samples were mounted on the
SEM for examination. The aluminum powders were identified by using the
X-ray dispersive energy device that is attached to the SEM. Figure 7.3 (a)
Shows the SEM image of one particle that was used for identification, and fig-
ure 7.3 (b) Shows data from the X-ray dispersive energy plot that identified

the particle as aluminum.

89

 

sample
C

 

 

15 cm
\ Sample B

 

 

Figure 7.2 Location of samples on the aluminum prepreg

tape.

90

1

1 51 4 ‘3 3 313131.”

 

Figure 7.3 (a) SEM image of one particle on a carbon fiber

from sample A.

91

 

‘x-RHV: 0 - 20 keU
Live: 755 Preset: 1005 Remaining: 255
Real: 1365 L157. Dead

 

 

F

 

 

 

 

 

 

 

l

A‘*‘ -— -AAA 4
“v

 

fi' '7‘.

 

< .0 KeU 5.1 >
F3: 16K - ch 138= 192 cts
NEMI3

 

 

Figure 7.3 (b) X-ray dispersive energy analysis identifying

the powder in figure (a) as aluminum.

92

Samples A: These samples were taken from a section of the prepreg that
looked gray to the naked eye which suggested that it might have the highest
concentration of aluminum powders. SEM examination at 200 times mag-

nifigation showed a moderate powder pick up as it is Shown in figure 7.4 (a).

Sample B: This sample had very little powder pick up as it is shown in
figure 7.4 (b). The SEM also showed that the particle size was very small

(under 3 microns).

Sample C: This sample showed the highest amount of powder pick up.
Figure 7.4 (c) Shows the increase in aluminum powder concentration as com-
pared to the other two figures that were taken at the same magnification from

different location.

b) Quantitative Analysis:

This analysis was based on SEM images that were taken from each
region at high magnification. A Size distribution analysis from each region
was done. Table 7.1and table 7.2 Show these results. The number of parti-
cles per unit fiber length was determined based on SEM images taken at mag-

nifications between 500x and 1000x. Figure 7.5 are examples of the SEM

93

images that were used in this analysis. Table 7.1 gives the results of this anal-

ySiS.

lSOum

 

Figure 7.4 (a) SEM image of powder pick up in Sample A.

95

21499.2019 xeaa isium

 

7.4 (b) SEM image of powder pick up in sample B.

96

n .m.
a :1.

.l m

k. "r"

E1

J
L

1411

I;

1..

 

Figure 7.4 (c) SEM image Showing powder pick up in

sample C.

97

 

 

 

 

 

 

Number Number Number Number Number
of of of of of
Sample particle particles particles particles particles
[1-2 111111 [2-3 um] [3-4 um] l4-5 um] [> 5 um]
A 3 9 14 20 10
B 2 7 0 0 O
C 4 10 16 21 13

 

 

 

 

 

 

Table 7.1 Powder size distribution from each sample.

 

98

 

Sample

Number of
particles

Average
particle size

Standard
deviation

A
A

 

 

 

 

 

 

56 4.43 i 1.54
B 9 2.50 i 0.89
C 64 4.89 i 1.83

 

Table 7.2 Average particle Size for each sample.

 

9 9

 

 

 

 

 

 

deation of
_ samples Average
above Magnifica- number 0f
Samples acked tion partrcles/
gowder unit length
level (100 11m)
A 7.5 cm 500 15
B 12.5 cm 1000 2
C 15.0 cm 1000 21

 

 

 

Table 7.3 Summary of the quantitative analysis.

100

 

Figure 7.5 (a) SEM image from sample A at high

magnification.

101

181410 20119 81.013111 3011M

 

Figure 7.5 (b) SEM image from sample B at high

magnification.

102

21413 201:9

 

Figure 7.5 (c) SEM image from sample C at high

magnification.

103

7. 2 Discussion:
7.2.1 Safety Factor:

The vacuum level that was reached prior to running the experiment was
27 inch Hg. This value corresponds to leaving 20.2 x10'3 moles of oxygen
inside the chamber. This amount of oxygen is capable of oxidizing 0.7 g of
aluminum which can cause a pressure build up of about 2 atm. This pressure
could produce a small explosion that could be contained mostly by the cham-
ber. However, the force that is applied to the top cover from the inside is well
above the force needed to lift the lid that is attached to the chamber by an o-
ring. AS a result, the danger of exposing the aluminum powders to air after
removing the top cover still existed. Even though, the principle gas load dur-
ing outgassing is water vaporthat clings to the surfaces of the evacuated sys-
tem, and most of the air is removed at the beginning of the pumping process
which means the oxygen content would actually be lower than the value pre-

dicted by the thermodynamics calculations, the experiment was Still not

totally safe.

The partial pressure of oxygen was, therefore, reduced further by pump-

ing and introdUcing argon one more time. Using the same computation

104

method, the new value for the number of the oxygen mOles is 2.02 x10'3.
This indicated that only 0.07 g of aluminum would be oxidized and the
pressure inside the chamber would be negligeable. This way, the experiment
could be run safely. This latter procedure is an additional safety measure dur-
ing the early experiments which can be eliminated in the future to speed up

the experiment if higher vacuum level can be obtained.

7.2.2 Heating of the Prepreg:

Heating of the prepreg tape was accomplished by placing it in a close
contact with the hot surface of the copper tube. The carbon fibers would con-
duct heat along the fiber direction from the point of contact to the rest of the
prepreg tape. The copper tube was heated to a temperature 220 °C which is
above the softening point of nylon to assure that enough heat was generated
at the point of contact to heat the entire length of the prepreg layer. At this
temperature of the heater, nylon was softened and made a sticky film that
covered the carbon fibers. Upon contact, the aluminum powders were
attached to the heated tape to form an aluminum prepreg tape. SEM images

Showed a large number of aluminum powders that were adhered to the sur-

face of the nylon-coated fibers.

The adhesion of these aluminum powders can be explained by three

105

mechanisms. First, powders were attached to the fibers because of the adhe-
sives forces that resulted from the sticky nylon coating due to the difference
in the surface energies between nylon and the aluminum powders. Nylon has
much lower surface energy than both aluminum and aluminum oxide. Upon
contact, the nylon spread over the aluminum powder surface to reduce its sur-
face energy. As a result, the aluminum powders were held to the fibers by the
spreading effect of nylon. Second, adhesion was achieved by plastic defor-
mation of the softened nylon around the aluminum particles as it was shown
in many particles in some of the SEM pictures at high magnifications. This
deformation is due to the impact of the fluidized aluminum particles on the
softened nylon as the hit the carbon fibers. If the Speed of these oscillating
particles inside the chamber is known, the magnitude of this plastic deforrna-
tion can be determined. Third, the adhesion could be due to the effect of the
electrostatic forces. Aluminum powders covered with a thin layer of alumi-
num oxide may carry static charges generated from powder contact with each
other. These charged powders are attracted to the carbon fibers because of
the net difference in the Static charge density between the powders and the
fibers. The effect of these forces depends on the thickness of the oxide layer
that covers the aluminum powder surface. More work could be done to esti-
mate the minimum oxide thickness that can make a 5 micron aluminum pow-

der hold a significant static charge density.

106

SEM images from the aluminum prepreg tape (figure 7.6) Showed most
of the fibers had a smooth appearance while few had axisymmetric configura-
tion of few nylon droplets. SEM examination of one sample of the original
prepreg tape that was not heated showed very similar appearance to the
heated one. This suggests two possible mechanisms. First, the heating of the
prepreg was not effective to soften the nylon coating and change its original
Shape. This theory was not supported by some of the SEM images that
Showed a clear deformation of the nylon around some powders (arrow in fig-
ure 7.4.c), which implied that nylon was softened due to the heat generated
from the point of contact with the hot copper tube surface. Therefore, the
heating of the prepreg, though it may not have been uniform, showed that the
heating method could work to make a sticky polymer tape. The second pos-
sible explanation is that the original nylon coating of the carbon fiber was not
uniform throughout the length of the tape due to either a significant nylon
powder loss during the handling of the original prepreg tape or to a low con-
centration of nylon powders that were deposited on the carbon fibers. A
reduced powder pick up concentration is a common feature in this process
when 5 micron polymer powders are used because Of the strong effect of the

cohesive forces.

 

107

7.2.3 Fluidization of Aluminum Powders:

The decrease in the fluidized powder concentration and the bed height
can be explained as follows: The sound pressure is the buoyant force that is
responsible for lifting the powders to a certain height. The sound pressure is
proportional to the amplitude of the sound wave. Maximum amplitude is
obtained at resonance which is determined according to the initial weight of
the powders on the rubber membrane. In this experiment, it was not known
whether the initial frequency and amplitude were close to the optimum values
since it was hard to visually set the resonance conditions through two trans-
parent chambers. In addition, as the weight of the packed bed was decreased
because of the powder caking on the glass wall and particle pick up by the
fibers, the initial conditions for a maximum sound pressure were changed.
Therefore, the particle concentration and the fluidized bed height were
decreasing with time as the experimental conditions were continuing to move

away from the original conditions.

Another big factor in the rapid decrease of the fluidized bed concentra-
tion is that the initial powder weight was less than 1 g. This small amount of
powder added to the difficulty of setting the resonance conditions since it was
hard to see the change in the amplitude of the vibrating rubber membrane

imposed by the weight of less than lg of aluminum powders. That means,

 

108

the changes in the amplitude were SO small that reaching the maximum value
at the natural frequency was very hard to see. Moreover, the small amount of
powder was reduced rapidly by powder caking on the glass walls of the
chamber. This did not allow enough time for the fluidization of the alumi-
num powders at the optimum conditions if they were ever reached during the

course of the experiment.

Agglomeration is a very common phenomenon with type C powders
mainly because of the effect of the interparticle and electrostatic forces. Van
der Waals forces act on particles that are smaller than 30 microns and less
than 100 angstroms apart. In polymer powders, electrostatic forces have a
big effect on the cohesiveness of small particles which can explain the
agglomeration problem that is frequently seen with these powders. The
degree of agglomeration depends on the charge to mass ratio which increases

with decreasing particle size.

In this experiment, the average particle size of the aluminum powders
that were used was 5 .5 microns. This makes the role of the interparticle
forces significant and must be taken into consideration. SEM examination of
a sample of the aluminum powders that did not take part in the fluidization
experiment showed that a large number of small size particle (l to 3 microns)

were attached to bigger particles (5 to 7 microns). It also showed few

109

clusters of particles that were about 20 to 30 microns. This proved that the
Van der Waals forces have a strong effect on the cohesiveness of these alumi-
num powders even before the fluidization process. On the other hand, the
SEM images of the aluminum prepreg tape showed that small particles were
still attached to each other or to bigger particles, but no large clusters were
seen attached to the fibers after the fluidization. This can be explained by two
reasons. First, the large particles stayed in the bottom of the bed and did not
come in contact with the fibers because of their larger mass. This means that
the sound pressure inside the chamber was not strong enough to overcome '
the gravitational forces caused by the weight of the large clusters, which
could imply that the fluidization of aluminum particles of about 20 microns in
size would not be possible with this process. Second, the absence of large
clusters of particles on the fiber surface in the aluminum prepreg tape could
be due to the fact that the sound pressure was capable of breaking these
clusters into smaller ones at the point of contact of two big particles since the
interparticle forces would be less effective. This could imply that the sound

pressure was partially successful in overcoming interparticle cohesive forces.

In the aerosolizer, smaller particles oscillate with larger amplitude. This
increases the distance travelled by each particle and therefore increases the
chances of interparticle collisions. During the collisions of small particles,

the interparticle distance is at minimum which makes the effect of the

 

110

interparticle forces even stronger. As a result, the interparticle cohesive
forces force the smaller particles to stick to each other or to bigger particles.
The effect of these forces is so strong that the sound pressure can not break
them apart. Therefore, in this case the aerosolizer could contribute to the for-
mation of small clusters. This effect is not obvious from this experiment
Since there was no clear difference in the number of small clusters (less than

10 microns) between the powders that were fluidized and those that did not.

The big parts of packed aluminum powders (1 to 2 m) that were stuck

 

to the glass wall can be explained by their inactive role during fluidization.
The sudden impact of the sound pressure lifted these clusters above the mem-
brane but they did not take part in the fluidization process Since they came in
contact with the wall of the small diameter chamber where they were
attached due to electrostatic forces. The glass wall could hold a static charge
and exert attractive forces on the aluminum powders. These aluminum pow-
ders could also have a small charge density from the thin aluminum oxide
layer formed on the surface of the aluminum powders. Therefore, electro-

static force are responsible for the wall caking.

There was no evidence of large clusters (1 to 2 mm) on the bottom of the
membrane which is different from the behavior of small polymer powders.

This could be related to the weaker effect of the electrostatic forces on the

111

aluminum powders, which implies that the sound pressure can overcome the
electrostatic forces between the aluminum particles because of the small
static charge density. This suggests that the aerosolizer has better fluidization

performance on metal powders.

7.2.4 Powder Pick up:

A visual examination Showed that the aluminum powder pick up was

not uniform throughout the prepreg tape. For example, region A looked gray

 

m“

which meant that it had the highest pick up, while region B looked the same
way as it did before fluidization with the aluminum powders. SEM images
showed some evidence that the powder pick up varied along the width of the
fluidized bed rather than the height. For example, region A and C were at dif-
ferent bed heights but had Similar powder pick up concentration in contrast to
region B which had very little pick up even though it was at a bed height
between A and C. This can be explained by three reasons. First, the amount
of the aluminum powders that was used was very small (less than 1 g). In
addition the packed powder bed on the bottom membrane was off centered

which gave location C the best powder trajectory toward the fibers.

Second, the average particle Size varied with the bed height. In table 7.1,

the smallest particles were observed near the center of the column in location

112

B, while region A and C which were close to the glass wall had an average
particle size between 4 and 5 microns. This could be related to the fact that
most of the powders were attracted to the glass wall because of the electro-
static effect. In the middle in the column, the smaller particles were seen
because the sound pressure was strong enough to keep them away from the
glass wall by making them oscillate in the middle of the bed. The sound
pressure has a stronger effect on these small particles because of their
reduced weight, which could suggest that the sound pressure is the dominant

force inside the bed for the smaller particles.

Third, SEM figures Showed that there was a significant powder loss dur-
ing handling after the prepregging experiment. Figure 7.7 shows some loose
powder particles on the specimen holder. This suggests that the adhesion of
the aluminum powders to the nylon-coated fibers was not strong. This can be
explained by at least two theories. First, the heating of nylon was not uni-
form throughout the entire prepreg tape. This resulted in some regions where
nylon was no softened enough to make the surface energy or the plastic
deformation mechanisms work to create a strong bond between nylon and the
aluminum powders. The powder loss occurred only in the regions where the
adhesion was achieved by electrostatic forces. This means that during han-
dling, the aluminum powders fell off the prepreg tape because the electro-

static forces alone were not strong enough to hold the aluminum powders to

 

113

the fibers.

 

114

 

Figure 7.6 SEM image showing the apperance of the heated

nylon coated fibers.

 

Figure 7.7 SEM image Showingpowder loss.

Chapter 8

8. Conclusions
Several conclusions can be drawn from this research:

1) The thermodynamics approach is a useful method for estimating the dan-

ger associated with the fluidization of fine aluminum powders. The potential
temperature and the pressure increase from oxidation inside the experimental
chamber can be determined from the vacuum level and the residual gas com-

position obtained during outgassing.

2) Heating nylon-coated carbon fiber prepreg tapes to a temperature above
the softening point of nylon created a sticky polymer host for fine aluminum
powders. The adhesion of aluminum particles to the carbon fiber was
achieved by making nylon serve in the role of a polymer binder. The adhe-
sion was not perfect in some regions, but it was potentially useful mecha- .

nism.

3) The fluidization of fine aluminum powders was successful by using the
acoustic energy coming off a speaker through a rubber membrane. The

agglomeration effect, frequently encountered in this process with fine

116

117

polymer powders, was not seen with the small aluminum particles that were
used during this Study. This is probably due to the weaker effect of the elec-
trostatic forces on these metal powders. The aerosolizer is more efficient
with aluminum powders which can be helpful if smaller particles (1 -3 um)

are to be considered.

4) The concentration of the fluidized powders was not uniform throughout

the bed. This phenomenon still needs to be investigated.

5) This study has proven that the powder process can be used to make fine

aluminum powder prepreg tapes.

More work is still needed before making metal-matrix composite materi-

‘ als by using this process. Some of this future work may include:

- Increase the allowable vacuum level inside the experimental chamber by
improving the seal at the top cover of the plexiglas tube. This will speed up

the testing process.

- Use prepreg tapes that are made from higher concentration of nylon pow-
ders. This will improve the quality of the polymer coating of the carbon

fibers.

 

118

- Use spread carbon fibers without nylon to see if the electrostatic forces can
make the aluminum powders stick to the carbon fibers. This will help charac-
terize the different forces that are active during aluminum powder coating of

carbon fibers.

- Investigate the aluminum oxide layer that covers the aluminum powder sur-
face. This will Show whether this oxide layer is thick enough to provide an

insulating coating for the aluminum powders.

- Increase the amount of the aluminum powders that is used during the
prepregging experiment (from less than 1 g to about 3 to 5 g). This will help

characterize the fluidization properties of the aluminum powders.

- Use two to three layers during each experiment to study the concentration of
powders inside the bed. The use of more layers inside the chamber at the
same time will help characterize the powder pick up in different locations

across the width of the fluidized bed.

- Use aluminum foil to cover some the glass chamber near the bottom where
most of the caking occurred. This will help reduce the rapid decrease in the

initial powder weight.

119

- Smaller aluminum powders (1-3 pm) should be investigated to see the
effect of the particle size on the bed height and the concentration of powders
in the aerosol. In addition, the use of some lighter metal powders such as
magnesium should be tried since a decrease in the powder density can help

the bed height and the control of powder concentrations.
- Investigate the use of a different binder that has a low melt viscosity.

- The prepreg tapes should be weighed before and after experiments where
the powder pick up and the volume fractions of the matrix and the fiber are to

be determined.

- Finally, the consolidation of several layers of aluminum prepreg tapes
should be the next step to make testable metal-matrix composite materials.
The layers should be heated in a vacuum fumace to burn ofl’ the nylon binder
and to drive off the adsorbed gases on the aluminum powder surface. Heat-
ing a mixture of the same nylon and aluminum powders that were used in this

experiment at 400 °C for 5 min reduced the nylon content to near 100%.

BIBLIOGRAPHY

BIBLIOGRAPHY

[l] Divecha, A.P., Fishman, S.G., & Karmarkar, J. Met., 2, (1981), 12.
[2] Chawla, K.K., Composite Materials, Springer-Verlag, New York, 1987.

[3] Taya, M. & Arsenault, R..l., Metal Matrix Composite, Pergamon Press,
1989.

[4] Kreider, K.G., Metallic Matrix Composites, Vol. 4, Academic Press,
New York, 1974.

[5] Mehan, R.L. & Bolon, R.B., J. Mater. Sci., 13, (1979), 2471.

[6] Chou, T.W., Kelly, A., & Okura, A., Composites, l_6_, ( 1985), 187.
[7] Arsenault, R.J., Mater. Sci and Eng., 65, (1984), 171.

[8] Webster, D., Metall. Trans., m, (1982), 1511.

[9] Okura, A., Trans. JSCM, p, (1985), 10.

[10] Christian, J.L., Forest, J .D., & Weisinger, M.D., Met. Prog., 91, No 5,
(1970), 113.

[11] Lewis, C.F., J of Met., 5, (1986), 33.
[12] Donomoto, T., Miura, N., & Miyake, N., “Ceramic Fiber Reinforced

Piston for High Performance in Diesel Engine,” SAE Technical Paper
No. 83052, (1983).

120

 

121

[13] Harrigan, W.C. Jr., Metal Matrix Composites: Processing and
Interfaces, Academic Press, (1991), 14.

[14] Bhagat, R.B., Metal Matrix Composites; Precessing and Interfaees,
Academic Press, (1991), 43.

[15] Mortensen, A., Solidification of Metal M atrix Compesites, TMS,
(1989), 1.

[l6] Rohatgi, P., J of Met., 4, (1991), 10.

[17] Rohatgi, P.K., Asthana, R., & Yarandi, F., Selidification of Metal Matrix
Cemposites, 51.

[18] Mortensen, A., Comie, J .A., & Flemings, M.C., J. of Met., _2_, (1988),
12.

[19] DeLannay, F., l of Mater. Sci., 2, (1987), 1.

[20] Defay, R., & Everett, D.H., Surface Tension and Absorption,
Lougrnaus, London, (1966). .

[21] Maruyama, B., & Rabenberg, L., Interfaces in Metal Matrix
Cempesites, Dhingra, A.K., & Fishman, S.G., (eds), AIME, (1986),
233.

[22] Long, J., & Sykes, K.W., Proc. R. 800., A215, (1952), 100.

[23] Maruyama, B., Ohuchi, F.S., & Rabenberg, L., Mat. Res. Soc. Symp.
Proc., Vol. 170, (1990).

[24] Okura, A., & Motoki, K., Comp. Sci. Tech., 2_4, (1985), 243.
[25] Khan, I.H., Metall. Trans., 2A, (1976), 1281.

[26] Zhang, G.D., Feng, S.R., & Carnie, T.A., Centrolled Intetphases in
Compesite Materials, Ishida, H., (ed.), Elsevier Science, (1990), 343.

[27] Li., Q., Blucher, T.T., & Comie, J .A., Centrelled Intetphases in .

122
Composite Materials, Elsevier Science, (1990), 131.
[28] Chemyshova, T.A., & Kobeleva, L.I., J. Mat. Sci., E, (1985), 3524.
[29] Baker, S.J., & Bonfield, W., J. Mat. Sci., Q, (1978), 1329.
[30] Gupta, v., Argon, A.S., & Comie, l.A., J. Mat. Sci., a, (1989), 2031.

[31] Scott, V.D, Trumper, R.L., & Yang, M., Comp. Sci. Tech., 42, (1991),
25.

[32] Aggour, L., et al., Thin Solid Films, fl, (1977), 97.

[33] Bouix, J ., et al., Less Comp. Met., it], (1986), 83.

[34] Baker, A.A., Mat. Sci. Eng., l_7_, (1975), 177.

[35] Kulkami, A.G., et al., J. Mat. Sci., 1_4_, (1979), 592.

[36] Lavemia, E.J., & Grant, NJ, Mat. Sci. Eng.,_28_, ( 1988), 381.
[37] Zaat, J .H., Ann. Rev. Mat. Sci., Q, (1983), 9.

[38] Amateau, M.F., J. Comp. Mater., [L0, (1976), 279.

[39] Ignatowitz, E., Aluminum, 50, (1974), 334.

[40] Miyase, A., & Piekarski, K., J. Mater. Sci., &, (1981), 251.
[41] Miyase, A., & Piekarski, K., J. Comp. Mater., fl, (1980), 160.
[42] Harper, J .M.E., & Cuomo, J .J ., Ann. Dev. Mater. Sci., L3, (1983), 413.

[43] Pepper, R.T., & Kendall, E.G., US Patent No. 3770488, November 6,
1973.

[44] Derby, B., & Wallach, E.R., J. Met. Sci. 18 (1984), 427.

,_,

123

[45] Calvo, F.A., Orena, A., & Gomez, J. M., J. Mater. Sci., 23, (1988),
2273.

[46] Guo, Z.X., & Ridley, N., Mater. Sci. Tech., 3, (1987), 945.

[47] Garrnoug, G., Paton, N.E., & Argon, A.S., Metall. Trans., 6A, (1975),
1269.

[48] Wilkinson, D.S., & Ashby, M.F., Acta Metall., 2_3, (1975), 1277.

[50] Islam, M.D., & Wallace, W., Adv. Mat. Manufact. Process, 3(1), (1988),
1.

[51] Everett, R.K., & Arsenault, R.J., M 1M t ' om .ite.' Pr cessin
m Interfaces, Academic Press, (1991), 20.

[52] Everett, R.K., Metal Matrix Cemposites; Processing apt! Interfaees,
Academic Press, (1991), 29.

[53] Paultnoi, K.I., & Chubarov, V.M., Stptcmre and Prepegies of
Composite Materials, (1979).

[54] Doble, G.A., & toth, 1]., Proc. ICCM-I, AIME, (1975), 775.

[55] Lioyd, D.J., J. Mat. Sci., l_9_, (1984), 2488.

[56] Gigerrenzer, H., et al., Proc. ICCM, AIME, Warrendale, (1978), 175.
[57] Iyer, S.R., Ph D Thesis, Michigan State University, 1991.

[58] Palmer, K.N., Particle Characterization in Technology, Vol. 1, CRC
Press, (1984), 135.

[59] Hertzberg, M., & Cashdollar, K.L., Introduction to Dust Explosions,
ASTM, Philadelphia, (1987), 5.

[60] Aluminum Association Handout, Recommendations for Storage and
Handling pf Alminm Pewders apgl Paste, TR-2.