.r r
.o. r
.o
.

 

3:35;';.‘ * ,

 

y

 

. .
.131!

 

w ..
$5.92.“
v 1w MS;

Alb

v.14.

1...... u...

flaunt
was. .5. x"
.3 11.1.3. flaw MP r

#5.... 2..
gram”
. a...

r.
1.1.
{:9}: .9

 

 

 

it 5.73.3! {infirm um.

i am IIIIIIIIIII||IIIIIIIIIIIIIIIIII IIIIIIIIIIIIIIIIIIIIII

sf. . 1293017718416

LIBRARY

Michigan State
Unlverslty

 

'—T

 

 

 

This is to certify that the

dissertation entitled

MICROWAVE PULTRUSION OF UNIDIRECTIONAL REINFORCED EPOXY
COMPOSITES USING SINGLE FREQUENCY AND VARIABLE FREQUENCY
PROCESSING TECHNIQUES

presented by

Aaron C. Smith

has been accepted towards fulfillment
of the requirements for

M.S. degree in Chemical Ergineering

 

 

Wat-MA

Major profegsor

Datefmfi’ Z271 Ifff

MS U is an Affirmatiw Action/Equal Opportunity Institution 0-12771

PLACE IN RETURN BOX to remove this checkout from your record.
To AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

I DATE DUE

DATE DUE

DATE DUE

 

ItE 1 I a.“

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1/99 chIRCID‘uDquG—p.“

MICROWAVE PULTRUSION OF UNIDIRECTIONAL REINFORCED EPOXY
COMPOSITES USING SINGLE FREQUENCY AND VARIABLE FREQUENCY
PROCESSING TECHNIQUES
By

Aaron C. Smith

A THESIS
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
MASTER OF SCIENCE
Department of Chemical Engineering

1 999

ABSTRACT
MICROWAVE PULTRUSION OF UNIDIRECTIONAL REINFORCED EPOXY
COMPOSITES USING SINGLE FREQUENCY AND VARIABLE FREQUENCY
PROCESSING TECHNIQUES
By

Aaron C. Smith

The pultrusion process is a continuous manufacturing method, which can be used
to produce reinforced plastics with a constant cross-sectional area through the length of
the product. The continuous microwave processing technique is highly desired for
processing large composite materials, which need even and fast heating such as long pipes
or panels. The general advantages of using microwave technology include: decreased
processing times, better control of temperature profiles within the composite material, and
enhanced mechanical properties.

The objective of this project was to obtain experimental data for microwave
pultrusion processing. Heating and curing profiles of the pultruded material were studied
at different processing conditions. The fixed frequency pultrusion tests were
accomplished using the best heating mode for the prepreg system at a certain processing
speed and input power. The relation between microwave processing rate, input power,
and extent of cure was studied by finding the optimum operating conditions. Mechanical
tests, such as flexural and tensile strength measurements, were conducted. A variable
frequency power source was used to achieve a more even temperature distribution across
the prepreg. The effects of fiber reinforcement on the efliciency of microwave heating

were studied. Concepts for an on—line monitoring system were explored.

To my wife, Jodi, for her unconditional love and encouragement throughout the duration
of this project. Her understanding for irregular hours and late nights did not waver. I am

truly blessed to have her in my life.

iii

ACKNOWLEDGMENTS

The author acknowledges Sigma Technology and Michigan State University for
their financial support. Appreciation is extended to Dr. Martin C. Hawley for his support
during this research project, for the intellectual discussions, and for a deeper
understanding of chemical engineering.

The author is gratefiIl for the CMSC and Chemical Engineering staffs. Their
kindness and hard work made graduate school less stressfiil. Thanks must be extended to
Mr. Michael Rich for his help with the CMSC instruments and for leading me to a great
career. Appreciation is given to Amy McMillan for preparing samples and charts and for
running analytical tests. The fellow colleagues are recognized for their assistance in the
laboratory: Mr. Phil Culcasi, Mr. Colin Durham, Mr. Bao Fu, Dr. Yong-Joon Lee, and

Mr. Yunchang Qiu. Thanks for the laughs and the deep office conversations.

iv

TABLE OF CONTENTS

LIST OF TABLES ................................................................................ viii

LIST OF FIGURES ................................................................................ ix

KEY TO SYMBOLS ............................................................................. xiii

INTRODUCTION .................................................................................. 1

CHAPTER 1

BACKGROUND .................................................................................... 2
1.1. Microwave Processing ............................................................... 2
1.2. Microwave Heating Mechanism ................................................... 3
1.3. Microwave Heating Advantages ................................................... 5
1.4. Microwave Applicators ............................................................. 6
1.5. Research Scope and Goals .......................................................... 8

CHAPTER 2

APPLICATION OF CONTINUOUS MICROWAVE PROCESSING .................. 10
2.1. Introduction ....................................................................... 10
2.2. Pultrusion ......................................................................... 11
2.3. Experimental Setup .............................................................. 14

2.3.1. Cavity .................................................................. 15
2.3.2. Calibration of Rollers ................................................ 17

2.4. Materials: Preparation and Discussion ....................................... 17
2.5. Goals .......................... . ................................................... 19

CHAPTER 3

CAVITY CHARACTERIZATION ............................................................ 20
3.1. Finding Electromagnetic Modes ................................................ 20
3 .2. Temperature Measurement ...................................................... 22
3.3. Heating Studies At Various Modes .............................................. 23
3.4. Preliminary Heating Studies in TE112 Mode .................................... 26
3.5. Conclusions ........................................................................... 27
CHAPTER 4
FIXED FREQUENCY PULTRUSION EXPERIMENTS ................................... 29
4.1. Background .......................................................................... 29
4.2. Experimental Setup ................................................................. 29
4.3. Variation in Pulling Speed ......................................................... 30
4.4. Variation in Input Power ........................................................... 31
4.5. Extent of Cure Measurement ...................................................... 31
4.6. Mechanical Properties Measurement ............................................. 38
4.6.1. Tensile Strength ......................................................... 38
4.6.2. F lexural Strength ........................................................ 41
4.7. Conclusions ......................................................................... 45
CHAPTER 5
VARIABLE FREQUENCY PULTRUSION EXPERIMENTS ........................... 46
5.1. Background On Variable Frequency Processing .............................. 46
5.2. Frequency Sweep Analysis ....................................................... 48
5.3. Input Power vs. Frequency ....................................................... 50

5.4. Temperature Measurement and Heating Studies Part I: 6-ply Graphite/Epoxy
Prepreg, Four Frequencies For V.F. Processing ...................................... 51

5.5. Heating Studies Part II: l8-p1y Graphite/Epoxy Prepreg, Change In Cavity
Dimension ................................................................................. 52

5.6. Heating Studies Part III: 11-ply Graphite/Epoxy Prepreg, Four Frequencies
and Time Intervals ........................................................................ 56

5.7. Heating Studies Part IV: 11-ply Graphite/Epoxy Prepreg, Four

Frequencies and Variation In The Time Interval ...................................... 60
5.8. Extent of Cure Measurement ...................................................... 62
5.9. Conclusions .......................................................................... 63
CHAPTER 6
FIBER REINFORCEMENT EFFECTS ON MICROWAVE PROCESSING
EFFICIENCY ...................................................................................... 64
6.1. Material Identification and Purpose of Study .................................... 64
6.2. Experiments .......................................................................... 65
6.3. Results and Discussion .............................................................. 67
6.4. Conclusions ........................................................................... 70
CHAPTER 7
ON-LINE MONITORING ....................................................................... 71
7.1. Possible On—line Monitoring Techniques ......................................... 71
7.2. Pultrusion Model for On-Line Monitoring ....................................... 73
7.3. Conclusions on On-Line Monitoring ............................................. 79
CONCLUSIONS .................................................................................. 8O
RECOMMENDATIONS ........................................................................ 82
BIBLIOGRAPHY ................................................................................. 86

vii

LIST OF TABLES

Table 1. Calibration of Rollers To Evaluate Possible Processing Speeds .................. 17
Table 2. Heating Modes for the Cavity with Teflon Die ..................................... 24
Table 3. Heating Modes for the Cavity with Teflon Die and Graphite/Epoxy Prepreg .. 24
Table 4. Heating Modes for the Cavity with Teflon Die and Glass/Epoxy Prepreg ...... 24

Table 5. Extent of Cure Measurement For Preliminary Heating of Six-Ply Graphite/Epoxy
Composite in TE“; Mode With Input Power Varied .......................................... 27

Table 6. Extent of Cure of 11-ply Prepreg With Variable Frequency Processing 62

Table 7. Heating Modes for the Cavity with Teflon Die and Glass/Epoxy Prepreg ...... 65

viii

LIST OF FIGURES

Figure 1. Microwave Pultrusion Experimental Apparatus at MSU ........................ 15
Figure 2. Schematic of Microwave Applicator at MSU ....................................... 16
Figure 3. Entrance To Microwave Cavity Through Teflon Die ............................. 16
Figure 4. Frequency vs. Cavity Length for 7 .00" +/- 0.001 " (TE modes) ................ 21
Figure 5. Frequency vs. Cavity Length of 7.00" +/- 0.001" (TM modes) ................. 22

Figure 6. Teflon Die Used for Heating Studies and Microwave Pultrusion Processing .. 23

Figure 7. Temperature Versus Time Results for Heating Studies of Six-Ply Hercules

AS4/3501-6 Prepreg at 100 Watts and 0.787 fi/rnin .......................................... 25
Figure 8. Temperature Versus Time Results for Heating Studies of Stationary Six-Ply
DA409/E250 Prepreg At Approximately 50 Watts ........................................... 25
Figure 9. Results For Preliminary Heating of Six-ply Graphite/Epoxy Composite In TE, ,2
Mode At Various Input Powers ................................................................. 26
Figure 10. Effect of Input Power on Extent of Cure ........................................ 34

Figure 11. Trendline for Extent of Cure as Function of Input Power (40-70 Watts) ... 34
Figure 12. Trendline for Extent of Cure as Function of Input Power (70-100 Watts) ...35
Figure 13. Effect of Pulling Speed on Extent of Cure ....................................... 36

Figure 14. Trendline For Extent Of Cure As Function Of Pulling Speed (0.281-0.562
fi/min) ............................................................................................... 36

Figure 15. Trendline For Extent Of Cure As Function Of Pulling Speed (0562-0733
fi/min) ................................................................................................ 37

Figure 16. Trendline For Extent Of Cure As Function Of Pulling Speed (0733-1. 17
fi/rnin) ................................................................................................ 37

Figure 17. Tensile Strength of Six-ply Graphite/Epoxy Prepreg Tape with A Pulling
Speed of 0.281 filmin and Variable Input Power .............................................. 39

ix

Figure 18. Tensile Modulus of Six-ply Graphite/Epoxy Prepreg Tape with A Pulling
Speed of 0.281 fi/rnin and Variable Input Power ............................................. 39

Figure 19. Tensile Strength of Six-ply Graphite/Epoxy Prepreg Tape with An Input
Power of 100 Watts and Variable Pulling Speed ............................................. 40

Figure 20. Tensile Modulus of Six-ply Graphite/Epoxy Prepreg Tape with An Input
Power of 100 Watts and Variable Pulling Speed .............................................. 40

Figure 21. Flexural Strength of Six-ply Graphite/Epoxy Prepreg Tape with A Pulling
Speed of 0.281 ft/min and Variable Input Power .............................................. 42

Figure 22. Flexural Modulus of Six-ply Graphite/Epoxy Prepreg Tape with A Pulling
Speed of 0.281 fi/min and Variable Input Power .............................................. 42

Figure 23. Flexural Strength of Six-ply Graphite/Epoxy Prepreg Tape with An Input
Power of 100 Watts and Variable Pulling Speed .............................................. 43

Figure 24. F lexura] Modulus of Six-ply Graphite/Epoxy Prepreg Tape with An Input
Power of 100 Watts and Variable Pulling Speed .............................................. 43

Figure 25. Six Ply Hercules AS4/3 501-6 Heated in TE”: mode at 100 Watts and 0.281
fi/min .................................................................................. 47

Figure 26. Six Ply Hercules AS4/3 501-6 Heated in TE,” mode at 100 Watts and 0.281
ft/min .................................................................................. 48

Figure 27. Percentage of Reflected Power At 2.00 to 3.00 GHz Processing Frequencies
For A Cylindrical Brass Cavity With Lc = 90 mm And (19 = 15.89 mm And Containing Six-
ply Hercules AS4/3501-6 Prepreg ............................................................... 49

Figure 28. Percentage of Reflected Power At 2.00 to 3.00 GHz Processing Frequencies
For A Cylindrical Brass Cavity With Lc = 135 mm And dp = 15.89 mm And Containing
Six-ply Hercules AS4/3501-6 Prepreg .......................................................... 50

Figure 29. Incident Power Versus Processing Frequency for Lambda Variable Frequency
Power Source (REF) .............................................................................. 51

Figure 30. Six-Ply Hercules AS4/3 501-6 Heated With Variable Frequency Processing
(2.4285 GHz, 2.444 GHz, 2.594 GHz, and 2.6140 GHz - 100 ms each) At An Average Of
70 Watts And A Pulling Speed of 0.281 fi/min ................................................ 52

Figure 31. Percentage of Reflected Power At 2.00 to 3.00 GHz Processing Frequencies
For A Cylindrical Brass Cavity With Lc = 90 mm And dp = 15.89 mm And Containing
Eighteen-ply Hercules AS4/3501-6 Prepreg ................................................... 53

Figure 32. Percentage of Reflected Power At 2.00 to 3.00 GHz Processing Frequencies
For A Cylindrical Brass Cavity With Lc = 90 mm And dp = 22.90 mm And Containing
Eighteen-ply Hercules AS4/3501-6 Prepreg ................................................... 54

Figure 33. Percentage of Reflected Power At 2.00 to 3.00 GHz Processing Frequencies
For A Cylindrical Brass Cavity With Lc = 93 mm And (IP = 22.90 mm And Containing
Eighteen-ply Hercules AS4/3 501-6 Prepreg ................................................... 55

Figure 34. Eighteen-Ply Hercules AS4/3501-6 Heated With Variable Frequency
Processing (2.1173 GHz, 2.4162 GHz, and 2.7014 GHz - 1000ms each) and A Pulling
Speed of 0.281 fi/min .............................................................................. 56

Figure 35. Eleven Ply Hercules AS4/3 501-6 Heated With Variable Frequency Processing
(2.1173 GHz - 1000ms, 2.4162 GHz - lOms, 2.6328 GHz - 1000 ms, 2.6809 GHz -
1000ms) and Using A Pulling Speed Of 0.281 fi/min ......................................... 57

Figure 36. Eleven Ply Hercules AS4/3501-6 Heated With Variable Frequency Processing
(2.4162 GHz - 1000 ms, 2.4281 GHz - 1000 ms, 2.6328 GHz - 5000 ms, 2.6809 GHz -
5000 ms) and Using A Pulling Speed Of 0.281 fi/min ........................................ 58

Figure 37. Eleven Ply Hercules AS4/3501-6 Heated With Variable Frequency Processing
(2.4162 GHz - 2000 ms, 2.4281 GHz - 2000 ms, 2.6328 GHz - 5000 ms, 2.6809 GHz -
5000 ms) and Using A Pulling Speed Of 0.281 ft/min ........................................ 59

Figure 38. Eleven Ply Hercules AS4/3 501-6 Heated With Variable Frequency Processing
(2.4162 GHz - 2000 ms, 2.4287 GHz - 2000 ms, 2.6328 GHz - 7000 ms, 2.6809 GHz -
7000 ms) and Using A Pulling Speed of 0.281 ft/min ......................................... 59

Figure 39. Eleven Ply Hercules AS4/3 501-6 Heated With Variable Frequency Processing
(2.4162 GHz - 2000 ms, 2.4287 GHz - 1000 ms, 2.6328 GHz - 5000 ms, 2.6809 GHz -
5000 ms) and Using A Pulling Speed of 0.281 ft/min ......................................... 61

Figure 40. Eleven Ply Hercules AS4/3501-6 Heated With Variable Frequency Processing
(2.4162 GHz - 2000 ms, 2.4287 GHz - 1500 ms, 2.6328 GHz - 5000 ms, 2.6809 GHz -
5000 ms) and Using A Pulling Speed of 0.281 fi/min ......................................... 61

Figure 41. Temperature Versus Time Results for Heating Studies of Stationary Six-Ply
DA490/250 Prepreg At Approximately 50 Watts ............................................. 66

Figure 42. Stationary Six-ply Glass/Epoxy Prepreg Heated in TEm Mode At 57 Watts
for Two Minutes ................................................................................... 67

Figure 43. Centerline Temperature for Six-ply Graphite/Epoxy Prepreg Heated in TE”;
Mode At 50 Watts and 0.281 fi/min ............................................................ 68

Figure 44. Six-ply Glass/Epoxy Prepreg Heat in TE”; Mode At 50 Watts and 0.281
fi/min ................................................................................................ 68

Figure 45. Stationary Six-ply Glass/Epoxy Prepreg Heated in TE”; Mode At 85 Watts
for Five Minutes .................................................................................... 69

Figure 46. Centerline Temperature for Six-ply Graphite/Epoxy Prepreg Heated in TE, ,2
Mode At 80 Watts and 0.281 ft/min ............................................................ 70

xii

KEY TO SYMBOLS

or [=] extent of cure

B [=] ratio of initial hardener equivalents to epoxide equivalents
8 [=] penetration depth

6 [=] partial derivative

80 [=] permitivity of free space (8.854 x 10'” F/m)
8’ [=] dielectric constant

8* [=] efi‘ective complex perrnitivities

e” [=] effective dielectric loss factor

ed” [=] loss factor due to the dielectric

f [=] frequency (Hz, MHz, or GHz)

A [=] wavelength (cm)

no [=] permeability of free space

11’ [=] permeability of material

w[=]2*1t*f (HZ)

1: [=] 3.14159

p [=] density (g/cm3)

O' [=] electrical conductivity (S/m)

of“ [=] ultimate strength of fiber (MPa)

on, [=] ultimate strength of composite (MPa)

Cmfu [=] matrix stress at the onset of fiber cracking (MPa)

xiii

tanOe [=] dielectric loss tangent

0 [=] angular direction in cylindrical coordinate system
c [=] speed of light (3.00 x 108 m/S)

°C [=] Celsius, a unit of temperature

cm [=] centimeter, a unit of length

cos [=] trigonometric cosine firnction

CF; [=] heat capacity of fiber (J/g_K)

Cpm [=] heat capacity of matrix (J/g__K)

D [=] diameter of brass cavity (cm or in)

dp [=] coupling probe depth (m)

DSC [=] Differential Scanning Calorimetry

E [=] electric field vector quantity

| E... I [=] magnitude of electric field inside the material
I, [=] 2.45 x 109 Hz

ft [=] feet, a unit of length

g [=] gram, a unit of weight

GHz [=] Gigahertz (109 Hz)

H [=] magnetic field vector quantity

I Hm I [=] magnitude of magnetic field

H [=] rate of heat generation per unit weight by the polymerization reaction (W/g)

H, [=] heat of reaction
Hm [=] heat of reaction for cured material (J/ g)

H... [=] heat of reaction for uncured material (J/g)

xiv

Hz [=] hertz, a unit of frequency

in [=] inch, a unit of length

j [=] imaginary number

I [=] joule, a unit of energy

J1(x) [=] Bessel Function

J ’1(x) [=] Henckel Function

k [=] thermal conductivity (W/cm_K), or kinetic rate constant
°K [=] Kelvin, a unit of temperature

I [=] number of firll-period variations of E1r with respect to 0
lbf [=] pound force

Lc [=] cavity length (cm)

Lp [=] coupling probe depth (m)

m [=] meter, a unit of length, or
number of half-period variations of E9 with respect to r

MHz [=] megahertz (10‘5 Hz)

rrrin [=] minute, a unit of time

mm [=] millimeter, a unit of length

MPa [=] megapascal, a unit of pressure or force

ms [=] millisecond, a unit of time

N [=] number of half-period variations of Er with respect to 2
P [=] microwave input power (W)

P,“ [=] power absorption per unit volume (W/cm3)

Pm’ [=] rate of heat generation per unit volume by absorbed microwave energy (W/cm3)

r [=] radius, or radial direction

R [=] universal gas constant (8.314 J/mol__K)
RP. [=] reflected power (W)

s [=] second, a unit of time

sec [=] second, a unit of time

sin [=] trigonometric sine firnction

t [=] time

T [=] temperature

T. [=] ambient temperature

Tc [=] curing temperature

TE [=] transverse electric

TM [=] transverse magnetic

T.W.T. [=] traveling wave tube

Vz [=] pulling speed

W [=] watts, a unit of power, or energy per time
Wf [=] weight fraction of fiber in composite
wml [=] weight fraction of matrix in composite
x [=] plane across width of prepreg

x1“. [=] mth root of Henckel Function for the TE-modes or
mth root of Bessel Function for TM-modes

z [=] axial distance or direction in brass cavity, or plane of pulling direction

INTRODUCTION

The pultrusion process is a continuous manufacturing method which can be used
to produce reinforced plastics with a constant cross-sectional area through the length of
the product. Pultruded samples are expected to exhibit good hydrolytic stability,
improved mechanical strength, and electrical insulation properties. The continuous
microwave processing technique is highly desired for processing large composite
materials, which need even and fast heating such as long pipes or panels. The general
advantages of using microwave technology include: decreased processing times, better
control of temperature profiles within the composite material, and enhanced mechanical
properties.

The objectives of this project are to obtain experimental data of microwave
pultrusion processing and to compare variable frequency and single frequency processing
methods. Temperature-time curves and extent of cure measurements will aid in finding
the optimized microwave pultrusion variables of input power and pulling speed.
Mechanical tests, such as flexural and tensile strength measurements, will be conducted.
A variable frequency power source will be used to determine if mode-switching heating
can provide more even heating in pultruded composites than with fixed frequency heating.
By processing continuous, unidirectional glass and graphite reinforced epoxy prepregs, it
is a goal to Observe an effect of the reinforcement on the efficiency of microwave
processing. Finally, the experimental data obtained will be applied to a pultrusion model
to begin the formulation of a computer program that will assist in the future integration of

on-line monitoring equipment for the microwave pultrusion apparatus.

Chapter 1. Background
1.1. Microwave Processing

Fiber reinforced polymer composites are widely used in the military, the aerospace
industry, and the sporting goods world. The primary way of processing these composites
is thermally. The autoclave process requires a relatively long time to perform. To
increase processing speed and to treat thick therrnosetting composites, industry has turned
to microwave processing. Microwave heating is an alternative method to conventional
thermal heating for the processing of materials. Compared to the thermal heating method,
microwave curing of composite materials is a faster and more direct heating method that
takes advantages of the dielectric properties of the material being processed.

An electromagnetic field interacting with a dielectric material, such as a composite,
heats from the inside out. The temperature profiles inside the composite and the
absorption of microwave energy are a strong function of the electromagnetic wave modes
resonating inside the cavity. [1] By changing the length of the microwave cavity and/or the
coupling probe depth, the waves are allowed to resonate inside the cavity, and the
reflected power is minimized. The cavity length controls what mode is excited and the
coupling probe, which attenuates the radio frequency signal, controls how much power is
reflected back into the circuit from the cavity. The general advantages of microwave
heating are rapid, inside-out heating, better control of the material's exotherm, selective
heating based on the lossy magnitude of the constituents in a material, and improved

mechanical properties of the composite.

1.2. Microwave Heating Mechanism

In conventional thermal heating, the surface of the material is heated and the
heating of the interior occurs by thermal conduction and by the exotherrn of the reaction if
the polymer is thermosetting. On the other hand, microwaves heat the bulk of the material
simultaneously and offer fast, selective, instantaneous, and relatively controllable heating.
Microwaves can penetrate into the bulk of the material and cause heating by coupling with
the material at the molecular level. Microwave energy excites dipole groups, which align
themselves with the electric field. After the electric field is removed, there is relaxation of
the dipoles, which causes heating at the molecular level. The dielectric constant is a
measure of the material's ability to store electrical energy. The dielectric loss factor
(Equation 1) corresponds to the material's ability to dissipate electrical energy as heat.
The effective dielectric loss factor consists of two terms: the loss factor due to the
dielectric and a term dependent on the conductivity and processing fiequency. The
electrical conductivity portion of Equation 1 is often large when a conductive material is

present, so as to drown out the dielectric term.

 

(1)

6'": 84 09+
08

The materials' dielectric properties play an important role in determining how well
radio frequency heating will work. The dielectric loss factor is a function of temperature
and extent of cure. The dielectric constant and loss factor are both implicit functions of
temperature by way of the relaxation time. The relaxation time is the time is takes for
dipolar molecules to reorient themselves afler they have been aligned with the electric

field. The relaxation time is a function of temperature through an Arrhenius relationship.

The penetration depth, 5, in Equation 2, of the electromagnetic waves is a function of the

dielectric properties. [2]

= LI—z—HII + (in — 1]“: (2)
2a) [1,;1' 3,5“ 5'

The penetration depth, which is traditionally known as the point in the thickness of
the sample where the power has decreased to 37% of its incident value, is inversely
proportional to the processing frequency. As frequency decreases, the Skin depth
increases, meaning the waves travel firrther into the tape than would be observed with
higher frequencies.

The dielectric power absorption per unit volume of the material can be calculated

using Equation 3, in which the power absorption is proportional to the dielectric loss

factor.
PM = 21tfeoe’tan53E (3)
where Pab, = Power absorption per unit volume
f = Frequency (Hz)
80 = permitivity of free space (8.854 x 10'14 F/m)
e’ = Dielectric constant
B = Electric field strength

tanfia = Dielectric loss tangent
The dielectric loss tangent is equal to the ratio of e”/e’, so it is apparent that the effects of
microwave heating are predominant in materials where the dielectric loss factor is high
relative to the dielectric constant. The dielectric loss factor is a function of the
temperature and frequency. For reactive systems, it is also a fiinction of the extent of

cure. The absorbed power can be set into Equation 4 to generate basic theoretical

temperature versus time curves. Ifthe percentage of power absorbed is the same when all
of the experimental conditions except input power level are kept the same, an increase in
input

dT/dt = P.t,./ (Cp*p) (4)
power would lead to an increase in power absorption per unit volume. The slope of the
temperature-time curves would then be directly proportional to the input power level. In
other words, increasing the microwave power will increase the slope of the temperature-
time curve, which is synonymous with a more rapid heating rate. Preliminary heating

studies on the composite material will attempt to prove this theory.

1.3. Microwave Heating Advantages

There are numerous advantages to using microwaves for the processing of
polymers and composites: decreased processing times, "inside-out" and selective,
molecular heating, control of the exotherrn by turning off microwave power, and improved
mechanical properties due to better resin/fiber interfacial bonding. Because thermal
heating of composites beats from the outside-in, the material on the outside surface could
degrade before the middle of a large-profiled part reaches the cross-linking temperature.
If the thermal conductivity of the resin decreases during the cure cycle, it becomes more
difficult to transfer the necessary heat to cure the inside of the part. Often, there is a post-
cure step when using conventional techniques, which increases processing times. Radio
frequency technology has been used in pultrusion lines for more than twenty years in the
form of a preheater for the wetted fibers. Chen a large frequency sweep is used for this

preheating step. By installing a single-mode microwave applicator in the pultrusion line

that operates at either a single frequency or a series of intelligently selected frequencies, it
is believed that processing times will decrease. By knowing the dielectric properties of the
materials and how each frequency will directly and instantaneously interact with the
composite, microwaves can properly and efficiently heat the material from the inside-out
and at the molecular level, thereby largely decreasing processing times.

For conventional processing, pultrusion dies have several temperature zones to
control the temperature and viscosity of the material in the die. The zones also control the
gelation point inside the die. As the resin begins to cure, crosslinking occurs which rids
the molecules of their dipolar nature. The microwave energy becomes selective in what it
heats in the resin/fiber mixture because only the monomer dipoles can align themselves
with the electric field, not the crosslinked polymer molecules. Ifthe reinforcement
material has better electrical properties than the resin, the fibers will absorb the majority of
the microwaves and help cure the resin at the interface through conduction. This transfer
of heat at the resin/fiber interface [3] will improve the bonding, and subsequently, the

mechanical properties.

1.4. Microwave Applicators

Four types of applicators are prevalent in the microwave industry: multimode,
single-mode, waveguide, and specialized applicators. Multimode applicators are favorable
because they can excite several electromagnetic modes at a single frequency, and they are
able to process a wide range of materials. On the downside, they are not particularly

energy efficient and can have trouble heating uniformly.

Single-mode applicators heat more efiiciently than multimode applicators. The
single electromagnetic mode excited in the applicator causes regions of the process
material to be heated in a highly localized fashion. "The heating uniformity inside lossy
materials is a strong firnction of the resonant mode." [4] If medium- to low-loss materials
are processed in the applicator, there is no single mode that will heat the large sample
uniformly. [4] The single-mode and multimode applicators are used to process low-loss
and medium-loss materials.

Either single mode heating or mode switching heating can be used in the process.
Proper mode switching heating provides a more uniform temperature profile and can help
alleviate the problems of processing low-loss materials in single-mode applicators. In
fixed frequency processing, adjusting the cavity length mechanically can change the
modes. With the variable frequency power source, the modes can be changed by
electronically tuning the fiequency, which is a much better approach than mechanical
tuning.

The third applicator that can be used to process materials with microwaves is a
waveguide. It is a rectangular or cylindrical hollow pipe made of a conducting material.
Materials are processed with a traveling wave, as opposed to a standing wave in the above
microwave applicator examples. Because the traveling wave is partially absorbed by the
material in the waveguide, the materials are often of high-loss so the waveguides are not
inordinately long.

The final applicators used in industry are specialized applicators such as part-
shaped cavities or horn applicators. Microwave energy is coupled into these units by way

of an iris or a series of irises. Part-shaped cavities are built to have the same dimensions as

the desired part, but the mold walls are the applicator by which the microwaves are
attenuated. Horn applicators concentrate microwave energy before it is delivered into a
larger chamber. Although the design aspect of the specialized applicators is more difficult,
the customer will save money on tooling costs if various dies are normally inserted into the

single- or multi-mode applicator.

1.5. Research Scope and Goals

The objectives of this project are to obtain experimental data of microwave
pultrusion processing and to compare variable frequency and single frequency processing
methods. There will be a fundamental study of the microwave pultrusion equipment,
aimed at providing the data for a pultrusion system design. It is also a goal of this project
to find the optimized microwave pultrusion variables of input power and pulling speed.

The beginning of this study includes system design and material selection. The
materials to be used in this study are continuous, unidirectional prepregs consisting of
glass fiber or graphite fiber and epoxy resin. The effect of the reinforcement on
microwave processing efficiency will be investigated through heating studies and extent of
cure measurements. The prepreg was purchased from suppliers. Hexcel provided samples
of their AS4/3 501-6 prepreg tape. It is an amine-cured epoxy resin with unidirectional
graphite fiber reinforcements. [5] The glass reinforced epoxy prepreg, DA409/E250-6”,
was purchased from Adhesive Prepregs for Composites Manufacturers. [6] Several plies
of the prepreg will be stacked together to form the composite prepreg tape for pultrusion.
Differential Scanning Calorimetry (DSC) will be used to determine the extent of cure of

the pultruded samples. The pultrusion tests will be accomplished using the best heating

mode for the prepreg system at a certain processing speed and input power. The Teflon
die will be used because of its transparency to microwaves. The relationship of
microwave processing speed, input power, and extent of cure will be investigated to find
the optimum operating conditions for microwave pultrusion processing. Pultruded
samples are expected to exhibit good hydrolytic stability, improved mechanical strength,
and electrical insulation properties. Mechanical tests, such as flexural and tensile strength
measurements will be conducted. A variable frequency power source will be used to
determine if mode-switching heating can provide more even heating in pultruded
composites than with fixed fi'equency heating. Finally, the experimental data obtained will
be applied to a pultrusion model to begin the formulation of a computer program that will
assist in the integration of on-line monitoring equipment for the microwave pultrusion

apparatus.

Chapter 2: Application of Continuous Microwave Processing
2.1. Introduction

Microwave heating is an alternative method to conventional thermal heating for
the processing of materials. Compared to the thermal heating method, microwave curing
of composite materials is a faster and more direct heating method that takes advantages of
the dielectric properties of the material being processed. In conventional thermal heating,
the surface of the material is heated and the heating of the interior is induced by thermal
conduction. On the other hand, microwaves heat the bulk of the material simultaneously
and offer fast, selective, instantaneous, and relatively controllable heating. Microwaves
can penetrate into the bulk of the material and cause heating by coupling with the material
at the molecular level. Microwave energy excites dipole groups and the relaxation of the
dipole groups causes heating at the molecular level.

The continuous microwave processing technique is highly desired for processing
large composite materials that need even and fast heating such as long pipes or panels.
The shape of the product is determined by continuously pulling the composite material
through a die to produce uniform profile parts. The idea of continuous microwave
processing is to pass the material through the microwave applicator. The applications of
continuous microwave processing in industry have been used to preheat materials or to
post—cure parts. [7] Asmussen describes a general purpose microwave applicator that is
commercially available. [8] The unique feature of this apparatus is that it allows for precise
tuning for mode selection and fine tuning within a mode selected. The problem, which
was not solved in using this apparatus, is how to seal the cavity when the material is being

continuously processed through the microwave cavity. The control of the microwave

10

leakage from the entry and exit ports of the applicator is one of the key points in
continuous microwave processing. The microwave leakage during continuous processing
of non-conductive materials has been controlled by attaching extended entry and exit ports
to the microwave cavity. [9] A method for controlling microwave leakage in pultrusion of
conductive materials has also been developed. [10]

One pultrusion application of processing materials continuously in a waveguide has
been patented. [11] The idea of using a waveguide as part of a pultrusion die is excellent
in terms of the simplicity of the system. However, the dimension of the waveguide has to
be in accord with the dielectric properties of the materials being processed in order to
create the suitable microwave field patterns. [12] This requirement has greatly limited the
wide application of the technique in industry due to the large variety of materials being
processed. A microwave transparent die will be used for the microwave pultrusion
process. Heating and curing profiles of the pultruded material will be studied at different
processing conditions. The current microwave pultrusion experimental apparatus [10,13]

developed at MSU will be the basis of this pultrusion system study.

2.2. Pultrusion

The pultrusion process is a continuous manufacturing method, which can be used
to produce reinforced plastics with a constant cross-sectional area through the length of
the product. The shape of the product is determined by continuously pulling the
composite material through a die to produce uniform profile parts. Pultruded composites
consist of reinforcing materials, a resin that binds the fibers together, and often a mat

material to improve the appearance of the composite’s surface and other ancillary

11

materials. [14] The key step in a pultrusion process is to control the interactions among
fiber, resin, and additives.

The general pultrusion process involves pulling continuous strands of glass or
graphite fiber from a creel station. To alleviate the build-up of static charge in the fibers,
the creel station is often in the form of steel shelving to ground the fiber spools. A shelf
can hold up to 20 roving packages. [15] The fibers can be passed through ceramic rings to
prevent the fibers from crossing paths and causing entanglements in the resin bath. Fiber
mats or woven rovings can be added to the outside of the part before the die to improve
the transverse strength. There is approximately 500 feet of mat per roll. [15] Surface veils
are also used to give a more finished look to the finished part. Fibers move through a set
of rollers at the entrance and exit of a resin bath to ensure complete wetting out of the
fibers. This is the most important step in establishing good mechanical properties. Slower
line speeds and higher resin bath temperatures give better fiber wet-out results.

The resin bath is metal and contains a plug so it can be drained and cleaned. After
the resin bath, the fibers pass through a set of rollers to rid the fibers of excess resin.
Preforrners before the die begin to align the wetted fibers into the desired profile. Cross
sectional wall thickness of the part should help estimate the line speed needed to make a
satisfactory product. Ifthe wall thickness is 3/ 16”, %-3/8”, or ‘/2-3/4”, then the average
line speed should be 32, 20, or 16 inches per minute, respectively. [15]

Platens or electrical strip heaters heat the metal die. Platens, although easy to
assemble, are only able to heat the top and the bottom of the die. They are usually placed
equidistantly down the die in the beginning, middle, and end. Platens are less efficient than

strip heaters, which will raise energy costs. [15] Zoned, electrical strip heating controls

12

temperatures along the width and the length of the die. The pultruder will have more
control over the optimization of the process parameters such as pulling speed and die
temperature. The conventional processing method is to heat the materials in the die
through thermal convection/conduction. Thermal heating is slow and requires a long die,
often causing uneven heating in large parts.

The preformers before the metal die are somewhat cool to prevent premature
gelation of the resin/fiber system. The die is approximately three feet long. The resin
viscosity initially decreases upon entering the heated die. AS the fibers conform to the
desired profile, the resin is allowed to fiirther penetrate fibers. The die’S temperature is
increased to ensure that a resin skin does not form on the die walls, causing undue shear
forces and poor finishes as the rollers pull the part through the die. [16] In the industrial
pultrusion apparatus, the heating zones are monitored and controlled by an experienced
operator. Pulling speed has a lot of influence on where the peak exotherm will be located
in the die. Ideally, one would like the pultruded part to be nearly cured before exiting,
thereby allowing it to transfer some of its heat to the die walls. If the exotherm peak
occurs to close to the die exit, residual heat in the center of the part could cause fractures
in the end product. [16]

The pulling section or mechanical pulling apparatus can be a set of rollers or a pair
of continuous belts containing pads that grip the pultruded product. Terms used are
hydraulic reciprocating (easy set up), electromechanical caterpillar (requires a large
number of pulling blocks), and ballscrew (high maintenance). [15] In a commercial
process, pulling speeds can range from 2-3 in/min to 10-15 ft/min. [16] If maximum

pulling speeds are desired, it is best to have multiple dies in the line so as to avoid

l3

incomplete fiber wet-out and curing. [16] The pulling force must be greater than the
fiictional force of the fibers and the shear force of the resin layer against the die wall and
the fibers dragging against the resin flowing back into the bath. [16]

A cut-off saw, wet or dry and automatic or manual, provides a method to produce
the pultruded members at the customer's specified length requirements. A wet saw has
water that cools and lubricates the blade at the same time it prevents harmfirl fiberglass
dust to travel through the air. A dry saw needs a dust collection system. Automatic saws
are programmed to cut at the appropriate lengths, while manual saws are run by the
operators themselves. [15] Cut parts are then sorted and stacked for finishing, packaging,

or shipping.

2.3. Experimental Setup

The experimental apparatus (Figure 1) at Michigan State University [10,13]
includes a power source that is either a single frequency (2450 MHz) source with a
magnetron or a variable fiequency (1700-43 00 MHz) power source with a microwave
power generator and TWT (traveling wave tube) amplifier. Circuit components include
directional couplers to send power to the cavity and to the incident and reflected power
meters, circulators to prevent any reflected power of damaging the power source, coaxial
connectors, coaxial cables, and resistors for dummy loads. Power meters were used to
measure the incident and reflected power, and Luxtron temperature units were used to

measure the surface temperature of the prepreg as it was pulled through the cavity.

14

Directional

    

 

 

 

 

 

 

 

 

 

 

 

 

 
 

 

 

 

 

 

 

 

Circulator Coupler
Sweep
Microwave 1 Oscillator
Source I
_ A
Dummy Oscilloscope '

 

 

 

 

 

 

Load

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1 1e”

Figure 1. Microwave Pultrusion Experimental Apparatus at MSU

2.3.1. Microwave Cavity

A Teflon die, as shown in Figure 3, which is transparent to microwaves, producing
a profile measuring 3.80 cm wide by 0.294 cm high was inserted into an opening in the
side of the 17.78 cm inner diameter, tunable, cylindrical, batch, brass microwave cavity.
The dies need to have sufficient dimensional stability and mechanical strength, just like the
ones used in conventional pultrusion. Ceramic dies are also transparent to microwaves.
Teflon dies have lower dielectric loss than the ceramic dies, but have less mechanical
strength. The cavity can be tuned for mode selection and fine tuned to minimize reflected
power by mechanically adjusting the sliding short height and coupling probe depth,
respectively. The microwave cavity in Figure 2 is modified for pultrusion processing such

that doubly corrugated chokes [9] are extended from the cavity to contain the

15

electromagnetic radiation. Conductive fins [3] are grounded to the cavity opening and in
turn, ground the induced current in the conductive graphite fibers when graphite/epoxy is
processed. Finger stock placed in the opening out of the jacket boxes provides additional

dampening of the induced current, thereby reducing microwave leakage.

 

 

Figure 2. Schematic of Microwave Applicator at MSU

 

Figure 3. Entrance To Microwave Cavity through Teflon Die

16

2.3.2. Calibration of Rollers

A Penta Power motor and a Baldor motor were chosen to pull the prepreg tape
through the system. The motors’ dial settings were calibrated to calculate possible
processing speeds (see Table 1). Past work at MSU suggested that the mechanical pulling
system be improved to allow for increased product throughput. [17] It was found that the
maximum obtainable processing speed was 4 cm/s (7.875 fi/min) with the Baldor motor.

Table 1. Calibration of Rollers to Evaluate Possible Processing Speeds

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Dial Setting Penta Power Speed (cm/s) (ft/min) Baldor Speed (cm/s) (ft/min)
'20 0.143 0.281 0.400 0.787
25 0.215 0.423
30 0.286 0.562 0.870 1.712
35 0.373 0.733
40 0.451 0.887 1.333 2.625
45 0.527 1.038
50 0.594 1.170 2.143 4.218
60 0.667 1.312 2.857 5.624
70 0.800 1 .575 2.857 5.624
80 1 .000 1 .969 4.000 7.874
90 1 .333 2.625 4.000 7.874

100 1 .333 2.625 4.000 7.874

 

 

 

 

 

 

2.4. Materials: Preparation and Discussion

The materials used in the study were a Hercules AS4/3501-6 graphite/epoxy
composite and a DA409/E250 glass/epoxy composite. The reason for performing
experiments on both systems is to compare and contrast the ability to process each
material. Processing speeds and efficiency of microwave heating are important factors
that are influenced by the type of reinforcement that is used. The electrically conductive
nature of the carbon fiber initially suggests that it will absorb more of the electromagnetic
radiation and dissipate the heat throughout the composite by conductance. Glass fibers

are less electrically conductive, but are inherently denser than the carbon fibers, leading

17

one to believe that they may provide better mechanical properties. An optimum product
could involve a mixture of these two reinforcements.

As was discussed before, the dielectric constant is a measure of the material's
ability to store electrical energy. The dielectric loss factor corresponds to the material's
ability to spread electrical energy as heat. The efi‘ective dielectric loss factor consists of
two terms: the loss factor due to the dielectric and a portion that deals with the
conductivity. The materials' dielectric properties play an important role in determining
how well radio frequency heating will work. The dielectric loss factor is a filnction of
temperature and extent of cure. The dielectric constant and loss factor are both implicit
functions of temperature by way of the relaxation time. The relaxation time is the time is
takes for dipolar molecules to reorient themselves after they have been aligned with the
electric field.

The University of Mississippi Pultrusion research group conducted pultrusion
experiments on an epoxy system with glass fiber and carbon fiber reinforcements. [18]
Using a thermally heated die, the temperature and extent of cure profiles across the die
were very similar for both material systems. When pultruding a composite using thermal
techniques, the resin's thermal conductivity is an important variable in determining the cure
inside the composite. The cured material on the outside of the pultruded part continues to
be heated for the sole purpose of making sure the exotherm in the center of the part is

enough to meet cure specifications.

l8

2.5. Goals

The Objectives of this project are to obtain experimental data of microwave pultrusion
processing and to compare variable fi'equency and single fi'equency processing. It is
thought that variable fiequency will provide a more even temperature distribution across
the width of the prepreg tape. It is also a goal of this project to find the optimized
microwave pultrusion variables Of input power and pulling speed. Extent of cure and
mechanical property (tensile and flexural strength and modulus) measurements will be
made to support process design. Comparison and contrasting of the effect of fiber
reinforcements on the efliciency of microwave processing will be made. Possible on-line

monitoring techniques will be investigated for future study.

19

Chapter 3: Cavity Characterization
3.1. Finding Electromagnetic Modes

The brass cavity was characterized to obtain the resonant modes by mechanically
changing the cavity length and the coupling probe depth. Normal modes in a right circular
cylinder consist of transverse electric (TE) and transverse magnetic (TM) modes.
Subscripts define which TE-mode is resonant in the cavity: 1 a number of filll-period
variations of E, with respect to 0, m a number of half-period variations of E9 with respect
to r, n a number of half-period variations of E, with respect to z. The subscripts are
defined equally for the TM-modes, but E is replaced by H, the magnetic field. Equation 6

shows the definition for the resonant fiequencies inside the cavity. [19]

r—-—I%I+I§%Jlil

where c = speed of light = 3.00 x 108 m/s

 

x1,“ = mth root of Henckel Function J‘|(x) = 0 for the TE-modes

xim = m‘11 root of Bessel Function Jt(x) = 0 for the TM-modes

From the equation above, a mode chart was constructed for the 7" brass cavity in
which the fi'equency in Hz was plotted versus the cavity length in centimeters. It was
assumed that the mode charts were independent of coupling probe depth, which assists in
minimizing reflected power. It was also assumed that the chokes on the entrance and exit
ports of the cavity did not affect the electromagnetic waves inside the cylindrical cavity
itself. The Teflon or ceramic die, albeit transparent to microwave radiation, was assumed
not to interfere with the resonant conditions of the cavity. There was no material inserted

into the cavity or the die. The x-axis is drawn from the 2.45x109 Hz point on the y-axis

20

because that was the processing frequency used by the single frequency power source. As
seen by Figures 4 and 5, one may obtain eight different modes in single frequency

processing by changing the cavity length a range of 10.5 cm.

 

6.0m0E+09

5.0WE+09

 

40000909 +TE011
+TEO12

TE111
—)(—TE112
+TE211
+TE212
—t-TE312
——-TE311

3.0ME+09

frequency. 1. Hz

2.0000er

 

 

 

1 0000909

  

0.0ME+OO
o

 

 

 

Figure 4. Frequency vs. Cavity Length for 7.00" +/- 0.001" (TE modes)

21

 

 

+TM011
+TMO10

W110
+TM210
+TM111
+TM012
—+—TM112
—.—TM211
——-—TM212

frequency, 1, Hz

 

 

 

 

Cavity Length. Le, em

 

Figure 5. Frequency vs. Cavity Length of 7.00" +/- 0.001" (TM modes)

Ifone chooses a fixed cavity height of 15.5 out, eleven modes can be excited over
the 2.00 to 3.00 GHz frequency range. Electronic tuning of frequency is faster than the
mechanical tuning of the cavity, giving rise to the idea of variable fiequency processing of
composite materials. Heating studies were conducted using a processing frequency of
2.45 GHz for each mode to find which electromagnetic mode most efficiently heated the

material.

3.2. Temperature Measurement
Fluorotropic temperature probes, which are transparent to microwaves, were
placed through holes in the die 8 mm apart across the width of the die and 21 mm apart

along the length of the die (Figure 6). The probes rested against the surface of the

22

prepreg. Temperature readings were sent fiom the Luxtron temperature meter to a

personal computer for data collection and analysis.

5.7 cm

 

 

 

I .l8mm
O
4.9cm . -
. 21mm
I .

5.1 cm

 

 

 

 

 

 

 

24.6 cm

Figure 6. Teflon Die Used for Heating Studies and Microwave Pultrusion Processing

3.3. Heating Studies At Various Modes

Heating studies were conducted using a processing frequency of 2.45 GHz for each mode
to find which electromagnetic mode most efficiently heated the material. Table 2 lists the
cavity dimensions used for the characterization studies. The Teflon die was in place when
the preliminary cavity dimensions were measured. Tables 3 and 4 list the heating modes
for the 17.78 cm cavity with the six-ply graphite/epoxy and glass/epoxy prepreg inserted
into the cavity, respectively.

Temperature versus time curves are shown in Figures 7 and 8 for the
graphite/epoxy and glass/epoxy systems, respectively. The glass/epoxy system was not
pultruded for the fill] range of process variables in the heating study because the
temperature rise was small over time, and an elevated temperature value was desired to be

able to choose the proper mode to continue studying.

23

Table 2. Heating Modes for the Cavity with Teflon Die

 

 

 

 

 

 

 

 

 

Mode # Mode Type Cavity Length, Lc, mm Probe Depth, Lp, mm
1 TM 011 70.5 8.80

2 TE 011 100.5 25.29

3 TM 111 119 13.15

4 TB 112 131 6.95

5 TE 311 152 13.32

 

 

Table 3. Heating Modes for the Cavity with Teflon Die and Graphite/Epoxy Prepreg

 

 

 

 

 

 

 

 

 

Mode # Mode Type Cavity Leggth, Lc, mm Probe Depth, Lp, mm
1 TM 011 70 8.80

2 TB 011 99 16.17

3 TM 111 115 21.57

4 TE 112 133.5 23.42

5 TE 311 151 13.49

 

 

Table 4. Heating Modes for the Cavity with Teflon Die and Glass/Epoxy Prepreg

 

 

 

 

 

 

 

 

 

 

Mode # Mode Type Cavity Legth, Lc, mm Probe Depth, Lp, mm
1 TM 011 70 ll

2 TB 011 10] 13.94

3 TM 111 117 15.75

4 TE 112 131 9.5

5 TE 311 152 16.33

 

24

 

 

 

 

 

 

 

 

 

 

 

 

 

 

80
70 ..
60 If
50 .I.
2 ' , -——TE11
:- 40- - ...» . er" “ ——TM111
a". FM -——TE112
5 30“ ---——T1vn11
.-
——TE311
20 ..
10 ~
0 t t t 1 i
It 50 100 150 200 250 300
-10
tlmo,t,uc

 

Figure 7. Temperature Versus Time Results for Heating Studies of Six-Ply Hercules
AS4/3501-6 Prepreg at 100 Watts and 0.787 fi/min

 

 

 

 

 

 

 

 

 

50 ..
45 I
40 ..
35I ,
30 , ' —Ta)11
o_ —TM111
...
E. 251 ’ ——TE112
~.“' . 4 . ,
r- 20 I. _TND11
—-TE311
15 ..
10 «r
5 ..
O i i t
0 100 200 330 400
tine,t,aoc

 

 

Figure 8. Temperature versus Time Results for Heating Studies of Stationary Six-Ply
DA409/EZSO Prepreg at Approximately 50 Watts

25

3.4. Preliminary Heating Studies In TE“; Mode

Three foot, six-ply Hercules AS4/3 501-6 prepreg tape samples were processed in a
TB“; mode cavity at 2450 MHz in the existing pultrusion apparatus with input powers of
50, 75, and 100 Watts and a pulling speed of0.281 ft/min. From the results in Figure 9, it
was observed that an increase in input power increased the steady state temperature value
of the material. Also, the linear portion of the temperature vs. time curve has a larger
slope for increasing input power, suggesting that higher power leads to decreased
processing times. Table 5 lists the relative values from the extent of cure measurements

conducted on DSC.

 

 

 

 

 

 

 

 

 

 

0
._. ——50w
9’- —75w
5 ———100w

2° ‘I

10 I

0 i r +
0 100 200 300 400 500 600

 

 

 

Figure 9. Results For Preliminary Heating of Six-ply Graphite/Epoxy Composite In TE, 12
Mode At Various Input Powers

26

Table 5. Extent of Cure Measurement for Preliminary Heating of Six-Ply Graphite/Epoxy
Composite in TE, 12 Mode with Input Power Varied

 

 

 

 

 

Input Power (Watts) Pulling Speed (ft/min) Extent of Cure
50 0.281 21.97%
75 0.281 96.66%
100 0.281 97.74%

 

 

 

3.5. Conclusions

The continuous microwave applicator was characterized to find which
electromagnetic modes could be excited during pultrusion. Mode charts for a 7" brass
cavity gave promising prospects for variable frequency processing because eleven modes
could be excited at a specified cavity height. Heating studies of the two prepreg materials
gave insightful results in establishing what mode heated the material most efficiently.

TE] 12 mode was chosen for the fixed frequency studies.

It appears that both the graphite/epoxy and glass/epoxy systems are able to be
processed in the microwave environment. It is shown in Figures 7 and 8 that there is a
more rapid temperature increase in the graphite/epoxy prepreg than the glass/epoxy
prepreg for the same heating mode. Graphite fibers absorb more electromagnetic energy
than the epoxy. In glass/epoxy systems, the matrix absorbs the majority of the energy.
There was a slower rate in temperature rise for the glass/epoxy because there was more
reflected power during processing. The energy was coupling efficiently with the
composite. The microwave cavity must be finer tuned when processing the glass/epoxy

system. Preliminary pultrusion experiments on the graphite/epoxy prepreg revealed that

27

 

the surface temperature rises more rapidly and to a higher temperature with an increase in
input power, supporting the hypothesis that higher microwave input power will lead to a
decrease in processing time. Promising data collected in this series of experiments set the

stage for more in-depth fixed frequency and variable frequency studies.

28

Chapter 4: Fixed Frequency Pultrusion Experiments
4.1. Background

Three foot long, 1 Mt" wide, six-ply Hercules AS4/3501-6 prepreg tape sammes
were processed in a TB“; mode cavity at 2450 MHz in the existing pultrusion apparatus.
Input powers of 40, 45, 50, 55, 60, 65, 70, 80, and 100 Watts were used to heat the tape
at a pulling speed of 0.281 ft/min. Pulling speeds with the corresponding dial settings of
20, 25, 30, 35, 40, 45, and 50 on the Penta Power motor (0.281 ft/min - 1.17 ft/min) were
used to process the tape while an input power of 100 Watts was selected.

From the heating studies, the TE“; mode was established as a center heating mode
where the electric field was the strongest along the central, axial (pulling direction) down
the tape (Figure 23). Optimizing the pultrusion variables of pulling speed and input power
is important in producing a high quality part. Even though high product throughput is
desired, good resin/fiber adhesion is paramount in achieving the mechanical properties

sought by the customer.

4.2. Experimental Setup

Six plies of three foot by 1 Mt" prepreg were cut parallel to the continuous fiber
direction. Next, they were stacked and sent through a pressurized roller to rid the tape of
any voids and air bubbles that were present from the stacking process. The tape was
manually fed through the Teflon die in the side of the cavity until just before the pulling
rollers. The circuit was closed by connecting the input coaxial line to the 100 Watt fixed
frequency power source and the output line to the coupling probe. The Luxtron

temperature probes were placed across the die width in four locations and calibrated to

29

ambient conditions. The probe tips touched the surface of the prepreg for accurate
temperature measurement. Temperature data was collected by electronic signals sent to a
PC. Temperature measurements began for a 30 second period to establish a steady state
reading. Afler the 30 second start-up, the pulling rollers and power source were
simultaneously turned on to their pre-determined positions. The reflected power meter
was monitored to insure that maximum power was being delivered to the composite.
Temperature readings halted once the back end of the prepreg exited the cavity. At this
time, the power source was turned off. Upon exiting the cavity, the tape traveled through
a set of pressurized rollers. The pultruded product was left to cool before thermal and

mechanical analysis samples were made and tested.

4.3. Variation in Pulling Speed

At a fixed pulling speed, the material's residence time in the cavity defines how
long the microwave energy has to heat the product to its curing temperature. If the
pulling speed is too fast for the power input level used, the material will exit the cavity
partially cured and a post-curing step will be needed. On the other hand, too slew of a
pulling speed may cause the resin to be overheated and degrade. "The location of the
exothermic peak depends on the speed of pulling the fiber-resin stream through the die."
[16] The exotherm peak will shift towards the die exit with an increase in pulling speed.
This will also increase the pulling force needed by the mechanical systems. By combining
temperature versus time curves and extent of cure measurements, one can determine the
point where the exotherm has ended and continuing with the addition of microwaves could

cause undue heating. The only chance for heat transfer when the tape has exited the

30

cavity is from the heat conduction through the resin and fiber, which tends to be low, and
by the remaining exotherm of the polymerization reaction. It is desired that the exotherm
be completed before the die exit because if the interior of the pultruded part is still at an

elevated temperature, interlaminar cracks may form within the product. [16]

4.4. Variation in Input Power

The issue of input power level must be considered. In the microwave pultrusion
energy balance (see Chapter 7), input power is an implicit variable in the heat generation
due to the power absorption term. In a published power absorption model [1], the five
parameters of interest are all explicit firnctions of input power. Also, to shorten
processing times and times for a material to reach a desired temperature, higher power
input is required, as suggested by the preliminary heating studies on the graphite/epoxy
prepreg. At a fixed pulling speed, too high of power may cause the resin to degrade and
the mechanical properties may suffer. Not enough power will result in a low-cured
product and the resin/fiber interfacial adhesion will be poor. Extent of cure and
mechanical property measurements will be related to pulling speed and input power in

order to find the optimum processing conditions.

4.5. Extent of Cure Measurement (Differential Scanning Calorimetry)

Because TE“; mode was used, the rriagnitude of the electric field was greater at
the center of the cavity and parallel to the pulling direction. Further examination of the
prepreg revealed it to be preferentially heated down the center of the tape. Relative values

of extent of cure were found using DSC. A sample from an uncured piece of

31

graphite/epoxy composite was heated 10°C/min to 350°C to establish a theoretical
maximum heat of reaction. When an epoxy matrix is heated, 100% of the possible
reactive sites do not react, so 100% extent of cure is not theoretically possible. Extent of
cure values reported below and in plots in Figures 10 and 13 are relative values and not
absolute. Samples were taken from the center of the prepreg tape. Samples were heated
10°C/min to 350°C to find the heat of reaction, which was compared to the heat of
reaction for the uncured composite. Equation 7 shows how the extent of cure was

calculated.

5

 

(7)

5

where or=extent of cure

HRc=heat of reaction of cured sample (J/ g)

HRU=heat of reaction of uncured sample (J/g)

The heat of reaction for the uncured composite sample was found to be 197.5 J/g.
Extent of cure increased with microwave power input up to 70 Watts and then leveled off
to approximately 98%. At these high levels of power input, the resin may have degraded,
resulting in what looks to be a "well-cured" product. There was negligible change in
extent of cure with powers above 70 Watts. Powers below 40 Watts, which gave 18%
extent of cure, had low heating potential and were neglected. Equation 8 simply reveals

why increasing input microwave power at the same pulling speed results in a higher extent

of cure.

(Cpft'wf +Cpm*(l—wf))*(Tc —7;)*(wm +wf)
P

 

t = (8)

32

Above, the time, t, for the temperature to rise to a curing temperature, T,, from the
ambient temperature, T,, is dependent on the weight fractions, w, and w;, and heat
capacities, Cpm and Cpf, of the matrix and fiber, respectively. The microwave power, P, is
inversely proportional to time so an increase in power leads to a decrease in time. If the
material's residence time is the same for varying input power, the exiting temperature of
the prepreg will be greater with increasing input power. The extent of cure is indirectly
related to the temperature through the kinetic parameters in the kinetic model. Equation 8
is a basic energy balance, which assumes that all the incident power delivered to the
microwave cavity is absorbed by the composite, causing heat generation within the
material from dipole relaxation.

Extent of cure increased somewhat linearly with increasing microwave power up
to 70 Watts and then leveled off to approximately 98%. The increase in power increases
the amount of heat generation inside the composite. Heat conduction is low perpendicular
to the fiber direction, resulting in a low-cured outer region of the product. There was
negligible change in extent of cure with powers above 70 Watts. At a pulling speed of
0.281 fi/min, power levels below 40 Watts had low heating potential and were neglected.
Figures 11 and 12 show trendlines with equations for extent of cure as a firnction of input

power at a constant pulling speed.

33

 

 

 

 

 

 

 

 

 

Figure 10. Effect of Input Power on Extent of Cure

 

 

0.9 ~>

0.3 i. y = 0.026x - 0.8687
2 -

0.7 «F R - 0.9887

0.6 «~
0.5 1»
0.4 ..
0.3 +
0.2 ~-
0.1 <»

0 + t t t + i
0 1 0 20 30 4O 50 60 70
hput Power (Watts)

Extent of Cure

 

 

 

 

 

 

Figure l 1. Trendline for Extent of Cure as Function of Input Power (40-70 Watts)

34

 

 

0.98

0.975 I
y = 55-053 - 0.0071x + 1.1924

0.97 4. R2= 1

.0
l

0.96 I

0.955 I

Extent of Cure

.0
8
+

0.945 «r

0.94 I

 

 

 

0.935 t 1 4. t
0 20 40 60 80 100

Input Power (Watts)

 

 

 

Figure 12. Trendline for Extent of Cure as Function of Input Power (70-100 Watts)
For pulling speeds between 0.5 fi/min and 0.7 ft/min, the extent of cure decreases

rapidly. For pulling speed ranges of 0.3-0.5 ft/min and 0.7-1.2 fi/min, the extent of cure
changed only 15% in each range. It is suggested to operate within these ranges for greater
control of extent of cure and better quality of product. It should be noted that line speeds
would increase with higher input power. The extent of cure dependence on pulling speed
is justified by the "residence time" of a differential slice of prepreg inside the cavity.
Figures 14, 15, and 16 show the extent of cure as a function of the pulling speed and give
trendlines for each of the three regions of interest: the high cure section of slow
processing speeds, the linear region of moderate pulling speeds for a low input power, and
high pulling speed region below the gel conversion point. Longer residence times will
allow the electromagnetic energy to interact with the composite for a longer period of
time. For slower processing speeds, the prepreg was cured across a greater width of the

tape. For processing conditions that produce a product with an extent of cure of 30% (gel

35

conversion for epoxy resin) or below, the microwave applicator could act as a preheating
device in a pultrusion line. If maximum extent of cure is desired, the microwave

pultrusion apparatus could be the sole heating element in the pultrusion line.

 

 

0.9 +
0.8 +
0.7 «~
,3 0.6 ..
° 0.5 -~

0.3 “r
0.2 1*
0.1 r

 

 

 

0.000 0.500 1 .000 1 .500
Pulling Speed (ll/min)

 

 

 

Figure 13. Effect of Pulling Speed on Extent of Cure

1
0.9 .. x
0.8 ..

0.7 ..
8
3 0.6 I y = 25999:? + 1.6135x + 0.7293
0 0.5 .. R’=1

 

 

 

" 0.3 --
0.2 ~~
0.1 ..

0 1r a
0.000 0.200 0.400 0.600
pulling speed (It/min)

 

 

 

 

 

 

Figure 14. Trendline for Extent of Cure as Function of Pulling Speed (0.281-0.562 ft/min)

36

 

 

0.800
0.700 «r
0.600 1+

0.500 4»

0400 4» y = -1 .55521 + 1.607
R2=1

extent of cure

0.300 I
0.200 1*

0.100 I

 

 

 

0.000 4 4 1 1
0.000 0.200 0.400 0.600 0.800 1 .000

pulllng speed (ll/min)

 

 

 

Figure 15. Trendline for Extent of Cure as Function of Pulling Speed (0.562-0.733 ft/min)

 

 

 

 

 

0.35
0.3 1
0.25 I
3
u 0.2 ~~
'0'
E 0.15 «1
0,1 .. y = 0030018 + 0.846811 + 0.056
R’ = 0.9938
0.05 «1
0 1 ea 1 1 1
0.000 0.200 0.400 0.000 0.000 1.000 1.200
pulllng speed (ft/min)

 

 

 

Figure 16. Trendline for Extent of Cure as Function of Pulling Speed (0733-1. 17 fi/min)

37

4.6. Mechanical Properties Measurements
4.6.1. Tensile Strength

Tensile coupons measuring approximately 17.78 cm (7 ") long and 0.9525 cm
(0.375") wide at the middle were cut from the cured, unidirectional graphite/epoxy
product. Three samples were tested for each set of processing conditions. The coupon
was shaped to specifications with a router while the sample was in a tensile bar metal
clamp. Tensile strength measurements were made using a pulling rate of 1.27 cm/min
(0.5"/min), and the samples were pulled parallel to the fiber direction. A 1000 lbf load cell
was used, in which the grips each held one inch of the tensile bar's ends. The load was
plotted against the strain to help calculate the tensile modulus. Three samples were made
for each processing condition. Coupon thickness and width in the middle were measured
at three points and averaged. Maximum tensile strength and load were measured and
tensile modulus was calculated through analysis of stress versus strain curves.

The results for the tensile tests were averaged for the three samples, and those
values are shown in Figures 17 through 20. Measurements typically varied from the
average by no more +/- 10%. Values for the tensile strength ranged from 138 to 188
MPa. The tensile strength was not affected by pulling speed or input power variations
because the samples were tested by pulling in the fiber direction. The tensile modulus was
calculated by performing a linear regression on the initial portion of the stress versus strain
curve. Average tensile modulus values ranged from 1100 to 1500 MPa, with values
varying +/- 15% from the average. Any anomalies in the curves were from experimental

error, such as the sample slipping in the tensile bar grips.

38

 

 

—. In.
J. 1
rr I

Tensile Strength (MPa)

 

 

 

0 20 40 60 80 100
Input Power (Watts)

 

 

 

Figure 17. Tensile Strength of Six-ply Graphite/Epoxy Prepreg Tape with A Pulling
Speed of 0.281 ft/min and Variable Input Power

 

1403.00

1200.00 ~~ /\
D

1003.00 ~>

 

800.11) «t

50000 «»

Tenslle Modulus (MPa)

200C!) --

 

 

0.00 1 41 1 1
O 20 40 60 80 100

In put Power (Watts)

 

 

 

 

Figure 18. Tensile Modulus of Six-ply Graphite/Epoxy Prepreg Tape with A Pulling
Speed of 0.281 ft/min and Variable Input Power

39

 

 

Tensile Strength (MPa)

 

 

.1.
T

O 0.2 0.4 0.6 0.8 1 1 .2
Pulllng Speed (mmln)

 

 

 

 

Figure 19. Tensile Strength of Six-ply Graphite/Epoxy Prepreg Tape with an Input Power
of 100 Watts and Variable Pulling Speed

 

 

 

 

 

16a)
1400 I
as! 1200 ..
‘; 1WD ~~
‘3 000
3
6m +
3
§ 400 ~~
)—
200 «I
O #1 1 1 1 1
O 0.2 0.4 0.6 0.8 1 1.2

Pulllng Speed (ltlmln)

 

 

 

Figure 20. Tensile Modulus of Six-ply Graphite/Epoxy Prepreg Tape with an Input Power
of 100 Watts and Variable Pulling Speed

For unidirectional composites containing fibers parallel to the pulling direction, the
samples fail by tensile rupture of the fibers. [16] This is followed or accompanied by the
debonding of the fiber/matrix interface parallel to the fibers. [16] Any final failures in the

composite are equal to (1-f)om.., where f is the volume fraction of fibers in the composite,

40

and on... is the ultimate strength of the matrix. If the matrix begins to fail while the fibers
are still bearing a portion of the load, then the composite's ultimate tensile strength can be
calculated fi'om Equation 9. [20]
6111 = fO'ru + “-00qu (9)
where Cf“ = ultimate strength of the fiber
on, = ultimate strength of the composite

emf“ = matrix stress at the onset of fiber cracking

4.6.2. Flexural Strength

Flexural strength tests were conducted on the Universal Testing System in
compliance with ASTM Standard D 790 - 84a. [21] The flexural strength and flexural
modulus were obtained for the tests. The support span-to-depth ratio was 16 to 1 for the
three-point loading machine. The pultruded composite samples were approximately 3.2
mm thick. Again, three samples were cut for each processing condition. The
recommended dimensions for test specimens were 25 mm wide and 80 mm long. The
support span was 50 mm, and the rate of cross-head motion was 1.3 mm per minute.

Figures 21 through 24 show the averaged results for the flexural test
measurements. Values were averaged for the flexural strength and ranged fi'om 25 to 120
MPa. Measurements typically varied from the average by no more +/- 10%. The flexural
strength of the material is thought to be the maximum fiber stress at failure on the tension
side of a flexural sample. [16] The flexural strength increased with increasing input power
and reached a maximum at approximately 60 Watts. At input powers greater than 60

Watts, the flexural strength decreased due to the brittleness of the specimens. F lexural

41

strength increased with increasing pulling speed up to a maximum of about 0.42 fi/min.
For faster pulling speeds, the specimens bent much more because the resin/fiber interfacial
strength was low and the resin was undercured. The flexural strength values for these

higher pulling speed experiments were a reflection of the non-homogeneity of cure across

 

 

 

 

 

the width of the prepreg.

140

120 1-
TI’
g 100 ..
E 00 ..
:3 60 --
E 1
a 40 .JL
E

20 ..
0 fl i i i
0 20 40 60 80 100
Input Power (W)

 

 

 

Figure 21. Flexural Strength of Six-ply Graphite/Epoxy Prepreg Tape with A Pulling
Speed of 0.281 ft/min and Variable Input Power

 

 

 

 

 

7.5 1400

a.

g 1200 -.

g 1000 «-

3

E 000»

e

3 400+

§ 200~~

2 0 1 5* 4 1
0 20 40 60 80 100

Input Power (Watts)

 

 

 

Figure 22. Flexural Modulus of Six-ply Graphite/Epoxy Prepreg Tape with A Pulling
Speed of 0.281 ft/min and Variable Input Power

42

 

8

 

8

9

 

Flexural Strength (MPa)
8 8

.8

O
I
1

 

 

O

 

L
U

0.2 0.4 0.6 0.8 1 1 .2
Pulllng Speed (itlmln)

O

 

 

 

Figure 23. Flexural Strength of Six-ply Graphite/Epoxy Prepreg Tape with An Input
Power of 100 Watts and Variable Pulling Speed

 

 

700

600 I

iii

Modulus of Elasticity (MPa)

 

 

 

0 0.2 0.4 0.6 0.8 1
Pulling Speed (ll/min)

 

 

 

Figure 24. F lexural Modulus of Six-ply Graphite/Epoxy Prepreg Tape with An Input
Power of 100 Watts and Variable Pulling Speed

43

Average flexural modulus measurements ranged from approximately 300 MPa to
1400 MPa, with values varying +/- 15% fiom the average. The modulus suffered from a
low-cured product at low input power levels or too high pulling speeds and from a brittle
product with a partially degraded matrix. When fibers are long and continuous, there is
larger statistical chance that there will be flaws in the fibers which could allow for failure
in the composite. [22] The resin acts like glue between the sensitive, dry fiber bundles.
When a load is applied, the resin increases the fibers' ability to carry that load by holding
together the surface imperfections in the fibers. [22] If the resin degrades with
overheating, the resin/fiber interfacial bond strength is at risk. The resin will not fill its
role of being glue and may become brittle just like the dry fiber bundles. It is important to
balance the processing conditions to ensure that the resin/fiber interface is strong, so the

resin can help improve the composite's strengths and moduli.

44

4.7. Conclusions

Extent of cure of the graphite/epoxy composite was established as a firnction of
input power and pulling speed for a microwave pultrusion apparatus. Extent of cure
increased with an increase in microwave power input and decreased with an increase in
pulling speed. At 100 Watts input power and a pulling speed of 0.281 fi/min, the center of
the six-ply graphite/epoxy prepreg was 98% cured.

Tensile strength and modulus were relatively unaffected by the change in process
variables. There was variability in the flexural strength and modulus with respect to input
power and pulling speed. Too high of power or too slow of pulling speed resulted in a
brittle matrix. Low input power and fast pulling speeds gave a product that had a

relatively uncured matrix and low interfacial bond strength.

45

Chapter 5: Variable Frequency Pultrusion Experiments

5.1. Background

Either a fixed frequency or variable frequency microwave power source can be
used to obtain uniform and fast microwave heating. When applying fixed fiequency
heating, one must rely on the thermal properties of the materials to dissipate heat
throughout the composite for a more even distribution of cure. During the single mode
heating studies, the prepreg tape was not heated evenly over the width. Some transverse
electromagnetic modes revealed that the tape was heated mOre in the center (Figure 25),
while others heated preferentially on the outside edges (Figure 26). Temperature
gradients for the single mode experiments reached as much as 40-45°C. Past results have
shown that an even cure distribution occurred after 15 minutes of single mode heating.
[17] These results and the need to increase process throughput led to the idea of variable
frequency processing of the prepreg, such that the cavity length would be kept constant
and processing fiequencies would be intelligently selected.

Either single mode heating or mode switching heating can be used in the process.
Proper mode switching heating provides a more uniform temperature profile. In fixed
frequency processing, the modes can only be changed by adjusting the cavity length
mechanically. With the variable frequency power source, the modes can be changed by
electronically tuning the frequency, which is a much better approach than mechanical
tuning.

In the variable fi'equency study, the 17.7 8 cm cylindrical brass cavity was
characterized, with a Teflon die and processing material inserted, for a fiequency range of

2.00 to 3.00 GHz. Heating studies for selected fi'equencies helped to intelligently select

46

those used in filrther processing studies. Prepreg thickness was varied to investigate if
sample size afi'ected microwave power absorption efficiency. Time intervals for each

processing frequency were varied to establish an optimum set of processing conditions for

a particular prepreg thickness.

 

p°Sition
Figure 25. Six Ply Hercules AS4/3501-6 Heated in TEnz mode at 100 Watts and 0.281
fi/min

Position 3 corresponds to the middle of the die, while positions 1 and 5 correspond

to the outer edges of the die. Each position is 8mm apart. The y-coordinate is the
processing time in seconds and can be compared to die length in conventional pultrusion

processing by considering a differential slice of prepreg and monitoring its temperature as

47

it passes through the heater. Then, by multiplying the processing time by the pulling

speed, die length would vary from two to three feet for the cases below.

0 ll ‘ejmeledwel

 

. 0
posIIiOn

Figure 26. Six Ply Hercules AS4/3501-6 Heated in TEou mode at 100 Watts and 0.281
fi/min

5.2. Frequency Sweep Analysis
For an initial set of cavity dimensions, the coupling probe was set at a depth of
15.89 mm and the cavity length (L,) was varied from 160 mm to 70 mm by 10 mm
increments. A Lambda variable fi'equency power source acted as an amplifier to a HP
Sweep Oscillator to enable the microwave input power to be increased to approximately

150 Watts. A process control program written by Qiu [23] measured the reflected power

48

for a frequency sweep of 2.00 to 3.00 GHz. Figure 27 is presented as an example of a
poor selection of cavity dimensions such that there are only a couple of fiequencies that
have below 10% reflected power. When high input power is desired, there is a greater

risk of the source shutting down due to the reflected power, even at 10-15% reflection.

 

0.6

 

0.5 ~~

0.4 w

0.3 ~-

0.2 .

16 Reflected Power

0.1 ..

 

A
I

1? 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9 11

 

 

 

-0.1
frequency. 1, GHz

 

 

 

Figure 27. Percentage of Reflected Power At 2.00 to 3.00 GHz Processing Frequencies
For A Cylindrical Brass Cavity With L0 = 90 mm And dp = 15.89 mm And Containing Six-
ply Hercules AS4/3501-6 Prepreg

Frequencies with minimum reflected power were individually selected to determine
their heating potential (Figure 28). A cavity length of 135mm and a coupling probe depth
of 15.89mm were used in the first study of variable frequency processing of a six-ply
Hercules AS4/3501-6 prepreg tape. A mode-switching control program was used to
intelligently select the appropriate frequencies. The process control program allowed the

operator to specify the time interval each fi'equency would heat the material. The time

49

intervals could be the same, or they could vary by simply changing the value before the

next experimental run. Initially, each frequency was given the same time interval value.

 

0.3

0.25 .. h (

 

O

N

1r
I

16 Reflected Power
.0
at

 

.0
d

 

 

 

 

 

 

 

0.05 .1 II (I

0 1 1 1 1% 1 1 a 1 1
2 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9 3
frequency, 1. GHz

 

 

 

 

 

 

Figure 28. Percentage of Reflected Power At 2.00 to 3.00 GHz Processing Frequencies
For A Cylindrical Brass Cavity With Lc = 135 mm And (1,, = 15.89 mm And Containing
Six-ply Hercules AS4/3501-6 Prepreg

5.3. Input Power vs. Frequency

For the variable frequency power source, power output varies with frequency
(Figure 29). For example, if the input power was set to 100 Watts, the frequencies with
low power output make the average input power less than 100 Watts. Also, when the
time interval is on the order of milliseconds, the fiequencies are changing very quickly, and

there is a relatively high proportion of time spent not heating the composite. Higher input

power is needed overall if the composite is to be cured in a short time scale.

50

 

 

_a ..a .at .a
d N b O a)
r I I I
r Y I T

lnputPowerMHOO
O
(D

.0 .0 .
a a:
1 _l._

 

0.2 1~

0 A A A J A A A A I
r r T I T Y T

2 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9 3
l'requency.l,Gl-lz

 

 

 

 

 

 

Figure 29. Incident Power Versus Processing Frequency for Lambda Variable Frequency
Power Source [23]

5.4. Temperature Measurement and Heating Studies Part I: 6-ply Prepreg, Four

Frequencies For V.F. Processing

By placing fluorotropic temperature probes across the Teflon die inside the

microwave cavity, it was found that temperature varied by less than 15 degrees across the
1 V4" prepreg tape. The tape was only heated to 42°C using 65 to 73 Watts input power
and a time interval of 100ms for each fiequency (Figure 30). The frequencies with low
reflected power were judged by measuring their heating capabilities. This was done by
heating a stationary prepreg sample in the cavity and recording data from the temperature
meter. The frequencies that had provided efficient heating were compared for their power
output potential from Figure 27. The frequencies with similar power outputs were chosen
out of this second set of characterization studies and were used for additional variable

frequency studies.

51

 

 

 

 

 

 

 

 

 

 

45
40 ..
35 ._
o
1—‘ 30 1»
5 25 . ——mlddle
E 20 ._ —0utermi1dle
% 15 I outer
)—
10 ..
5 ..
0 1 1 1 1
0 100 200 300 400 500
tine,t, sec

 

 

 

Figure 30. Six-Ply Hercules AS4/3 501-6 Heated With Variable Frequency Processing
(2.4285 GHz, 2.444 GHz, 2.594 GHz, and 2.6140 GHz - 100 ms each) At An Average Of
70 Watts And A Pulling Speed of 0.281 fi/min

5.5. Heating Studies Part H: lS-ply Prepreg, Change In Cavity Dimension

A new set of experiments was conducted in which an 18-ply piece of prepreg tape
was processed in the cylindrical brass cavity. It was thought that the thickness of the six-
ply tape was not efficiently interacting with the electromagnetic radiation put out by the
power source. The coupling probe depth ((1,) was changed, and the cavity length was
again varied. At a cavity length of 93 mm and a coupling probe depth of 22.90 mm, it was
found that several modes could be excited. Figure 33 shows there were thirteen
frequencies having below 10% reflected power.

Figures 27, 31, 32 and 33 will show the natural selection of the final set of cavity
dimensions that was used for many of the variable frequency experiments. Cavity length,
coupling probe depth, and prepreg thickness afl‘ected the frequency sweep charts. By
changing the prepreg thickness from 6-p1y to 18-ply, more of the incident power was

absorbed by the material, resulting in more possible modes that could be excited with

52

minimal reflected power (RP. < 10%). Increasing the coupling probe depth seven
millimeters decreased the reflected power for the possible modes and the peaks were
tighter, giving a better idea of the processing frequency needed. Finally, the cavity height
was increased three millimeters. The reflected power was decreased slightly, and there
was a tighter range of possible modes around the center frequency of 2.45 GHz, which

was theoretically the most efficient processing fiequency in the swept range.

 

0.25

0.2 + /

 

.0
..a
0'1

 

 

O
.a
#1

 

St Reflected Power

 

 

 

0.05 .. I]

A A L 4L A A A J K
I T I T I I

fifif f

2 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9 3
frequency, 1. GHz

 

 

 

 

 

 

 

Figure 31. Percentage of Reflected Power At 2.00 to 3.00 GHz Processing Frequencies
For A Cylindrical Brass Cavity With Lc = 90 mm And 11,, = 15.89 mm And Containing
Eighteen-ply Hercules AS4/3 501-6 Prepreg

53

 

0.25

 

0.2 1r

.0
.5

16 Reflected Power
.0
a:
3,—

 

 

 

 

 

 

 

 

”I II

2 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9 3
frequency. 1. GHz

 

 

 

 

Figure 32. Percentage of Reflected Power At 2.00 to 3.00 GHz Processing Frequencies
For A Cylindrical Brass Cavity With L, = 90 mm And d, = 22.90 mm And Containing
Eighteen-ply Hercules AS4/3 501-6 Prepreg

54

 

 

0.25 +

0.2 «-

0.15 1-

0.1 +

% Reflected Power

 

 

 

 

0.05 <~

 

 

 

-0.05

 

 

frequency, 1, Sit

 

 

Figure 33. Percentage of Reflected Power At 2.00 to 3.00 GHz Processing Frequencies
For A Cylindrical Brass Cavity With L, = 93 mm And dp = 22.90 mm And Containing
Eighteen-ply Hercules AS4/3501-6 Prepreg

Input power was raised to approximately 100 Watts, and the time interval was set
to 1000ms for each frequency. Each of the three fiequencies heated the composite for
approximately one second before switching to the next one selected. Temperature versus
time profiles (Figure 34) revealed that again the center temperature was higher than the
outside of the tape. The tape was too thick to exhibit large heating rates with the small
input power used, so the tape thickness was decreased to 11 plies. It was then suggested
that the time interval used to switch among the frequencies not be constant, but that the
time in which the outside heating frequencies are attenuated be increased. This would
allow the outside edges to be heated to a temperature closer to that in the center of tape,

resulting in a more even distribution of cure.

55

 

 

 

-—center
outside
order-middle

 

 

 

 

 

 

 

 

 

0 11 1111 1 .1
0 50100150200250300350400450
tine.t.ssc

 

 

 

 

Figure 34. Eighteen-Ply Hercules AS4/3 501-6 Heated With Variable Frequency
Processing (2.1173 GHz, 2.4162 GHz, and 2.7014 GHz - 1000ms each) and A Pulling
Speed of 0.281 ft/min

5.6. Heating Studies Part HI: ll-ply Prepreg, Four Frequencies and Time Intervals
The third set of experiments involved heating eleven-ply Hercules AS4/ 3 501-6
prepreg tape. Four fiequencies (2.1173 GHz, 2.4162 GHz, 2.6328 GHz, and 2.6809
GHz) were used to process the tape at 0.281 fi/min and 50 Watts input power. 2.1173,
2.6328, and 2.6809 GHz were deemed as outside heating frequencies from previous
heating studies. 2.4162 GHz had the highest heating potential out of the four fi'equencies;
therefore, the time interval was set to lOms, while the other three were set at 1000 ms.
Figure 35 shows the heating profiles for this set of processing conditions. The composite
did not heat at a rapid rate or to high temperature. 2.1173 GHz in TE“ mode was
deemed to be a poor heating frequency by running single frequency heating experiments

on static prepregs. It was replaced by 2.4267 GHz in TE“; mode for the remainder of the

56

variable frequency heating studies. This change established two center heating fiequencies
and two outside heating frequencies. Another modification made was the input power
being raised to approximately 100 Watts. Ideally, the goals were to obtain rapid heating

to a high temperature and an even temperature distribution across the composite prepreg.

 

 

 

 

 

 

 

 

 

 

 

50

45

40 .-
U 35 .
t-'
g 30 ~— I—outq-rniddle
l 25 -—oenter
a 20 fl outer-middle
9 —outside
I" 15 ..

10 <—

5 .-

0 . r 1 . I

0 100 200 300 400 500 600

time,t,see

 

 

 

Figure 35. Eleven Ply Hercules AS4/3 501-6 Heated With Variable Frequency Processing
(2.1173 GHz - 1000ms, 2.4162 GHz - lOms, 2.6328 GHz - 1000 ms, 2.6809 GHz -
1000ms) and Using A Pulling Speed Of 0.281 fl/min

Runs were conducted in which the center heating frequencies (2.4162 GHz and
2.4267 GHz) were set at equal time intervals (1000ms and 2000ms) and the outside
heating frequencies (2.6328 GHz and 2.6809 GHz) were set at time intervals of 5000ms

and 7000ms.

57

 

 

 

--— outer-middle
-— middle
—— miter-middle

 

 

 

 

 

 

 

0 100 200 300 400 500
the, t,sec

 

 

 

Figure 36. Eleven Ply Hercules AS4/3 501-6 Heated “nth Variable Frequency Processing
(2.4162 GHz - 1000 ms, 2.4281 GHz - 1000 ms, 2.6328 GHz - 5000 1115, 2.6809 GHz -
5000 ms) and Using A Pulling Speed Of 0.281 ft/min
By increasing the center frequency time interval, the heating rate of the tape

increased, but the temperature distribution was not uniform. This can be seen by
comparing Figures 36 and 37. Increasing the outside heating frequency time interval was
detrimental to heating and was discontinued. The temperature gradient across the width

of the prepreg increased. Figure 38 shows the results of for an increase in time interval for

the outside heating fiequencies.

58

 

 

 

—- outer-middle

— middle

 

outer-middle
— outside

 

 

Temperature, T, C

 

 

 

 

 

 

 

 

Figure 37. Eleven Ply Hercules AS4/3501-6 Heated With Variable Frequency Processing
(2.4162 GHz - 2000 ms, 2.4281 GHz - 2000 ms, 2.6328 GHz - 5000 ms, 2.6809 GHz -
5000 ms) and Using A Pulling Speed Of 0.281 ft/min

 

 

 

. —- aster-middle
— centre

—- mummiddle

 

 

 

 

 

 

 

 

 

 

Figure 38. Eleven Ply Hercules AS4/3501-6 Heated Vlfrth Variable Frequency Processing
(2.4162 GHz - 2000 ms, 2.4287 GHz - 2000 ms, 2.6328 GHz - 7000 ms, 2.6809 GHz -
7000 ms) and Using A Pulling Speed of 0.281 fi/min

59

5.7. Heating Studies Part IV: ll-ply Prepreg, Four Frequencies and Variation In
The Time Interval

The final three experiments involved setting the following time intervals for the
appropriate fiequencies: 2000ms for 2.4162 GHz, 1000, 1250, and 1500 ms for 2.4267
GHz, and 5000 ms for 2.6328 and 2.6809 GHz. Since 2.4267 GHz supplied the most
efficient heating, it was paired with the lowest time interval in hopes of better controlling
the heating rate and temperature distribution of the composite prepreg. Microwave input
power was approximately 100 Watts for the three trials. Figures 39 and 40 show the two
heating profiles for a change in the time interval for the frequency 2.4267 GHz from 1000
ms to 1500ms.

When the time interval was changed from 1000 ms to 1500 ms, the heating of the
composite was more rapid - approximately half the time to reach steady state. The
temperature gradient across the prepreg was less pronounced. The increase in the time
interval also allowed the composite to be heated an additional 10°C. The temperature
distribution across the pultruded part varied from a few degrees to 20°C, which is much
improved from the 40°C temperature gradient on the prepreg from center to edge with

single frequency processing at 100 Watts input power.

60

 

 

 

 

 

 

 

 

 

 

90

80

70 (I

60 1
U
r-’ 5° 1 —outer-middle
5 ,0 ( —..m.1
E" outer-middle

30 -

20 -

10 <

0 t t t 1

0 100 200 300 400 500
time, t, see

 

 

 

Figure 39. Eleven Ply Hercules AS4/3501-6 Heated With Variable Frequency Processing
(2.4162 GHz - 2000 ms, 2.4287 GHz - 1000 ms, 2.6328 GHz - 5000 ms, 2.6809 GHz -
5000 ms) and Using A Pulling Speed of 0.281 ft/min

 

 

 

o

i-7 60

5 so _

I W

g. 40 outer-middle
[-

 

 

 

-r 00 'n Os )0 n l‘ V
-— -- N N M M v V In n
“0.1.506

 

 

 

Figure 40. Eleven Ply Hercules AS4/3 501 -6 Heated With Variable Frequency Processing
(2.4162 GHz - 2000 ms, 2.4287 GHz - 1500 ms, 2.6328 GHz - 5000 ms, 2.6809 GHz -
5000 ms) and Using A Pulling Speed of 0.281 ft/min

5.8. Extent of Cure Measurement

Extent of cure was measured for the product from the pultrusion of an ll-ply
prepreg heated with the variable frequency power source. Temperature versus time
curves are displayed in Figure 37. The processing frequencies and their corresponding
time intervals and input powers used were 2.4162 GHz (2000 ms and 90 Watts), 2.4281
GHz (2000 ms and 87 Watts), 2.6328 GHz (5000 ms and 88 Watts), 2.6809 GHz (5000
ms and 74 Watts). DSC samples were taken from five points across the width of the
prepreg to investigate the distribution of cure. The processing speed for the experiment
was 0.281 ft/min. The heat of reaction of the uncured sample Hercules AS4/3501—6 was
197 .5 J/g fiom DSC. Table 6 lists the results of the extent of cure study for variable
frequency processing.

Table 6. Extent of Cure of 1 l-ply Prepreg With Variable Frequency Processing

 

 

 

 

 

 

 

Position Heat of Reaction (J/g) Extent of Cure (%)
Outside - Left 158.3 19.85
Middle/Outside - Left 142.2 28.0
Center 125.0 36.71
Middle/Outside - Right 130.5 33.92
Outside - Right 153.6 22.23

 

 

 

62

 

The graphite/epoxy composite prepreg was successfirlly heated using variable
frequency processing technology by setting individual time intervals for the processing
frequencies as opposed to having the same time interval for all four frequencies. Pulling
speed was kept constant at 0.281 ft/min for all experimental runs. Higher microwave
input power will allow more rapid heating, heating to a higher temperature, and ultimately,

faster processing speeds.

5.9. Conclusions

Variable frequency processing provided an even temperature distribution across
the prepreg. Thicker samples provided more efficient heating in the variable frequency
study. The cavity was characterized by measuring reflected power versus frequency for
various cavity lengths and coupling probe depths. Frequencies with the highest heating
potential and similar power output were selected to perform curing studies. Time
intervals were increased for the outer edge heating frequencies to allow for even
temperature distribution across the tape. Better control of microwave heating was
obtained by varying the time interval for the frequency with the highest heating potential.
Variable frequency microwave pultrusion processing of composite materials looks to be

promising in providing an even cure distribution across the profile.

63

Chapter 6. Fiber Reinforcement Effects On Microwave Processing Efficiency
6.1. Material Identification and Purpose of Study

The reinforcing of composites has benefits, which are two-fold. One is that the
fibers absorb the majority of the load when the composite undergoes some kind of tensile
or flexural stress. The second benefit is that the thermal conductivity of the fibers is
typically different than that of the resin, allowing the energy to couple preferentially into
the fiber or resin. The materials used in the study were a Hercules AS4/3501-6
graphite/epoxy composite and a DA409/E250 glass/epoxy composite. The reason for
performing experiments on both systems is to compare and contrast the ability, or
efficiency, of microwaves to process each material.

As was discussed before, the dielectric constant is a measure of the material's
ability to store electrical energy. The dielectric loss factor corresponds to the material's
ability to dissipate electrical energy as heat. The effective dielectric loss factor consists of
two terms: the loss factor due to the dielectric and a portion that deals with the
conductivity. The materials' dielectric properties play an important role in determining
how well radio frequency heating will work. The dielectric loss factor is a function of
temperature and extent of cure. Jow, et a1. document that the dielectric loss factor
increased with increasing temperature until the gelation point was reached. The dielectric
loss factor decreased suddenly during the gel and then continued to decrease as the matrix
fully cured. [24] Springer and Lee examined the microwave interaction with both glass
and graphite fiber reinforced epoxy composites. [25] They listed the dielectric properties
of both material systems, and the graphite/epoxy composite was far superior over the

glass/epoxy - both uncured and cured. [25] If the graphite/epoxy system has a higher

64

dielectric constant, it is expected that it will absorb microwave energy more efficiently
than the glass/epoxy system. For the electrical energy the material system does absorb,
the graphite/epoxy system, with a much higher loss factor, is expected to take that
electrical energy and dissipate it as heat more efficiently. Carbon fibers have a thermal
conductivity value an order of magnitude higher than glass fibers [26], suggesting that
thermal heat transfer will be better than a composite with glass fiber from the inside to the
outside of the composite part. It is then predicted that the graphite/epoxy system will heat

more rapidly than the glass/epoxy system.

6.2. Experiments

The glass/epoxy system was processed in a single-mode microwave applicator.
Three foot, six-ply DA409/E250 prepreg tape samples were prepared. The cavity was
characterized with the material inserted into the Teflon die to find what electromagnetic
modes could be excited (Table 7), followed by heating studies for each mode (Figure 41).
The most efficient heating mode, namely TEm, was chosen to further conduct tests on the
6-ply glass/epoxy prepreg tape samples.

Table 7. Heating Modes for the Cavity with Teflon Die and Glass/Epoxy Prepreg

 

 

 

 

 

 

 

Mode # Mode Type Cavity Length, Lc, mm Probe Depth, Lp, mm
1 TM 011 70 11

2 TB 011 101 13.94

3 TM 111 117 15.75

4 TE 112 131 9.5

5 TB 311 152 16.33

 

 

 

 

 

65

 

 

 

 

 

 

 

 

 

50 L.

451

40 .11

35 1 r

30 . ' ——TBJ11
". ' ——TM111
)—
6 25 " —TE112
5 "w -_—r~n11
r- 20 1.

—Tr.-311

15 «—

10 I

5 1.

0 1 1 1

0 100 200 300 400
tlme,t,sec

 

Figure 41. Temperature versus Time Results for Heating Studies of Stationary Six-Ply
DA490/250 Prepreg at Approximately 50 Watts

The glass/epoxy prepreg was processed in a TB“; mode cavity at 2450 MHz in the
existing pultrusion apparatus. The tape was not pulled through the cavity to see just how
well the material would beat in the microwave environment. The processing time was
varied from one minute to 15 nrinutes. For the 7" cavity, the processing time was
assumed to correspond to the residence time of a differential slice of prepreg. In the fixed
frequency experiments for graphite/epoxy samples (see Chapter 4), the fixed pulling speed
for variable input power experiments was 0.281 filmin. When the glass/epoxy was heated
for two minutes, the equivalent pulling speed to give the same residence time was 0.292

fi/min.

6.3. Results and Discussion

Figures 42 and 43 show the heating results for a six-ply glass/epoxy and graphite/epoxy
prepreg, respectively, at approximately 50 Watts. The centerline temperature at 120
seconds after the power was turned on was 31.38°C for the glass/epoxy system and
41.29°C for the graphite/epoxy system. The extent of cure was measured on DSC for the
glass/epoxy samples. For the sample heated in TE"; mode at 57 Watts for two minutes,
the material had below 10% cure. With an idea of how well the glass/epoxy heated with
microwaves, an attempt was made to pultrude a six-ply sample at 50 Watts and 0.281

ft/min. Figure 44 shows that there was only a 5°C temperature increase over the tape.

 

 

 

 

 

 

 

35
30 .t
25
0
..-

" 20 it —-—midde
% —ouer-middle
g, 15 .1- lUiIidl
.—

10 I
5 ..
0 fl

 

 

0 50 100 150 200
tine,t,sec

 

 

 

Figure 42. Stationary Six-ply Glass/Epoxy Prepreg Heated in TE“; Mode at 57 Watts for
Two Minutes

67

 

 

Temperature, T, C
8

 

 

0 ¢ 1‘ 4 t t 4
O 50 100 150 200 250 2!!) 350

tune, t, soc

 

 

Figure 43. Centerline Temperature for Six-ply Graphite/Epoxy Prepreg Heated in TE";
Mode at 50 Watts and 0.281 ft/min

 

 

 

odor-triads
-—-—— rn'ddle
ouside

 

 

 

 

 

Tomporlun, ‘l', c
a

 

 

O i t + A.
O 100 200 300 400 500
time. t, s

 

 

 

Figure 44. Six-ply Glass/Epoxy Prepreg Heat in TE“; Mode at 50 Watts and 0.281 fi/min

68

Power was increased to 85 Watts for processing the glass/epoxy samples because
power absorption was low. The residence time was also increased in attempt to identify a
point in time where the exotherm of the polymerization reaction would make the
temperature rise more rapidly with time. Figure 45 shows the temperature versus time
curves for points across the tape. The temperature did continue to rise for the span of five
minutes, but the slope of the temperature-time curve did not increase even with the
increase in power. This could be due to poor tuning of the cavity, which would lead to
more reflected power and ineflicient coupling with the material. Extent of cure of the
above prepreg was measured to be approximately 70%. Figure 46 reveals that for similar

power levels, the centerline temperature of graphite/epoxy is 40°C greater than that of the

 

 

 

 

 

 

 

 

 

glass/epoxy sample.
so
45 .t
40 it
w «-
0
o-’ J
2- 3° -——m’ddle
g 25 *’ -—-ouer-niudle
g. 20 1* ——OLISHQ
l- 15 ‘-
10 4+
5 ..
0 i . +
o 100 200 300 400
time, t, soc

 

 

 

Figure 45. Stationary Six-ply Glass/Epoxy Prepreg Heated in TEuz Mode at 85 Watts for
Five Minutes

69

 

 

5‘88

.1
I’

Ternperature,T,C
a a .8 a :8

 

A

O
1
r

 

 

O
.1

 

100 150 200 250

tlmo. t, sec

0
s«

 

 

 

 

 

Figure 46. Centerline Temperature for Six-ply Graphite/Epoxy Prepreg Heated in TE, 12
Mode at 80 Watts and 0.281 ft/min

6.4. Conclusions

Glass/epoxy and graphite/epoxy prepregs were evaluated as to their
reinforcement's effect on the efficiency of microwave heating. The graphite/epoxy system
had a higher dielectric constant, suggesting it would absorb microwave energy more
efficiently than the glass/epoxy system. For the electrical energy that it did absorb, the
graphite/epoxy system, with a much higher loss factor, seemed to dissipate the electrical
energy as heat more efficiently with evidence from the temperature-time curves. The
glass/epoxy will heat to similar temperatures as the graphite/epoxy, but it will require a
longer processing time. It is concluded that the graphite fiber / epoxy heats more
efficiently in a microwave environment and heats more rapidly than the glass fiber / epoxy

system.

70

Chapter 7: On-Line Monitoring

7.1. Possible On-line Monitoring Techniques

Possibilities for an on-line monitoring system are being explored. Possible
variables to be monitored are temperature, dielectric constant, dielectric loss factor, and
extent of cure. Extent of cure cannot be monitored directly. It would be deduced based
on the properties of heat capacity, dielectric constant, and dielectric loss. It would be
possible to correlate these properties with the temperature as long as the chemical reaction
taking place in this process is exothermic. At the completion of the reaction, the exotherm
subsides, indicating completion of cure. However, measuring the temperature of the
material that is moving through the die is difiicult. The degree of cure is related through
the temperature profile of the material with respect to time and the dielectric loss,
dielectric constant, and heat capacity in the batch study. Then, one can position a probe at
the die exit location where values for the dielectric constant and dielectric loss can be
achieved.

The other on-line monitoring method can be the measurement of surface
temperature to ensure complete cure. The assumption is that the material would be heated
from the inside to the surface. If the surface temperature of the exiting material is
predefined as the criteria to denote complete curing, (which incidentally is possible to
arrive at fi'om the batch curing studies with appropriate assumptions and calculations) one
can use simple probes to monitor the temperature of the exiting material. By being able to
observe certain characteristics of the composite during processing, one can get a better

idea on how to control gelation before the material enters the die. Based upon the

71

outcome of the examination of a possible on-line monitoring system, a strategy will be
developed to implement the desired on-line monitoring and control system into the
pultrusion apparatus.

Previous work by Dr. Jinder Jow, et a1. [24], included a monitoring method in a
batch microwave process. The brass cavity dimensions were adjusted to allow resonation
of the 2450 MHz wave. Initial dielectric measurements were taken for the uncured epoxy
sample. The sample was heated with a specified input power, and microwave transparent
fluorotropic probes were used to monitor the internal temperature of the sample. Upon
completion of the cure cycle, the circuit was reconfigured to include the low-power sweep
oscillator in place of the 100 W source in order to begin diagnostic measurements using
the sweep fiequency method. Temperature, dielectric constant, and dielectric loss were
monitored during convective cooling. Then DSC tests were run to find the extent of cure.
From the results of the batch studies, various curves can be generated:

0 Dielectric loss versus temperature at different extents of cure
0 Dielectric constant versus temperature at different extents of cure
0 Dielectric loss versus extent of cure

o Extent of cure versus microwave processing time

72

It is known that both the dielectric constant and the dielectric loss will increase
with increasing temperature. The dipole relaxation time is related to temperature through
an Arrhenius relationship. If temperature increases, the relaxation time decreases and
subsequently increases the dielectric loss factor. Also, the dielectric properties will
decrease with increasing extent of cure. This is because the dipoles become less mobile as
crosslinking increases. Jow found that at approximately 40% cure, the dielectric
properties became less dependent on temperature and more dependent on extent of cure.
[24]

If extent of cure is related through the temperature profile of the material with
respect to time and the dielectric loss, dielectric constant, and heat capacity in the batch
study, a fluorotropic temperature probe can be positioned at the die exit. It can be initially
assumed that if the surface temperature is approximately equal to the consolidation
temperature, then complete cure has occurred. The monitored temperature can give
values for the dielectric constant and dielectric loss factor by interpolating the curves from

the batch study. DSC will again evaluate the extent of cure.

7.2. Pultrusion Model for On-Line Monitoring

A pultrusion model is proposed that will relate the necessary variables to establish an on-
line monitoring system. It begins with a two-dimensional energy balance for a rectangular
profile part in Equation 10, [27] where p, Cp, and k are the density, heat capacity, and

thermal conductivity of the composite, respectively.

73

ar_ 50:07] a
017

+—(k%)+pH+P,; (10)

, V
pC c3c

, :3; ' 25
His the rate of heat generation per unit weight by the polymerization reaction.

Pm’ is the rate of heat generation per unit volume by the absorbed microwave energy.

. 6a
Hz—H 11
at r ( )

where H, is the heat of reaction = 197.5 J/g from DSC.

The Cartesian coordinate plane for the rectangular profile is such that z is in the
pulling direction and x is across the width of the prepreg. Note that in Equation 10, VZ is
the pulling speed of the prepreg through the die. It can be assumed that the heat transfer

in the axial direction is negligible relative to the heat transfer across the prepreg, therefore
making the energy balance one-dimensronal ( 6— : 0 ). The rmcrowave power absorption
2

term in the energy balance is dependent on the processing frequency, dielectric loss factor,

and the electric field strength inside the material.
P,” = nfoeoe"lEml2 (12)
where fo = 2.45 x 109 Hz and so = 8.854 x 10'12 F/m (permittivity of free space). The loss

factor is broken up into two terms, one for the loss factor due to the dielectric, and one for

the electrical conductivity.

N H a
6‘ =6} +5 (13)

0

The magnitude of the electric field inside the material is a complex calculation, and
the power absorption model by Wei [1] gives an accurate account of the power dissipation

of microwave energy inside a composite. If it is assumed that all of the microwave

74

 

incident power is absorbed into the composite, then the electric field strength inside the
material will be similar to the electric field strength inside the cavity. The electric field
strength inside the cavity for the transverse electric (TE) and transverse magnetic (TM)
modes can be calculated by taking the cross product of the vector equations given in
Equations 14-19 for the TE modes and Equations 20-25 for the TM modes below. [19]
The modes in a cylindrical cavity can be modeled by a computer program in order to
establish the magnitude of the electric field at a specified cavity height. This can help by
finding the optimum sample height in the cavity and by being able to visualize where the
“hot spots” will be in the cavity as it heats the pultruded part.

For the TE-modes,

 

E, = —IJ’£:") sin(19)sin(k,z) (14)
E, = —J',(k,r)cos(10)sin(k,z) (15)
E. = 0 (16)
H, = %J',(k,r)cos(l€)cos(k3z) (17)
H, = -l%£’-I($£)-sin(19)cos(k3z) (18)
Hz = %J,(k,r)cos(16)sin(k3z) (19)

75

For the TM-modes,

 

E, = —%—J',(k,r)cos(10)sin(k3z) (20)
_ £1 11(k1') . .

E, .. I k krr sm(119)sm(k,z) (21)

E2 = %J,(k,r) cos(10)cos(k3z) (22)

H = —Ifl£¥2sin(lfl)cos(k3z) (23)

H9 2 —J', (klr) cos(16)cos(k3z) (24)

H = o (25)

where k1: 2x1m/D, k3 = nit/L, k2 = k,2 + k32, and A = 21t/k.

The documented kinetics [28] for the Hercules AS4/3501-6 composite will serve
as a basis for this section. The gel point for the graphite/epoxy system occurs at an extent
of cure of 30% and separates the two prevalent kinetic expressions. When the conversion
is less than 30%, Equation 26 defines the kinetics. Equation 27 gives the kinetics for the
conversion when it is greater than the gel point.

601

E = (k, + kza)(1— a)(B — a) (26)
ii: k,(1-a) (27)
k, = 2.101 x 109 exp(-8-07 >< 10%“) (28)
k, = —2.104 x 109 exp("7-78 x 10%“) (29)

76

k, = 1.960 x 105 exp(-5-66 X 10%“) (30)

where R = 8.314 J/mol K

B = 0.47 = ratio of initial hardener equivalents to epoxide equivalents.

The dielectric properties of the Hercules AS4/3501-6 composite have been reported in the
literature [25]. The effective complex permitivities of the uncured material along and
perpendicular to the fiber direction are 8* = 1-j25000 and 8* = 33.0 - j53.3, respectively.
The real part of the complex permittivity equation is the dielectric constant, while the
imaginary part is the loss factor.

The electrical properties also come into play in the energy balance for the material
system. A proposed energy balance to model the microwave processing of composite
materials [27] can be modified into a pultrusion model. One must take the pulling speed,
Vz, as small incremental steps in the pulling direction, z. Then the chain rule can be
utilized to establish an energy balance equation that is time dependent for temperature and
time dependent for extent of cure. Originally, the derivative of temperature with respect
to time was equated with a two-dimensional dependence of temperature on location in the
composite and a reaction term. Substituting the pulling speed on the left side, the
temperature-time curves are now dependent on the processing speed.

The reaction term is broken down into two terms. The first is the heat of reaction,
which is multiplied by the derivative of the extent of cure with respect to time. This is the
heat generated inside the composite by the polymerization reaction. The second term
takes into account the electromagnetic interaction of the microwaves with the material

system. Pm' is equivalent to the rate of heat generation per unit volume by the absorbed

77

microwave energy. A power absorption model must be employed here to accurately
calculate the amount of heat generated. Pm' is a fimction of the processing frequency,
permittivity of free space, which is a constant, the square of the electric field strength
inside the material, and the effective dielectric loss factor. A five-parameter power
absorption model was proposed by Wei, et. al in order to simulate the "power dissipation
inside a fiber reinforced composite during microwave processing in a tunable resonant
cavity." [1] The five parameters involved in the model are the power dissipation due to the
electromagnetic (EM) waves from the side and the magnitudes and the polarization angles
with respect to the fiber direction of the incident TEM waves at the top and bottom of the

composite.

78

7.3. Conclusions on On-Line Monitoring

It is possible to integrate an on-line monitoring system and process control
equipment to a microwave pultrusion line. First, the curves from the batch studies, i.e.,
dielectric data and temperature-time curves, will be obtained. Then, a two-dimensional
energy balance with heat transfer, polymerization reaction, and microwave power
absorption terms will be employed to find the temperature dependence on the profile size,
dielectric properties, and pulling speed. Next, electric and magnetic fields should be
modeled by computer software to calculate where the "hot spots" will be inside the cavity
for the various transverse electromagnetic modes that can be excited. Finally, a previous
microwave power absorption model [1] will be used to define the electric field magnitude
inside the composite. Computer simulations could then model what would happen in a
microwave pultrusion apparatus with a chosen set of processing variables. The on-line
monitoring system would allow the operator to easily control the process variables to

optimize the pultruded product characteristics.

79

CONCLUSIONS

The microwave pultrusion apparatus was assembled. The 17.78 cm inner
diameter, tunable, cylindrical, batch, brass microwave cavity was characterized by finding
the different electromagnetic modes for different cavity dimensions. Heating studies were
conducted for each of the modes to find how rapidly each could heat the prepreg and
where the electric field magnitude was large. Fixed-frequency experiments at 2450 MHz
were run in TE“; mode. The microwave input power was varied from 40 to 100 Watts,
and the pulling speed was varied from 0.281 to 1.17 ft/min. Extent of cure of the
graphite/epoxy composite was established as a function of input power and pulling speed
for a microwave pultrusion apparatus. Extent of cure increased with an increase in
microwave power input and decreased with an increase in pulling speed. At 100 Watts
input power and a pulling speed of 0.281 ft/min, the center of the six-ply graphite/epoxy
prepreg was 98% cured.

Tensile strength and modulus were relatively unaffected by the change in process
variables. Flexural strength measurements revealed there to be maxima in strength and
modulus for certain processing conditions. Too high of power or too slow of pulling
speed resulted in a brittle matrix. Low input power or fast pulling speeds gave a product
that had a relatively uncured matrix and low interfacial bond strength.

Variable frequency processing provided an even temperature distribution across
the prepreg. The cavity was characterized by measuring reflected power versus frequency
for various cavity lengths and coupling probe depths. Frequencies with the highest heating

potential and similar power output were selected to perform curing studies. Thicker

80

samples provided more efficient heating in the variable frequency study by absorbing a
greater percentage of the incident power. Time intervals were increased for the outer
edge heating frequencies to allow for even temperature distribution across the tape.
Better control of microwave heating was obtained by varying the time interval for the
frequency with the highest heating potential. Variable frequency microwave pultrusion
processing of composite materials looks to be promising in providing an even temperature
distribution across the profile.

Glass/epoxy and graphite/epoxy prepregs were compared and contrasted as to
their effect on the efficiency of microwave heating. The graphite/epoxy system had a
larger dielectric constant, suggesting it would absorb microwave energy more efficiently
than the glass/epoxy system. The graphite/epoxy system, with the larger loss factor,
converted dissipated the electrical energy as heat more efficiently with evidence from the
temperature-time curves. It is concluded that the graphite/epoxy would heat more rapidly
than the glass/epoxy system.

Possibilities for an on-line monitoring system were explored. Possible variables to
be monitored are temperature, dielectric constant, dielectric loss factor, and extent of cure.
The degree of cure is related through the temperature profile of the material with respect
to time and the dielectric loss, dielectric constant, and heat capacity in the batch study.
Then, one can then position a probe at the die exit location where values for the dielectric
constant and dielectric loss can be achieved. A pultrusion energy balance in conjunction
with a five-parameter power absorption model could help provide a basis for an on-line

monitoring system and/or a process control program for microwave pultrusion processing.

81

RECOMMENDATIONS

Microwave processing was applied to a continuous manufacturing process for
composite materials. After studying glass fiber/vinyl ester composites, Min Lin suggested
that in future microwave pultrusion research, one should include other resin/fiber systems
so the technology can branch out to new applications. He also recommended improving
the system for high-speed pultrusion by modifying the cavity design, installing a new
pulling mechanism, and integrating a high power source. Finally, he suggested researching
new microwave transparent die materials and combining a part-shaped cavity and
pultrusion die, where the walls of the die would be the microwave applicator. [29]

Experimental data on the microwave pultrusion of graphite/epoxy and glass/epoxy
systems was collected. Higher RPM pulling rollers were placed in the line to increase the
maximum pulling speed to 240 cm/s from 3 cm/s. Heating and curing profiles of the
pultruded material were studied at different processing conditions. To find the optimum
operating conditions for microwave pultrusion processing, the relation between processing
rate, power, and extent of cure was examined. A variable frequency power source capable
of 150 Watts input power was used to achieve a more even temperature and cure

distribution across the prepreg.

82

It is recommended that the following issues concerning the pultrusion project be
addressed in firture work:

0 A higher microwave power source is needed to process larger-profiled parts and
to increase processing speeds.

0 New mechanical pulling equipment capable of the higher line speeds.

0 New die material that has a longer life and is more transparent to microwaves
than both Teflon and ceramic.

o Thermally heated die at die exit of microwave cavity to see if single-mode
applicator is a viable alternative to a radio frequency preheater.

0 Parallel plate dielectric analysis machine at die-exit to assist in on-line
monitoring of material properties.

0 Combine part-shaped cavity ideas with a pultrusion die to make the die walls the
actual microwave applicator.

0 Computer program relating all process variables and material properties for
inclusion in a process control program and/or on-line monitoring system.

0Modification of an existing larger single-mode applicator to operate at 915 MHz
for the continuous processing of materials. The larger cavity will be usefirl in the high
power experiments.

It is proposed that the existing microwave technology available at Michigan State
University be integrated into a Vermont Pultrusion Test System available from Vermont
Instrument Company, Inc. The laboratory set-up would provide a means by which to
compare and contrast conventional and microwave pultrusion processing, or to study the

two techniques in conjunction. By selecting a resin system, one may study the optimum

83

cure conditions, microwave input power, die oven temperature, and pulling speeds. A
polyester or vinyl ester resin would be mixed with the appropriate catalysts and additives
in the disposable container. The glass rovings would be led through the pultrusion line to
ensure proper alignment of the fibers during processing. The rollers would be controlled
electronically and have a maximum pulling speed of 75 cm/min (2.46 fl/min). The VICO
bench pultruder would produce 3 mm rods. The experiment station would include glass
rovings that pass through a ceramic ring, a resin fiber-wetting system complete with
disposable resin bath containers and roller bars, an aluminum oxide ring which would act
as the prefonner, and a VICO preheater, bench pultruder, and postcure oven. A VICO
mechanical torsion tester would be available to measure the product's shear modulus and

its fracture energy in shear.

84

BIBLIOGRAPHY

85

10.

11.

12.

13.

14.

15.

16.

BIBLIOGRAPHY

. Wei, J. and MC. Hawley, and J. Asmussen, Jr., "Power Absorption Model For

Microwave Processing of Composites in a Tunable Resonant Cavity," ASM/E SD
ACCE, Chicago, (1992).

Bruce, R.W., Material Research Society Symposium Proceedings, 124, 3 [Microwave
Processing of Materials], (1988).

 

Wei, 1., Dissertation for the Doctor of Philosophy, Chapter 7, Department of Chemical
Engineering, Michigan State University, (1992).

. Wei, J ., Shidaker, T., and Hawley, M.C., “Recent Progress in Microwave Processing

 

of Polymers and Composites,” TRIP, Vol. 4, No. 1, pp. 18-24, (1996).
Hexcel Carbon Fibers, Product Data Summary, Hercules AS4/3501-6, (1998).

Adhesive Prepregs for Composites Manufacturers, Product Data Summary,
DA409/E250, (1998).

Munson-McGee, Opportunities for Innovation: Polymer Composites, NIST GCR
90-577-1, pp. 121 (August, 1990).

Pat. 4,792,772, I. Asmussen (1988).

L. VanKoughnett and J. G. Dunn, J. Microwave Power. _8_ (1), 101 (1973).

Pat. 5,406,056, M. C. Hawley, J. Wei, V. Adegbite, and Y. Chang (1995).

Pat. 2,245,893, A. Cooper and J. M. Methven (1990).

Ghaffariyan, SR. and J. M. Methven, Mat. Res. Soc. Symp. Proc, 189, 135 (1991).
Pat. 5,470,423, M. C. Hawley, J. Wei, V. Adegbite, and M. Lin (1995).

Meyer, Raymond W., Handboole Pultrusion Technology, Chapman and Hall,
New York, (1985).

Martin Pultrusion Group. Pultrusion Equipment Buyer’s Guide. [Online] Available
http://www.ne-ohio.net/pultrude/equip.htm, May14, 1998.

 

P.K. Mallick, Fiber-Reinforced Composites: Materials, Manufacturing,_and Design,
Marcel Dekker, Inc, New York, 1988, pp. 184-185, 201-204, 345-351.

86

17.

18.

19.

20.

21.

22.

23.

24.

25.

Lin, M. and MC Hawley, "Preliminary Tests of the Application of Continuous
Microwave Technique to Pultrusion," 38th SAMPE International Smposium, (1993).

Shanku, R., Vaughan, J .G., and J .A. Roux, “Dielectric and Thermal Cure
Characterization of Resins Used in Pultrusion,” 42"d International SAMPE
Symposium, Pre-publication copy, (1997).

Montgomery, C .G., Technique of Microwave Measurements, McGraw-Hill Book
Company, Inc, (1947).

Hull, D. and T.W. Clyne, An Introduction to Composite Materials, Cambridge
University Press, pp. 158-171, (1996).

 

ASTM Standard D 790 - 84a: Standard Test Methods for Flexural Properties of
Unreinforced and Reinforced Plastics and Electrical Insulating Materials.

Swanson, S.R., Introd i n Desi 11 an Anal sis with Advanced Com osite
Materials, Prentice Hall, pp. 91-97, (1997).

Qiu, Y. and Hawley, M. C., “Computer Controlled Variable Frequency Microwave
Processing of Graphite/Epoxy In A Single Mode Cavity,” DALI, 1998.

low, J ., Hawley, M. C., et. al., "On-line Measurement of Temperature- and Cure-
dependent Dielectric Properties During Single Frequency Microwave Curing and
Comparison with Thermal Curing of Epoxy Resins", Conference On Emergiiig
Technologies In Magda]; _A_iC_hI;Z, 1987.

Lee, WI, and Springer, G.S., "Interaction of Electromagnetic Radiation with Organic
Matrix Composites," Journd of Cgmposite Materials, Vol. 18, pp. 357-386 (1984).

26. Han, CD, and D.S. Lee, “Development of a Mathematical Model for the Pultrusion

Process,” Polymer Engineering and Science, Vol. 26, No. 6, pp. 393-404, (1986).

27. Wei, J. and MC. Hawley, "Modeling and Controlling During Microwave and Thermal

28.

Processing of Composites, " WSjmposium Proceedings, 269, 439. (1992).

Lee, W.I., Loos, AC, and Springer, G.S., Journal of Composite Materials, Vol 16,
pp. 510, (1982).

29. Lin, M., Master’s Thesis, Chapter 6, Department of Chemical Engineering, Michigan

State University, (1995).

87

"I1111111111111111111115