Cure cycle optimization of

composites by dielectric
measurements
M. UNGARISH, R. JOSEPH, J. VITTOSERand S. KENIG

(Armament Development Authority, Israel)

Cure cycle monitoring by dielectric loss factor was investigated for prepreg
composites based on epoxy, epoxy-phenolic, polyimide and phenolic
matrices. Dielectric loss measurements were carried out to establish optimal
cure conditions for the various polymer composites with respect to testing
rates, curing temperature and timing for pressure application. For each
polymer composite studied, initial laboratory experiments were carried out
in a temperature-programmable oven to determine the onset of polymer
flow, the temperature range where resin viscosity is appropriate for pressure
application and the duration of the cross-linking reaction. Subsequently,
autoclave processing was conducted using similar conditions accompanied
by in situ dielectric loss monitoring. Dielectric cure monitoring of room
temperature aged, epoxy-based prepreg indicated that curing of aged
prepregs is possible provided a flow region exists. Flexural properties of
specimens prepared from composite laminates processed under optimized
conditions substantiated the adequacy of in situ dielectric loss measure-
ments for cure monitoring of prepreg composites.

Key words: composite materials; cure; dielectric loss factor; processing

conditions; flow range; cross-linking

High performance thermosetting composites are
usually supplied in a 'prepreg' form where the fibres
(short or continuous) are pre-impregnated with the
appropriate resin and curing agent in a 'B-stage' state.
To slow down the cross-linking reaction, the prepreg is
stored under low temperature conditions. Prior to
processing, the cold prepreg is warmed up to ambient
conditions, placed in a proper mould and then cured
under appropriate temperature and pressure condi-
tions. For optimal mechanical properties pressure is
applied at a predetermined temperature to squeeze the
excess resinous material and to improve the fibre/
matrix wetting and adhesion prior to complete
moulding and cross-linking. It was found that in most
cases pressure should be applied when the viscosity of
the polymerizable resin is at a minimum level.
Dielectric cure monitoring has been widely recognized
as a viable method for investigating changes occurring
in thermosetting systems during the polymerization
process. Being one among a few techniques available
for studyinl~ molecular properties in both liquid and
solid states l, it has been used for analysing complex
resin systems and monitoring cure response of materi-
als under the actual production environment 2. The
range of systems studied is wide 3-5 with an increasing
tendency towards the use of microdielectrometry 6'7. In
0010-43611901060481-06 ~) 1990 Butterworth-Heinemann Ltd
COMPOSITES.VOLUME 21. NUMBER6. NOVEMBER1990 481
spite of the variety of experimentation on the subject, it
has been pointed out that data of practical value for
cure monitoring and control are scarce 1.
The theory of dielectric measurements is well
established TM.Typically, the loss factor, expressing
changes in the dissipated energy when a capacitor
holding the material of interest is subjected to an
externally imposed alternating electric field, is
recorded during a cure cycle. Since the ability of the
polar groups to adjust to the alternating polarity of the
field is related to the freedom of movement on the
molecular level, changes in the physical and chemical
state of the resin are reflected in the value of the loss
factor.
As a resin is subjected to heating at a defined rate, the
curve of the loss factor vs. time exhibits two peaks2'3:
the first one is attributed to solid/liquid transition in the
material, which softens and begins to flow. The second
peak appears when the cross-linking reaction becomes
dominant, overtaking the tendency of the losses to
increase as the temperature rises. Thus, chemical
changes resulting in cross-linking bring about a
decrease in the loss factor. Consequently, from the loss
factor curves, it is possible to detect the stage of resin
flow and hence determine when to apply pressure.
 200 500
Furthermore, one can extract valuable information on o Temperature
the duration and kinetics of the polymerization 175 450
reaction.
400
The purpose of the present work is to study the dielec- 150
350
tric loss factor for different families of materials and 125
define their characteristics in an attempt to establish 300
the conditions for pressure application during curing in loo 250
an industrial autoclave environment and thus improve x
200
the final composite quality. 75
150
5O
EXPERIMENTAL 100
25
Material systems 5O

Four different polymer composite prepregs were
studied. The first was composed of a 120°C curing
epoxy and glass fabric (Narmco 3203/1581). The second
was an epoxy-phenolic matrix and glass fabric (Hexcel
F161/1581) cured at 160°C. The third consisted of a
polymide resin and unidirectional graphite fibre (Hitco
V378A) with a maximum curing temperature of 185°C.
The fourth prepreg system was a combination of
phenolic resin and carbon fabric (Hexcel 4C1008),
recommended for curing at 165°C. Prior to use, the
cold prepregs were warmed up to ambient conditions in
their sealed containers, to avoid moisture pick-up.
Specimen preparation and testing procedure
For dielectric measurements, two plies of prepreg for
each composite were isolated by alternating layers of
Tedlar R and glass fabric from two metallic plates
measuring 20 x 30 mm. The arrangement served as the
capacitor for the dielectric measurements. To reduce
external interference, the whole capacitor was wrapped
in aluminium foil. The capacitor was connected to a
Hewlett-Packard 4262A LRC meter which measured
the dielectric loss factor and capacitance during cure
monitoring. A frequency of 1 kHz was employed which
was found to suit the investigated prepreg composite
systems. Other frequencies were examined, but failed
to distinguish between the flow range and the reaction
peak.
Prior to in-process cure monitoring in an autoclave,
laboratory measurements were performed in a
temperature-controlled oven to study the dielectric
behaviour as a function of heating rates and a variety of
cure cycles. Subsequently, composite laminates were
prepared in an autoclave under optimal conditions
using the dielectric monitoring system.
I I I I ] I I 0
25 50 75 100 125 150 175 200
Time (min)
Fig. 1 Loss f a c t o r (D) variation for curing epoxy/glass prepreg
( N a r m c o 3203/1581) in a l a b o r a t o r y s e t - u p at a h e a t i n g rate o f
1oc m i n -1
plateau resulting from continuing flow and decreasing
viscosity. At 98°C the loss factor resumes its steep rise.
Thus, a 'flow range' or 'softening range' between
82-98°C is defined. A second peak was obtained at
114°C followed by a sharp drop in loss factor values.
This peak can be attributed to the polymerization
reaction. In the curing cycle depicted in Fig. 1, the
temperature is held at 12&C for 30 min prior to
cooling. It is evident that the drop in the loss factor
values is very rapid and in only a few minutes the curve
of the loss factor (D) vs. time (t) reaches an asymptote
indicating termination of the cross-linking reaction.
Following the initial oven experiment, an autoclave
was employed to prepare samples for mechanical
properties testing, and monitoring the cure response by
dielectric measurements. Heating was carried out at
I°C min -t to 85°C, where a 20 min soak time was
allowed. Pressure was raised to 310 kPa starting at
82°C. A final dwell time of 60 min was used at 124°C -
prior to cooling. The cure cycle is presented in Fig. 2.
The values of D during the 85°C dwell show a slow
decrease, confirming that flow takes place in this range
as expected from the preceding laboratory experiment.
The polymerization peak appears at a temperature of
107°C compared with 114°C. This can be attributed to
the additional dwell at 85°C. The cross-linking reaction
200 200
o Temperature
• D x1000
175
150 50

Mechanical properties were evaluated by three-point o

125
bending according to ASTM D-790.
too 00!
RESULTS AND DISCUSSION ~- 75
o.1
Epoxy~glass composite prepreg
5C 5O
A typical loss factor curve for the 120°C curing epoxy
composite is depicted in Fig. 1. The composite capaci- 25
tor was subjected to a constant heating rate of l°C I I I I f I I
min -1 in a temperature-programmable oven. Tempera- 25 50 75 100 125 150 175 200
ture variation during the cure cycle is also shown, using Time (min)
the same time scale for convenience. The first peak due Fig. 2 Loss factor (D) variation for curing epoxy/glass prepreg
to the onset of flow appears at 82°C, followed by a ( N a r m c o 3203/1581 ) in an autoclave

482 COMPOSITES. NOVEMBER 1990

 200 900 to 122°C compared with the previous slower autoclave
o Temperature ,~
cure cycle, as anticipated from the laboratory experi-
.ox,ooo /1 _8oo ment depicted in Fig. 3. As evident, the cross-linking
reaction was carried out to completion within 30 min as
predicted by the laboratory experiment.
The mechanical properties obtained as a result of this
-- 500 o~
100 \ second autoclave cure are superior to those achieved in
the first: a flexural strength of 580 MPa vs. 470 MPa
-- 4 0 0
and a modulus of 23 GPa vs. 18 GPa, respectively.
-

- 0 - 200 Epoxy-phenolic~glass composite prepreg

25 -100 The second system studied was the epoxy-phenolic
0 I I I I I 100
one. Laboratory oven experiments were first per-
10 20 30 40 50 60 70 80 90 formed to establish the cure conditions. Results for two
Time (min) different heating rates, at I°C min-1 and 2°C min-1, are
Fig. 3 Loss factor (D) variation for curing epoxy/glass prepreg shown in Figs 5 and 6, respectively. A local minimum in
(Narmco 3203/1581 ) in a laboratory set-up at a heating rate of the loss factor is detected in the 55-93°C range, indicat-
2oc min -1 ing softening and minimum visosity. The polymeriza-
is carried out to completion after several minutes tion peak is exhibited at 152°C. As can be seen, results
following the second peak, indicating that a 60-min for both heating rates are nearly identical. The only
dwell is too long and confirming the initial laboratory difference is the duration of the softening range as a
result of the different heating rates.
results.
200 6000
To compensate for the dwell during pressure applica-
o Temperature
tion and in an attempt to speed up the cure cycle, an 175 • D x 1000
increased heating rate was used. Fig. 3 shows the loss 5000
factor vs. time curve obtained in a laboratory oven 150
experiment conducted with a constant heating rate of
4000
2°C rain-1. In this case a 87-103°C flow range is °~,~125
observed. The second peak (polymerization peak)
appears at 118°C, indicating the importance of heating 1oo / 3000
o

X
time. The asymptotic D values are reached 30 min after
attaining the peak, indicating that such a period should
° / i
2000
be allowed during the dwell at 118°C. One can 50
conclude here that a faster heating rate raises the
1000
temperature of the polymerization peak, indicating that 25
the main reaction occurs, probably, at a somewhat
0 I I I
higher temperature. 0 50 100 150 200 250 300

Based on these results a faster autoclave cure was used. Time (min)
Results of dielectric monitoring are given in Fig. 4. Fig. 5 Loss factor(D)variation forcuring epoxy-phenolic/glass
During the dwell at 85°C D values remain constant, prepreg (Hexcel F16111581)in a laboratoryset-up at a heating
indicating that flow occurs. At this range pressure was rate of1°C m i n - I
applied. The polymerization peak was shifted upwards
200 6000
200 500 o Temperature
175 AD xl000
o Temperature - 450 --5000
175
• D x 1000
- 400 150
150
-350
G --4000
U °125
o
125
- 300 o o
o ~ 100 --3000~
o
X
~ 100 -250 x Q
& 75
X 75 200 ~ E 2000
E
150 50
50
100 25
, , ?oo
25 50
0b I
5O 100 150 200 250 300
I ; I 1 I I I 0
25 50 75 100 125 150 175 200
Time (rain)
Time (min)
Fig. 6 Loss factor (D) variation for curing epoxy-phenolic/glass
Fig. 4 Loss factor (D) variation for curing epoxy/glass prepreg prepreg (Hexcel F161/1581) in a laboratory set-up at a heating
(Narmco 3203/1 581 ) in an autoclave rate of 2 ° C min -1

COMPOSITES. NOVEMBER 1990 483

The dielectric losses attain an asymptote approximately 200
o Temperature
20O0

90 min after the occurrence of the polymerization peak, • D x 1000

175 1750
indicating completion of the cross-linking reaction. It
could be concluded that the chemical reaction is mainly 150 1500
temperature-dependent. The different heating rates
o
have little effect on reaction onset or duration. In the 125 125o
epoxy-phenolic matrix system, the process window is
wider than the 120°C epoxy one and, consequently, 3 1o0 1000 x
P
processing conditions could be chosen based on second-
E 75
ary considerations. Based on these results, the auto- oJ
F-
75O

clave cycle employed consisted of heating at a rate of

5O 500
2°C min -1 to 60°C followed by a lower heating rate of
I°C min -1 to 93°C, which is the upper limit of the 25 250
softening range as shown above. Pressure of 207 kPa
was applied at 75°C, which is well within flow range. I I I I I I I I 0
50 100 150 200 250 300 350 q00 q50 500
Finally, fast heating to 160°C and a dwell of 90 min at Time (min)
this temperature were used. Fig. 8 Lossfactor(D)variation forcuring polyimide/graphite
prepreg (Hitco V378A) in a laboratory set-up at a heating rate of
Fig. 7 shows the results obtained during monitoring of Ioc rain -~
this cure cycle. In accordance with the laboratory might not suffice for pressure build-up in a large
experiments, the D values remain constant in the autoclave. Consequently, an extended softening range
softening range where pressure is applied. The reaction is recommended. This may be achieved by employing
peak appears immediately at the beginning of the final lower heating rates (a longer time span from the cure
dwell, as required, and an asymptotic dielectric loss is start to the reaction peak). Fig. 9 shows the results
attained after 90 min, representing cross-linking obtained with a heating rate of 0.75°C min-~. As
termination. Specimens from laminates prepared using evident, the softening ranged has been extended to
the latter curing conditions gave flexural strength of over 30 min, remaining in almost the same temperature
570 MPa and a modulus of 25 GPa. These values are in 200 2000
agreement with the best data published by the manu- O Temperature
• D x 1000
facturer. 175 -1750

150 -1500
Polyimide/graphite composite prepreg ?
125 1250
Fig. 8 describes the changes in the dielectric loss factor o
o

obtained in a laboratory oven experiment with a D

100 1000
o

constant heating rate of l°C min -~. An apparent x

softening range between 100-124°C appears as a 75 750

shoulder in the curve, followed by a steep rise in D
values to the polymerization peak at 186°C. An asymp- 50 - 500

tote in the dielectric loss factor is observed 90 min after

the peak, demonstrating completion of the cross- 25' -- 250

linking reaction. I I I 0
0 50 100 150 200 250 300 350 400 450 500
For polyimide matrices, high pressures in the range of T i m e (rain)
689-1034 kPa are needed for consolidation. As indi- Fig. 9 Loss factor (D) variation for curing polyimide/graphite
cated by Fig. 8, a softening range lasting only 15 min is prepreg (Hitco V378A) in a laboratory sat-up at a heating rate of
obtained under a heating rate of 1o C m i n -1, which 0.75°C min -1
200 6000 200 4000
o Temperature
o Temperature
u D x 1000
175 _ • D x 1000 175 3500
5000

150 150 3000

4000 o ~
o
~ 125 125 2500 o
oo
3 100 3000~ ~ 100 2000
x

~- 75 ~ 75 1500
2000
5O 50 1000
1000 25
25 500

I I I I I 0
0 50 100 150 200 250 30O 0~ 50 100 150 200 250 300 350 400 450 500
Time (min) Time (rain)

Fig. 7 Loss factor (D) variation for curing epoxy-phenolic/glass Fig. 10 Loss factor (D) variation for curing polyimide/graphite
prepreg (Hexcel F161/1581 ) in an autoclave prepreg (Hitco V378A) in an autoclave

484 COMPOSITES. NOVEMBER 1990

range of 97-124°C while the reaction peak occurs at the 150 1000
Ageing time (days)
same temperature as before. A0
• 23
Based on the laboratory data, an autoclave cure cycle 125 [] 63
• 250 800
was devised where the pressure is applied at 105°C 0 508
during 30 min at this temperature. A dwell of 90 min at ~I00
186°C was used prior to cooling. The dielectric loss 6oog
curve using the above conditions is given in Fig. 10.
~ 75
Flow is indicated by the constant loss factor values X

during the first dwell, as required for pressure applica- q00 c~

tion. The polymerization peak appears at 186°C and ~ 5o
k-
the reaction is completed, as indicated by the asymp-
totic behaviour of the loss factor curve. In this case, an 2OO
25
average flexural strength of 770 MPa and a modulus in
the range of 60-69 GPa were obtained, in good agree- --o I ol
ment with the manufacturer's specifications. 10 20 30 40 50 60 70 80 910 100
Time (rain)
Phenolic~carbon composite prepreg Fig. 13 Loss factor (D) variation for epoxy/glass prepreg
(Narmco 3203/1581 ) aged for various times at room temperature
Fig. 11 depicts the loss factor vs. time curve obtained in
a laboratory oven experiment for the carbon/phenolic 0.5°C min-1. The resulting flow range was wide (150
prepreg. An apparent short flow range between min). Pressure was applied at 85°C. The polymerization
80-96°C is visible, when a heating rate of 2°C min-I is peak takes place at 162°C as was the case for the
used. A polymerization peak occurs at 162°C. heating rate of 2°C min-]. A flexural strength of
310 MPa and a modulus of 21 GPa were obtained,
Fig. 12 depicts the results of dielectric cure monitoring
substantiating the choice of processing conditions used.
in an autoclave using a lower average heating rate of

2001 •° TemperatureD
x 1000 -2500 Curing of aged prepreg composites
To complement the present study and demonstrate the
175
1150 -2000 usefulness of dielectric cure monitoring and curing
control, aged specimens of epoxy-based prepreg
(Narmco 3203/1581) were characterized. Ageing was
~'~ 125~
[ 1500 ! carried out at room temperature for various periods of
time. As can be seen from Fig. 13, the maximum D
values decrease with ageing time compared with the
unaged case. However, samples aged to 63 days exhibit
I-
1 500
a softening range at 82-98°C and a polymerization peak
at 120°C, the same temperatures as the unaged
prepreg. Samples aged for longer times than 90 days
5250 ~ L ~ I i I 1 did not show a softening temperature and their poly-
, , , I , gO merization peak was shifted to lower temperatures.
00 50 100 150 200 250 300 350 400 The above experimental results indicate that dielectric
Time (min) cure monitoring of aged epoxy prepregs could be used
Fig. 11 Loss factor (D) variation for curing phenolic/carbon for curing control of partially cured prepregs provided a
prepreg (Hexcel 4C1008) in a laboratory set-up at a heating rate softening range still exists.
of 2°C min -1
200 1250
o Temperature CONCLUSIONS
• D x 1000
175
The dielectric loss factor variation during curing of
1000
15C epoxy, epoxy-phenolic, polyimide and phenolic
composite prepregs exhibit a common behaviour.
~ 12E
750
Upon heating the loss factor approaches a constant
o value representing the flow range where pressure
X should be applied to consolidate the composite. The

°/ / \ 500 ~ flow range precedes the polymerization peak, which is

~- 75 followed by a second plateau indicating termination of
1- the cross-linking reaction. By changing the heating rate
50 I
250 and maximum curing temperature the flow range could
be varied to suit the duration of pressure build-up.
Good agreement has been demonstrated between the
0 ~ I I 0
0 50 100 150
200 250 300 350 400 450 dielectric loss factor curves obtained in a laboratory
Time (min) controllable oven and in situ monitored results
Fig. 12 Loss factor (O) variation for curing phenolic/carbon recorded in an autoclave. Thus, a few relatively simple
prepreg (Hexcel 4C1008) in an autoclave and inexpensive laboratory experiments could be

COMPOSITES. NOVEMBER 1990 485

performed to establish the appropriate heating rates
and temperature rise for optimal final mechanical
properties. Since the variation of the loss factor
depends, among other factors, on the initial state of
cross-linking of the prepreg composite, the possibility
arises to use aged prepregs that have already been
partially cross-linked for component preparation.
ACKNOWLEDGEMENT
The authors wish to acknowledge the contribution of
Mr Y. Zur for performing the experimental work and
the technical assistance of Mr R. Shavit.
486 COMPOSITES. NOVEMBER 1990
REFERENCES
1 Kranbuehl, D.E. Developmentsin ReinforcedPlastics, vo15
(Elsevier, London and New York, 1986)pp 181-202
2 Banmgartner, W.E. and Ricker, T. SAMPEJ 19 No 4 (1983)p 6
3 Lawless,G.W. PolymerEngng Sci 20 No 8 (1980)p 546
4 Venkataswamy,K. and Waack, R. Plastics Engng (February
1988) p 43
5 Holmes,B.S. JAppl Polym Sci 35 (1988)p 1399
6 Senturia, S.D., Sheppard, N.F., Lee, I-LL. and Marshall, S.B.
SAMPEJ 19 No 4 (1983)p 22
7 Bistrap, W.W. and Senturia, S.D. PolymerEngng Sci 29 No 5
(1989) p 290
AUTHORS
The authors are with R A F A E L , A r m a m e n t Develop-
ment Authority, PO Box 2250, Haifa, 31021, Israel.