ARTICLE IN PRESS

POLYMER
TESTING
Polymer Testing 26 (2007) 803–813
www.elsevier.com/locate/polytest

Test Method

Temperature-dependent elastic moduli of epoxies measured by

DMA and their correlations to mechanical testing data
Shiqiang Denga,, Meng Houb, Lin Yea
a
Centre for Advanced Materials Technology, School of Aerospace, Mechanical and Mechatronic Engineering,
The University of Sydney, NSW 2006, Australia
b
Cooperative Research Centre for Advanced Composite Structures, 361 Milperra Road, Bankstown, NSW 2200, Australia
Received 26 March 2007; accepted 9 May 2007

Abstract

Measurements of temperature-dependant elastic moduli of two cured epoxy systems and two silica–epoxy nanocomposites
were conducted using dynamic mechanical analysis (DMA) methods with different loading fixtures. Comparisons of the results
were made with those obtained from tests with different loading modes and from mechanical testing results under various
temperatures. Discrepancies were found between the temperature-dependant elastic moduli measured by DMA with different
loading modes and varying test conditions, such as machine compliance, drive amplitude, loading frequency and contact
stresses, although the specimen-to-specimen deviations were relatively small between specimens tested with the same fixture
and similar test parameters. However, the elastic moduli obtained from DMA measurements can be used to represent the
temperature dependence of elastic modulus once corrections are made based on the static mechanical testing data obtained at
ambient temperature. Although there were discrepancies in magnitude between the original temperature-dependant elastic
moduli determined by DMA and those measured by mechanical tests, very good agreement was found for the two epoxy
systems and two silica–epoxy nanocomposites after the DMA data were shifted.
r 2007 Elsevier Ltd. All rights reserved.

Keywords: Dynamic mechanical analysis (DMA); Temperature-dependant modulus; Epoxy

1. Introduction joining various engineering structures. Most cru-

cially, structural components in relation to the use
The temperature-dependant elastic modulus of a of epoxies are commonly utilized under various
polymer material such as cured epoxy resin is one of temperature conditions. Therefore, the temperature-
the most important material parameters in structur- dependant moduli of a cured epoxy resin are
al analysis and design, especially for stiffness-based important values to be determined using reliable
design, as epoxies are often used as matrix materials measuring methods. Mechanical tests are often
for fibre-reinforced composites or as adhesives for carried out under various temperatures to determine
Corresponding author. Tel.: +61 2 93512153; temperature-dependent properties but such ap-
fax: +61 2 93513760.
proaches, although straightforward and reliable,
E-mail address: [email protected] are costly in both materials and time, and often
(S. Deng). limited to selected temperatures.

0142-9418/$ - see front matter r 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.polymertesting.2007.05.003
 ARTICLE IN PRESS
804 S. Deng et al. / Polymer Testing 26 (2007) 803–813

Dynamic mechanical analysis (DMA), or dy- ture in order to incorporate the temperature-
namic mechanical thermal analysis (DMTA), is a dependant elastic moduli into finite-element analysis
very versatile technique which can provide a models. However, although DMA may provide a
convenient and sensitive testing system for rapid simple approach for obtaining the moduli at
determination of thermo-mechanical properties of temperatures other than room temperature, the
polymers and polymer-based materials as a function validity of such manipulation obviously needs to be
of frequency, temperature or time, using only small assessed carefully.
amounts of the material. Measurements using In the present study, the temperature-dependent
DMA can be conducted either by applying a small moduli of two cured epoxy systems and two
oscillating displacement to a specimen to measure silica–epoxy nanocomposites were experimentally
the resulting force or by applying a periodic load to measured by DMA with different loading fixtures
measure the resulting deformation or by applying a and by mechanical tests at various temperatures. The
constant load (or a displacement) to obtain creep main objectives were to attain a better understanding
(or relaxation) data. Commonly, DMA can also of the validity of the elastic modulus measured by
provide ramp loading which is similar to that DMA and to establish possible correlations between
offered by a material testing machine. A number DMA and mechanical testing data.
of material properties can be measured by means of
DMA, such as the glass transition temperature (Tg), 2. Experiments
the coefficient of thermal expansion (CTE) and,
most often, visco-elastic properties with the help of 2.1. Materials
various fixtures such as single/dual-cantilever, three-
point bend, tension/compression and shear sand- Two epoxy systems were used in the current
wich [1]. Conventionally, DMA measures the study, HexFlow RTM6 (Hexcel, USA) and Ara-
storage (elastic) modulus, the loss (viscous) modulus ldite-F (Ciba-Geigy, Australia). The former is a
and damping. Depending on the loading mode, the monocomponent epoxy resin which was specifically
elastic modulus may represent shear, tensile or developed to fulfil the requirements of the aerospace
flexural modulus. Although the values of Tg and space industries in advanced resin transfer
measured by DMA are fairly reproducible, the moulding (RTM) processes. It is a premixed epoxy
absolute values of the elastic modulus are known to system suitable for service temperatures from 60
often show large discrepancies between specimens up to 180 1C. The standard curing cycle for RTM6
and different loading clamps with only minor resin is to cure the resin in mould at 160 1C for
changes in experimental conditions, regardless of 75 min, followed by freestanding post-cure at 180 1C
the type of DMA instrument used [2]. Particularly, for 120 min (ramp: 1 1C/min). The second resin is a
elastic moduli measured by DMA often do not diglycidyl ether of bisphenol A (DGEBA) epoxy
agree well with those obtained from mechanical resin, which was mixed with a hardener (Piperidine,
testing methods [3], although in principle they Sigma Aldrich) at a ratio of 100:5 by weight and
should be the same. Furthermore, for measurements then cured at 120 1C for 16 h. An SiO2 nanoparticle
using DMA, in addition to systematic errors due to dispersed epoxy (Nanopox XP 22/0516, Hanse-
instrument design, machine compliance and evalua- Chemie AG, Germany) was used as a modifier to
tion of measured data, the elastic modulus has also mix with Araldite-F to form silica–epoxy nanocom-
been found to be influenced by specimen span-to- posites with the nanosilica content of 6 and 8 wt%,
thickness ratio, drive amplitude, specimen align- respectively. Detailed procedures for making the
ment, loading clamps and contact stresses [3,4]. In nanocomposites have been described previously
most cases, the elastic moduli measured by DMA elsewhere [8].
are utilized only for screening material properties Another polymer material, polycarbonate, was
for the purposes of quality control, research and also tested to demonstrate the effects of loading
development, and establishment of optimum pro- modes on the elastic moduli as measured by DMA.
cessing conditions [5,6]. In a recent study by
Hobbiebrunken et al. [7], storage moduli of an 2.2. Dynamic mechanical analysis
epoxy resin measured by DMA using a dynamic
tension mode were calibrated using the static A dynamic mechanical analyser (TA DMA 2980)
mechanical testing data obtained at room tempera- was used for measuring the temperature-dependant
 ARTICLE IN PRESS
S. Deng et al. / Polymer Testing 26 (2007) 803–813 805

elastic moduli of the two cured epoxy resins and two 2.3. Mechanical tests
silica–epoxy nanocomposites. Tests were conducted
with three different loading fixtures, single/double The cured pure Araldite-F and its silica–epoxy
cantilever and three-point bending modes [1]. The nanocomposites were machined into tensile and
support span was 17.39 mm for the single cantilever flexure specimens, which were then tested at five
loading mode, 34.78 mm for the double cantilever different temperatures, 50, 0, 23, 50, and 70 1C.
mode and 50 mm for the three-point bending mode. The tensile specimen was dumbbell shaped as
Temperature scanning from low to high was required by ASTM D 638 and ISO 527, with a
performed with a heating rate of 2 1C/min at an cross-section of 10  6 mm in the gauge length. An
oscillation frequency of 1 Hz. The oscillation extensometer with a gauge length of 50 mm was
amplitude of displacement was kept at 40 mm for attached on the tensile specimen surface for
most specimens unless otherwise specified. determining the axial strain. A loading rate of

3500 1.4

3000 Three-point bending 1.2

Double cantilever beam 1.0

2500
Modulus [MPa]

0.8
2000 Single cantilever beam

Tan δ
0.6
1500
0.4
1000
0.2

500 0.0

0 -0.2
20 40 60 80 100 120
Temperature [°C]

Fig. 1. Storage moduli of Araldite-F epoxy measured by DMA with three different modes.

5500
5000
4500
4000
Modulus [MPa]

3500 Three-point bending

3000
2500
Single cantilever beam
2000
1500
1000
500
0
-80 -60 -40 -20 0 20 40 60 80 100 120 140 160 180 200 220 240
Temperature [°C]

Fig. 2. Storage moduli of RTM6 epoxy measured by DMA with two loading modes.
 ARTICLE IN PRESS
806 S. Deng et al. / Polymer Testing 26 (2007) 803–813

5 mm/min was selected for all tensile tests. The silicone rubber moulds and following the recom-
flexure test specimens had nominal dimensions of mended curing cycle from the material manufac-
80  12  4 mm. A supporting span of 50 mm and a turer. After de-molding, the top surfaces of the
loading rate of 2 mm/min were used for all the cured specimens were milled to remove surface
flexural tests. At least five specimens were tested in roughness and defects using a grinding machine
both tests for each group at different temperatures. with an SiC wheel. Tensile tests for the cured RTM6
All tensile and flexural specimens were tested in as- resin specimens were conducted at six temperatures,
cured condition without post-curing. Flexural tests 54, 25, 0, 23, 50, and 71 1C.
could not be conducted at 70 1C due to the material A universal material testing machine (Instron
softening, and only four temperatures were at- 5567) was used for all mechanical tests. Tests at low
tempted. and elevated temperatures were carried out in an
The dumbbell-shaped tensile specimens of pure environmental chamber, in which accurate control
RTM6 were fabricated by casting the resin into of the testing temperature was maintained. For the

3000
Mechanical testing data
Three-point bending
(ASTM D5023)
2500
Storage Modulus [MPa]

2000

Single cantilever
1500

1000

500

0
20 40 60 80 100 120 140
Temperature [°C]

Fig. 3. Storage moduli of polycarbonate measured by DMA with two different loading modes.

3500

3000 Three-point bending (amplitude = 40 μm)

2500
Modulus [MPa]

Double cantilever (amplitude = 10 μm)

2000

1500
Single cantilever (amplitude = 20 μm)
1000

500

0
20 30 40 50 60 70 80 90 100 110 120
Temperature [°C]

Fig. 4. Storage moduli of Araldite-F epoxy measured by DMA with three different loading modes.
 ARTICLE IN PRESS
S. Deng et al. / Polymer Testing 26 (2007) 803–813 807

low temperature testing, liquid nitrogen was used to distribution within the specimen. The elastic moduli
create a low-temperature environment by a tem- were derived from the linear portion of the
perature controller that regulated the liquid nitro- stress–strain curves for both tensile and flexure
gen flow blowing into the chamber according to the tests, and the average values were used to establish
temperature required. Once a specimen was correlations with DMA data.
mounted and the chamber door was closed, at least
20 min soaking time was maintained to allow the 3. Results and discussion
temperature in the chamber to become homoge-
neous. For the tests of specimens at elevated 3.1. Elastic moduli measured by DMA
temperatures, the chamber was gradually heated
up to the designated temperature and then held at 3.1.1. Effect of loading modes
that temperature for at least 20 min before testing Fig. 1 shows the measurement results using DMA
began, in order to ensure uniform temperature for the cured pure Araldite-F epoxy measured with

6000 0.9

0.8
5000
0.7

4000 0.6
Modulus [MPa]

0.5

Tanδ
3000
0.4

2000 0.3

0.2
1000
0.1

0 0
-100 -50 0 50 100 150 200 250
Temperature [°C]

Fig. 5. Storage moduli of RTM6 resin measured by DMA with three-point bending loading mode at an amplitude of 40 mm.

3500
Amplitude = 20 μm
3000
Amplitude = 40 μm

2500
Amplitude = 60 μm
Modulus [MPa]

2000

1500

1000

500

0
30 40 50 60 70 80 90 100
Temperature [°C]

Fig. 6. Storage moduli of Araldite-F epoxy resin measured by DMA with three-point bending mode and different amplitudes.
 ARTICLE IN PRESS
808 S. Deng et al. / Polymer Testing 26 (2007) 803–813

three different loading modes, single cantilever, carbonate measured by the same DMA instrument,
double cantilever and three-point bending. It can shown in Fig. 3. The polycarbonate specimens were
be seen clearly that discrepancies exist in absolute supplied by TA instruments as sample materials for
values of the storage (elastic) moduli measured with DMA, which have a cross-section of 12  3 mm2.
different loading modes, although values of the The test conditions used for these specimens were
glass transition temperature (peak points of damp- similar to those for testing epoxies. In Fig. 3, the
ing) measured from the three modes are almost the storage moduli of the material measured by the
same. The results measured by DMA for the cured single cantilever and three-point bending modes
RTM6 demonstrate a similar situation as shown in obviously do not agree well with each other,
Fig. 2, in which clear differences can be seen although the absolute values measured using the
between the storage moduli measured with the three-point bending mode are nearly the same
three-point bending and single cantilever modes. as those listed in ASTM D 5023 for the same
Another example is the storage moduli of poly- material [5].

5500
5000
4500
Mechanical testing Data
4000 (Tension)
Modulus [MPa]

3500 Shifted DMA data

3000
2500
2000
1500
Original DMA Data
1000 (Single cantilever)
500
0
-80 -40 0 40 80 120 160 200
Temperature [°C]

5500
5000
4500
4000
Modulus [MPa]

3500 Original DMA data

(Three-point bending)
3000
2500
2000
1500
Mechanical testing data
1000 (Tension) Shifted DMA data

500
0
-80 -40 0 40 80 120 160 200
Temperature [°C]

Fig. 7. Comparison of elastic moduli of RTM6 resin measured by DMA with tensile testing results measured at six temperatures.
 ARTICLE IN PRESS
S. Deng et al. / Polymer Testing 26 (2007) 803–813 809

3.1.2. Effects of test parameters ical testing methods using the same material. As the
With the same mode and with similar testing mechanical testing data at room temperature are
conditions, the repeatability of the elastic moduli normally available or can be easily obtained
measured by DMA is normally good. As shown in experimentally, corrections can be readily made
Figs. 4 and 5, the storage moduli measured by based only on the available mechanical testing data.
DMA for both the epoxies with the same loading
clamps and the same loading parameters (ampli- 3.2. Correlations of elastic moduli between DMA
tude, contact force and machine compliance) under and mechanical tests
all three loading modes show relatively consistent
results between identical specimens, although dis- 3.2.1. RTM6 resin
crepancies are apparent. However, if the loading Comparisons of elastic moduli measured by
parameters vary, the repeatability and reproduci- DMA and those by mechanical tests (tension) are
bility of storage moduli become less good. As shown shown in Fig. 7. It can be seen that at different
in Fig. 6, with the same loading mode of three-point temperatures the original DMA data measured
bending, the absolute values of storage moduli using both the single cantilever and three-point
measured with different oscillation amplitudes (20, bending clamps did not agree well with the Young’s
40 and 60 mm) are dissimilar. If other variables were moduli obtained from the tensile tests. However,
considered, such as specimen span-to-thickness when a vertical shifting operation was performed
ratio, specimen alignment and contact stresses based on the elastic modulus at ambient tempera-
[3,4], there would be evident uncertainty about the ture (23 1C) measured from tensile tests, good fit
elastic moduli measured by DMA. Therefore, could be observed for the DMA data to the tensile
correction is required using a reliable method in testing results, particularly for the storage moduli
order to obtain consistent elastic moduli measure- measured using a three-point bend mode. The
ment by DMA. One solution is possibly the use of vertical shifting of the DMA data was straightfor-
standard materials with known temperature-depen- ward. The DMA storage modulus at the ambient
dant moduli over a wide range of temperatures. temperature was first compared with the elastic
However, the responses of dissimilar materials to modulus measured by mechanical testing at the
temperature variation may be different, so that same temperature. All DMA data were then added
correction made using a given standard material is or subtracted by the difference of the two moduli, so
possibly invalid for another. Hence, one effective that a shifted temperature-dependent moduli curve
approach to solving this inconsistency is perhaps to was produced (Fig. 7).
establish the correlation between the elastic moduli Aluminium alloy specimens were also tested by
measured by DMA and those obtained by mechan- DMA with both single-cantilever and three-point

300000

Single cantilever
250000

200000
Modulus [MPa]

150000

100000 Three-point bending

50000

0
0 20 40 60 80 100 120 140 160 180 200
Temperature [°C]

Fig. 8. Storage moduli of an aluminium alloy measured by DMA with two loading modes.
 ARTICLE IN PRESS
810 S. Deng et al. / Polymer Testing 26 (2007) 803–813

bending modes in order to identify the consistency eliminate clamping effects [3–5], it is reasonable to
of the DMA for testing materials other than believe that the results measured by DMA with a
polymers. A comparison of the test results is shown three-point bending mode are closer to the mechan-
in Fig. 8. It can be seen that the storage moduli ical testing data than either the single-cantilever or
measured using the three-point bending mode were the dual-cantilever modes.
similar to the elastic moduli of the material
(70–72 GPa) in the temperature range tested, while 3.2.2. Araldite-F resin
the DMA values measured using the single-canti- Storage moduli of the cured Araldite-F resin
lever mode deviated greatly from the elastic moduli measured by DMA with the three-point bending
of the aluminium alloy because of the clamping mode also showed good agreement with the
effects. As the three-point bending mode can mechanical testing results obtained at different

4500

Original DMA data 4000

3500

3000
Modulus [MPa]

Flexure testing data 2500

2000
Shifted DMA data
1500

1000

500

0
-80 -60 -40 -20 0 20 40 60 80 100
Temperature [°C]

4500

Original DMA data 4000

3500

3000
Modulus [MPa]

2500
Tensile testing data
2000 Shifted DMA data
1500

1000

500

0
-80 -60 -40 -20 0 20 40 60 80 100
Temperature [°C]

Fig. 9. Storage moduli measured by DMA compared with (a) flexure and (b) tensile testing results of Araldite-F resin measured at
different temperatures.
 ARTICLE IN PRESS
S. Deng et al. / Polymer Testing 26 (2007) 803–813 811

temperatures when DMA data were shifted verti- shifting was done based on the room-temperature
cally towards the results of the flexure and tensile data, though there were some deviations at a low
tests at room temperature, as shown in Fig. 9. temperature ( 50 1C).

3.2.3. Silica– epoxy nanocomposites 3.3. Validity of temperature-dependent elastic moduli

Storage moduli measured by DMA for silica– measured by DMA
epoxy nanocomposites with SiO2 contents of 6 and
8 wt% also exhibited close correlations to the elastic In principle, DMA works in the same way as a
moduli obtained from tensile and flexure tests. As material testing machine, with differences in loading
demonstrated in Figs. 10 and 11, the storage moduli scale, deformation range and specimen size. Indeed,
of the two nanocomposites, measured by DMA DMA can be considered as a miniature testing
with a three-point bending mode, showed relatively machine. However, for a normal universal material
good fit with mechanical testing data if vertical testing machine, the stiffness of the loading frame

4500

Original DMA data

4000

3500

3000
Modulus [MPa]

Flexure testing data 2500

2000 Shifted DMA data

1500

1000

500

0
-80 -60 -40 -20 0 20 40 60 80 100
Temperature [°C]

4500
Original DMA data
4000

3500

3000
Modulus [MPa]

2500
Tensile testing data
2000 Shifted DMA data

1500

1000

500

0
-80 -60 -40 -20 0 20 40 60 80 100
Temperature [°C]

Fig. 10. Storage moduli of 6% silica nanocomposite measured by DMA compared with mechanical testing results: (a) three-point bending
and (b) tensile tests.
 ARTICLE IN PRESS
812 S. Deng et al. / Polymer Testing 26 (2007) 803–813

6000
Original DMA data
5000
Modulus [MPa]

4000

3000

Flexure testing data 2000

Shifted DMA data

1000

0
-80 -60 -40 -20 0 20 40 60 80 100
Temperature [°C]

6000

Original DMA data

5000

4000
Modulus [MPa]

3000

Tensile testing data

2000
Shifted DMA data

1000

0
-80 -60 -40 -20 0 20 40 60 80 100
Temperature [°C]

Fig. 11. Storage moduli of 8% silica nanocomposite measured by DMA compared with mechanical testing results: (a) three-point bending
and (b) tensile tests.

and the clamps as well as the dimensions of the tal conditions, as already demonstrated by the
loading frame are much greater than those of the DMA results shown in this study. Consequently,
specimens to be measured. All of these factors the resultant data may not be consistent if different
would ensure that the universal testing machine is DMA machines, dissimilar clamps or varying test
much stiffer than the test material, so that the parameters are used. To resolve the apparent
influence of the instrument compliance on test discrepancies in results between different studies,
results is negligible. The loading frame in a DMA the experimental conditions must be kept consistent
instrument is normally much smaller than that of a for all measurements and the absolute values should
universal testing machine, so that the influence of also be verified. However, in reality, it is difficult to
the instrument compliance on the test results of do so because of the variations in materials,
DMA cannot be ignored. In fact, the elastic moduli specimen dimensions and test requirements. There-
measured by DMA are influenced not only by the fore, the correction method described in this study
instrumentation compliance but also by experimen- provides a simple yet effective approach for
 ARTICLE IN PRESS
S. Deng et al. / Polymer Testing 26 (2007) 803–813 813

determining temperature-dependant moduli without extended to polymer-based composites, but further

spending enormous amounts of materials and time. confirmation of the validity will be needed.
The advantage of using the proposed method is that
it can exclude the influence of factors such as
References
different DMA machines, dissimilar clamps or
varying test parameters, as these factors affect only [1] TA Instruments, 2980 Dynamic Mechanical Analyzer Opera-
the magnitudes of elastic modulus but not the tor’s Manual, 1997.
temperature dependency. [2] A.R. Wedgewood, J.C. Seferis, Error analysis and modelling
of non-linear stress strain behaviour in measuring dynamic
mechanical properties of polymers with the Rehovibron,
4. Conclusions
Polymer 22 (1981) 966–991.
[3] P. Lee-Sullivan, D. Dykeman, Guidelines for performing
Although there are discrepancies between the storage modulus measurements using the TA Instruments
temperature-dependent elastic moduli measured by DMA 2980 three-point bend mode. I. Amplitude effects,
DMA and those measured by normal mechanical Polym. Test. 19 (2000) 155–164.
tests, good agreement has been confirmed for two [4] Q. Shao, P. Lee-Sullivan, Guidelines for performing storage
modulus measurements using the TA Instruments DMA 2980
epoxy systems and two nanocomposites after DMA three-point bend mode. II. Contact stresses and machine
data were corrected based on the static mechanical compliance, Polym. Test. 19 (2000) 239–250.
testing data obtained at a convenient temperature [5] ASTM D5023-99, Standard test method for measuring the
(e.g., the ambient temperature). Therefore, without dynamic mechanical properties of plastics using three-point
running expensive mechanical tests at all tempera- bending, 1999.
[6] ASTM D5024-95a, Standard test method for measuring the
tures, it is possible to determine temperature- dynamic mechanical properties of plastics in compression.
dependent elastic moduli of polymers using DMA. [7] T. Hobbiebrunken, B. Fiedler, M. Hojo, S. Ochiai, K.
The three-point bending mode was found to be Schulte, Microscopic yielding of CF/epoxy composites and
the most suitable for measuring temperature-depen- the effect on the formation of thermal residual stresses,
Compos. Sci. Technol. 65 (2005) 1626–1635.
dent moduli by DMA, as it can eliminate clamping
[8] S. Deng, L. Ye, K. Friedrich, Fracture behaviours of epoxy
effects and provide the most analogous approach to nanocomposites with nano-silica at low and elevated tem-
normal mechanical testing methods. It is anticipated peratures, J. Mater. Sci., 2007, available online on the journal
that the proposed treatment of DMA data may be website.