Polymer thermal analysis
Polymer thermal analysis
Published on: Mar 4, 2016
Transcripts - Polymer thermal analysis
Encyclopedia of Polymer Sceince and Technology
Copyright c 2005 John Wiley Sons, Inc. All rights reserved.
THERMAL ANALYSIS OF POLYMERS
Thermal analysis is defined by the International Confederation of Thermal Anal-ysis
and Calorimetry (ICTAC) (1,2) as “a group of techniques in which a property
of a sample is monitored against time or temperature while the temperature of
the sample, in a specified atmosphere, is programmed.” In practice, the temper-ature
of the oven that contains the sample actually is programmed, while the
temperature of the sample in some cases may differ from the programmed tem-perature.
Exothermic or endothermic reactions or phase transitions in the sample
subjected to the programmed temperature variation may cause variations in the
temperature between the sample and oven up to several degrees.
The more common thermal analysis (TA) methods are listed in Table 1
(3). Acronyms used for the techniques are included in the right-hand column.
The present discussion will be limited to the major techniques that are used to
characterize polymer samples. These are differential thermal analysis (DTA), dif-ferential
scanning calorimetry (DSC), thermogravimetry (TGA), thermomechani-cal
analysis (TMA), dynamic mechanical analysis (DMA), and dielectric analysis
(DEA). The most popular technique for polymer applications is DSC followed by
TGA (along with its derivative, DTGA). The use of the other methods is not as
widespread as DSC and TGA but is gaining in popularity. The realization of the
potential for TMA as a valuable tool for polymer characterization is comparatively
recent. DMA and DEA were originally developed by rheologists over 50 years ago
and have been adopted by thermal analysts more recently.
In addition to the individual techniques cited above, there are situations
where individual samples are subjected to a common atmosphere and thermal en-vironment.
This is referred to as a concurrent analysis. An example of this is the
combined DTA and TGA measurement, where two separate measurement devices
share the same oven. In some cases two or more measurements are performed on
the same sample. There is also the possibility of interfacing several techniques,
such as TGA combined with some form of evolved gas analysis such as gas chro-matography,
Mass Spectrometry (qv), or infrared spectroscopy (or some combina-tion
of these) (see CHROMATOGRAPHY, AFFINITY; VIBRATIONAL SPECTROSCOPY). This
situation is referred to as ‘coupling’ or a coupled technique. Since the possibilities
for interfacing several analytical techniques are quite large, concurrent or coupled
techniques will not be considered further in this presentation.
Prior to the advent of DSC (ca 1964–65), DTA was first used in the field of
geology to study clay mineralogy and pyrometry. The evolution of DSC has its
2 THERMAL ANALYSIS OF POLYMERS
Table 1. Principal Thermoanalytical Methodsa
Property Technique Acronym
Mass Thermogravimetry TGA, TG
Apparent massb Thermomagnetometry TM
Volatiles Evolved gas detection EGD
Evolved gas analysis EGA
Radioactive decay Emanation thermal analysis ETA
Temperature Differential thermal analysis DTA
Heatc or heat fluxd Differential scanning calorimetry DSC
Dimensions Thermodilatometry TD
Mechanical properties Thermomechanical analysis TMA
Dynamic mechanical analysis DMA, DMTA
Acoustical properties Thermosonimetry (emission) TS
Electrical properties Thermoelectometry (resistance) DETA, DEA
Optical properties Thermooptometry (spectroscopy)e TPA
aFrom Ref. 3.
bChange induced by an imposed magnetic field gradient.
dHeat flux DSC.
eAbsorption, fluorescence, Raman, etc. Nonoptical forms of spectroscopy, eg, NMR, ESR, M¨ossbauer,
etc, are also applicable.
origins in DTA, driven by applications in polymers and the timely development of
techniques, instruments, and advanced software. In recent years, DSC has largely
supplanted DTA for use in polymers, except for applications at high temperatures.
Most commercial DSC instruments have a temperature range from −150◦C up to
750◦C, which is quite adequate for most polymer applications.
Differential Thermal Analysis
Description of the Method. As shown in Table 1 and Figure 1, DTA mea-sures
the differential temperature between a sample and a reference pan, which
are closely matched thermally and arranged symmetrically within the oven. As
the sample goes through the programmed dynamic temperature change, there is
no temperature difference until the sample undergoes an exothermic or endother-mic
chemical reaction or change of physical state. In the case of an exotherm,
the sample’s temperature will increase, while in the case of an endotherm, it will
decrease. In either case, the thermal event will be recorded as the sample tem-perature
departs from the baseline and then returns to the base line when the
THERMAL ANALYSIS OF POLYMERS 3
Fig. 1. General principle of operation for DTA and schematic of a DTA apparatus (4). The
subscripts represent: s-sample, r-reference, i-initial, f-final.
4 THERMAL ANALYSIS OF POLYMERS
Table 2. Some Applications of DTA and DSC to Polymers
Melting ranges Release of strains
Rates of crystallization and reaction Purity determination
Degree of crystallinity Quality control
Glass-transition phenomena Phase diagrams
Heat capacity Energy storage
Enthalpy of transitions Hazards evaluation
Identification—Fingerprints Mesophase transitions
Thermal and oxidative stability Nucleation of crystals
Analysis of copolymers and blends Catalysis
Nucleation phenomena Thermal conductivity
reaction or transformation is complete. Also, a change in the heat capacity of the
sample will show as a change in slope (eg, Glass Transition (qv)) in the T ver-sus
time or temperature plot. Modern DTA instruments generally use matched
thermocouples as sensors, one each in contact with the sample or its container
and the reference material or its container. The output of the differential signal
is amplified and sent to a data acquisition system. Further details about DTA
instrumentation and the appropriate experimental parameters can be found in
several references by Wendlandt (4), Gallagher (Ref 5, Chapt. 1), and Wunderlich
(Ref. 5, Chapt. 2). Some consideration of sample preparation, the effect of experi-mental
parameters (eg, rate of heating and cooling), and methods for enthalpy and
temperature calibration will be considered in the subsequent discussion of DSC.
Applications of DTA for Polymers. Table 2 (Ref. 5, Chapt. 1) describes
some of the many applications of DTA and DSC. Both DTA and DSC can be used
to determine the temperature of the transitions, cited in Table 2. However, the
DSC peak area, in addition, gives quantitative calorimetric information (heat of
reaction, transition, or heat capacity). DTA can only do so when calibration with
a standard material allows the quantitative conversion of T to heat flow and,
ultimately, heat of transition (H) or heat capacity (Cp). Also, the response of
DTA with increasing temperature may be affected by poor heat transfer in the
system, detector sensitivity, etc (4). For these reasons, when there is a choice
between DSC and DTA, DSC is the preferred method. The illustrations shown
below for applications of DSC in characterization of polymers also generally apply
for DTA, with the limitations mentioned above. Therefore, DTA applications will
not be considered here. Illustrations of polymer applications for DTA can be found
in the Thermal Analysis section by Bacon Ke (6) in the previous edition of this
Differential Scanning Calorimetry
Instrumentation. The development and introduction of DSC in 1964 (7)
was a major innovation in thermal analysis. Several competitive commercial in-struments
were developed shortly afterward. Basically, there are two types of DSC
instruments. The Perkin-Elmer version is called the “power compensation” DSC.
THERMAL ANALYSIS OF POLYMERS 5
Fig. 2. A schematic representation of a power compensation DSC instrument and its
Zahra and Zahra (8) reviewed a recent version of this type of DSC instrumen-tation.
Figure 2 provides a schematic representation of the power compensation
The concept of operation of this instrument is based on keeping the temper-ature
of R and S the same. This is achieved by placing the temperature sensors
(platinum resistance thermometers) in a bridge circuit. Any change in tempera-ture
in either the sample or in the reference (by virtue of any exo- or endother-mic
event) is immediately compensated for by an equivalent amount of current
flow required to drive a heater to keep them at the same temperature. Thus, the
6 THERMAL ANALYSIS OF POLYMERS
integral of the power input during the transition (or the heat capacity change) is
equal to the energy difference (H) supplied to the sample or the reference during
the particular event. The event would be endothermic or exothermic depending
on whether the current travels to the sample or the reference pan, respectively.
This results in an exothermic peak pointing downward and an endothermic peak
pointing upward. This is in conformity with the thermodynamic convention, but
differs with the sign convention for the DTA and the DSC “heat flux” for instru-ments
made by other manufacturers, as discussed below. For DTA, ICTAC (2)
recommends plotting “an upward deflection as a positive temperature differential
and a downward deflection as a negative temperature differential with respect to
the reference.” In other words, exo is upward and endo is downward.
The second type of DSC unit operates in a “heat flux” mode. This type of
instrument is offered by TA Instruments, Mettler-Toledo, Setaram, Netszch, and
other manufacturers. Its operation is similar to that of DTA (ie, it generates a
T signal). However, the associated hardware and software, carefully integrated
into the system, quantitatively converts T to H and compensates for other
deficiencies, such as the temperature dependence of thermal transport and sen-sor
sensitivity. Figure 3 shows two types of heat-flux calorimeters along with a
schematic diagram taken from Boerio-Goates and Callanan (9). Figure 3a repre-sents
the Boersma-type instrument used by TA Instruments. In this unit, single
thermocouples are in good thermal contact with the sample and the reference
holders. In the Tian–Calvet-type calorimeter, shown in Figure 3b, a thermopile is
used. The latter tends to give greater temperature sensitivity and better temper-ature
averaging over the entire sample and reference containers. The schematic
representation in Figure 3c notes the temperatures and thermal resistances that
are necessary for mathematical analysis of the heat-flux DSC. The analysis is de-scribed
by Wunderlich (5). Each type of DSC gives satisfactory enthalpy data (10)
with an accuracy of around 1–2%. Some heat-flux DSC instruments can operate at
temperatures up to 1500◦C, while the power-compensation instrument operation
is limited to 725◦C.
Calibration and Standards. Thermal analysis methods are not absolute
and calibration is needed to record the correct abscissa value of temperature T
(in Kelvin) and time t (in seconds or minutes). On the ordinate, calibration is
necessary for the amplitude of deflection, T, expressed as the difference in tem-perature
(in Kelvin) for DTA or as heat flux, dQ/dT (in joules per second or watts)
for DSC. Each instrument manufacturer provides methods and standard materi-als
for these calibrations. In addition, ICTAC, in collaboration with the National
Institute of Standards and Technology (NIST), has developed a series of materials
as calibration standards for DSC/DTA. These reference materials can be used to
calibrate both the temperature scale (K, abscissa) and heat flow (J/g, ordinate) on
the basis of the integrated area under the curve. Figure 4 shows the heat flow–
temperature relationship for various solid–solid and solid–liquid melting stan-dards.
Table 3 lists the solid1-to-solid2 transitions, melting points, and Curie tem-peratures
of various pure metals, and also their transition enthalpies (J/g) (11).
Figure 5 illustrates temperature calibration curves for both DTA/DSC and
TGA using high purity nickel as a magnetic material (Curie point = 358.4◦C).
Unlike Tm, the Curie point temperature is practically unaffected by the changing
of the heating rate between 1 to 20◦C/min.
THERMAL ANALYSIS OF POLYMERS 7
Fig. 3. Representation of a two heat-flux calorimeter showing (a) Boersma thermocouple
placement and (b) the Tian–Calvet design. The schematic diagram (c) is appropriate for
analysis of the response of both types of calorimeters. Symbols in (c) are: subscript T refers
to temperature, R refers to reference; the temperatures of the block, sample, container,
reference, and reference containers given by TB, TSC, TR, TRC, respectively; capital R refers
to heat transfer resistance in the instrument (9).
Hakvoort (16) and also Della Gatta and Barczynska (17) advocate certain
solid-state first-order transitions for use as subambient temperature standards.
For relatively low temperature ranges, such as the glass transitions (Tg) for elas-tomers,
Tan and Sabbah (18) have proposed the melting transitions of various
organic compounds. Certified reference materials for heat capacity calibration,
which are recommended by International Union of Pure and Applied Chemistry
(IUPAC), are also available (Ref. 5, pp. 46 and 353–356). Two ASTM standards
document procedures for temperature (ASTM E967) and heat-flow (ASTM E968)
calibration in DSC.
Operating Parameters for DSC/DTA.
Sample Size. In order to avoid temperature gradients inside a sample, a
small sample size is preferred. A small sample size also gives better resolution.
8 THERMAL ANALYSIS OF POLYMERS
Fig. 4. Typical enthalpy calibration curve for DSC. Salt (Solid1–solid2 standard);
Metal (melting standard) (14).
Table 3. Some Reference Materials for DTA and DSC
Melting pointsa Solid1–solid2
Material Temperature, ◦C Enthalpy, J/g Material Temperature, ◦C Enthalpy, J/g
Hg −38.8344 11.5 Cyclohexane −83.1 Unavailable
In 156.5985 28.42 KNO3 127.7 50.48
Sn 231.928 59.2 KClO4 299.5 99.32
Pbc 327.502 23.16 Ag2SO4 412 59.85
Zn 419.527 112.0 SiO2 573 12.1
Al 660.323 400.1 K2SO4 583 48.49
Ag 961.78 104.7 K2CrO4 665 51.71
Au 1064.18 63.7 BaCO3 810 89.00
Cu 1084.62 205.4 SrCO3 925 133.23
aInternational Temperature Scale 1990 for temperatures, and Ref. 12 for the enthalpies.
bNBS–ICTA Certificates GM-758, GM-759, GM-760 for the recommended temperatures, and Ref. 13
for the enthalpies.
cDowngraded to a secondary temperature point.
THERMAL ANALYSIS OF POLYMERS 9
Fig. 5. Thermomagnetometry (TM)/DTA curves for accurate determination of Tc for mag-netic
On the other hand, sensitivity of the instrument increases with increasing
sample size. The choice of sample size for DSC/DTA depends on the composi-tion
of the sample. For pure polymers, 3–10 mg is sufficient. For heterogeneous
materials, such as blends and filled polymers or those containing several other
compounding ingredients (such as elastomer vulcanizates), larger samples in the
range of 10–20 mg may be necessary. However, for characterization of all materi-als
in an air or oxygen environment, thin samples weighing only 0.2–0.5 mg are
recommended in order to minimize diffusional effects with respect to oxygen or
air, inhibit secondary reactions, and improve reproducibility. In the case of hetero-geneous
materials it is advisable to run multiple samples in order to verify that
the data are representative of the composition.
Sample History. A factor that is characteristic of polymers is the depen-dence
of morphology on thermal history. Significant variations in the degree of
crystallinity as well as temperatures for glass transition (Tg), crystallization (Tc),
and crystalline melting (Tm) can be obtained by varying the crystallization and
annealing temperatures, or by changing cooling and heating rates. Additional
types of sample history may be due to the viscoelastic character of the polymer.
Elastomer samples that have been pressed, molded, or extruded may exhibit ap-preciable
elastic “memory,” which eventually may result in severe distortion of
the sample during the TGA, DSC, and DMA experiments. This would result in
anomalous data. Another type of time-dependent sample history variation can be
encountered with thermoplastic block copolymers. Here changes in dimensional
stability and/or polymer transition temperature may occur because of changes
in the morphology of the material during the thermal measurement. This might
10 THERMAL ANALYSIS OF POLYMERS
result from changes in the phase distribution resulting from aggregation of the
hard or soft block segments. A further complication is physical aging or enthalpy
relaxation that occurs in the amorphous phase due to the tendency of a polymer in
a nonequilibrium state to spontaneously shift toward the equilibrium state. This
is a complex subject that is not considered here but is covered in some detail by
Chartoff in Reference 5.
In order to obtain a representative DSC curve for the polymer sample that
is consistent and reproducible, it is necessary to destroy the prehistory. This is
achieved by preheating the sample above the crystalline melting temperature in
an inert atmosphere, holding it there for a few minutes (generally, 5 min) and then
cooling slowly to the start temperature. If the crystallization rate is slow, it may
be necessary to hold the polymer (ie, anneal) at a temperature below the melting
point for some time in order for crystallization to go to completion.
In certain cases, whenever a special kind of prehistory is imparted to a poly-mer
(eg, either through processing or aging, such as in fiber and film processing
from the melt), it is important to study the prehistory. Here it may be desirable not
to destroy the prehistory because the DSC response of the material may be corre-lated
with a process parameter, thus extending the usefulness of the DSC curve.
In this case, one should always record the first heat DSC/DTA curve followed by
a second run using the reheat procedure noted above.
Base Line. Selecting a proper base line for integrating the area under the
curve is essential for enthalpy determinations. In the absence of detailed infor-mation
about overall error in different alternative modes, a simple straight-line
approach is adequate. However, for partial area measurements in kinetics, a more
accurate base line may be needed. The following texts may be consulted for further
information on this topic: Wendlandt (4), Brown (19), and Wunderlich (11). Many
modern instruments have the ability to manipulate the base line. In the case of
irreversible transformations without any weight loss, a simple rerun of a sam-ple
(after the transformation), under the same conditions should give a suitable
base line. All current computerized instruments can subtract one base line from
another that is generated under identical experimental conditions.
Thermal Transport. Heat transfer to the sample can be affected by the
following factors: (1) physical arrangement of the sample and reference relative
to the furnace; (2) the choice of sensor, its size and position; and (3) the type
of construction materials. These factors determine the thermal coupling. They
also influence the shape and size of the peak for a thermal event. The better the
thermal coupling between the sample and the reference, the smaller the original
T signal and the faster the return to the base line, thus improving resolution.
A higher thermal conductivity environmental gas, such as helium, will result in
better thermal coupling. Placing a metal lid on the sample pan also markedly
improves the thermal properties of the system by helping to press the sample into
good contact with the pan and reducing heat flow to the environment.
Effect of Heating Rate. The thermal lag between the sample and the sensor
increases as the heating rate increases and, generally, the resolution of a transi-tion
(or other thermal event) decreases. However, the amplitude of the T signal
increases with heating rate because the transition takes place in a shorter period
of time and the rate of change is greater. With modern instruments, amplifica-tion
of the T signal at a lower heating rate is not a problem, but it increases
THERMAL ANALYSIS OF POLYMERS 11
Table 4. Compromises in Operation and Construction for DTA and DSC
Parameter Maximum resolution Maximum sensitivity
Sample size Small Large
Heating rate Slow Fast
Sample–reference Linked Isolated
Particle size Small Large
Atmospherea High conductivity Low conductivity
aIn the absence of any reaction.
the signal-to-noise ratio. Therefore, compromises must be made. Typical heating
rates for DSC experiments are in the 10–20◦C/min range. For unknown polymer
samples, a range of heating rates with different amounts of sample should be
explored. This should provide information on the optimum conditions for detec-tion
and resolution of minor thermal events and would be a background study for
establishing a routine procedure for subsequent analyses. The effects of different
parameters on resolution and sensitivity are summarized in Table 4 (Ref. 5, p. 87).
DSC Applications in Polymers
The numerous applications of DSC in characterizing polymeric materials are quite
varied. Therefore, it is only possible to cast a glimpse on the subject. Figure 6 shows
a schematic DSC curve (20) for a pure polymer as it is progressively heated from a
low temperature to a high temperature in an oxidative (A) or inert (B) atmosphere.
The actual temperature, for a given polymer to undergo all of the thermal events
shown in Figure 6, depends on the structure of the polymer. For an elastomer, a
low temperature transition (glass-transition temperature Tg) may be as low as
−140◦C, whereas for a thermoplastic polymer, Tg values of room temperature or
higher may be more common. As noted before, the DSC experiment should be
started at a temperature at least 20–50◦C lower than the region of interest and
continued to at least 20◦C higher than the highest transition temperature. The
elevated temperature at which most carbon-chain polymers degrade is usually
less than 500◦C.
Starting from the lowest temperature, the first discontinuity observed is the
glass transition, which appears as a shift of the base line corresponding to the
heat capacity difference of the sample before and after the transition. There are
various methods to mark the exact location in the curve where the glass-transition
temperature Tg is recorded. This will be discussed below. The magnitude of base
line shift (Cp) during the glass transition is related to the concentration of amor-phous
chains in the sample and is affected by other factors such as molecular
mobility and free volume.
As the temperature increases, there may be a cold crystallization peak (Tc,
exotherm), followed by a crystalline melting peak (Tm, endotherm ). The intensity
of the cold crystallization peak depends on the prehistory of the sample. It will not
be present at all if the sample were cooled under controlled conditions from the
melt state or if it were annealed at a fixed temperature below Tm so as to complete
12 THERMAL ANALYSIS OF POLYMERS
Fig. 6. Schematic DSC curves showing different transitions and reactions of a polymer,
ranging from low temperatures to progressively higher temperatures. A, in oxygen; B, in
the crystallization process. If the sample is heated above Tm and then cooled slowly
to a low temperature, and scanned again, the enthalpy or the area under the melt-ing
curve will relate directly to the amount of crystalline material in the polymer
or the degree of crystallinity. Up to this point, the transitions mentioned are due
to physical changes and they are reversible. Physical transitions are not accom-panied
by any weight change and are not affected by the environment (nitrogen
or oxygen). Thus, they cannot be monitored using TGA, since there is no weight
At a still higher temperature (150–220◦C), if the polymer is an elastomer,
formulated with all the vulcanizing ingredients or uncured thermosetting type
monomer, an exotherm should occur indicating the curing reaction (not shown in
Figure 6). DSC enthalpy determination (from the area under the exotherm) is one
of the methods for determining the degree of cure of an elastomer formulation or
a thermoset and can be used as a quality control tool.
At still higher temperatures (250–500◦), the polymer undergoes degradation,
resulting in main-chain scission, cross-linking, cyclization, and loss of volatile
fragments. In an inert atmosphere, the degradation pattern may be endothermic,
exothermic, or both, whereas in oxygen or air, it is always exothermic. As expected,
degradation occurs at a much lower temperature in oxygen than in nitrogen. The
THERMAL ANALYSIS OF POLYMERS 13
DSC degradation patterns in nitrogen and oxygen have been used to identify many
polymers (21). Many authors, including Goh (22) and Savasc¸i and Baysal (23) have
attempted to use oxidation exotherms to characterize the oxidative stability of
elastomers, and, subsequently, to evaluate the effectiveness of added antioxidants.
DSC curves in the degradation regime may tend to be erratic because of the loss
of mass that may occur during the degradation event.
Specifying Tg by DSC. All Amorphous Polymers (qv) become stiff and ulti-mately
brittle like a glass, if sufficiently cooled. The temperature region where the
physical transition from a rubbery to a glassy state takes place is called the “Glass
Transition (qv).” It is characterized by a step change in Cp. As shown in Figure 7,
the glass transition covers a wide range of temperature (10–15◦Cfor pure, uncross-linked
polymers and wider for blends, filled, and cross-linked polymers). A defined
point in this temperature range is called the glass-transition temperature or Tg.
Fig. 7. Representative curve of heat flow versus temperature in the region of the Tg of a
polymer:To, temperature of departure from the base line;Te orTeo, temperature of intercept
of extrapolated base line and tangent to steepest slope; T0.5, temperature of 50% transition
(A = B); Tp, temperature of maximum slope (peak of derivative plot); Tf, temperature at
completion of the transition (24).
14 THERMAL ANALYSIS OF POLYMERS
Figure 7 shows five characteristic temperatures associated with a glass tran-sition
on the DSC curve. All five have been cited in various places in the literature
as Tg, often without mention of the specific location actually picked (24). The
assignment of Tg by thermal analysis methods was the topic of a recent ASTM
symposium (25). To, the “onset” temperature, defines the point at which the first
deviation from the base line on the low temperature side is observed. To is very
subjective and often difficult to determine because of the base line slope; and it
is not reproducible. Te or Teo (extrapolated onset, also called the fictive temper-ature)
is the temperature at the intersection of the extrapolated base line and
tangent taken at the point of maximum slope. It is generally reproducible and is
the most quoted value in many early publications on thermal methods. T0.5, the
endotherm half-height (also called half-vitrification) is the temperature preferred
in recent years for specifying Tg. The autoanalysis feature available with all mod-ern
thermal analysis equipment makes this determination easy and reproducible.
It has better reproducibility than the extrapolation method and locates a tempera-ture
where the thermal transition has approximately reached the inflection point.
ASTM E1356 and ASTM D3418 recommend either Te or T0.5 for specifying Tg.
Wunderlich (Ref. 5, p. 278) advises that Tg should be measured as the sample
is progressively cooled, rather than heated. The advantage here is that the sample
is in thermal equilibrium at the start of the measurement and enthalpy relaxation,
often encountered in Tg measurements, is avoided. However, instrumental draw-backs,
such as precise control of the cooling rates in some DSC models, preclude
general use of cooling curves for determining Tg. If Tg is obtained by progressive
heating of a cooled sample, it is advisable to cool the sample somewhat faster than
the subsequent heating rate to avoid enthalpy relaxation. This is also discussed
by Wunderlich (5). The heating or cooling rate and the location on the DSC curve
where Tg is taken always should be specified along with the reported Tg value.
Another method for locating Tg is by taking the derivative of the DSC glass-transition
curve. (The acronym DerDSC for “derivative DSC” is used here.) The
derivative curves exhibit a peak corresponding to the maximum slope of the en-dothermic
step in Cp accompanying Tg. In Figure 7, Tp is very close to T0.5, the
midpoint value. This is illustrated in Figure 8 for natural rubber (26). Note the
small endotherm at the high temperature end, presumably due to enthalpy relax-ation.
The derivative curves may also be helpful in locating multiple transitions
close in temperature and not resolved by the Tg curve (Ref. 25, p. 229). The disad-vantage
of this method is the high signal-to-noise ratio for the DerDSC curves.
For amorphous thermoplastic polymers, Tg at the extrapolated onset is tech-nically
more significant, since it defines the initial temperature for the loss of struc-tural
properties (eg, modulus) as the polymer softens through the glass-transition
region. Thus, it defines both a low temperature limit for the processing of amor-phous
thermoplastic polymers and an upper use temperature. For elastomers,
the temperature for useful elastomeric properties lies above the glass-transition
region, (to the right of the Tg curve) shown in Figure 7. Therefore, Tf, the tem-perature
at the completion of the glass transition, should be technically more
significant (26). If the elastomer is cooled beyond this point it enters the glass-transition
region and starts losing its elastic properties, becoming progressively
stiffer as the temperature decreases. As noted above, ASTM E1356 describes the
test method for determining the glass-transition temperature by DSC or DTA.
THERMAL ANALYSIS OF POLYMERS 15
Fig. 8. DSC glass transition of natural rubber showing T0.5 and Tp; heating rate 10◦C/min
Degree of Cure Using Tg. For many thermosets, Tg can be correlated with
the conversion or degree of cure (Ref. 5, Chapt. 6). Therefore, simultaneously with
the heat of reaction, measured by enthalpy, Tg can be used as a measure of the
degree of cure. In the final stages of cure, where Tg is actually the more sen-sitive
measure, it may be preferred to the residual heat of reaction, H1, as a
measure of the degree of cure. This may happen when the glass transition im-mediately
precedes the small residual exotherm. This is illustrated in Figure 9
(27) for epoxy–anhydride samples, previously cured to apparent completion at dif-ferent
temperatures. Note that the exotherms are facing downward, as would be
expected for a power compensation DSC instrument; Hc is the exothermic heat
of reaction for 100% conversion, expressed as heat per mole of reacting groups
(kcal/mole or kJ/mole) or per mass of material (cal/g or J/g) at the time t. The
fractional degree of conversion is given by
αl =Hl/Hc (1)
where α1 is the fractional conversion or extent of reaction and H1 the heat
generated up to time t1.
The measurement of the degree of conversion by enthalpy measurement ap-plies
to elastomer cure also, since the vulcanization reaction is highly exothermic.
Unlike for thermosets, Tg is not a measure of degree of cure for elastomers. Sircar
discusses this and numerous other aspects of the characterization of elastomeric
materials using thermal analysis in chapter 5 of Reference 5.
Melting in Semicrystalline Polymers.
Traditional DSC Methods. Many commercially available thermoplastics
[eg, polyethylene (PE), polypropylene (PP), poly(ethylene terephthalate) (PET),
16 THERMAL ANALYSIS OF POLYMERS
Fig. 9. DSC scans at 8◦C/min on epoxy–anhydride samples previously reacted to apparent
completion at indicated temperatures. Note increasing Tg followed by residual exotherm for
samples cured below Tg∞. Diminishing Tg at highest cure temperature suggests possible
various nylons, and numerous others] and some elastomers [eg, polychloroprene,
known commercially as neoprene (CR), cis-1,4 -poly(butadiene) (BR), ethylene–
propylene–diene polymer (EPDM)] are morphologically semicrystalline. These
are virtually two-phase systems where a crystalline phase and an amorphous
phase coexist. However, the crystalline phase and the amorphous phase are not
THERMAL ANALYSIS OF POLYMERS 17
Fig. 10. Four characteristic temperatures for melting transitions. In addition, the peak
area (heat of transition) and the time scale of the experiment should be recorded (11).
completely homogeneous. The crystals generally are of different sizes and forms
and the amorphous phase near the crystallite interfaces is constrained and less
mobile than the amorphous chains far from the crystallite boundaries. DSC is a
useful tool for characterizing these semicrystalline materials.
Other important characteristics of Semicrystalline Polymers (qv) are that
melting takes place over a range of temperatures (in some cases, over a range
of 100◦C or more) and supercooling always occurs, so that there is a hysteresis
between the melting and crystallization cooling curves. Semicrystalline polymers
melt at characteristic temperatures when heated in DSC. Wunderlich (Ref. 5,
p. 277) describes reasons for designating the melting temperature Tm at different
locations of the melting curve (Fig. 10). If there is a large temperature gradient
within the sample, it has been observed empirically that the peak temperature
often represents the actual melting temperature. Also, a sample with a broad
melting range and a small melting rate (as are the semicrystalline polymers with
crystals of different perfection, shape, and size) should be better characterized
by its melting peak temperature, since Tp represents the temperature where the
largest fraction of the sample melts. This is the most frequently quoted Tm for
semicrystalline polymers. Lastly, the temperature for recovery of the base line,
Te, is a function of the design of the instrument. For a polymer with a very broad
melting range, the instrument lag becomes negligible and Te would indicate the
melting temperature of the most perfect, highest melting crystals of the sample.
Another important characteristic of melting for semicrystalline polymers
using DSC is the enthalpy or the value of the heat of fusion,Hf. It should be noted
that the value ofHf, determined by DSC, denotes only the amount of crystallinity
18 THERMAL ANALYSIS OF POLYMERS
present in the sample and not the inherent value of the enthalpy of fusion of a
fully crystalline polymer, Hu. Methods for determining Hu are described by
Starkweather and co-workers (28). If the enthalpy of fusion of the fully crystalline
polymer is known, the degree of crystallinity of an unknown polymer sample can
be determined as follows:
Degree of crystallinity (%)=(Hf/Hu)×100 (2)
As in the case of the glass transition, another important item to be reported is the
heating or cooling rate. Equilibrium melting should be independent of heating
rate, but rarely do polymer crystals represent equilibrium crystals, nor do they
melt in DTA or DSC under equilibrium conditions. Under such nonequilibrium
conditions, the time scale of a first-order transition (such as melting) is an impor-tant
experimental parameter. It should be noted again that the magnitude of the
step change in the base line during the glass transition relates to the amorphous
content of the polymer. Similarly, the magnitude (area) of the melting transition
enthalpy relates to the degree of crystallinity of the polymer. The degree of crys-tallinity
of a semicrystalline polymer is an important parameter that correlates
with most major engineering mechanical properties including elastic modulus,
strength, and elongation to break. DSC is the most efficient method for accurately
measuring the degree of crystallinity.
Modulated Differential Scanning Calorimetry. A relatively recent devel-opment
in DSC instrumentation allows heating or cooling a sample under a con-stant
underlying rate while simultaneously superimposing a sinusoidally varying
time–temperature wave. This is shown in Figure 11 for the modulated differential
scanning calorimetry [MDSC, TA Instruments; Sauerbrunn and co-workers (29)].
As described below, the accepted name for this method has since been changed
to modulated temperature DSC, or MTDSC, after this term was introduced in
1996. A similar technology [dynamic DSC or step scan DSC (SCDSC) developed
by Perkin-Elmer] is based on a series of isothermal holds, each followed by a linear
heating segment (ramp) instead of the sinusoidal cycle of MTDSC. Schawe (30)
compares the two methods (MTDSC versus DDSC) using the glass transition as
an example. The results are virtually identical.
A major advantage of the MTDSC technique and analysis is that the total
heat flow rate can be separated into two other signals by deconvolution of the
raw data (Fig. 11b). One of these is a reversible signal, in phase with the mod-ulated
heating rate, and the other a nonreversible signal out of phase with the
modulated heating but dependent on the rate of heat flow. Examples of polymer
thermal events that are nonreversible are molecular relaxation, cold crystalliza-tion,
evaporation, thermoset cure, elastomer vulcanization, and decomposition.
The utility of having two response components is demonstrated in Figure 11c for
PET, where the glass transition and the melting endotherm are reversible but the
endothermic enthalpy relaxation at Tg and the exothermic cold crystallization are
not. Another advantage of MTDSC is that once the instrument is calibrated for
the amplitude and period, heat capacity data can be obtained directly in the same
run by measuring the modulated heat flow signal, and dividing by the amplitude
of the heating rate.
THERMAL ANALYSIS OF POLYMERS 19
Fig. 11. Examples of MDSC data: (a) typical temperature–time profile; (b) raw data for
an MDSC scan of quenched PET; (c) Deconvolution and analysis of the curve (29).
20 THERMAL ANALYSIS OF POLYMERS
Fig. 12. Modulated DSC curve for Makroblend UT-400, an impact-modified polycarbon-ate/
poly(ethylene terephthalate) blend. Curve A shows the conventional DSC curve; curve
B shows the heat capacity extracted from the reversing component of the signal; curve C
shows the modulus of the same material measured by DMA (Hale and Bair, in Ref. 5).
An interesting analysis of a blend of PET/polycarbonate (PC), using MTDSC,
was reported by Hale and Bair (Ref. 5, p. 804) (Fig. 12). The conventional DSC
curve (A) has all the indications of PET, but PC cannot be detected without the de-convoluted
reversing curve (B) that indicates PC is present by virtue of its Tg that
occurs in the same temperature range as the cold crystallization of PET. Further
confirmation of the presence of PC is obtained from the dynamic mechanical stor-age
modulus (curve C). The dynamic modulus technique will be discussed later in
In the MTDSC technique, Cp (heat capacity) data from the convoluted MDSC
is replaced with the term Cp
∗, called a complex heat capacity. Cp
∗ can then be
THERMAL ANALYSIS OF POLYMERS 21
further deconvoluted into an out-of-phase Cp
, and an in-phase (Cp
relative to the imposed temperature modulation. The vector sum of these com-ponents
is equal to complex heat capacity, the data traditionally generated by
MDSC. For most polymers, Cp
is very small. In those cases, Cp and Cp
∗ will be
virtually identical. However, a phase correction in MDSC should be relevant for
time-dependent heat capacity phenomena. Aubochon and Gill (31) present a good
comparison of MDSC and MTDSC with illustrations. The attention of interested
readers is also directed to the papers on MDSC and MTDSC presented at the 11th
ICTAC conference (1996), and later published in three volumes of the Journal of
Thermal Analysis (1997). NATAS, the North American Thermal Analysis Society,
has sponsored symposia on MDSC at each of its meetings for the past several
DSC Modifications and Simultaneous Techniques. DSC and other ther-mal
analysis instrumentation have undergone many modifications and develop-ments
in recent years. Among these innovations is the coupling of various meth-ods.
In this context the terms parallel, concurrent, and simultaneous should be
defined. Parallel techniques use separate samples, each in its own unique thermal
environment. Concurrent or combined techniques also use separate samples, but
the experiment is carried out in a common atmosphere and thermal environment.
Simultaneous techniques use the same sample in the same atmospheric and ther-mal
environment. Such methods are increasing in popularity. While a detailed
treatment of this topic is outside the scope of this discussion, a brief list will be
presented in order to provide input for those new to the subject area. Interested
readers should consult the chapter by Gallagher (5) and the additional references
Some of the simultaneous instruments available commercially are as follows:
TGA/DTA/DSC, TGA/DTA/FTIR, DTA(DSC)/EGA(EGD), and DSC/FTIR. Also,
high pressure DSC and photo-DSC instruments are available commercially. In
addition, individual researchers have used simultaneous DSC/XRD (X-ray diffrac-tion),
DSC/EGA/XRD, and DSC/TRXRD (time-resolved X-ray diffraction). Exper-iments
with parallel, combined, and simultaneous techniques help to affirm the
conclusions drawn from a single technique and very often offer definitive clues to
the actual mechanisms taking place during thermal analysis.
Thermal conductivities can be measured using DSC. Chiu and Fair (32) irre-versibly
modified a DSC cell to determine thermal conductivity. Sircar and Wells
(33) developed a modification of this approach, which allowed the use of the same
cell for conventional DSC work as well as for the measurement of thermal con-ductivity.
Marcus and Blaine (34) discuss the possibilities for using MTDSC for
measuring thermal conductivity.
The terms, thermogravimetry (TG) and thermogravimetric analysis (TGA) are syn-onymous.
Both ICTAC and International Union of Pure and Applied Chemistry
(IUPAC) accept either of them. This is because of the early use and popularity of
the term TGA, as well as an interest in avoiding verbal confusion with the glass
transition Tg. There may also be potential problems with computer searching.
22 THERMAL ANALYSIS OF POLYMERS
Fig. 13. Typical arrangement for the components of a TGA instrument (3).
Thermogravimetry involves the continuous recording of mass versus tempera-ture
or time as a sample is heated in a furnace with a controlled environment.
The sample may be heated at a constant rate or held at an isothermal tempera-ture.
Madorsky (35) and Jellinek (36) authored major books dealing with thermo-gravimetry
of polymers. The era of modern automated thermogravimetry started
with the introduction of the electrobalance by Cahn and Schultz (37). Other com-petitors,
such as DuPont, Mettler, and Perkin-Elmer, introduced their products in
Description of a TGA Instrument. The components of the instrument
are the microbalance, the furnace, the programmer controller, and a computer or
data acquisition system. Typical arrangements of the components for TGA are
shown in Figure 13. The sample can be linked to the balance in three different
ways: (1) above the balance, (2) below the balance, and (3) beside the balance as a
horizontal extension to the beam.
Arrangement 1 is the most common. However, arrangement 3 is actually pre-ferred
because it minimizes the heat effects of the furnace thatmay be encountered
in arrangement 2 and is less influenced by the flow patterns of the gases within the
balance and furnace chamber. It also lengthens the balance lever arm, increasing
sensitivity, and minimizes the problem of condensation of volatiles on the sample
support. In order to minimize thermal expansion of the balance lever arm, quartz
is frequently the beam material of choice.
The Balance. Figure 14 shows a schematic of the Cahn electrobalance (38)
most frequently used in thermogravimetry work (arrangement 1). A photodetec-tion
system monitors the beam position. If the beam moves from the horizontal,
enough current flows to the torque motor to move the beam back to its original
THERMAL ANALYSIS OF POLYMERS 23
Fig. 14. The Cahn electrobalance (19).
position. The restoring force generated by the current is proportional to the change
in mass. A sensitivity of 1 μg is easily achieved. The sensitivity and total capacity
of the balance have an inverse relationship. A commercial instrument with high
sensitivity (ELCHEMA of Potsdam, New York) has a sensitivity of 0.1 μg and an
operating range of up to 100 μg.
The Furnace and Controller. The heating elements of the furnace are most
often resistance heaters. Nichrome and Kanthal elements are used for the temper-ature
range up to 1000–1200◦C. This range is sufficient for most polymer appli-cations.
Aluminum sample pans are often used for low temperatures up to 600◦C.
For higher temperatures, platinum pans are frequently used as sample containers.
For higher temperatures in the range up to 1500–1700◦C, molybdenum disilicide
(Super Kanthal) or silicon carbide (Globar) is used for heating elements, and plat-inum
or platinum alloys for sample pans. Ceramic refractories such as mullite
or alumina are needed to contain the controlled atmosphere. Only a few man-ufacturers,
such as Linseis, Netzsch, and Setaram, market instruments for use
above 1700◦C. They are expensive and perform poorly below 300◦C. Such high
temperatures usually are unnecessary for polymers.
Most instrument manufacturers offer several furnace options in interchange-able
modules to cover a broad range of temperatures. Thermal shock, oxidation,
and vaporization shorten the life span of the resistance elements and refractories
so that they may need occasional replacement. The higher the temperature of
operation, the more severe this problem is. To enhance TGA instrument lifetime,
Gallagher (5) recommends current limiting devices, thermocouple protection cir-cuits
and conservative ratings. Also, short times at elevated temperatures and less
cycling to high temperatures will help. The use of dedicated computers and micro-processor
chips has markedly improved the effective sensitivity and temperature
24 THERMAL ANALYSIS OF POLYMERS
control of modern instruments. Speyer (39) describes many of the fundamental
considerations for the proper design of TGA equipment.
For checking uniform heating rates, the temperature (T) versus time plot (t)
is suggested. Wendlandt (40) found that replacing T in the above plot with the
derivative dT/dt shows temperature fluctuations better than those of the temper-ature
versus time plot. For further details, refer to the chapter by Gallagher (5).
Major Factors Affecting Thermogravimetry. Mass and temperature
are the two most important parameters that must be determined accurately for
proper characterization of a material by TGA. Table 5 (Ref. 5, p. 24) shows all of
the factors that need to be taken into account for accurate determination of either
mass (left-hand side) or temperature (right-hand side). However, it is easier to
deal with corrections for mass than it is for temperature. The following discussion
summarizes these factors.
Mass. The factors that may affect mass determination are buoyancy, atmo-spheric
turbulence, condensation and reaction, electrostatic and magnetic forces,
and electronic drift. Buoyancy corrections arise because the sample is in a lighter
gas environment as the temperature of the environment increases. This shows up
as a greater weight for the sample. The correction for this is minimal and, except
for most accurate or demanding work, it can be neglected. If needed, there are
suggestions for both instrumental and software adjustments that will take care of
the correction. Atmospheric turbulence depends on the specific gas, its flow rate
and pressure, as well as geometrical considerations.Wendlandt (4) discusses these
factors in detail and suggests necessary compromises.
Evolved volatile products may condense on the cooler parts of the sample
suspension system, thus altering the measured mass loss. The direction of flow
of the evolved gases must be manipulated to protect the vulnerable parts of the
balance mechanism, temperature sensor, and furnace windings from chemical
attack. Also, in case the TGA system becomes contaminated from the products
of the high temperature decomposition, subsequent experiments (particularly at
high temperature) may be affected. An easy solution for this is to heat the system
to around 600◦C in an oxidizing atmosphere. This will eliminate most organic
condensation products. If there is a chemical reaction with a component in the en-vironment,
there may be specific effects on the reversibility and the chemical equi-librium
involved. Examples of this are the decomposition of carbonates, sulfates,
hydroxides, and hydrates in the presence of CO2, SO3 and water, respectively. The
partial pressure of the gases present will determine the extent of the respective
reactions. Consequently, the degree of decomposition will be determined by the
ability to sweep away the decomposition products.
Table 5. Major Factors Affecting Thermogravimetry
Buoyancy Heating rate
Atmospheric turbulence Thermal conductivity
Condensation and reaction Enthalpy of the process
Electrostatic and magnetic forces Sample, furnace, and sensor arrangement
Electronic drift Electronic drift
THERMAL ANALYSIS OF POLYMERS 25
Catalytic reaction of the sample pan with the environment can also affect the
environment. Moreover, potential reactions of a sample with the sample holder
must be considered. For example, a silica crucible will lower the decomposition
temperature of CaCO3 by about 600◦C in the presence of the self-generated at-mosphere
of CO2. There is potential for reactions of materials such as alkalis and
alkaline earth carbonates to react with ceramic or metal parts of the sample holder.
The form or packing of the sample is another factor that will determine excess
of the degree of contact with gaseous environment and the rate of escape of the
volatile products. Thus, decomposition may be diffusion controlled and dependent
on the sample geometry instead of chemically controlled and dependent only on the
sample composition. Packing will also affect the thermal gradients in the sample.
Thus, results for a single “lump” of a sample, rather than a powder, may differ.
Low humidity of the surroundings and purge gases can generate electrostatic
forces that can cause the sample holder to be attracted to or even stick to the
nearest wall, thus disturbing the mass signal. This effect tends to be seasonal
(dry winter) and there are simple remedies, such as wiping glass surfaces with
conducting liquids or surfactants and providing a ground connection through thin
metallic film on the glass surface. If moisture can be tolerated, the gas stream can
be humidified to eliminate electrostatic charges or, if it is not too hot, the outer
wall surface can be wrapped with a moist covering. The electronic drift problem
usually is relatively minor and only becomes a problem for experiments lasting
for several hours.
Temperature. As mentioned above, accurate temperature control is more
of a problem in TGA than is weighing mass accurately. Referring again to the five
factors affecting accurate temperature measurement as listed on the right-hand
side of Table 5, geometrical placement of the sensor, the sample, and the furnace
are particularly important. The sample should be as near to the thermocouple as
possible in order to sense accurately the temperature of the sample, but should
not affect the measurement of weight. It should not undermine effective control
of the programmed rate of heating and cooling by virtue of the heating or cooling
generated by any exothermic and endothermic reactions that take place. Also, the
products of decomposition may have a corrosive effect on the thermocouple and
balance parts. Thus, these volatile products should not come in contact with sys-tem
components. Some of these requirements are contradictory, and compromises
must be made.
The most common configurations for the placement of the sample and the
thermocouple are shown in Figure 15 (Ref. 5, p. 30). This emphasizes the contra-dictions
in the above requirements and the uncertainty in measuring the actual
temperature of a sample. Very often, thermal runaway reactions caused by highly
exothermic processes (e. g., oxidative decomposition of polymers) show as foldback
curves (where the temperature increases and then decreases as heat is released)
in the mass or weight % versus temperature plot. There is no completely satis-factory
remedy for this. Using a high thermal conductivity atmosphere (helium)
reduces the response time for temperature control and permits better heat dissi-pation
or absorption by the sample but cannot completely alleviate the problem.
Also, at a high flow rate, the high thermal conductivity gas will increase the heat
loss from the furnace causing an increase in the power level necessary to obtain
the same temperature.
26 THERMAL ANALYSIS OF POLYMERS
Fig. 15. Possible placements of the thermocouple relative to the sample in TGA (Gal-lagher,
in Ref. 5).
Sartorius-Werke GmbH developed a clever solution (41). Here the balance
and the sample chamber are completely separate, linked magnetically with a sim-ple
suspension (Fig. 16), thus alleviating the problem of corrosive gases coming in
contact with the balance mechanism. A modern version of this “Thermosuspension
balance” is available from Netzsch.
A number of manufacturers supply TGA instruments capable of operation
in high vacuum or high pressure. A TGA with a magnetic suspension system is
available from Rubotherm (Bochum, Germany). The magnetic suspension sepa-rates
the high pressure sample environment from the delicate balance. The system
can operate up to 1000◦C or 450 atm. The readers should note that the pressure
and temperature limits cited throughout this article represent the manufacturers’
specifications, rather than any practical limits determined and recommended by
the editor or the authors. Operation at reduced pressure is particularly valuable
for determining the adsorption or absorption characteristics of materials. High
pressure instruments may be used for studies of combustion, pyrolysis, catalysis,
Mass and Temperature Calibrations and Standards.
Calibration of Mass. Calibration of mass is conducted by weighing a stan-dard
mass (over 1 μg) at a controlled temperature. Room temperature is prefer-able,
since buoyancy and aerodynamics add to the uncertainty when the experi-ment
is conducted over a wide range of temperatures. Any changes due to these
factors go into a blank or background correction.
For experiments over an extended period of time, such minor factors as fluc-tuations
in room heating and air conditioning, exposure to sunlight, drafts, and
other sources of temperature, and variable heat transfer from the furnace should
be taken into consideration. These factors, along with vibration stability, should
be considered when choosing a location for a TGA instrument. Generally, all TGA
equipment gives precise data with only infrequent checks of the calibration with a
standard mass. The starting mass for a TGA experiment is generally larger than
that for DTA/DSC and is in the range of 6–10 mg for pure polymers and around
15 mg for filled or blended polymers.
THERMAL ANALYSIS OF POLYMERS 27
Fig. 16. Sartorius Model 4201 magnetic suspension balance. 1, Magnet and coil; 2, beam;
3, beam fulcrum; 4, suspension ribbon; 5, beam magnet; 6, suspension magnet; 7, observa-tion
window; 8, upper glass body; 9, lower glass body; 10, sample pan (41).
For a simple calibration of weight loss, the weight loss of a standard material
can be checked under reproducible conditions of sample mass, packing, heating
rate, sample holder configuration and atmosphere type, flow, and pressure. The
TGA of calcium oxalate monohydrate (CaC2O4·H2O) is often used as a standard for
the calibration of mass loss in thermogravimetry. This is due to three well-resolved
steps in its thermal decomposition.
The three steps in Figure 17 are (1) the loss of H2O to form anhydrous ox-alate,
(2) the loss of CO to form the carbonate, and (3) the loss of CO2 to form
CaO. The observed weight losses, determined from the integral curve (TGA), may
be compared with the theoretical values obtained from the known stoichiometry
of the process. Potential differences in the sample temperatures indicated by the
28 THERMAL ANALYSIS OF POLYMERS
Fig. 17. TGAandDTGAcurves for the thermal decomposition of calcium oxalate (CaC2O4,
H2O) in argon at 20◦C/min (3).
TGA curve can arise when the starting masses differ significantly among the ex-periments.
Plots of weight versus time should reveal the linearity of the heating
rate in the range of the experiments. The derivative TGA curve (DTGA), shown
in Figure 17, enhances resolution and gives a “fingerprint” of the decomposition
process. DTGA curves always should accompany a TGA analysis. Such curves
also are important for kinetic studies, since they are related to the actual rate
of the reaction. Weight loss is, however, more readily evaluated from the inte-gral
curve. There is no international standard method for TGA, because all of
the instrumental and experimental variables noted above cannot be fixed for dif-ferent
instruments and different laboratories. However, for a given instrument,
standards for mass loss under carefully controlled conditions can be adopted for
routine checks of reproducibility.
Temperature Calibration. As mentioned, temperature control and calibra-tion
of TGA equipment is more difficult. Not only are the sample and the thermo-couple
a finite distance apart, so as not to interfere with the weight measurement,
but also the heat transfer between sample and oven in TGA is usually across an
air or an inert gas gap. In addition, sample masses are often larger than those in
DTA/DSC. There may be a large temperature gradient inside the sample. There
are also other factors noted in the right-hand side of Table 5 that would affect
Wunderlich (5) recommends that the actual sample temperature should be
measured by using an external thermocouple in contact with the sample in the
THERMAL ANALYSIS OF POLYMERS 29
crucible and then running a heating scan. The thermocouple adds additional
weight to the balance pan. Therefore, the mass information is lost. Two methods
commonly used to measure sample temperature in relation to the sensor temper-ature
are (1) magnetic standards and (2) “fusible links.” They have been found to
work equally well in a comparison study (42).
In the “fusible link” method a small weight hangs at the end of a metal link.
The link takes the normal position of the sample. When the link melts, there is a
sharp break in the TGA curve. The temperature at the break point is the Tm of
the metal, which should be known beforehand. The second method uses the Curie
temperature (Tc) of a ferromagnetic material. A pellet of the reference material is
placed in the field of a magnet, placed outside the balance. Because of the effect
of the magnet, the actual weight measured by the thermobalance is either less or
more than the true weight. As the Curie temperature of the standard material
is reached, the magnetic effect vanishes and an equivalent increase or decrease
of weight is recorded. A nice feature of using magnetic materials is that there is
no interaction between different samples and multimetal sandwiches of magnetic
standards can be used in a single experiment to obtain several calibration points.
Representative curves for a series of magnetic standards are shown in Figure 18
(43). The heating and flow rates used for calibration should be the same as those
chosen for subsequent experiments.
A series of magnetic reference metals are available from NIST. Gallagher (5)
lists the selected metals and alloys for potential use as magnetic standards. The
ICTAC Committee for Standardization has conducted an extensive round-robin
study to establish the calibration procedures and recommended temperatures for
various standards (44). An inert gas (N2, Ar) whose thermal conductivity ap-proximates
that of the gas to be used in subsequent experiments should be used
instead of oxygen, since these metals are liable to be oxidized. An ASTM method
describes the practice for evaluating the temperature scale for thermogravimetry
(ASTM E914). A recent proposal (Ref. 5, p. 48) is to use the ignition temperature
of selected organometallic compounds for temperature calibration. There is an
abrupt weight loss at the ignition temperature of these compounds. The tempera-ture
of ignition can be previously calibrated using DSC or DTA, which again has
been calibrated using metallic standards to conform to the International Tem-perature
Scale. ICTAC recommendations should be followed in reporting TGA
The present trend in thermal analysis is toward smaller samples and fur-naces,
as well as faster heating and cooling rates in order to obtain minimum
turnaround times. As these trends continue, temperature calibration will remain
a major consideration and probably increase in importance. For recommendations
on the presentation of data, the reader should consult ICTA recommendations (2).
Application of TGA/DTGA in Polymers
The principal applications of TGA/DTGA in polymers are (1) determination of the
thermal stability of polymers, (2) compositional analysis, and (3) identification of
polymers from their decomposition pattern. Also, TGA curves are used to deter-mine
the kinetics of thermal decomposition of polymers and the kinetics of cure
30 THERMAL ANALYSIS OF POLYMERS
Fig. 18. Representative curves for a series of magnetic standards: (a) Thermomagne-tometry
(TM) curves using Perkin-Elmer TGS-1 and their materials and recommended
temperatures. The values in parentheses were the observed values prior to calibration. (b)
DTM (derivative) curves after conversion of the sensor voltage to actual temperature (43).
THERMAL ANALYSIS OF POLYMERS 31
where weight loss accompanies the cure reaction (eg, as in condensation polymer-izations,
such as cure of phenolic resins). The latter will not be discussed here.
Thermal Stability of Polymers. Thermogravimetry has become a general
method for comparing the thermal stability of polymers. In comparing thermal
stability, it should be remembered that TGA measurements only record the loss
of volatile fragments of polymers, caused by decomposition. TGA/DTGA cannot
detect any chemical changes or degradation of properties caused by cross-linking.
Examples of Degradation (qv) involving cross-linking without weight loss are the
degradation of styrene-co-butadiene rubber (SBR) and butadiene rubber (BR) or
their blends, although at high temperature they decompose into fragments with
loss of weight that can be detected by TGA/DTGA. Small sample sizes and low
heating rates prevent any dramatic influence of reaction heat that accumulates
faster at high heating rates. However, there are some exceptions when small sam-ple
size and low heating rate do not yield satisfactory results.
A strict comparison of thermal stability can only be made when the sam-ples
are evaluated under identical conditions. Figure 19 shows an example of the
stability determination of poly(vinyl chloride) (PVC) and polystyrene (PS) and a
Fig. 19. TGA curves for PVC and PS films for 1:1 (by weight) blend of PVC and PS: 1,
predicted behavior for the blend based upon the PVC and PS curves; 2, behavior observed
experimentally for the blend (45).
32 THERMAL ANALYSIS OF POLYMERS
50:50 blend of the two polymers by weight (45). The two inside curves are for (1)
the predicted behavior for the blend, based on the neat PVC and PS curves, and (2)
the actual experimental curve for the blend. Under thermal decomposition con-ditions,
there is some possibility that any of the components or their degradation
products can interact with one another causing either (1) greater stability for the
blend, (2) less stability for the blend, or (3) no change in stability. The case of PVC
and PS seems to be one where the blend is more stable than the individual pure
polymers (Fig. 19). The shift of curve 2 to higher temperatures as compared to
curve 1 indicates that some interaction between the components is taking place
leading to mutual stabilization. This may be due to reducing the kinetic chain
length (and rate) for dehydrohalogenation of HCl in the PVC component. It will
also be dependent on the degree of mixing of PVC and PS in the blend. Examples
of polymer blends that are less stable than the pure polymers, to name a few,
are PVC/PMMA poly(methyl methacrylate) and (PVC)/PVA poly(vinyl acetate).
PS and PMMA blends show no evidence of interaction; thus the thermal stability
of their blends neither increases nor decreases.
Compositional Analysis of TGA/DTGA.
Copolymers. If different segments of the copolymer chain have different
thermal stabilities, it is possible to quantitatively analyze certain polymers by
TGA. The classical example is the polyethylene-co-(vinyl acetate) polymer or EVA.
When EVA is thermally degraded in an inert atmosphere, the first product emitted
at about 350◦C is acetic acid. The rest of the hydrocarbon chain decomposes at a
higher temperature (430◦C). This is shown in Figure 20 (Ref. 5, p. 870). The weight
loss at 350◦C corresponds to the amount of vinyl acetate present in the copolymer.
The amount of vinyl acetate can be calculated from the weight loss as follows:
%vinyl acetate=%weight loss×(1.43) (3)
Rubber Vulcanizates. One of the common uses of TGA/DTGA analysis in
the rubber industry is to determine the composition of a rubber formulation, either
cured or uncured. The general TGA/DTGA procedure for elastomer vulcanizate
analysis is illustrated in Figure 21 (46). The standard test method for composi-tional
analysis by TGA (47) provides a method for determining four arbitrarily
defined components: (1) highly volatile matter, (2) matter of medium volatility, (3)
combustible material, and (4) ash, left after oxidative decomposition of inorganic
components. The materials fitting this description in the case of elastomer vulcan-izates
are depicted in Figure 21 and described in Table 6. The definition of each
component is based on its relative volatility. The success of the method depends on
each component having a different thermal stability range in an inert atmosphere
and in an oxidizing atmosphere. The standard method for compositional analysis
by thermogravimetry is described in ASTM E1131.
The analysis is performed by first taring the microbalance and then introduc-ing
the sample. The initial weight is set after establishing an inert atmosphere.
Next the desired heating program is started and the specimen weight continuously
monitored in the recorder. The ordinate setting can be either weight in milligrams
or percentage of the original sample weight. As noted in Table 6, the first weight
loss (∼150–350◦C) is due to highly volatile matter. For a vulcanized rubber, this
THERMAL ANALYSIS OF POLYMERS 33
Fig. 20. Weight loss curves of several ethylene/vinyl acetate copolymers. The ethy-lene/
vinyl acetate composition is shown next to each curve. The heating rate was 10◦C/min
(Hale and Bair, in Ref. 5).
generally consists of the amount of oil added as a processing agent and a small
amount (∼1–2%) of decomposed fragments of the vulcanizing agent (sulfur per-oxide),
vulcanization accelerators and antioxidants, emulsifiers, etc. Moisture, if
present, evolves at a lower temperature. The second fraction (∼350–550 or 600◦C)
represents the amount of polymer decomposed. After the plateau of the second or
“medium volatile matter” weight loss has been established (∼600◦C), the atmo-sphere
is changed from inert to oxidative.
In the case of an elastomer vulcanizate that contains Carbon Black (qv), the
sample is first cooled to ∼300 to 400◦C before changing to air or oxygen. This is to
ensure that the initial oxidation atmosphere is at a lower temperature than the
carbon black oxidation temperature. The modified procedure often permits the
identification of carbon blacks from the weight-loss profile (48). The combustible
material refers to oxidizable material (eg, carbon black, graphite), which is not
volatile in an inert atmosphere at temperatures up to 750◦C. In the absence of
nonblack fillers, the ash is mostly composed of zinc oxide, which is a component in
most vulcanization recopies. The analysis is complete when a weight-loss plateau,
corresponding to the residual sample mass, is established. Several reviewers have
stressed the use of smaller sample sizes (for better resolution, lower heating rate)
and an inert atmosphere of high thermal conductivity (eg, helium). In using small
34 THERMAL ANALYSIS OF POLYMERS
Fig. 21. Schematic TGA and DTGA curve for compositional analysis of a rubber vulcan-izate
Table 6. Definitions of Different Fractions in Thermogravimetric Analysis of
Highly volatile matter Refers to moisture, polymer, diluent, oil, plasticizer,
emulsifiers (eg, in styrene–butadiene rubbers), curatives
(sulfur, accelerator), antioxidants, antiozonants, and other
low boiling components (approx. 300◦C or lower)
Medium volatile matter Refers to medium volatility material such as processing oil,
processing aid, elastomer, resin (used as curing agent), etc.
In general, these materials degrade at 300 to 750◦C.
Combustible material Refers to oxidizable material, not volatile (in the unoxidized
form) at 750◦C or some stipulated temperature dependent
on the material (eg, carbon black, graphite, etc).
Ash Refers to nonvolatile residues in an oxidizing atmosphere
which may include metallic oxides, filler or inert reinforcing
material (eg silica). In the absence of nonblack fillers, the
ash is composed of zinc oxide which is a component in most
vulcanizates. A small amount of ash (1%) may be due to
aFrom Ref. 47.
THERMAL ANALYSIS OF POLYMERS 35
samples it may be useful to characterize multiples to ensure that the results are
representative of bulk material. The residue may be analyzed to determine the
nature and amount of inorganic components by using X-ray fluorescence.
Factors that influence the accuracy of the component analysis procedure
are the following. The above discussion assumes that there is no high boiling oil
present as a processing aid that can interfere with detecting clearly the initial
breakdown point of the polymer and no mineral filler (which may have water of
crystallization or other constituents that escape around the temperature of the
“highly volatile matter”). Also, it is important to note that the determination of
carbon black by the suggested method is accurate only for hydrocarbon elastomers,
which degrade cleanly in an inert atmosphere at temperatures below 500◦C Elas-tomers
with a heteroatom in the molecule leave a char residue, which volatilizes
along with carbon black during the oxidation stage. Resin curing agents also will
interfere with polymer determination. Graphite, if present, can be determined by
oxidation at a higher temperature (∼800◦C). These and numerous other consid-erations,
as well as possible remedies, have been discussed in considerable detail
by Sircar (49).
Despite the above-cited difficulties, TGA-DTGA analysis remains a useful
tool for the compositional analysis of rubber vulcanizates. It is a straightforward,
reasonably accurate process and is much faster than classical extraction methods.
Also, it is an excellent quality control tool for determining possible weighing errors
and reproducibility for elastomer batches in a production setting. Thus, TGA-DTGA
is the method of choice for compositional determinations for cured and
uncured elastomer compounds, as well as for many other polymer compositions.
Identification of Elastomers. Sircar and co-workers (Ref. 5, p. 1278) de-veloped
a unique and very useful DSC and TGA/DTGA protocol for identifying
elastomers. It involves using the measurement of Tg and the degradation pat-terns
for many elastomers and their blends in both nitrogen and oxygen. On the
basis of this, they classified elastomers into three principal groups according to
their degradation characteristics in nitrogen: (1) those exhibiting an endotherm
(eg, polyethylene, EPDM, butyl rubber); (2) those exhibiting an exotherm (eg, SBR,
BR, NBR); and (3) those exhibiting multiple peaks, both exotherm and endotherm
(eg, NR, IR). For details on using the protocol, the reader is referred to the above
Controlled Rate Thermogravimetry. Gallagher (5) describes controlled
rate thermal analysis. In the case of mass loss, the rate of change of temperature is
controlled by a preset rate of weight loss or change in partial pressure. Gallagher
describes three different modes of controlled rate TGA and also dynamic TGA
with isothermal rest periods. In the first mode, the operator chooses a fixed rate
of weight loss (eg, X%/min) and the temperature programmer maintains that rate
of change, sometimes by both heating and cooling at different times. The second
mode is similar to the first mode, except that the sample is always dynamically
heated. TA Instruments, Inc. (50) markets a commercial software package called
“Hi-Res TGA.” It operates so that the heating rate approaches a preset maximum
during periods of no mass loss. As the mass change is sensed, however, the heating
rate automatically slows depending upon two parameters set by the operator (res-olution
and sensitivity). The aim is to improve resolution while minimizing the
total time for the experiment. The third controlled rate option is called stepwise
36 THERMAL ANALYSIS OF POLYMERS
Fig. 22. Weight loss curves for a 72:28 ethylene/vinyl acetate copolymer. Curve A was
obtained at 10◦C/min rate of heating. Curve B was obtained using Autostepwise software,
which automatically slows down the heating rate when a sharp change in weight loss is
detected, and increases the rate when little change in weight loss is observed (from Hale
and Bair, in Ref. 5).
(or autostepwise) isothermal measurement and involves one or more isothermal
steps as the weight loss reaches a threshold level. The necessary software and in-strumentation
are available through Seiko Instruments (currently marketed by
An example of autostepwise TGA shows an improvement in resolution for
weight-loss curves of EVA samples. Without the autostepwise rate control it is
evident (Fig. 20) that the end point for the acetic acid evolution is not distinct.
The acetic acid weight loss curves are not completely level before the residual
polymer fraction starts decomposing. As shown in Figure 22, the resolution can
be much improved by using the High-Res or autostepwise TGA technique enabling
more quantitative information. The figure shows an initial weight loss of 18.4%,
corresponding to a vinyl acetate content of 26.3% versus the known value (supplied
by the manufacturer) of 28%.
Simultaneous or Combined Techniques. Many of the simultaneous or
combined techniques that include TGA as a component have been described while
discussing DSC/DTA techniques. Others with TGA as a component are TGA-EGA
or EGD, TGA-MS, TGA-FTIR, TGA-GC-MS, TGA-MS-MS, and TGA-APCI-MS.
APCI stands for atmospheric pressure chemical ionization. More details are
THERMAL ANALYSIS OF POLYMERS 37
available in the reference by Gallagher (5). The other acronyms, not mentioned
in Table 1, are as follows: mass spectrometry (MS), Fourier transform infrared
Thermomechanical analysis (TMA) measures the deformation of a material con-tacted
by a mechanical probe, as a function of a controlled temperature program,
or time at constant temperature.TMAexperiments are generally conducted under
static loading with a variety of probe configurations in expansion, compression,
penetration, tension, or flexure. In addition, various attachments are available to
allow the instrument to operate in special modes, such as stress relaxation, creep,
tensile loading of films and fibers, flexural loading, parallel-plate rheometry, and
volume dilatometry. The type of probe used determines the mode of operation of
the instrument, the manner in which stress is applied to the sample, and the
amount of that stress.
TMA Instrumentation. For accurate measurements of the change in di-mension
of small samples, the device generally used by most manufacturers of
TMAis called a linear voltage differential transformer (LVDT). This device, as con-figured
in a Perkin-Elmer TMA module, is shown in Figure 23. For distinguishing
between expansion and contraction, the sign of the output voltage changes with
the direction of motion. The device can readily detect changes of the order of 0.1
μm. Testing for linearity of the range of output voltage of an LVDT (mV/μm) is
frequently performed by measuring gauge blocks of a sheet metal. For accurate
work, it is necessary to provide compensation for the dependence of the output on
temperature. The instrument shown in Figure 23 is typical of a single rod module,
based on LVDT. Contemporary TMA instruments use closed loop electromagnetic
circuits to accomplish constant loading and positioning of the probe as compared
to older instruments where mechanical placement of the weights was used. The
instrument in Figure 23 is reported to be capable of detecting changes as small
as 3 nm and can operate in the temperature range −170 to 1000◦C. The resolu-tion
is around 0.1 μm, when the instrument is operated carefully over a range
of temperatures. Temperature-sensitive components such as the LVDT generally
are isolated from the high temperature environment of the sample.
Optical instruments that use laser interferometry have much greater sensi-tivity
with an accuracy of 0.02 μm or 1/32nd of the laser wavelength used. The
surfaces for the sample and the reference specimens need to be polished for this
kind of accuracy. Also, the high temperature limit is governed by the loss of reflec-tion
from the mirrored surfaces and is around 700◦C, as compared to a maximum
of 2000◦C for the LVDT-based instruments.
For determining the coefficient of linear expansion, it is a common practice
to generate L/L0 curves with and without the sample. The two curves are sub-sequently
subtracted to compensate for the expansion of the sample holder. For
convenience, it is best to use a standard reference material, eg, an isotropic metal
such as aluminum or platinum (52), as the sample holder. In Figure 24, we see
that, compared to other isotropic materials, quartz (fused silica) has a very low
thermal expansion coefficient (0.6 × 10−6/◦C). It is also easily formed. If the upper
38 THERMAL ANALYSIS OF POLYMERS
Fig. 23. Schematic representation of a typical thermomechanical analyzer (TMA), Perkin-
Elmer Model TMA 7, based on an LVDT (3).
THERMAL ANALYSIS OF POLYMERS 39
Fig. 24. TMA thermal expansion curves for platinum, alumina, and fused quartz (52).
temperature limit (1100◦C) is acceptable, then use of a quartz sample holder may
permit ignoring the small correction for the linear thermal expansion of the sam-ple
holder, with very little effect on accuracy. If the upper temperature limit is
over 1100◦C, a high temperature material such as alumina or platinum is the
material of choice and the correction is substantial. Another way to reduce the
correction is to use the instrument in a differential fashion. Both the optical and
the LVDT instruments can be used in this fashion with the reference (a standard
material) pushrod connected to the coil and the sample rod to the core. To ob-tain
the correction accurately, the sample and the reference should have the same
length. It is also necessary to maintain thermal equilibrium by keeping the heat-ing
and cooling rate low. This also avoids a thermal gradient inside the larger-sized
samples typically used for TMA. It is necessary to allow sufficient time to equili-brate
if programming several isothermal steps is the temperature profile followed.
40 THERMAL ANALYSIS OF POLYMERS
A temperature calibration procedure for TMA has been proposed (53–55) and
subsequently included as an ASTM method (Test Method for Temperature Cali-bration
of Thermomechanical Analyzers, E1363-90). It uses a penetration probe
and the melting temperature of one or more standard materials. Pure metals with
sharp melting points are the standards often used. An open DSC pan may be used
to contain the calibrant material. Another potential material would be the selected
shape memory alloy, reported to be reproducible to ±1◦C (56). Several reviews on
temperature calibration for TMA have been published based on ASTM efforts in
this area (54,55). Sircar (26) suggests that, when used for elastomer evaluation,
temperature calibration for TMA should be conducted with low melting liquids as
in DSC. For calibration of the expansion, one manufacturer’s manual (TA Instru-ments)
recommends aluminum for calibrating the linear expansion parameter.
Other calibration standards suggested for the linear coefficient of thermal expan-sion
(CTE) are lead (57) and copper (58).
Sample Preparation and Procedure. Sample preparation for CTE mea-surement
is somewhat more rigorous for TMA than for either DSC or TGA. Sam-ples
with flat parallel faces (∼7 mm2 × 1 to 2.5 mm in thickness) may be used.
The larger the sample thickness, the greater will be the possibility of thermal
gradients developing within the sample. ASTM D696 describes a recommended
procedure for determining the coefficient of linear expansion of plastics. A general
test method for solids (Test Method for Linear Expansion of Solid Materials by
Thermomechanical Analysis, E831-86) is also available. Not all polymers and, par-ticularly
some elastomers, are amenable toTMAstudy. For example, a flat surface,
to allow intimate contact of the sample and the probe, cannot be cut from a gum
elastomer of high molecular weight. Maurer (59) has recommended hot pressing,
which works for lower molecular weight gum rubbers. Silica powder also has been
used to level the surface of gum elastomers (60). For penetration measurements
the surface does not need to be smooth.
For Thermosets (qv), a flat surface can be achieved by molding, machining,
or melt-pressing the sample. If the sample is punched, the burred edges must be
removed or the sample placed burred edge up. Prime (Ref. 5, p. 1447) advises that
the sample be conditioned in the TMA apparatus by heating to just above Tg and
cooling with the loaded probe in place. This helps in relieving the internal stress,
removes thermal history (eg, the effects of physical aging), and allows the sample
to conform to the probe. For samples that are not fully cured, care must be taken
not to advance the degree of cure while following this procedure.
For Tg studies in elastomers, the ASTM E831-86 procedure must to be mod-ified
(26). After the probe is zeroed at room temperature, the sample is placed
on the TMA platform and cooled to −120◦C under liquid nitrogen purge at 50
mL/min. After equilibration at −120◦C for 10 min, a 0.5–1-g weight (10 mN) or
equivalent force is placed on the top of the pan. This places the probe in contact
with the hardened elastomer sample. The thickness of the sample is measured
and the probe zeroed. The sample is ramped at a low rate of heating (2–5◦C/min)
under 50 mL/min of nitrogen purge from −120◦C to 100◦C. For presentation of
data, the ICTAC recommendations (2) are followed.
The above procedure differs from ASTM E831 in two important ways. First,
the weight should be added only after the elastomer sample is rigid (below Tg)
to avoid compression; and second, a small weight of only around 0.5 g or less is
THERMAL ANALYSIS OF POLYMERS 41
necessary to ensure intimate contact. Larger weights (1–3 g, as recommended for
plastics by ASTM E831) may again compress the elastomer as it reaches tempera-tures
above Tg. A static weight of 10 g or more or the equivalent force is generally
used with a penetration probe.
Probe Configuration for TMA. Typical probe configurations for TMA
have been described by Maurer (59) and Earnest (25). Figures 25a to 25e are from
Earnest (25), whereas 25f and 25g are fromWendlandt (4). The coefficient of linear
expansion uses a probe of greater surface area and is operated under negligible
load (Fig. 25a). A high loading converts it to a compression probe for elastomeric
materials and the results differ with load. Figures 25b, 25c, and 25f show various
penetration probes. The hemispherical probe can also be used as an expansion
probe. Ennis and Williams (61) used the hemispherical probe to determine linear
Fig. 25. Examples of probe geometry or operation modes for TMA: (a) from Neag
(62); two Perkin-Elmer TMA sample probes for penetration and dilatometry (b) from
42 THERMAL ANALYSIS OF POLYMERS
expansion of thermosets and claim better accommodation of probe misalignment
with this type of probe. Figures 25d and 25e are for tensile and flexure modes,
respectively. For anisotropic materials, the determination of linear expansion does
not give the volume expansion by a simple calculation. However, TMA also can
be used to determine volume expansion by using a fluid to surround the sample
in a cylinder and piston arrangement, as shown in Figure 25g. This is similar to
a dilatometer. Another method uses a vitreous silica dilatometer (ASTM E228).
Special instrumentation has been designed to determine the coefficient of thermal
expansion of fibers using TMA (Ref. 5, p. 1974). For “fingerprinting” of fibers,
TMA offers more detail than either DSC or DTA. Therefore, TMA is the preferred
technique if the study is limited to only a single method.
Derivative TMA. An example illustrating the use of different probe con-figurations
for determining linear expansion in both the glassy and the rubbery
states, as well as Tg of chloroprene rubber (neoprene, CR) (62), is presented in Fig-ure
26. Note the use of the derivative signal (DTMA) for both the loaded expansion
(penetration) and tension measurements. The derivative plot aids in locating the
transitions or relaxations, which are not prominent, and also defines a repro-ducible
transition point in cases where a diffuse transition is encountered. DTMA
of tension measurements is a sensitive technique for evaluating polymer blends.
Several transitions are reproducibly detected by this method when other modes
may fail. The Tgo values are the glass-transition temperatures reported in the lit-erature.
They correspond to the inflection point of the DSC curves (Teo) in Figure
7. Tgd is slightly different and corresponds to the peak of the derivative DSC curve
(see Tp in Figs. 7 and 8). Sircar (26) determined Tg values for eight elastomers
using TMA (expansion and penetration probes), DSC, and DMA. He found that
Tg(expansion) and Tg(penetration) have somewhat different values and the latter
is closer to that determined by DSC. The differences in the Tg values are due to
the differences in the temperature dependence of the different thermal properties
being measured by different techniques: specific heat capacity in the case of DSC,
linear expansion with the TMA expansion probe, and modulus in the case of the
TMA penetration probe.
Illustrative Examples for TMA.
Expansion Mode. Figure 27 (64) illustrates the calculation of the CTE from
the slope of the TMA curve of an epoxy printed circuit board and the use of the
derivative (DTMA) to directly record the coefficient of linear expansion (α). Ther-mosetting
polymers, especially when highly filled, can absorb moisture from the
atmosphere. Moisture acts as a plasticizer and has the effect increasing α and
lowering Tg. The process is reversible. While moisture in polar polymers is often
tightly bound, desorption may occur on heating during the measurement. This will
be discussed at greater length below. In this case the measurement will reflect the
sum of thermal expansion and hygroscopic contraction. Purely thermal expansion
then can be measured in a second run after the sample has been thoroughly dried.
Penetration Mode. TMA in the penetration mode can be used to evaluate
the degree of cure of thermoset materials (64). A large load (typically, 10–100 g or
0.1–1.0 N) is employed for these experiments. The probe measures the softening
associated with the large decrease in modulus as the temperature approaches
Tg. The softening point correlates well with Tg (65,66). The penetration probe
referred to in Figure 28 is a small-radius (0.48 mm) hemispherical probe. The
THERMAL ANALYSIS OF POLYMERS 43
Fig. 26. TMA free expansion, indentation, and tension curves for polychloroprene (neo-prene)
vulcanizates; 5◦C/min, 50-g load in indentation and tension.—, l versus tempera-ture;
-˙-˙-, d(l/dT) versus temperature (63).
44 THERMAL ANALYSIS OF POLYMERS
Fig. 27. Linear expansion before and after the glass-transition range using TMA-DTMA.
Linear expansion coefficient α from TMA trace requires measurement of the slope; DTMA
allows direct recording of α (64).
smaller amount of penetration for sample A indicates that this is more highly
cross-linked of the two samples.
Flexure Mode. Figure 29 illustrates TMA data for a two-layer insulation
wire coating. Using the knife-edged three-point bend flexure probe (67) accentu-ates
the softening that occurs at the glass-transition temperature. It is remark-able
that the softening of both the outer layer of polyester and the inner layer of
poly(amide–imide) were easily detected.
Degree of Cure. As shown in Figure 27, the Tg is taken as the extrapolated
intersection for the glassy and rubbery state curves. Typically, thermal expansion
in the rubbery state is about three times that in the glassy state. The Tg deter-mined
by TMA (using either an expansion or penetration probe), is often related
to the degree of cure for thermosetting polymers, as was mentioned in the dis-cussion
of DSC and illustrated in Figure 9. As the degree of cure increases, Tg
increases and the coefficient of linear expansion (α) decreases. Thus, a relation-ship
between Tg and conversion, independent of the temperature of cure, exists
for most thermosetting systems. This will be discussed further later in this pre-sentation.
Also, a direct proportionality exists between the deflection temperature
under load (ASTM D648) and the TMA Tg determined in either the expansion or
THERMAL ANALYSIS OF POLYMERS 45
Fig. 28. TMA penetration tests to detect relative states of cure of a thermoset polyester.
The more highly cross-linked material softens at a higher temperature, undergoes less
penetration, and is more abrasion-resistant. – – – No load; 50-g load (64).
the penetration mode. Sykes and co-workers (68) describe a fixture to measure a
TMA heat distortion temperature (HDT) as defined in ASTM D648. For printed
circuit board materials, Lamoureux (66) observed a correlation between Tg by
TMA expansion and softening temperature determined by TMA penetration (as
described above by Sykes and co-workers). For an elastomer compound, Brazier
and Nickel (63) found a correlation between Tg, determined by TMA (expansion
and penetration mode) and Gehman rigidity temperature.
Unlike thermosetting materials, the Tg of elastomers does not increase sig-nificantly
with the degree of cure. However, the penetration temperature under a
fixed load will increase with cure and can be used as a measure of the degree of
cure of an elastomer compound. This is illustrated in Figure 30 (69) and is based
on the fact that the elastomer modulus increases with cure.
Dynamic Mechanical and Dielectric Analysis
Dynamic mechanical (DMA) and dielectric analysis (DEA) methods are used both
to study molecular relaxation processes in polymers and to determine inherent
46 THERMAL ANALYSIS OF POLYMERS
Fig. 29. Use of the TMA knife-edged flexure probe to determine the softening point of
both the outer and inner coatings of a motor-winding wire (67).
mechanical and dielectric properties. DMA measurements involve mechanical ex-citation
by imposing a small cyclic deformation to a sample and measuring the
stress response. DEA involves imposing a cyclic (or alternating) electric field to a
sample and observing the electric polarization or current induced in the sample
(see DYNAMIC MECHANICAL ANALYSIS; DIELECTRIC RELAXATION).
DMA methods are widely used by thermal analysts to determine the vis-coelastic
properties of polymers for a number of purposes (see VISCOELASTICITY).
The primary application of these techniques to the study of polymeric solids and
melts is well documented. Excellent general discussions covering the subject are
provided in References 70–72. Linear Amorphous Polymers (qv) exist in a number
of characteristic physical states depending on the time scale and temperature of
measurement. These are illustrated in Figure 31 in terms of an arbitrary modulus
function and are classified as glassy, leathery, rubbery, rubbery flow, and viscous
(73). All linear amorphous polymers exhibit these five physical states when they
THERMAL ANALYSIS OF POLYMERS 47
Fig. 30. Evaluation of the degree of cure of an elastomer by TMA penetration probe;
sample is a carbon black filled rubber stopper (69).
are observed over a wide range of time or temperature. Polymers that are either
cross-linked or crystalline do not exhibit the rubbery flow and viscous responses
Region 2 in Figure 31 corresponds to the Glass Transition (qv), which serves
as an important benchmark or corresponding state for viscoelastic response. The
particular physical state that the polymer exhibits during a DMA measurement
reflects the extent to which the rate of molecular motion or relaxation compares
to the test rate at a given temperature. In general, the modulus decreases either
with decreasing test rate or increasing temperature. In fact, there is a strong
relationship between time and temperature.
In most commercial dynamic mechanical instruments a steady-state alter-nating
strain is applied such that
ε (u)=ε0 eiωu (4)
is the generalized sinusoidal strain function in complex notation, where ε0 is the
strain amplitude, ω is the radian angular frequency of the sine wave (rad/s), and
u is the historical time, u ≤ t (u = 0 at t = 0). An analysis of the cyclic stress and
strain provides that the steady-state ratio of stress to strain is (70,73) a complex
quantity having both in-phase and out-of-phase components.
48 THERMAL ANALYSIS OF POLYMERS
Fig. 31. Diagram of master curves for modulus and loss tangent for a typical amorphous
polymer showing five characteristic physical states (73).
=E∗ =E +iE (5)
Here σ(t) is the stress at time t and ε is the strain; E∗ is the complex dynamic
mechanical modulus, E is the ratio of in-phase stress to the applied strain, and
E is the ratio of out-of-phase stress to strain. The out-of-phase stress leads the
strain by 90◦. Further, it can be shown that E is related to the mechanical energy
stored per cycle and E is related to the energy converted to heat through viscous
dissipation. As a result, E is referred to as the storage modulus and E is called
the loss modulus. The material loss factor or loss tangent is
tan δ =E
representing the ratio of energy dissipated to energy stored per cycle of defor-mation.
In a physical sense the storage modulus is related to the stiffness of the
material and the loss modulus is reflected in the damping capacity of the material.