Você está na página 1de 15

View Article Online / Journal Homepage / Table of Contents for this issue

June, 1980 FUNDAMENTALS AND APPLICATIONS OF DTA AND DSC 217

Fundamentals and Quantitative Applications of DTA


and DSC
The following are summaries of ten of the papers presented at a Meeting of the Thermal
Methods Group held on November 15-16th, 1979, a t Stoke Poges, near Slough, Berkshire.
Introduction
Published on 01 January 1980. Downloaded by Brown University on 30/10/2014 06:50:54.

C. J. Keattch
Honorary Secretary, R S C A nalytical Division Thermal Methods Group, P.O. Box 9, L y m e Regis, Dorset,
DT7 3BT
Following on the success of the meeting held at Fulmer Grange on November 23-24, 1978,
entitled “Fundamentals and Applications of Thermogravimetry,” it was decided to hold a
further meeting, in 1979, a t the same venue, on “Fundamentals and Quantitative Applications
of DTA and DSC.” The remit given to the invited speakers was that they should present an
overview of their subject area rather than describe original work. All authors were invited to
provide summaries and it is a pleasure to note that this invitation was accepted unanimously.
These summaries are published below and any enquiries for further information should be
addressed to the individual authors.
It may also be of interest that the Thermal Methods Group has planned a further meeting at
the same venue for November 13-14, 1980, entitled “Fundamentals and Applications of
Thermomechanical Methods, including Thermodilatometry” and that other titles for further
meetings are under consideration.

Differential Thermal Analysis and Differential Scanning Calorimetry:


Similarities and Differences
R. C. Mackenzie
Macaulay Institute f o r Soil Research, Craigiebuckler, A berdeen, A B 9 2Q J
The Nomenclature Committee of the International Confederation for Thermal Analysis
(ICTA) have defined differential thermal analysis (DTA) as1 :
“A technique in which the temperature difference between a substance* and a reference
material is measured as a function of temperature whilst the substance and reference
material are subjected to a controlled temperature programme.
The record is the differential thermal or DTA curve: the temperature difference should be
plotted on the ordinate with endothermic reactions downwards and temperature ( T )or time
(t) on the abscissa increasing from left to right”;
and differential scanning calorimetry (DSC) as1 :
“A technique in which the difference in energy inputs into a substance and a reference
material is measured as a function of temperature whilst the substance and reference
material are subjected to a controlled temperature programme.
Two modes, power-compensation DSC and heat-flux DSC, can be distinguished depending
on the method of measurement used.”
It might thus seem that the only similarities are in the use of a substance and reference
material in each instance: in reality, however, one can recognise two end-members, conven-
tional DTA and power-compensation DSC, with a rather grey intermediate area. The last
point is well illustrated by the definition of quantitative DTA,l namely, “this term covers
those uses of DTA where the equipment is designed to produce quantitative results in terms of
energy and/or any other physical parameter.”
The Two End-members
In conventional DTA, the basic equipment would be based on something like the system
shown diagrammatically in Fig. l(a), where AT is the difference between Ts and TR,the
temperatures of the sample and reference material, respectively. Let us assume, for the
* In this and %hesubsequent definition substance should be interpreted as substance andlor its reaction
products.
View Article Online

218 FUNDAMENTALS AND APPLICATIONS O F DTA AND DSC Anal. Proc.

r-----’
- 4
Pt sensors
0
I
* O
Published on 01 January 1980. Downloaded by Brown University on 30/10/2014 06:50:54.

ATand T
T
registering Energy Individual
regulator + controller
system heaters
(a1 (6)
Fig. 1. (a) Block diagram of equipment suitable for DTA.2 (b) Basic specimen holder system for
power-compensation DSC (Perkin-Elmer arrangement).

present argument, that the junctions of the difference thermocouple are, as in the classical
system, a t the centres of the specimens [see Fig. 2(a)]: then, because of heat transfer through
the specimens, change in thermal diffusivity with temperature, etc., the curve is qualitative or
a t best semi-quantitati~e.~p~y~ Although the critical points of the peak have little physical
significance, the peak is very little distorted from the theoretical shape and this is indeed the
best arrangement for qualitative DTA.6 The system and a typical curve for an endothermic
reaction are shown diagrammatically in Fig. 3 ( a ) .
In power-compensation DSC [Fig. l(b)] the sample and reference material are supplied with
separate heaters and are maintained a t nominally the same temperature by a servo system,
operated by the platinum resistance thermometers, supplying different amounts of heat to each
specimen. As the parameter recorded is the difference in power inputs to the heaters, as
d(AQ)/dt [Fig. 2(f)],an endothermic peak appears on the positive side of the ordinate [Fig.
3 ( b ) ] and the curve is essentially quantitative for specific heat, heat of reaction, etc. (Fig. 4).
The two end-members are thus distinct but possible overlap occurs in the intermediate
sphere.

DTA Ts - T R = AT

LEU Heat-flux DSC

U3
-E Heat- flux DSC

uu
PQm
Power-compensation DSC d Q s l d t - d o Rl d t = d(AQ)/dt

-
Fig. 2. The DTA DSC series described in the text: T , temperature: S, sample; R, reference
material ; A, difference; dq/dt, spontaneous thermal flux ; dQ/dt, compensatory thermal flux ;
i\, thermcouple; 3 , thermopile; and p o l , heater (adapted from SestSk et a1.7.
View Article Online

Jww, 1980 FUNDAMENTALS AND APPLICATIONS O F DTA AND DSC 219


Quantitative DTA and Heat-flux DSC
In 1925 the enthalpy change of a reaction was first related to the area enclosed by the
deviation it caused on the heating curve,‘ and in 1945 SpeiP derived an expression
mAH = ghSt2ATdt
Published on 01 January 1980. Downloaded by Brown University on 30/10/2014 06:50:54.

(where m is sample mass, AH enthalpy change, g a geometrical constant, h thermal conduc-


tivity and the integral term is peak area), which indicates that sample mass, enthalpy change
or thermal conductivity can be determined from peak area when the other factors are con-
~ t a n t . This
~ relationship, the basis of quantitative DTA, is independent of heating rate5 but
applies only to a curve for temperature difference against time: on a curve for temperature
difference against temperature, peak area is approximately proportional to heating rate.1°
Many simplifications were introduced in obtaining this expression and more rigorous theories
have later been developed by Sewellll and Wilburn,G but these reduce essentially to the above
if special precautions are taken.

Controlled

rAT1 d(AQ1
dt

Fig. 3. Diagrammatic representation of systems for,


and endothermic peaks obtained by, (a) conventional
DTA, and ( b ) , power-compensation DSC; symbols as in
Fig. 2 (adapted from SestAk et aZ.s).

The optimum specimen-holder system for quantitative DTA measurements was shown by
Wilburn6 to consist of two separate metal containers enclosed in an air space and sitting on the
thermocouple junctions [see Fig. 2(b)j-the system which, as the thermal conductivity of the
sample can be neglected, was earlier used by Schwiete and Zieglerl2 for so-called “dynamic
differential calorimetry.” A development of this system is that of Boersma,13who showed
that introduction of a controlled heat leak between the sample and reference holders [see Fig.
2(c)] rendered the system quantitative for energy measurements : the theories behind these
various systems have been discussed by Rouquerol and Boivinet14 and by SestAk et al.3
If a sample is surrounded by a thermopile, as in the Tian-Calvet calorimeter,14heat flux can
be measured directly. Consequently, two thermopiles surrounding the sample and reference
material and connected in opposition-as in the differential Calvet system [see Fig. 2(e)]-un-
questionably give heat-flux DSC. A simpler system suggested by Petit et al.,14s15which again
represents heat-flux DSC, is to measure the heat flux between the sample and the reference
material, dqA/dt, by a thermopile having all “hot junctions” in contact with the sample and all
“cold junctions” in contact with the reference [see Fig. 2(d)].
Where is the Dividing Line?
From the above descriptions, there are at least three possible DSC systems [Figs. 2(d), (e) and
(f)]and three derived from DTA [Figs. 2(a), ( b ) and (c)], the latter grading into DSC. All of
these systems are now found in commercial instruments : thus, many manufacturers produce
View Article Online

220 FUNDAMENTALS AND APPLICATIONS O F DTA AND DSC Anal. Proc.


~

Endot he rrn Fusion peak

Second-order
(glass) transition
Published on 01 January 1980. Downloaded by Brown University on 30/10/2014 06:50:54.

Deflection directly
proportional
to specific heat

IsotherrnaI
Isothermal

I I I I I I I I I IIIIIIIII 11111111I111111111 1 I I I I I I I I I I I I I I I I I I I 11111IIII I I I 1 1 1 1 1 1 II

Ternperature/K

Fig. 4. Typical power-compensation DSC curve for a polymer, as obtained with the Perkin-Elmer
system, showing sample attributes that can be directly measured.

equipment that could be represented by Figs. 2 ( a ) and 2 ( b ) , that in Fig. 2(c) is produced by
Du Pont (their “DSC cell”), Fig. 2 ( d ) by Mettler (the 2000 series), Fig. 2(e) by Setaram and
Fig. 2(f)by Perkin-Elmer and Rigaku. The question thus arises as to where the dividing line
should come. The above discussion and the diagrams in Fig. 2 indicate that the systems in
Figs. 2(a) and 2 ( b ) are DTA and those in Figs. 2 ( d ) , 2(e) and 2(f) are DSC, leaving that in Fig.
2(c) as an intermediate. I n view of Boersma’s theory,13 there is certainly justification for
considering this variant as just falling into the DSC field-and the Nomenclature Committee of
ICTA have taken this view in their fourth rep0rt.l However, an argument could possibly be
made for leaving the system within the purview of DTA. In Fig. 2, therefore, the indication
of the two fields (on the left of the diagram) is shown to overlap a t this point, with a suggestion
that the Boersma system falls within the DSC field. c
In conclusion, the reader will have noted that only endothermic reactions are considered
above; similar conclusions may also apply to exothermic reactions, but the position is not
clear, as such reactions may well require different measures to avoid interference with pro-
portionality by such parameters as thermal conductivity or diffusivity.
References
1. Mackenzie, R. C., Thermochim. Acta, 1979, 28, 1.
2. Mackenzie, R. C., and Mitchell, B. D., i n Mackenzie, R. C., Editor, “Differential Thermal Analysis,
Volume 1, Fundamental Aspects,” Academic Press, London and New York, 1970, p. 63.
3. SestAk, J., Holba, P., and BArta, R., Silikdty, 1976, 20, 83.
4. Cunningham, A. D., and Wilburn, F. W., in Mackenzie, R. C., Editor, “Differential Thermal Analysis,
Volume 1, Fundamental Aspects,” Academic Press, London and New York, 1970, p. 36.
5. Mackenzie, R. C., and Mitchell, B. D., i n Mackenzie, R. C., Editor, “Differential Thermal Analysis,
Volume 1, Fundamental Aspects,” Academic Press, London and New York, 1970, p. 101.
6. Wilburn, F. W., P h D Thesis, University of Salford, 1972.
7. Andrews, D. H., Kohman, G . T., and Johnston, J., J . Phys. Chenz., 1925, 29, 914.
8. Speil, S., Tech. Pap. B u r . Mines, Wash., 1945, No. 664, 1.
9. Kostomaroff, V., and Rey, M., Silic. Ind., 1963, 28, 9.
10. Proks, I., Silikdty, 1961, 5, 114.
11. Sewell, E. C., “Theory of Differential Thermal Analysis,” Research Notes, Building Research Station,
Watford, 1952-56.
12. Schwiete, H. E., and Ziegler, G . , Ber. Dtsch. Keram. Ges., 1955, 35, 193.
13. Boersma, S. L., J . Am. Ceram. Soc., 1955, 38, 281.
14. Rouquerol, J., and Boivinet, P., in Mackenzie, R. C . , Editor, “DifferentialThermal Analysis, Volume
2, Applications,” Academic Press, London and New York, 1972, p. 23.
15. Petit, J . L., Sicard, L., and Eyraud, L., C.R. Acad. Sci., 1961, 252, 1741.
View Article Online

June, 1980 FUNDAMENTALS AND APPLICATIONS OF DTA AND DSC 22 1


Criteria for Quantitative DTA and DSC
F. W. Wilburn
Pilkington Bros. Ltd., Research Laboratories, Lathom, Ormskirk, Lancashire L40 5 U F
Thermal effects are produced in both DTA and DSC by heat changes within the sample under
examination. Thus the area under a DTA peak must be related in some way to the heat
change producing the effect. Indeed, in DSC, heat energy is supplied to either sample or
Published on 01 January 1980. Downloaded by Brown University on 30/10/2014 06:50:54.

reference material to nullify any temperature difference which would have been produced by
the heat effect.
In addition, DTA and DSC peaks vary in shape depending on the type of reaction mechanism
involved.
These two observations have led to the development of theories of DTA from which it has
been claimed to be possible to relate area to the heat effect and to determine the reaction
mechanism involved from the peak shape.
Theories of DTA
Theories for DTA can be broadly divided into two types: those which relate peak area to
the heat of reaction and those which relate peak shape to reaction mechanism.
Theories of the first type utilise heat transfer equations to develop relationships between the
heat of reaction, H , the thermal conductivity, A, density, p, and specific heat, C,, of the sample
and the area of a DTA peak. Such equations1-12 have limited application as some of the
assumptions made are not valid, and often the heat transfer via the holder and the reaction
mechanism are not considered.
Theories of the second type manipulate various reaction mechanisms to show that various
relations exist between peak temperature and heating rate, between peak shape and reaction
mechanism from which the “activation energy’’ and other constants of reaction equations may
be determined. However, these theories do not include the influence of heat transfer and
these, as those mentioned earlier, have very limited use.13--18
In a practical DTA experiment the production of a thermal effect is affected by the reaction
mechanism and the heat transfer characteristics of the cell system. Equations that include
both heat transfer and reaction equations are often intractable, but it is possible using finite
difference techniques to calculate continuously the temperature profile across the sample and
reference materials within various types of cylindrical “holder” as the peripheral temperature
is increased linearly with time. It is also possible to vary the physical properties A, C, and p
with temperature while the reaction proceeds. Further, a reaction equation can be introduced
into the finite difference equations, which may take the form

If dor/dt is plotted against T as T increases linearly with time, t , a curve similar to a DTA
curve is produced. In fact, this is the DTA curve for a substance reacting according to equa-
tion (1) if heat transfer has no effect on the resultant curve. The DTA curve produced from
the finite difference equations under similar conditions includes the effect of heat transfer, and
thus a comparison of this curve with that produced from equation (1) shows the effect of heat
transfer. Indeed, any shape distortion can be attributed to heat transfer.
Two types of DTA cell system have been investigated by this technique.
Type A [in which the sample and reference materials are contained within cylindrical wells
within a highly conducting metal block ( A high) and temperatures are measured a t the centres
of sample and reference materials]. It is found that the peak shape distortion is minimal, and
that the peak temperature is little different from that found using equation (1) when the
difference temperature is plotted against sample (not reference) temperature. Peak shape
distortion increases with the sample radius (i-e., increasing sample size). The area of a DTA
peak is related to the heat of reaction as follows:
A = mH/4h, .. .. .. .. - - (2)
where A, = sample thermal conductivity, m = sample mass and H = heat per unit mass.
View Article Online

222 FUNDAMENTALS AND APPLICATIONS O F DTA A N D DSC Anal. PYOC.


Thus the heat of reaction cannot be found unless the thermal conductivity, As, is known.
Calibration is possible only if the conductivity of the material used as a calibrant is the same as
that of the test material.
Type B (in which the sample and reference materials are contained in crucibles in a gaseous
space and in which temperatures are measured below the crucibles). I n this type of cell, peak
shape distortion is evident even on small samples and the peak temperature is always
higher than that found from equation ( l ) ,even when the difference temperature is plotted
against sample temperature. The area under the peak is related to the heat of reaction as
Published on 01 January 1980. Downloaded by Brown University on 30/10/2014 06:50:54.

A = mH/2;\,
where AH is the conductivity of the gaseous space, i.e., the area is independent of the sample
physical properties. Thus this cell is ideal for quantitative use.
The base line of a practical DTA curve is often a t a different level before and after a peak due
to changes in the physical properties of the sample during the reaction which produces the peak,
and thus it is difficult to determine the base line to be used for area determination. By careful
consideration of the results of the finite difference theory a constructional method is proposed
(Fig. 1) by which it is possible to define a more realistic and accurate base line by which area
can be measured rather than that drawn across the “base” of the peak.

T, Time

Fig. 1. Base line construction from computed results in order to define


the area under a DTA peak. The area is always too large when constructing
straight line (3).

Conclusion
The influence of cell design factors on a DTA peak is outlined in summary from results
gained using finite difference methods involving heat transfer and reaction equations.
It has been shown that the cell design plays an important role in determining whether the
cell can be used for qualitative or quantitative purposes, and that other equations which do
not involve both heat transfer and reaction equations have very limited use.
References
1. Vold, M. J., A n a l . Chem., 1949, 21, 683.
2. Borchardt, H. J., and Daniels, P., J . A m . Chem. SOC.,1957, 79, 41.
3. Borchardt, H. J., J . Inorg. Nucl. Chem., 1960, 12, 252.
4. Blumberg, A. A , , J . Phys. Chem., 1959, 63, 1129.
5. Speros, D. M., and Woodhouse, R. L., J . Phys. Chem., 1963, 67, 2164.
6. Padmanabhan, V. M., Saraiyo, S. C., and Sundaram, A . K., J . Inorg. Nucl. Chewa., 1960, 12, 356.
7. David, D. J., A n a l . Chem., 1964, 36, 2162.
8. Arens, P. L. “A Study on the DTA of Clays,” Druk Excellors Foto-Offset, s’Gravenhage, 1951.
9. Boersma, S. L., J . A m . Ceram. SOC.,1955, 38, 281.
10. Sturm, E., J . Phys. Chem., 1961, 68, 1939.
11. Pacor, P.,A n a l . Chim. Acta, 1967, 37, 200.
12. de Jong, G., J . A m . Ceram. SOC.,1957, 40, 42.
13. Kissinger, H. E., Anal. Chem., 1957, 29, 1702.
View Article Online

June, 1980 FUNDAMENTALS AND APPLICATIONS OF DTA AND DSC 223


14. Piloyan, G. O., and Norikova, 0. S., Nature (London), 1967, 212, 1229.
15. Reich, L., J. Inorg. Nucl. Che.m., 1966, 28, 1329.
16. Freeman, R. S., and Carroll, B., J . Phys. Chem., 1958, 62, 394.
17. Reed, R. L., Weber, L., and Gottfried, B. S., I n d . Eng. Chem. Fundam., 1965, 4, 38.
18. Murray, P., and White, J . , Trans. BY. Ceram. Soc., 1955, 54, 204.

Calibration of Differential Thermal Analysis Equipment; the ICTA


Tem perat ure Standards
Published on 01 January 1980. Downloaded by Brown University on 30/10/2014 06:50:54.

E. L. Charsley
Consultancy Service, Stanton Redcroft, Copper Mill Lane, London, S W17 OBN
Differential thermal analysis (DTA), in common with the other thermal methods of analysis,
is a dynamic technique and the results obtained are dependent on the experimental conditions
used, e.g., heating rate, sample mass and also on the type of equipment. In order to be able to
make useful comparisons of the results obtained by different workers there is clearly a need for
standard materials. This paper discusses the development of DTA temperature standards,
covering the range from sub-ambient temperatures to 1000 O C , by the Standardisation Com-
mittee of the International Confederation for Thermal Analysis (ICTA).
The objectives of the Committee, founded in 1965, were to provide DTA temperature
standards that would : provide a common basis for relating independently acquired data ;
provide a means for comparing and calibrating all available instrumentation regardless of
design; and provide a means for relating thermoanalytical data to physical and chemical
properties determined by conventional isothermal procedures.
I t was intended that by including curves given by standard materials, obtained under the
conditions of the particular study, it would be possible for a reader to relate the subject matter
to the performance of his own instrumentation, to evaluate the quality of the published data
and hence the conclusions derived from them.1,2
The Committee first considered standards for the range ambient to 1000 "C. Although
nielting- or freezing-points have traditionally been used as temperature standards for calibra-
+
tion under near equilibrium conditions it was decided that solid I solid I1 first order phase
transitions would be preferable for use in DTA. The prime reasons were that not all equip-
ment could be used with samples that melted and that a number of melting-point standards,
e.g., lead or zinc, would attack thermocouples and sample holders by means of alloy formation.
For metals melting a t high temperatures, a completely inert atmosphere would be required
which would be beyond the capability of a number of DTA systems. Additionally, changes in
physical properties, e.g., heat capacity or specific volume, would be likely to be lower with
solid I + solid I1 transitions and the latter would be less likely to be affected by impurities.
Amongst the criteria for the systems to be selected were chemical stability and inertness, the
characterisation of the transition by equilibrium thermodynamic methods and their avail-
ability commercially in high purity at a cost that would not inhibit large scale distribution.
Of the 112 systems considered 12 were chosen for the 1st International Test Programme and
the results of this programme, published in 1968l led to the selection of eight provisional
standards and the establishment of a 2nd International Test Programme in 1969. This 2nd
programme involved 34 experienced workers, in 13 countries, and employed 28 different types
of commercial or "own-design" equipment. After requests by a number of workers, two low
temperature metal melting-points, indium and tin, were also included in the programme.
Before obtaining the provisional standard in bulk, specimens of each material from the
various sources were evaluated to determine which showed the best thermal behaviour, i.e.,
well defined peak shape, absence of spurious peaks, etc. Portions were distributed to partici-
pants in the programme, who were supplied with detailed instructions asking them to run the
samples under the normal operating conditions of the equipment a t a heating rate between 4
and 10 "C min-l (with cooling at the same rate if possible) in a dry nitrogen atmosphere.
Samples were not to be diluted or pre-treated in any way, apart from potassium nitrate, which
was to be cycled once to 150-160 "C before use. At least two runs were to be carried out on
each sample. The accuracy of the temperature measuring thermocouple was required to be
known, and if calibration was necessary the use of recognised melting-point standards was
preferred.
View Article Online

224 FUNDAMENTALS AND APPLICATIONS O F DTA AND DSC Anal. Proc.


Based on the experience of the First Programme, the extrapolated onset and peak tempera-
tures were selected for measurement as shown in Fig. 1. The departure point or first detect-
able deviation from the base line had been found to give poor reproducibility, being highly
dependent on the judgement of the operator. The results of the Second Programme were
published in 1971, and are summarised in Table I, together with the reported equilibrium
temperature values. The ranges and standard deviations represent the consistency that can
be achieved by experienced workers using the same samples and different equipment. Indi-
vidual workers may achieve higher precision. No material was found to give unusually
Published on 01 January 1980. Downloaded by Brown University on 30/10/2014 06:50:54.

precise or imprecise results and therefore no material was rejected. The mean onset tempera-
ture was found to be closest to the equilibrium thermodynamic transition temperature. The
results of the cooling studies confirmed that irreproducible supercooling took place with a
number of materials, so that the standards were recommended to be used in the heating mode
only. Participants were also asked to measure the areas of the DTA peaks in terms of "C mg-1
in order to enable the Committee to evaluate the potential use of the materials as peak-area
standards. A very wide spread of results was obtained, even when each set of results was
related to a common material (potassium sulphate) and further work is required in this area.,

TABLE
I
INTERNATIONAL
RESULTSOF SECOND TESTPROGRAMME
Measured temperatures/"C
r
Extrapolated Equilibrium
Material onset Peak value/ "C
KNO, 128 f 5 135 & 6 127.7
(112-149) (126-160)
In 154 6 159 46 157
( 140-1 62) (140-1 7 1)
Sn 230+ 5 237 f 6 231.9
(217-240) (226-256)
KClO, 299+ 6 309 f 8 299.5
(280-310) (296-330)
&,SO4 424 7 433 &- 7 430
(400-439) (405-452)
SiO, 571 f 5 574 f 5 573
(552-58 1) (560-5 88)
K'2S04 582 7 588 f G 583
(560-598) (575-608)
K,CrO, 665 f 7 673 f 6 665
(640-678) (656-692)
BaCO, 808 f 8 819 f 8 810
(783-834) (800-841)
SrCO, 928 f 7 938 f 9 935
(905-948) (910-96 1)

Participants in this programme were also asked to prepare and run 4 : 1 mixes of silica -
potassium sulphate, using the two standards supplied. The mixture was found to be a reason-
able system for testing the ability of instruments under a particular set of experimental
conditions to resolve two transitions occurring about 10 "C apart.
The ICTA Certified Reference Materials are issued by the United States National Bureau of
Standards as 3 sets: GM-758: KNO,, In, Sn, KClO,, Ag2S0,; GM-759: KClO,, Ag,SO,, SiO,,
K,SO,, K,CrO,; GM-760: SiO,, K,SO,, K,CrO,, BaCO,, SrCO,. The sets are accompanied by a
detailed certificate of over 60 pages on the test procedures and results, including an analysis of
the influence of experimental parameters.
In 1973 the Committee established a 3rd International Test Programme to evaluate tem-
perature standards below 100 "C. It was decided to use liquid samples because these samples
could be handled by the majority of low-temperature DTA units and considerably more
+
information was available on low temperature solid liquid transitions than on solid 1 solid +
2 transitions. It was recognised that impurities would have a greater effect on the former, but
this was considered not to be a significant problem, as a particular batch of a material and not
the niaterial itself was being certified. On the basis of work by the Committee four samples
with well defined melting-points were chosen. One of them, cyclohexane, also gave a solid
View Article Online

June, 1980 FUNDAMENTALS AND APPLICATIONS O F DTA AND DSC 225


1 + solid 2 transition. The fourteen laboratories participating in the programme were given
essentially the same instructions as in the 2nd Programme and were asked to determine
extrapolated onset and peak temperatures. For calibration of the sample measuring thermo-
couple, under near-equilibrium conditions, distilled water and triply distilled mercury were
suggested.
The results of the 3rd Programme are shown in Table 11. All investigators obtained well
defined peaks and good reproducibility. The smaller over-all deviations in this programme
compared with those for the high temperatures were attributed, at least in part, to the fact that
Published on 01 January 1980. Downloaded by Brown University on 30/10/2014 06:50:54.

fewer different types of instrument were used. The four materials were certified by ICTA and
are available from the NBS as set GM-757, together with a 38-page certificate.
TABLEI1
OF THIRD
RESULTS TESTPROGRAMME
INTERNATIONAL
Measured temperatures/K
r
Extrapolated Equilibrium
Material onset Peak value/K
1,2-Dichloroethane 237.3 f 2.0 241.6-4.3 237.5
(235-2 43) (236-252)
Cyclohexane (t) 187.1 f 3.5 190.9f 4.0 186.3
(184-1 92) (185-196)
Cyclohexane (m) 278.0 & 1.1 280.2 f 1.9 279.9
(276-280) ( 276-2 83)
Phenyl ether 298.6 f 2.2 301.9 6 2.3 300.1
(293-3 03) (299-307)
o-Terphenyl 328.2 f 2.2 331.1 f 3.1 329.4
(325-332) (327-339)

In order to cater for the needs of polymer chemists, the Committee also turned its attention
to the development of standards for glass transition temperature measurement. Two
materials were initially chosen, namely polypropylene and polystyrene, and tests led to
rejection of the former because of its lack of stability on storage. However, a particular
sample of polystyrene developed by the Rubber and Plastics Research Association showed
both good stability and homogeneity, and formed the basis of the 4th International Test
Programme in 1973, in which 24 laboratories using eight different types of equipment took
part. Participants were asked to measure, to the nearest 0.1 O C , the three temperatures
T a ,Tb and T,, as defined in Fig. 2, at four different heating rates within the ranges 4-6, 8-12,
16-24 and 30-50 "C min-l.

Extrapolated onset
temperature

I Peak temperature 1 Ta Tb Tc
Temperature Temperature
Fig. 1. Specified points for DTA Fig. 2. Specified points for glass transition
temperature measurement. temperature measurement.

The programme led to certification of the material, unweighted mean values of 104.4 &
1.5 "C and 107.5 & 1.7 "C being obtained for the defined points T b and T,. The point T , was
rejected for certification because of its relative lack of reproducibility and the subjective nature
of its determination. The Certified Reference Material is distributed by the NBS under
View Article Online

226 FUNDAMENTALS AND APPLICATIONS O F DTA AND DSC Anal. Proc.


number GM-754. The accompanying certificate includes a detailed study of the dependence
on heating rate of the glass transition and the influence of instrumental parameters on the
selected temperatures.
The Committee is also active in other thermal analysis fields and is in the process of preparing
temperature standards for the calibration of thennobalances. In the area of DTA, potential
standards are still being sought for temperature calibrations above 1000 “C.

References
Published on 01 January 1980. Downloaded by Brown University on 30/10/2014 06:50:54.

1. McAdie, H. G., in Schwenker, R. F., and Garn, P. D., Editors, “Thermal Analysis,” Volume 1,
Academic Press, New York, 1969, p. 695.
2. McAdie, H. G., i n Wiedemann, H. G., Editor, “Thermal Analysis,” Volume 1,Birkhauser Verlag, Basel,
Switzerland, 1972. D. 591.
3. McAdie, H. G., in BGzas, I . , Editor, “Thermal Analysis,” Volume 1, Akademiai Kiado, Budapest,
1975, p. 251.

Measurement of Heat Capacity


P. G. Laye
Department of Physical Chemistry, University of Leeds, Leeds, LS2 DJT
Thermal analysis has shown considerable scope for the measurement of heat capacity. There
is a range of commercial equipment and the experimental aspects of the technique are simple,
albeit deceptively so. The technique is applicable to solids and liquids and only small amounts
are needed, usually 10-50 mg. For comparison, the classical technique is adiabatic calori-
metry, in which a sample is heated electrically and the temperature rise measured. In
various forms this technique covers the temperature range from nearly 0 to 1800 K and
the results provide a reference with which we may compare those of thermal analysis.
Precisions of better than 0.1% have been reported and in those laboratories which have access
to primary physical standards, high accuracy can be expected. In contrast, thermal analysis
is more restricted in its temperature range, from about 100 to 1000 K, and is not an absolute
technique. Almost invariably comparative measurements are made on the sample and a
standard substance for which the heat capacity has been well established usually via adiabatic
calorimetry.
Theoretical Considerations
The theory of thermal analysis depends on a description of the temporal and spacial distribu-
tion of heat in the sample and its surroundings. The effects of assumptions which are made in
the mathematical treatment may often be minimised in instrument design or embraced in an
appropriate calibration. An analysis set out by Gray1 more than a decade ago is useful for
indicating the basis of the thermal analysis technique, although the analysis itself makes
sweeping assumptions. The model is that of a sample and a reference assumed to be a t
uniform temperatures T , and Tr, respectively. The exchange of heat between the sample or
reference and the surroundings is represented by Newton’s law, dq/dT = AT/R, where AT is
the temperature gradient and R is the thermal resistance. Two extreme techniques are con-
sidered : differential thermal analysis and power-compensated differential scanning calori-
metry. In the former instance the steady-state signal, T , - T,, established when both the
sample and reference are subjected to the same linear temperature increase, is proportional to
the difference between the heat capacity of the sample and reference. The proportionality
involves the thermal resistance which may be difficult to reproduce from experiment to
experiment. In power-compensated differential scanning calorimetry the relationship between
the steady-state signal and heat capacity does not involve the thermal resistance. On the
basis of this analysis the use of power-compensated differential scanning calorimetry appears to
be advantageous and probably most thermal analysis values have been obtained via this
technique. However, from the experimental standpoint the distinction between many of the
current thermal analysis techniques is now much less marked.
View Article Online

June, 1980 FUNDAMENTALS AND APPLICATIONS O F DTA AND DSC 227


Practical Considerations

The basis of the comparative technique is three experiments involving (i)an empty container,
(G)the same container and the sample and (iii)the container and the standard substance. In
each experiment the instrument signal is recorded for the same steady-state initial temperature,
T I ,during a programmed linear temperature rise and for the same final temperature, T,, when
the instrument is returned to the isothermal mode of operation. Each experiment is carried
Published on 01 January 1980. Downloaded by Brown University on 30/10/2014 06:50:54.

out with the same reference, almost invariably an empty container, and with the same
instrument settings. The standard substance is usually synthetic sapphire, which is available
in the form of discs which fit the aluminium or gold pans used as the containers. The heat
capacity of the substance is obtained directly from the relative magnitude of the signals. The
procedure has been described in considerable detail by O’NeilP in the context of power-
compensated differential scanning calorimetry but may be applied more generally to other
instrumental techniques. Indeed, many of these techniques are now classified under the
generic title of “differential scanning calorimetry” even though they do not involve power
compensation. The alternative approach in which the heat capacity is derived from the
enthalpy increment He(T,) - He(T,) is much less common but with substances of known heat
capacity it forms the basis of a calibration procedure.
There are aspects of this apparently simple experimental procedure which require more
detailed scrutiny. Fundamental to the success of the technique is that the base line of the
instrument should be reproducible. Curvature, which was marked with some earlier instru-
ments, need not undermine the reliability of the results, as it is the relative magnitudes of the
signals that are important. Uncertainty in the base line is shown by an inability to reproduce
the relative magnitudes of the isothermal signals a t the beginning and end of the experiments.
Arbitrary corrections may be applied but of concern is the inadequate representation of
instrument behaviour afforded by the simple analysis. Attention to experimental detail may
minimise the problem. The base of pans should be flat : as obtained from the manufacturers
this is not always so. The pans should be placed in the apparatus in precisely the same posi-
tion in each experiment. I n some equipment, domed covers may be fitted over the pans to
reproduce the emissivity. In some circumstances the use of different pans for the sample and
standard substance may be unavoidable, which introduces a correction for the difference in the
heat capacity of the two pans. A significant improvement can be made with some equipment
by thermostating the sample holder a ~ s e m b l y . ~
A potential source of error lies in the assignment of temperature. A calibration graph
should be established for the instrument using several standard substances of known melting
temperatures, under experimental conditions similar to those adopted for the heat capacity
measurements. A single point calibration can, in some circumstances, lead to significant
error in the assignment of temperature. Another aspect is thermal lag whereby the tempera-
ture of the sample lags behind the nominal instrument temperature. The effect has been
examined by Richardson and ~ o - w o r k e r sand
~ ~ although
~ the theory was developed for power-
compensated differential scanning calorimetry the principles are applicable to other differential
scanning techniques. The nature of the sample, its geometry, the interface between the
sample and pan and pan and heating block all contribute to the thermal lag. The effect may
be assessed from the signal trace when the instrument is returned from the scanning to the
isothermal mode of operation a t the final temperature. Generally, thermal lag has a small
effect on the apparent values of heat capacity. The exception is in the regions of some
transitions, e.g., glass transitions, where the heat capacity may be more strongly temperature
dependent and errors of several per cent. may arise. The existence of thermal lag between the
pan and heating block reinforces the need to ensure that the pans are flat and used in the same
position. The point is well illustrated in a recent paper6 where it is also shown that difficulties
may arise from the use of silicone oils, ostensibly to reduce thermal lag. Slow scanning speeds
(< 5 K min-l) should be used unless special circumstances prevail. The consequent reduction
in sensitivity can, if necessary, be offset by increasing the mass of the sample. Although this
in turn increases the thermal lag in the sample itself, generally the effect of slow scanning
speeds is to reduce the total thermal lag. The nature and geometry of the sample should be a
prime consideration : when poorly packed the sample may show significant thermal lag but if
pressed may be atypical due to changes in the morphology. Extensive decomposition,
View Article Online

228 Anal. Proc.


F U N D A M E N T A L S A N D APPLICATIONS O F D T A A N D DSC

evaporation or sublimation will invalidate the measurements and as a routine check the pan
and sample should be weighed before and after each experiment.
The comparative technique incorporates a calibration of the instrument, which should be
reproducible and independent of the nature of the sample. Variation of the calibration with
the magnitude of the signal is deleterious and may be examined using sapphire discs of different
masses. The effect may be minimised by using appropriate masses of sample and standard
substance to obtain similar signal magnitudes. Reproducibility in the scanning speeds must
be confirmed experimentally, as any variation will have a direct effect on the magnitude of the
Published on 01 January 1980. Downloaded by Brown University on 30/10/2014 06:50:54.

signal.

Conclusion
The thermal analysis technique permits heat capacities to be obtained easily with an un-
certainty of about 1-3%. This was so even for early power-compensated differential scanning
calorimeters, but over the restricted temperature range from ambient to about 500 K. Con-
siderable care is necessary to increase the precision to better than 1%. The advent of second-
generation power-compensated differential scanning calorimeters and other precise quantita-
tive equipment has increased the ultimate precision and moreover extended the temperature
range over which measurements may be made. The improved reproducibility of the base lines
with modern instruments allows greater temperature intervals to be scanned so that deter-
minations may be made over a wider range of temperatures in one experiment.
A survey of the literature over the last 5 years serves to indicate the greater range of
equipment which is currently in use for heat capacity measurements. A wide range of sub-
stances have been investigated, with the study of polymers being particularly rewarding. A
significant change has been the replacement of the traditional chart recorder with its restrictive
influence on the ultimate precision of the technique. In its place have come data aquisition
systems and computer processing, which facilitate the detailed quantitative assessment of the
data needed to obtain the high precision required for many present day applications.

References
1. Gray, A. P., in Porter, R. F., and Johnson, J . M., Editors, “Analytical Calorimetry,” I’lenum, New
York, 1968, p. 209.
2. O’Neill, M. J . , A n a l . Chem., 1966, 38, 1331.
3. Barrall, E. M., 11, and Dawson, B., Thermochim. Acta, 1974, 8, 83.
4. Richardson, M. J., and Rurrington, P., J . Thermal Anal., 1974, 6, 345.
5 . Richardson, M. J., and Savill, N. G., Thermochim. Acta, 1975, 12, 213.
6. Yuen, H. K., and Yosel, C. J . , Thermochim. Actn, 1979, 33, 281.

Entha Ipy Determination


M. J. Richardson
National Physical Laboratory, Tcddington, Middlesex, T W11 OL W
The ordinate in a DSC trace is a power (albeit a power difference) and the abscissa a time (the
co’mmon representation of the latter as a temperature is due to the good approximation to
linear heating (or cooling) rates of modern calorimeters; these, however, work equally well
isothermally, as with reacting or crystallising systems, where time is the relevant axis.]
Integration of a power versus time curve gives an energy and this, expressed as joules (prefer-
ably), corresponds to some enthalpy ( H ) change. Exactly uhat change depends on the con-
struction used to define the area (integrand) and the remainder of this paper will seek con-
structions that give thxmodynamically meaningful quantities.
Graphical Methods
The soundest method is obviously that which gives heat capacity (C,) versus temperature
curves as described a t this meeting by Laye.l Unfortunately, this can be a tedious procedure
for those without computing facilities and the need can be circumvented for certain applica-
tions. Consider, for example, the melting process shown schematically in Fig. 1. The full line
View Article Online

June, 1980 FUNDAMENTALS AND APPLICATIONS O F DTA AND DSC 229


represents a complete “run” from Tl (in the solid, s) to T , (in the liquid, 1). I t typifies condi-
tions when no care has been taken in setting up the instrument-gross curvature and disparity
in levels for isothermal values. Any kind of “estimated” base line has little significance, but a
meaningful base line is very often immediately available by simply returning to Tl and re-
running to T,. All liquids supercool to some extent and it is the rule, rather than the excep-
tion, that the supercooled liquid (broken line, Fig. 1) can be obtained. Let the empty pan
base line be as shown. H1(T2)- Hs(Tl)is proportional to the shaded area ABDG (A,) and
H1(T2)- H1(Tl) to the dotted area AJFG(A,), so that AH(T,) = H1(Tl) - Hs(Tl) = K ( A l -
Published on 01 January 1980. Downloaded by Brown University on 30/10/2014 06:50:54.

A,) = K ( A , - Ab),where A , and A b are the areas CDE and AJCB, respectively, in Fig. 1. In
this particular instance, therefore, no empty pan base line is needed (provided that the area to
enthalpy conversion factor, K , is known ; see below) and instrumental curvature is automatic-
ally eliminated. The resultant quantity is AH(Tl) (Fig. 4),but there is no reason why
“steady-state” temperatures T,, T,, etc., cannot be used to define the course of the A H ( T )
versus T curve, which can then be extrapolated to Tm. Pre-melting effects (Fig. 4) can be
monitored in this way; these are not, of course, peculiar to DSC and are mentioned here only in
a cautionary way to indicate that even with well defined enthalpies there are still uncertainties
as to the true value of AH(Tm). Returning to Fig. 1, two further general points can be made.
Although extrapolation from high to low temperatures is a general experimental possibility the
converse is not-but a very common procedure with polymers (where pre-melting can start
many tens of degrees below the final Tm) is to extrapolate some “guesstimated” low-tempera-
ture base line t o T,; not surprisingly, there are wide variations in reported values of AH(Tm)
for these materials. At Tm a sudden power surge is demanded and instrumentation is such
that this is spread over a finite time so that the peak temperature, for example, may appear
several degrees above the equilibrium T m value. This is purely an artefact and there is gener-
ally no question of superheating. Various complex treatments2 seek to decompose the
resultant peak area into AH(Tm) and heat capacity contributions, but these are not needed
with the simple treatment given here.
The area conversion factor, K , must be known from separate calibration experiments that
define an area corresponding to a known enthalpy change-the most convenient way is to run
the empty pan followed by the pan plus calibrant over a given temperature range just as
described by Laye.l
Heat Capacity and Enthalpy Curves
Although the simple “internal base line” procedure outlined above has widespread applica-
tions (solid - solid phase changes can be similarly treated), the only truly general method for
determining enthalpy changes is that which transforms DSC curves to C, versus T curves, as
described by Laye,l followed by integration. Subsequent remarks which apply to C, versus
T curves will often have wider significance than in a purely DSC context, but over-all they must
be considered specific to DSC because no other type of calorimetry approaches the unusually
wide range of heating and cooling rates of this technique.
One of the most striking consequences of transformation to true C, curves is the frequency
with which linear C, versus T regions, covering wide ranges of temperature, are found. When
this is the case, heats of fusion at T , can be derived directly from
AH(Tm>= [H1(T2) - He(T1)I - [H1(T2) - H l ( T m ) I - [Hs(Tm) - Hs(TJ1
where the first term in square brackets is the over-all experimental quantity and the remaining
two are estimated using equations of the form Cp = a + bT; the AH(T,) thus obtained has
been corrected for pre-melting effects (Fig. 4). Attention to the behaviour of the reference
state (or states, solid and liquid in the example above) can transform many enthalpy changes
into meaningful isothermal heats of transition or reaction. A problem can often be aggravated
by the fact that the DSC output is the derivative of enthalpy rather than H itself; fortunately
modern computational techniques are such that transformation to H versus T curves can
readily be accomplished using some convenient, but reproducible, state as an arbitrary
enthalpy zero.
An Illustration
A good example of the value of enthalpy changes in clarifying a complex structural problem
is given by Figs. 2-4, which are for 4-cyano-4’-n-hexoxybiphenyl, a liquid crystal-forming
View Article Online

230 FUNDAMENTALS AND APPLICATIONS O F DTA AND DSC Anal. Proc.

I
w
D I

..-
I
30

20
1.
Y
Published on 01 January 1980. Downloaded by Brown University on 30/10/2014 06:50:54.

7
\

On 10

280 320 360

Timehemperature Temperatu re/K

Fig. 1. Using a supercooled liquid Fig. 2. The heat capacity of M18 (see text).
t o define an enthalpy of fusion. arrows indicate heating or cooling (i
10 K min-I).

material (BDH M18). The solid curve in Fig. 2 refers to solution-grown crystals and shows the
crystal (s) + nematic liquid (n) --f isotropic liquid (1) transitions a t about 335 and 350 K,
respectively. On cooling (broken curve), the 1 -+n transition occurs with virtually no super-
cooling (this is general and these transitions are not amenable to the treatment exemplified
by Fig. l),but there is then much supercooling, to give a good definition of Cp(n),before the
main crystallisation event a t about 310 K. On re-heating (dotted line) there is a fairly large
exothermic effect between 300 and 315 K, but otherwise the second melting curve is similar to
the original Obviously something happens when the original melt-crystallised product is re-
heated, but exactly what is not obvious from the DSC curves (Fig. 2) themselves. If, however,
enthalpy changes are plotted with respect to Hl(370 K) = 0, all is clarified (Fig. 3): melt
crystallisation gives a structure with a heat of fusion of 14 J g-l (Figs. 3 and 4)less than that
of the original, the latter is reformed in the phase change occurring over the range 300-315 K,
as shown by the closure of the enthalpy loop in Fig. 3.
Figs. 2 4 illustrate the many advantages of computer treatment when using the full instru-
mental potential. Heating and cooling curves (i.e., Cp data) superimpose in regions away from
thermal “events” (in itself a good test of quantitative behaviour). Crystallisation, when it
eventually occurs, is often very rapid so that range switching is essential to keep the signal on
scale-this is easily dealt with in a computer program to give a smooth final C, versus T curve.

Ixtrapolated from T >360 K

b
7
1
Y
-100
-b -o-l -----yAH(s-+ I) Pre-melting

z
5
/
I AH
I
7 50-
I 3
X Q AH(n + I )
I I
F7 I
1

2 -200 I
L- _-- -I
0
280 320 360
Temperature/K
280 320 360 Fig. 4. Enthalpies of fusion and transition M l S .
Tem perature/K Details as for Fig. 2.

Fig. 3. Enthalpy changes for M18 when


H(370 K) =0. Details as for Fig. 2.
View Article Online

June, 1980 FUNDAMENTALS AND APPLICATIONS OF DTA AND DSC 231


Several references have been made to the temperature dependence of AH. That this is a
real and important effect is clear from Fig. 4. Suppose that melt crystallisation gave the same
product as the original ( i e . , solid and broken curves superimpose below about 300 K), then
+
from Fig. 4 measured values of A H (nematic solid) on crystallisation and “re-melting”
should be 98 and 108 J g-l, respectively, and even the crudest of base-line constructions could
not fail to reveal this direct manifestation of the relationship AH(T) = f(T). Nevertheless,
the converse is often implied. Consider the melting of two forms (I and 11)of a given material,
one of which shows a low-temperature phase change AHt; the equality, or otherwise, of
Published on 01 January 1980. Downloaded by Brown University on 30/10/2014 06:50:54.

+
AHt(1) AHm(I) and AHm(II) is often used to discuss the relative energies of the phases,
whereas in reality only over-all enthalpy changes (Le., including heat capacity terms) are
significant for such a discussion. “Inclusion of C, terms” is another way of saying that sets of
transitions can be compared, but only after reduction to the same temperature. Thus, in
Fig. 4 it is valid to observe that a t 320 K AH(s + 1) = 109.6 J g-l can be decomposed into
AH(s+n) + +
AH(n-4) = 104.8 4.8 J 8-l; in contrast, addition of AH(s+n, Tsn) AH +
( n 4 , Tnl)gives a thermodynamically meaningless quantity.
Although not yet explicitly stated, a key quantity in the transformation of the directly
accessible enthalpy change H,(T,) - H,(T,), where x and y refer to two different phases, into
an isothermal quantity AH(x-+y, T1) is JTeC,(y)dT. In Fig. 4,for example, y = 1, the iso-
T*
tropic liquid, and it is clear that some accuracy is needed because C,(1) is not available below
360 K and a lengthy extrapolation is required. Fortunately, the equation Cp(l) = a + bT
holds to above 400 K and the extrapolation can be made with some confidence. On the
other hand, if AH(s+n) is sought, the portion of the Cp(n) curve (Fig. 2) available in direct
heating experiments is not adequate for other than an extrapolation of, perhaps, 10 K [a
larger scale graph shows much “pre-melting,” as evidenced by gross curvature in the C,(n)
versus T curve as Tnl is approached]. Fortunately, as already mentioned above, the cooling
curve gives an excellent definition of Cp(n) well below the pre-melting region-again emphasis-
ing the benefits of a full use of instrumental capabilities.
It is hoped that this brief discussion has given some idea of the potential uses of DSC-
derived enthalpy changes. Present-day instrumentation is such that these changes can be
routinely determined with an accuracy of &l%. It is unfortunate if this is later lost by lack
of care in the subsequent resolution of an over-all enthalpy change into isothermal and heat
capacity terms.
References
1. 1980, 17, 226.
Laye, P. G., Anal. PYOC.,
2. Heuvel, H. M., and Lind, K. C. J. B., Anal. Chem., 1970, 42, 1044.

Você também pode gostar