8 ANALYTICAL PROCEEDINGS, JANUARY 1988, VOL 25 New Developments and Applications of Thermal Analysis The following are summaries of six of the papers presented at the Annual General Meeting of the Thermal Methods Group held on November 13-14th, 1986, at the Bonnington Hotel, Southampton Row, London WCI. Some Applications of Thermal Analysis to Cement Hydrates S. M. Bushnell-Watson, H. D. Winbow and J. H. Sharp Department of Ceramics, Glasses and Polymers, University of Sheffield, Sheffield S I0 2TN Hydrates are of general interest, but are technologically important as the products of the hydration of cements. During our investigations into the chemistry of hydration of several special cements. we have used thermal methods, especially DTA, extensively. These techniques provide a sensitive method for the detection of hydrates and are particularly valuable when phases amorphous to X-rays are formed.Problems of identification arise, however, because the tem- peratures at which dehydration reactions are observed vary markedly with the amount of the hydrate present and procedural variables associated with the apparatus used. Overlapping peaks are frequently encountered and are not always clearly resolved. It is essential, therefore, to use thermal methods in conjunction with other techniques, of which X-ray powder diffraction (XRD) and scanning electron microscopy are perhaps the most important. DATA and TG curves have frequently been obtained as an integral part of the study of the formation of C-S-H gel and Ca(OH)2 in Portland cement pastes and the hydration of ciment fondu [high-alumina cement (HAC) with an A1203 content of approximately 40%].We have extended the application of these techniques to investigate the hydration of refractory HAC with 5&80% A1203, and magnesia - phos- phate cements which are used as repair materials because of their rapid hardening properties. Usually the combination of XRD and DTA leads to a reliable identification of the hydrates formed and at least a semi-quantitative estimation of their amounts. We have, however, made several intriguing observa- tions that may be of interest to others besides cement chemists. Magnesia - phosphate cementsl.2 are formed by the reactions between magnesia and ammonium dihydrogen phosphate (ADP) in the presence of sodium tripolyphosphate: Schertelite MgO + 2NHjH,PO,+ 3HzO=(NH4)zMg(HP01)2.4HIO MgO + (NHj):Mg(HPOj):.4H20 + 7H20 = 2NTljMgPOj.6Hz0 Struvite Schertelite and struvite are both readily identified as reaction products by means of XRD. When a sample of pure struvite is heated, it loses five molecules of water at around 100 "C and the last molecule at about 23OoC,3..J as shown in Fig. 1A. When struvite is diluted with an excess of quartz sand (as i n a mortar or concrete), usually only five molecules of water are lost at around 100 "C (Fig. lB), but sometimes all six molecules are apparently lost simultaneously (Fig. 1C). The bar charts on the figure represent the XRD peak intensities of struvite (stj, schertelite (sc) and ADP (a). Reaction products containing both struvite and schertelite (and sometimes unreacted ADP as well) give rise to a double endotherm (Fig.lD), but often the DTA curve is more complex, such as that shown in Fig. lE, which is clearly an exotherm superimposed on an endotherm or endotherms. We believe that the presence of the exotherm is evidence for the following reactions: MgO + (NH4)2Mg(HP04)2.4HIO = NH4MgP04.HIO + 3Hz0 Dittmarite MgO + NH4H2POA = NH4MgPOj.HIO A \ I / - - - 1501 D \ I Fig. 1. DTA curves and XRD bar charts of (A) pure struvite (ATat half-sensitivity) and (B-E) various magnesia - phosphate cement mortars hydrated for 3 h at 25 "CANALYTICAL PROCEEDINGS, JANUARY 1988. VOL 25 9 The formation of dittmarite by these routes, in addition to that from struvite. explains the enhanced intensity of the endother- mic peak at 256 "C.The refractory calcium aluminate cements are similar to ciment fondu in that the major active component is CaA1204 or CA (C = CaO, A = A1203, H = H,O), and the principal hydration products, which vary with temperature and duration of hydration, are CAHlo, C2AH8, C3AHhr AH, gel and crystalline AH3. Problems in resolving the peaks at 100-160 "C (due to AH, gel and CAHIO) and at 270-330°C (due to AH3 and C3AH6) are well known to thermal analysts.5 A less intense peak is usually present in the DTA curve of refractory calcium aluminate cements at around 190°C and this some- times appears as a doublet (Fig. 2). Although this peak is 33 1 Fig. 2. DTA curves and XRD bar charts of hydration products of Secar 51 (25 g ) + CaCO: ( 5 g) + H,O (15 ml) at 10°C after (A) 7 h.(B) 1 d , (C) 2 d and (D) 5 d frequently attributed to the dehydration of C;AHH, we have alreaiy suggested that it can also indicate the presence of C4ACH11,6 where C = C 0 2 . In refractory cements with intermediate Al2O3 contents (5&60%), gehlenite, C2AS, is present as a minor phase which hydratesrito form stratlingite ~ C2ASH8. When HAC is hydrated in the presence of LiCl (which acts as a spectacular accelerator of the set) we have observed the presence of C4AHI3 in the hydrated paste by XRD. Both of these hydrates give endothermic peaks on heating, which can occur at around 190°C when they are present in only small amounts. The presence of double peaks or shoulders on a main peak (as shown in Fig. 2) in this temperature region is, therefore, to be expected.It is essential to use XRD or some other technique to establish which phase or mixture of phases is causing the thermal effect and it seems certain that errors of attribution have been made in the published literature. On the other hand, the bar charts shown in Figs. 2 and 3 indicate that the XRD peaks of these fresh cement pastes are relatively weak. The DTA curves, represent- ing amorphous as well as crystalline hydrates, provide much additional information. The DTA curves shown in Fig. 2 also show an interesting observation made repeatedly in our work and also by others,'.g but generally disregarded. The expected doublet at 270-330 "C attributed to the presence of AH3 and C3AHh is sometimes observed as a triplet. We strongly suspect, from comparison of the relative intensities of our XRD and DTA peaks due to gibbsite, that amorphous AH3 in addition to crystalline gibbsite is formed in fresh HAC pastes.This could account for the relative weakening of the peak at 279-290°C in Fig. 2, as the hydration products crystallise. The DTA curves shown in Fig. 3, however, indicate that sometimes it is the middle peak 300 300 Fig. 3. DTA curves and XRD bar charts of hydration products of ( A ) Secar 51 with w : c = 0.5 at 30°C after 5 . 6 and 7 d. (B) Secar 71 with w : c = 0.5 at 50°C: after 1 . 2 and 3 h and (C) Secar 71 with w : c = 0.5 at 40 "C after 7, 16 and 24 h that disappears with increasing duration of hydration. As it is most unlikely that an amorphous form of AH3 would give a DTA peak at a higher temperature than a crystalline poly- morph, an alternative explanation must be sought.Possible explanations are that the peaks are due either to the presence of two crystalline forms of AH3 and one of C3AH6 or to gibbsite and two forms of C3AH6, one of which would seem to be an amorphous prototype formed during the conversion process. Work is continuing to try to resolve this matter. References 1. El-Jazairi, B., Concrete, 1982, 12. 2. Abdelrazig, B. E . I., Sharp. J. H., Siddy, P. A , . and El-Jazairi, B., Proc. Br. Ceram. Soc., 1984, 35, 141. 3. Paulik, J . , andPaulik, F . , Proc. 4th Int. Conf. ThermalAnalysis, Budupest, 1974, 3 , 789.10 ANALYTICAL PROCEEDINGS, JANUARY 1988, VOL 25 4. 5 . Abdelrazig, B. E. I.. and Sharp. J . H., Thermochim.Acta, in the press. 1985, 93, 613. Wilburn, F. W., Keattch, C . J., Midgley, H. G., and Charsley, E. L.. "Recommendations for the Testing of High Alumina Cement Concrete Samples by Thermoanalytical Techniques," Thermal Methods Group. Chemical Society, London, 1975. 6. 7. Bushnell-Watson, S. M. and Sharp. J . H., Thermochim Actu. Murat. M., in "Proceedings of the International Seminar on Calcium Aluminates. Turin. 1982." Polytechnic0 di Torino. Turin, 1982, pp. 59-84. Midgley, H. G., personal communication. 8. Arrhenius-Right or Wrong? D. Dollimore Department of Chemistry, University of Toledo, Toledo, OH, USA The Arrhenius equation is used as a "corner stone" in kinetic analysis in spite of the fact that a sizeable fraction of results collected, especially in solid-state decomposition, show deviant behaviour.Frost and Pearson1 list a series of deviant behaviour patterns, but the most common deviant behaviour is several linear regions in the plot of log (rate constant) against reciprocal temperature (in degrees Kelvin), or even a continu- ous curve which can be regarded as an infinite collection of linear regions. The relationship A' T log K = - + constant where K is a specific reaction rate. T is the temperature in degrees Kelvin and A ' is a constant, was first put forward by Hood.' It was opposed by Harcourt and Essen,3 who claimed the relationship was where rn is positive. the form we now recognise as the Arrhenius equation: K = constant X Arrheniusj used the van't Hoff relationship' as the basis for E RT l n K = - - + l n A or where A is the pre-exponential factor and exp (-EIRT) is proportional to the number of molecules possessing energy E in excess of the average energy characteristic of all molecules in the system at the temperature T. There are difficulties in thoughtfully applying this equation to polymers and certain other systems because of an inability to identify the mole in the definition of the units of E as kilojoules per mole.This arises because in the calculation of E from the Arrhenius plot the slope is divided by the appropriate value of the gas constant R. In making the application of the Arrhenius equation to solid-state reactions one is further hampered by the absence of a concentration term and the consequence of the K = A e-ERT dependence of the kinetic laws on geometric factors controlling the process at a reaction interface.This leads to a nomenclature where one deals with the fraction decomposed ( a ) when the specific reaction rate ( k ) is defined as d a dr - = kf(@ where r is the time andf(a) some function of a.0 For this to be an acceptable rate constant, then as CY + 0, f ( a ) + 1 and daidr = k . However. many solid-state kinetic laws instead show that as a + 0, then f ( a ) + 0 when daidr = 0 and k is not a specific reaction rate constant. This difficulty is removed by normalis- ing at a = 0.5.7 The occurrence in many instances of Arrhenius plots showing two or more linear regions8 or a curve.9 however, demonstrates an absolute need for rising tempera- ture methods of establishing kinetics to be portrayed in the form of an Arrhenius plot of log k against ZIT.This is often ignored in the application of these methods, which are often implicitly based on there being only a single set of Arrhenius parameters. 1. 2. 3. 4. 5. 6. 7. 8. 9. References Frost. A. A.. and Pearson. R. G.. "Kinetics and Mechanism." Wiley. New York. 1953, p. 23. Hood, J . J . . Phil. Max.. 1878, 6, 731: 1885. 20. 323. Harcourt. A. V.. and Essen. W.. Phil. Truris. R. Soc. London, Ser. A . 1895, 186. 187; 1913. 212. 187. Arrhenius, S . . Z . Phys. Chem.. 1889. 4. 226. van't Hoff. J . H.. "Etudes de Dynamique Chemique." Muller. Amsterdam, 1884. Dollimore. D . , Heal. G. R.. and Krupay. R. W., Thermochim. Actci. 1978, 24. 293. Fatemi. N.. Whitehead. R.. Price. D.. and Dollimore. D.. Thermochirn.Acta., 1986. 104, 93. Dollimore. D.. and Rodgers. P. F.. Thermochim. Actu. 1979. 30, 273. Mikhail, R . Sh., Dollimore. D.. Kame], A. M.. and El-Nazer. N. R.. J. Appl. Chem. Riotechriol.. 1973. 23. 419. Use of Thermal Analysis in Coal Energy Studies Alexandra de Koranyi British Gas Corporation, Research and Development Division, Michael Road, London S W6 2AD Owing to the depletion of oil and gas resources that is occurring throughout the world, there is renewed interest in more efficient use of the vast reserves of coal that exist worldwide. There are four main areas of coal utilisation: pyrolysis or carbonisation, combustion, gasification and liquefaction. In order to use coal efficiently, it is important to understand the complex structure and properties of coal.Experimental techniques to study and define these properties quantitatively are still being developed, while thermoanalysis is continuing to contribute to coal research and process analysis. Coal Characterisation Coal contains mostly carbon, together with ash or mineral matter, in addition to tars, hydrogen and volatiles. The most important minerals are the clay minerals and carbonates. Coal ranking depends on a knowledge of its proximate and ultimate analysis. Proximate analysis of coal is a determination of the moisture, ash and volatile matter content of the coal sample,' and is obtained by heating the coal under a set of standard conditions. Proximate analysis of coals routinely usedANALYTICAL PROCEEDINGS, JANUARY 1988, VOL 25 11 to take about 1 d to complete but the use of thermal analysis has reduced this time to 1 h.Ultimate analysis is an absolute measure of the elemental composition of coal.' Evaluation of mineral matter in coal is important and is derived indirectly from the ash content remaining after high-temperature combustion. Low-temperature ashing of coals removes the organic carbon from coal preferentially by reacting the coal in an oxygen plasma at temperatures up to 250 "C without major alteration of the inorganic content. The minerals are then determined directly by thermal analysis. Sulphur found in coal ash has many detrimental effects on fuel properties.' A common sulphide mineral in coal is pyrite (FeS2), which may be determined using thermomagnet- ometrya2 In this technique, a magnet is sited within a thermobalance which is used to measure the magnetic attrac- tion of iron in pyrite.When combined with thermogravimetry (TG), thermomagnetometry is a useful method of analysis. However, when siderite (FeC03) is present in coal together with pyrite, the use of evolved gas analysis3 will determine umambiguously the pyrite content by measuring directly the SO2 evolved from it. The clay minerals and carbonates found in coal exhibit strong endothermic decompositions4 and can be analysed by differential scanning calorimetry (DSC), which can also potentially measure char reactivity. Differential thermal analysis (DTA) and differential ther- mogravimetry (DTG) are used to establish the behaviour of mineral matter on heating in different atmospheres.Multiple atmosphere DTA and TG will separate many thermal analysis peaks, affording better identification. Pyrolysis (Carbonisation) Pyrolysis of raw coals in an inert atmosphere leads to a loss of the so-called volatile matter and leaves behind a solid, highly porous char. TG and DTG5 have been used in tests developed recently to assess the temperature a coal has been subjected to in a gasifier, and also to "fingerprint" individual coal devolatil- isation characteristics. Kinetics of carbonisation can also be studied. Emanation thermal analysis (ETA) is a thermoanaly- tical method based on the rate of release of z2oRn by heating samples previously labelled with radium parent isotopes.6 When used in conjunction with DTG and porosity measure- ments, detailed information on coal structural changes occur- ring7 during pyrolysis can be obtained.Low-temperature carbonisation (below 700 "C) has been used to make smokeless solid fuel for domestic consumption. High-temperature car- bonisation (above 900'C) gives a less reactive char of lower porosity. This process has been used for centuries for the production of metallurgical coke. When a coal is heated rapidly, a higher proportion of volatiles can be released. This process of rapid heating of coal in an inert atmosphere is known as flash pyrolysis or, when carried out in a hydrogen atmosphere, flash hydropyrolysis. These methods have been proposed for conversion of coal to gaseous and liquid fuels. Numerous products can also be derived from coals. In fact, next to combustion, carbonisation is the greatest use made of coal.Combustion Of the four areas of coal utilisation, the largest is combustion. Combustion may be defined as high-temperature oxidation of carbon and hydrocarbons to carbon dioxide and water with an accompanying release of heat.8 Combustion is used mainly for electricity generation. The most recent method of combustion is fluidised bed combustion (FBC). FBC produces a higher heat release rate and heat transfer rate within the bed than other combustion systems' and, therefore, lower temperatures can be used, resulting in the need for less coal to sustain combustion due to efficient burning. Boiler costs and corrosion are also reduced and a wider variety of fuels can, therefore, be burned. In combustion systems for power generation, the emission of oxides of sulphur is an environmental problem which can be solved by the addition of limestone in FBC.Thermal methods of analysis can be used to monitor the efficiency of these coal combustion processes.9 Potentially a more attractive process for power generation is FBC at high pressure, but this process is still in the experimental stage. Ash layers can be monitored by DTG and TG. DTG can also measure the dissociation features of coal while databanks can be developed for assessment of coal combustion characteristics. Gasification Over the past 200 years, gasworks have been commonly seen throughout Western Europe producing gas from carbonisa- tion. Gasification combines the thermal decomposition of coal with reaction of the resulting char with reactive atmospheres yielding fuel gas.Coal gasification results in complete conver- sion of all the carbon in coal to gas products. Gasification can be performed in various ways and the resulting combustible gases can be of low, medium or high calorific value (CV). Medium CV gas (ca. 300BTU per standard cubic foot) can be used for power generation and other industrial uses, while high CV gas (1000 BTU per standard cubic foot) is interchangeable with natural gas and is used as the basis of substitute natural gas (SNG). Research into the fundamental aspects of coal gasification, such as the kinetics of gasification, has improved efficiency. However, further improvements in performance can be obtained by the use of catalysts. Again, research in this area of coal gasification is taking place with the aid of thermogravi- metric techniques. Liquefaction Coal liquefaction processes involve the addition of a solvent prior to heating the coal to the desired temperature, generally 40G50O"C.l Solvent extraction is a mild form of chemical conversion.The yield of extract is enhanced by temperature and the presence of hydrogen. The underlying problem in liquefaction is to increase the hydrogen to carbon ratio in the products cheaply and various processes are being developed for this purpose. Commercial liquid fuels can be produced by such rapid pyrolysis processes as the US Cogas process, or indirectly by gasification to a synthesis gas, followed by appropriate catalytic processes, as in South Africa's SASOL plant. Mineral matter present in coal can act as a catalyst in any of the utilisation processes. In the liquefaction process, the rate of liquefaction increases directly with the concentration of mineral matter,3 depending on the composition.Using mul- tiple atmosphere DTA and TG, coals can be assessed for chemical properties and reactivity to determine their stability as a source of liquid fuels. Fundamental Research Even though gas - carbon reactions have been a part of our industrial economy for decades, a basic understanding of the reaction mechanisms and kinetics involved has lagged far behind their practical use, However, much research is now going on in these areas, increasing our understanding of gas - solid interactions. Models to describe mass transport and the heterogeneous chemistry which occurs during coal pyrolysis and gasification are being developed. One of the most complete models on microscopic pore diffusion is the pore tree structure model developed recently by Simons.* ( J Such models reflect in detail the essence of over-all char reactions. Mechanisms have also been developed to describe gasification or combustion interactions, such as the shrinking core model11 and the two-dimensional coal gasification or DICOG model. l 2 These put theories on the solid basis of a mathematical model. Thermal methods of analysis can be useful in elucidating these12 ANALYTICAL PROCEEDINGS, JANUARY 1988, VOL 25 interactions by conferring a sound experimental footing to theory and indicating directions for further expansion and validation of the models.The concepts of active surface area and active site concentra- tions are of equal importance in understanding the kinetics of coal gasification and combustion. Thermal analysis can deter- mine the fine pore structure of a char or the total surface area accessible to reactants. Intrinsic reactivity of a coal char is related to the number of active sites which can be measured by various methods, such as in the recent method developed by Causton and McEnaney13 involving oxygen chemisorption followed by temperature-programmed desorption. More work needs to be done in the area of reactivity of different coal chars with oxygen, steam and hydrogen, and also in high-pressure gasification where reactivity data are sparse. The thermobalance is especially useful in obtaining reactivity data as gasification reaction rates can be measured directly under differential conditions in a constant, well defined environment during reaction. Conclusions The potential use of thermoanalytical techniques in the field of coal characterisation and reaction is enormous.We have seen in this brief review only a small part of the usefulness of these methods in the area of coal energy studies. However, with further work and increased understanding of coal utilisation, the vast resources of this fossil fuel can be used with greater efficiency for the benefit of mankind. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. References Speight, J. G., “The Chemistry and Technology of Coal,” Marcel Dekker, New York, 1983. Aylmer, D. M., and Rowe, M.W., Thermochim. Acta, 1984, 78, 81. Warne, S. St. J., Bloodworth, A. J . , and Morgan, W. J . , Thermochim. Acta, 1985, 93, 741. Tarrer, A. A . , Quin, J. A., Pitts, W. S . , Henchy, J . P., Prather, J . W., and Styles, G. A., in Ellington, R. T., Editor, “Liquid Fuels from Coal,” Academic Press, New York, 1977, p. 45. Cumming, J . W., and McLaughlin, J . , Thermochim Acta, 1982, 57, 3. Balek, V., Thermochim. A m , 1977, 22, 1. de Koranyi, A., and Balek, V., Thermochim. Actu, 1985, 93, 757. Pitt, C. J . , and Milliard, G. R., Editors, “Coal and Modern Coal Processing-An Introduction,” Academic Press, London, 1979. Mikhail, S. A., Thermochim. Acta, 1985, 93, Suppl., 116. Simons, G. A . , “19th Symposium (International) on Combus- tion,” Combustion Institute, 1982, p.1067. Laurendeau, N. M., Prog. Energy Combust. Sci., 1978, 4,221. Smoot, L. D., Prog. Energy Combust. Sci., 1984, 10, 350. Causton, P., and McEnaney, B., Fuel, 1985, 64, 1447. ~~~ A New Approach to Thermal Analysis: Simultaneous Microcalorimetry and Thermogravimetry Measurement; Simultaneous TG - DTA up to 2700K Francis Pithon SETARAM, 7 rue de I‘Oratoire, 69300 Caluire, France Introduction A new range of thermal analysers are now available. They have been generated by long experience and know-how in this field. The main feature of these instruments is a combination of a scanning microcalorimeter with a symmetrical balance and symmetrical simultaneous TG - DTA up to 2700 K. The automation of the sample’s environment (the tempera- Mass-variation monitoring by symmetrical microbalance 1 1 Energy monitoring by heat flux transducer Fig.1. The calorimeter microbalance Sa m p le ” c r uci b I e Reference ” c ru ci bl e heat flux transducers Fig. 2. Apparatus for simultaneous microcalorimetry and thermo- gravimetry ture and the gas) has led to better reproducibility and to the automatic control of complex experimental conditions. Simultaneous Scanning Microcalorimetry and Thermogravimetry: the TG - DSC 111 Calvet microcalorimeters provide a cylindrical space surroun- ded by heat flux transducers, which monitor all the heat transfer, thus giving accurate direct access to the heats of reaction, transformation and the heat capacity. A new design has been developed, in which the sample, instead of standing inside the space, is suspended below a symmetrical micro-ANALYTICAL PROCEEDINGS, JANUARY 1988, VOL 25 - - 13 -400 2 -200 2 - O 3 2 . 0) m r 0 balance (Fig.1). The high performances of both microcal- orimetry and symmetrical microgravimetry are now available simultaneously in a single simple investigation (Fig. 2). This is of great interest, as illustrated by the applications in the field of catalysis, adsorption, oxidation, gas solid reaction, decomposi- tion, pyrolysis dehydration, etc. The importance of the information given by the recording of the mass variation during a DSC scan is illustrated by the oxidation of coal (Fig. 3). The heat flow versus temperature + H20 - - - - - - - - 30 t 0.8 0.7 0.6 TS) 0.5 2 0.4 4 0.3 0.2 5 0.1 7 0 10 c 4 - 5 exo / / 1 2 0 0 1 I I 1 I I I i00 200 300 400 500 600 Temperatu rePC Fig.3. Conditions: 0.932 mg; 10 K min-1; oxygen 0.9 1 h-1 Mass change during a DSC scan for oxidation of coal. graph has two exothermic peaks, and the simultaneous TG curve indicates that the first reaction corresponds to a mass increase, which means that there is chemisorption of oxygen on coal up to 700K. It is followed by a combustion with an important mass loss. Further information can be obtained by analysing the evolved gases, as illustrated in Fig. 4, represent- ing a catalyst investigated under a flow of hydrogen. The evaporation of water, followed by the reduction of the nickel oxide, is monitored through the endothermic and exothermic reactions and the mass variation. During the whole process, the water evolved is measured by means of a mass spectrometer.Evaporation NiO + H2-+ Ni 2 -of adsorbed H20 1 1 1 I I I I I I I Tern peraturePC Fig. 4. Analysis of evolved gases for a catalyst under a flow of hydrogen. Conditions: NiO - Si02 catalyst (3 + 1, rnlm); 10°C rnin-l 80 180 280 380 480 580 680 780 Simultaneous Symmetrical Computer-controlled Thermal Analyser TAG 24 TG and DTA are widely used for quality control and research purposes and the latest developments in TG - DTA equipment, and their practical consequences for the users, are considered below. Symmetrical Design Symmetrical microbalances with a horizontal beam are those with the highest sensitivity. They have excellent long-term stability when the sample is suspended from the beam. The horizontal position is controlled by an opto-electronic system, and the mass changes are adjusted by electro-magnetic compensation.The best TG instruments are founded on this principle, but it is possible to achieve even better stability, better sensitivity and better resolution by having a dual furnace (Fig. 5). In this instance the thermal analyser is fully symmetrical, and it can monitor very small mass variations, down to a few micrograms. Symmetrical balance, very low mass detection Secondary vacuum facility ]lmir-- -- Active gas Symmetrical furnaces, buoyancy compensations #-I---- Simultaneous TG - DTA facility for gas analyser (GC and MS) Fig. 5. TAG 24: T G - DTA. simultaneous and symmetrical The main advantages are the compensation of the buoyancy zffect and the modifications and changes in the gas flow.The difference between a monofurnace (single furnace) and a symmetrical furnace (dual furnace) assembly is measured by a test in which the temperature of both furnaces is scanned, followed, after cooling, by scanning of one of them. The test is performed under a flow of 1 1 h-1 of argon and a heating rate of 1OKmin-1 from room temperature to 1300K. There is no active sample in the crucible and, on both sides of the balance, the same crucible holder is hung with its simultaneous DTA transducer. The two corresponding TG curves are plotted on the same diagram (Fig. 6). The monofurnace curve has a non-linear apparent mass variation of 1.5mg, compared with a few micrograms with the symmetrical dual furnaces. The compen- sation is very good and, with a blank correction, a resolution as low as 1 pg is achieved by means of the symmetrical thermal analyser .I I 1 .o F 5 I- 0.5 0 I I I 1 I I 0 200 400 600 800 TemperatureiOC Fig. 6. T G curves obtained using the TAG 24. T G - DTA configuration: empty crucible: argon at 1 1 h-I; 10°C min-' Simultaneous TG - DTA up to 2700K There is a need for information on the behaviour of materials at high temperature, above 2000K, and only a few instruments could go above this value. The two main reasons that limited the temperature of the thermal analysers to 2000 K were that platinum melts and alumina softens above this value. It is therefore not possible to use these basic materials, which are very reliable, especially as far as the platinum - rhodium thermocouples are concerned. A new instrument, the temperature of which can be scanned up to 2700K, has been developed.It uses graphite furnace technology and the DTA transducers are made of tungsten - rhenium alloy. This simultaneous TG - DTA thermal analyser monitors the behaviour of material at very high temperature. Phase transformation, melting, evaporation and decomposi- tion reactions are investigated. Its high performance is14 /- ANALYTICAL PROCEEDINGS, JANUARY 1988, VOL 25 22 illustrated in two fields of application: high temperature metallurgy (Fig. 7), with the melting of a rhodium sample inside a zirconia crucible; and ceramic research (Fig. S), with the melting of alumina inside a molybdenum crucible. ~ ~~ - - - - - - - I I 1 42 38 34 30 26 22 Temperatu rePC Fig.7. Melting of rhodium in a zirconia crucible Automatic Control of the Sample's Environment The TG and DTA signals can be meaningless if the tempera- ture and the surrounding atmosphere of the sample are not controlled. If this control is important, it is also essential to reproduce it, as this is one of the determining factors in the reproducibility and precision of the technique. It is achieved by use of a microprocessor controller, in which the experimental conditions are kept in memory. They are programmed through a keyboard, the most important ones being the temperature and the gas flow. The temperature of the sample in the system to be investigated is programmed following the requirements of the test. It can be either a simple or a very complex programme.The gas environment is controlled by the proces- sor, which drives eight electromagnetic valves. They allow a wide range of automation of the atmosphere inside the space where the experiment is carried out. Different surrounding conditions can be programmed, such as vacuum, purge, very high vacuum, inert gas and reactive gases at different flow- rates. The fact that the programme is kept in memory provides a high reproducibility of the experimental conditions. Fig. 8. Melting of alumina in a molybdenum crucible Conclusion Thermal analysis has made a great leap forward in recent years, although its basic principles were known many years ago. There are different reasons for this revival, one being the fact that manufacturers have integrated modern technological developments and now provide a new generation of instru- ments which are automated and reliable.Thermal Analysis in Circuits Manufacture C. A. Smith Liquid Crystal Displays, E. E. V. Co. Ltd., Chelmsford, Essex Polymeric materials are increasingly being used in the electron- ics industry. Phenolic and epoxy resins are used in laminates for printed circuit boards, acrylate materials are used as dry film solder resists and photoresists, polytetrafluoroethylene and polystyrene are used as substrates for microwave circuitry and polyimide film is used in flexible circuitry. Thermal analysis techniques are extremely useful for charac- terising these materials, for the determination of kinetic data on polymerisation reactions and for locating phase transitions t I Cure exotherm I Q, Thermal 0 - decomposition sl 1 , I I I I 1 I" 50 100 150 200 250 300 Temperatu re/"C Fig.1. DSC curve of a B-stage epoxy glass prepreg 4 l T ,/-\ Standard prepreg I I , I I I 50 100 150 200 250 300 Tern peratu re/"C Fig. 2. DSC curves of standard and no-flow epoxy glass prepregs such as the glass transition in amorphous polymers. The techniques used include differential scanning calorimetry (DSC) and thermomechanical analysis (TMA). Epoxy Glass Prepreg Prepregs are pre-engineered laminating materials used in the manufacture of fibre-reinforced composites. 1 Those used in multi-layer printed circuit board manufacture, where they act both as an insulating layer and a bonding material between the inner layers that carry the circuitry, consist of a glass cloth pre-impregnated with a B-stage epoxy resin.This is a partially cured, vitrified resin requiring only heat to cure fully. B-stagingANALYTICAL PROCEEDINGS, JANUARY 1988, VOL 25 + 15 is carried out to provide a material that is easy to handle and simple to process. The degree of B-stage cure is determined using DSC, from the glass transition temperature or from the enthalpy of reaction of the uncured resin when it is heated. To make a measurement, the epoxy resin is first removed from the glass cloth and a weighed amount (20 mg) is placed in an aluminium pan. This is sealed and placed in the calorimeter alongside an empty reference pan. The sample is heated at 10"Cmin-1 to 300 "C and this gives the curve shown in Fig. 1 ~ which is a plot of heat flow against temperature.The resin first undergoes a glass transition which appears as an endothermic peak at 60°C. This may be used to determine the degree of cure, as the temperature of this transition will increase with the degree of cure. The actual curing reactipn starts at 100°C and is denoted by a small exothermic rise followed by the main exotherm, which reaches a maximum at 160°C . It is the enthalpy, A H . of this reaction that has been used by Smith'J to characterise prepregs for use in multi-layer printed circuit board manufacture. AH can be used as a measure of the degree of cure as the heat of reaction decreases with a decrease in the amount of uncured resin left. AH is inversely proportional to the degree of cure. It is also related to the amount of flow achievable in the laminating press.Prepregs with A H 3 100 Jg-1 are preferred. These have good flow properties and always result in successful lamina- tion. Prepregs with AH < 100 J g-1 do not flow easily when heated and this results in delamination, which appears as air pockets within the printed circuit board. AH is used as an incoming goods quality control test on all batches of prepreg. In the case of one major supplier, AH correlates very well with their own test based on the scaled flow method developed by Bell Laboratories and Western Electric in the USA.4 In this test prepreg is characterised (scaled) by its flow in a small test press, which is related to the thickness of the cured composite. The higher the scaled flow number, the higher the flow and A H .corresponding to a lower degree of B-staging (cure). Other tests for epoxy - glass prepreg used in multi-layer printed circuit manufacture have been reviewed by Schiffers and Carreman.6 t r I I transition decomposition I Another example concerns no-flow epoxy prepreg, used in the manufacture of flexi-rigid multi-layer circuits, to laminate polyimide film and epoxy glass laminate together. In this instance a flow inhibitor has been added to the resin to reduce the amount of flow when it is heated. This also results in the epoxy becoming less brittle. The DSC curve is slightly different (Fig. 2) and the heat of reaction is lower. The experimental procedure for determining AH is the same as that used for standard prepregs except that the sample must be corrected for the glass cloth, which is not easily separated from the resin.The test is also run in a nitrogen atmosphere to reduce drifting of the DSC curve. The temperature and the time required to laminate epoxy glass laminate to polyimide film using no-flow prepreg is extremely critical. If the resin is not fully cured, de-smearing the holes in the printed circuit board using chromic acid after drilling will result in etch back along the bonding layer. Over-curing results in decomposition of the resin and delami- nation of the circuit board. Samples of no-flow prepreg were bonded to polyimide film and cured in a lamination press at temperatures ranging from 110 to 205 "C for 1 h. The degree of cure was then determined from the enthalpy of reaction of uncured resin using DSC.AH was found to decrease with increasing lamination temperature up to 170°C. This was confirmed by measurement of the glass transition temperature which reaches a maximum at 170 "C. Laminating for 1 h at 170 "C cures the no-flow prepreg resin to a level comparable to the degree of cure in the epoxy glass laminate, without decomposing the epoxy resin. Part of the problem is that even when fully cured the glass transition temperature of no-flow prepreg is 20°C lower than that of cured epoxy glass laminate and standard prepreg. This results in disproportionate etching during the de-smearing process. Epoxy Glass Laminate A second group of materials used in printed circuit manufac- ture are the base laminates. Phenolic - paper composites have been used for some time, but modern high-quality circuits for military and aerospace applications use epoxy glass laminates.Methods for characterising laminates have included measure- ment of the decomposition temperature using TMA and TGA and measurement of the glass transition temperature using TMA and DSC.7 0.15 t t c 0 v) C m X .- a 5 0, a J I I I I 50 100 150 200 Temperatu re/"C Fig. 4. TMA curve of an epoxy glass laminate To determine the glass transition and thermal decomposition temperatures by DSC, a small section of laminate is placed in an aluminium pan in the calorimeter alongside an empty reference pan and heated at 10 "C min-1 to 300 "C. This gives the DSC curve shown in Fig. 3. The glass transition temperat- ure for a fully cured epoxy glass laminate is 130 "C, and appears as a discontinuity in the heat capacity.The decomposition temperature appears as an exothermic rise at 2200°C. Both values are dependent on the degree of cure. Exothermic peaks at 160 "C indicate residual chemical reaction owing to incom- plete cross-linking of the epoxy resin. Glass transitions, blistering, delamination and decomposi- tion can all be determined by TMA. In TMA the expansion of the resin is measured under an applied load as the temperature is raised with time. The result is a plot of expansion against temperature. Samples are annealed beforehand to relieve stress present in the polymer, using an accelerated heat-up rate, to a temperature below the decomposition temperature. The actual test is made at a heating rate of 10"Cmin-1 to 300 "C.A typical TMA plot for a glass epoxy laminate is shown in Fig. 4. At the glass transition temperature the rate of expansion changes, and this is shown as a step in a plot of the16 ANALYTICAL PROCEEDINGS, JANUARY 1988, VOL 25 I 1 I 1 I 1 0 100 200 300 400 Tern Deratu re/"C Fig. 5. mask Change in enthalpy with degree of curve for dry film solder first derivative of expansion against temperature. Blistering, delamination and decomposition are all shown as rapid increases in expansion. Solder Mask Two types of solder mask are in common use in the printed circuit industry. Dry film solder mask is a photopolymeric coating, often based on an acrylate polymer, which is applied as a film to the printed circuit board after manufacture to prevent solder bridging between conductors when components are wave soldered into place.The polymer is photoimaged and cured by photochemical free radical reactions that are initiated by exposure to ultraviolet light. However, heating in a conveyorised oven and further exposure to ultraviolet light is often necessary to cure the polymer fully. Infrared heating may be used. 1 1 I I I I I I I 50 100 150 200 250 300 Tern peraturei'c Fig. 6. DSC curve of a screen printable liquid epoxy solder mask Liquid epoxy solder masks are also used and these are applied to the printed circuit board by screen printing. The pattern is produced using a stencil. They have the advantage that they are much cheaper and only require heating at 120 "C for 30 min-1 h in an oven to cure fully.They do, however, lack the definition that can be achieved using dry film solder masks. Recent advances in polymer chemistry have led to the development of screen printable solder masks that can be photoimaged and cured using ultraviolet radiation. This gives the definition of dry film solder mask, without the cost. DSC curves of dry film solder mask are shown in Fig. 5 , in which the enthalpy of the major peak is an indication of the degree of cure. The same principles used to determine the degree of cure for epoxy resin prepregs can be used here. Any trace of an exotherm on the DSC curve is evidence that the processing conditions are not curing the solder mask correctly. Fig. 6 shows a DSC curve of a screen printable liquid epoxy solder mask. I I I 1 I 100 200 300 400 Tern peratu re/"C Fig.7. DSC curves of dry film acrylate solder masks DSC curves may also be used to differentiate between similar solder masks. Fig. 7 shows two dry film solder masks, one of which has an endothermic peak at 150 "C, indicating an additive not present in the other material. Both solder masks have cure exotherms near 240"C, showing them to be chemically similar. References 1 . 2. 3. 4. 5. 6. 7. Molyneux, M., Composites, 1983, 14, No. 2, 87. Smith, C. A . , Circuit World, 1985, 12, No. 1, 29. Smith, C. A., GECJ. Res., 1985, 3, 162. Bloechle, D. P., Circuit World, 1982, 9, No. 1, 8. Schiffer, W., Circuit World, 1986, 12, No. 3, 4. Carreman, J., Circuit World, 1985, 11, No. 4, 32. Smith, C. A . , Circuit World, 1986, 12, No. 2, 62. Quantitative Differential Scanning Calorimetry-from Heat Capacities to Free Energies M.J. Richardson National Physical Laboratory, Teddington, Middlesex TWI I OL W When the Thermal Methods Group was founded in 1965, thermodynamic standards for the subsequent application of calorimetry was a specialised technique practised in only a few differential scanning calorimetry (DSC) to more conventional laboratories throughout the world. Because of sample require- materials, potentially transforming the calorimetric scene, ments, operational complexities and the time scale of days, or "Potentially" must be emphasised because the majority of even weeks, measurements tended to be made on very well reported applications of DSC use it only as a fairly crude defined, highly purified materials rather than those which were characterisation technique that does not utilise the full met in day to day usage.The former should have provided the potential of the several instruments that are now availableANALYTICAL PROCEEDINGS, JANUARY 1988, VOL 25 c 30 I Y cs, >. 0 r I 2 t. .- g 2 0 - m V m a c I 10 17 - - commercially. This was understandable before microcomput- ers became widespread: successful quantitative operation requires much simple, but tedious, mathematics and computa- tional aid is essential, but with this, any routine measurement is readily transformed into one of thermodynamic significance. This paper will briefly review the steps needed to produce free energy curves. Calibration: Temperature Quantitative work is possible only if both ordinate and abscissa are correctly calibrated.The latter is a time or, more usefully for the scanning mode, a temperature which may be directly measured or derived from the known heating or cooling rate. Whatever procedure is used, the final value refers to a sensor that is not in direct contact with the sample so that there is thermal lag between the two. The conventional temperature calibration seeks to overcome this by calibration with materials of known melting (or transition) temperature at the relevant heating rate (commonly 10 or 20 K min-1). This gives a unique calibration graph that clearly cannot be valid for all materials or even for different masses and/or geometries of the same material. This difficulty can be overcome by using the “enthalpy lag” (when the final programmed temperature is reached) to generate a thermal lag for individual runs.This procedure has the great advantage of being valid for cooling as well as heating-because of supercooling, conventional calib- rants do not freeze at well defined temperatures. Calibration: Specific Heat The ordinate is a differential power or temperature in power-compensation or heat-flux DSC, respectively. In both instances the magnitude is proportional to the difference in “heat capacity” between the sample and reference cells. “Heat capacity” includes all processes demanding or generating energy so that, in addition to the heat capacity itself, there may be heats of fusion or transition, recrystallisation and annealing 40 t T, 373.0 340 360 380 Tern pe ra t u re!K Fig.1. biphenyl (2.OCB). Heating rate, 10 K min--’ Heat capacity of solution-grown crystals of 4.4’-ethoxycyano- phenomena and chemical reactions. These processes super- impose peaks or troughs on a normally monotonous heat capacity curve and special treatment may be required for local disruptions to pseudo steady-state conditions. Fortunately, a common calibration is generally adequate, whatever the sample behaviour. Subtraction of data for the empty pan from that of the pan + calibrant (mass m,, specific heat cpc) gives the calibrant signal S, and KS, = rnccpc where K is the ordinate to heat capacity conversion factor. The most widely used calib- rant is a-alumina, although benzoic acid is useful at low temperatures. Once K has been established the procedure is reversed and cps (subscript s = sample) becomes the unknown.K is not normally recorded as such because cps = (m,c,,S,)/ (m,S,) - KS,/m, and the whole calculation is normally carried out in one operation. However, the behaviour of K is a useful indication of instrumental performance-how it is affected by instrumental settings and temperature (both ambient and programmed) and, for a given set of conditions, its day to day reproducibility. A variant of the above procedure is to measure an urea that corresponds to a known enthalpy change. Heats of fusion are popular candidates for this procedure, but an area due to heat capacity effects alone is equally valid. Whatever quantity is chosen, great care must be taken to ensure that the area measured corresponds to a thermodynamically meaningful quantity.Most “base lines” that are described in the literature do not meet this condition. Specific Heat and Enthalpy Changes A modern DSC gives heat capacities (cp) (Fig. 1) that are accurate to kl% over most of the temperature range of the 0 r I 0 -100 2 - 0 m 0 - 5 -200 P - 300 I I 1 I 320 340 360 380 Tern peratu re/ K Fig. 2. Enthalpy changes calculated from Fig. 1 using H , (390 K) as the reference state; the monotropic liquid crystal phase is also shown. _ _ -, Idealised curve; - - - -, liquid extrapolated assuming cp, = a + bT instrument. Integration leads, in turn, to enthalpy - tempera- ture curves (Fig. 2) which, whenever possible, should always be referred to a reproducible and well defined state-in Fig. 2 this is the isotropic liquid at 390 K.Heats of fusion [AH( r>] follow after extrapolation of data for the liquid; for molten 2.OCB this is mathematically simple because a linear cp - T relation- ship holds. A similar form of equation is valid for the solid, and the broken lines in Fig. 2 show how an idealised H - T curve I I A T X X k T- Fig. 3. Detail for the calculation of entropy changes (see text)18 ANALYTICAL PROCEEDINGS, JANUARY 1988, VOL 25 would appear for 2.OCB in the absence of pre-melting and instrumental rate effects. In a correctly calibrated DSC the only effect of changes in the heating rate should be an apparent displacement of the enthalpy step, the magnitude at T, remaining unchanged. “Apparent” displacement should be emphasised because the effect is not due to superheating, as might be inferred from Fig.2, but to the finite time needed to transfer heat of fusion to the sample. Although it is not difficult to obtain AH( T,) from Fig. 2, entropy changes need some care in their derivation and this is considered below. Entropy Changes The formal calculation of entropy (S) changes from heat capacity data is by summation of terms in c,AT/T, (Fig. 3). This procedure is only valid in regions of thermodynamic reversibility where results should be independent of both the sign and magnitude of the rate of change of temperature (compliance with this requirement is a good test of the over-all performance of a DSC). When the reversibility criterion is not met, as in regions of melting or, especially, crystallisation (supercooling often delays this by tens of degrees), the apparent entropy change may decrease (melting) or increase (crystallisation) with increase in rate of change of temperature.The true, reversible entropy change must be somewhere in between and a good approximation may be obtained by extrapolating heating data to zero rate. However, this proce- dure loses the convenience of rapid experimentation and a satisfactory compromise is to calculate the entropy change along the idealised (broken) curves of Fig. 2 using standard thermodynamic procedures. Additional, low-temperature phase changes can be similarly treated. It is very useful to list both this calculated (reversible) entropy change and the observed (irreversible) value so that the correction to the latter can be monitored as a function of sample size, heating rate, etc. Pre-melting effects are neglected in this calculation but they are usually very small compared with the over-all value. Temperature + Fig. 4. the definition of AG( T ) Schematic free energy curves for the 2.OCB system showing Entropies of fusion [AS( 7‘)J are obtained by extrapolation of data for the molten material just as for AH(T) (Fig. 2). The reversibility correction can increase AS by up to 1% for a 10 K min-1 heating rate but the correction for supercooling can be an order of magnitude larger. Free Energy Changes Previous sections have shown how DSC measurements are used to derive AH(7‘) and AS(7‘). The corresponding Gibbs free energy follows from AG( 7’) = GI( 7‘) - G,( 7‘) = AH( 7‘) - TAS(T), and Fig. 4 demonstrates how GI functions as the reference state. The full curve in Fig. 5 shows AG( 7‘) for the solution-grown crystals of 2.OCB to which the data of Figs. 1 and 2 refer; here T, = 373.0 K and AH(373.0) = 117.6 J g-1. A different structure, with T , = 375.4K and AH(375.4) = 100.6 J g-1, is obtained by crystallisation from the melt in the DSC. This phase (broken line, Fig. 5 ) becomes metastable below 360.OK, accounting for the very slow transition to the solution crystallised form that is observed at room tempera- ture-many months are required for completion. Cyanobi- phenyls are particularly important for their liquid crystalline behaviour, 2.OCB is monotropic, the nematic - isotropic liquid transition [T, = 363.0K, AH(363.0) = 4.1 J g-I] is only observed in the supercooled liquid because the nematic phase is metastable with respect to either of the solid forms (Fig. 5 ) . 20 0, 7 - 10 I- U 4 - Nematic 0 TemperatureiK Fig. 5 . AG(T) for the several forms of 2.OCB Polymorphism is an extremely widespread phenomenon. Thermodynamics indicates nothing about the kinetics of phase changes but the information that it gives about their absolute stabilities is of great importance when properties are in- fluenced by crystal structure, the solubility of a drug, for example. DSC gives the required information with only minimal, soundly based assumptions.