Thermogravimetry and Derivative Thermogravimetry 21 interface through the solid,reaction being caused by strain of bonds at the interface between the reactant and product solids.In other cases the rate of reaction is thought to be controlled by the rate of transfer of heat to or from the reacting interface.Because of this spread of reaction over time and the fact that temperature is always rising with respect to time. the rea appears tocovera spread of temperature.For this reasona careful definition of"decomposition temperature"has to be made Figure 5 shows a typical mass loss in a decomposition experiment.The obvious definition would seem to be where the mass loss is steepest,which corresponds to the peak temperature T in the DTG plot.However,this is me ely the point where reaction is fastest and does start of not represent the e.g.where bonds in the compound egin to reak.The position of T will depend upon the sample size,packing.and heat flow properties.The point T is the initial temperature or onset temperature, but is not easy to identify and depends on the sensitivity of the balance and the amount of drift or"noise seen.There may be traces ofimpurities. which decomp ose or promote some decomposition ahead of the main reaction.A definition c art of reaction is th extrapolated o temperature Te This requires drawing of tangents to the curve at the horizontal baseline and the steepest part of the curve and marking their intersection.For a reaction that starts very slowly and only speeds up later,T.and T will be very different and a more satisfactory point would be sho vn as temp eratur where the fraction n reactedais equal to.5,i.e. 70.0s ,important in kinetic studies,is when the reaction is half over.that is,when the fraction reacted THERMOGRAVIMETRY ANG MASS TEMPERATURE IN DEGREE CELSIUS Figure5 Definition of the decomposition temperature ona TG cur
Therrnogravimetry and Derivative Thcrmogravimetry 21 interface through the solid, reaction being caused by strain of bonds at the interface between the reactant and product solids. In other cases the rate of reaction is thought to be controlled by the rate of transfer of heat to or from the reacting interface. Because of this spread of reaction over time and the fact that temperature is always rising with respect to time, the reaction appears to cover a spread of temperature. For this reason, a careful definition of “decomposition temperature” has to be made. Figure 5 shows a typical mass loss in a decomposition experiment. The obvious definition would seem to be where the mass loss is steepest, which corresponds to the peak temperature Tp in the DTG plot. However, this is merely the point where reaction is fastest and does not represent the start of reaction, e.y. where bonds in the compound begin to break. The position of Tp will depend upon the sample size, packing, and heat flow properties. The point Ti is the initial temperature or onset temperature, but is not easy to identify and depends on the sensitivity of the balance and the amount of drift or “noise” seen. There may be traces of impurities, which decompose or promote some decomposition ahead of the main reaction. A better definition of start of reaction is the extrapolated onset temperature T,. This requires drawing of tangents to the curve at the horizontal baseline and the steepest part of the curve and marking their intersection. For a reaction that starts very slowly and only speeds up later, T, and Ti will be very different and a more satisfactory point would be shown as temperature where the fraction reacted a is equal to 0.05, i.e. To.os. Another definition of reaction temperature, important in kinetic studies, is when the reaction is half over, that is, when the fraction reacted THERMOGRAVIMETRY INITIAL BASELINE MASS m in mg I TO To 5 TEMPERATURE IN DEGREE CELSIUS Figure 5 Dejinition of the decomposition temperature on a TG curve
22 Chapter 2 a=0.5;this is Tos.To show the complete temperature range for reac Although the sample may be decomposing at a temperature which is characteristic of the compound,the shape of the decomposition curve will be affected by many factors.The particle size may control the rate of diffusion of a reactant or product.Sometimes large sized particles split so violently that pieces jumpou of the sing a spurious mass loss record.The rate of flow of heat may be controlled by the type and size of the crucible and by the particle size and degree of packing and mass of the sample.Large samples tend to have a temperature gradient through them leading to early decomposition of the outer part and d decompo n of the c entre part.A thin film of powder gives the lowest comm nd fin shing temperature,follc ed by a thick fm a fine powder,and then by compressed pellets of the material at the highest temperature.For these reasons,small sample sizes are recommen- ded,although it has been shown that grinding may alter the crystalline structure and the course of the decomposition.It should be remembered that all reac ctions in an enthalpy change and the heat relea sed will heat or coo he samp slightly relative to th programmed d furnace temperature.Many of the solids investigated will be poor conductors of heat,which will exacerbate the effect.This is another good reason for small samples,keeping a thin layer of sample as near as possible to the crucible temperature.The relative sizes of the sample crucible and furnace will also cha e the eheat flow.The material of th cible will have a effect,metals like platinum being good conductors of heat,while alumina is a much poorer conductor. The reaction may be reversible with respect to the product gas,so a varying pressure of product gas around the sample would affect the kinetics of the decomposition and thus the shape of the decomposition curves.The shape of a ontainer may have a slight effect. or dee conta n ay ca se the atmosphere at the sample surface to differ if th gas evolved is not the same as the purge gas.If the decomposition product gas does build up in deep crucibles,or deliberately closed crucibles,this is referred to as a "self-generated"atmosphere.If the intention is to carry out decomposition into pure product gas,deliberately flowing over the sample,to produce anq m effect,then that is a different experi ment to the one with an inert purge gas The effec t of changing the purge gas may be very large.This is illustrated below in the example on oxysalt decomposition,where the product may react with oxygen in air,but an entirely different product is left if nitrogen is used.In the case of the
22 Chapter 2 a = 0.5; this is To.5. To show the complete temperature range for reaction, two more temperature values may be added. These are Tf the final temperature (again difficult to pick out accurately) and To the extrapolated offset temperature. Although the sample may be decomposing at a temperature which is characteristic of the compound, the shape of the decomposition curve will be affected by many factors. The particle size may control the rate of diffusion of a reactant or product. Sometimes large sized particles may split so violently that pieces jump out of the crucible, causing a spurious mass loss record. The rate of flow of heat may be controlled by the type and size of the crucible and by the particle size and degree of packing and mass of the sample. Large samples tend to have a temperature gradient through them leading to early decomposition of the outer part and delayed decomposition of the centre part. A thin film of powder gives the lowest commencing and finishing temperature, followed by a thick film of a fine powder, and then by compressed pellets of the material at the highest temperature. For these reasons, small sample sizes are recommended, although it has been shown that grinding may alter the crystalline structure and the course of the decomposition. It should be remembered that all reactions involve an enthalpy change and the heat released will heat or cool the sample slightly relative to the programmed furnace temperature. Many of the solids investigated will be poor conductors of heat, which will exacerbate the effect. This is another good reason for small samples, keeping a thin layer of sample as near as possible to the crucible temperature. The relative sizes of the sample crucible and furnace will also change the heat flow. The material of the crucible will have an effect, metals like platinum being good conductors of heat, while alumina is a much poorer conductor. The reaction may be reversible with respect to the product gas, so a varying pressure of product gas around the sample would affect the kinetics of the decomposition and thus the shape of the decomposition curves. The shape of a container may have a slight effect. Shallow or deep containers may cause the atmosphere at the sample surface to differ if the gas evolved is not the same as the purge gas. If the decomposition product gas does build up in deep crucibles, or deliberately closed crucibles, this is referred to as a “self-generated’’ atmosphere. If the intention is to carry out decomposition into pure product gas, deliberately flowing over the sample, to produce an equilibrium effect, then that is a different experiment to the one with an inert purge gas. The effect of changing the purge gas may be very large. This is illustrated below in the example on oxysalt decomposition, where the product may react with oxygen in air, but an entirely different product is left if nitrogen is used. In the case of the
Thermogravimetry and Derivative Thermogravimetry 23 decomposition of anhydrous calcium,strontium and barium oxalates the gas evolved is CO and the carbonate is formed.This is true whether air of nitrogen is used.However,the CO is oxidised by oxygen in the air as a surface reaction on the surface of the oxalate-carbonate mixture.This reaction is exothermic produce s local heating of the ren reactant,causing a slight acceleration of its decomposition.No such effe appears for a nitrogen atmosphere,so the decomposition traces differ slightly when different atmospheres are used. The rate of heating will have a maior effect on the result.because.at a high heating rate,the temperature recorded moves to higher values while th reaction is taking e.the tion appe ars to be ata higher temperature.This effect is illustrated in Figure 11 below In some high temperature experiments a solid present may become volatile and sublime out of the crucible.This will have to be accounted for in the description of the mechanisms taking place.However,in unfortu- nate cas me of the solid may redeposit on the support rod or wire for the crucible in sligl c oler regions. This will ca e an appare gain,partially cancelling the mass los and invalid dating the results.The support system should be examined frequently for this effect and any deposits cleaned off thoroughly. In summary.experiments should be carried out with high thermal capacity furnaces and with small,lightweight crucibles.The rate of heat- ing should b t reaction an take pl e over of ang temperature.The sample size should as sma an should bo spread thinly on the base of the crucible.A difficulty arising is that if a sample is coarse-grained then only a few particles may be included and may not be representative of the sample if,say,minerals or concrete are being tested.In this case the sample should be ground ina pestle and rtar t powder a well xed before sampling From the above remarks it may be seen that many factors affect the observed results.This means that although thermogravimetry can ident- ify a substance from its decomposition temperature it should not be thought of as a"finger print"method like spectroscopy.In that technique peak will always be at the same position in the sp ectrum,regardless of make of the instrument or sample A consequence of this is that all experimental variables should be reported thoroughly to allow for variations between instrument and experimental conditions.A rigid set of rules for making reports has been drawn up and should always be adhered to.It is important to ensure that someone reading the report would be able to repeat the experiment exactly and get.as near as possible,the same res lt
Thermogravimetry and Derivative Thermogravimetry 23 decomposition of anhydrous calcium, strontium and barium oxalates the gas evolved is CO and the carbonate is formed. This is true whether air of nitrogen is used. However, the CO is oxidised by oxygen in the air as a surface reaction on the surface of the oxalate-carbonate mixture. This reaction is exothermic and produces local heating of the remaining reactant, causing a slight acceleration of its decomposition. No such effect appears for a nitrogen atmosphere, so the decomposition traces differ slightly when different atmospheres are used. The rate of heating will have a major effect on the result, because, at a high heating rate, the temperature recorded moves to higher values while the reaction is taking place slowly, i.e. the reaction appears to be at a higher temperature. This effect is illustrated in Figure 11 below. In some high temperature experiments a solid present may become volatile and sublime out of the crucible. This will have to be accounted for in the description of the mechanisms taking place. However, in unfortunate cases, some of the solid may redeposit on the support rod or wire for the crucible in slightly cooler regions. This will cause an apparent mass gain, partially cancelling the mass loss, and invalidating the results. The support system should be examined frequently for this effect and any deposits cleaned off thoroughly. In summary, experiments should be carried out with high thermal capacity furnaces and with small, lightweight crucibles. The rate of heating should be low so that reaction can take place over a narrow range of temperature. The sample size should be as small as possible and should be spread thinly on the base of the crucible. A difficulty arising is that if a sample is coarse-grained then only a few particles may be included and may not be representative of the sample if, say, minerals or concrete are being tested. In this case the sample should be ground in a pestle and mortar to a fine powder and well mixed before sampling. From the above remarks it may be seen that many factors affect the observed results. This means that although thermogravimetry can identify a substance from its decomposition temperature it should not be thought of as a “finger print” method like spectroscopy. In that technique a peak will always be at the same position in the spectrum, regardless of the make of the instrument or size of sample. A consequence of this is that all experimental variables should be reported thoroughly to allow for variations between instrument and experimental conditions. A rigid set of rules for making reports has been drawn up and should always be adhered to. It is important to ensure that someone reading the report would be able to repeat the experiment exactly and get, as near as possible, the same result
24 Chapter 2 REPORTING THERMOGRAVIMETRY RESULTS Recommendations for reporting thermal analysis results,including ther mogravimetry results,have been made by the Standardisation C ommit tee of ICTAC and appear in standards such as ASTM E 472(1991).Not all of the points listed may be known,if commercial equipment is used, but as much as possible should be reported. A.Properties of the Sample (1)Identification of all substances (with %if an inert diluent is used). (2)Give the source of all materials used(if commercially obtained,give rer,grade and pu ty) just "from the bottle",but might include grinding,sieving,pre- drying in an oven,surface modification,etc.). (4)Give physical properties if known (particle size,surface area or porosity). B.Experimental Conditions (1)State the apparatus used(manufacturer and model name or numb- 2Deebe the thermal treatmentduring the rn(inta teperature final temperature.rate of heating if linear or full details if not linear). (3)Identify the sample atmosphere(flow rate,pressure,composition and purity).Remember that cylinders of so called pure inert gas, nitroge ,conta a carbon samp in such a ntrogen try holdi observe th mass loss!).Also state if static (zero flow),self-generated (not to be encouraged)or dynamic(flowing)atmosphere is used. (4)State the dimension,geometry and material of the sample holder (crucible).Also give the method of loading.e.g.tipped into the crucible on the bala or weighed out on a separate balance and tapped down on a hard surface. (5)Give the sample mass. C.Data Acquisition and Manipulation Methods Note:much of this may be covered by just quoting the apparatus used
24 Chapter 2 REPORTING THERMOGRAVIMETRY RESULTS Recommendations for reporting thermal analysis results, including thermogravimetry results, have been made by the Standardisation Committee of ICTAC and appear in standards such as ASTM E 472 (1991). Not all of the points listed may be known, if commercial equipment is used, but as much as possible should be reported. A. Properties of the Sample (1) Identification of all substances (with %, if an inert diluent is used). (2) Give the source of all materials used (if commercially obtained, give manufacturer, grade and purity). (3) List the history of the sample before it was sampled (this might be just “from the bottle”, but might include grinding, sieving, predrying in an oven, surface modification, etc.). (4) Give physical properties if known (particle size, surface area or porosity ). B. Experimental Conditions (1) State the apparatus used (manufacturer and model name or number and if modifications have been made). (2) Describe the thermal treatment during the run (initial temperature, final temperature, rate of heating if linear or full details if not linear). (3) Identify the sample atmosphere (flow rate, pressure, composition and purity). Remember that cylinders of so called pure inert gas, such as “white spot” nitrogen, contain traces of oxygen (try holding a carbon sample in such a nitrogen flow at 1000°C and observe the mass loss!). Also state if static (zero flow), self-generated (not to be encouraged) or dynamic (flowing) atmosphere is used. (4) State the dimension, geometry and material of the sample holder (crucible). Also give the method of loading, e.g. tipped into the crucible on the balance or weighed out on a separate balance and tapped down on a hard surface. (5) Give the sample mass. C. Data Acquisition and Manipulation Methods Note: much of this may be covered by just quoting the apparatus used
Thermogravimetry and Derivative Thermogravimetry 25 (1)Give the software version of the computer program for data log ging as supplied by the manufacturer.If known,give details of the signal conditioning routines used.Alternatively give details of self- developed versions. (2)Refer to equations used to process data (3)Give the frequency of sampli ing the rea ings (4)Indicate any filtering,smoothing or other signal conditioning used (5)Give the calibration methods used,if known(for temperature and massl (6)If DTG is reported,give the differentiation method. D.Results (1)State whether the abscissa scale is time or temperature or mV thermoco EM(th shouldnrase from left to right) scale should indicate mass s with ass losses plotted downwards(other scales s ch as I action al reaction,or molecular composition may be used if stated explicitly). (3)Reproduce all of the original records(not just the derived plots). (4)Detail any attempts made to identify intermediate or final prod- ucts. (5)Where possible,identify each the ermal event(mass loss)seen (6)Try to read off the"decomposition temperature"for each reaction step as explained above. Presentation of Results The results may be reported directly as mass of the sample varying with temperature or time,i.e.as m versus T.Thus a mass loss appears as a downwards curve.Instead of mass in mg,the scale may be converted into percent of the original mass.An alternative is to convert the results into a percentage mass loss.The scale is sometimes presented in molecular mass ful if the hanism of de composition is being studied.Partic larly in the study of kinetics,another type of presentation is used:th fraction reacted a.If the original mass is m,the final mass after the reaction has finished is m and the mass at any time is m,then a fraction reacted may be defined as: a=(m,-m)(m.-mg) (1) The fraction will be between 0 and 1.It should be noted that,if a
Thermogravimetry and Derivative Thermogravimetry 25 (1) Give the software version of the computer program for data logging as supplied by the manufacturer. If known, give details of the signal conditioning routines used. Alternatively give details of selfdeveloped versions. (2) Refer to equations used to process data. (3) Give the frequency of sampling the readings. (4) Indicate any filtering, smoothing or other signal conditioning used. (5) Give the calibration methods used, if known (for temperature and (6) If DTG is reported, give the differentiation method. mass). D. Results (1) State whether the abscissa scale is time or temperature or mV thermocouple EMF (the scale should increase from left to right). (2) The ordinate scale should indicate mass with mass losses plotted downwards (other scales such as fractional reaction a, or molecular composition may be used if stated explicitly). (3) Reproduce all of the original records (not just the derived plots). (4) Detail any attempts made to identify intermediate or final prod- (5) Where possible, identify each thermal event (mass loss) seen. (6) Try to read off the “decomposition temperature” for each reaction ucts. step as explained above. Presentation of Results The results may be reported directly as mass of the sample varying with temperature or time, i.e. as rn uersus T. Thus a mass loss appears as a downwards curve. Instead of mass in mg, the scale may be converted into percent of the original mass. An alternative is to convert the results into a percentage mass loss. The scale is sometimes presented in molecular mass units, useful if the mechanism of decomposition is being studied. Particularly in the study of kinetics, another type of presentation is used: the fraction reacted a. If the original mass is m, the final mass after the reaction has finished is rn, and the mass at any time is m, then a fraction reacted may be defined as: The fraction will be between 0 and 1. It should be noted that, if a