Evaluating the performance of polymeric roofing materials with thermal analysis.
Thermal analysis is not widely used in the roofing industry but is gaining popularity (refs. 2-21). Thermoanalytical techniques can be used to monitor a wide array of material characteristics. Some of the applications include enthalpy, weight-loss, thermal stability, coefficient of thermal expansion (CTE) and the glass transition temperature (T[.sub.g]). Thus, these techniques can assist in selecting roofing materials for a particular application. For example, the T[.sub.g] is an important characteristic that should be considered for the cold temperature performance of roofing membranes. Below T[.sub.g] the material will be rigid and hard. Yet, above T[.sub.g] the material will be flexible. Generally, the strength of polymeric materials above the glass transition temperature is lower than the strength below T[.sub.g]. Other properties that vary with T[.sub.g] are thermal expansion coefficient, heat capacity and thelectric constant (refs. 22-26)
There are four main thermoanalytical techniques that are commonly used to determine and monitor the changes in a roofing membrane. They are thermogravimetry (TG), differential scanning calorimetry (DSC), thermomechanical analysis (TMA) and dynamic mechanical analysis (DMA). In this article, a brief overview of these thermoanalytical techniques is presented. In addition, the data obtained by the various techniques on EPDM and PVC roofing membranes are presented.
Thermogravimetry measures the change in mass of a material as a function of time at a determined temperature (i.e., isothermal mode) or over a temperature range using a predetermined heating rate. Essentially, a TG consists of a microbalance surrounded by a furnace. A computer records any mass gains or losses. Thus, this technique is very useful in monitoring heat stability, and loss of components (e.g., oils, plasticizers or polymers).
DSC is widely used in providing valuable information on chemical and physical properties of materials. The DSC technique measures the amount of energy (or heat) absorbed or released as the material is heated, cooled or held at an isothermal temperature. A DSC thermal curve shows the amount of heat evolved or absorbed as a function of temperature or time. This technique yields thermodynamic data such as enthalpy, and specific heat, as well as kinetic data.
The shape and appearance of DSC curves can give a clue as to the type of transition taking place. Generally, first-order transitions such as melting give distinct peaks. These peaks can be integrated and a value for enthalpy (AH) can be determined. In the case of second order transitions such as the glass transition, a step-wise increase in heat capacity is observed. This is also detected in the DSC by a step change in baseline slope. In the case of heavily plasticized materials (e.g., roofing swnples) a broad transition is obtained and the step change is difficult to detect.
Calibration of a DSC instrument is important to obtain useful data. Normally, the melting points of pure standards are used. It is best to use a standard that will have a transition in the same temperature range as the samples being studied.
The glass-transition temperature may be determined by taking the middle of the change in baseline (half-height method). This method requires establishing a tangent line. This step is facilitated by using the first derivative of the heat-flow signal. T[.sub.g] values obtained by DSC are generally different from those obtained by dynamic techniques such as DMA since DSC is a static technique.
DMA (refs. 27-31)
The DMA technique measures the stressstrain relationship for a viscoelastic material. A real and imaginary component of modulus can be obtained by resolving the stress-strain components: [sigma] = [epsilon] E'sin ([omega]t+ [epsilon]o E"cos ([omega]t)
The storage modulus, E', is defined as E' = ([sigma]o/[epsilon]o) cos [delta] and is a measure of recoverable strain energy in a deformed body. The loss modulus, E", is associated with the dissipation of energy as heat due to the deformation of the material and is defined as E" = ([sigma]o/[epsilon]o) sin[delta]. The ratio E"/E' yields the loss tangent or damping factor (tan [delta]) which is the ratio of energy lost per cycle to the maximum energy stored and therefore recovered, per cycle.
A typical dynamic mechanical analysis curve shows either E', E" or tan[delta] plotted as a function of time or temperature. In general, the most intense peak observed for either E" or tan [delta] in conjunction with a relatively pronounced drop in E' corresponds to the glass transition. However, to prove that this relaxation event is a T[.sub.]g, a DMA multiplexing experiment to establish the activation energy would be required. The Tg may be affected by the crosslink density or degree of crystallinity and is directly related to the amorphous region within a polymer.
Care should be taken when reporting the glass-transition temperature obtained by DMA. The transition temperature determined by DMA (or other dynamic techniques) is not only heating-rate dependent but also frequency dependent. In addition to heating rate and frequency, the mechanical/rheological property E', E" or tan [delta]) used to determine the T[.sub.g] must also be specified. It has been found that the E" peak maximum at I Hz corresponded closely with the T[.sub.g] obtained from volume-temperature measurements (ref. 20).
TMA (refs. 28 and 32)
TMA, as defined by ASTM E473-85, is a method for measuring the deformation of a material under a constant load as a function of temperature while the material is under a controlled temperature program. The measuring system consists of a linear voltage differential transformer (LVDT) connected to the appropriate probe. Various probes are available and the measurements can be done in either compression, expansion, penetration, flexure or in tension mode. It is this variety of probes which allows for the measurement on samples of different configurations. Any displacement of the probe generates a voltage which is then recorded. The dimensional change of a simple with an applied force is measured as a function of time or temperature. The plot of expansion (or contraction) vs. temperature (or time) can then be used to obtain T[.sub.g], the coefficient of thermal expansion (CTE), softening temperature and Young's modulus.
The change in linear dimension as a function of temperature can be described by the following:
[Mathematical Expression Omitted]
where [alpha][.sub.1] is the coefficient of linear expansion, and L[.sub.1] and L[.sub.2] are the lengths of the specimen at temperatures (or time) T[.sub.1] and T[.sub.2], respectively. If the difference between T[.sub.2] and T, is relatively small, then the equation can be represented by: L[.sub.2] - L[.sub.1] = L[.sub.1][alpha][.sub.1](T[.sub.2-T.sub.1]) or it can be rewritten as: [alpha][.sub.1] = (1/L[.sub.1])/([delta]L//[delta]T)
Therefore, the slope of the curve of length vs. temperature yields [alpha][.sub.1]L[.sub.1] and the coefficient of linear thermal expansion is obtained by dividing by L[.sub.1].
There are some drawbacks with thermomechanical analysis. Proper calibration is required to obtain reliable and reproducible data. Other sources of errors include slippage of the probe on the specimen and specimens undergoing creep in addition to length changes.
The various experimental conditions are outlined in references 5-7, and 10, 11 and 13.
Results and discussion
Figures I and 2 contain the derivative weight loss (DTG) as a function of temperature of two EPDM roofing membranes. These EPDM membranes had been heat-aged at 100[degree]C for up to 28 days. It is quite apparent from these curves that one sample figure 1) is more stable than the other figure 2). The observed weight loss figure 2) between 200[degree]C and 400'C is due to loss of oils. This loss significantly affected the in-service performance of this membrane, e.g., severe shrinkage was observed.
A similar experiment was carried out on some PVC membranes. The results are shown in figures 3 and 4. As can be seen, the differences between the two are more subtle than in the case of the EPDM membranes. The sample shown in figure 4 has actually lost 4% of its original weight while the one in figure 3 has remained stable. This loss could be attributed to loss of plasticizer.
The results obtained by oscillating DSC on some PVC and EPDM roofing membranes are shown in figures 5-8. It has been found that it is generally difficult to measure the T[.sub.g] using conventional DSC. However, this was not the case using oscillating DSC. The deconvoluted curves are very similar to what would be obtained by conventional DSC. The Cp curve allows for an easy separation of the T[.sub. g] component from the rest of the curve. Thus, one can more easily determine if the changes in the baseline can be attributed to the T[.sub.g].
DMA curves showing E" results are shown in figures 9 to 12. Once again, one can clearly discern between a thermally stable membrane and an unstable one. In the case of the EPDM, it is obvious that the peak maximum shown in figure 9 is not significantly changing, even after being exposed for 28 days at 100[degree]C. This is definitely not the case for the EPDM sample shown in figure 10. This sample has a T[.sub.g] which shifts from approximately -70[degree]C in the as received mode to -40[degree]C after 28 days at 100[degree]C. Similar observations are recorded for PVC samples. Figure 11 contains the E" curve for a PVC sample exposed to 100[degree]C for up to 28 days. As can be seen, the T[.sub.g] (approximately 35[degree]C) does not appear to change. The data for another PVC sample is shown in figure 12. There it is quite obvious that the T[.sub.g] has shifted significantly. Changes in T[.sub.g] can be attributed to various factors (loss of oils or plasticizers, crosslinking, etc.). Since below the Tg the membrane is stiff, DMA allows the membrane manufacturer/user to determine some low temperature behavior of these materials.
A typical TMA curve for an EPDM membrane is shown in figure 13. From such a curve, one can obtain the coefficient of thermal expansion (CTE) and determine the glass-transition temperature. A plot of CTE vs. heat-aging days figure 14) clearly demonstrates how stable one membrane is vs. the other.
Thermal analysis shows much promise in providing quick and reliable data regarding the stability of polymeric roofing membranes. It has been shown how the relative stability of PVC and EPDM membranes can be determined by TG, oscillating-dsc, DMA and TMA. Each technique provides information which is complimentary to the other. Furthermore, this data can be of assistance when trying to understand why a membrane is exhibiting peculiar behavior.
In the future, one would hope that these techniques be incorporated into the relevant membrane standards. In Canada, the next version of the PVC standard will be addressing this issue. Another aspect to be considered is the use of these techniques in service-life-prediction. Using a kinetic approach it may be possible to determine the approximate service-life of some of these materials. Work in this area, has recently been initiated at NRC and data will hopefully be forthcoming in the near future.
[1.] "Performance testing of roofing membrane materials, " Recommendations of the Conseil International du Batiment pour la Recherche I'Etude et la Documentation (CIB) W.83 and Reunion International des Laboratories d'Essai et de Recherche sur les Materiaux et les Constructions RILEM) 75-SLR Joint Committee on Elastomeric, Thermoplastic and Modifted Bitumen Roofing, RILEM, Paris, France, November 1988. [2.] C G. Cash, Single ply roofing technology: ASTM STP 790, pp. 55-64, (1982). [3.] D. Backenstow and P. Flueler, "Thermal analysis for characterization, " Proceedings, 9th Conference on Roofing Technology, National Roofing Contractors Association, Rosemont, IL, pp. 54-68, April 1987. [4.] G.D. Gaddy, WJ. Rossiter, Jr, R.K. Eby, Roofing research and standards development: ASTM STP 1088, pp. 3 7-52, 1990). [5.] O. Dutt, R.M. Paroli, N.R Mailvaganam and R.G. Turenne, "Glass-transitions in polymeric roofing membranes, " Proceedings, 1991 International Symposium on Roofing Technology, 495-501, 1991), F Kocich, editor [6.] R.M. Paroli, O. Dutt, Polymeric materials science and engineering, 65, 362-363 1991). [7.] R.M. Paroli, O. Dutt, A.H. Delgado, M. Mech, Thermochimica Acta, 182, 303-317, 1991). [8.] G.D. Gaddy, W.J. Rossiter, Jr, R.K. Eby, "The application of thermal analysis to the characterization of EPDM roofing membrane materials after exposure in service, " Proceedings, 1991 International Symposium on Roofing Technology, 502-507, 1991, F Kocich, editor [9.] G.D. Gaddy, WJ. Rossiter, Jr and R.K. Eby, ASTM STP 1136, 168-175, 1991). [10.] R.M. Paroli, O. Dutt, A.H. Delgado and H.K. Stenman, Journal of Materials in Civil Engineering 5(l), 83-95,(1993). [11.] J.J. Penn and R.M. Paroli, "Evaluating the effects of aging on the thermal properties of EPDM roofing materials," Proceedings of the 21st North American Thermal Analysis Society (NATAS) Conference, Atlanta, Georgia, Sept. 13-18, 1992, pp. 612-617. [12.] Oba, and F. Bjork, Polymer Testing, 12, 35-56 (1993). [13.] M. Paroli, TL. Smith and B. Whelan, "Shattering of unreinforced PVC roof membranes: Problem phenomenon, causes and prevention," NRCA/NIST Tenth Conference on Roofing Technology, April 22-23, 1993, pp. 93-107. [14.] James J. Penn, Ralph M. Paroli, Thermochimica Acta, 226, 77-84 1993). [15.] Rodriguez, O. Dutt, R.M. Paroli and N.P. Mailvaganam, Materials and Structures, 26 (160), 355-361 1993). [16.] Oba, Flat roofs: Investigation of heat welding techniques for polymer-modifted bituminous roofing membranes, " dissertation, Royal Institute of Technology, Sweden 1994). [17.] Oba, and M.N. Partl, Empa-forschungs- und Arbeitsbericht No. 13616: FE 147'135. Swiss Federal Laboratories for Materials Testing and Research, Switzerland (1994). [18.] Oba, and. A. Roller, Characterization of polymer-modified bituminous roofing membranes using thermal analysis, Materials and Structures, in press (1994). [19.] Ralph M. Paroli and James Penn, Assignment of the Glass Transition, ASTM STP 1249, R.J. Seyler, Ed., American Society for Testing and Materials, Philadelphia, 1994, pp. 269-276. [20.] Paroli, R.M., Dutt, O., Smith, T.L. and Whelan, B.J., Roofing research and standards development: Volume 3, ASTM STP 1224, Thomas J. Wallace and Walter J. Rossiter, Jr., Eds., American Society for Testing and Materials, Philadelphia, 1994, pp. 139-147. [21.] Ralph M. Paroli, O. Dutt, Ana H. Delgado and William Lei, Polymeric materials science and engineering, 72, 378-379 (1995). [22.] Feldman, D., Polymeric building materials, Elsevier. Science Publishers Ltd., New York, 1989. [23.] Billmeyer, F. W., Textbook of polymer science, 3rd Ed., John Wiley and Sons, New York, 1984. [24.] Young, R.J., Introduction to polymers, Chapman and Hall, New York, 1981. [25.] Bikales, N.M., Mechanical properties of polymers, John Wiley and Sons, New York, 1971. [26.] Flory, P.J., Principles of polymer chemist , Cornell University Press, 1967. [27.] Murayama, T., Dynamic mechanical analysis of polymeric material, Elsevier Scientific Publishing Company, New York, 1978. [28.] Wendlandt, W. Wm., Chemical analysis 19: Thermal analysis, 3rd Edition, John Wiley and Sons, 1986. [29.] Campbell, D. and White, J.R., Porymer characterization: Physical techniques, Chapman and Hall, New York, 1989. [30.] Crompton, T.R., Analysis of polymers: An introduction, Pergamon Press, 1989. [31.] Skrovanek, D.J. and Schoff, CK, Progress in organic coatings, 16, 135-163 1988). [32.] Riga, A.T, Neag, M.N. (editors), Materials characterization by thermomechanical analysis, ASTM STPI 136, 1991.