J. Anal. Appl.

Pyrolysis 71 (2004) 2746

Pyrolysis combustion ow calorimetry

Richard E. Lyon a, , Richard N. Walters b
a

Fire Safety Branch AAR-440, Federal Aviation Administration, William J. Hughes Technical Center, Atlantic City International Airport, Atlantic City, NJ 08405, USA b Galaxy Scientic Corporation, Egg Harbor Township, NJ, USA

Abstract A method for evaluating the combustibility of milligram samples is described. Pyrolysis-combustion ow calorimetry (PCFC) separately reproduces the solid state and gas phase processes of aming combustion in a nonaming test by controlled pyrolysis of the sample in an inert gas stream followed by high temperature oxidation of the volatile pyrolysis products. Oxygen consumption calorimetry is used to measure the heat of combustion of the pyrolysis products. The maximum amount of heat released per unit mass per degree of temperature (J g1 K1 ) is a material property that appears to be a good predictor of ammability. 2003 Elsevier B.V. All rights reserved.
Keywords: Polymer; Fire; Flammability; Thermal analysis; Combustion; Calorimetry; Fire hazard; Heat release rate; Oxygen consumption

1. Background The rate at which heat is released by a burning material is the single most important parameter determining its hazard in a re, particularly in an enclosed space such as a building, a ship, or an aircraft cabin [13]. Several different bench scale methods have been developed for measuring heat release rate during aming combustion of materials, products, and components [4,5]. These bench scale re calorimetry methods require replicate samples on the order of a 100 g each and the results are highly dependent on the ignition source [6], sample thickness [7], sample orientation [8], ventilation [9], and edge conditions [8]. All of these factors combine to make the test data conguration dependent and obscure the effect of material properties and composition on burning behavior.

Corresponding author. Tel.: +1-609-485-6076; fax: +1-609-485-6909. E-mail address: richard.e.lyon@faa.gov (R.E. Lyon).

0165-2370/$ see front matter 2003 Elsevier B.V. All rights reserved. doi:10.1016/S0165-2370(03)00096-2

28

R.E. Lyon, R.N. Walters / J. Anal. Appl. Pyrolysis 71 (2004) 2746

The desire for a quantitative analytical laboratory test that correlates re behavior or ame test performance with material properties has been the motivation to relate thermogravimetric analyses to ammability [1013]. To date, most thermogravimetric investigations of ammability have relied on a single thermal stability parameter (e.g. char yield or thermal decomposition temperature) to relate the chemical composition of a material to its re or ame test performance (e.g. char yield vs. limiting oxygen index). Individually, these thermal stability parameters have found limited success as material descriptors of ammability and their interrelationship in the context of aming combustion has remained obscure until recently, when it was shown that a particular combination of thermal stability and combustion parameters could correlate re behavior [14]. Laboratory methods that directly measure heat release rate of milligram-sized samples of materials have been developed to study the effect of material properties and chemical composition of materials on combustibility under controlled conditions. Susott [1517] was the rst to measure the heat of combustion of organic materials using transient heating and oxygen consumption. In Susotts method milligram-sized samples of forest products (foliage, wood, stems, and bark) were pyrolyzed in an inert gas stream at a constant rate of temperature rise and the pyrolysis gases were combined with oxygen prior to entering a catalytic reactor. Pyrolysis was conducted under inert gas ow to prevent oxidation of the char residue. The rate of oxygen consumption was determined from the electrolytic oxygen generation rate using a null-balance, closed-loop technique. Only qualitative information was obtained for the dynamics of the fuel generation process because the oxygen consumption signal was distorted by the instrument, and therefore, was not synchronized with the mass loss history of the sample. Pyrolysis residue/char fraction was measured by weighing the sample before and after the test and the heat of combustion of the char was determined separately. Sample heating rates were limited by the dynamic capability of the oxygen generator to less than 16 K min1 . The rate of heat released by complete oxidation of the pyrolysis gases during thermal decomposition of the sample was calculated from the measured oxygen consumption rate using an average gross heat of combustion for forest products or 14.0 0.5 kJ g1 O2 . The water produced in aming combustion is in the gaseous state so the relevant heat of combustion is the net value obtained by subtracting the heat of vaporization of water from the gross heat of combustion. Converting Susotts gross heat of combustion to a net value gives 13.3 0.5 kJ g1 O2 which is equivalent to the currently accepted value 13.1 0.7 kJ g1 O2 used in oxygen consumption re calorimetry [5,1820] as determined from data on a wide range of organic solids and polymers [21,22], as well as liquids and gases [23]. Susott did not measure mass loss during the pyrolysis-combustion experiment but normalized the total heat of combustion of the volatiles to the starting sample mass on an ash-free, dry fuel basis. The gross heat of combustion of the char determined in separate oxygen bomb calorimetry experiments [24] was found to be relatively independent of the fuel type at 32.0 0.9 kJ g1 for the 43 typical forest fuels tested. Parker [25,26] claims to have improved on Sussots non-aming dynamic combustion method by using a step-change in temperature to pyrolyze the sample at a heating rate considered to be more typical of the surface conditions in a re. In Parkers device samples of cellulose are inserted into a preheated, nitrogen-purged furnace resulting in a rapid uncontrolled temperature change and subsequent pyrolysis. An impinging stream of oxygen is mixed with the inert gas-pyrolyzate stream in a catalytic reactor. Combustion gas con-

R.E. Lyon, R.N. Walters / J. Anal. Appl. Pyrolysis 71 (2004) 2746

29

centrations (O2 , CO2 , H2 O) are measured and oxygen consumption is used to calculate the heat release rate of volatile fuel combustion. Although mass loss is not measured directly during the test, transient oxygen depletion and combustion gas generation were used to calculate the mass loss rate of cellulose from its known chemical structure. Babrauskas and Parker [27] later rened Parkers device for the measurement of the fuel/oxygen ratio of re gases. Reshetnikov et al. [28,29] used an experimental arrangement similar to Parker [25,26] in their study of the gas phase oxidation kinetics of polymer decomposition products. In Reshetnikovs technique milligram polymer samples are decomposed isothermally in an inert gas stream and excess oxygen is added to the volatile fuel stream at various temperatures to effect oxidation. A gas chromatograph (GC) with ame ionization detector (FID) was used to sample the fuel stream prior to mixing with oxygen in order to separate and identify the individual products of decomposition. Gas phase oxidation kinetics of the fuel species are computed from the measured oxygen consumption history using isothermal methods of analysis. Reshetnikovs device measures the rate of combustion (oxidation) of the fuel gases but not the rate at which these gases are produced by the decomposing solid as it is heated, the latter being the rate limiting process in a re. Lasers have been used to pyrolyze polymers for ammability studies by Price [30] and Angel [31] and the pyrolysis gases analyzed by mass spectrometry and laser induced uorescence of hydroxyl radicals, respectively. In combination with thermochemical calculations of the heat of combustion of the fuel species, laser pyrolysis methods can provide an estimate of the total heat released by the polymer. Commercial thermogravimetric analyzers (TGA) have been used to thermally decompose milligram samples under controlled heating and environmental conditions for aming and nonaming combustion studies. Coupling the TGA to an evolved gas analysis (EGA) detector which is synchronized with the mass loss signal provides dynamic (rate) capability during the test and allows for interpretation of the transient evolved gas data in terms of the decomposition kinetics of the solid using established methods of nonisothermal analysis [3235] and thermal degradation models for charring and noncharring polymers [36]. Flaming combustion in a TGA was used by Gracik et al. [37,38] in combination with evolved gas (CO, CO2 ) analysis to study the ammability of ber reinforced polymer composites. In Graciks test a 50 mg sample is heated at a rate of 20 K min1 in air in a commercial TGA until ignition occurs. Flaming combustion of the sample in the TGA produces carbon dioxide (CO2 ) and some carbon monoxide (CO) which are measured and used to calculate the heat release rate [39]. An advantage of CO2 /CO generation calorimetry compared with oxygen consumption calorimetry as a measure of combustion heat is the higher sensitivity and lower cost of the CO2 /CO detector(s) but the method is more sensitive to fuel type. Graciks method is essentially a scaled down version of the early Factory Mutual Research Corporation re calorimeter [40] but uses a controlled heating rate instead of a controlled heat ux to force gasication of the solid. Nonaming combustion of volatile fuel products generated in a TGA was used by Kifer [41] who studied re retarded polymers and wood using a high temperature FID to burn the pyrolysis gases generated in a nitrogen-purged TGA. The FID signal is proportional to the total carbon in the pyrolysis gases, so that the integrated value, which was assumed to be

30

R.E. Lyon, R.N. Walters / J. Anal. Appl. Pyrolysis 71 (2004) 2746

proportional to the total heat release, was related to the ammability of the material. Walters and Lyon [42,43] and Inguilizian [44] determined the heat release rate of milligram samples using controlled heating in a TGA to thermally decompose the polymer and determined the heat release rate of the solid from the product of the maximum mass loss rate in the TGA and the net heat of combustion of the volatile pyrolysis products. Walters and Lyon used oxygen consumption calorimetry to measure the heat of combustion of the pyrolysis products while Inguilizian calculated the heat of combustion from the oxidation thermochemistry of the primary fuel species identied by gas chromatography-mass spectrometry (GC-MS). The present method of pyrolysis-combustion ow calorimetry (PCFC) seeks to improve upon laboratory pyrolysis-combustion methods by providing dynamic capability for solids without the need to measure mass loss rate during the test. Because the PCFC combustibility test requires milligram samples rather than the kilogram samples of re calorimetry, it is microscale by comparison to aming heat release rate tests.

2. Principle of operation PCFC measures the rate at which the heat of combustion of the fuel gases is released by a solid during controlled pyrolysis in an inert gas stream. The fuel gases are mixed with excess oxygen and completely oxidized at high temperature and the instantaneous heat of combustion of the owing gas stream is measured by oxygen consumption calorimetry. The rate at which combustion heat Q is liberated per unit time t in the PCFC or in a re is limited by the rate at which the thermally decomposing solid releases fuel because the gas phase combustion reactions in the ame are rapid by comparison. Thus, the heat release rate dQ/dt (W) is proportional to the mass generation rate (g s1 ) of volatile fuel, i.e. the mass loss rate of the solid: d mv (t) d ms (t) 0 0 c (t) d Qc (t) = hc Q = hc (1) ,v (t) ,v (t) dt dt dt
0 (J g1 ), the enthalpy (heat) of where a superscripted dot indicates the time derivative; hc ,v complete combustion of the volatile pyrolysis products of mass mv ; and ms is the instantaneous residual mass of solid. In a re volatiles are generated at the surface of the material over a range of temperatures and a distribution of molecular weights and atomic composi0 varies over the mass loss history and cannot tions are produced [45], so that in general hc ,v be treated as a constant in Equation 1. In many cases low molecular weight organic and inorganic (e.g. HCl, SOx ) species are cleaved from the polymer backbone and released in the initial stages of fuel generation followed by higher molecular weight organic compounds in the intermediate and latter stages [45,46]. For materials that form a carbonaceous char during the fuel generation process the instantaneous atomic composition of the volatiles will necessarily differ from the atomic composition of the original material [17], with hydrogen being evolved in a secondary, high temperature decomposition step [13]. Consequently, a constant heat of combustion for the volatile fuel that is equal to the heat of combustion of the solid cannot be assumed except for the few polymers that thermally decompose by depolymerization exclusively to monomer (e.g. polymethylmethacrylate, polyoxy-methylene,

R.E. Lyon, R.N. Walters / J. Anal. Appl. Pyrolysis 71 (2004) 2746

31

0 (t) must be determined poly(-methylstyrene) or to a single, known species. Thus, hc ,v continuously during the course of the fuel generation process to obtain dQc (t)/dt from the 0 during fuel generation is problematic and mass loss rate. Continuous determination of hc ,v 0 have been used instead. time consuming so average [4244] values of hc ,v Thornton [23] made the experimental observation that the net heat of complete combustion of typical organic molecules per mol of oxygen consumed is relatively constant at C = 419 19 kJ mol1 O2 = 13.1 0.6 kJ g1 O2 , and is essentially independent of the chemical composition of the combusted material. Sussot [15], Huggett [21], Babrauskas [22], and Walters [47] later conrmed this result for a wide range forest products, chemical compounds, and organic polymers, thereby establishing oxygen consumption as the preferred method for determining the heat released in aming combustion. Thus, to a good approximation ( 5%) the net heat of complete combustion of the volatile degradation products, regardless of chemical composition, is: 0 mv (t)hc ,v (t) = C m O2 (t)

(2)

where mO2 (t) is the mass of oxygen consumed. Substituting the time derivative of Equation 2 into Equation 1: d Q(t) d mO2 (t) d ms (t) 0 c (t) = hc Q =C =C m O2 (t) ,v (t) dt dt dt (3)

c (t) (W) resulting from complete shows that the instantaneous heat release rate of the solid Q and instantaneous combustion of the volatile decomposition products can be determined from the product of the mass loss rate and heat of combustion of the fuel, or more simply 0 m O O2 . Equation 3 shows that from the mass consumption rate of oxygen m O2 (t) = m 2 the rate at which heat is released by combustion of the fuel gases during polymer thermal degradation can be obtained by measuring the mass of oxygen consumedand this result is independent of the composition of the fuel. The total heat of combustion of the volatiles (J) after the pyrolysis process is complete and the rate of oxygen consumption returns to zero is simply the time integral of Equation 3: Qc =
0

c (t)dt = C Q
0

O2 (t)dt m

(4)

While Equation 4 provides a means for relating the total oxygen consumed to the total heat released by combustion of the pyrolysis products, the heat release history of the polymer (Equation 3) requires that the oxygen consumption measured downstream be synchronized with the fuel generation (mass loss) history of the sample in the pyrolyzer. Unfortunately, the ow of combustion gases from the pyrolyzer (or burning surface) to the oxygen analyzer through PCFC cold traps and/or scrubbers distorts or smears the oxygen signal. Distortion of the oxygen consumption signal does not effect the total (time integrated) heat of combustion [15,25,28], but the heat release rate is no longer related to the fuel generation rate of the sample [37,42]. Heat release rates computed from the product of mass loss rate and heat of combustion (Equation 3) did not compare well with those obtained from the oxygen consumption history in the PCFC unless the oxygen signal was deconvoluted to correct for instrumental broadening. A simple mixing model for the pyrolysis-combustion ow

32

R.E. Lyon, R.N. Walters / J. Anal. Appl. Pyrolysis 71 (2004) 2746

calorimeter (see Appendix A) was conrmed by experiment and gives the heat release rate of the solid (W) in terms of the change in oxygen concentration measured at the detector: c (t) = CF + d Q dt (5)

where and F are the gas stream density and volumetric owrate, respectively, and is the response time of the instrument. Dividing Equation 5 by the initial mass of the sample m0 (g) gives the specic heat release rate in units of W g1 . c (W g1 ) = q c (t) d Q CF + = m0 m0 dt (6)

If the sample is pyrolyzed at a constant rate of temperature rise (K s1 ) and thermal decomposition to inert or char fraction, and/or volatile fuel occurs in a single step then the maximum specic heat release rate of the sample is related to the pyrolysis kinetic parameters [48,49]: c (J g1 K1 )
max q c CF = m0

d dt

=
max

o E hc ,s a

eRT2 p

(7)

where Ea is the global activation energy for pyrolysis; Tp , the temperature of maximum mass 0 =(1)ho is loss rate; e, R are the natural number and gas constant, respectively, and hc ,s c,v the heat of combustion of the fuel gases per unit initial mass of solid. The derived expression for the maximum specic heat release rate normalized for heating rate on the right hand side of Equation 7 is a collection of thermochemical properties that has the units (and signicance) of a heat release capacity c and depends only on the composition of the material [14]. The following sections describe the construction, calibration, and operation of a pyrolysis combustion ow calorimeter that measures the heat release rate, total heat, and heat release capacity of polymers using only milligram samples.

3. Instrument construction Fig. 1 is a schematic diagram showing how the component processes of aming combustion are reproduced in pyrolysis-combustion ow calorimetry. The apparatus is based on Susotts original concept [1517] of linear programmed heating of milligram samples in an inert (non-oxidizing) atmosphere to separate the solid state and gas phase processes of aming combustion as normally occur in a re. In particular, char formation in a re occurs at or below the burning solid surface in contact with a fuel rich/oxygen decient gas layer [46]. In the present device the sample is heated using a linear temperature program and the volatile thermal degradation products are swept from the pyrolysis chamber by an inert gas and combined with excess oxygen in a tubular furnace at ame temperatures to force complete non-aming combustion (oxidation) of the fuel. Combustion products CO2 , H2 O, and acid gases are scrubbed from the gas stream and the transient heat release rate is calculated from the measured ow rate and oxygen concentration after correcting for ow dispersion as per Equation 6. Time-integration of the PCFC heat release rate gives the heat of complete

R.E. Lyon, R.N. Walters / J. Anal. Appl. Pyrolysis 71 (2004) 2746

33

Fig. 1. Schematic of aming combustion and pyrolysis combustion ow calorimetry.

combustion/oxidation of the pyrolysis gases and the char yield is measured by weighing the sample before and after the test. If the pyrolysis is conducted in air so that the organic char is completely oxidized during the high temperature hold period, time-integration of the oxygen consumption signal gives the net heat of complete combustion of the solid as would be determined in a high pressure oxygen bomb calorimeter [50]. 3.1. Pyrolyzer Initial studies using a commercial TGA to pyrolyze the polymer samples [42,43] were unsuccessful due to condensation of the thermal decomposition products in the TGA cell and in the heated transfer line. Smearing of the output signal (heat release rate) was also observed because of dilution of the pyrolysis gases with nitrogen in the large mixing volume of the TGA cell. Moreover, the maximum heating rate capability of the TGA (100200 K min1 ) is at the lower range of the surface heating rates in res [48]. For these reasons a temperature-controlled pyrolysis chamber was designed that could be continuously purged with gas, coupled directly to the combustion furnace, and accept a commercial probe pyrolyzer (Pyroprobe 1000/2000, CDS Analytical) to thermally decompose the sample. This arrangement provided consistent temperature and minimum dead-volume with the probe in place for the experiment as shown in Fig. 2. The probe pyrolyzer body is 6.4 mm in diameter and contains a 3 mm diameter, 25 mm long platinum resistance coil that heats the sample at a constant rate in the reported range = 2010 3 20103 K s1 . At the highest heating rate the temperature history of the sample approximates a step change to a preset temperature and in this mode can be used to study the isothermal pyrolysis kinetics of liquids and solids by pulsed heating [51,52]. Mass transfer efciency from the heated pyrolysis chamber to the combustor was studied for a few polymers (polyethylene, polyetheretherketone (PEEK), and KEVLAR) to determine the minimum temperature necessary to maintain all of the pyrolysis products in the gaseous state entering the combustor. Mass transfer efciency was calculated as the ratio of the time-integrated heat release rate (total heat of combustion) of the pyrolysis gases at 0 (T) to the maximum value obtained in the experiments, i.e. mass cell temperature T, i.e. hc ,s 0 (T)/h0 (max). The results of these studies are plotted in Fig. 3 as the transfer efciency = hc ,s c,s

34

R.E. Lyon, R.N. Walters / J. Anal. Appl. Pyrolysis 71 (2004) 2746

Fig. 2. PCFC pyrolyzer and pyrolysis chamber.

mass transfer efciency versus pyrolysis chamber temperature. The data in Fig. 3 indicate that high molecular weight thermal degradation products are generated during pyrolysis that vaporize at temperatures approaching the decomposition temperature of the polymer. Loss of these low volatility fuel products by condensation between the pyrolyzer and combustor reduces the peak heat release rate and total heat release unless the pyrolysis chamber temperature is held to within a few degrees of the onset (1%) weight loss temperature of the sample. Mass ow controllers measure and regulate the ow of nitrogen, oxygen, or air, depending on the experiment. To prevent condensation of high molecular weight decomposition products on the walls of the pyrolysis chamber it is held several degrees below the onset decomposition temperature of the sample determined in a separate TGA experiment at moderate (1020 K min1 ) heating rate. 3.2. Combustor The combustor is a coiled, 5 m length of 6.35 mm OD Inconel tubing having a wall thickness of 0.89 mm, a coiled length of 24 cm, and an outer coil diameter of 5 cm as shown in cross-section in Fig. 4. The gas stream from the pyrolyzer passes through the

Fig. 3. Effect of pyrolysis cell wall temperature on mass transfer efciency of pyrolysis products for PEEK, KEVLAR, and polyethylene.

R.E. Lyon, R.N. Walters / J. Anal. Appl. Pyrolysis 71 (2004) 2746

35

Fig. 4. Combustor cross-section and temperature distribution for 900 C setpoint.

coiled combustion tube contained in a cylindrical ceramic furnace capable of maintaining a maximum temperature of 1200 C. The ceramic furnace is surrounded with 5 cm of mineral wool insulation and the entire assembly is enclosed in a 3 mm thick cylindrical aluminum shell. The combustion tube length was selected to provide a residence time of approximately 60 s at a volumetric ow rate of 100 cm3 min1 in order to completely oxidize the pyrolyzate stream. Published studies of the oxidation of the products of aming combustion indicated that a residence time of 60 s at 1000 C was required to completely oxidize the largest size soot particles observed real res [27]. However, gaseous pyrolysis products and re gases should be completely (> 99%) oxidized in less than a few seconds at the nominal 900 C combustor temperature, as deduced from high temperature oxidation kinetics of methane [53] and volatile polymer pyrolysis products [28]. The temperature distribution along the length of the Inconel tubing coil was measured using a shielded thermocouple probe positioned at several locations along the inside surface of the coil with nitrogen owing through the coil at 100 cm3 min1 . These experiments were repeated for various set point temperatures. A nearly symmetric parabolic temperature distribution about the coil midpoint location, x = 0, as shown in Fig. 5 for the nominal 900 C set point. Temperature distributions for set point temperatures from 500 to 1000 C were similar. 3.3. Combustion gas scrubbing The gas stream exiting the combustor contains nitrogen, combustion products, and residual oxygen. This combustion gas stream enters two 6 mm diameter Teon tubes, 20-cm long, connected in series and tightly packed with anhydrous calcium sulfate (DrieriteTM ) and sodium hydroxide coated silica (AscariteTM ), respectively, to remove any H2 O, acid gases, and CO2 from the sample stream. After cooling and scrubbing, the combustion gas stream contains only nitrogen and the residual oxygen which was not consumed in the combustion reactions. Removing the combustion products from the gas stream ensures that they do not dilute the oxygen concentration and prolongs the life of the oxygen analyzer. The owrate of the scrubbed combustion gas stream (100 cm3 min1 ) is measured by a mass owmeter and continuously recorded by the data acquisition system.

36

R.E. Lyon, R.N. Walters / J. Anal. Appl. Pyrolysis 71 (2004) 2746

Fig. 5. Heat release rate signal for square wave methane calibration pulses before and after signal deconvolution.

3.4. Data acquisition The mass ow rate of pure oxygen entering the nitrogen-pyrolyzate stream is set by the mass ow controller assuming ideal gas behavior. The mass owrate of oxygen in the scrubbed sample stream after combustion of the volatile polymer degradation products is determined from the measured mass ow rate of the gas stream and the concentration of oxygen measured by a high speed commercial oxygen analyzer. For the present work a zirconia oxygen analyzer (Panametrics Series 350) was used having a 90% response time of less than 1 s and an accuracy 1% of full scale (020% O2 , v/v). Temperatures of the pyrolysis probe, pyrolysis chamber, combustor, and gases are monitored continuously during the test as well as gas and sample stream ow rates and oxygen concentration. Data are acquired at 5 Hz on a personal computer during the experiment using a multichannel data acquisition board and software (National Instruments). To determine the heat release rate from oxygen consumption the mass ow rate, temperature, and oxygen concentration of the gas stream before and after combustion of the pyrolysis products are measured. 4. Experimental 4.1. Materials Polymer samples were unlled, natural, or virgin-grade commercial resins obtained from Aldrich Chemical Company, Scientic Polymer Products, or directly from manufacturers. Methane, oxygen, and nitrogen gases used for calibration and testing were dry, >99.99% purity grades obtained from Matheson Gas Products, Welco, and Praxair. 4.2. Instrument calibration A heating rate calibration of the pyrolysis probe was performed using a ne thermocouple in the quartz sample tube to measure the actual rate of temperature rise of the sample at

R.E. Lyon, R.N. Walters / J. Anal. Appl. Pyrolysis 71 (2004) 2746

37

the nominal programmed rate. Heat release rate calibrations were performed by metering methane gas directly into the pyrolyzer in a continuous stream and measuring the oxygen depletion of the scrubbed gas stream after complete combustion. Steady-state oxygen depletion was in quantitative agreement with expected values using the net heat of combustion and ow rate of methane and the stoichiometric fuel/oxygen ratio. Square-wave fuel pulses were generated using a syringe pump containing methane to test the dynamic response of the ow calorimeter to an instantaneously applied ( = ) heat release rate. A characteristic time = 6 for the instrument was obtained from the dynamic response of the oxygen depletion signal to the step change in methane ow (heat release rate) as per Equation A6 in the Appendix A. Equation 6 with = 6 s was then used to deconvolute the oxygen consumption history for the square wave methane heat release rate pulse with the results shown in Fig. 5. It is seen that the square wave pulse is reasonably reproduced by the deconvolution procedure with only a slight overshoot at the step changes because of signal noise. 4.3. Test procedure Polymer samples were dried for at least 8 h at 80 C in a convection oven and held in a dessicant chamber until testing. A dry sample weighing 1.0 0.2 mg is placed into a pre-weighed, thin-walled, quartz capillary tube which is 2.5 mm in diameter and 1012 mm long. The sample and tube are weighed on a microbalance to an accuracy of 2 g to determine the initial, mo , and nal, m, sample mass after subtracting the weight of the quartz tube. The quartz tube containing the sample is inserted into the heating coil of the pyrolysis probe and the probe is inserted into the pyrolysis chamber and sealed. The pyrolysis chamber is equilibrated at a temperature a few degrees below the onset degradation temperature of the sample as determined in separate TGA experiments at a heating rate in the range of 10 K min1 . A constant rate of temperature rise (ramp) is used to heat the sample from the initial pyrolyzer temperature to a maximum temperature Tmax where it is held for a period of time. The parameters relevant to polymer ammability are measured using anaerobic pyrolysis and a soak temperature that is well above the thermal decomposition range of the solid to force complete thermal decomposition. Under these conditions Equation 7 applies and the maximum heat release rate normalized for heating rate has physical signicance [48,49] as the capacity of a material to release its heat of combustion in a re. Thus, the standard ammability test in the pyrolysis-combustion ow calorimeter involves heating the sample at a constant rate of temperature rise (258 K min1 , typically) in nitrogen owing at 82 cm3 min1 to a maximum temperature Tmax = 930 C and holding the sample at Tmax for between 10 and 120 s to effect complete pyrolysis. The volatile pyrolysis products generated during the temperature ramp are swept from the pyrolyzer by the nitrogen purge gas and 18 cm3 min1 of pure oxygen is added to the pyrolyzate/nitrogen gas stream at the inlet to the combustor. Combustion products carbon dioxide, water, and acid gases (if any) exiting the combustor are removed by the scrubbers and the dry nitrogen and residual oxygen pass through a owmeter and oxygen analyzer. Deconvolution of the oxygen consumption signal is performed numerically during the test and the heat release rate, heat release capacity, and total heat of combustion are calculated and displayed. The quartz tube is weighed after the test to determine the residual mass of the sample.

38

R.E. Lyon, R.N. Walters / J. Anal. Appl. Pyrolysis 71 (2004) 2746

The total heat of combustion of the solid, as opposed to just the pyrolysis gases, can be measured by switching the purge gas from nitrogen to air after the temperature ramp, or by using air as the purge gas during the ramp and hold cycles. Thus, selective thermoxidative degradation of the solid or char can be accomplished and the oxygen consumption can be measured to determine the heat of complete combustion since any residual organic material including carbonaceous char, will be completely oxidized in air at 930 C. Experiments were conducted in which dry air was the purge gas in the pyrolyzer during the heating cycle so that the sample was thermally degraded in the presence of oxygen and the remaining carbonaceous char was oxidized during a 2 min hold period at 930 C. The quartz tube is weighed after the test to determine the residual mass of the sample, if any.

5. Results and discussion All results are the average of three to ve replicated determinations for each polymer sample. Data on polymers from different sources was averaged to obtain generic values of the combustion parameters. The repeatability of heat release rate measurements for a single operator is estimated to be 3% and accuracy is 7%, with the majority of the error associated with moisture pickup during the weighing and handling of small (1 mg) samples and the noise in the deconvoluted oxygen consumption signal. Repeatability and accuracy of total heats of combustion obtained by time integration of the heat release rate is 5%. 5.1. Heat release rate Experiments were conducted in which the heat release rate of 1 mg samples of several common polymers was measured over a range of nominal heating rates from 0.1 to 10 K s1 . Fig. 6 shows the results of these experiments plotted as the maximum specic heat release rate versus heating rate. Below a nominal heating rate of about 5 K s1 the maximum specic heat release rate is proportional to the heating rate as per Equation 7 with slope equal

Fig. 6. Specic heat release rate versus heating rate for 1 mg samples of several common polymers.

R.E. Lyon, R.N. Walters / J. Anal. Appl. Pyrolysis 71 (2004) 2746

39

Fig. 7. Specic heat release rate data for several common polymers at a heating rate of 258 K min1 (horizontally shifted for clarity).

to the heat release capacity of the sample. Above a nominal heating rate of about 5 K s1 the specic heat release rate is no longer proportional to the heating rate due to thermal lag of the sample and quartz tube. In practice a nominal heating rate of 5 K s1 is used for the experiment and the maximum heat release rate (c.f. peak height in Fig. 3) is divided by the actual heating rate in the test ( = 4.3 K s1 ) to obtain the heat release capacity. Fig. 7 shows experimental heat release rate data for several polymers heated in nitrogen at 258 K min1 to a maximum temperature of 930 C and held at the maximum temperature for 10 s. Included in this data are polyethylene (PE), polypropylene (PP), polystryene (PS), acrylonitrilebuatdienestyrene terpolymer (ABS), polymethylmethacrylate (PMMA), polyethyleneterephthalate (PET), polyetheretherketone (PEEK), and polybenzimidazole (PBI). The maximum specic heat release rates of these commercial polymers varies by more than a factor of 10 and the magnitude is in qualitative agreement with their ignition resistance [54]. To calibrate the PCFC and test the derived ammability relationship (Equation 7), heat release capacities were measured in the PCFC at = 4.3 K s1 (258 K min1 ) and compared with values obtained for the same samples by TGA-gas chromatography/mass spectrometry (TGA/GC-MS) [44,55] at a nominal heating rate of 10 K min1 . In the TGA/GC-MS the o (t) at peak mass loss rate was computed from the therinstantaneous heat of combustion hc ,v mochemistry of the major species identied by mass spectroscopy. The heats of combustion so calculated were multiplied by the instantaneous maximum fractional mass loss rate in the TGA to obtain the maximum specic heat release rates as per Equation 3. Dividing the maximum heat release rate by the initial sample mass and heating rate ( = 10 K min1 ) gives the heat release capacity by TGA-GC/MS for comparison to PCFC values at = 258 K min1 . Results for the same samples tested by the two different test methods are listed in Table 1. The average weighted difference between the heat release capacities obtained by the two different techniques (i.e. PCFC and TGA/GC-MS) is 9% for the 14 polymers in Table 1. 0 ), and Table 2 is a partial listing of heat release capacities (c ), total heat release (hc ,s char yield () of the hundreds of polymers tested in our laboratory by PCFC for which the Chemical Abstract Service (CAS) registry numbers were available [56]. Data for generic

40

R.E. Lyon, R.N. Walters / J. Anal. Appl. Pyrolysis 71 (2004) 2746

Table 1 Comparison of heat release capacities by PCFC and TGA-GC/MS Polymer Heat release capacity (J g1 K1 ) PCFC ( = 258 K min1 ) Polyethyene (PE) Polyproplylene (PP) Polystyrene (PS) Poly(-methyl)styrene (PMS) Polyoxyphenylene (POP) Polyhexamethyleneadipamide (PA66) Polyethyleneterephthalate (PET) Polycarbonate (PC) Polymethylmethacrylate (PMMA) Poly(p-aramide) (KEVLAR) Polyoxymethylene (POM) Polyetheretherketone (PEEK) Polyphenylenesulde (PPS) Polyimide (PI) 1600 1391 1198 730 553 494 393 390 376 292 261 180 156 29 TGA/GC-MS ( = 10 K min1 ) 1422 1338 1302 695 635 509 407 470 360 207 233 222 118 31

polymers from different sources were averaged to obtain the values in Table 2 with the exception of PMMA for which the range of values is given. Polymers are listed in descending order of heat release capacity c which span the entire range measured to date for polymeric solids. Heat release capacities can vary by 20% between polymers from different sources, e.g. compare Tables 1 and 2 and polyimides in Table 2. Heat release capacity variations probably reect real differences in thermal stability, backbone chemical structure, molecular weight, chain defects, end groups, thermal processing history, and additive package between 0 in Table 2 is the heat of complete polymers from different sources. Total heat release hc ,s combustion of the pyrolysis gases per unit initial mass of polymer. Low total heat release relative to the heat of complete combustion measured by oxygen bomb calorimetry (see Table 3) indicates that a fraction of the pyrolysis products were not oxidized in the PCFC test (e.g. acid gases or char). The theoretical relationship between heat release capacity and ammability is an active area of research [48,49] that is being driven by empirical data showing that low heat release capacity is a good predictor of ignition resistance and low heat release rate in aming combustion [56]. 5.2. Heats of complete combustion Fig. 8 shows heat release rate history for oxidative pyrolysis of the polycarbonate of bisphenol-A (PC), a polymer that has a single heat release rate peak and forms 2025% char when pyrolyzed under anaerobic conditions. The total oxygen consumed in this experiment is proportional to the net heat of complete combustion of the polymer as determined in an oxygen bomb calorimeter [24,50]. The net heat of complete combustion of polycarbonate calculated from oxygen consumption is the time integrated heat release rate in Fig. 8. Table 3 shows that the net heats of complete combustion measured by PCFC are within 2% of literature values [22,47] on average for the ten polymers listed in Table 3.

R.E. Lyon, R.N. Walters / J. Anal. Appl. Pyrolysis 71 (2004) 2746 Table 2 Heat release capacity, total heat released, and char yield of selected polymers Polymer (ISO abbreviation) Polyethylene (LDPE) Polypropylene (PP) Polyisobutylene (PIB) Polystyrene (PS) Isotactic polystyrene (PS-iso) Polyhexamethylene sebacamide (PA610) Poly-2-vinylnaphthalene (PVN) Polyvinylbutyral (PVB) Polylaurolactam (PA12) Poly -methylstyrene (PMS) Polyhexamethylene dodecanediamide (PA612) Poly(acrylonitrile-butadiene-styrene) (ABS) Bisphenol-A epoxy/phenoxy-A (EP) Polyethyleneoxide (PEOX) Polyhexamethylene adipamide (PA66) Polyvinylalcohol 99% (PVOH) Polycaprolactone (PCL) Dicyclopentadienyl bisphenol cyanate ester Polycaprolactam (PA6) Polybutyleneterephthalate (PBT) Polyethylmethacrylate (PEMA) Polymethylmethacrylate (PMMA) Polyepichlorohydrin Poly-n-butylmethacrylate Poly-2,6-dimethyl-1,4-phenyleneoxide (PPO) Polyisobutylmethacrylate Polyethylmethacrylate (PEMA) Polycarbonate of bisphenol-A (PC) Polysulfone of bisphenol-A (PSU) Polyethyleneterephthalate (PET) Bisphenol E cyanate ester Polyvinylacetate (PVAC) Polyvinylideneuoride (PVDF) Polyethylenenaphthylate (PEN) Poly(p-phenyleneterephthalamide) (KEVLARTM ) Bisphenol A cyanate ester Tetramethylbisphenol F cyanate ester Poly(styrene-maleicanhydride) (SMAH) Epoxy novolac/phenoxy-N (EP) Polynorbornene Bisphenol-M cyanate ester Polychloroprene Polyoxymethylene (POM) Polyacrylic acid (PAA) Poly-1,4-phenylenesulde (PPS) Liquid crystalline polyarylate (LCP) Polyetheretherketone (PEEK) Polyphenylsulphone (PPSU) Polyvinylchloride (PVC) CAS number 9002-88-4 25085-53-4 9003-27-1 9003-53-6 25086-18-4 9008-66-6 28406-56-6 63148-65-2 25030-74-8 52014-31-7 26098-55-5 9003-56-9 001675-54-3 25322-68-3 32131-17-2 9002-89-5 24980-41-4 1355-71-0 25038-54-4 26062-94-2 9003-42-3 9011-14-7 24969-06-0 9003-63-8 25134-01-4 9011-15-8 9003-42-3 24936-68-3 25135-57-7 25038-59-9 47073-92-7 9003-20-7 24937-79-9 24968-11-4 308069-56-9 1156-51-0 101657-77-6 9011-13-6 028064-14-4 25038-76-0 127667-44-1 9010-98-4 9002-81-7 9003-01-4 9016-75-5 70679-92-4 29658-26-2 25839-81-0 9002-86-2 HR capacity (J g1 K1 ) 1676 1571 1002 927 880 878 834 806 743 730 707 669 657 652 615 533 526 493 487 474 470 376-514 443 412 409 406 380 359 345 332 316 313 311 309 302 283 280 279 246 240 239 188 169 165 165 164 155 153 138 Total HR (kJ g1 ) 41.6 41.4 44.4 38.8 39.9 35.7 39 26.9 33.2 35.5 30.8 36.6 26.0 21.6 27.4 21.6 24.4 20.1 28.7 20.3 26.4 23.2 13.4 31.5 20.0 31.3 26.8 16.3 19.4 15.3 14.7 19.2 9.7 16.8 14.8 17.6 17.4 23.3 18.9 21.3 22.5 16.1 14 12.5 17.1 11.1 12.4 11.3 11.3

41

Char (%) 0 0 0 0 0 0 0 0.1 0 0 0 0 3.9 1.7 0 3.3 0 27.1 0 1.5 0 0 4.8 0 25.5 0 0 21.7 28.1 5.1 41.9 1.2 7 18.2 36.1 36.3 35.4 2.2 15.9 6 26.4 12.9 0 6.1 41.6 40.6 46.5 38.4 15.3

42 Table 2 (Continued )

R.E. Lyon, R.N. Walters / J. Anal. Appl. Pyrolysis 71 (2004) 2746

Polymer (ISO abbreviation) Polyetherketone (PEK) Novolac cyanate ester Polyetherimide (PEI) Poly-1,4-phenyleneethersulfone (PESU) Polyacrylamide Polyetherketoneketone (PEKK) Poly(m-phenylene isophthalamide) (NOMEX) Poly-p-phenylenebenzobisoxazole (PBO) Polyimide (PI) Polybenzimidazole (PBI) Polytetrauoroethylene (PTFE) Polyamideimide (PAI) Hexauorobisphenol-A cyanate ester Polyimide (PI) Polyimide (PI) Polyimide (PI)

CAS number 27380-27-4 173452-35-2 61128-46-9 25667-42-9 9003-05-8 74970-25-5 24938-60-1 852-36-8 105030-42-0 25928-81-8 9002-84-0 42955-03-3 32728-27-1 26023-21-2 79062-55-8 87186-94-5

HR capacity (J g1 K1 ) 124 122 121 115 104 96 52 42 38 36 35 33 32 29 14 13

Total HR (kJ g1 ) 10.8 9.9 11.8 11.2 13.3 8.7 11.7 5.4 6.7 8.6 3.7 7.1 2.3 6.6 3.4 2.9

Char (%) 52.9 51.9 49.2 29.3 8.3 60.7 48.4 69.5 57 67.5 0 53.6 55.2 51.9 57 52

Table 3 Net heats of combustion of selected polymers by PCFC and oxygen bomb calorimetry Polymer Net heat of combustion (kJ g1 ) PCFC Polyethylene Polystyrene Polyetheretherketone Phenolic triazine Polycarbonate Poly(p-aramide) Polybutyleneterephthalate Polymethylmethacrylate Polyethyleneterephthalate Polyoxymethlene 44.1 40.1 30.9 29.5 29.1 28.1 26.3 25.0 23.2 15.0 Oxygen bomb 43.3 39.8 30.2 29.8 29.8 27.8 26.7 24.9 21.8 15.9

Fig. 8. Heat release rate for oxidative pyrolysis-combustion of polycarbonate.

R.E. Lyon, R.N. Walters / J. Anal. Appl. Pyrolysis 71 (2004) 2746

43

6. Conclusions Pyrolysis-combustion ow calorimetry is a viable method for determining combustion parameters of materials when only milligram samples are available for testing. Measured quantities include the specic heat release rate under simulated re conditions, the heat of combustion of the fuel gases, the heat of combustion of the solid, and a derived quantity called the heat release capacity. While PCFC is a quantitative technique, the generic polymer heat release capacities in Table 2 are qualitative because of real differences in polymer samples from different sources. Polymer heat release capacities (Table 2) span two orders of magnitude and ongoing work in our laboratory suggests that the heat release capacity is a reliable indicator of re hazard.

Acknowledgements The authors are indebted to Taline Inguilizian, Huiqing Zhang, and Stanisalv Stoliarov of the University of Massachusetts for valuable contributions in testing and analysis. Certain commercial equipment, instruments, materials and companies are identied in this paper in order to adequately specify the experimental procedure. This in no way implies endorsement or recommendation by the Federal Aviation Administration.

Appendix A In the pyrolysis-combustion ow calorimeter (Fig. 1) the oxygen concentration is measured several meters downstream from the point at which the pyrolysis gases are generated, mixed with excess oxygen in the heated manifold, and enter the combustor. As the oxygen depletion prole moves through the instrument mixing and diffusion of the ow stream in the combustion tube and scrubbers tends to broaden and attenuate the peak. These effects must be corrected to synchronize the fuel generation history of the sample with the oxygen consumption history measured at the downstream detector. If the oxygen signal distortion is assumed to be the result of perfect mixing in an idealized dead volume V0 , which may be distributed over the ow path, the mass of oxygen in the mixing volume at any time t is: mO2 = V0 [O2 ] = V0 [O2 ]out (A1) where is the gas density and [O2 ] is the instantaneous volume fraction (concentration) of oxygen in the gas stream which, if V0 is well mixed, is equal to the oxygen concentration leaving V0 and entering the detector, so [O2 ] = [O2 ]out . For a constant volumetric ow F the change in the mass of oxygen in the mixing volume over any time interval, dt, is:
out out mO )dt = (F[O2 ]in F[O2 ]out )dt dmO2 = (mO 2 2

(A2)

in where m in O2 and m O2 are the mass ow rates of oxygen into and out of the well mixed volume V0 . Differentiating Equation A1 with respect to time and equating the result to the

44

R.E. Lyon, R.N. Walters / J. Anal. Appl. Pyrolysis 71 (2004) 2746

differential form of Equation A2:

0 in m O2 m 0 in O2 = F[O2 ] F[O2 ] O2 m

= F([O2 ]0 [O2 ]out ) V0

d[O2 ]out dt

(A3)

O2 is the mass consumption where m 0 O2 is the baseline mass owrate of oxygen and m rate of oxygen relative to the baseline. Making the substitutions, = [O2 ]0 [O2 ]out and = V0 /F, the specic heat release rate in terms of the oxygen concentration at the detector is: d c (t) C m Q O2 (t) = K + dt (A4)

where K = CF (a constant with units of W). Equation A4 synchronizes the oxygen signal with the heat release history of the solid. Inverting Equation A4 with time variable of integration: (t) = 1 K
t

c ()d exp((t )/)Q

(A5)

shows that the oxygen depletion measured at the detector (t) is distorted (convoluted) by an exponential apparatus function. The convolution integral (Equation A5) can be solved exactly for a heat release rate history having an analytic form. An experimentally convenient c (t) = 0 for < t < 0, and, Q c (t) = Q 0 heat release rate history is a step change, i.e. Q c (a constant) for t 0, which when substituted into Equation A5 and solved: (t) = 0 Q c K
t

exp((t )/)u()d =

0 c Q (1 exp((t/))) K

(A6)

0 shows that approaches an equilibrium value Q c /K as t , and that t = when equals 0 63% of this steady-state value for a constant heat ow Q c . This result and the step change heating history is the basis of the calibration procedure for the PCFC (see Fig. 5).

References
[1] V. Babrauskas, R.D. Peacock, Heat release rate: the single most important variable in re hazard, Fire Safety Journal 18 (1992) 255272. [2] C.P. Sarkos, R.G. Hill, Evaluation of aircraft interior panels under full scale cabin re conditions, Proceedings of the AIAA 23rd Aerospace Sciences Meeting, Reno, NV, January 1417, 1985. [3] J.G. Quintiere, V. Babrauskas, L. Cooper, M. Harkelroad, K. Steckler, A. Tewarson, The role of aircraft panel materials in cabin res and their properties, DOT/FAA/CT-84/30, June 1985. [4] V. Babrauskas, From bunsen burner to heat release rate calorimeter, in: V. Babrauskas, S. Grayson (Eds.), Heat Release in Fires, Elsevier, New York, 1992, pp. 729. [5] M. Janssens, Calorimetry, in: SFPE Handbook of Fire Protection Engineering, second ed (Chapter 3.2), Society of Fire Protection Engineers, Boston, MA, 1995 (Chapter 3.2). [6] M.J. Scudamore, P.J. Briggs, F.H. Prager, Cone calorimetrya review of tests carried out on plastics for the association of plastic manufacturers in Europe, Fire and Materials 15 (1991) 6584. [7] V. Babrauskas, Comparative rates of heat release from ve different types of test apparatuses, Journal of Fire Sciences 4 (2) (1986) 148158.

R.E. Lyon, R.N. Walters / J. Anal. Appl. Pyrolysis 71 (2004) 2746

45

[8] T. Kashiwagi, T.G. Cleary, Effects of sample mounting on ammability properties of intumescent polymers, Fire Safety Journal 20 (1993) 203225. [9] A. Tewarson, Generation of heat and chemical compounds in res, SFPE Handbook of Fire Protection Engineering, second ed (Chapter 34), Society of Fire Protection Engineers, Boston, MA, 1995 (Chapter 34). [10] E.M. Pearce, Y.P. Khanna, D. Raucher, Thermal analysis of polymer ammability, in: E.A. Turi (Ed.), Thermal Characterization of Polymeric Materials, Academic Press, Orlando, FL, 1981, pp. 793843. [11] R.L. Hassel, Evaluation of polymer ammability by thermal analysis, American Laboratory 9 (1) (1977) 3547. [12] T.D. Gracik, G.L. Long, Prediction of thermoplastic ammability by thermogravimetry, Thermochimica Acta 212 (1992) 163170. [13] D.W. Van Krevelen, Properties of Polymers, Elsevier, Amsterdam, 1990, pp. 627653. [14] R.E. Lyon, Heat release capacity, Presented at the Fire & Materials Conference, San Francisco, CA, January 2224, 2001. [15] R.A. Susott, F. Shazadeh, T.W. Aanerud, A quantitative thermal analysis technique for combustible gas detection, Journal of Fire and Flammability 10 (1979) 94104. [16] R.A. Susott, Thermal behavior of conifer needle extractives, Forest Sciences 26 (3) (1980) 147160. [17] R.A. Susott, Characterization of the thermal properties of forest fuels by combustible gas analysis, Forest Sciences 28 (2) (1982) 404420. [18] M.L. Janssens, Measuring heat release by oxygen consumption, Fire Technology 27 (1991) 234. [19] M. Janssens, W.J. Parker, Oxygen consumption calorimetry, in: V. Babrauskas, S.J. Grayson (Eds.), Heat Release in Fires (Chapter 3), Elsevier, New York, 1992, pp. 3159 (Chapter 3). [20] Standard test method for heat and visible smoke release rates for materials and products using an oxygen consumption calorimeter, ASTME 1354-90, ASTM Fire Test Standards, third ed., American Society for Testing of Materials, Philadelphia, PA, 1990, pp. 803817. [21] C. Huggett, Estimation of rate of heat release by means of oxygen consumption measurements, Fire and Materials 4 (2) (1980) 61. [22] V. Babrauskas, Heat of combustion and potential heat, in: V. Babrauskas, S.J. Grayson (Eds.), Heat Release in Fires (Chapter 8), Elsevier, New York, 1992, pp. 207232 (Chapter 8). [23] W.M. Thornton, Philosophical Magazine 33 (1917) 196. [24] Standard test method for gross caloric value of coal and coke by the adiabatic bomb calorimeter, ASTMD 2015-85, ASTM Fire Test Standards, third ed., American Society for Testing of Materials, Philadelphia, PA, 1990, pp. 222229. [25] W.J. Parker, Prediction of the heat release rate of wood, Ph.D. Thesis, The George Washington University, April 1988. [26] W.J. Parker, Determination of the input data for a model of the heat release rate of wood, in: J.R. Mchaffey (Ed.), Mathematical Modeling of Fires, ASTM STP 983, American Society for Testing and Materials, Philadelphia, PA, 1987, pp. 105115. [27] V. Babrauskas, W.J. Parker, G. Mulholland, H. Twilley, The phi meter: a simple, fuel-independent instrument for monitoring combustion equivalence ratio, The Review of Scientic Instruments 65 (7) (1994) 23672375. [28] S.M. Reshetnikov, I.S. Reshetnikov, Polymer Degradation and Stability 64 (1999) 379385. [29] S.M. Reshetnikov, G.A. Prevozchikov, USSR 285663, 01/02/1998. [30] D. Price, F. Gao, G.J. Milnes, B. Eling, C.I. Lindsay, P.T. McGrail, Laser pyrolysis/time-of-ight mass spectrometry studies pertinent to the behavior of ame retarded polymers in real re situations, Polymer Degradation and Stability 64 (1999) 403410. [31] S.M. Angel, In situ ame chemistry by remote spectroscopy, in: R.E. Lyon (Ed.), Fire Resistant Materials: Progress Report (DOT/FAA/AR-97/100), 1998 (DOT/FAA/AR-97/100). [32] W.W. Wendlandt, Thermal Methods of Analysis, second ed (Chapter V.C), Wiley, New York, 1974 (Chapter V.C). [33] R.E. Lyon, An integral method of nonisothermal kinetic analysis, Thermochimica Acta 297 (1997) 117124. [34] B. Wunderlich, in: E.A. Turi (Ed.), Thermal Characterization of Polymeric Materials (Chapter 2.D), Academic Press, New York, NY, 1981, p. 112 (Chapter 2.D). [35] J. Flynn, L.A. Wall, General treatment of the thermogravimetry of polymers, Journal of Research of the National Bureau of StandardsA. Physics and Chemistry 70A (6) (1966) 487.

46

R.E. Lyon, R.N. Walters / J. Anal. Appl. Pyrolysis 71 (2004) 2746

[36] R.E. Lyon, Pyrolysis kinetics of char forming polymers, Polymer Degradation and Stability 61 (2) (1998) 201210. [37] T.D. Gracik, G.L. Long, Heat release and ammability of a small specimen using thermoanalytical techniques, in: R.E. Lyon, M.M. Hirschler (Eds.), Fire Calorimetry, Proceedings from the 50th Calorimetry Conference, Gaithersburg, MD, DOT/FAA/CT-95/46, 1995. [38] T.D. Gracik, G.L. Long, U.A.K. Sorathia, H.E. Douglas, A novel thermogravimetric technique for determining ammability characteristics of polymeric materials, Thermochimica Acta 212 (1992) 209217. [39] A. Tewarson, R.F. Pion, Flammability of plastics-I. Burning intensity, Combustion and Flame 26 (1976) 85103. [40] A. Tewarson, Experimental evaluation of ammability parameters of polymeric materials, in: M. Lewin, S.M. Atlas, E.M. Pearch (Eds.), Flame Retardant Polymeric Materials (Chapter 3), Plenum Press, New York, 1982, pp. 97153 (Chapter 3). [41] E.W. Kifer, L.H. Leiner, Thermal evolution analysis of some organic materials, in: Analytical Calorimetry, Plenum Press, New York, 1974, pp. 199205. [42] R.N. Walters, R.E. Lyon, A microscale combustion calorimeter for determining ammability parameters of materials, presented at NIST Annual Conference on Fire Research, Gaithersburg, MD, October 30, 1996. [43] R.N. Walters, R.E. Lyon, A microscale heat release rate device, presented at Society of Plastics Engineers Annual Technical Conference (ANTEC 96), Indianapolis, IA, May 59, 1996. [44] T.V. Inguilizian, Correlating polymer ammability using measured pyrolysis kinetics, Masters Thesis, University of Massachusetts (Amherst), January 1999. [45] N. Grassie, Products of thermal degradation of polymers, in: J. Brandrup, E.H. Immergut (Eds.), Polymer Handbook third ed (Chapter II), Wiley, New York, 1989, pp. 365386 (Chapter II). [46] C.F. Cullis, M.M. Hirschler, The Combustion of Organic Polymers, Claredon Press, Oxford, England, 1981. [47] R.N. Walters, S.M. Hackett, R.E. Lyon, Heats of combustion of high temperature polymers, Fire and Materials 24 (5) (2000) 245252. [48] R.E. Lyon, Solid state thermochemistry of aming combustion, in: A.F. Grand, C.A. Wilkie (Eds.), Fire Retardancy of Polymeric Materials, Marcel Dekker, New York, 2000, pp. 391447. [49] R.E. Lyon, Heat release kinetics, Fire and Materials 24 (2000) 179186. [50] Standard test method for heat of combustion of hydrocarbon fuels by bomb calorimeter (High Precision Method), ASTM-D238, American Society for Testing of Materials, Philadelphia, PA, 1988. [51] T.B. Brill, P.J. Brush, K.J. James, J.E. Shepherd, K.J. Pfeiffer, T-Jump/FT-IR spectroscopy: a new entry into the rapid, isothermal pyrolysis chemistry of solids and liquids, Applied Spectroscopy 46 (6) (1992) 900911. [52] J.C. Jones, W.O. Stacey, Use of the pyroprobe in pyrolysis experiments, Fuel 65 (1986) 454. [53] W.M. Hefngton, G.E. Parks, K.G.P. Sulzmann, S.S. Penner, Studies of methane oxidation kinetics, in: Proceedings of the 16th Symposium (International) on Combustion, Combustion Institute, 1976, pp. 9971010. [54] Themoplastics and thermosets: ammability data, Plastics Digest, 17 ed., vol. 1, D.A.T.A. Business Publishing, Englewood, CO, 1996, pp. 773889. [55] H.Q. Zhang, P.R. Westmoreland, R.J. Farris, Understanding the re behavior of polymers, presented at the Cooperative University of Massachusetts-Industry Research on Polymers Meeting, Fire Safe Polymers and Polymer Composites Cluster, Amherst, MA, May 11, 2000. [56] R.N. Walters, R.E. Lyon, Molar group contributions to polymer ammability, Journal of Applied Polymers and Science 87 (3) (2003) 548563.