Thermal Decomposition Kinetics of PBAN-Binder and Composite Solid Rocket Propellants

Center for Thermal Analysis, Department of Chemistry, University of Utah, Salt Lake City, Utah 84112 Thermogravimetric analysis at heating rates between 0.5 and 10C min1 has been used to study the decomposition kinetics of the PBAN binder and three propellants based on ammonium perchlorate (AP) with PBAN, HTPB, or BAMO-AMMO binders. Thermal runaway was observed for the PBAN and HTPB propellants at heating rates faster than 4.5 and 3C min1, respectively. The multistep decomposition kinetics were analyzed by an advanced isoconversional method that showed that the effective activation varies with the extent of decomposition of the studied systems. For PBAN the activation energy increases from 100 to 200 kJ mol1 throughout the polymer decomposition. In propellants the major mass loss step primarily relates to AP decomposition accompanied by some degradation of the polymer binder. For the PBAN and HTPB propellants this step shows respective increases in the activation energy from 60 to 180 kJ mol1 and from 100 to 230 kJ mol1. The major decomposition step of the BAMO-AMMO propellant has a fairly constant activation energy of 120 kJ mol1. 1999 by The Combustion Institute

THORSTEN SELL, SERGEY VYAZOVKIN, and CHARLES A. WIGHT*

LIST OF ABBREVIATIONS Al aluminum powder (particle size) AP ammonium perchlorate (particle size) BAMO-AMMO Bis-azidomethyloxetane/ azidomethyl-methylazidocopolymer Desmodur N-100 crosslinking agent, hexane 1-6 diisocyanate homopolymer (C8H12N2O2)x DOA plasticizer, dioctyladipate or bis (2 ethylhexyl)adipate GAP glycidyl azide polymer HTPB hydroxy terminated polybutadiene PBAN polybutadiene-acrylic acidacrylo nitrile terpolymer INTRODUCTION The thermal decomposition of solid rocket propellants is a complex process that may involve various chemical (solid- and gas-phase reactions as well as reactions of gaseous products with the solid) and physical (phase transitions, diffusion, adsorption, desorption) phenomena. In a rocket
*Corresponding author. Email: wight@chemistry.chem. utah.edu 0010-2180/99/$see front matter PII S0010-2180(99)00036-X

motor, extreme conditions are predominant leading to temperatures of 2000 3500 K, pressures of 510 MPa, and estimated heating rates as high as 106 K s1. To simulate these conditions, pyrolysis studies of solid rocket propellants and polymeric binders should be conducted at heating rates greater than 100C s1 [13]. Unfortunately the extracted kinetic information is limited because of the high uncertainty in measuring the temperature and reaction rate. Slow heating rate experiments allow one to get a better insight into the reaction kinetics and mechanisms. However, the relevance of these data to the fast heating rate and high temperature conditions of combustion is not obvious [4]. The slow heating rate data may happen to be irrelevant if the reaction mechanism changes with temperature. However, the difference in the temperature regions is not the sufcient condition of the change in the reaction mechanism. For instance, Vyazovkin and Wight [5] studied thermal decomposition of ammonium dinitramide at heating rates of 102 to 101 C s1 by using regular thermal analysis techniques and at a heating rate of 107 C s1 by using laser pyrolysis technique. Both techniques have shown formation of the same gaseous products that suggests that the decomposition mechanism does not change in spite of the dramatic change in the temperature. In such a situation, the reaction kinetics and mechanisms derived from the slow heating rate experiments
COMBUSTION AND FLAME 119:174 181 (1999) 1999 by The Combustion Institute Published by Elsevier Science Inc.

DECOMPOSITION OF PBAN AND ROCKET PROPELLANTS

TABLE 1 List of Ingredients Used in the Composite Propellants Tested Ingredient BAMO-AMMO: BAMO-AMMO polymer GAP plasticizer AP (200 m) AP (20 m) Al (5 m) Desmodur N-100 Triphenyl tin chloride Weight % 7.58 11.99 49.60 12.40 18.00 0.42 0.01 Ingredient PBAN-AP: PBAN AP (2 m) HTPB-DOA-AP: HTPB (R45M) DOA AP (2 m)

175

Weight % 29.76 70.24

23.94 5.81 70.25

can be used to predict decompositions under conditions of combustion [6]. However, reliable predictions cannot be made without extracting reliable kinetic information from slow heating rate data. This study focuses on obtaining reliable kinetic information on the thermal decomposition PBAN and three different solid propellants (Table 1). Experimental Section Samples of PBAN, PBAN-AP, HTPB-DOAAP, and BAMO-AMMO propellants were kindly supplied by the Thiokol Corp. The chemical composition of the propellants is shown in Table 1. The thermogravimetric analysis (TGA) experiments were carried out using a Rheometrics 1000M thermobalance. To reduce thermal gradients and self-heating the experiments were performed on small samples (PBAN 2.9 mg, PBAN-AP 0.8 mg, HTPB-DOA-AP 0.9 mg, BAMO-AMMO 1.0 mg). Samples of the propellants were placed in open aluminum pans and heated in a owing atmosphere of nitrogen (100 ml min1). The instrument was programmed for heating at constant rates from room temperature to 590C. Kinetic Analysis of TGA Data The kinetics of heterogeneous decompositions of solids are customarily described by the basic kinetic equation d kT f dt (1)

f ( ) is the reaction model, which describes the dependence of the reaction rate on the extent of reaction. The value of is experimentally derived from the global mass loss in TGA experiments. In most cases the temperature dependence of k ( T ) can be satisfactorily described by the Arrhenius equation, whose substitution into Eq. 1 yields E d A exp f dt RT

(2)

where represents the extent of reaction (0 1), t is time, k ( T ) is the rate constant, and

where E is the activation energy and A is the preexponential factor. To evaluate E and A in Eq. 2, one has to separate the temperature and conversion dependencies of the reaction rate. Here, a commonplace approach is force tting of experimental data to assumed reaction models. When applied to decomposition data obtained under nonisothermal conditions, the model-tting method gives highly uncertain values of the Arrhenius parameters [7]. Another major aw of this approach is that usually the model-tting methods produce a single pair of Arrhenius parameters for the whole process. However, the thermal decompositions of solid materials are known to involve multiple steps that are likely to have different activation energies. Then, the contributions of these steps into the overall decomposition rate measured by TGA should vary with both T and . This means that the effective activation energy determined from TGA experiments will also be a function of these two variables. The model-tting methods are simply incapable of accounting for this degree of complexity. The above-mentioned problems can be avoided by using model-free isoconversional methods

176 [8 10]. These methods allow the activation energy to be evaluated without making any assumptions about the reaction model. Additionally, the isoconversional methods evaluate the effective activation energy as a function of the extent of conversion, which allows one to explore multistep kinetics [11]. Vyazovkin developed an advanced isoconversional method [12], which is based on the assumption that the reaction model is independent of the heating program, T ( t ). According to this method, for a set of n experiments carried out at different heating programs, the activation energy is determined at any particular value of by nding the value of E that minimizes the function E

T. SELL ET AL.

Fig. 1. TGA curves for decomposition of PBAN.

i1 j i

n n

J E , T i t J E , T j t

(3)

Henceforth, the subscript denotes the values related to a given extent of conversion. In Eq. 3, the integral J E , T i t

exp

E dt RT i t

(4)

is evaluated numerically for a set of experimentally recorded heating program, T i ( t ). The minimization procedure is repeated for each value of to nd the dependence of the activation energy on the extent of conversion. An advantage of the advanced isoconversional method is that it can be applied to study the kinetics under arbitrary temperature programs (e.g., under a linear heating program distorted by self-heating) [12]. To our knowledge, this is the rst work to apply such a method to thermal decomposition of composite rocket propellants. RESULTS PBAN Polymer Binder Thermal decomposition experiments on the neat PBAN binder were performed at constant heating rates of 0.9, 1.8, 2.7, 4.6, 7.3, and 9.1 C min1. The representative TGA curves are shown in Fig. 1. The PBAN polymer shows three major steps of mass loss. The steps are strongly overlapped. The rst step covers a

temperature region from 110 to 220 C and involves a mass loss of 8%. The second step demonstrates a mass loss of 15% in the temperature region of 220 380 C. The thermal decomposition noticeably accelerates in the third step, in which the sample loses 75% of mass in a temperature region of 380 500 C. After decomposition we found a small amount of char that accounted for 1% of the initial mass. There is also a minor step of the mass loss (2%) with the maximum rate at 100C. This step is hardly noticeable at faster heating rates. Figure 2 compares the E -dependence against the experimentally observed mass loss curve. The three decomposition steps are associated with variations in the E -dependence.

Fig. 2. TGA curve for decomposition of PBAN at heating rate of 0.9C min1 (upper trace) compared to the dependence of the activation energy on the extent of decomposition conversion (lower trace). Open symbols represent literature data [13].

DECOMPOSITION OF PBAN AND ROCKET PROPELLANTS

177

Fig. 3. TGA curves for decomposition of PBAN-AP composite propellant.

The rst step ( 0.08) is characterized by a small activation energy of 50 70 kJ mol1. The second step ( 0.08 0.25) is characterized by an increase in E from 100 to 260 40 kJ mol1. In the third step ( 0.25 0.99) the activation energy rapidly decreases from 260 40 kJ mol1 to a practically constant value of 200 30 kJ mol1. PBAN-AP Composite Propellant The composite PBAN-AP propellant was studied at constant heating rates of 0.5, 0.9, 2.3, 3.2, 4.1, 4.5, 7.3, 8.9, and 9.0 C min1. The TGA experiments showed (Fig. 3) thermal runaway at the heating rate of 4.5C min1 and faster. At slower heating rates the TGA curves show two major steps of mass loss. As in the case of neat PBAN, we can also see a minor mass loss (2%) with maximum rate at 100C. It precedes the rst major step that shows a fast mass loss of around 70% in the temperature region of 100 350C. The second step occurs above 350C and demonstrates a slow mass loss of 15%. The slower heating rate (4.5C min1) decompositions resulted in formation of a char residue that accounted for 15% of the initial mass. The E -dependence was evaluated for the slow heating rate decompositions, which do not undergo thermal runaway. The results are presented in Fig. 4. The major decomposition step ( 0.8) shows a rise in the activation energy from 20 to 180 kJ mol1. The slow second step ( 0.8) starts with an abrupt drop in the

Fig. 4. TGA curve for decomposition of PBAN-AP at heating rate of 0.9C min1 (upper trace) compared to the dependence of the activation energy on extent of decomposition (lower trace).

activation energy. The average activation energy for this step is around 40 kJ mol1. HTPB-DOA-AP Composite Propellant The thermal decomposition of HTPB-DOA-AP composite propellant was examined at constant heating rates of 0.5, 0.9, 1.5, 1.8, 2.3, 2.5, 2.7, 3.2, 3.6, 4.5, 6.8, and 9.0 C min1. Thermal runaway was observed at the heating rates faster than 3 C min1 (Fig. 5). At slower heating rates, the TGA curves show 3 major steps of the mass loss. During the rst step the propellant loses 8% of its mass in the temperature range of 100 230C. The second step demonstrates a fast mass loss of 70% at temperatures between 230 and 380

Fig. 5. TGA curves for decomposition of HTPB-DOA-AP composite propellant.

178

T. SELL ET AL.

Fig. 6. TGA curve for decomposition of HTPB-DOA-APpropellant at heating rate of 0.9C min1 (upper trace) compared to the dependence of the activation energy on extent of decomposition (lower trace).

Fig. 7. TGA curves for decomposition of BAMO-AMMO composite propellant.

C. The third step (T 380C) shows a very slow decomposition accompanied by 5% of loss in mass. When the temperature reached the nal value of 590C, the samples had less than 10% of the initial mass remaining. The E -dependence was determined for decompositions conducted at heating rates slower than 3C min1. The results are shown in Fig. 6. The rst step of decomposition ( 0.08) exhibits an activation energy of about 100 kJ mol1. The activation energy suddenly drops at the beginning of the second step (0.08 0.9), which shows an increase in E from 100 to 230 kJ mol1. The third step exhibits an abrupt change of the activation energy to very low values that could not be reliably evaluated. BAMO-AMMO Composite Propellant The BAMO-AMMO composite propellant was decomposed at constant heating rates of 0.9, 2.3, 4.5, 7.2, and 9.0 C min1. The results are presented in Fig. 7. The TGA curves show three major steps of the mass loss. As in previous cases, the rst step is preceded by a minor mass loss with the maximum rate around 100C. During the rst step the propellant loses 15% of its mass in the temperature range 150 250 C. In the second step ( T 250320 C), the propellant loses more than 50% of its mass. The third step ( T 320C) is a slow decomposition

accompanied by a small mass loss of 4%. By 590C about 4% of the propellant mass remained. The thermal decomposition of this propellant is characterized by the E -dependence shown in Fig. 8. The rst stage of decomposition shows an increase of the activation energy from 100 to 160 kJ mol1 at 0.2. At the beginning of the second step the activation energy suddenly drops to 80 kJ mol1. The activation energy for the second step ( 0.2 0.9) is practically constant (120 20 kJ mol1) at 0.3. The transition to the third step of decomposition is characterized by some increase in activation energy that is followed by the rapid drop at

Fig. 8. TGA curve for decomposition of BAMO-AMMO composite propellant at heating rate of 7.2C min1 (upper trace) compared to the dependence of activation energy on the extent of decomposition (lower trace).

DECOMPOSITION OF PBAN AND ROCKET PROPELLANTS 0.95. The third step has an activation energy of 20 kJ mol1. DISCUSSION The neat PBAN as well as PBAN-AP and BAMO-AMMO propellants showed a minor initial mass loss (2%). This minor step is absent in the thermal decomposition of HTPBDOA-AP propellant. The values of the activation energy related to that step are too small (20 50 kJ mol1) for a chemical reaction (bond breaking). Based on this fact and on the proximity of the maximum rate of mass loss to 100C, we assume that this initial step is associated with vaporization moisture from the above systems. As was noticed, the evaluated E -dependencies sometimes show abrupt drops in the activation energy. We feel that generally they represent a computational artifact. Note that these drops occur at the point of transition from one mass loss step to another. The reaction rate at this point markedly decreases and does not vary much with temperature. The low temperature sensitivity of the reaction rate manifests itself in a small value of the activation energy. PBAN Polymer Binder The rst mass loss step ( 0.08) shows the activation energy (50 60 kJ mol1) that is uncharacteristically small for the thermal decompositions. This step may result from vaporization of a residual organic solvent or some other species of low molecular mass. The second decomposition step ( 0.08 0.25) demonstrates an increasing E -dependence. It suggests that this step is likely to involve two decomposition pathways [11]. With increasing the temperature, the pathway having a smaller activation energy (100 kJ mol1) is taken over by the pathway that has a greater activation energy (200 kJ mol1). Given the fact that the third decomposition step ( 0.25 0.99) is strongly overlapped with the second one and that the activation energy for the third step is 200 30 kJ mol1, it is reasonable to assume that the third and second steps may involve a common reaction pathway. Therefore

179

the overall kinetics of PBAN decomposition appears to be determined by two overlapping processes with the respective activation energies of 100 and 200 kJ mol1. Rao and Radhakrishnan [13] studied the thermal decomposition of cured PBAN by using TGA at heating rates between 2 and 50 C min1. The use of the Ozawa method [9] resulted in a E -dependence (Fig. 2) that has a similar shape as that obtained by us. The somewhat higher values of the activation energy may be explained by the effect of curing that is likely to result in formation of a more stable polymer network. However, it may also be due to the limitations of Ozawas method, which uses an oversimplied approximation for the temperature integral (4) that holds only for ideally linear heating program and cannot account for appreciable self-heating. Cohen et al. [3] used Xe lamp radiation to surface pyrolyze PBAN. By plotting the logarithm of the mass loss ux against the reciprocal surface temperature, they found the activation energy of 70 kJ mol1. The validity of this value is obviously limited by an arbitrary assumption of zero-order kinetics. The value is too small to represent a chemically controlled decomposition. Besides, Cohen et al. [3] did not report the extent of pyrolysis reached in their experiments. This is a crucial factor because according to the present results the activation energy varies with the extent of PBAN degradation. Overall, we cannot meaningfully compare our results with those obtained by Cohen et al. [3]. This also applies to the data of Rao and Radhakrishnan [14] who measured the total amount of C2 hydrocarbons released from PBAN, which was pyrolyzed at different temperatures. Under assumption of rst-order kinetics, they obtained the activation energy of 52 kJ mol1. PBAN-AP Composite Propellant The thermal decomposition of the PBAN-AP propellant does not show the mass loss between 110 to 220 C that we observed for the neat PBAN binder. The major mass loss step starts at 200C and accounts for about 80% of the mass loss. It should be kept in mind that the PBAN-AP propellant is mostly (70%) composed of AP. When heated at 5C min1, neat

180 AP completely decomposes in the temperature region of 240 350C [7]. At this heating rate, neat PBAN decomposes at the temperatures between 200 and 480 C. However, by 350C neat PBAN loses only about 10% of its mass (Fig. 1). We may therefore expect that the rst step of the propellant decomposition represents the decomposition of all AP and about onethird of PBAN. The E -dependence determined for the rst and major step of the propellant decomposition (Fig. 4) is suggestive of a complex reaction mechanism. Note that this dependence shows an increase in the activation energy that is similar to that observed for the PBAN decomposition at 0.25 (Fig. 2). However, the maximum value of the activation energy for the propellant decomposition is lower (180 kJ mol1) than that for the decomposition of neat PBAN (260 kJ mol1). On the other hand, the maximum activation energy for the propellant decomposition is markedly higher than that for decomposition of neat AP. According to Vyazovkin and Wight [7] the thermal decomposition of AP is characterized by a complex E dependence for which the maximum value of the activation energy does not exceed 120 kJ mol1. Therefore, the maximum activation energy of the PBAN propellant decomposition is likely to characterize the reaction between PBAN and decomposition products of AP. Taking into account that the PBAN propellant begins to decompose at the temperature that is 40C lower than the decomposition temperature of neat AP, we should ascribe the initial smaller activation energy (80 kJ mol1 at 200C) to the thermal decomposition of PBAN. Therefore, the overall kinetics of the rst and major decomposition step can be reduced to two parallel pathways, which are decomposition of individual PBAN and reaction of PBAN with decomposition products of AP. The slow rate and low activation energy for the second decomposition step ( 0.8) suggest that further decomposition may be limited by diffusion [11]. HTPB-DOA-AP Composite Propellant The rst decomposition step ( 0.05) has the maximum rate at 150C. The corresponding activation energy is about 100 kJ mol1. Note

T. SELL ET AL. that the mass loss is practically equal to the content of the DOA plasticizer in the propellant. It does not seem to be unreasonable to assume that the rst step relates to the vaporization or decomposition of the plasticizer. We cannot, however, rule out that the HTPB may also contribute to this rst step. Unfortunately, pure HTPB was not available for this study to clarify this situation. The temperature region of the second and major mass loss step (0.1 0.9) practically coincides with the temperature region of AP decomposition [7]. During this step the propellant loses 80% of its mass. Based on these facts we may conclude that this step involves the thermal decomposition of all AP as well as of 10% of HTPB. The activation energy for this step increases from 100 to 280 kJ mol1. Because the average activation energy for decomposition of the propellant is signicantly greater than that for decomposition of pure AP [7], we may expect that the overall kinetics of this step is determined by decomposition of HTPB or/and by its reaction with the products of AP decomposition. The low rate and small activation energy (we were unable to reliably determine its value) of the third step suggest that the decomposition of the residual polymer is likely to be diffusion-controlled. Ninan and Krishnan [15] used TGA to study the kinetics of thermal decomposition of HTPB binder at heating rates of 1100 C min1. To evaluate Arrhenius parameters, they tted data to the reaction-order model. The reported values [15] of the activation energy correspond to a reaction of zero order and randomly vary from 80 to 150 kJ mol1 depending on the heating rate as well as on the computational method used. These values are noticeably smaller than those found by us. However, the reliability of the reported values [15] is rather questionable because of the awed nature of the modeltting methods [7]. Cohen et al. [3] studied the kinetics of the surface pyrolysis of HTPB. Assuming zero-order kinetics, they found the activation energy of 71 kJ mol1. Rao and Radhakrishnan [14] measured the total amount of C2 hydrocarbons that was released from pyrolyzed HTPB at different temperatures. By assuming rst-order kinetics for this process, they obtained the activation

DECOMPOSITION OF PBAN AND ROCKET PROPELLANTS energy of 53 kJ mol1. We have already questioned the validity of such values in our discussion of PBAN data. BAMO-AMMO Composite Propellant Interpretation of the overall kinetic data for this propellant is extremely difcult. The propellant contains seven components, including two polymers (BAMO-AMMO and GAP), two sizes of AP, aluminum, and curing agents. The overall kinetics for this propellant may be determined by decomposition of individual polymers and AP, as well as by the reaction of the polymers with decomposition products of AP. Additionally, the overall decomposition kinetics may be complicated by curing of the polymers. Although the nature of the mass loss step is unclear, the second step occurs in the same temperature region as the temperature decomposition of neat AP [7]. This step represents the major mass loss of 70%. Since this amount exceeds the overall amount of AP in the propellant, we may conclude that the thermal decomposition of AP is accompanied by decomposition of the polymers. The activation energy for the second step remains fairly constant at 120 20 kJ mol1 (Fig. 8). The constancy of the effective activation energy suggests that the overall kinetics of this step is limited by a single reaction. For the thermal decomposition of neat AP the activation energy varies within the limits of 80 120 kJ mol1 [7]. Therefore, the Edependence (Fig. 8) observed for the propellant decomposition is likely to represent the reaction of AP decomposition products with the polymer binder. The third decomposition step ( 0.95) has a low activation energy and reaction rate that are characteristic of diffusion control. CONCLUSIONS As a result of this study we found that the thermal decompositions of the PBAN binder and PBAN, HTPB, and BAMO-AMMO propellants follow multistep kinetics. The use of the advanced isoconversional method allows one to

181

observe variation of the effective activation energy with the extent of decomposition. The major mass loss step for decomposition of propellants is the thermal decomposition of AP accompanied by partial degradation of the polymer binder. Comparison of the activation energies for this process with the activation energies for decomposition of individual AP or/and binder suggests that the overall kinetics of the major mass loss is determined by the reaction between binder and decomposition products of AP. This research is supported by the Ballistic Missile Defense Organization and the Ofce of Naval Research under MURI Contract No. N00014-951-1339. T.S. has been supported by a scholarship from Deutsche Akademischer Austauschdienst (DAAD). REFERENCES
1. 2. 3. 4. 5. 6. Arisawa, H., and Brill, T. B., Combust. Flame 106:144 (1996). Arisawa, H., and Brill, T. B., Combust. Flame 106:131 (1996). Cohen, N. S., Fleming, R. W., and Derr, R. L., AIAA J. 12:212 (1974). Price, E. W., J. Spacecraft Rockets 20:320 (1983). Vyazovkin, S., and Wight, C. A., J. Phys. Chem. A 101:7217 (1997). Vyazovkin, S. V., Levchik, G. F., Goryachko, V. I., Vyazovkina, A. I., and Lesnikovich, A. I., Thermochim. Acta 215:315 (1993). Vyazovkin, S., and Wight, C. A., Int. Rev. Phys. Chem. 17:407 (1998). Friedman, H., J. Polym. Sci. C 6:183195 (1964 65). Ozawa, T., Bull. Chem. Soc. Japan 38:1881 (1965). Flynn, J. H., and Wall, L. A., J. Res. Nat. Bur. Standards 70A:487 (1966). Vyazovkin, S., Int. J. Chem. Kinet., 28:95 (1996). Vyazovkin, S., J. Comput. Chem. 18:393 (1997). Rao, M. R., and Radhakrishnan, T. S., J. Appl. Polym. Sci. 41:2251 (1990). Rao, M. R., and Radhakrishnan, T. S., Propellants, Explosives, Pyrotechnics 20:32 (1995). Ninan, K. N., and Krishnan, K., J. Spacecraft Rockets 19:92 (1982).

7. 8. 9. 10. 11. 12. 13. 14. 15.

Received 5 August 1998; revised 25 February 1999; accepted 10 March 1999