Macromolecules zyxwvu 1980,13, 1163-1167 Kinetics of Thermal Degradation of zyx Poly[bis(2,2,2-trifluoroethoxy)phosphazene] Martel Zeldin,* Won Ho Jo,+ and Eli M. Pearce Polytechnic Institute zyxwvu of New York, Brooklyn, New York 11201. Received May zyx 14, 1979 1163 ABSTRACT Poly[bis(2,2,2-trifluoroethoxy)phosphazene] was depolymerized at 350 "C under vacuum. Volatile products were characterizedby infrared, 31P NMR, and gas chromatographiechemical ionization mass spectral techniques; the predominant product was the cyclic trimer. A random scission-partial unzipping mechanism was supported by fitting molecular weight change-weight loss data to theoretical curves. An average zip length of 35 chain units was found. An activation energy of 43 kcal/mol was observed from isothermal TGA data at different temperatures. A similar value was obtained from dynamic TGA at different heating rates. A reaction order of 0.8 was found and interpreted in terms of quasi-first-order kinetics. Although synthetic techniques for the preparation of polyphosphazenes are well established and numerous studies of their physical and dynamic-mechanical prop- erties have been reported,'-12 the mechanism of thermal degradation of this class of substance is still unresolved. MacCallum and Tanner13 have shown that pyrolysis of polyphosphazenes leads to a mixture of ring and low mo- lecular weight chain oligomers. Allcock et al.14J5 analyzed the decomposition products of polyphosphazenes by gel permeation chromatography (GPC), solution viscosity, and mass spectrometry. Their data on poly[dialkoxy- phosphazenes] qualitatively support both a random cleavage and unzipping mechanism. However, a depo- lymerization process involving chain-end initiation followed by unzipping or other mechanisms could not be precluded. Recently, Valaitis and Kyker16reported a study of the thermal degradation of a polyphosphazene containing a mixture of trifluoroethoxy and (octafluoropentany1)oxy substituents by isothermal TGA and GPC. The data were interpreted in terms of a random polymer scission process; however, differentiation between this mechanism and one involving continuously random scission and the determi- nation of certain important kinetic parameters (e.g., av- erage zip length, reaction order) by application of theo- retical treatments were not accomplished. Furthermore, the presence of oligomers and multiple substituents in the polymer make evaluation of activation parameters difficult to determine and interpret. In the present study we report the kinetics of thermal decomposition of poly[bis(2,2,2-trifluoroethoxy)phos- phazene] by following molecular weight changes with re- action conversion. The experimental data are fitted to equations derived by MacCallum" and support a mecha- nism involving a random chain scission process followed by a partial unzipping of the fragments to give cyclic ol- igomers. In addition to the calculation of the average zip length, the reaction order, which is implied from curve fitting, has been determined experimentally. Additionally, the activation energy has been obtained from both iso- thermal and dynamic thermogravimetric analysis. Theory Kinetic equations based on MacCallum's theoretical treatment for thermal depolymerization are given below. According to theory, depolymerization mechanisms can be divided into two groups categorized by the nature of the initiation reaction, Le., random scission along the polymer molecule backbone or initiation at the polymer chain ends. Each category can be further divided into (a) initiation In partial fulfillment of the requirements for the Ph.D. degree in Polymer Science and Engineering at the Polytechnic Institute of New York. 0024-9297/80/2213-1163$01.00/0 followed by partial unzipping and (b) initiation followed by complete unzipping. Relationship 1 is generally applied for any polymer degradation, where W is the weight of the polymer at time t, zyxw N is the number of molecules in the sample at time t, Dp is the number-average degree of polymerization of the sample at time t, and m is the re- peat-group molecular weight. Differentiation of (1) with respect to t gives W zyxw = N(&)m ( 1) - 1 dW d(DP) - CW =N- + Dp - m dt dt dt -- For random initiation followed by an incomplete un- zipping process and assuming that the depolymerization follows first-order kinetics with respect to sample weight dN/dt = k W/m (3) -d W/dt = k WZ (4) where 2 is defined as the zip length. Combination of (1)-(4) followed by integration gives zyx M = f(1- C)/(C + f, (5) where M = Dp/Dpo, f = Z/Go, and C is the fractional conversion, ( Wo - W)/ W,. For random initiation followed by complete unzipping, 2 is defined as some multiple of & (eq 6), where b is a Z = b(&) (6) parameter related to the polydispersity of the polymer. Combining (3), (4), and (6) and integrating (2) gives (7) 1 - c = M(b/b-l) For terminal initiation followed by partial unzipping -dN/dt = 0 (8) M=l-C (9) Finally, for terminal initiation followed by complete unzipping -cW/dt = kN (10) Combining (l), (21, (4), and (10) and integrating gives M=l (11) Representative theoretical curves from (5), (7), (9), and (11) are summarized in Figure 1 (insert), which is a plot of the fractional number-average molecular weight as a function of fractional conversion. Line ACG corresponds to degradation by a chain-end initiation followed by a complete unzipping process (eq 11). Since initiation is the rate-determining step, no change in molecular weight with -- From (2), (4), and (8) and integration one gets 0 1980 American Chemical Society