IN SEARCH OF A MECHANISM FOR CYANATE CURE KINETICS F. LÓPEZ-SERRANO † Dpto. de Ing. Química, F. Q., Univ. Nacional Autónoma de México, CP 04510. MÉXICO †, J.E. PUIG‡, E. MENDIZÁBAL‡ ‡ Dptos. de Ing. Química y Química, CUCEI, Univ. de Guadalajara, CP 09340. MÉXICO lopezserrano@unam.mx ; puig@mail.udg.mx ; lalomendizabal@hotmail.com INTRODUCTION This is a continuation of a work presented recently (López-Serrano et al., 2010) and the purpose is to continue analyzing conversion-against-time data (Zhao and Hu, 2007), of NCOCH2 (CF2) 6 CH2 OCN(2,2,3,3,4,4, 5,5,6,6,7,7 dodecafluorooctanediol dicyanate ester) at temperatures from 140 to 190°C, in order to search for an apparent reaction mechanism to describe the reaction. Cyanate esters are important because they belong to a new thermosetting resins generation, attractive to many current polymer applications (e.g. microelectronics, aerospace, etc). Despite the fact that the resin and its technology are more than 20 years old, fundamental aspects related to curing, kinetics and modeling, still evoke great interest and new hypotheses continue to emerge (Nair et al., 2000). PROPOSED APPROACH Consider the reaction scheme: x, . = (x)(1-x) (x) , x(0) = 0; y = x k = (x), n = (x) (1b) where x is the fractional conversion, y its measurement, (1a) x, . is the time derivative of x, and (x) [or (x)] is the function which describes the dependency of the apparent rate constant (s -1 ) (or reaction rate order) on x. Following the integro-differential (ID) approach (López- Serrano et al., 2007), one fits the experimental data with the smooth analytic curve y(t), to obtain y, . (t) and y, .. n(t) = -{ [1-y(t)] (t) take the derivatives of Eq. 1a with [(x), (x)]  constant (drift assumption in estimation theory), and solve for the resulting equation pair (k, n) time- evolution: y, .. (t)}/y, . 2 k(t) = (t) (2a) y, . (t)/[1-y(t)] n(t) (2b) From the plot triplet [n(t), k(t), y(t)], the rate constant and order plots (x) and (x), respectively, for x  [0, 1], follow (at y = 1 there is lack of observability, meaning that Eq. 2 does not have a solution). With the statistics fitting report it is also possible to obtain the parameters certainty estimation along the experimental run (López-Serrano et al., 2007). RESULTS AND DISCUSSION The estimated dimensionless rate constant obtained by the proposed approach are depicted in Fig. 1. Here, it can be seen that, at each temperature, the overall rate constant (solid lines) shows an independent behavior only at low conversion values. For comparison the results obtained by Zhao and Hu (2007) are also presented (dashed lines). It is clear that, for the ID approach, all the runs present a similar shape-behavior but this does not occur for the Zhao and Hu proposal. Fig. 1. Estimated dimensionless rate constant (solid lines) compared with Zhao and Hu (2007) results (dashed lines). The estimated values for the reaction order, for each temperature, are presented in Fig. 2, where one can appreciate that, opposed to the single function proposed by Zhao and Hu (black line), a different function depicts the reaction order for each temperature. Analyzing the overall rate constant trajectory (or the reaction order), a function K (or n) was used to fit Fig. 1 (or Fig. 2) behavior, with excellent description up to 65% (or 40%) conversion, and that is when observability with certainty is obtained. These functions are depicted as follows: K(x) = k 1 +k 2 x m n(x) = k (3a) a + k b x + k c x 2 The values for all the parameters appearing in Eq. 3 were evaluated and Arrhenius and activation energies were obtained for k (3b) 1 and k 2 (not shown). Using the functions depicted by Eq. 3, the conversion evolution is T-138 309