Rheokinetics of ring-opening metathesis polymerization of norbornene-based monomers intended for self-healing applications.
Thermosetting polymers and their composites, while used in a wide variety of applications, are susceptible to cracking damage of various forms (e.g., matrix microcracking, interlaminar fracture, fiber-matrix debonding). Traditionally, the repair of cracks requires some type of manual intervention. However, a new class of self-healing materials is emerging that is able to sense the cracking damage, react to that damage to initiate healing, and restore the original properties of the material .
White et al.  first demonstrated the self-healing concept in an epoxy polymer composite in 2001. In that work, a microencapsulated healing agent (endo-dicyclopentadiene [endo-DCPD]) and a chemical trigger (Grubbs catalyst) were embedded in an epoxy matrix. Upon damage, the developed cracks rupture the embedded microcapsules, releasing healing agent into the crack plane through capillary action. The ring-opening metathesis polymerization (ROMP) of the healing agent was triggered by contact with the embedded catalyst, bonding the crack faces. Recovery of up to 75% in fracture toughness was reported.
Subsequent studies using the same microencapsulated healing agent examined the effect of healing time , catalyst form [4, 5], fatigue loading , and microcapsule size and concentration  on the healing efficiency. The concept was also demonstrated in the repair of delamination damage in woven fiber-reinforced systems where the microencapsulated healing agent was mixed with the epoxy matrix that was used to impregnate the fiber perform . A detailed overview of the microencapsulation process is given in Ref. 9.
In these previous studies, it is shown that understanding the ROMP kinetics of the microencapsulated monomer and the chemical trigger is of great importance for the optimization of self-healing materials. If the ROMP reaction is too fast, there is a possibility that gelation of the monomer will occur before it travels along the entirety of the crack, or that rapid polymerization around the exposed catalyst particle will quench the reaction before the majority of the catalyst is dissolved into the monomer. On the other hand, if the reaction is too slow, some of the monomer could diffuse into the matrix before it polymerizes resulting in incomplete coverage of the fracture plane, as seen in Ref. 10.
The main requirements that the healing agent must meet in order for the self healing concept to be successful are long shelf life, low viscosity, low shrinkage upon polymerization, high reactivity at room temperature, and strong adhesion to the matrix and reinforcement. Following these criteria, different norbornene based ROMP monomers, as well as some of their mixtures, initiated by the highly functional group tolerant Grubbs Ru based catalyst, were chosen for examination.
The ROMP of the healing agent triggered by the catalyst is characterized by a significant and often fast change of viscosity with time. The study of the time dependent rheological behavior of polymeric materials during cure is a useful aid in characterizing polymerizations. This field of research, called rheokinetics, aims to evaluate the effects that experimental conditions (e.g., test temperature, catalyst concentration, shear rate) have on a polymerization reaction and to model and predict the change in viscosity with time by monitoring the deformation and flow (rheology) of the material under study .
Several review articles of rheokinetics and chemorheological methods and models are available that are helpful in understanding the kinetics of rheological changes during polymerization; see for example Refs. 11-14. Typically, the effects of shear rate and filler content on the viscosity development are separated from the cure effects by separate tests. Test methods include both isothermal and nonisothermal experiments and are typically measured in either dynamic or steady shear modes using a variety of test geometries including parallel plate, cone and plate, eccentric disc, and sliding plate configurations. A variety of models have been proposed that have focused on the determination of the cure effects (viscosity development as a function of time and temperature) which are summarized in Ref. 12 and range from simple empirical models to various Arrhenius models to detailed models based on free volume analysis, gelation, and probabilistic methods.
In this study, the effect of cure on the viscosity of candidate ROMP based healing agents are investigated using isothermal steady shear rheokinetics experiments. The influence of several different variables (e.g., test temperature, catalyst concentration, and catalyst form) on polymerization initiation and propagation rates are considered to develop more effective monomers and blends for use in self-healing composite materials.
MATERIALS AND METHODS
Three different monomers, as well as one mixture of two of these monomers, were chosen as candidate healing agent systems for evaluation: 5-ethylidene-2-norbornene (ENB), endo-DCPD, exo-DCPD, and, finally, a 1:1 volume mixture of endo-DCPD:ENB. Figure 1 shows the chemical structure of the three pure monomers. ENB and the endo-isomer of dicyclopentadiene are commercially available, and were purchased from Sigma-Aldrich, while exo-DCPD was synthesized from endo-DCPD following the procedure detailed in the following section. A low concentration, 10 mg per mL, of untreated fumed silicon dioxide (CAB-O-SIL[R]) was added to the pure ENB monomer to increase the initial viscosity enough to be confidently measured within the sensitivity of our instrument.
[FIGURE 1 OMITTED]
First generation Grubbs catalyst, (PC[y.sub.3])[.sub.2](Cl)[.sub.2]Ru = CHPh, (Sigma-Aldrich) was used as the ROMP trigger. Figure 2 summarizes the ROMP reaction for endo-DCPD, which is similar to the ROMP reaction of the other monomers intended as healing agents. Previous studies show different advantages with the use of ENB and exo-DCPD, such as a decrease in the amount of Grubb's catalyst needed for the ROMP reaction (therefore, cost optimization), and also much faster kinetics [15-17].
Organic Synthesis of exo-DCPD
The two isomer forms of DCPD, endo-DCPD and exo-DCPD, can be seen in Fig. 1. Recent NMR studies have shown that the faster polymerization kinetics for exo-DCPD, compared with endo-DCPD, when initiated by the Grubbs' catalyst, is primarily due to steric interactions . Since the exo-DCPD could not be purchased commercially, it had to be synthesized from the kinetic endo-DCPD.
[FIGURE 2 OMITTED]
The synthesis procedure for creating exo-DCPD from endo-DCPD selected for this experiment was a slight modification of a procedure that was first demonstrated by Nelson and Kuo . In this more annotated procedure, a complex sequential organic synthesis scheme is used to rearrange the structure of the endo-DCPD molecule. In the first reaction of the synthesis, endo-DCPD is slowly added to a stirred solution of hydrobromic acid (HBr) under nitrogen, heated to 75[degrees]C and allowed to stir overnight. Reversal in the addition order of the reagents will result in polymerization of the endo-DCPD. The resultant product of the first reaction is an HBr adduct in the exo-configuration as seen in Fig. 3. The rearrangement most likely occurs from the non-classical carbocation intermediate of the first reaction. Isolation of the HBr adduct followed the published paper as stated without encountering any difficulties. The second reaction is used to perform a dehydrohalogenation using potassium hydroxide as base (KOH), affording exo-DCPD as depicted in Fig. 4. Though the procedure can be summarized into simple diagrams there are several intermediary-processing steps that must be performed to obtain pure exo-DCPD, and those interested should refer to reference . To ensure that the second reaction had reached completion, gas chromatography analysis was used to monitor the ratio of starting materials to the reaction product. Using the gas chromatography data, a plot was created to chart the progress of the reaction as seen in Fig. 5. In this plot, it can be seen that after about 60 hr the reaction reached an equilibrium plateau where the reaction could be stopped. The dehydrohalogenation reaction most likely follows an El mechanism as evidenced by the fact that stronger bases than KOH did not affect the rate of the reaction. In addition, for effective enhancement of the rate of the reaction, it was necessary to super heat the reaction by maintaining a "hard reflux."
[FIGURE 3 OMITTED]
The results of exo-DCPD synthesis were tracked throughout the course of the process using nuclear magnetic resonance ([.sup.1.H] NMR) analysis. These NMR spectra provided detailed data concerning how well each reaction had transformed the starting materials. In Fig. 6, a NMR spectrum of the HBr reaction product is shown with the critical peaks labeled. The sizes of the halide peaks from the NMR were compared with the peaks obtained from the product of the KOH reaction seen in Fig. 7. Through this comparison it can be seen that the endo-DCPD was virtually eliminated and the halide content had been significantly reduced before the final distillation. The total synthesis gave a high purity yield of 32.4 g (12.3%) for use in the rheokinetics experiments. The NMR of the final product showed no traces of residual endo-DCPD or halide compounds as shown in Fig. 8. The lower than expected yield of the large-scale reaction was due to thermal decomposition of the product in the reaction pot. The decomposition occurred because too much time was spent trying to remove the residual endo-DCPD using the cracking method described in the original paper .
[FIGURE 4 OMITTED]
[FIGURE 5 OMITTED]
[FIGURE 6 OMITTED]
[FIGURE 7 OMITTED]
The ROMP kinetics of the different monomer systems were investigated using a classical cone and plate viscometer (Brookfield CAP 2000+ Viscometer). The variables studied were catalyst concentration, monomer system, test temperature, and, to a lower degree, catalyst size and morphology.
Table 1 summarizes the range used for catalyst concentration (mg catalyst per mL healing agent) and isothermal test temperatures for each of the four healing agent systems evaluated. For each experiment, the Grubbs' catalyst was poured into a 15-mL vial and weighed using a 0.01 mg analytical balance. Next, the healing agent was injected into the vial using a 10-mL syringe. The endo-DCPD:ENB mixtures were mixed in a separate container using a volume ratio of 1:1 before being injected into the vial containing the catalyst. As soon as the healing agent was injected into the vial, the time was started on the viscometer's data acquisition system, and then the monomers and catalyst were vigorously shaken in the vial for a few moments before pouring a thin film onto the viscometer's preheated plate. The cone was then lowered onto the thin film of polymerizing material. The cone speed was set to 40 rpm for all cases which corresponded to a constant shear rate of 133 sec[.sup.-1]. The diameter of the plate is 50.0 mm, the diameter of the cone is 24.0 mm, and the angle of the cone is 1.8[degrees].
[FIGURE 8 OMITTED]
A small number of experiments using crushed Grubbs' catalyst were conducted to see the influence that crystal size has in the ROMP reaction of the monomers. To reduce the crystal size, a small amount of catalyst was placed in an electric pulverizing ball mill, or "wiggle bug" (Crescent Amalgamator Model #3110B), and milled for approximately 30 sec just prior to use. Subsequent NMR results on the catalyst before and after milling under the same experimental conditions did not show a change, indicating that the milling process does not cause chemical decomposition of the catalyst.
RESULTS AND DISCUSSION
In all cases the healing agent candidates showed a superlinear increase in viscosity with time during the ROMP reaction. The general qualitative trend on the viscosity behavior with time, regardless of the test conditions (see Table 1), was observed; Fig. 9 shows a typical result. Because of the great difference in the kinetics of the ROMP reaction for the four different healing agent systems, different experimental parameters (e.g., concentration, temperature) were used for each of the four cases. For example, because of the fast polymerization observed with the ENB monomer, the experiments were limited to low catalyst concentrations (between 0.05 mg/mL and 0.5 mg/mL), because concentrations higher than about 1 mg/mL resulted in the monomer polymerizing in the mixing vials before measurements could be taken using the viscometer. On the other hand, for the endo-DCPD samples, the polymerization was limited to higher concentrations (between 5 mg/mL and 30 mg/mL), otherwise the polymerization rate would be too slow for reasonable measurement lengths, especially for the low test temperature cases.
Following principles of rheokinetics developed by Cioffi et al. [11, 19], it was observed that these curves could be well fitted by two power laws, as shown in Fig. 9, and in Eq. 1 below:
[FIGURE 9 OMITTED]
[upsilon] = [a.sub.1][t.sup.[b.sub.1]] for t < [t.sub.t]
[upsilon] = [a.sub.2][t.sup.[b.sub.2]] for t > [t.sub.t] (1)
where [upsilon] is absolute viscosity, [a.sub.1] and [a.sub.2] are constants, t is time from the initial mixing of the monomer and catalyst, [b.sub.1] and [b.sub.2] are an experimentally determined viscosity exponents, and [t.sub.t] is an experimentally determined transition time.
According to Cioffi , the viscosity increase during some chain growth polymerization reactions has a power law dependence with two, sometimes even three, distinguishable regions. Because of the sensitivity issues with the viscometer used in this study, only the latter two regions can be observed in the results, which are labeled as Region I and Region II in Fig. 9, respectively.
The transition time ([t.sub.t]), which divides Region I from Region II, was chosen to maximize the sum of the correlation coefficients ([R.sup.2]) of the least squares best fit to the data, which was observed to be the approximate time at which the viscosity value reached about 10 poise for all cases. The time difference ([DELTA]t) between the points at which the viscosity reaches 10 poise and when it reaches 150 poise (see Fig. 9) is also recorded for comparison purposes. The [t.sub.t] parameter is a measure of the ROMP reaction's initiation rate, and seems to dependant mainly on the concentration and test temperature, while [DELTA]t is related to how fast the healing agent system polymerizes once this process is started, which seems to be dependant heavily on the healing agent system used.
[FIGURE 10 OMITTED]
Initially, the time exponents ([b.sub.1] and [b.sub.2]) of the least squares best fits of the curves were intended to be used to evaluate the kinetics of the ROMP reactions, but results did not seem to follow the expected trends (i.e., the exponent for Region I would sometimes be higher than that of Region II). This was due to the scattered nature of the measurements caused by decreased instrument sensitivities at low viscosities in the initial measurements in Region I, clearly shown at lower viscosities when plotted on a log-log scale in Fig. 9b.
Figure 10 shows the typical viscosity history for the four different candidate healing agents at different catalyst concentrations. The ENB at the lowest catalyst concentration of just 0.06 mg/mL had the fastest initiation time, [t.sub.t], at 249 sec and a [DELTA]t value of 46 sec. The exo-DCPD (1.00 mg/mL catalyst loading), 1:1 mixture of endo-DCPD:ENB (1.67 mg/mL), and endo-DCPD (10.00 mg/mL) healing agents shown in Fig. 10 had corresponding [t.sub.t] values of 344, 355, and 1063 sec and [DELTA]t values of 30, 124, and 248 sec, respectively. It is interesting to point out how endo-DCPD is much slower than the other three systems, yet it uses ~170 times more catalyst than ENB. Another feature observed in Fig. 10 is how the apparent slopes of the curves in Region II change depending on the healing agent system used, quantified by the [DELTA]t parameter.
[FIGURE 11 OMITTED]
The effect of catalyst concentration, at a test temperature of 25[degrees]C, on the ROMP kinetics of the four different healing agent systems can be observed in Fig. 11, which plots [DELTA]t and [t.sub.t] versus concentration in a log-log scale. The lines represent the power law best fits for the data. The data appear relatively linear on the log-log scale, indicating the power law dependence of the kinetic parameters on concentration. The transition time, [t.sub.t], and [DELTA]t for the endo-DCPD cases are the highest; therefore, endo-DCPD has the slowest polymerization kinetics and requires the largest catalyst concentration out of the four systems. Figure 11 also shows that the 1:1 endo-DCPD:ENB mixture has faster kinetics than endo-DCPD, but slower kinetics than those observed for either exo-DCPD or ENB. It can also be observed that ENB and exo-DCPD have similar [DELTA]t versus concentration behavior, indicating that they polymerize in a similar fashion once the ROMP reaction is started, but the time it takes for the ROMP reaction to be initiated (i.e., [t.sub.t]) is higher for the exo-DCPD than the ENB.
Figure 12 shows the effect that the test temperature has on the ROMP kinetics, by plotting [DELTA]t and [t.sub.t] versus test temperature on a log-log scale for different healing agent systems at various catalyst concentrations. The catalyst concentrations ranged from 0.08 mg/mL for the ENB to 10 mg/mL for the endo-DCPD. As expected, both [DELTA]t and [t.sub.t] decrease as the isothermal test temperature increases. It is important to note that the pure endo-DCPD test samples could not be tested at the lowest isothermal test temperature of 5[degrees]C because their viscosity would not show any significant increase, even with relatively high concentration levels, and the samples would eventually diffuse into the air. The exo-DCPD samples were not tested in these experiments because of the limited supply of the synthesized material. Again, the lines represent the power law best fit of the data for each monomer system, indicating the power law dependence of the kinetic parameters on temperature.
[FIGURE 12 OMITTED]
Figures 13a and 13b show Scanning Electron Microscope (SEM) images (at 200X magnification) of the Grubbs' catalyst crystals as received from Sigma-Aldrich, and of the Grubbs' catalyst powder after pulverization using the electric ball mill, respectively. Comparing both images, it can be observed that the Grubbs' catalyst particles were greatly reduced using the mill; therefore, a large increase in the surface area was achieved. Figure 14 shows the results obtained using Grubbs' catalyst as received, at a 2 mg/mL concentration, and using the more reactive crushed Grubbs' catalyst, at a 1 mg/mL concentration, for two test samples of the 1:1 endo-DCPD:ENB mixture. It can be observed that for the crushed Grubbs' catalyst case the reaction was much faster than that of the "as received" catalyst, even though a lower concentration was used. It is important to point out how the Grubbs' size and morphology seem to have more influence on the initiation time ([t.sub.t]) of the ROMP reactions, than on the [DELTA]t (see Fig. 14). Clearly, the kinetic parameters found in this study depend both on the polymerization kinetics and the dissolution kinetics. When preparing the samples in the vial, it was also observed that the speed and quality of the dissolution of the crushed catalyst in the monomer was significantly improved. This improved solubility is attributed to the increase in the surface area obtained by milling the catalyst. This is consistent with recent work by Jones et al., where the dissolution kinetics for Grubbs' catalyst of different size and morphologies in decalin was examined by ultraviolet/visible absorbance . Jone's experiments indicated that, as expected, smaller crystals provide faster dissolution kinetics. While it is not possible to measure the dissolution rate of Grubbs' catalyst in the ENB and DCPD systems investigated in the present work because of the competing polymerization reaction, the differences in solvating ability between the two isomers of DCPD and ENB are considered negligible because the three monomers are all very similar hydrocarbons with negligible polarities.
[FIGURE 13 OMITTED]
[FIGURE 14 OMITTED]
The ROMP kinetics of the norbornene-based candidate healing agents, and their parameters [DELTA]t and [t.sub.t], show a strong dependence on the monomer used, the catalyst concentration, catalyst crystal size, and test temperature. The healing agent system that showed the fastest kinetics was ENB, the system that showed the slowest kinetics was the endo-DCPD, while the mixture of ENB and endo-DCPD had kinetics parameters between the pure ENB and pure endo-DCPD. Previous results on the difference of the kinetics of both endo- and exo-isomer forms of DCPD  were confirmed by showing that the exo-DCPD's polymerization reaction was much faster (i.e., lower [t.sub.t] and [DELTA]t) and required much lower catalyst loading levels. Also, by reducing the catalyst crystal size through milling, faster kinetics can be obtained.
Requirements for effective self-healing depend on the type of damage and loading conditions, as well as the temperature and environment of the healing process. However, in general, increased kinetics at ambient (and lower) temperatures is desirable. By increasing the reaction kinetics, higher frequency fatigue damage modes may be repaired (see for example Ref. 6), and monomer diffusion into the matrix resulting in incomplete coverage of the fracture plane can be avoided (provided that crack filling happens much faster than gelation which is typically expected to be the case). In addition, for practical applications, it is necessary to use very low catalyst loadings due to cost constraints. For these reasons, ENB and exo-DCPD rich systems are attractive potential self-healing agent candidates for autonomic damage repair applications.
1. R. Wool, Nature, 409, 773 (2001).
2. S.R. White, N.R. Sottos, P.H. Geubelle, J.S. Moore, M.R. Kessler, S.R. Sriram, E.N. Brown, and S. Viswanathan, Nature, 409, 794 (2001).
3. E.N. Brown, N.R. Sottos, and S.R. White, Exp. Mech., 42, 372 (2002).
4. J. Rule, E.N. Brown, N.R. Sottos, S.R. White, and J.S. Moore, Adv. Mater., 72, 205 (2005).
5. A. Jones, J.D. Rule, S.R. White, N.R. Sottos, and J.S. Moore, Chem. Mater., 18, 1312 (2006).
6. E.N. Brown, S.R. White, and N.R. Sottos, Compos. Sci. Technol., 65, 2474 (2005).
7. E.N. Brown, S.R. White, and N.R. Sottos, J. Mater. Sci., 35, 1703 (2004).
8. M.R. Kessler, N.R. Sottos, and S.R. White, Compos. Part A: Appl. Sci. Manuf., 34, 743 (2003).
9. E.N. Brown, M.R. Kessler, N.R. Sottos, and S.R. White, J. Microencapsul., 20, 719 (2003).
10. M.R. Kessler and S.R. White, Compos. Part A: Appl. Sci. Manuf., 32, 683 (2001).
11. M. Cioffi, K.J. Ganzeveld, A.C. Hoffmann, and L.P.B.M. Janssen, Polym. Eng. Sci., 42, 2383 (2002).
12. P.J. Halley and M.E. MacKay, Polym. Eng. Sci., 36, 593 (1996).
13. A. Ya. Malkin and S.G. Kulichikhin, Adv. Polym. Sci., 101, 217 (1991).
14. A. Ya. Malkin and S.G. Kulichikhin, Rheokinetics, Huthig & Wepf, Heidelberg, Germany (1996).
15. J.K. Lee, S.J. Hong, X. Liu, and S.H. Yoon, Macromol. Res., 12, 478 (2004).
16. X. Liu, J.K. Lee, S.H. Yoon, and M.R. Kessler, J. Appl. Polym. Sci., in press.
17. J.D. Rule and J.S. Moore, Macromolecules, 35, 7878 (2002).
18. G.L. Nelson and C.L. Kuo, Synthesis, 105, 105 (1975).
19. M. Cioffi, A.C. Hoffmann, and L.P.B.M. Janssen, Polym. Eng. Sci., 41, 595 (2001).
G. E. Larin, N. Bernklau
Department of Mechanical Engineering, University of Tulsa, Tulsa, Oklahoma 74104-3189
M. R. Kessler
Department of Materials Science and Engineering, Iowa State University, Ames, Iowa 50011-2300
Department of Chemistry and Biochemistry, University of Tulsa, Tulsa, Oklahoma 74104-3189
Correspondence to: M.R. Kessler; e-mail: firstname.lastname@example.org
Contract grant sponsors: U. S. Army Research Laboratory and the U.S. Army Research Office; contract grant number: W911NF0510540.
TABLE 1. Experimental setup. endo-DCPD ENB 1:1 endo-DCPD:ENB exo-DCPD Catalyst concentration 3-30 0.05-0.5 1.67-4 0.125-1 range (mg/mL) Test temperature range 15-40 5-40 5-40 25 ([degrees]C)
|Printer friendly Cite/link Email Feedback|
|Author:||Larin, G.E.; Bernklau, N.; Kessler, M.R.; DiCesare, J.C.|
|Publication:||Polymer Engineering and Science|
|Date:||Dec 1, 2006|
|Previous Article:||A novel fabrication method for the mold insert of microlens arrays by hot embossing molding.|
|Next Article:||Natural rubber-polyurethane block copolymers: nonlinear structural variations with NCO/OH ratio.|