Chapter 9:Predicting Thermooxidative Degradation 369 cracks and transverse ply cracks coalescing to form larger ply-level cracks. For this reason,the resin-or matrix-dominated properties are of primary interest for modeling the high-temperature aging response. The mechanistic approach of modeling discrete damage,such as trans- verse ply cracks [58,89],provides accurate assessments of the effective response for a given damage state but has limitations on the number of discrete cracks or amount of discrete damage that can be represented. Therefore,mechanistic damage modeling is often reserved for analyzing local details and small components,or it can be used as a local model in a global analysis.Ultimately,for modeling large structures,phenomeno- logical approaches of continuum damage mechanics [4,100]can be used to represent the evolution of effective local damaged properties through the constitutive relationships,whereby widespread dispersed damage can be more easily represented. 9.2 Experimental Characterization and Observation The research presented here focuses on PMR-15 resin,carbon fibers,and PMR-15/carbon fiber composites.Although other HTPMC material systems are of interest,the open literature contains very little data for all but the PMR-15 material.Other materials may behave significantly different from the PMR-15 material system,however the modeling methodology given here should be applicable to modeling thermal oxidation in other polymer systems.Thermooxidative aging experiments,which include the effects of both physical and chemical aging,were conducted on the neat polymer and composite specimens.Model development and experiments focus on thermo- oxidative degradation while hydrolytic degradation will be addressed in subsequent research efforts.The spatial and temporal variability of the oxidized material and the ensuing mechanical properties were monitored to determine material parameters for the implemented models. Determination of model parameters for the multiscale modeling effort entails experimental characterization of the composite constituents,namely the fiber and the matrix polymer.However,characterization of the fiber and matrix is not sufficient to predict the behavior of the composite due to the critical fiber-matrix interface/interphase that develops during the cure process.Although a direct measure of the properties in the interface/ interphase region is challenging,indirect measures of the influence of the fiber-matrix interphase based on experimental observation of the com- posite behavior can be used to predict the model parameters for the interphase
cracks and transverse ply cracks coalescing to form larger ply-level cracks. For this reason, the resin- or matrix-dominated properties are of primary interest for modeling the high-temperature aging response. The mechanistic approach of modeling discrete damage, such as transverse ply cracks [58, 89], provides accurate assessments of the effective response for a given damage state but has limitations on the number of discrete cracks or amount of discrete damage that can be represented. Therefore, mechanistic damage modeling is often reserved for analyzing local details and small components, or it can be used as a local model in a global analysis. Ultimately, for modeling large structures, phenomenological approaches of continuum damage mechanics [4, 100] can be used to represent the evolution of effective local damaged properties through the constitutive relationships, whereby widespread dispersed damage can be more easily represented. 9.2 Experimental Characterization and Observation The research presented here focuses on PMR-15 resin, carbon fibers, and PMR-15/carbon fiber composites. Although other HTPMC material systems are of interest, the open literature contains very little data for all but the PMR-15 material. Other materials may behave significantly different from the PMR-15 material system, however the modeling methodology given here should be applicable to modeling thermal oxidation in other polymer systems. Thermooxidative aging experiments, which include the effects of both physical and chemical aging, were conducted on the neat polymer and composite specimens. Model development and experiments focus on thermooxidative degradation while hydrolytic degradation will be addressed in subsequent research efforts. The spatial and temporal variability of the oxidized material and the ensuing mechanical properties were monitored to determine material parameters for the implemented models. Determination of model parameters for the multiscale modeling effort entails experimental characterization of the composite constituents, namely the fiber and the matrix polymer. However, characterization of the fiber and matrix is not sufficient to predict the behavior of the composite due to the critical fiber–matrix interface/interphase that develops during the cure process. Although a direct measure of the properties in the interface/ interphase region is challenging, indirect measures of the influence of the fiber–matrix interphase based on experimental observation of the composite behavior can be used to predict the model parameters for the interphase. Chapter 9: Predicting Thermooxidative Degradation 369
370 G.A.Schoeppner,G.P.Tandon and K.V.Pochiraju 9.2.1 Fiber Oxidation Degradation The effects of service temperatures and oxidation on the mechanical pro- perties of high-temperature polymer composites are primarily manifested in the polymer-dominated properties,namely the transverse properties (perpendicular to the fiber direction)and the shear properties.However, the fiber-dominated properties may be affected by degradation of the fiber and deterioration of the fiber-matrix interface.This may be particularly true for composites with glass fibers [44]in which magnesium and sodium can leach out from the fiber and possibly contribute to polymer/interface degradation.It is reported that graphite fibers containing significant amounts of sodium and potassium as contaminants are less thermooxidatively stable than graphite fibers with very low alkali metal contents [39,43].Studies conducted by Bowles and Nowak [12]indicate that extreme oxidative erosion of the Celion 6000 graphite fiber occurs at elevated temperatures in the presence of the polyimide matrix. Bowles [8]has investigated the effects of different fiber reinforce- ments on thermooxidative stability of various fiber-reinforced PMR-15 composites.The ceramic Nicalon and Nextel fibers were found to drastically accelerate thermal oxidation of the corresponding composites because of the active fiber-matrix interface.Compared to polyimide matrices,rein- forcing carbon fibers [60]are usually far more stable at the elevated temperatures considered.Studies conducted by Wong et al.[126]indicate that IM6 carbon fibers are more stable than the G30-500 fibers within the first 600 h during thermal oxidation at 371C.The oxidation rate of the IM6 fibers then increases substantially,leading to complete decomposition of the fibers during the final stage.The sudden increase in the oxidation rate in the IM6 fibers implies the possibility of change in degradation mechanisms. The three PAN-based carbon fibers of interest to the present body of work are two low modulus carbon fibers T650-35 and G30-500 that are typically used in HTPMCs,and one intermediate modulus carbon fiber, IM7.Isothermal aging studies were recently conducted [105]on these three types of fibers to study their oxidation behavior.Both sized and unsized fibers were exposed to different elevated temperatures for varying time periods in an attempt to understand the influence of the fiber sizing/ coupling agent on their thermooxidative stability.Note that aging of the bare fibers may not necessarily be representative of the behavior of the in situ fibers embedded in the matrix,because the exposed surface area of the fibers in the composite is only a very small percentage of the total surface area of the fibers.Thermal degradation was quantified by the amount of weight loss measured,while degradation of mechanical properties was
9.2.1 Fiber Oxidation Degradation The effects of service temperatures and oxidation on the mechanical properties of high-temperature polymer composites are primarily manifested in the polymer-dominated properties, namely the transverse properties (perpendicular to the fiber direction) and the shear properties. However, the fiber-dominated properties may be affected by degradation of the fiber and deterioration of the fiber–matrix interface. This may be particularly true for composites with glass fibers [44] in which magnesium and sodium can leach out from the fiber and possibly contribute to polymer/interface degradation. It is reported that graphite fibers containing significant amounts of sodium and potassium as contaminants are less thermooxidatively stable than graphite fibers with very low alkali metal contents [39, 43]. Studies conducted by Bowles and Nowak [12] indicate that extreme oxidative erosion of the Celion 6000 graphite fiber occurs at elevated temperatures in the presence of the polyimide matrix. Bowles [8] has investigated the effects of different fiber reinforcements on thermooxidative stability of various fiber-reinforced PMR-15 composites. The ceramic Nicalon and Nextel fibers were found to drastically accelerate thermal oxidation of the corresponding composites because of the active fiber–matrix interface. Compared to polyimide matrices, reinforcing carbon fibers [60] are usually far more stable at the elevated temperatures considered. Studies conducted by Wong et al. [126] indicate that IM6 carbon fibers are more stable than the G30-500 fibers within the first 600 h during thermal oxidation at 371°C. The oxidation rate of the IM6 fibers then increases substantially, leading to complete decomposition of the fibers during the final stage. The sudden increase in the oxidation rate in the IM6 fibers implies the possibility of change in degradation mechanisms. The three PAN-based carbon fibers of interest to the present body of work are two low modulus carbon fibers T650-35 and G30-500 that are typically used in HTPMCs, and one intermediate modulus carbon fiber, IM7. Isothermal aging studies were recently conducted [105] on these three types of fibers to study their oxidation behavior. Both sized and unsized fibers were exposed to different elevated temperatures for varying time periods in an attempt to understand the influence of the fiber sizing/ coupling agent on their thermooxidative stability. Note that aging of the bare fibers may not necessarily be representative of the behavior of the in situ fibers embedded in the matrix, because the exposed surface area of the fibers in the composite is only a very small percentage of the total surface area of the fibers. Thermal degradation was quantified by the amount of weight loss measured, while degradation of mechanical properties was 370 G.A. Schoeppner, G.P. Tandon and K.V. Pochiraju
Chapter 9:Predicting Thermooxidative Degradation 371 measured using single-filament tests on unaged and aged fibers.Finally, surface morphology changes were monitored in an attempt to relate physical and surface changes to the decreases in mechanical performance as a function of aging Weight loss studies Figure 9.2 shows that the thermal degradation (as measured by weight loss)of unsized T650-35 carbon fibers is substantial for moderate temp- eratures above 316C.It is seen that there is a large (~40%)weight reduction of T650-35 carbon fibers after 2,000 h of aging at 343C.At the lower temperature of 232C,the weight loss is not significant and the weight loss as a function of time follows approximately a linear relation- ship.However,as the aging temperature is increased to 343C,the T650-35 fibers are observed to lose weight rapidly.There seems to be a change in slope in the weight loss curve at around 750 h of aging,possibly signifying a change of degradation mechanism.Typically,the weight loss data for neat polymer specimens are normalized with respect to the specimen surface area,since oxidative weight loss occurs from losses on exposed surfaces and edges of the specimen.Note that carbon fibers have an extremely large surface area due to their small diameter(typically 4-8 um) and large aspect ratio(typically >1,000). 50 0.1 PMR-15 40 T650-35 e0.08 341 006 343℃ 316℃ 232℃ T650-35 0 0 5001000150020002500 要o2月 05001000150020002500 Time,hrs Time,hrs Fig.9.2.Weight loss of T650-35 carbon Fig.9.3.Normalized weight loss of fiber as a function of aging temperature T650-35 fiber and PMR-15 resin Figure 9.3 shows a comparison of the normalized weight loss of T650- 35 carbon fiber and PMR-15 neat resin at 343C.Note that the normalized fiber weight loss is almost negligible compared to that of neat resin over the entire 2,000 h of aging time.Thus,even though the carbon fiber loses a significant weight fraction with aging at elevated temperatures,the weight
measured using single-filament tests on unaged and aged fibers. Finally, surface morphology changes were monitored in an attempt to relate physical and surface changes to the decreases in mechanical performance as a function of aging. Weight loss studies Figure 9.2 shows that the thermal degradation (as measured by weight loss) of unsized T650-35 carbon fibers is substantial for moderate temperatures above 316°C. It is seen that there is a large (∼40%) weight reduction of T650-35 carbon fibers after 2,000 h of aging at 343°C. At the lower temperature of 232°C, the weight loss is not significant and the weight loss as a function of time follows approximately a linear relationship. However, as the aging temperature is increased to 343°C, the T650-35 fibers are observed to lose weight rapidly. There seems to be a change in slope in the weight loss curve at around 750 h of aging, possibly signifying a change of degradation mechanism. Typically, the weight loss data for neat polymer specimens are normalized with respect to the specimen surface area, since oxidative weight loss occurs from losses on exposed surfaces and edges of the specimen. Note that carbon fibers have an extremely large surface area due to their small diameter (typically 4–8 µm) and large aspect ratio (typically >1,000). Fig. 9.2. Weight loss of T650-35 carbon fiber as a function of aging temperature T650-35 fiber and PMR-15 resin Figure 9.3 shows a comparison of the normalized weight loss of T650- 35 carbon fiber and PMR-15 neat resin at 343°C. Note that the normalized fiber weight loss is almost negligible compared to that of neat resin over the entire 2,000 h of aging time. Thus, even though the carbon fiber loses a significant weight fraction with aging at elevated temperatures, the weight -0.02 0 0.02 0.04 0.06 0.08 0.1 0 500 1000 1500 2000 2500 Weight loss/surface area, g/cm2 Time, hrs PMR-15 T650-35 343o C 0 10 20 30 40 50 0 500 1000 1500 2000 2500 % Weight reduction Time, hrs T650-35 343o C 232o C 316o C Chapter 9: Predicting Thermooxidative Degradation Fig. 9.3. Normalized weight loss of 371
372 G.A.Schoeppner,G.P.Tandon and K.V.Pochiraju loss normalized by the fiber surface area is negligible compared to that of the resin.Weight loss studies on sized fibers [105]indicate that the sizing/coupling agent is released within a short time period (~24 h)of aging beyond which the weight loss trend is similar to the corresponding unsized fibers. Mechanical properties Single fiber specimens were tested at room temperature in tension using the single-filament test.Figures 9.4 and 9.5 show the normalized failure strength and failure strain,respectively,of unsized T650-35 carbon fibers aged at 343C.The strengths and failure strains are normalized with respect to their corresponding unaged values,such that the observed decreases are a reflection of the reduction resulting from isothermal aging.A minimum of ten fibers were tested for each condition considered,and the standard deviation is shown as the error bar across the measured mean values.There is some scatter in the strength and failure strain data which is typically encountered in single fiber testing as failure is sensitive to the presence of flaws over the fiber gage length.Fibers aged at the elevated temperature of 343C show a large decrease in the strength after 1,000 h of aging.The significant decrease in mechanical strength signifies that the carbon fibers should not be treated as static entities when composites containing these fibers are aged at this temperature for extended periods.However,the reported data are a worst case scenario in which all surfaces of the fibers are exposed during aging.In situ fibers of the composite have only a small fraction of their total surface area exposed to the oxidizing environment and should,therefore,suffer less degradation.Further,test data from single-filament testing [105]indicate that application of fiber sizing may 1.4 1.4 1.2 T650-35 aged at 343C 1.2 T650-35aged@343℃ 0.3 8 0.6 04 urens amje pazyeuoN 0.4 0.2 0 235547 764 235 547 764 100 Aging Time,hrs Aging Time,hrs Fig.9.4.Normalized strength of unsized Fig.9.5.Normalized failure strain of T650-35 carbon fibers aged at 343C unsized T650-35 fibers aged 343C
loss normalized by the fiber surface area is negligible compared to that of the resin. Weight loss studies on sized fibers [105] indicate that the sizing/coupling agent is released within a short time period (∼24 h) of aging beyond which the weight loss trend is similar to the corresponding unsized fibers. Mechanical properties Single fiber specimens were tested at room temperature in tension using the single-filament test. Figures 9.4 and 9.5 show the normalized failure strength and failure strain, respectively, of unsized T650-35 carbon fibers aged at 343°C. The strengths and failure strains are normalized with respect to their corresponding unaged values, such that the observed decreases are a reflection of the reduction resulting from isothermal aging. A minimum of ten fibers were tested for each condition considered, and the standard deviation is shown as the error bar across the measured mean values. There is some scatter in the strength and failure strain data which is typically encountered in single fiber testing as failure is sensitive to the presence of flaws over the fiber gage length. Fibers aged at the elevated temperature of 343°C show a large decrease in the strength after 1,000 h of aging. The significant decrease in mechanical strength signifies that the carbon fibers should not be treated as static entities when composites containing these fibers are aged at this temperature for extended periods. However, the reported data are a worst case scenario in which all surfaces of the fibers are exposed during aging. In situ fibers of the composite have only a small fraction of their total surface area exposed to the oxidizing environment and should, therefore, suffer less degradation. Further, test data from single-filament testing [105] indicate that application of fiber sizing may Fig. 9.4. Normalized strength of unsized T650-35 carbon fibers aged at 343°C unsized T650-35 fibers aged 343°C G.A. Schoeppner, G.P. Tandon and K.V. Pochiraju Fig. 9.5. Normalized failure strain of 372
Chapter 9:Predicting Thermooxidative Degradation 373 result in an improved performance of the carbon fiber in the unaged condition but could result in some loss of mechanical performance after aging at elevated temperatures.Failure is sensitive to the presence of flaws,and it is likely that the sizing/coupling agent on the fiber helps to decrease the influence of the fiber surface flaws resulting in a slightly better performance in the unaged condition. Surface characterization Scanning electron microscope (SEM)microstructural studies were con- ducted to determine the carbon fiber surface morphology changes during long-term isothermal aging.Figure 9.6a compares the SEMs of unsized T650-35 carbon fiber in the unaged condition with that of the fiber aged for 1,052 h at 343C in Fig.9.6b.No significant visible damage or surface morphology changes are observed,although some minimal amount of pitting is visible on the previously smooth fiber surfaces after 1,052 h of aging.Thus,the fiber surfaces provide little indication of the deterioration in the mechanical performance of the aged fiber. Pitting Mowitme (a) (b) Fig.9.6.SEMs showing surface morphology of unsized T650-35 carbon fibers in (a)the unaged condition and(b)aged for 1,052 h at 343C Anisotropic diffusivity It has been established that the fibers account for only a minimal amount of the total weight loss for HTPMCs.However,the role of fibers in the composite oxidation process,in particular their role in transporting oxygen into the composite,is not fully understood.Extruded PAN carbon fibers exhibit a preferred orientation in which the graphitic layers may extend for thousands of angstroms and extend straight for hundreds of angstroms parallel to the fiber axis providing high-strength and high-stiffness along
result in an improved performance of the carbon fiber in the unaged condition but could result in some loss of mechanical performance after aging at elevated temperatures. Failure is sensitive to the presence of flaws, and it is likely that the sizing/coupling agent on the fiber helps to decrease the influence of the fiber surface flaws resulting in a slightly better performance in the unaged condition. Surface characterization Scanning electron microscope (SEM) microstructural studies were conducted to determine the carbon fiber surface morphology changes during long-term isothermal aging. Figure 9.6a compares the SEMs of unsized T650-35 carbon fiber in the unaged condition with that of the fiber aged for 1,052 h at 343°C in Fig. 9.6b. No significant visible damage or surface morphology changes are observed, although some minimal amount of pitting is visible on the previously smooth fiber surfaces after 1,052 h of aging. Thus, the fiber surfaces provide little indication of the deterioration in the mechanical performance of the aged fiber. Fig. 9.6. SEMs showing surface morphology of unsized T650-35 carbon fibers in (a) the unaged condition and (b) aged for 1,052 h at 343°C Anisotropic diffusivity It has been established that the fibers account for only a minimal amount of the total weight loss for HTPMCs. However, the role of fibers in the composite oxidation process, in particular their role in transporting oxygen into the composite, is not fully understood. Extruded PAN carbon fibers exhibit a preferred orientation in which the graphitic layers may extend for thousands of angstroms and extend straight for hundreds of angstroms parallel to the fiber axis providing high-strength and high-stiffness along Chapter 9: Predicting Thermooxidative Degradation 373