4. PAG6 nanocomposite specimens tested at decreasing deflection amplitude exhibited nearly linear enhance- ‘ment in fatigue life on log scale with deerease in sur- face temperature and maximum stress amplitude at higher amplitudes of defection. 5. Fatigue fracture surface morphology of PAGS nano- composites revealed more rubbery behavior and a large seale void formation with increase in clay content REFERENCES 1 EV. Onmonde and M. Hiroaki, Chemical Economic Hand- book Report, 380.0800 (2007). 2. E, Carlson and K. Nelson, Nylon Under the Hood: A History of Innovation, Literature from EI dupont de Nemours and Company, New Jersey, USA (2003), 3. M. Alexandre and P. Dubois, Mater. Sci. Eng., 28, 1 (2000). 4, S. Sinha Ray and M. Okamoto, Prog. Polym. Sei., 28, 1539 (2003), 5. CE, Powell and G.W. Beall, Curr. Opin. Solid State Mater Sci, 10, 73 (2006). 6. QH. Zeng, A.B. Yu, G.Q. Lu, and DR. Paul, J. Nanosci Nanorechnol., 5, 1574 (2005), 7.1L Constable, J.G. Williams, and DJ. Bums, J. Mech. Eng. Sei, 12, 20 (1970), 8.S, Rabinowitz and P. Beardmore, J. Mater. Sci., 9 81 (1974), 9. RJ. Crawford and P.P. Benham, Polymer, 16, 908 (1975). 10. V. Bellenger, A. Teharkhtchi, and Ph, Castaing, Int. J. Fa- rigue, 28, 1348 (2006), 11. P.K. Mallick and Y. Zhou, J. Mater. Sci., 38, 3183 (2003), 12. S.C. Bellemare, MLN. Bureau, J. Denault, and J. Dickson, Polym. Compos., 25, 433 (2004). 13, A. Ramkumar and R. Gnanamoorthy, Compos, Sci. Technol., 68, 3401 (2008), 14, TD, Fores and D.R. Paul, Polymer, 44, 3945 (2003). 15. X. Liu, Q. Wu, and L.A. Berglund, Polymer, 43, 4967 (2002). 16. F. Chavarria and D.R. Paul, Polymer, 45, 8501 (2004. 17. K, Wang, $. Liang, R. Du, Q. Zhang, and Q. Fu, Polymer, 45, 7953 (2004). 18, PJ. Yoon, D.L. Hunter, and D.R, Paul, Polymer, 44, 5323 (2003). 19. H. Hedicke, H. Wittich, C. Mebler, F. Gruber, and V. Alt stadt, Comipos. Sei. Technol., 66, 571 (2006). 20, ML. Kohan, Nylon Plastics, Wiley, New York (1973), 21. L. Shen, LY. Phang, L. Chen, T. Liu, and K. Zeng, Poly- mer, 45, 3341 (2004), 22. B. Lin, A. Thumen, H.P. Heim, G. Scheel, and U. Sundar- raj, Polym. Eng. Sei., 49, $24 (2009) 23. AM, Freudenthal, Imroduction to the Mechanies of Solis, Wiley, New York (1996) 24, A. Tomkins and W.D. Biggs, J. Mater. Sci., 4, 932 (1969). 25. RJ. Crawford, Plastics Engineering, 3rd ed.. Pergamon Press, Oxford (1987). 26. G.M. Kim, S. Goerlitz, and G.H. Michler, J. Appl. Polym Sei 108, 38 (2006)FIG, 14. Macroscopic fatigue fracture surface appearance of (a) PAGS, (8) PAGSCNI, and (¢) PAG6CNS tested in laboratory environment [Color figure can be viewed in the online issue, which is available at wileyonlinlibrary com] scopic crack was found initiated at the comer in all the samples where the stress concentration is the highest, but the overall fracture surface texture of the specimens was effectively influenced by the percentage of clay content. No significant stress whitening was observed in all the specimens. PA66 specimens exhibited a flat relatively macroscopic featureless surface, near the comer, where the crack was initiated and propagated steadily. A bubbly surface in the middle region may be due to fast fracture with a viscous flow. PAG6CNI specimens revealed a slightly perturbed surface near the edges and a pull out region with viscous flow at the terminal fracture region. ‘The macroscopic fracture surface of PASSCNS specimens exhibited a more perturbed surface with regular bands near the fracture-initiated zone, macroscopic beach lines on the edge connected to the crack-initiated corner. This indicates further that the crack is propagating from bottom left comer to right fast and slowly to upward. An exten- sive void formation with more rubbery behavior was observed in the middle region. A regular large-scale void formation with void coalescence and secondary cracks was observed in the terminal fracture region. The severe rise of specimen temperature above 7, due to self heating resulted in more rubbery behavior in nanocomposite specimens with increase in clay content Figure 15a illustrates the initiation of crack in the patchy structure near the top left corner and then propa- gating toward the bottom right comer (discontinuous growth bands can be seen perpendicular to the arrow). As the crack is propagating, an extensive void formation with DOI 10.1002/pe an increased viscous flow was observed. Figure 15b shows a significant viscous flow with large-scale void for- ‘mation near the terminal fracture. This infers that the crack propagation in nanocomposite specimens was gov- emed by the void formation and viscous flow of material. CONCLUSIONS Based on the investigations, following conclusions are dra 1. An improved fatigue life for nanocomposites compared ‘with pure polymer, 2. PAG6 nanocomposites revealed higher induced stress due to improved storage modulus and extensive hyste- rotic heating due to high loss modulus. 3. A significant cyclic softening in nanocomposite speci- ‘mens compared with pure PAGS specimens subjected to, constant deflection amplitude. However, cyclic soften- ing is less prominent in PAG nanocomposite speci- ‘mens than PAG6 specimens at higher temperatures. FIG. 15. Microscopic fatigue fracture morphology of PAGGCNS (a) i of crack from voids and (b) large-scale void formation and signif- icant viscous flow of material near the terminal fracture (arrow indicates direction of erack propagation) POLYMER COMPOSITES—2010 19856mm oF 3mm eo | 21 Sm 2 Seo Fao x» ' a0, foot! bon No. of cycles FIG. 11. Variation of specimen surface temperature during cyclic load ing for PAGOCNS specimens at end deflection amplitudes of 6 mm, 4.5 ‘mm, 3 mm, and 1.5 ram. specimen temperature could be due to mechanical insta- bility of the material (Fig. 12). ‘A nearly linear trend is observed among fatigue life, specimen surface temperature, and maximum steady stress amplitude at higher end deflection amplitudes (Fig. 13). Material exhibited significant cyclic softening at higher cyclic deflection amplitudes. An observed proportional drop in fatigue life with increase in deflection amplitude could be attributed to an increase in inelastic component of strain and mechanical damping, with an increase in deflection amplitude. Fractography A protective skin layer in the injection-molded poly- mer components delays the fatigue crack initiation/propa- @ 70 ga 2 gn 2 a0 § 2 Bn 10 o 1 100 10000 1000000 No. of cycles FIG. 12, Evolution of flexural stress amplitude at outer layer during cyelic loading for PAGGCNS specimens at end deflection amplitudes of 6 ‘mm, 4.5 mm, 3 mm, and 1.5 ram. 1984 POLYMER COMPOSITES—2010 FIG, 13, Patigue life for PAGSCNS specimens along with steady state stress amplitude and steady state temperature at end deflection amphi tudes of 6 mm, 45 mm, 3 mim, and 1.5 mn. gation. Hence, in the injection-molded samples, more likely fatigue cracks develop from within the bulk of the material. ‘The cracks initiate through slip of molecules or at the areas of weak spherulitie boundaries in crystalline regions, and in voids formed during viscous flow in amor- phous regions of pure polymers [25]. The mechanisms of deformation and fracture of polymer clay nanocomposites differ from pure polymer as they contain various. struc- tural entities like individual silicate platelets, tactoids with different degree of intercalation, and larger particles. The amount of these structural entities inside the polymer clay nanocomposites depend on the amount of clay, type and amount of surfactant used for organophilization. Organo- philization influences both matrix/fller interaction and the internal adhesion of silicate platelets. The deformation process would be directly influenced by the orientation, along with the dispersion of the silicate platelets as well as the tactoids. ‘The main deformation mechanism observed by Kim et al, [26] is microvoid formation either within the intercalated tactoids through microcavitation or in the vicinity of the layered silicates in the matrix, for intercalated and exfoliated nanomorphology systems, respec- tively. Microvoid formation takes place more uniformly and homogeneously throughout the deformed specimen of exfoliated nanocomposites. As a result, well-improved mechanical properties can be achieved in comparison with the mixed nanomorphology system. Hence, it can be inferred in the case of nanocomposite specimens subjected to cyclic loading that the fatigue cracks may develop in the voids formed inside the intercalated tactoids because the exfoliated nanocomposite specimens contain substan- tially higher proportions of polymer chains with restricted ‘mobility at the clay surface. Influence of clay content on the fracture morphology is evident from the difference in morphological features revealed by PAG6, PAGGCNI, and PAGGCNS specimen fracture surfaces at low magnification (Fig. 14). Macro- DO! 10.1002/pe8 8 PAGECNS (87.7 °C) B PABECNI (60.1 °C) 52°C) 8 ‘Steady stress amplitude (MPa) 20 "1000 700000 770000 No, of eycles to faire FIG, 10. Fatigue life of PAGS, PAGSCNI, and PAGGCNS specimens along with steady stat stress amplitude under fully reversed constant amplitude deflection of 6 mm, dominant energy dissipation (Joss modulus) in that tem- perature region. In region Il, which is at a slightly higher temperatures, PA66 nanocomposite specimens demon- strated lesser drop in stress amplitude compared with that of PA66, despite its “high” specimen temperature. This is due to a larger drop in energy dissipation for nanocompo- sites and moderate drop for pure PA66. The region I and region I] are influenced primarily by temperature-assisted changes in storage and dissipated energies, whereas region III is dominated by mechanical instability. In region III, a nonlinear cyclic softening is observed while the heat generated by the cyclically softened material is reaching equilibrium with heat dissipation. As cyclic deformation proceeded a steady state stress— strain relation was observed in the fully softened state ‘The dissipated energy in driving the specimen through a complete cycle remains constant and dynamic strain re- covery processes balance the cyclically applied strain to give a constant cyclic inelastic strain range. This inelastic strain and mechanical damping are the most important pa- rameters in the evaluation of fatigue damage of polymers [8]. In the steady state, the width of the stress-strain loops, at zero stress is greater than the width of initial cycles, indicating an increase in the inelastic strain range due to the cyclic softening (Fig. 6). This illustrates that the fa- tigue damage per cycle is greater in the fully softened state than in the initial condition. This effect can be more clearly visualized in the next set of experiments on PAGGCNS samples with varied amplitudes of cyclic deflection. Tomkins and coworker [24] observed that the breadth of the hysteresis loop, i.e., inelastic strain range ‘was directly proportional to total strain range and hence arrived at a Coffin-Manson constant and exponent for fa- tigue life for PA66 subjected to completely reversed strain at constant amplitude, In this study, PAG6 nanocomposite samples exhibited improved fatigue life in spite of equal or greater stress drop with higher temperature rise in the transition region than pure PAGS samples. It shows that the presence of organoclay in PA66 reduced the influence of the changes induced both in mechanical response and molecular pack- ing in the transition region on the fatigue lite. At labora- tory condition, in spite of exhibiting higher stress ampli- tude associated with more temperature rise, PAG6CNI specimens displayed more than twice the fatigue life com- pared with PAG6 samples (Fig. 10, Table 3). Whereas, PAG6CNS samples exhibited an enhancement in fatigue life compared with PA66CN1 samples. These results forti- fied the Bellemare et al. [12] findings that dispersed sheet- like inorganic nanoparticles in polyamide matrix enhanced the inherent material resistance to crack initiation. The enhancement in fatigue life of PAG6CNS samples confined to a marginal amount over PAG6CNI samples and a signif- icant increment in scatter may be due to existence of domi- nant intercalated structure in PA66CNS samples compared with the exfoliated structure in PAG6CN1 samples ‘The mechanism of fatigue fracture is primarily gov- emed by applied deflection and frequency because of the influence of these two parameters on the hysteretic heat- ing. A steady rise in specimen surface temperature before reaching steady state, was observed for PAGGCNS speci mens at higher deflection amplitudes, ic., 6 mm, 4.5 mm, and 3 mm, whereas no perceptible increase in specimen surface temperature was observed for PAGGCNS speci- mens at lower deflection amplitude, ie, 1.5 mm (Fig, 11). This indicates that the temperature rise in PA6GCNS specimens is an exponential function of cyclic deflection because temperature rise was not observed at lower deflection amplitudes and was almost linear at higher deflection amplitudes. At deflection amplitudes of 6 mm, 4.5 mm, and 3 mm, cyclic softening induced in the mate- rial is primarily due to rise in surface temperature of the specimen and is nearly proportional to that rise in temper- ature. In the case of deflection amplitude of 1.5 mm, a marginal cyclic softening was observed with out rise in TABLE 3. Maximum stress and specimen surface temperature at Ist and 10,000th cycles, and fatigue life for PAG, PAGSCNI, and PAGSCNS samples subjected to constant cyclic end deflection of 6 mm, Ist eyele 10.0001h eyete Material ‘Maximom siess (MPs) Temperature ©) Maximum stress (MPa) Temperature (©) _ Fatigue life (No, of eyeles) PAGO 395 239 20031 + 3717 PAGSCNI 652 267 445805 = 6538 PAGGCNS 704 283 54477 = 10541 OI 10.1002/pe POLYMER COMPOSITES—2010 1983100 © € jo i * PAGE “Pasecn “a Paeecns 2 110 100000 10000 100000 No. ofeyelee FH. 7. Variation of specimen sufce temperature wit constant esc loading for PAS6, PAGSCNI, and PAG6CNS, cyclic loading. The temperature rise is more pronounced, ive., about 17% higher in PAG6CNS samples and slightly more pronounced, ie., about 7% higher in PA66CN1 compared with PA66 samples. This can be due to increased energy dissipation with increase in clay content in PA66 (Fig. 5b). The energy dissipated, Q per cycle per unit volume, at deformations with constant amplitude of strain, &% is Q = zE"e}, where E” is the loss modulus of material [10]. A significant equilibrium temperature pla- teau existed after the transition region ruled out the possi- ble thermal failure in all the tested samples in this study. The PA66CNI and PAG6CNS samples exhibited ~10% and 18% higher respective maximum stress ampli- tudes compared with PA66 samples during first cycle 8), This is because of higher storage modulus of PAG6 nanocomposites than pure PAG6, which was observed at all temperatures in DMTA results as well (Fig. 5a). All samples tested under identical conditions 1 10 100. No. of eycles 4000 10000 FIG. 8. Evolution of flexural stress amplitude at outer layer with con- stant eyelic loading for PAGS, PAGGCN1, and PAGSCNS. 1982 POLYMER COMPOSITES—2010 100 %0 2 B70 $ 60 Boo Po 20 PAGS BSOCN 9 30 40 60 60 70 80 ‘Stress amplitude (MPa) FIG, 9, Evolution of Hexural stress amplitude and temperature at the ‘outer layer in transition stage with consiamteyelic loading exhibited ~60% drop in stress amplitude irrespective of their magnitudes of initial and steady stress amplitudes during the first thousand cycles (Table 3). This cyclic softening in the transition region can be attributed to drop in storage modulus with temperature rise. PAG6CNI and PAGGCNS samples exhibited higher stress amplitude of about 12% and 19% in cyclic steady state region even with ‘greater hysteretic heating compared with PAG6 samples. Figure 9 shows the variation of specimen temperature related to the stress amplitude. In region I and region Il, only a part of the dissipated energy by the specimen can be abstracted by the surroundings and the remaining energy is contributed to heating up the specimen, As the amount of dissipated energy is becoming equal to the amount of heat energy abstracted by the surroundings, the specimen temperature becomes stable. The regions I and TI are distinguished by a shift in the derivative, i, the rate of change in temperature with respect to change in cyclic amplitude, in response to constant deflection cyclic flexure (Table 2). The regions I and I corresponds to the glassy and rubbery behavior of PAGS nanocomposites and the derivatives depend on the relation between the dynamic modulus and temperature. In region I, which is at lower temperatures, PA66 nanocomposite samples exhibited more softening compared with PA66 sample due to the larger self heating prevailed because of its TABLE 2. Derivatives, ie, the rate of change in specimen temperanure With respect to change in stress amplitude, inthe linear cyclic softening domain for PAGS, PAGGCN1, and PAGSCNS. Derivative Material Region 1 Region I PA. o74s 0897 PAG6CNI 0737 1046 0.008, 1.106 PAG6CNS DOI 10.1002/pe14000, @ 12000 Froooe |PASECN $1000 >ABBCNS 3 8000} 3 so B so00 ca 2000 {5000 80-0800 Temperature ((C) 700 (b) 600 Fs00 = PASECN: 3400 B00 PASGCN 8200 AE 100 160-100-800 80 —«100~—«180 Temperature (°C) 02 (c) 0.10} 0.08 e PAGSCN: 0.06) * Pag 0.04 PABECN) 0.02 so -10 $00 «60100 ~—«180 Temperature (°C) FIG 5, Dynamic mechanical thermal analysis curves of (a) storage ‘modulus, (b) loss modulus, and (e) tan 3 for PAG6, PAGOCNI, and PAGGCNS at 1 He. tures, whereas an increased values for peaks associated with loss modulus and tan 6 are observed. The peaks of tan 3, for all the samples, falls within the usually reported temperature range of 70-80°C for DAM specimens. DOI 10.1002/pe Flexural Fatigue Behavior Ductile polymers subjected to cyclic loading at con- stant amplitude of strain exhibit four regimes of cycli stress-strain behavior; incubation, transition, cyclic steady state, and crack propagation terminated by fatigue frac~ ture. Some of the ductile polymers that are semicrystalline like polyamides do not exhibit cyclic incubation. This is ue to the presence of regions of high-local stress even at relatively low-applied body stresses [8]. Hence, in this study, each specimen exhibited three distinct regions, as the cyclic softening initiated in the first cycle and pro- ceeded through the transition region. ‘The maximum stress endured by a specimen made of PAGGCNS during completely reversed, cyclic loading at constant amplitude is shown in Fig. 6. Assuming the spec~ imen is near equilibrium during deformation, a linear viscoelastic specimen, the flexural stress was calculated from the measured values of resisting force offered by the specimen, for a particular end deflection, using Eg. / [23], Der o= (Dery 1 where M is the moment of force and is given by the rela- tionship P()*L, P being the force, L is the length of the specimen under flexure, / is moment of inertia, z is the distance from the neutral axis, ¢, i the stress, is eharac- teristic time, and D is the operator. Strain, « = Kz was calculated considering 2; w= "8S, where 2 is the distance from neutral axis, w is the deflection, x is the distance from the fixed end, and wo is the end deflection. All the material elements at any given fixed distance from the neutral axis, until initiation of a crack, undergo equal ‘magnitudes of tension and compression A significant rise in surface temperature with an equi- librium plateau is observed for all samples (Fig. 7). The surface temperature rise is due to the hysteretic heating exhibited by the polymer samples when subjected to a 60) 4 22 é = Book —eistoycle 6 + 10th cycle -40 + 100th cycle = 100th cycle 60h + 100001h cycle + 50000th cycle 80) 4 2 0 2 4 6 Strain (%) FIG. 6, Flenural stress-siran at the outer layer of PASSCNS sp under fully reversed eyelic loading with an end deflection of 6 mm. POLYMER COMPOSITES—2010 1981Intensity (a.u) 10 15 20 25 30 20¢ FIG, 3. XRD pattems of PAGS, PASGCN1, and PAGECNS (vertically shifed for clarity), its corresponding positions remaining unchanged. ‘The decrease in a peak intensity indicates the disturbance of the perfect arrangement of hydrogen atoms due to the addition of organoclay. There is no evidence for the exis- tence of 7 crystalline phase in significant quantity—a minor peak is observed at 26 = 13.6°. No perceptible increase in 7 phase can be seen in XRD patterns with increase in clay loading. Liu et al. [15], however, observed a significant y, at a 20 value of about 21.8° and a weaker v2 peak at 20 = 13.6° for 5% organoclay in PAG6. The degree of crystallinity of PAG6, PAG6CNI, and PAGGCNS samples was calculated by deconvolution of the XRD patterns (Table 1). A significant reduction in the degree of crystallinity is observed in PAG6 nanocomposite samples compared with that of pure polymer samples, similar to that observed by Chavarria and coworker [16] and Shen et al. [21] The crystallinity measures obtained by XRD match very well with the results obtained from DSC (Table 1). All DSC scans, irrespective of clay loading, exhibited only one sharp melting point at about 261°C, confirming the existence of a single x crystalline phase (Fig. 4) [19]. However, DSC scan of PAG6CNI shows a kink at about 252°C, unlike PAG6, indicating the presence of traces of crystalline phase. The presence of 7 crystalline phase can- not be confirmed using the results of DSC scan of PAGGCNS. Liu et al. [15] and Lin et al. [22] observed that 7 crystalline phase is present along with the exfol ated clay platelets of MMT in PA66, whereas Hedicke et al. [19] observed the absence of 7 crystalline phase when a mixture of intercalated and exfoliated structure of MMT is found in PA66, which is consistent with our find- ings. Our findings indicate that phase is thermodynami cally more stable than y phase. However, the trace amount of 7 phase is observed in PAG6CNI because of the fact that the exfoliated clay platelets must have pro- ‘moted the nucleation and growth of y phase by a small 1980 POLYMER COMPOSITES—2010 TABLE | Crystallinity results of PAGS, PAGOCNI, Ceystainity 6) Material XRD psc eo) psc PA66 Ma aa 261.7 PA@GCNI 244 249 20019 PA@GCNS 72 286 26017 amount. PA66 has a melting temperature (Tq) of 261.7°C, whereas Tq, for nanocomposites occurs at a slightly lower temperature. This phenomenon may be related to reduc~ tion in crystallite size in the presence of organoclay [22 In the case of PAGGCNI, the reduction in Ty is 0.8°C compared with that of pure PAGS. The crystallite size is not influenced by the intercalated organoclay present in PAG6CNS to the same extent as that of exfoliated organo- clay present in PAG6CNI. ‘The storage modulus, loss modulus, and tan 5 associ ated with dynamic mechanical thermal analysis are pr sented in Fig. Sa-c, respectively. The influence of clay on storage and loss moduli and tan 6 is significant over the entire range of temperature. For example, at 30°C, a sig nificant improvement in storage modulus of about 26% for PA66CNI and 35% for PA66CNS is observed. ‘The overall increase of storage modulus per unit weight of clay exhibited by PAG6CNI is higher than that of PAGGCNS. Chavarria and coworker [16] observed similar trend in the case of tensile properties for both PAG as well as PAG6 nanocomposites. The reason for this behav- ior could be attributed to the exfoliated and intercalated clay nanostructure observed in PAG6CN1 and PAG6CNS. respectively. A significant drop in storage modulus occurred for PAG6, PAGGCNI, and PAGGCNS at 7, because of the presence of significant quantity of amor- phous phase (~70%). As the clay loading is increased, the T,, obtained from storage modulus plots, and the peaks of loss modulus and tan 6 shifted toward lower tempera- 26089°C Heat Flow (W/a) © Bo 0 $50 200 250 300 Temperature (°C) FIG. 4. DSC heating scans of PAS6, PAGSCNI, and PAGSCNS. DOI 10.1002/pelinear sider cal FIG. 1. An in-house developed experimental setup for flexural fatigue tests. [Color figure can be viewed in the online issue, whic is available at wileyoalinslibrary.com.] loading of a specimen. A tension-compression load cell, linear displacement transducer, and an infrared tempera- ture sensor were used to measure the flexural force, deffection, and temperature of the specimen, respectively. A humidity sensor was fitted near the specimen to mea- sure the temperature and relative humidity of the cham- ber. Programming and logic circuit (PLC) controller and dala acquisition system were used to control the motion of the servomotor and to acquire data from various sen- sors, respectively. To avoid absorption of the samples were taken out of a desiccator (or a vacuum oven) only when an experiment is conducted and weighed to verify whether the moisture absorption is less than that of 0.2% (DAM requirement). A specimen is suitably held in a fix- ture for a flexural fatigue test and is enclosed in chamber to avoid heating or cooling of the specimen from an external source. All flexural fatigue experiments were conducted under completely reversed, deflection-con- trolled mode using a custom built table-top test rig at a frequency of 2 Hz. The first set of experiments was con- ducted at a constant end-deflection amplitude of 6 mm for specimens with a clay content of 0, 1, and 5%. The last set of experiments was conducted to determine the effect of hysteretic heating on PAG6CNS, at constant end-deflee tions of 4.5 mm, 3.0 mm, and 1.5 mm. All fatigue tests were conducted until failure or 10° cycles, whichever occurs first. The weight of a sample was measured after fatigue test and stored in desiccator, until fractographic studies are conducted. The average value and the standard deviation of the test results were reported in fatigue life studies. The fracture surfaces were studied using optical and high-resolution scanning electron microscope. joisture. RESULTS AND DISCUSSION Material Characterization The XRD @26 = 2.75 to 10°) patterns of PAGS, PAGGCN1, PAG6CNS, and organoclay samples were used to evaluate the dispersion of clay in a polymer matrix (Fig. 2). Since the reflections corresponding to crystalline Do! 10.1002/pe regions of PAG6 occur at 20 = 20,33” and 23.02”, the fol- owing discussion only applies to XRD pattem associated with organoclay. The organoclay exhibited two basal reflections; a well-defined peak at 20 = 3.36° and a weak broad peak at 20 = 6.7° corresponding to the basal inter- platelet spacing of door = 26.3 A and doo: = 13 A, respectively. The XRD pattern of PAGGCNI lacks basal reflection peaks observed in organoclay, indicating that the clay platelets are randomly dispersed in the polymer matrix, i.c., clay platelets are exfoliated. A broad peak was observed in XRD patter of PAG6CNS at around 20 = 4.6°, and it is identified as the reflection correspond ing to doo with a spacing of 19-A.[I7, 18], an increase of 6 A compared with that of pire organoclay, aiid usu ally PA66 molecules ate found between two clay platelets 13, 4], which is referred to as intercalation. Further, a peak corresponding to a reflection corresponding t0 doo} is not observed indicating that such a peak must occur at an angle less than 20 = 2.75°. Since the XRD pattern of PAGGCNS exhibits a spread out nature together with a weak reflection at 26 = 4.6°, one may conclude that the organoclay is both intercalated and exfoliated, However, the present findings further confirm the findings of Cha- varria and coworker [16] and Hedicke et al. [19], ic. the intercalated and exfoliated structure of organoclay dis- persed in PAG6 matrix. ‘The XRD patterns (20 = 10° to 30°) corresponding to that of PA66, PAG6CNI, and PAG6CNS samples (Fig. 3) exhibited two significant diffraction peaks at about 20 20.33° and 23.02", respectively, consistent with that of triclinic crystal structure (2 phase), namely, the 2; peak corresponding to reflection of successive planes (100) and a peak corresponding to reflections of successive planes of (010) or (110) [20]. PAGS is usually found in the stable a phase rather than less stable 7 phase. As the clay load- ing is increased, a decrease in the intensity of 2 peak and an increase in intensity of 2 peak are observed with = 2 z FIG. 2. XRD pattems of organoclay, PAGS, PASECNI, and PASACNS at low angles (intensity of organoclay is scaled for ela). POLYMER COMPOSITES—2010 1979response of polymer components. Crawford and coworker [9] performed cyclic uniaxial and bending tests on several thermoplastics. It was evident that the two modes of fail- ure were related to the magnitude of loss tangent of the material. In a recent study, Bellenger et al. [10] observed thermal and mechanical fractures at a frequency of 10 Hz, whereas only mechanical fractures were noticed at a fre- quency of 2 Hz. in PAG6/glass fiber composites, Mallick and coworker [11] reported fatigue failures at low stress levels and thermal failures at high stress levels at a fre- quency of 1 Hz, whereas fatigue failures alone at all stress levels at 0.5 Hz frequency. Initiation of failure by fatigue was observed at agglomerated nanoparticles in PAG matrix. Bellemare et al. [12] concluded that fatigue life increases at constant stress amplitude and decreases at constant strain amplitude. The improvement in fatigue life during tension-tension fatigue tests was credited to the decrease in the amplitude of displacement of macromole- cules. An increase in initiation life and decrease in propa. gation life of a crack was found and was attributed to an increase in “yield” stress (stress at which permanent set occurs), which favors the formation of micro voids ahead of the crack tip. Ramkumar and Gnanamoorthy [13] con- cluded from the axial tension-tension fatigue studies that incorporation of clay in PA6 decreased the strain ampli- tude compared with pure PAG, which results in less self heating and drop in dynamic modulus for a given constant cyclic stress amplitude. Many components made of plastics in service are sub- jected to flexural loading conditions in displacement con- trol. The fatigue loading behavior of plastics is different under defiection-controlled and load-controlled mode, i.e., for example, the extent of hysteretic heating (dissipation) will be different for both the modes, and hence the fatigue life, Therefore, it is necessary to study the cyclic loading behavior under deflection-controlled mode. The amount of clay and its nanostructure in polymer nanocomposite also influence the fatigue life. This article describes the influ- ence of clay content on the fatigue life of a polymer nanocomposite under deflection-controlled mode, EXPERIMENTAL Materials and Characterization The PAG6/hectorite nanocomposites with 1% organo- clay by weight (PAG6CN1) and 5% organoclay by weight (PAG6CNS) were prepared from commercially available pure PAG6 pellets and organically modified layered sili- cate hectorite clay (organoclay) by melt compounding technique. The PAG6 pellets were dried in a vacuum oven for 6 h at 80°C before the melt blending process. A coro- tating twin-screw extruder comprising of a screw diameter of 25 mm and length to diameter ratio of 48 was used for ‘melt compounding. The barrel temperatures were set as 260, 265, 270, 275, 280, and 285°C and the screw speed 1978 POLYMER COMPOSITES—2010 was s jon of silicate tat 150 rpm, The shear delami platelets from mineral aggregates occurs due to the shear stresses generated by high melt viscosity of polyamide. The resulting nanocomposite in the melt-like state was extruded, cooled, pelletized, and dried in vacuum oven for about 6 h at 80°C. ‘The dried pellets were injection molded to a cantilever specimen, The geometry of the specimen is according to ASTM D671 standard. The injection pressure was set at 12.5 MPa and the temperature was set at 280°C, After injection molding, the dry-as-molded (DAM) specimens were weighed to the nearest 0.001 g and sealed in poly- ethylene bags. The sealed bags were kept in desiccators in case of short-term storage and were kept in vacuum oven at 30°C for long-term storage to prevent the speci- mens from absorbing moisture. ‘The injection-molded dry plate surfaces of PAGS, PAG6CNI, and PAG6CNS, and organoclay powder sam- ples were characterized by X-ray diffraction (XRD) using Cu Kz radiation (= 1.5418 A) at an operating voltage of 35 KV and a current of 25 mA. These specimens were scanned from 20 = 2° to 10° to verify the exfoliation of clay, and 20 = 10° to 30° to determine the crystallinity, ‘The scan rate for entire range was I‘/min. XRD scans from 10° to 30° were loaded into commercial peak-fiting software to separate crystalline and amorphous peaks. The degree of crystallinity (X.) was determined by the ratio of sum of crystalline areas to the total area under curve [14] To study the influence of the layered silicates on ther- mal transitions and to confirm the degree of erystallinity obtained by XRD, the melt behavior of PAG6, PAG6CNI, and PAG6CNS was recorded using a differential scanning calorimeter (DSC). All DSC experiments were performed from 30 to 300°C at heating rates of 10°C/min in a nitro- gen environment with a sample mass of about 10 mg ‘The crystallinity of sample (X.) was calculated by taking AID, for PA66 as 195.9 J/g [15, 16]. Dynamic mechanical thermal analysis (DMTA) was performed under three point bending mode to determine the variation of visco- elastic properties such as storage moxtulus, loss modulus, and mechanical loss factors with temperature, and as a result, to understand its fatigue behavior. The storage modulus and damping factor of specimen of dimension 40 10 X 4 mm were recorded in the temperature range of 150°C to 150°C at a heating rate of 2°C/min, The data was obtained at frequency of 0.1, 1, 2, and 10 Hz. ‘The glass transition temperature (7) was obtained as per ASTM E.1640-04 standard. Flexural Fatigue Tests A flexural fatigue test rig, equipped with an environ- mental chamber, was developed to understand the micro mechanisms associated with the failure of polymeric materials under cyclic loading and is shown in Fig. 1. A precision ball screw and nut mechanism provides neces- sary reciprocating motion required for cyclic flexural OI 10.1002/peInfluence of Organoclay on Flexural Fatigue Behavior of Polyamide 66/Hectorite Nanocomposites at Laboratory Condition Mallina Venkata Timmaraju,’ R. Gnanamoorthy,’ Krishna Kannan’ ‘Department of Mechanical Engineering, Indian Institute of Technology Madras, Chennai 600 036, India Indian Institute of Information Technology Design and Manufacturing (IlITD&M), Kancheepuram, Indian Institute of Technology Madras Campus, Chennai 600 036, India Polyamide 66/hectorite nanocomposites exhibit supe- rior mechanical properties compared with pure poly- mers and are promising for structural applications. action results revealed reduced degree of ith i in organoclay content. Flex- ural fatigue characteristics of polyamide 66/hectorite nanocomposites containing different quantities of clay content were investigated, under deflection control mode, using a custom built flexural fatigue test rig. Addition of organoclay improved the moduli of the ma- terial. An enhanced resistance to cyclic softening was noticed at high temperatures with the incorporation of ‘organoclay. Nanocomposite samples exhibited a signit- icant improvement in fatigue life compared with pure Polymers; however, the degree of enhancement is governed by the nanostructure of organoclay in poly- mer matrix. The fatigue life of nanocomposite sam- ples is strongly affected by specimen temperature and induced stress. Macroscopic fracture surfaces changed from flat featureless structure to a highly perturbed structure with increase in organoclay con- tent. POLYM, COMPOS., 31:1977-1986, 2010. © 2010 Society of Plastics Engineers INTRODUCTION Polyamide (nylon) is a commonly used engineering plastic and is produced in large volumes [1]. In the poly- amide family, Polyamide 66 (PA66) is the most used ther- moplastic in automotive industry due to its favorable properties, iic., high strength and stiffness over wide range of temperatures, excellent toughness, and good chemical resistance [2]. Initially, they started replacing Correspondence 1: R. Gnaoamoonthy; e-mail: gmoorthy@iitm ain ‘Contract grant sponsor: GITAM University DOI 10.1002/pe.20997 Published online in Wiley Online Library (wileyoalinelibray.com), (© 2010 Society of Plastics Engineers POLYMER COMPOSITES—2010 metals in less critical components such as speedometer gears, bearings, bushings, ete. to reduce cost of the com- ponent and to minimize overall weight of a vehicle. Later, these plastics were used in applications that require resist- ance to heat such as radiators, fuel systems, air intake manifolds, etc. by reinforcing the plastic with microfillers like minerals and glass. Plastics filled with microfillers (about 20-60% by weight) showed an increase in density compared with that of pure polymers. Polymers filled with nanofillers (2-5% by weight) such as organically modified hectorite, MMT (Montmorillonite), etc. exhibited 1a decrease in material density compared with that of poly- mers filled with microfillers. Such plastics also demon- strated significant improvement in mechanical properties good thermal stability, reduced liquid and gas permeabil- ity, and improved flame retardency compared with pure polymers [3-5], and hence are used in automotive appli- cations such as engine covers, oil reservoir tanks, fuel hoses, and so forth [6]. Many engineering components experience cyclic load- ing and fail by fatigue. These components undergo hyster- esis heating due to their high damping and low thermal conductivity, which makes fatigue behavior of polymers quiet different from th: metal. Consta- ble et al. [7] subjected many thermo beams to cyclic bending tests and showed the existence of bimodal failure mechanisms. The first mode of failure, which is referred to as cyclic thermal softening was quantitatively related to loss compliance, specimen geometry, frequent and magnitude of the cyclic load. The second mode of failure is pertinent to fracture due to incremental crack propagation from microscopic cracks or defects produced during the manufacturing process or during service condi- tions. Axial strain controlled fatigue tests carried out by Rabinowitz and coworker [8] on a wide variety of poly- mers revealed the strong influence of thermomechanical on the eyclic history and ambient service environmet[22] Rite D. An investigation ofthe heat generated during cyclic lading of two sassy polymers. Mech Mater 200032131 47 {23} Suresh’s. Fatigue of materials. Cambridge: Cambridge University press: 1998, {2a} Freudenthal AM. Introduction to the mechanics of solid: New York John ‘Wiley & Sons ne: 1960. [25 FHljn. Fatigue damage, cack growth and fe prediction. Chapman & Hall In 1997. MY. Timmarjue a/incernational Journal of Fatigue 33 (2011) 541-548 {25} Erber Hysteresis and fatigue. Ann Phys 1993:224:157-92. [27] Tombins B, Biggs WD. Low endurance fatigue in metals and polymer. Pat stress/strain relationships. Mater Si 1969;4:532-8 [28] Kim GM, Coerta , Micher CH, Defonmalon mechanism of nylon Glayered silicate nanocomposites; role of the layered silicate. J Appl Poy Sei 2oos: 038-88[MY Timmaraj eta International journal of Fatigue 33 (2071) 541-548, sa the fatigue samples are subjected to cyclic flexure under varied surrounding media, Further, a variation in the hysteretic energy dissipation in air and mist Suggests a possible relation between hysteretic energy dissipation and fatigue life for PASGCNS samples based on the previous observations in polymers and metals [7,25]. Although the empirical relation arrived between fatigue life and hysteretic energy dissipation for PAGSCNS samples in air could able to offer the fatigue life under varied surrounding media, but this value is significantly higher than the experimental values (Fig, 10). The results other than 6 mm deflection amplitude in mist ‘were not considered while predicting the empirical relation be- ‘cause of a change in material of the specimen resulted due to mois- ture absorption, 3.3, Fractography In air fatigue tested samples did not exhibit significant stress ‘whitening, whereas in mist, PAGGCN specimens revealed an in- crease in stress whitening with increase in weight percentage of clay loading. The microstructures of failed fatigue specimens re- vealed distinct deformation and fracture mechanisms because of large difference between specimen temperature and glass transi- tion temperature (T,) in air and mist at high deflection amplitudes (Fig. 4). ‘The fracture surfaces of PAGGCN specimens exhibited an it creased fibrillated structure with increase in clay content [13]. The agglomerated particles present with some extent of orienta- tion to the injection direction in the matrix, appearing in the form of intercalated tactoids may act as stress concentration zones in PAGGCN [28], The nanoscale cavities formed from either splitting ‘or opening of bundles of the intercalated tactoids would join to- gether to form microvoids, which are responsible for crack initia- tion, when the specimen is subjected to continued cyclic deflection. The fatigue fracture surfaces of PAGGCNS samples failed due to large scale voids formed in the crack initiation region after extensive plastic deformation and a patchy structure with fine dimples pointed towards the fracture origin along the path of crack propagation in air (Fig. 11a). The extensive plastic defor- ‘mation around the large scale voids in the crack initiation zone reveals the rubbery behavior of the material, where the tempera- ture is much higher than T, [13]. This rubbery behavior delays the ‘macroscopic crack initiation near corner, where there is a higher stress concentration, Hence, a significant increase in fatigue life with an increased cyclic steady state region was observed for PAGGCN samples in air compared with mist, indicating an in creased initiation life. These observations are in good agreement ‘with the findings of Bellemare et al, [10], i. an increase in initi- ation life for PAG(clay nanocomposite samples subjected to con- stant axial cyclic stress amplitude in air compared with pure PAG samples, Instead, a relatively flat surface without a significant plastic deformation and shallow dimples at the crack initiation site was observed in PAGGCN5 samples tested in mist (Fig, 11b). In addition, the wide shallow dimples along the path of crack Propagation indicate a brittle fracture with glassy behavior as the temperature of the specimen during the test is much below , of the specimen [13]. PAGGCNS specimens tested in air revealed clear undulated crack growth bands perpendicular to the direc- tion of crack propagation along with secondary cracks in the crack propagation region (Fig. 11¢). The crack growth bands were dis- torted and deformed by the voids resulted from the organoclay platelets. PAGSCNS specimens tested in mist revealed a cleavage type of fracture during crack propagation with hazy crack growth bands (Fig. 11d). 4. Conclusions A clear evidence of the effect of surrounding media on the per formance of PAG6/clay nanocomposites during cyclic flexure was found from the fatigue studies carried out in laboratory air and ‘water mist under deflection control mode. Higher energy dissipa- tion per cycle in mist compared to that in air at cyclic steady state Jed to a significant drop in fatigue life for both pure and nanocom- posite samples. The present study showed that both relations based on inelastic strain amplitude and hysteretic energy dissipa~ tion are appropriate for nanocomposite samples in air, but not extendable to mist. An increase in the amount of clay content improved the fatigue life in both air and mist in a similar way. However, the rubbery behavior increased the fatigue life in air ‘compared with glassy behavior in mist. Acknowledgement ‘Support for one of the author (MVT) from GITAM University is acknowledged. References {1 Constable 1, Willams JC, Burns DJ. Fatigue and eyeic thermal softening of| ‘thermoplastic j Mech Eng Se 176 12:20-0. 2] Carlson E,Nelion K- Nylon under the hood: 3. history of Innovation. Delaware: Literature from El dupont de Nemours and Company: 2003, [3] Alexandre Mf, Dubois P. Polymer layered sicate nanocomposites: preparation, properties, and uses of new cass of material, Mater Sc Eng 200028: 1-63, 1] Sinha Ray's, Okamoto Mc Pelymerfayered silicate nanocomposites: a review {fom preparation te processing. Prog Palym Sel 2003:28:1539-641. [5] Srinath G, Gnanamoorthy R ffet of organoclay on the two bedy abrasive ‘wear characteristics of polyamide & nanocomposites.) Mater Set 2007:42; 520-3. 16] Zeng QH. Yu AB, Lu GQ. Paul DR. Cly-based polymer nanocomposite: research and commercial development. J Nanosci Nanotechnol 2005:5: 1574-92. 17] Rabinowitz 5, Beardmore P. Cyclic deformation and facture of polymers.) Mater Sei 1978:9:81-99. {8 Crawiord RJ. Benham PP. Some fatigue characteristics of thermoplastics, Polymer 1975;16:908-14, {9} BellengerV, Teharknch A, Castaing Ph. Thexmal and mechanical fatigue of AGBjslas bers composite materia. Int atigue 2006;28:1348-52 [10 Beliemare SC, Bureau MIN, Dena, Dickson J Fatigue crack initiation and propagation in poWamide-S and in polyamide-§. nanocomposites, Polym {Compos 2004:25:433-41 [14] Ramkumar A: Granamoorthy R. Axial fatigue behavier of polyamide-6 and polvamide-6 nanocomposites at room temperature. Compos Sci Technol 200868340165, [1 Rajeesh. KR, Gnanamaorthy R. Velmunugan R.Fifect of humidity on the indentation hardness and flexural fatigue behavior of polyamide © ‘anocomposite. Mater Sti Eng A 2010;527:2826-30. [13] Timmaraj MV, Gnanamoortny R. Kannan K. Influence of organoclay on Mlexural fatigue behavior of polyamide G6fhectorte clay: nanecomposits ¢oi:t0.1002/pe20007, [14] Wang G, Lio H, Saintiee N. Mai ¥. Cyclic fatigue of polymer nanocomposites fing Fail Anat 20082 10:2035-40, [15] Ratner 8, Barash NL. Efect of iquid media on fatigue strength of polyamides MekhanikaPolimeroy 1905:1:124-7, [16] E-Hakeem HA Culver LE. Environmental dynamic fatigue crack propagation in nylon 66. ft J Fatigue 1979:1-133-40, [17] Heatle JWS, Mirae M. The fex fatigue of polyamide and poyester fibers Part 1: the influence of temperature and Humidity. J Mater Set 1981-26: 2861-7 [18] ChavariaF, Paul DR. Comparison of nanocomposites based on nylon 6 and nylon 66, Polymer 2008;45:8501—15, [19] Hedicke i, Witten, Mehler , Gruber, Alstade V. Crystallization behavior of polyamide 6 and polyamide G& nanocomposites. Compos Sei Technol 2006;66'571"5. (20) Brusselle-Dupend Ny, Lal D, Feaugas XGulgon M, Clavel ML Mechanical behavior of 2 semicrystalline polymer before necking. Part | characterization of uniaxial behavir. Polym Eng Sci 2001;41:66~76 (21) Lai D. Yakimets 1 Guigon Mz A noalinear viscoelastic model developed for stimi-crystallin polymer deformed at_small strains with losding. and unloading paths. Mater Sei Eng A 2005;405:265-71546 MY. Timmaraj eo /Interatona journal of Fatigue 3 (2011) 541-548 10” TB PAGSCNS nav 3378456 mm ls 1 PAGECNS in mist 6 mm t a 3 3g : sU(e)=0.584(2Np7 0.545 g Bg we ‘ee No, of reversals to failure (2N) Fig. 10. No.of reversals to allure versus hysteretic energy dissipation for PAGECNS specimens subjected to constant cyclic end deflection of 3,375.4 and 6 mm in air and 6 nm alone in mist (average values of hysteretic energy disipaion pet tle at cyclic steady state and fatigue fe were considered) sipated hysteretic energy loss per cycle may be regarded as main contributor to the fatigue damage process taking place in each cy- cle [25]. Further, determination of the part of the total strain that is causing the damage is not straight forward for the polymeric mate- rials because both elastic and viscoelastic components contribute to the total strain, Therefore, a possible correlation between energy dissipation and fatigue damage for ductile polymers was proposed [7]. The Coffin-Manson type expression given in Eq. (5) directly links the hysteretic energy clssipation per cycle, the accumulation of material damage, and the average number of loading cycles leading to failure [25.26] ‘aU(e)) = DIN (el 6) where AU() is the average hysteretic eneray dissipation per cycle, Nj the average fatigue life, and D and d are the proportionality constant and index of power law similar to material ductility coef ficient and material ductility exponent in Coffin-Manson relation. It is not necessary to resort to double power interpolations because the relation based on hysteretic energy dissipation is appropriate for all regimes of fatigue [26] linear log-log relation was exhib- ited by PABGCNS in air between the average hysteretic energy dis- sipation per cycle and numberof cycles to failure as well (Fig. 10), relationship is in good agreement with the previous findings (25-27), PAGGCNS samples exhibited nearly equal magnitude of inelastic strain range in both air and mist for a constant cyclic end deflection {otal strain amplitude) at the same parameters of study (Fig. 3). However, the fatigue life exhibited by PAGSCNS samples tested i air is six times more compared with samples tested in mist at the same frequency (Fig, 6). It implies that the fatigue life cannot be predicted using a single Coffin-Manson type relation when Fig 11. Fatigue facture surfaces of PABECNS samples subjected to constant end deflection amplitude of 6 mm. (a) Large scale voids in the cack initiation region with extensive deformation and patchy structure along the path of erack propagation fr specimen tested in air: (b at structure atthe crack inition and along the path of rack propagation with rarely found shallow dimples for specimen tested in mist (c) wll deformed CBs perpendicular tothe direction of erack propagation separated by econdary cracks for specimen tested in air and (4) mild OCBs with a bite facture surface for specimen tested in mist (unidiretional arrows indicate direction of crack propagation).(ML, Tinmoraju et a Intemational Journal of Fatigue 33 (2011) 51-548 545 ay 1 700 To000 No. of cycles Fig. 7. Evolution of stress amplitude at the extreme outer layer of PASECNS specimens with respect to continued cyic end deflection of 375,45 and 6 mm in 7000000 gradual decrease in induced stress amplitude for the specimens subjected to lower deflection amplitudes indicates a continuous softening (ie. decrease in load bearing capacity of the specimen) ddue to absorption of moisture from the surrounding mist. This ‘was further confirmed by a significant increase in weight of fatigue samples atthe end of the test. A decrease in deflection amplitude ‘of the specimen increases the duration of test and hence permits 2 significant amount of diffusion of moisture, The diffusion of moisture is slow in circulated mist at ambient temperature, and is further slow in still air at ambient temperature and humidity. Hence, softening due to absorption of moisture, in terms of varia- tion in peak stress during the cyclic steady state, was not observed during the tests at constant deflection amplitude of 6 mm in mist and at all deflection amplitudes in air. ‘A steady state cyclic stress-strain relation existed in the fully softened state is significantly influenced by the environment dur- ing cyclic flexure at constant end deflection of 6 mm, The energy dissipated in driving the specimen through a complete cycle re- mains constant and dynamic strain recovery processes balance the cyclically applied strain to give a constant cyclic inelastic strain range in this cyclic steady state regime [7] Hysteresis laops in the cyclic steady state regime exhibit a consistent increase in inelastic strain and hysteretic energy dissipation with increase in deflection amplitude in air (Fg. 8). The effect of temperature on modulus is evident in the hysteresis loops in terms of decrease in the slope of the loops with increase in cyclic end deflection, Strain-based approaches employ a linear log-log relation be- tween elastic or plastic strain amplitude and fatigue life based on the findings of Basquin inthe high cycle fatigue regime, and, Coffin. and Manson in the low cycle fatigue regime. The Coffin-Manson ‘equation together with modified form of the Basquin equation leads to the equation of the strain life curve i.e., the fatigue resis- {ance relationship in terms of the total strain amplitude is given by the Eq, (4) [23.25} ty _ Bey, Aty zat s ‘where «is the total strain range, «is the elastic strain range, pis the plastic strain range, 6; is the fatigue strength coefficient, E is the Young's modulus of the material, 2N;is the number of cycles G any? + Nye Np) + e/(2Npy a 2%” =" 6 mm(-87 °C) 2o|---> 4.5mm(~71 °C) 3 mm(~62 °C) 40 | 1-5mm(-20 °C) ° i, 20 0 i ao a eS strain (4) Fig. 8. Cyclic steady state stess-sain curves at the extreme outer layer of PABBCNS specimens subjected to constant cyclic end deletions of 15, 3,45 and mm in ai, to failure, bis the fatigue strength exponent, ¢;is the fatigue ductil- ity coefficient and c is the fatigue ductility exponent. The Coffin— ‘Manson (plastic) model is considered here because of the presence of significant inelastic strain in terms of viscous strain that can influence the fatigue life similar to that of the plastic strain because both are dissipative in nature. Therefore, the plastic strain term, é in the equation is replaced with a more appropriate term, inelastic strain, ¢, The variation of the elastic, inelastic and total strain ampli- tudes with the number of reversals to failure for PAGGCNS in air is shown in Fig. 9. The coefficients and exponents were evaluated by a best fit technique. Though the strain-based approach is essentially introduced for thermal and low cycle fatigue, this approach showed better results in high cycle fatigue regime as well. This could be due to the presence of a substantial component of inelastic strain even in high eycle fatigue regime resulted from the viscoelastic nature of the polymers. On the microscopic level, the irreversible nature of the micro- inelastic deformation caused by each cycle of loading is associated with the dissipation of strain energy (hysteretic energy). This dis- 10 © Elastic strain © Inelastic strain 10} 4 Total strain § (82/2)=7.551(2N)°** (a f2)=(06,/2)4(d6/2) (0¢,/2)=0.665(2N,)°** 10' 10° 10° 10° 10° 10° No, of reversals to failure (2N) Fig 8. The variations of the elastic, inelastic and total strain amplitudes a the extreme outer layer as Funetions of the mumberof reversals to falae based on E (4forPAGOCNS specimens in ar (average values of elastic inelastic and total stain amplitudes at cyci steady state and fatigue hie were considered).g 2 PASS in air a PASSCNI in air = PASSCNS in air = PASE in mist -S PABSCNI in mist -S— PABECNS in mist Temperature (Cy 3 8 3 1 701001000 No. of cycles e000 +0000 Fig. 4 Evolution of surface temperature for PASS, PASGCNT and PASGCNS specimens with respect to continued cycic end deflection of 6am in ar [13] and A decrease in peak stress with increasing cycles before reaching the cyclic steady state is termed as cyclic softening; described phe- nomenologically as a decrease in material resistance to inelastic strain with reversed deformation [7). Region I in Fig. 5 illustrates tthe cyclic softening exhibited by PA6, PAGECNT and PAGBCNS in air and mist. Region Il in Fig. 5 illustrates the cyclic steady state exhibited by the samples. However, cyclic steady state is signifi- cantly longer in air compared with that in mist. The induced stress amplitude in mist is decreased by 25%, in contrast to the 65% de- crease in stress amplitude in air [13]. A large decrease in stress amplitude i. significant cyclic softening in air is due to a signifi- cant rise in temperature of the specimen. This greater cyclic soften- ing in air leads to lower hysteretic energy dissipation in cyclic steady state (Fig, 3). A near isothermal condition of the specimen in mist restricted the cyclic softening in PAGS and PASECNI. Whereas, a noticeable cyclic softening and a clear region of cyclic steady state exhibited by PAGGCNS is due to a perceptible increase in the temperature of the specimen in transient state. The fast frac~ ture stage is clearly evident for each specimen in Region Ill in Fig. 5 PAG6CN in mist PAGCNS in mist 1 0 1001000 No. ofeycles 10000 Fig. 8. Evolution of stress amplitude atthe extreme outer layer of PASS, PABGCNY ind PABGCNS specimens with respect to continued cyclic end deflection ofS mm in 2 [13] and mis. MY. Timmaraju eo,/nternationel Journal of Frgue 33 2011) 541-548 in terms of sharp decrease in stress amplitude to zero in both air and mist. The effect of the humidity, etc. is to alter the boundary condi- tions associated with the specimen, in that, the surrounding media changes the heat transfer coefficient, and obviously, the evolution of the temperature of the sample. In other words, the effect of humidity (Wt% of mist in air in this case) of surrounding media and specimen temperature on fatigue life cannot be separated ‘while other test parameters like frequency and amplitude of alter- nating deflection and temperature of surrounding media are fixed The humidity of surrounding media influences the fatigue life through the specimen temperature only. One can explain such a correspondence between the surrounding media and CSS temper- ature as follows: polymeric materials are dissipative in nature ‘when subjected to cyclic loading. During an initial set of (i. first 1000) cycles, an imbalance between the hysteretic heat generated in the specimen and heat loss to the surrounding media results in specimen temperature rise (Fig. 4). The softening of specimen asso- cated with rise in specimen temperature balances heat generation and loss in order to attain the cyclic steady state (CSS) (Fig. 5). Therefore, in Fig. 3, the differences in specimen temperatures at CSS in air and mist are ditectly correspond to the variation in envi- ronment. Further, a significant difference in energy dissipation of specimens at CSS in air and mist led to a major variation in fatigue life between the two environments (Fig. 6). PAGGCN samples exhibited higher fatigue lives even with increased induced stress amplitude in air and mist compared with PAGS samples for the ‘same applied deflection amplitude. However, the amount of in- ‘crease in fatigue life with incorporation of organoclay depends ‘on the quantity and its nanostructure in polymer matrix. However, the specimen material itself changes due to the signif- icant influence of hygroscopic nature of the PAGG in the case of ‘specimens exhibiting sufficiently longer fatigue lives resulted from low deflection amplitudes at high humidity levels. PA66 absorbs ‘moisture from the surrounding media and exhibits a large decrease in strength and modulus with a simultaneous increase in ductility. ‘The absorbed moisture strongly influences the fatigue behavior of PAG6 as wel [15,16]. Hence, each specimen is weighed after the fa- tigue test to find the amount of moisture absorbed. None of the specimens revealed significant moisture absorption after fatigue tests in both air and mist at end deflection of 6 mm. A significant variation in fatigue behavior with decrease in deflection amplitude in mist is shown in Fig. 7. After 10,000 cycles, in spite of constant deflection amplitude and temperature being nearly the same, a © PABECNS in mist [Eo i gy PREC mit PAE nist r je PABECNS nar Hi Pasedit in ai re Pass ai "000 Tatoo Tene No, of eytes to failure ig. 6, Fatigue life for PAB, PABECNI and PABECNS specimens subjected to fll ‘reversed constant cj end defection of mm in air [13] and ms.(MY. Tmmarju et a/tarernationa Journal of Fatigue 33 (2011) 541-548 58 Relative Intensity (a.u,) 207) Fig.2. RD pattems of organoclay, PA, PAGGCNI and PAGECNS at Iw ante. ‘element in parallel to an elastic spring is capable of describing the mechanical behavior of polymeric materials adequately (20,21), and therefore, such a model is used to determine the inelastic strain, The strain associated with the Maxwell spring is elastic strain, whereas the strain associated with the dashpot is viscous strain (dissipative in nature). The Maxwell spring is much stiffer than the elastic spring. Therefore, the breadth of the hysteresis loop at zero stress indicates an inelastic strain, which is approxi- mately equal to the strain associated with that of the dashpot. ‘The viscous strain is referred hereafter as inelastic strain. In ductile polymers like PASS, the initial cyclic stress-strain re- sponse is unstable with regard to constant cycling; both the cyclic hysteresis and inelastic deformation undergo significant changes prior to fatigue crack initiation and propagation [7]. The rate of hysteretic energy dissipated per unit volume of the material, Q during constant strain amplitude fatigue is given by Eq, (1) [9,22] meg a ‘where ¥cis the number of strain cycles per unit time, is the loss modulus and sy is the cyclic strain amplitude, In ambient air, the heat dissipated by the polymer fatigue specimen neither rises the temperature of the specimen alone (adiabatic condition) nor lose to the surroundings alone (isothermal condition) Hence, the time rate of change of temperature ofthe polymer fatigue specimens gi- vven by Eq, (2) [23} dt _myE' HA a pe pet a ‘where p is the mass density, cis the specific heat, H is the heat transfer coefficient for loss of heat from the specimen surface to the surroundings, A and Vare the surface area and volume of the fa- tigue test specimens, and T and To are the instantaneous specimen temperature and the ambient temperature. The dynamic mechani- ‘al thermal analysis results at 1 Hz for the similar nanocomposite system as that of the present study revealed a significant increase in storage and loss moduli and a shift in T, towards lower temper- atures with increase in the amount of organoclay [13] The stresses at the outer layer are calculated from moment, based on the formulation for linear viscoelastic solid and the strains are determined from deflection [13,24]. The area enclosed by the stress-strain hysteresis lop indicates the hysteretic energy dissipated per cycle per unit volume and is given by Eq. (3). Fo ‘where ois the instantaneous stress and ¢ is the instantaneous strain. The area of stress-strain hysteresis loop decreases and the inelastic strain increases with decrease in peak stress due to a raise in specimen temperature and mechanical instability of material ‘with respect to continued cyclic deflection, Fig. 3 illustrates the s nificant changes between initial and steady state stress-strain hys~ tetesis loops for PAGBCNS specimens subjected to constant cyclic fend deflection of 6 mm in both air and mist. Similar behavior was ‘observed in PAGS and PAGGCN1 as well, The hysteretic energy dis- sipated during the first cycle in both air and mist is neatly equal However, the amount of hysteretic energy dissipated during cyclic steady state compared with first cycle is significantly less in aie and more in mist, Table 1 gives the fractional areas of cyclic stress-strain hysteresis loops ie, ratios of area of cyclic stress- strain hysteresis loop in cyclic steady state to first cycle, for PAGE, PAGCNI and PAGCNS samples, The heat generated in specimen is much higher than the heat loss when the surrounding medium is air. Hence a significant rise in the temperature of the specimen was observed in air before reaching cyclic steady state at deflection amplitude of 6 mm [13] ‘On the contrary, the specimens exhibited a near isothermal condi- tion in mist with only a marginal rise in surface temperature of the specimens other than PAGGCNS before reaching cyclic steady state in mist because of rapid extraction of heat from the specimen sur= face by the water mist (Fig. 4). A slight increase in temperature of PAGBCNS specimens during first 1000 cycles in mist is due to a per~ ceptible imbalance between the hysteretic energy loss and heat loss to mist. However, an increase in the amount of rise in surface ‘temperature of the specimen with increase in organoclay content ‘was observed in both environments because of an increase in loss ‘modulus of PA66CN with increase in the amount of organoclay. ® |_—sstytein air OF. ast cycle in mist sok" C85 mai(-87 °C) [--- CSS in mist(~27 °C)} g* Eo z & 0 a) 0 2 ae eee ae Stain (%) Fi, 3. Intl and steady state cyte stress-strain curves a the extreme outer ayer Of PASGCNS specimens subjected to constant cyclic end deflection of mm in ai table ato of 23s of sess-strai hysteresis loops at eye steady state wo steele for PAGO, PABST and PASGCNS in ir and mist environments. Specimen material Factonal area i steady sate ir Mis Come SORiether) ES ESM Pate eeu ues aes a ueae eur tulssu Fig. 1. An in-house developed flexural fatigue experimental setup consists of an environmental chamber.men by the envitonment during the cyclic loading. Alternatively, investigations on notched PAGG specimens did not show any potential change in the crack growth behavior with change in sur- rounding media from air to water, as long as there is no significant variation of water content in notched specimens of PAS6 [17]. In notched specimens, the cyclic deformation confines to the crack tip region and the bulk of the specimen acts as heat sink. Whereas, the entire specimen experiences the cyclic deformation in un- notched specimens, results in large amount of heat dissipation Hence, a possible variation in fatigue behavior in un-notched specimens cannot be neglected with change in surrounding media, The present study aimed at understanding the influence of envi- ronment on flexural fatigue behavior of polyamide 66jclay nanocomposites (PAG6CN) containing 0, 1 and 5% weight of organoclay by conducting the deflection controlled flexural fatigue tests in still air at laboratory condition and in circulated water (deionised) mist. 2. Experimental Commercially available PAGG pellets and organically mocified hectorte clay (organoclay) were dried in an even for Ghat 80°C and melt compounded in a co-rotating twin screw extruder with LUD ratio of 48. Two types of nanocomposites with 1% organoclay by weight (PAGGCN1) and 5% organoclay by weight (PAG6CNS) were prepared. The extruder was operated at a screw speed of 150 rpm. Alter extrusion, the pellets weve dred ina vacuurn oven for 6 h at 80°C. The dried pellets were injection molded into stan- dard uniform strength cantilever fatigue specimens (ASTM D671). The pressure and temperature used for injection molding were 1255 MPa and 280°C respectively. The dry as molded (DAM) spec- imens were weighed using an electronic balance of accuracy 0.0001 g after injection molding, and stored in a vacuum chamber at 302 1 °Cin sealed covers. X-ray diffraction (XRD) studies were carried out using iron radi- ation (= 1.9373) at 30KV, 207A to evaluate the degree of organoclay exfoliation. The diffractograms were scanned in the 20 ange of 210", aa scanning rate of 1"/min. A table-top flexural fatigue test ig, containing an environmental chamber around the specimen holder was developed to study the variation in fleural fatigue behavior of polymeric materials with change in surround- ing media from still air at laboratory condition to circulated water ‘mist (Fig. 1). A tension-compression load cell, linear displacement transducer, and an infrared temperature sensor were used fo mea- sure the flexural force, displacement and temperature ofthe spec- ‘imen, respectively. A hygrothermal sensor was fitted near the specimen to measure the temperature and relative humidity of ‘chamber, A single stage compressor and pump supply the com pressed air and water respectively to the mist spray gun. All exper= iments were conducted under completely reversed deflection control made at a frequency of 2 Hz. Experiments were conducted in still air (30+ 1°C and 60 5RH) and in circulated water mist (261°C and 100%RH) at constant end deflections of 1.5, 3, 4.5 and 6 mm, and, 3.75, 4.5 and 6 mm, respectively. A near isothermal condition of the specimen is achieved in water mist, therefore, could able co observe the effect of organoclay on fatigue behavior ‘of PAGGCN specimens in the absence of significant specimen temperature rise. Fatigue tests were conducted up to failure or 10° cycles, whichever occurs first. A minimum of three specimens ‘were tested at each condition and the average and standard deviation values of these test results were reported in fatigue life studies. The weight of the specimen was measured before and after the fatigue test. Fatigue fractured surfaces were investigated using scanning electron microscope to understand the effect of surrounding media on failure mechanism, 3. Results and discussion 3.1. Characterization of materials The XRD patterns of PAGS, PAGSCN1, PAGGCNS, and organoclay are shown in Fig. 2. Organoclay exhibited two basal reflections; a ‘well defined peak at 20 = 457° and a weak broad peak at around 209° correspond to respective basal inter-platelet spacing of {iggy = 24.3 And dogo ~ 12.35 A. The XRD pattern of PABBCNI lacks basal reflection peaks observed in organoclay, indicating that the clay platelets are randomly dispersed in the polymer matrix ie, the clay platelets are exfoliated. Whereas, both basal reflection peaks of organoclay appeared in the XRD pattern of PABBCNS sam- ple with a shift ie, at 29 = 2.63° and 5.26”, indicating an increased basal spacing (00 1) of about 18 A This increased basal spacing has, resulted from intercalation of PAG6 molecules into the gallery spacing of organaclay platelets, The present findings are in good agreement with the observations of Chavarria and Paul [18] and Hedicke et al [19] indicating a mixture of intercalated and exfoli- ated structure in melt compounded PAGE nanocomposites with ~5 we% montmorillonite organoclay. 3.2, Flesural fatigue behavior The mechanical behavior associated with polymeric material such as PAGS is viscoelastic in nature. A three parameter solid is uused to describe the hysteresis loops associated with the flexural fatigue. Such a three parameter model consisting of one MaxwellEffect of environment on flexural fatigue behavior of polyamide 66/hectorite nanocomposites Mallina Venkata Timmaraju*, R. Gnanamoorthy*, K. Kannan? *Depertment of Mechanical Engineering Indian Insite ofTechology Made, Chena-600 036, indi nie Insite of formation Techoouy Design and Manufacturing (UITDBA Kandheepurar. I Madvas Campus. Cunn-600 036. nda ARTICLE INFO ApsTRACT ‘ie istry Receive 17 June 2010 Received in reise form 15 October 2010 Accepted 19 October 2010 ‘Available onine 23 October 2010, Keynorts Polym max composts Senin controlled fatigue Fexualloading Environmental eects Mirorcony Polymer nanacomposites are promising for structural applications due to their high modulus and good corrosive resistance. The influence of environment on flexural fatigue behavior of polyamide 66/hectorite nanocomposites has been investigated instil air and circulated water mist environments, High hyster- tic energy dissipation per cycle and high induced stress in mist lead to a significant drop in fatigue life In mist compared with that in ait. The stain and hysteretic energy dissipation per eycle against fatigue lie curves exhibited linear relationships oo deformation and fracture mechanisms in air and mist. 1-log scale im air. The failure analysis revealed a distinct © 2010 Elsevier Ltd. Al eights reserved, 1. Introduction Thermoplastic components in automotive and other applica- tions are often exposed to varied ambient service conditions. The effect of surrounding conditions on the performance of these com- ponents varies depending on the loading condition and exposure duration, These components also experience variable loads in ser~ ‘vice and fail due to fatigue. The components made of polymers subjected to variable loading undergo hysteretic heating due to their high damping and low thermal conductivity, which makes the fatigue behavior of polymers different from that of metals. ‘Thermoplastic components subjected to cyclic loading exhibit a bimodal failure mechanism [1]. The first mode of failure is by dis tortion (thermal failure), because of localized rises in temperature. ‘The second mode of failute is pertinent to fracture (mechanical failure) due to incremental crack propagation from microscopic cracks or defects produced during the manufacturing process or by service conditions. The transition in failure mode depends on loss compliance of the material, specimen geometry. frequency and applied stress. Polyamide 66 (PASS) is the widely used thermoplastic due to their versatile nature, easy to mold, high strength and stiffness over a wide range of temperatures, excellent toughness, and good chemical resistance compared with other polymers [2], Further, © Corresponding author Tel lan: 191 44 2257 4601 Email adress pmoorthyitm.acin(R.Cnanamocrhy) polyamidejorganoclay nanocomposites (PACN) containing 2-5% ‘lay demonstrated a significant improvement in mechanical prop- erties together with thermal stability, reduced liquid and gas per- meability, improved flame retardency compared with pure polymers [3-5]. Therefore, in recent days, PACN are being used far automotive underhood applications like engine cavers, ail res- ervoir tanks, ete. and under consideration for various applications by automotive and other industries (6). Polyamide (PA) specimens subjected to cyclic loading revealed a strong influence of thermomechanical history an the eyclic re- sponse of polymer components [7,8], Further, a marginal increase in fatigue strength along with reduced thermal softening was attributed to the stress gradient exist through the thickness of the specimen. Similar to pure polymers, reinforced polymers are also susceptible to fatigue, however the mechanism of failure dif fers due to the existence of complex state of stress. PAGB/short ‘lass flber composites exhibited bimodal failures with a significant increase in test frequency irtespective of the strain amplitude [9]. Recent investigations on PACN demonstrated improved fatigue characteristics compared with pure PA and further revealed that the transition in mode of failure depends on test frequency, the thickness of the specimen, type of load and parameter under con~ trol (stress or strain) {10-14} Previous investigations on dry un-notched PA specimens and PA fibers revealed a variation in fatigue strength and a strong depen- dence of flex fatigue behavior, respectively, with variation in sur- rounding media {15.16}. The vatiation in fatigue strength was attributed to the difference in the amount of cooling of the speci