FATIGUE OF ADHESIVELY BONDED

FATIGUE OF ADHESIVELY BONDED THERMOPLAST Alisa Biichman, Arieh Sidess, Hannah Dodhik RAFAEL, P.O.Box 2250, Haifa 31021, Israel

on the material properties: the t\pe of adlie treatment and adliesive, on test conditions; mean stress, R ratio, and on external varia temperature, Long temi performance of structural adliesives is a humidity. Fatigue resistance complex problem especially when fatigueeither and empirically by evaluating the number failure as a function of stress amplitude (Sdurability are involved. Static and dynamic behaviour of theplastic mechanistic method by defining fatigue ductile and tougli adliesives on metal and propagation a function of stress intensity adlierends were tested using an empirical approach. as The objective of tliis work was to provide fatigue use of metal adherends and thernio- plastic adlierends adhesiveofjoints for quantitative and qualitat enabled to distinguish between the contribution the pure prediction studies by conducting cyclic fatig adliesive and the joint in fatigue. Lap shear joints were control and practical adhesive joints. tested and the failure was analj'zed. Results showed that all joints had a relatively low endurance limit (30% of UTS). The tough adhesive had a higlier endurance limit but was more sensitive to liigh dynamic loads. Defect nucleation was probabty the dominant mechanism ofExperimental fatigue in the adliesive joints. A model predicting the Materials life term was developed. It can be concluded that and specimen preparation fatigue performance is a severe problemThe in adliesive type of adliesive joint used in the expe joint durability. double lap shear (DLS) specimen obtaining mode II frature under static and fatigue tes Introduction to differentiate the pure behaviour of the ad fatigue from that of the adlierent. Aluminum adherends treated with cliromic anodization Adhesive joints offer many potential advantages over and loaded. Than,intwo kinds of thermoplas conventional mechanical fasteners. Recent development were tested a Lexan 9034 ( composites and light weight materials processing havepolycarbonate, led polyetherimide, Ultein 1000 (PEI), both pro to increase use of structural adhesives in the aerospace, General Electric. Two adliesives of the sam automotive, sport and appliance industries as structural (polyurethanes) were employed: components. Despite of the wide application of adliesive two part ductile adliesive Uralite joints, a relatively limited amount of dataan onmifilled long term durability of different adlierends, adhesives, of Hexcel stressUSA, modes, and a tough filled tvvo part enviromnents or strength fracture criterions, areof available. product Helmitin Germany. The surface Adhesive failure is usually a result of crack SiCgrowth. blastingBeing (38 mesh). subjected to cyclic fatigue stress is therefore a particularly The specimens were bonded under consta severe condition for the durability of an adhesive joint. MPa) using a special jig and cured at room Such conditions are common in many engineering for 48 hours. The nominal bond tliickness w applications. The adhesi^'e joint may fracture even though and the over lap was 2.8 cm^. The specime the applied cyclic stress amplitude is well below cured at the 50°C for 15 days and aged in ai ultimate strength of the adliesive joint. Adhesive joint before testing. behavior is significantly different from that of metals or pure polymers which fail in fatigue only by crackand Fatigue tests Static initiation and growth. Adhesive joints also exliibit other modes of damage including, crazing, interfacial failure, Experiment included static and fatigue test debonding, void growth and multidirectional cracking [1]both adhesives. adlierends and The objecti (see Figure 2a). The futigue resistance of the(ajoint depends test shear to failure test according to AS

Abstract

was to measure the shear strength of the The joins filled in order tough to adhesive is more sensitive define the parameter of 100% stress forhigh the cyclic stresses loading. than the ductile one. The end The crosshead speed of the Instron machine T700 was is 3.62 times higher than that of U-312 mm/min. Calculating the strains of both adliesives at The dynamic tests were conducted at room in nearly temperature similar values (deviation of ±15% according to ASTM D-3116 (empirical approach) stresses at atfailure a were totally different. Th sinusoidal tension-tension cyclic frequency the of failure 30 Hzcriterion and in fatigue of the adhesiv stress ratio R=0.1. Five samples were tested failureforstrain. each stress level. All tests were performed under condition of constant load with an MTS 880 servohydraulic testing macliine. of failure b. Morphology The effect of test frequency on fatigue behaviour was also evaluated. Mode of fracture was observed visually The mode and of failure did not change during with Scanning Electron Microscope (SEM). The data tests. SEM niicrographes of U-3125 failure obtained in the fatigue experiments wasshowed analyzed and that atahigher fatigue stresses and model was developed which predicts fatigue lifeonly of the cycles a few initiation points of defects adhesive joints. but the damage they caused was severe an oriented. At lower stresses and higher num Results and Discussion many defect initiations were observed but t destructive. A mild defect was characterize and small cracks, while a severe defect sho and entangling of the adliesive layer, torn t Static Shear Tests deformations (figs. 2a,b). SEM niicrographe showed a the more brittle mode of failure. At h Static tests were conducted in order to determine stresses and number of cycles, the fatig ultimate stress (UTS) to which the fatigue loads were lowsurface on the fractured were in a parallel related. Table I suimnarizes the static results for both the stress direction), and their edges were adlierends. All failures propagated cohesively thougli the composed of several secondary layers. The T700 adliesive layer and interfacially at the U-3125 defects was relatively low. At lower stresse adliesive/adlierend bond. nmnber of cycles, significantly more defects the ridges appeared in different angles Fatigue Shear Tests- Al Adherendsand direction. The edges of the ridges were sha and showed pure shear failure of 45°. Altho a. Mechanical test results. seemed brittle, a closer observation of the ductile features of dimples spread all over t The first experiments were conducted at 50%number of the of defects in T700 compared ultimate static stress (UTS). The nmnbertotal of cycles to higlier, probably the presence of fille failure were determined to be 50,000 winch is within the due to the voids which originated nucleation of de standard range of 2x10 ^ to 1x10 ^ cycles to failure. 3a,b). Accordingly, higher and lower loads were applied and number of cycles to failure were recorded. The effect resultsof material parameters c. The were plotted on a S-N curve (stress amplitude, S,vs. number of cycles, N). Both adhesives showed an samples, large diviations from the For a few endurance limit at 30% of their static ultimate shear values were observ^ed. Closer examination strength. samples revealed two sources for these res An interesting phenomena observed during fatigue test layerthe of adliesive (0.15-0.2 nun. compared of the ductile adliesive (U-3125) was a gradual decrease and the presenceofof large voids in the adlie the load close to the failure altliougli a constant was affectedload the number of cycles to failure by a applied. Thus, the failure was not iniiiiediatly observed in order of magnitude. This indicates that fatig metals. It can probably be a result of stress relaxation more sensitivein to geometrical changes or m the adhesive during fatigue. No such phenomena was test (where deviation is only than any static observed in the tougli adliesive (T700). The luglier sensitivity of fafigue results prob In order to compare both adhesives the higher stresses (S) were of defects and cracks at i nucleation normalized to their static ultimate stressborders (So) andof voids due to stress concentration presented as a fimction of N (Fig. 1). The results show statistically higlier concentration of defects that both adliesives have the same endurance limit (30%). adliesive layers. The dynamic movement du

prevents recovery and leads to an accelerated It should progress be noted that an interesting phen towards failure concerning the shape of the loading wave during the fatigue test with plastic joints. W d. The effect, of frequency joints the Sin. wave was clear and sharp, f joints the Sin. wave was composed of coin Adliesives in structural components experience waves (adynamic small wave on top of the main wa forces at various frequencies. Thus, the effect effect was of different also observed in the pure adliere frequencies was tested for both adhesives (15,30tension. and 40 A possible explanation is dynamic Hz.) at 40% of UTS. The resuhs are presented in stress table 2.is directly transferred into joints the The results show an opposite effect of frequency on the are obsen^ed, wlule in p phase differences two adliesives: in the case of the tough adliesive, higher viscoelastic behaviour causes the stress to frequency increases fatigue strength. The fatigue occurs in along the joint obtaining phase differences the elastic region, so that increase in frequency results in above mentioned phenomena. increase of toughness [3] (less sensitivity to fatigue). In the case of the ductile adhesive fatigue occurs in the b. Morphological observation viscoelastic region enabling recover)^ at the rest periods, at lower frequency recovery is more effective andanah'sis fatigue of the fracture surface of U-3 SEM strength increases. showed parallel lines of crack propagation stress direction. The edges of the stripes in behaviour (drowi and torn adliesive and sm Fatigue Shear Tests - Plastic Adherends a fine secondary structure of fatigue striatio the fatigue load, the denser are the cracks a. mechanical test results the finder is the secondary structure. observation of the fracture surface of The results of applied stress as function SEM of number of cycles to failure were presented on S-N similar curves to forthat bothof Al joints as the failure is co adliesive. adliesives and adlierends. The endurance limit obtained for both adliesives on plastic adherends was close to that of Al adlierends (28% of UTS). Examination of Model the S-N for Predicting Fatigue curves showed that it is actually composed of two combined curves. The upper curve (lughFig. loads) is less the fatigue results of the of 5 presents sensitive to fatigue than the lower curve.PEI Theand twoU-3125 cun'cs / PC. Adopting an empirica result from two different fatigue mechanisms observed in the number of cycles to fa relation betiveen the failure locus. At higli stresses the failure occurs in the certain level of applied stress (S) is presen adliesive wlule at lower stresses in the adlierends. This assumption was proved by conducting fatigueNtests f = c on ( l / the S) (1) unbonded adlierends (tension-tension) and comparing the S-N curves. The adherend's curve fitted directly vsdth the c and m are material constants. lower cmve (Fig. 5). The transformation Where between the two leads to the followin curves for the T700 / PEI joint was less Logaritlunic pronounced regression than constants for the U-3125 / PC joint. This result is due to the presented in table 3. these constants the fatigue performa mechanical similarit}' in properties of theUsing adhesive T700 predicted and it's adlierend PEI compared to the other joint.to any given level of load. The em compared to the experimental results is sh In order to compare both joints the stresses (S) were There is good agreement between them. T normalized to their static ultimate stress (So) and that the presented as function of the number of cycles to lifetime failure of the tough adhesive T700 stress level is liiglier (N), (Figs. 4,5). The results show that both joints have an than that of the ductile constant c), wlule the rate of change in endurance limit of 28% of UTS and T700(see is more cycles a fimction of stress level is similar sensitive to fatigue at higlier stress compare to as U-3125, adhesives. but these differences are minor compared to those obtained for the Al adlierends. The endurance limit of T700 is 1.7 times higher than Conclusions that of U-3125 (on the plastic adherends). The variations from AI joints results from weaker adliesion toThe theevaluation plastic of adliesive joint strength in adlierend and the additional effect of fatigue on the plastic investigated with various adlierends and ad adherend itself. an empirical approach.

Table 1: The of results of the static tests for the Fatigue performance proved to be a severe problem joints.to cracks ; and adhesive joint durability due to its sensitivityother defects. Adherent (surfaceAdhesive UTS (MPa) A similar endurance limit was obtained for all joints treatment) (~ 30% of the ultimate strength). This limit is mostly acid U-3125 Al (Chromic etch) 6.5±1 (a) * important in engineering design as it detennines a safety T700 Al (Chromic acid etch) 19.5±1.1 (c) factor of 3-4 for designing structures subjected to dynamic U-3125 4.6±0.3 (a) Lexan (SiC abraded) stresses. T700 6.8+0.3 (c) Ultem (SiC abraded) There is a major difference in the fatigue behaviour *a-interfacial failure, c-coliesive failure between the two adhesives tested. The tough adliesive T700 had an endurance limit 3.6 times higher than that ofresults, nmnber of cycles, a Table 2: Fatigue the ductile adliesive on Al adherends, and 1.7 times frequencies on plastic adherends. Adliesive U-3125T700 The fatigue perfomiance of the tough adhesive at high Frequency stresses is lower than that of the ductile adhesive but its 40 Hz. 142 K 336 K predicted lifetime is longer. 30 Hz. is 141 K 362 K The strain to failure in fatigue for both adhesives Hz. is strain. 829 K 39 K similar thus a reasonable fatigue failure 15 criteria The fatigue behaviour depends significantly on material, geometrical and experimental factors such as 3: adhesive and Table Material constants of the model. adlierend materials, surface treatment, presence of cracks Adliesive U-3125 T700 and voids, adliesive layer tliickness and fatigue frequency. Constant Morphological observations indicated that at low fatigue 2.568. 10 stresses nucleation of defects dominates while atchigh 1.504. 10" stresses localized damage in the adliesive layer m is the 5.4367 main 6.6112 cause for failure. An emperical model for predicting fatigue failure was developed.

Acknowledgment The authors wish to thank Ms. Irit Liran for her teclinical assistance, Mr. Y. Cohen for operating the MTS, and Dr. N. Nir for helpfiil discussions tlirougliout this research.

Bibliography 1. J. A. Sauter, "fatigue of polymers" in Encyclopedia of as Materials, Vol. II, 1680 (1989). 2. J. P. Trotigiioii, Plasty- a Kaucuk, 11-12, 0.7161 (1992). 3. N.Y. Chung, R. Yuuki, H. Isliikawa, S. Nakaiio, Key Eng. Mat, 51-52, 185 (1990).

m 3125/3119 T70OILFSO0

Key Words: Fatigue, adhesive joints, shear, model.