WELDING RESEARCH

NOVEMBER 2017 / WELDING JOURNAL 429-s

Introduction Energy saving, emission reduction,and lightweighting have recently de-veloped into major concerns in the au-tomobile industry. Lightweight poly-mers, such as carbon-fiber-reinforcedpolyamide 66 (Cf/PA66), are good can-didates for fabricating automobilecomponents (e.g., bumpers and body-in-white) because of their advanta-geous combination of light weight andmechanical performance (Refs. 1–3).Similar to metal component fabrica-tion, complex plastic components of-ten consist of small parts. Among all the bonding techniquesavailable, ultrasonic welding (UW) isan efficient method (Refs. 4, 5) be-

cause it avoids the disadvantages oftraditional techniques, such as exten-sive surface preparation, long curingtime, and extra weight increase (Refs.6, 7). This technique is fast, energy ef-ficient, suitable for mass production,and offers a good cosmetic quality(Refs. 8, 9). The quality of the ultrasonicallywelded Cf/PA66 composite is usuallyassessed by a static tensile test (Refs.10, 11). However, the Cf/PA66 com-posite is increasingly used in produc-ing structural parts, such as intakemanifolds or energy mounts, that aresubjected to complex and repeated me-chanical loading (Refs. 12, 13). There-fore, the fatigue property of weldedplastics has become a serious issue,

and different researchers have begunto study this problem. Tsang et al. (Ref. 14) discussed thefatigue behavior of vibration-welded,glass-fiber-reinforced nylon 6 and ny-lon 66 under high and low pressureconditions. Although welding pressureaffected fatigue strength, all weldedspecimens exhibited similar fatiguebehavior under the tension-tensionload ratio of R = 0.1. Additionally,Lockwood et al. (Ref. 15) investigatedthe effect of service temperature onthe fatigue strength of vibration-welded, glass-reinforced nylon 6. In-creasing the temperature led to a signif-icant decrease in both tensile strengthand fatigue life, but the endurance ratioof the fatigue strength to static tensilestrength was approximately 45% re-gardless of the temperature under sinu-soidal constant amplitude tension-tension load at a stress ratio of R = 0.1. According to the aforementionedstudies, welding pressure and servicetemperature can affect the fatigueproperty of the joint (Refs. 14, 15). Inaddition to these factors, the jointconfiguration can function as an influ-ential element. A shear lap configura-tion without an energy director is rec-ommended to manufacture large andcomplex structures in the automobileindustry, even though energy directorsare beneficial to high heating efficien-cy and weld quality (Refs. 16, 17).However, molding energy directors ona large weld is difficult and can compli-cate the molding process. Shear lap joints are efficient, easy tooperate, and without strict require-ments in part dimensions. However,compared with the butt joint, thestress condition for a lap joint configu-ration is more complex. Sharifimehr et

Effects of Preheat Treatment on the Ultrasonic Welding of Carbon­Fiber­Reinforced

Polyamide 66 CompositeThe mechanical properties, temperature evolutions, and microstructures of ultrasonic­

welded joints at different preheating temperatures were investigatedBY Q. ZHI , X.-R. TAN, AND Z.-X. LIU

ABSTRACT Microstructure and mechanical properties of ultrasonically welded carbon­fiber­reinforced polyamide 66 composite in a shear lap configuration were studied by analyz­ing and comparing the welds made from preheated and nonpreheated workpieces. Pre­heat treatment of workpieces slightly affected tensile strength, whereas its influence onthe fatigue property was significant for the joints made with optimal welding time. Thecomparable tensile strengths were attributed to the similar welding areas, and thefatigue property was correlated with the temperature profile during welding. Thetransient temperature curves during ultrasonic welding, along with the fourier transforminfrared spectroscopy (FTIR) analysis of the composite, showed that the decrease in thejoint’s fatigue property was mainly caused by the decomposition of the composite at ahigh temperature (above the decomposition point of PA66; i.e., 375°C) in the middle ofthe upper workpiece near the periphery of the horn. Combination analyses of the timeto obtain a stable maximum welding area and decomposition occurrence time in themiddle of the upper workpiece showed that the suitable preheating temperature for ul­trasonic welding of carbon­fiber­reinforced polyamide 66 fell in the range of 95° to145°C. This finding was attributed to the avoidance of decomposition in the compositeand the decrease in the joint’s temperature gradient during ultrasonic welding. The jointswelded with preheated workpieces of 125°C exhibited the highest endurance limit,which increased by 30% compared to the nonpreheated workpieces.

KEYWORDS • Ultrasonic Welding • Preheat Treatment • Fatigue Property • Microstructure

Zhi 201734.qxp_Layout 1 10/10/17 3:08 PM Page 429

WELDING RESEARCH

WELDING JOURNAL / NOVEMBER 2017, VOL. 96430-s

al. (Ref. 18) reported that the fatigueproperty of welded lap shear speci-mens was inferior to that of the buttjoint due to stress concentration.Therefore, the fatigue property of thelap shear joint must be improved. There are numerous methods avail-able to improve weld fatigue propertyincluding grinding treatment (Ref. 19),mechanical impact treatment (Ref. 20),and preheat treatment (Ref. 21). Pre-heat is one of the most common meth-ods applied for pretreatment; it reducesthe temperature difference and im-proves thermal stress in the joint, there-by improving fatigue performance (Ref.21). Moreover, preheating can minimizethe moisture in the specimen, whichcan significantly affect the mechanicalproperty of polyamide (Ref. 22). In this study, the potential of pre-heat treatment to achieve desirablemechanical properties of ultrasonic-welded 30 wt-% Cf/PA66 compositewas investigated. The mechanicalproperty, temperature evolutions, andmicrostructures of the ultrasonicallywelded joint subjected to different pre-heating temperatures were tested andanalyzed.

Material and Methods

Injection­Molded Carbon Fiber Composite

Commercial polyamide 66 and car-bon fiber (24 K, T300 type, Toray Car-bon Magic Co. Ltd.) with a length of 2mm and a diameter of 7 m were used.

The fibers were firstcleaned with a concen-trated solution of nitric acid and thensurface pretreated using 8% diglycidylether of bisphenol solution in acetone.Both the polyamide 66 and pretreatedcarbon fiber were dried at 80°C undervacuum condition for 3 h prior to use infabricating the carbon fiber/ polyamide66 composite with 30 wt-% fiber. A twin-screw extruder with two sep-arate inlets was used to mold the 30wt-% Cf/PA66 composite with dimen-sions of 132 × 38 × 4.0 mm. The carbonfiber was homogenously distributedwithin the polyamide matrix. Allcoupons were stored in an ambient lab-oratory environment (25°C and 50%relative humidity) and dried in a vacu-um oven at 80°C for 48 h before weldingto completely remove moisture in thespecimen.

Preheat Procedure The samples were heated to a presettemperature and stored for 30 min ina furnace to achieve homogenousheating prior to welding. To maintainthe preheating temperature, a warmair blower was turned on when thesamples were fixed. A thermocouplewas used to measure the temperatureof the workpiece, after which weldingwas conducted.

Ultrasonic Welding

The UW process was performed us-ing a KZH-2026 multifunction machinewith a nominal power of 2.6 kW, nomi-

nal frequency of 20 kHz, and nominalamplitude of 25 m (Weihai KaizhengUltrasonic Technologies Co. Ltd., Chi-na). The welding setup used in thisstudy is schematically shown in Fig. 1. The machine was equipped with adata acquisition system combined witha pressure sensor, horn-displacementsensor, and timer, which were integrat-ed in the controller of the UW machine.The welding pressure (i.e., gauge pres-sure), welding energy, and horn dis-placement were recorded online using aPC as a function of time by the data ac-quisition system. The final horn dis-placement, welding energy, weldingtime, welding pressure, hold time, anddelay time were also displayed in thecontrol panel during UW. The couponswere held in place using a fixture toavoid motion during welding. The machine had three weldingmodes: energy, time, and horn dis-placement. The values of welding ener-gy, welding time, and horn displace-ment for the three modes were presetto control the welding process. Theworkpieces were then welded usingthe nominal power of the machine.When the welding energy, weldingtime, or horn displacement reachedthe preset values for the selected weldmode, the ultrasonic wave oscillationwas stopped. Thus, weld quality wascontrolled by the preset values in eachselected welding mode. When the timemode was selected, the values of thedelay time, weld pressure, hold time,and ultrasonic time were preset prior

Fig. 1 — Schematic for ultrasonic welding of injection­molded carbonfiber/polyamide 66 composite (dimensions in mm).

Fig. 2 — Sketch of the temperature measurements duringthe ultrasonic welding of injection­molded, 4­mm­thicklapped carbon fiber/polyamide 66 composite without an en­ergy director (dimensions in mm).

Zhi 201734.qxp_Layout 1 10/10/17 3:09 PM Page 430

WELDING RESEARCH

NOVEMBER 2017 / WELDING JOURNAL 431-s

to welding. When ultrasonic triggeringwas performed, the horn was pressedonto the workpieces for 2 s and thenultrasonically vibrated until the presettime was reached. The welded work-pieces were held for 3 s to solidify themolten material. All specimens werewelded using a 7075 aluminum hornwith a diameter of 18 mm.

Temperature Measurement

To analyze the influence of preheat-ing temperature during UW on tensileand fatigue properties, the tempera-ture evolutions at the faying interfacenear the periphery of the horn (loca-tion 1) and in the middle of the upperworkpiece near the periphery of thehorn (location 2) were measured. Fig-ure 2 shows the experimental setupfor temperature measurements. Asmall hole with a diameter of 1.0 mmwas drilled at the side of the middle-upper workpiece. Another hole was

drilled at 0.2 mm from the bottomsurface. Two K-type thermocoupleswere embedded into the two smallholes and secured with epoxy com-pound to fix the thermocouples. Thetemperature sampling rate was at aninterval of 50 ms, and the temperatureprofiles at these two locations wererecorded as a function of time using adata acquisition system during UW.

Shear Tensile and Fatigue Tests

The shear tensile and fatigue testswere performed by an MTS 810 servo-hydraulic testing machine at ambienttemperature. To minimize the bendingstresses inherent in the test, two fillerplates were attached to both ends ofthe specimen using masking tape toaccommodate for sample offset. Forthe shear tensile test, the specimenswere loaded at a stroke rate of 2.0mm/min. Three replicates were per-formed, and the average joint strength

was reported. For fatigue test, load-controlled sinusoidal tension-tensionfatigue tests were carried out using asinusoidal waveform and in tension-tension mode. The cyclic frequencywas 30 Hz, and sinusoidal load cycleswere applied with the load ratio R =0.1. The fatigue test was either run un-til fatigue failure or terminated whenthe number of cycles reached 5 × 106

(runout). During the load controlledfatigue tests, the axial strain wasmeasured. Maximum and minimumstrain values were recorded for eachload cycle, whereas full stress-straincycles were recorded at fixed time in-tervals. Three joints were tested ateach maximum load.

Characterization of the Material

The morphologies of the welds afterthe fatigue tests (i.e., the fracturedworkpieces and faying surfaces) werecharacterized by scanning electron mi-croscopy (SEM, JSM 6700F). All sam-ples were sputter-coated with plat-inum for 50 s before the SEM analysisto induce conductivity. Thermogravi-metric analysis of Cf/PA66 was per-formed by TG/DTG techniques (a Lin-seis STA PT 1600). The heating ratewas 5°C/min under air, and the initialsample weight was approximately 10 mg. The chemical structure of the poly-mer surface was characterized usingthe fourier transform infrared spec-troscopy (FTIR) Perkin Elmer Spec-trum One. Pellets made of Cf/PA66powder were diluted in KBr, and theFTIR spectra were recorded between4000 and 400 cm−1. The powder wasobtained by grinding in an agate mor-tar with pestle under an infrared lamp. Dynamic mechanical analysis wascarried out to investigate the effects oftemperature and frequency on the vis-coelastic properties of the Cf/PA66composite. Specimens with dimen-sions of 38 × 8.5 × 4 mm were subject-ed to three-point bending with a spanlength of 20 mm. An oscillating forcewas applied (maximum of 4 N) to yielda constant deflection amplitude of 30m. Measurements were conductedover the temperature range of 23° to200°C with a heating rate of 2°C/minand under fixed frequencies (1, 2, 5,and 10 Hz). The moduli at 20 kHz wereextrapolated by time-temperature

Fig. 3 — Effect of welding time on peak load and welding area of joints fabricatedwith various preheated workpieces.

Table 1 — Fatigue S­N Curve Data for Ultrasonic­Welded Cf/PA66 with Various Preheating Temperatures

Preheating Coefficient A Exponent Endurance Limit (MPa) R2

Temperature (MPa) b Based on 5 × 106 Cycles (°C)

25 10.5 –0.12 10 0.94 75 11.5 –0.14 8 0.98 125 8.43 –0.09 13 0.97 175 10.82 –0.15 6.6 0.93

Zhi 201734.qxp_Layout 1 10/10/17 3:10 PM Page 431

WELDING RESEARCH

WELDING JOURNAL / NOVEMBER 2017, VOL. 96432-s

superposition.

Results and Discussion

Effect of Preheat Treatment onJoint Tensile Strength

Three preheating temperature levels(i.e., 75°, 125°, and 175°C) were select-ed to investigate the effect of preheattreatment on the tensile strength ofthe ultrasonic-welded Cf/PA66. Thenonpreheated joint (25°C, ambienttemperature) was also evaluated forcomparison. Figure 3 presents the in-fluence of different preheating temper-atures on the peak load and weldingarea of the joints with different weld-ing times under 0.15 MPa. Both the preheated and as-welded(i.e., nonpreheated) welds followed asimilar trend as the welding time var-ied. In particular, the peak load initial-ly increased with the welding time andthen decreased, whereas the weldingarea increased first and then ap-proached a plateau. As expected, theoptimal welding time for the joint withmaximum ultimate tensile strengthdecreased with increasing preheatingtemperature. The optimal weldingtime for workpieces preheated at vari-ous temperatures (25°, 75°, 125°, and175°C) were 2.1, 2.1, 1.5, and 1.3 s, re-spectively. The maximum peak loads ofall-welded workpieces were compara-ble mainly due to the similar weldingareas (Ref. 23). The failure mode of the joints wasalso analyzed. Visual examinations ofthe fractured joints indicated that thejoints fabricated with the optimal weld-ing time, as well as those that exceededthe optimal welding time, were likely toexhibit workpiece breakage regardless ofthe preheating temperature. By con-trast, the joints with insufficient weld-ing areas (less than the optimal weldingtime) likely displayed interfacial failure— Fig. 4A. Interfacial failure occurredmainly because the insufficient weldingarea could not bear the tensile force,breaking through the weld. With a sufficient welding area, thestress distribution of the welded jointduring the tensile test under a givenstatic load was modeled with the finiteelement method using the commercialsoftware ANSYS 15.0. The welding arearemained virtually constant at 450

mm2, and the thickness of the weldjoint was approximately 0.15 mm basedon the microstructural observation ofthe ultrasonic-welded joints. Thus, acylinder with a diameter of 24 mm anda height of 0.15 mm was used to simu-late the weld area of the ultrasonic-welded joint, and placed directly in theweld zone to present the weld joint. For simplification, the Cf/PA66composite with 30 wt-% fiber was con-sidered an isotropic material mainlybecause of the homogenous carbonfiber distribution in the matrix. Thetype of element used for analysis wassolid-brick 8 node 185, and a freemesh (smart size 1) was used formeshing. To simplify the simulationsof the weld joint in favor of the ANSYScapacity for solving these long-form-ing path parts, the lower workpiecewas considered fixed in the x (widthdirection), y (length and tensile direc-

tion), and z directions (thickness direction), whereas the upper work-piece was fixed in the x and z direc-tions as boundary conditions. Thestress distribution during the tensiletest at the faying interface is present-ed in Fig. 4B. When the welding areaapproached a plateau and carriedforce, the stress was concentrated se-verely along the vicinity of the weld re-gion during the tensile test. The work-piece was likely to break at the maxi-mum stress position in the region ofstress concentration.

Effect of Preheat Treatment onthe Joint Fatigue Strength

A shorter welding time is preferredwhen the tensile strength of the jointis similar because of its efficiency andbecause a longer time would causethermal decomposition of the material

Fig. 4 — Failure modes and stress distribution during tensile testing of the ultrasonic­welded Cf /PA 66 composite.

Fig. 5 — Fatigue life data for the lap­welded Cf /PA66 composite with various preheatingtemperatures. Each arrow denotes one runout specimen.

A B

Zhi 201734.qxp_Layout 1 10/10/17 3:14 PM Page 432

WELDING RESEARCH

NOVEMBER 2017 / WELDING JOURNAL 433-s

(Ref. 24), which is deleterious to themechanical properties of the joint. Ac-cording to the results in Figs. 3 and 4,all joints made with the optimal weld-ing time exhibited workpiece break-age, and the corresponding tensilestrengths were 36, 35, 37, and 32MPa, respectively. The tensilestrengths for joints with preheatingtemperatures of 75° and 175°C wereslightly lower than those of the as-welded joint and the joint from theworkpieces preheated at 125°C.

The ‘‘strength-life equal rank as-sumption’’ (Refs. 25, 26) proposes thata material with great tensile strengthwill possess a long fatigue life. The jointsmade with their optimal welding timewere adopted to investigate the fatigueproperty of the ultrasonic-weldedCf/PA66 composite. Notably, thermalfatigue occurred when the surface tem-perature in the lap section during thefatigue test approached or exceeded theglass transition temperature (Tg) (Refs.14, 27, and 28).

Therefore, the temperature in thelap section during the fatigue test wasmeasured by an infrared thermometerand fluctuated within 25° and 33°C,which was below Tg of the Cf/PA66composite (73°C). This characteristicconfirmed the absence of thermal fa-tigue, and all specimens failed by me-chanical fatigue. Figure 5 presents theresults from the fatigue test of thewelded specimens. The S–N curves (stress vs. numberof cycles to failure) were fitted with aBasquin equation of the followingform (Ref. 14):

Smax = ANbf (1)

where Nf is the number of cycles tofailure, and A and b are coefficients ob-tained from the experimental data.The coefficients, endurance limits, andcorrelation coefficients (R2) are listedin Table 1. These coefficients are strict-ly fitting parameters that are validwithin the high-cycle testing rangeand do not necessarily possess a physi-cal meaning. Values of A varied withthe preheating temperature, and theexponent b showed comparable values.The S–N curves of different preheatingtemperatures were nearly parallel toone another. As shown in Fig. 5 and Table 1,stress decreased with the number ofcycles under all conditions. The pre-heating temperature exerted a com-plex effect on the fatigue life and fa-tigue strength of the joint. Comparedwith the as-welded joint, the en-durance limit decreased from 10 MPaof the as-welded joint to 8 and 6.6MPa for the joints with preheatingtemperatures of 75° and 175°C, re-spectively. By contrast, the preheatingtemperature of 125°C improved fa-tigue performance. The joints after the fatigue testshowed a fracture mode of workpiecebreakage, which usually occurred inthe upper workpiece. This was proba-bly because the upper workpiece dissi-pated more heat than the lower work-piece. The fracture surfaces of the bro-ken workpieces with various preheat-ing temperatures were examined, andthe results are shown in Fig. 6. The morphologies of the fracturedworkpiece surfaces preheated at 25°and 125°C were similar, and bothmacrostructure and microstructure

Fig. 6 — Morphology of the fractured surface for upper workpieces with various preheat­ing temperatures after the fatigue test.

Fig. 7 — Temperature histories at locations 1 and 2 for joints made with various preheat­ing temperatures: A — 25C; B — 75C; C — 125C; D — 175C.

A

C D

B

Zhi 201734.qxp_Layout 1 10/10/17 3:13 PM Page 433

WELDING RESEARCH

WELDING JOURNAL / NOVEMBER 2017, VOL. 96434-s

appeared to be compact. By contrast,for workpieces preheated at 75° and175°C, some pores distributed in themiddle of the workpieces. The pres-ence of pores in the workpiece may de-crease the mechanical property ofCf/PA66. Thus, the tensile strengthsfor joints with preheating tempera-tures of 75° and 175°C were slightlylower than those of workpieces with-out pores. These characteristics may berelated to the temperature history ofthe joint during welding, which will bediscussed in the following section.

Effect of Preheat Treatment onthe Temperature Profile

Based on the aforementioned re-sults, preheating exerted a more pro-found effect on the fatigue propertythan on the tensile strength of the joint.The maximum tensile strength forjoints made with preheated and nonpre-heated workpieces ranged within 32 to37 MPa, whereas the fatigue life differedsignificantly. The difference in the ten-sile strength and fatigue property wasprobably correlated with the tempera-ture history of the joint. To investigatethe effect of the preheating temperatureon the material and joint, the tempera-ture distribution at locations 1 and 2(Fig. 2) were measured. The results areshown in Fig. 7A–D. The thermogravimetric analysis ofthe Cf/PA66 composite with 30 wt-%carbon fiber was conducted to show thethermal property of Cf/PA66. The deriv-ative weight curve is the differential of

the weight curve,which denotes thechange in theweight loss rate.The mass loss andthe derivativeweight of the com-posite as a func-tion of tempera-ture are presentedin Fig. 8. The de-composition ofPA66 started at ap-proximately 375°C(5% weight loss),and the rate of de-composition in-creased with theincreasing temper-ature. According tothe temperatureprofiles shown in Fig. 7, the tempera-ture at locations 1 and 2 initially in-creased with the welding time and de-creased after the vibration stopped.The temperature at location 1 in-creased faster than that at location 2,which was mainly because of existingfrictional heat and viscoelastic dissipa-tion at the faying interface, whereasonly viscoelastic dissipation was pres-ent in location 2. A comparison of the temperaturehistories of joints made of workpiecespreheated at 25° and 75°C revealedthat the temperature in location 2 in-creased more rapidly, and the peaktemperature exceeded the thermal de-composition temperature of the

Cf/PA66 composite. This quick increasewas likely related to the viscoelasticproperty of the material. Figure 9shows the storage and loss moduli ofthe Cf/PA66 composite as a function oftemperature at 20 kHz. The maximum loss modulus oc-curred at about 73°C, which is Tg. Thecorrelation between the loss modulusand viscoelastic heat could be expressedas the following equation (Ref. 29):

where Q is the average power dissipat-ed, = 2f, f is the applied frequency, is the maximum of the strain, and Eis the loss modulus. The welding timeof the ultrasonic process was short

Q = ��2E''2

(2)

Fig. 8 — Weight and derivative weight curves for the carbonfiber/polyamide 66 composite with 30 wt­% fiber.

Fig. 9 — Temperature dependence of the storage and loss moduliof the Cf /PA66 composite at 20 kHz.

Fig. 10 — FTIR spectra of the Cf /PA66 composite at the middle ofthe fractured upper workpiece.

Derivative

Zhi 201734.qxp_Layout 1 10/10/17 3:13 PM Page 434

WELDING RESEARCH

NOVEMBER 2017 / WELDING JOURNAL 435-s

(several seconds), and the temperatureof the material increased rapidly. Thisduration was too short for the materi-al to transform its state during weld-ing, so the initial loss modulus ofCf/PA66 (after preheating and beforeUW) was applied as the loss modulusof the composite during welding. For the joint made with the work-pieces preheated at 75°C, a large lossmodulus was observed, indicating greatviscoelastic dissipation within the mate-rial. Therefore, the temperature in loca-tion 2 increased faster than that in loca-tion 1. The energy conducted to theweld zone decreased because more ener-gy was dissipated by the material. Thus,the temperature rising rate at the fayinginterface (i.e., location 1) decreased (Fig.5B). As the preheating temperature fur-ther increased to 125°C, the loss modu-lus decreased, and the temperature atlocation 2 decreased below the decom-position point. The loss moduluschanged slightly with a preheating tem-perature of 175°C. However, the initialtemperature of the workpiece was high,so decomposition temperature was easi-ly reached or exceeded under ultrasonicvibration. Based on the temperature profilesin Fig. 7B and D, it can be reasonablyassumed that the pores in the upperworkpiece (shown in Fig. 6) were at-tributed to the thermal decompositionof the polyamide 66 composite. To val-idate this assumption, the materials inlocation 2 for joints subjected to vari-ous preheating temperatures were

characterized using FTIR. The resultsare presented in Fig. 10. Absorption bands around 3300and 3400 cm−1 in Cf/PA66 correspond-ed to hydrogen-bonded N-H stretchingand free N-H stretching, respectively(Ref. 30). The peaks at 1650 cm−1 wereascribed to C = O stretching (Ref. 31).Comparing these spectra, the intensi-ties of absorption peaked at 3300 to3400 cm−1, and 1650 cm−1 changedwith the preheating temperature. Theintensities of the peaks were similarfor 25° and 125°C preheating tempera-tures, and decreased gradually for pre-heating temperatures of 7° and 175°C.These characteristics indicated thatthe number of hydrogen-bonded N-Hgroups decreased, demonstrating thatthe polyamide was partly thermallydecomposed (Ref. 30). This character-istic would cause a loose structure anddecrease the properties of the Cf/PA66composite (Ref. 32). As observed fromFig. 6, a relatively large amount ofpores formed during welding, especial-ly for a short duration of 1.3 s. Thisphenomenon will be elaborated uponlater in this study.

Effect of Preheat Treatment onthe Faying Surface

The different temperature profilesduring welding could produce a pro-found influence on the structure andproperties of the joint, which wereclosely correlated with the structure ofthe weld joint. Thus, the faying sur-

faces of the joints after fatigue tensiletesting were examined. The results arepresented in Fig. 11. Some small voids were observed inthe faying interfaces of the workpiecessubjected to 25° and 125°C preheating,whereas the small voids grew intolarge pores (marked as porous region)for those treated at 75° and 175°C. Aporous region appeared within theweld region. The microstructure of theweld region was relatively compactwith a small quantity of voids, whichwas likely due to slight decompositionor shrinkage cavities (Ref. 31). The car-bon fibers at the faying interface reori-ented themselves toward parallel dis-tribution to the weld joint. The mi-crostructure of the porous region wasloose with large pores and cracks.Combining the temperature profileand microstructure shown in Figs. 6and 7, the porous region was probablycaused by the thermal decomposition.In addition, the porous region wasmainly concentrated within the weldregion, which could be explained bythe simple illustration in Fig. 12. As noted from the aforementionedresults, the faying interface exhibitedthe maximum temperature, which in-creased through friction and viscoelas-tic dissipation. The melted material atthe faying interface flew along theweld joint bilaterally under the weldforce. Some voids appeared in the weldjoint, as shown in Fig. 12A. For jointsmade with workpieces pretreated at 25° and 125°C, the temperature in themiddle-upper workpiece was below thedecomposition temperature, and noobvious pores were observed. Thus,the weld formation process was simi-lar to that shown in Fig. 12A, and noevident porous region was observed atthe faying interface. Figure 12B shows the weld forma-tion for joints with temperaturesabove the decomposition point in themiddle of the upper workpiece. Thematerial in the middle of the upperworkpiece produced some voids due topolyamide decomposition, but thevoids could not have been instantly ex-pelled out. The voids grew into largepores, and the melt fluidity increasedwith the increase in temperature. Thepressure in these pores was large, andthe large pores extended to the fayinginterface, which squeezed the meltedmaterial out. As the welding proceed-

Fig. 11 — Morphology of the fractured faying surface for welded workpieces with variouspreheating temperatures after fatigue testing.

Zhi 201734.qxp_Layout 1 10/10/17 3:13 PM Page 435

WELDING RESEARCH

WELDING JOURNAL / NOVEMBER 2017, VOL. 96436-s

ed, an increasing amount of the mate-rial was squeezed out. As a result, aloose porous region appeared in theweld region, and a large amount ofpores was produced in the middle ofthe material. Figure 12C shows the microstructureof the joint made with the workpiecespreheated at 175°C, thereby confirmingthe assumption on weld formation.These results indicated that a high tem-perature (i.e., above the decompositiontemperature of 375°C) within the upperworkpiece was detrimental to the weld.The temperature within the upper work-piece mainly depended on the viscoelas-tic dissipation of the material. As a re-sult, the suggested preheating tempera-ture should focus on those ranges wherethe loss modulus of the material is rela-tively small. According to the above results, thejoints fabricated from workpieceswithout preheating and subjected to apreheating temperature of 125°Cshowed superior microstructure andmechanical properties. In theory, thelarger temperature gradient in the ma-terial is more likely to introduce ther-mal stress concentration (Ref. 33).Consequently, pretreatment at 125°Cis supposed to be an effective methodin terms of efficiency and long-termservice. To validate this theory, the mi-crostructures of these joints subjectedto 1.75 kN after 1 × 106 cycles were ex-amined. Notably, the stress concentra-tion at the final solidification zone(weld end) was the most severe posi-tion due to the inhomogeneous heat-ing (Ref. 34, 35). Thus, the microstruc-tures of the weld ends with and with-out preheating are shown in Fig. 13.Cracks were present in the upperworkpiece after 1 × 106 fatigue cyclesfor the joint without preheating,whereas no visible cracks were pro-

duced for the jointfabricated from thepreheated work-piece of 125°C. With the preheattreatment, the tem-perature gradientsin the thickness di-rection (perpendicu-lar to the tensile di-rection) and lengthdirection (tensile di-rection) were small-er during coolingcompared with the nonpreheated weld(Fig. 7A and C). Accordingly, the tem-perature gradient at location 1 for pre-heating temperature of 125°C wassmaller than that at 25°C, which mini-mized thermal stress and hinderedcrack initiation and propagation.These results confirmed that preheat-ing at 125°C was a suitable method toimprove the fatigue life of ultrasoni-cally welded Cf/PA66 composite.

Selection of Preheating Temperature

Preheat treatment significantly in-fluences the fatigue property of ultra-sonic-welded joints. Therefore, select-ing the optimal preheat temperatureshould be a priority. According to aprevious study (Ref. 23) and experi-mental results, the fatigue property ofthe joint is closely related to thenugget size and decomposition in theupper workpiece. Thus, an optimalpreheating process should provide thejoint with a stable maximum weldingarea and avoid decomposition at loca-tion 2, which means the required timeto obtain a stable maximum weldingarea (t1) should be smaller than thetime for decomposition at location 2(t2). Extensive experiments (with various

preheating temperatures) were conduct-ed, and the corresponding temperatureprofiles were recorded. The correlationbetween preheating temperature andt1/t2 is shown in Fig. 14. The overall tendency of these timesdecreased with the increase in pre-heating temperature. The decrease rateof t2 was quicker than that of t1 main-ly because a certain duration at a hightemperature (above melting tempera-ture) was required to obtain a stablemaximum welding area. A distinct characteristic presentedin Fig. 14 shows that a peak and valleyoccurred in the curves of t1 and t2 inthe range of 55° to 95°C, respectively.This phenomenon was closely relatedto the loss modulus of the material. Atthis preheating temperature range(55° to 95°C), the loss modulus in-creased to its maximum, leading to themaximum viscoelastic dissipation inthe material, which accelerated thetemperature increase at location 2 anddecreased t2. More heat was dissipated in thematerial, so a longer welding time wasrequired for the faying interface to ob-tain a stable maximum welding area.Therefore, three intersection points di-vided the preheating temperaturerange into four regions: R1 (25° to55°C), R2 (55° to 95°C), R3 (95° to

Fig. 12 — Schematic diagrams showing the effect of preheattreatment on the formation of the weld region.

A B

C

Zhi 201734.qxp_Layout 1 10/10/17 3:13 PM Page 436

145°C), and R4 (145° to 175°C). In the R2 and R4 ranges, t2 wassmaller than t1, indicating that de-composition occurred before the sta-ble maximum welding area ap-proached. Thermal decomposition re-sulted in a porous region in the joint,which was detrimental to the fatigueproperty of the joint. For the R1 andR3 ranges, t2 was larger than t1, sug-gesting that slight or no decomposi-tion occurred in the joint when thestable maximum welding area wasreached, which was applicable temper-ature ranges for the preheating. As dis-cussed in the previous section, a high-er preheating temperature could de-crease the temperature gradient in the

joint and the cracking tendency. Ac-cordingly, the preheating temperaturefor ultrasonic-welded, carbon-fiber-reinforced polyamide 66 should be se-lected from R3 (95° to 145°C).

Conclusions 1) With preheating at differenttemperatures, the joints made withthe optimal welding time presentedsimilar tensile strengths and exhibitedworkpiece breakage, which was attrib-uted to the similar weld areas. The op-timal welding time for the preheatingtemperatures of 25°, 75°, 125°, and175°C were 2.1, 2.1, 1.5, and 1.3 s, respectively.

2) When the temperature in themiddle of the upper workpiece nearthe periphery of the horn exceededthe decomposition point of the com-posite (375°C) during welding for thejoints made of workpieces with differ-ent preheating temperatures, the com-posite decomposed and the fatigueproperty of the joint deteriorated. 3) The joint welded with preheatedworkpieces of 125°C exhibited thehighest endurance limit, which in-creased by 30% of that without pre-heating. The good fatigue performancewas attributed to the avoidance of de-composition in the composite as wellas the decrease in the temperature gra-dient in the joint during ultrasonicwelding. 4) The suitable preheating tempera-ture for ultrasonic-welded, carbon-fiber-reinforced polyamide 66 waswithin the range of 95° to 145 °C.

1. Lyu, M.-Y., and Choi, T. G. 2015. Re-search trends in polymer materials for usein lightweight vehicles. International Jour-nal of Precision Engineering and Manufactur-ing 16: 213–220. 2. Jansson, A., and Pejryd, L. 2016.Characterisation of carbon fibre-reinforcedpolyamide manufactured by selective lasersintering. Additive Manufacturing 9: 7–13. 3. Jeon, K. S., Nirmala, R., Navamatha-van, R., and Kim, H. Y. 2013. Mechanicalbehavior of electrospun nylon 66 fibers re-inforced with pristine and treated multi-walled carbon nanotube fillers. Ceramics In-ternational 39: 8199–8206. 4. Levy, A., Le Corre, S., and Fernandez

WELDING RESEARCH

NOVEMBER 2017 / WELDING JOURNAL 437-s

Fig. 14 — Relation between preheating temperature and t1/t2.

Fig. 13 — Microstructure of the weld­end position for joints subjected to 1.75 kN after 1 × 106 cycles: A — 25°C; B — 125°C.

References

BA

Zhi 201734.qxp_Layout 1 10/10/17 3:13 PM Page 437

Villegas, I. 2014. Modeling of the heatingphenomena in ultrasonic welding of ther-moplastic composites with flat energy di-rectors. Journal of Materials ProcessingTechnology 214: 1361–1371. 5. Fernandez Villegas, I. 2015. In situmonitoring of ultrasonic welding of ther-moplastic composites through power anddisplacement data. Journal of ThermoplasticComposite Materials 28: 66–85. 6. Iqbal, H. M. S., Bhowmik, S., andBenedictus, R. 2017. Study on the effect ofsurface morphology on adhesion proper-ties of polybenzimidazole adhesive bondedcomposite joints. International Journal ofAdhesion & Adhesives 72: 43–50. 7. Fiore, V., Calabrese, L., Proverbio, E.,Passari, R., and Valenza, A. 2017. Saltspray fog ageing of hybrid composite/met-al rivet joints for automotive applications.Composites: Part B 108: 65–74. 8. Chinnadurai, T., and Arungalai Ven-dan, S. 2017. Thermal and structuralanalysis of ultrasonic-welded PC/ABSblend for automobile applications. Journalof Thermal Analysis and Calorimetry 127:1995–2003. 9. Arungalai Vendan, S., Chinnadurai,T., Senthil Kumar, K., and Prakash, N.2015. Investigations on mechanical andstructural aspects of ultrasonic hybridpolymer mixture welding for industrial ap-plications. The International Journal of Ad-vanced Manufacturing Technology 16: 13, 14. 10. Jiao, J., Wang, Q., Wang, F., Zan, S.,and Zhang, W. 2017. Numerical and exper-imental investigation on joining CFRTPand stainless steel using fiber lasers. Jour-nal of Materials Processing Technology 240:362–369. 11. Fernandez Villegas, I., Valle Grande,B., Bersee, H. E. N., and Benedictus, R.2015. A comparative evaluation betweenflat and traditional energy directors for ul-trasonic welding of CF/PPS thermoplasticcomposites. Composite Interface 22: 717–729. 12. Launay, A., Maitournam, M. H.,Marco, Y., and Raoult, I. 2013. Multiaxialfatigue models for short glass fibre rein-forced polyamide — Part II: Fatigue life es-timation. International Journal of Fatigue47: 390–406. 13. Mortazavian, S., and Fatemi, A.2015. Fatigue behavior and modeling ofshort fiber reinforced polymer composites:A literature review. International Journal ofFatigue 70: 297–321. 14. Tsang, K. Y., DuQuesnay, D. L., andBates, P. J. 2008. Fatigue properties of vi-bration-welded polyamide 6 and polyamide

66 reinforced with glass fibres. Composites:Part B 39: 396–404. 15. Lockwood, K. T., Zhang, Y., Bates, P.J., and DuQuesnay, D. L. 2014. Effect oftemperature on fatigue strength of vibra-tion welded and unwelded glass reinforcedpolyamide 6. International Journal of Fa-tigue 66: 111–117. 16. Srajbr, C., Tanasie, G., Dilger, K.,and Bohm, S. 2011. Active thermographyfor quality assurance of joints in automo-bile manufacturing. Welding in the World55: 90–97. 17. Khoy, Y., Chien, L.-H., Chang, B. C.,and Liu, S.-J. 2000. Effects of the shape ofthe energy director on far-field ultrasonicwelding of thermoplastics. Polymer Engi-neering and Science 40: 157–167. 18. Sharifimehr, S., Fatemi, A., Cha, S.C., Bae, M.-K., and Hong, S.-H. 2016. Fa-tigue behavior of AHSS lap shear and buttarc welds including the effect of periodicoverloads and underloads. InternationalJournal of Fatigue 87: 6–14. 19. Fu, Z., Ji, B., Kong, X., and Chen, X.2017. Grinding treatment effect on rib-to-roof weld fatigue performance of steelbridge decks. Journal of Constructional SteelResearch 129: 163–170. 20. Harati, E., Svensson, L.-E., Karls-son, L., and Widmark, M. 2016. Effect ofhigh frequency mechanical impact treat-ment on fatigue strength of welded 1300MPa yield strength steel. InternationalJournal of Fatigue 92: 96–106. 21. Sanchez-Cabrera, V. M., Rubio-Gonzalez, C., Ruiz-Vela, J. I., and Ramirez-Baltazar, C. 2007. Effect of preheatingtemperature and filler metal type on themicrostructure, fracture toughness and fa-tigue crack growth of stainless steel weldedjoints. Materials Science and Engineering A452–453: 235–243. 22. Chaichanawong, J., Thongchuea, C.,and Areerat, S. 2016. Effect of moisture onthe mechanical properties of glass fiber re-inforced polyamide composites. AdvancedPowder Technology 27: 898–902. 23. Matsuoka, S.-I. 1995. Ultrasonicwelding and characteristics of glass-fiberreinforced plastic: Comparison betweenthe paper-making method and the impreg-nation method. Journal of Materials Pro-cessing Technology 55: 427–431. 24. Wang, X., Guo, D., Chen, G., Jiang,H., Meng, D., Yan, Z., and Liu, H. 2016.Thermal degradation of PA66 during lasertransmission welding. Optics & Laser Tech-nology 83: 35–42. 25. Barnard, P. M., Butler, R. J., andCurtis, P. T. 1988. The strength-life equalrank assumption and its application to the

fatigue life prediction of composite materi-als. International Journal of Fatigue 10:171–177. 26. Hwang, W., and Han, K. S. 1986.Cumulative damage models and multi-stress fatigue life predication. Journal ofComposite Materials 20: 125–153. 27. Tsang, K. Y., and DuQuesnay, D. L.2005. Fatigue strength of vibration-weldedunreinforced nylon butt joints. Polymer En-gineering and Science 45: 935–944. 28. Mortazavian, S., and Fatemi, A.2015. Fatigue behavior and modeling ofshort fiber reinforced polymer composites:A literature review. International Journal ofFatigue 70: 297–321. 29. Tolunay, M. N., Dawson, P. R., andWang, K. K. 1983. Heating and bondingmechanisms in ultrasonic welding of ther-moplastics. Polymer Engineering and Science23: 723–726. 30. Kim, K. T., and Jo, W. H. 2011.Non-destructive functionalization of mul-ti-walled carbon nanotubes with naphtha-lene-containing polymer for high perform-ance nylon 66/multi-walled carbon nan-otube composites. Carbon 49: 819–826. 31. Liu, W., Zhang, S., Chen, X., Yu, L.,Zhu, X., and Feng, Q. 2010. Thermal be-havior and fire performance of nylon-6,6fabric modified with acrylamide by pho-tografting. Polymer Degradation and Stabili-ty 95: 1842–1848. 32. Shaoquan, W., Shangli, D., Yu, G.,and Yungang, S. 2017. Thermal ageing ef-fects on mechanical properties and barelyvisible impact damage behavior of a carbonfiber reinforced bismaleimide composite.Materials and Design 115: 213–223. 33. Ma, H., Qin, G., Wang, L., Meng, X.,and Chen, L. 2016. Effects of preheat treat-ment on microstructure evolution andproperties of brazed-fusion welded joint ofaluminum alloy to steel. Materials and De-sign 90: 330–339. 34. Lei, Z., Li, B., Ni, L., Yang, Y., Yang,S., and Hu, P. 2017. Mechanism of thecrack formation and suppression in laser-MAG hybrid welded 30CrMnSiA joints.Journal of Materials Processing Technology239: 187–194. 35. Krektuleva, R. A., Cherepanov, O. I.,and Cherepanov, R. O. 2017. Numerical in-vestigation of residual thermal stresses inwelded joints of heterogeneous steels withaccount of technological features of multi-pass welding. Applied Mathematical Model-ling 42: 244–256.

WELDING RESEARCH

WELDING JOURNAL / NOVEMBER 2017, VOL. 96438-s

QIAN ZHI, XIN­RONG TAN, and ZHONG­XIA LIU (liuzhongxia@zzu.edu.cn) are with the School of Physics and Engineering, Key Lab of Material Physics,Zhengzhou University, Zhengzhou, China.

Zhi 201734.qxp_Layout 1 10/10/17 3:13 PM Page 438