Evolution of Interfacial Shear Force during … of Interfacial Shear Force during Ultrasonic Al Ribbon Bonding Masaya Ando1,+, Masakatsu Maeda 2and Yasuo Takahashi 1Division of Materials

Evolution of Interfacial Shear Force during Ultrasonic Al Ribbon Bonding

Masaya Ando1,+, Masakatsu Maeda2 and Yasuo Takahashi2

1Division of Materials and Manufacturing Science, Graduate School of Engineering, Osaka University, Suita 565-0871, Japan2Joining and Welding Research Institute, Osaka University, Ibaraki 567-0047, Japan

An Al ribbon was bonded to SiO2 substrate with a 60 kHz ultrasonic wedge bonder. The shear force applied at the interface between theribbon and the substrate was measured with a piezoelectric load-cell. Simultaneously, the vibration amplitude at the tip of bonding tool wasmonitored with a laser-Doppler vibrometer. It was suggested from experimental results that the maximum interfacial shear force was 6.4 timeslarger than the bonding force, i.e., the friction coefficient at the interface could be significantly high during ultrasonic bonding. The evolution andthe transmission of the interfacial shear force were discussed, based on numerical simulations. The purpose of the present study is to reveal theevolution of interfacial shear force at the interface during ultrasonic ribbon bonding. [doi:10.2320/matertrans.MD201207]

(Received December 17, 2012; Accepted February 12, 2013; Published April 5, 2013)

Keywords: ultrasonic ribbon bonding, interfacial shear force, adhesion, constraint, measurement

1. Introduction

In recent years, power electronics devices play importantrole in efficient energy use.1) Application of the devices isspreading to most of equipments which use electric power.Above all, the power electronics devices used in infrastructurehave to endure a heavy load because the devices control veryhigh electric power. On the other hand, every joint in powerelectronics devices needs to be extremely reliable becausesingle disconnection would make entire system wrong.Ultrasonic bonding is extensively applied in packaging ofthe devices. To ensure the reliability of bonding, it is essentialto control the bonding process based on detailed knowledgeof the bonding mechanism. Therefore, the bonding mecha-nism has to be understood correctly. The mechanism ofultrasonic bonding has been studied both experimentally andnumerically. A few studies about interfacial frictional stateduring wire bonding and ball bonding have been reported.2­4)

Shah et al. conducted experiment to derive ultrasonic frictionpower during Al wedge-wedge bonding.5) They measuredultrasonic force using integrated piezoresistive microsensorand estimated ultrasonic power force from measured values.However, the Al wire used in their study was very thin (fine).In case of thick wire or ribbon bonding, the result can bedifferent. Moreover, the integrated piezoresistive microsensormay not be adequate to monitor interfacial force. Suzuki et al.investigated interfacial states during ultrasonic Al ribbonbonding by numerical simulation.6,7) It was suggested that theinterfacial shear force can be over 10 times higher than thebonding force. But there is no experimental report about that.The knowledge of interfacial frictional states will help tounderstand ultrasonic ribbon bonding. The purpose of thepresent study is to reveal the evolution of the interfacial shearforce at the interface during ultrasonic ribbon bonding.

2. Experimental Procedure

Figure 1 schematically illustrates the ultrasonic bondingapparatus used in the present study. High purity

(99.99mass%) Al ribbon was bonded to SiO2 substrate bythe ultrasonic bonding method.8) The thickness and the widthof Al ribbon were 0.2 and 1mm, respectively. The thicknessof SiO2 substrate was 0.82mm. The specimens were cleanedwith ultrasonic acetone bath just before the bonding tests.The ultrasonic vibration was applied to the upper surface ofAl ribbon through the bonding tool as illustrated in Fig. 1.The bonding tool was made of tungsten carbide. Themeasuring system was made of SUS304 stainless steel.The vibration direction was in parallel to the longitudinaldirection of Al ribbon. The ultrasonic frequency f was60 kHz. The bonding force FB was also applied to the uppersurface of Al ribbon. The bonding force was 7.0N, whichwas perpendicular to the surface of Al ribbon. The ultrasonicpower P was changed in the range of 0.5­3.0W. The bondingtime t was 400ms (the period when the ultrasonic vibrationwas inputted).

The interfacial shear force F is produced during ultrasonicbonding at the interface between the ribbon and the substrate.The interfacial shear force is transmitted from the lateral sideof the substrate to the jig (push rod) which fixes the substratein the horizontal direction and then it can be measured by thepiezoelectric load cell. The measuring system was designedand made of SUS304 stainless steel, based on numerical

Fig. 1 Schematic illustration of ultrasonic bonding apparatus.

+Graduate Student, Osaka University, Corresponding author, E-mail:[email protected]

Materials Transactions, Vol. 54, No. 6 (2013) pp. 911 to 915Special Issue on Nanojoining and Microjoining©2013 The Japan Institute of Metals and Materials

simulations. The system is illustrated in Fig. 1. In otherwords, we measured the average gloss shear force transmittedfrom the lateral side of the substrate to the piezoelectricload cell, because it is very difficult to measure the actualinterfacial shear force directly. As illustrated in Fig. 1, thepush rod fixed the load cell well to the substrate by a bolt,i.e., the interfacial shear force was actually transmitted tothe load cell by the elastic vibration. Simultaneously, thevibration amplitude "x at the tip of the bonding tool wasmonitored with a laser-Doppler vibrometer. The samplingpermissible precisions of the piezoelectric load cell andthe laser-Doppler vibrometer were less than 90 kHz and2.5MHz, respectively.

3. Numerical Simulation Procedure

It is necessary to understand how the interfacial shear forceat the bonding interface is transmitted to the load cell throughpush rod made of stainless steel. In other words, we have tosolve the problem whether the shear force can be transmittedto the load cell precisely or not. The numerical analysisconcerning the elastic deformation of the measuring systemwas carried out. It is very important to estimate the accuracyof the measuring system and to secure the experimentalresults. The numerical method was carried out by using thefinite element method.9) Figure 2 shows the mesh patternof the measuring system, together with SiO2 substrate,Al ribbon and the bonding tool. Numerical simulations ofdeformation of Al ribbon and the measuring system (push rodand stage) which fixes SiO2 substrate were conducted as aproblem of two-body which consist of two sub-areas.10) Onesub-area is Al ribbon and another is the measuring system.It was assumed that Al ribbon was a visco-plastic body andthe measuring system with the substrate was a perfect elasticbody. It was also assumed that the frictional slip occurredbetween Al ribbon and SiO2 substrate and that the substratewas fixed well at the both lateral sides to the push rod.At first, the free boundary condition was assumed at theinterface between the bottom side of the substrate and thestage.

The visco-plastic deformation of Al ribbon was calculatedaccording to the previous study.7) The elastic strain of Alribbon was ignored because non-linear calculation withlarge deformation was mainly conducted in the numericalanalysis. The vibration ratio between the upper and the lowerside of Al ribbon was changed from 1.0 (for the free slipcondition) to 0.0 (for fixed boundary condition). The reactionforces in the x and y direction were given from the Alribbon to the SiO2 substrate at the bond-interface due to thefrictional slip which occurred at the interface between Alribbon and the substrate. Because the elastic displacementof SiO2 substrate in the x direction was much less thanthe lower side displacement of Al ribbon, the slip amountwas nearly equal to the lower side displacement of Alribbon, i.e., the elastic displacement of substrate could beneglected.

The frictional slip may occur at the interface between thebottom side of the substrate and the stage. The boundarycondition of the free slip was changed to consider thefrictional slip. The average friction coefficient at the interface

between the substrate and the stage was changed from 0.0(free slip) to 0.5. By taking into account the frictional slipbehavior at the interface between the substrate and the stage,we could estimate the loss of the interfacial shear force whichwas transmitted to the load cell. The material constants ofdeformation were given from literature.11,12)

The force given to the Al ribbon by the bonding toolis reduced due to Al plastic deformation and the frictionalslip at the bond-interface. The reduced force is transmittedto the upper side of SiO2 substrate first. Solution of two-body problem makes it possible to give the distribution ofthe reduced forces (shear force and normal force) at theinterface in the x and y directions. The measuring systemhas to transmit the reduced force to the load cell correctly.This depends on the friction behavior at the interfacebetween the substrate and the stage and also the structureof the push rod. We investigated the dependence, basedon the numerical simulation using the model illustrated inFig. 2.

As illustrated in Fig. 2, the mesh division of Al ribbon isconsiderably finer than that of SiO2 substrate and themeasuring system. It is because the elastic deformation ofSiO2 can be solved with not so much fine mesh division, i.e.,the elastic strain of the measuring system (elastic body) ismuch less than plastic strain of Al ribbon. In the numericalmodel, Al ribbon has 12 elements with 25 nodes at the bond-interface.7) On the other hand, the SiO2 substrate has only 2elements with 5 nodes at the bond-interface. The nodal forcesof 25 nodes of Al ribbon were assigned to these of 5 nodes.The stress distribution was comparable between Al ribbonbottom side and SiO2 substrate upper side. The sum of theforces assigned in the x direction was assumed to be theinterfacial shear force produced at the bond-interface duringbonding. That is, the two-body problem between Al ribbonand the measuring system with the substrate was numericallysolved. The friction between SiO2 substrate and the stagewas taken into account as stated above. The force detectedby the load cell was estimated, based on the numericalsimulations.

Fig. 2 Mesh pattern of the bonding tool, Al ribbon, SiO2 substrate and themeasuring system.

M. Ando, M. Maeda and Y. Takahashi912

4. Results and Discussions

Figure 3 shows the stress distribution at the bond-interfaceof Al ribbon under the condition of bonding pressure PB of28MPa and temperature T of 300K. The bonding pressurePB = 28MPa is comparable to the bonding force FB = 7Nin the experiment. This is a calculated result when theultrasonic vibration is applied from the left to the right handside. The displacement rate in the x direction was givenby 1.88 © 10¹2m s¹1 at the upper side of Al ribbon as aboundary condition due to the ultrasonic vibration trans-mitted from the bonding tool. The Al ribbon was assumed tobe pressed uniformly in the y direction by the bonding tool,i.e., the displacement rate in the y direction at the upper sideof Al ribbon was assumed to be uniform so that the bondingpressure could be equal to the given value (28MPa). Thevibration ratio Rd = Ab/Ao was assumed to be 0.99999,where Ao is the amplitude at the top (upper side) of Al ribbonand Ab is the average amplitude at the bottom side of Alribbon contacted to the bond-interface. As seen in Fig. 3,the shear stress ¸xy is much higher than the normal stress ·yat the center of the bond-interface. The normal stress iscompressive in the front area of the vibration direction (righthand side) but it is tensile at the back area (left hand side),even if the bonding tool presses the Al ribbon. We canestimate from Fig. 3 that the gross interfacial shear force FI

is approximately 33.4N.Because of the force balance at the bond-interface, the

substrate is compressed by FB = 7N and is shear-deformedby FI = 33.4N. The measuring system consists of substrateand stage, push rod, bolt and fixing sticker illustrated inFig. 1. They were assumed to be a perfect elastic body.Because the substrate was fixed in advance by the push rodand the sticker under a large compression, the measuringsystem was elastically deformed under the fixing compres-sion before the bonding force was applied to the substrate.The elastic deformed state of the measuring system wasdefined as the initial state. Figure 4 shows the elasticdisplacement of the measuring system from the initial state,i.e., Fig. 4 represents the elastic deformation due to FB andFI, although the fixing sicker is neglected in Fig. 4. Theelastic displacement in Fig. 4 is magnified 140 times. Thepush rod and the load cell undergo a slight elastic deflectionby compressive and shear forces FB and FI. Table 1 shows

the calculated values of the gross force FL at the load cell. Ifthere is no friction between SiO2 substrate and the stage (thefriction coefficient ® = 0), FL is slightly larger than FI. Itwould be because a moment is applied to the load cell by thedeflection. As the value of ® becomes larger, FL decreases.Experimentally, ® is supposed to be less than 0.5. Therefore,it is suggested that the measuring system designed in thepresent study is adequate to monitor the interfacial shearforce, although it may contains an underestimation.

Figure 5 shows the experimental results of (a) interfacialshear force and (b) tool-tip vibration under the condition ofFB = 7.0N and P = 2.0W. The interfacial shear force shownin Fig. 5(a) was measured by the load cell. If the bondingprocess (adhesion) is successfully produced, the evolution inthe shear force during bonding can be divided into threestages: the initial stage, the middle stage and the final stage.In the initial stage, the shear force increases rapidly and thevibration amplitude is large and dose not decrease. In themiddle stage, the shear force keeps increasing but theamplitude begins to decrease gradually. On the other hand, inthe final stage, the shear force and the amplitude are kept tobe constant under a decreased state. If the adhesion is brokenby the shear force, then they exhibit a sharp fluctuation.Because the adhesion is not enough and locally formed inthe initial stage, the fluctuation is often observed until theadhesion is formed well. Then the amplitude graduallydecreases into the final stage.

Figure 6 shows the interfacial shear force and the vibrationin the initial stage of Fig. 5 until t = 1.5ms. The amplitudereaches maximum at t µ 0.3ms. This is due to a characteristicof the ultrasonic transducer. In contrast, it takes about 1.2msbefore the shear force becomes maximum. These resultsindicate that at first the ribbon slides on the substrate almostfreely. Adhered areas are formed locally at the bondinginterface and they expand rapidly in the initial stage. In themiddle stage, the shear force increases gradually. Corre-sponding to this change, the vibration amplitude at the tool

Fig. 4 Elastic deformation (calculated result) of the measuring system at® = 0.1.

Table 1 Calculated result of gross force at the load cell.

Friction coefficient, ® Gross force at load cell, FL/N

0 35.5

0.1 32.9

0.5 22.5

Fig. 3 Stress distribution (calculated results) at bond-interface of Al ribbonbottom side.

Evolution of Interfacial Shear Force during Ultrasonic Al Ribbon Bonding 913

tip is suppressed. It is considered that the adhered areasconstrain the vibration and the expansion of the adhered areaincreases the constraint in the middle stage.

In the final stage, the shear force and the amplitude arekept steady. The motion of the ribbon in the final stage isstrongly constrained by grown adhered areas in the bonding

interface. The maximum shear force is approximately 40N,which is 5.7 times larger than the bonding force. This valueis extremely high compared with the friction coefficient ofnormal frictional state. It is, therefore, concluded that thefriction coefficient can be significantly high at ultrasonicribbon bonding because of adhesion at the bonding interface.Also, it was suggested in the previous study that the frictionalslip occurs in the final stage against such a high frictionalstate.7)

Figure 7 shows the result of (a) interfacial shear force and(b) tool-tip vibration under the condition of FB = 7.0N andP = 3.0W. The maximum shear force is approximately 35N,which is smaller than that of P = 2.0W in spite of highultrasonic power. The amplitude is not suppressed completelyand is unsteady. Under this condition, the ribbon deformslargely and often fractures on the way of bonding, beforesufficient adhered areas are grown up. It can be said thatthe adhered areas are not expanded effectively by the highultrasonic power, i.e., increasing the ultrasonic power doesnot always lead to a good bonding process.

Figure 8 shows the interfacial shear force at t = 400msunder the condition of FB = 7.0N and P = 0.5­3.0W.Because the condition of P = 0.5W was not enough for agood bonding, sometimes the bonding was not achievedunder P = 0.5W. This suggests that adhered areas are notformed sufficiently because of low ultrasonic power. Asthe ultrasonic power increases, the interfacial shear forceincreases. At P = 2.5W, the average shear force reachesapproximately 45N, which is 6.4 times larger than thebonding force, as seen in Fig. 8. The increment of the shearforce becomes smaller until P = 2.5W. It is suggested thathigher ultrasonic power not only facilitates adhesion butalso causes the bond-interface destruction. At P = 3.0W, the

Fig. 6 Initial stage of Fig. 5 until t = 1.5ms, (a) interfacial shear force and(b) tool-tip vibration.

Fig. 7 Experimental results of measurement, (a) interfacial shear force and(b) tool-tip vibration under the condition of FB = 7.0N and P = 3.0W.

Fig. 5 Experimental results of measurement, (a) interfacial shear force and(b) tool-tip vibration under the condition of FB = 7.0N and P = 2.0W.

M. Ando, M. Maeda and Y. Takahashi914

interfacial shear force decreases although high ultrasonicpower is applied, as already explained with Fig. 7. It can besaid that the excessive ultrasonic energy is just consumed bya large plastic deformation of the ribbon. The measurementof the gross shear force during ultrasonic bonding is veryhelpful to understand the bonding progress state.

The bottom side of Al ribbon may be recrystallized duringultrasonic bonding because it is worked strongly and thetemperature rises. We have confirmed in the previous study13)

that thick Al wire is recrystallized during ultrasonic bonding.We will observe the microstructure of bond-interface of Alribbon bonding.

5. Conclusion

It is very important to understand the interfacial frictionalstates during the Al ribbon bonding. The interfacial shearforce and the vibration of the tool-tip during bonding weremeasured. The main results are shown as follows.(1) The friction coefficient of the interface during ultrasonic

ribbon bonding can be significantly high because of astrong constraint at the interface.

(2) As the adhesion of the interface progresses, theinterfacial shear force increases and the tool-tipvibration amplitude decreases.

(3) Ultrasonic power contributes not only to adhesion butalso to destruction of the interface.

(4) When the bonding process is successfully produced, thebonding can be divided into three stages: the initialstage, the middle stage and the final stage. In the initialstage, the interfacial shear force increases rapidly andthe vibration amplitude is high. In the middle stage, theshear force increases slowly and the vibration amplitudeis suppressed. In the final stage, the shear force and thevibration amplitude are steady.

REFERENCES

1) Y. Takahashi: Trans. Inst. Electron. Inform. Commun. Eng. C 92 (2009)581­594.

2) M. Mayer, J. Schwizer, O. Paul, D Bolliger and H. Baltes: Adv.Electron. Packag. 26 (1999) 973­978.

3) M. Hizukuri and T. Asano: Jpn. J. Appl. Phys. 39 (2000) 2478­2482.4) H. Gaul, A. Shah, M. Mayer, Y. Zhou, M. Schneider-Ramelow and H.

Reichl: Microelectron. Eng. 87 (2010) 537­542.5) A. Shah, H. Gaul, M. Schneider-Ramelow, H. Reichl, M. Mayer and Y.

Zhou: J. Appl. Phys. 106 (2009) 013503.6) Y. Takahashi, M. Maeda, S. Suzuki and Y. Ohyama: Quart. J. Jpn.

Weld. Soc. 29 (2011) 138s­141s.7) Y. Takahashi, S. Suzuki, Y. Ohyama and M. Maeda: J. Phys. Conf. Ser.

379 (2012) 012028.8) Y. Yoneshima, H. Kitamura, M. Maeda and Y. Takahashi: Proc. 14th

Symp. on Microjoining and Assembly Technology in Electronics (Mate2008), Feb., 2008, Yokohama, Japan, 14 (2008) pp. 347­350.

9) Y. Takahashi and M. Inoue: J. Electron. Packag. 124 (2002) 27­36.10) K. Washizu, H. Miyamoto, Y. Yamada, Y. Yamamoto and T. Kawai:

Finite Element Method Handbook II, (Baifukan, Tokyo, 1983)pp. 765­768.

11) H. Toshio et al.: Chronological Scientific Tables 2012, (Maruzen,Tokyo, 2012) p. 391.

12) H. Toshio et al.: Chronological Scientific Tables 2012, (Maruzen,Tokyo, 2012) pp. 394­395.

13) M. Maeda, S. Kitamori and Y. Takahashi: Sci. Technol. Weld. Join. 18(2013) 103­107.

Fig. 8 Dependence of interfacial shear force on ultrasonic power att = 400ms.

Evolution of Interfacial Shear Force during Ultrasonic Al Ribbon Bonding 915