Deformation Behavior of Thick Aluminum Wire during Ultrasonic Bonding

Masakatsu Maeda1, Yasuhiro Yoneshima2,+, Hideki Kitamura2,+,Keita Yamane2,+ and Yasuo Takahashi1

1Joining and Welding Research Institute, Osaka University, Ibaraki 567-0047, Japan2Graduate School of Engineering, Osaka University, Suita 565-0871, Japan

The deformation behavior of thick Al wires and the expansion behavior of the bond area during ultrasonic wedge bonding to Al­Si, Si andSiO2 substrates were measured simultaneously in detail with a high-speed measuring system. The deformation of the wire by the application ofthe bonding force is completed immediately. The deformation restarts by the application of the ultrasonic vibration. The deformation induced byapplying the bonding force consists of only elastic component, whereas that by ultrasonic vibration consists of only plastic component. The Alwire is not work-hardened by the plastic deformation during application of ultrasonic vibration. The adhered area expands to the directionperpendicular to the ultrasonic vibration. The evolution of the wire deformation behavior and the expansion of the adhered area show an intimatecorrelation with each other. [doi:10.2320/matertrans.MD201210]

(Received December 17, 2012; Accepted February 21, 2013; Published April 12, 2013)

Keywords: high-speed process monitoring, deformation of wiring material, expansion of adhered area, ultrasonic wedge bonding, powerelectronic devices

1. Introduction

Ultrasonic bonding is commonly used in connectingwiring materials to pads on semiconductor chips and toouter-circuits.1­5) The technique forms solder-free directinterconnections at considerably low temperature and withina very short processing time.1) Thus, sound interconnectionsformed by ultrasonic bonding perform high reliability at hightemperatures. This feature makes the technique irreplaceablefor wiring electronic devices used at high temperatures andunder high thermal cycle conditions, e.g., heavy-duty powerelectronics. Power electronics is becoming more importantthan ever as one of the key technologies to improve energyefficiency of electric systems. A representative success can beseen in regenerative braking system installed in hybridvehicles.6) In this way, application fields of power electronicsare broadening rapidly and the demands on them are gettingseverer. To fulfill the demands, the wires are getting thicker.Each wire interconnection must guarantee extremely highreliability, since only one failure in the connections willhalt the entire system. To achieve such a high reliability inmanufacturing, in-situ monitoring of the bonding processesand rapid feeding back of the judgments to the processesare required.7­9) Therefore, correct understanding on thephenomena occurring at the bond interface during ultrasonicbonding is required for precise control of the process.

Up to now, several approaches to the ultrasonic bondingmechanism have been reported. Harman and Albers havereported that the adhesion always initiates around theperimeter of the surfaces in contact.1) Their results indicatethat the frictional sliding occur only at the periphery, insteadof the entire contact area. Lum et al. have investigated theadhesion behavior between Au4) or Al5) wires and Cusubstrate in detail and proposed a model which describes theratio of sliding and stationary area. Their model predicts thatthe entire interface sliding is achieved by setting the bondingparameters with low bonding force and high ultrasonic power

as a function of the static friction coefficient of the interface.Shah et al. have implemented in-situ measurement of shearforce applied to the vicinity of a bonding pad duringultrasonic bonding and derived the ultrasonic frictionpower.9) Their results showing four stages of interfacialsliding provide clear insight of ultrasonic bonding behavior.However, the reports cited above have used fine wires. Inultrasonic bonding of thick wires, Maeda et al. have reportedthe enhanced formation of oxide particles at the periphery ofthe bond area, which was not seen in bonding of fine wiresso far.10)

The present study has been implemented to elucidate thecontact area expansion behavior induced by deformation ofthick Al wire during ultrasonic bonding using high-speedmeasurement and observation techniques: a 2560 kHz laser-Doppler vibrometer, a 50 kHz laser displacement sensor and ahigh-speed video microscope.

2. Mechanism of Adhesion and Expansion of AdheredArea

Generally, ultrasonic bonding proceeds at temperatureslower than the melting points of both wire and substratematerials, i.e., in solid state. Solid state bonding process canbe divided into the following two steps. One is the step toform chemical bond. This step requires adhesion-preventinglayers such as adsorption layers and oxide layers to beremoved from the surfaces to be bonded. The other step is tomake the two bodies to be bonded approach to each otherinto an intimate distance sufficient for metallurgical bondformation. In this step, the gaps originated from theroughness of the surfaces prevent the approach. Therefore,two bodies going to be bonded need to be deformed locally inthe vicinity of the interface to fill the gaps.11­13)

In ultrasonic bonding, the former step is achieved byscrubbing the bonding surfaces with each other. This step isnamed the slipping step. In the early stage of ultrasonic wirebonding, the wire slides on the substrate with a very narrowcontact area. The roughness of the surfaces will affect the+Graduate Student, Osaka University

Materials Transactions, Vol. 54, No. 6 (2013) pp. 916 to 921Special Issue on Nanojoining and Microjoining©2013 The Japan Institute of Metals and Materials

stress applied by the frictional relative motion. In the case ofsuccessful bonding, however, the stress is sufficiently large tosmash and disperse the adhesion-preventing layers, exposingfresh surfaces in the vicinity of the contacted area. Thecontacting fresh surfaces of the wire and the substrate willadhere immediately.

The latter step proceeds predominantly by plastic deforma-tion of the wire, since it is generally softer than the substrate.Takahashi et al. have indicated that the faying area between awire and a substrate increases by folding the side surfaces ofthe wire on to the substrate.14,15) Therefore, the step is namedthe folding step. The deformation is induced by the stressarisen at boundaries between adhered and non-bonded areas.The ultrasonic motion of the wire at adhered areas of theinterface is strongly constrained, whereas the non-bondedareas tend to keep scrubbing. Shear, tensile, compressive andtorsional stresses will arise at their boundaries depending onthe angle between the direction of ultrasonic vibration andthe boundary line. The deformation of the wire during thisstep smashes the adhesion-preventing layers of the wire andthe substrate, revealing fresh surfaces of them. Although theencountered fresh surfaces of the wire and the substrate adhereimmediately, the probability of the encounter is graduallysuppressed in the final stage of ultrasonic wire bonding dueto the residues of adhesion-preventing layers accumulated inthe periphery of the adhered areas.10)

The deformation of the wire during the folding step is verylarge. The cross-section of the wire changes from circular tobottom-flat shape. Thus, the deformation of the wire maybe disturbed by work-hardening. On the other hand, thedeformation will be enhanced by frictional heating. Harmanand Albers have argued that the heat does not affect thebonding behavior by demonstrating successful ultrasonicbonding at liquid nitrogen temperature.1) On the contrary,Krzanowski and Murdeshwar estimated the thermal effectby measuring the width of dislocation-loop free zone in thevicinity of grain boundaries,3) which indicated that the effectis equivalent to annealing at 523K for 90ms. Maeda et al.also showed recrystallized microstructure of the Al wiredeveloped in the vicinity of the bond interface.10) Thesereports suggest that the Al wire is heated by friction up toits recrystallization temperature during bonding. Heating upto such a high temperature lowers the yield stress of Alsignificantly. Recrystallization also suppresses the yield stresswhich is once increased by work-hardening. In addition tothe effects of the frictional heat on the deformation behaviorof the wire, alternative deformation of the materials byultrasonic vibration may deform them with relatively lowstress due to the Bauschinger effect.

Thus, the progress in interfacial adhesion during ultrasonicbonding is explained by two mechanisms taking placesequentially, namely the slip-and-fold mechanism. In theearly stage of bonding, the adhesion proceeds by the slippingmechanism, since it can proceed without pre-existing adheredareas. Once an adhered area which can endure the stressapplied by the ultrasonic vibration is formed, the expansionof the area proceeds predominantly by the folding mecha-nism. The expansion rate of the adhered area will be high atthe beginning of the folding step and then the rate will begradually lowered as the adhered area expands.

3. Experimental Procedure

The ultrasonic bonding apparatus prepared for the presentstudy is schematically illustrated in Fig. 1. The apparatus wasequipped with a 60 kHz ultrasonic transducer, a linear motorfor applying various bonding force, a laser-Doppler vibrom-eter, a laser displacement sensor, and a high-speed digitalvideo camera. The wedge tool attached to the transducer wasmade of cemented carbide and had a V-grooved chuckingface. The wedge tool holds a wire by applying the bondingforce perpendicular to the wire axis. The laser-Dopplervibrometer monitored the ultrasonic vibration behavior at thetip of the wedge-tool with the sampling rate of 2560 kHz.Unfortunately, the amplitude of vibration at the interfacebetween the wire and substrate was quite difficult to monitor.Although the amplitude at the interface is known to be lowerthan that at the tip of the wedge tool, the position wasselected as the monitorable position at which the vibrationbehavior is in the closest relation with that at the interface.The laser displacement sensor measured the displacement ofthe wedge tool to the direction perpendicular to the bondinterface with the sampling rate of 100 kHz. Since the wedgetool is sufficiently rigid, the displacement can be relatedsolely to the change in the height of the wire. Thedeformation of the wire is represented by this value. Itshould be noted that, however, the actual elastic or plasticstrain at each part of the wire is not homogeneous. The high-speed video microscope observed the expansion behavior ofthe adhered area from the back of the substrate with the framerate of 1000 fps (1 kHz). Thus, a cumulative photogene of60-cycles-motion was recorded on one frame of the video.The frame rate was obviously insufficient. Unfortunately, itwas the highest at this moment with a good image resolutionto analyze the adhered area.

In general wire bonding, fully annealed wires are used inorder to enhance the expansion of the adhered area and tosuppress the damage of the semiconductor substrates. In thepresent study, on the contrary, as-drawn (i.e., work-hardened)Al wires of which diameter and nominal purity were 300 µmand 99.9%, respectively, were used in order to slow down thedeformation behavior of the wire to a level detectable with

wed

ge

too

l

linear motorforce loader

bondingforce

horn

transducer

chassis

Al wire(vibrated)

transparent substrate (fixed)

stage

ultrasonicvibration

laserdisplacement

sensor

laser-Dopplervibrometer

high-speedvideo microscope

arrayedLED lamps

Fig. 1 Schematic illustration of the ultrasonic bonding apparatus used inthe present study.

Deformation Behavior of Thick Aluminum Wire during Ultrasonic Bonding 917

the measuring systems described above. On the other hand,three types of substrates were prepared, namely Al­Si, SiO2

and Si substrates. The Al­Si substrates were 1.5-µm-thickAl-1mass%Si alloy films deposited on 0.6-mm-thick Sisubstrates, which are similar to pads on power electronicdevices. The SiO2 substrates were high-purity SiO2 glassdisks of which diameter and thickness were 20.91 and3.00mm, respectively. They were employed to make itpossible to observe the adhesion behavior at the interfacefrom the back of the substrates. The Si substrates were 1.00-mm-thick Si single-crystal chips cut into a size of 5.0-mm-square. The Si substrates were first used in the preliminarystudy. Although the combination of Al wire and SiO2

substrate is different from that of practical wire andelectrode-pad on a semiconductor chip, the deformation ofthe wire and expansion of adhered area during the ultrasonicbonding process will be fundamentally the same as describedin the previous chapter. Therefore, the knowledge to predictand control the deformation behavior of wire in practicalcombination of materials can be obtained by measuring thebehavior of those employed in the present study.

Bonding experiment was implemented in ambient temper-ature and atmosphere using a bonding sequence described inFig. 2. A constant bonding force in the range from 1.0 to7.0N was applied to the wire at first. The wire was keptpressed on the substrate for 300ms to damp the vibrationgenerated by the impact of the bonding force. Then, theultrasonic vibration was applied for 200ms. The ultrasonicpower was set at a constant value between 0.25 and 3.0W.Finally, the bonding force was unloaded 1000ms afterfinishing the application of the ultrasonic vibration. This longholding time was introduced to homogenize the temperaturedistribution in the wire caused by application of ultrasonicvibration. The changes in the vibration behavior, wire height,and bond area were monitored with the laser Dopplervibrometer, laser displacement sensor, and high-speed videomicroscope, respectively, throughout the bonding operation.

4. Results and Discussion

Figure 3 shows a typical displacement behavior of thewedge tool to the direction perpendicular to the bondinterface during ultrasonic bonding of Al wire to Si substrateunder the ultrasonic power and the bonding force of 2.0Wand 3.0N, respectively. The wedge-tool moves immediately

toward the substrate for "zF by applying the bonding force.The displacement does not proceed further after achieving"zF, i.e., the wedge-tool is kept at a constant position until theultrasonic vibration is applied. By applying the ultrasonicvibration, the displacement of the wedge-tool restarts. Thedisplacement rate is high in the beginning and graduallylowered down. The displacement stops after achieving "zmax

(= "zF + "zU) even if the ultrasonic vibration is still applied.In some bonding conditions with short application timeof ultrasonic vibration, ultrasonic vibration ends before thedeformation saturates. In such cases, the displacement stopsat the time ultrasonic vibration ended. At the end ofultrasonic vibration, the wedge-tool does not spring backtoward the balancing point with the bonding force (z = "zF).This phenomenon indicates that the displacement induced byultrasonic vibration ("zU) consists of only plastic deforma-tion of the wire, i.e., it does not contain elastic component.This deformation behavior is clearly different from thatoccurs by increasing static bonding force. When the bondingforce is unloaded, the wedge tool springs back for "zE.The value of "zE corresponds to the elastic component ofdeformation conserved in the wire after the application ofultrasonic vibration.

Figure 4 shows the dependence of "zF, "zU and "zE onthe bonding force at a constant ultrasonic power of 1.0W.The absolute values of "zF and "zE increase proportionatelywith the bonding force, obeying the Hooke’s law. Thus,the deformation behaviors at these stages consist of elasticcomponent. On the other hand, the absolute value of "zUincreases at first and then turn to decrease by increasing thebonding force, showing a maximum deformation ("zUmax) at3.0N. Although the amplitude of ultrasonic vibration will belarge at a condition with a low bonding force, the frictionalforce applied to the wire and the substrate by ultrasonicscrubbing is weak. Thus, it will require a long time to removethe bond-preventing surface layer and to form an initialadhesion, which is required for macroscopic wire deforma-tion by folding. At a condition with a high bonding force, thefrictional force will be large enough to remove the bond-preventing layer. However, the large frictional force alsoconstrains the scrub motion. To form an initial adhesion, the

300 ms

Bonding time, t / ms

P

Bon

ding

forc

e, F

/ N

F

0 500 1000 1500

0

0

200 ms

Ultr

ason

ic p

ower

, P

/ W

<=0.25 P / W 3.0<=

1.0 F / N 7.0<=<=

Fig. 2 Ultrasonic bonding sequence adopted in the present study.0 500 1000 1500 2000

-200

-150

-100

-50

0

50

Bonding time, t / ms

Pos

ition

of t

he w

edge

-too

l, z

/ μm

load bonding force

start ultrasonic vibration

stop ultrasonic vibration

unload bonding force

ΔzF

ΔzU

ΔzE

Al / Si, 2.0 W, 3.0 N

Fig. 3 Displacement behavior of wedge-tool during ultrasonic bonding ofAl wire to Si substrate under the condition of 2.0W, 3.0N.

M. Maeda, Y. Yoneshima, H. Kitamura, K. Yamane and Y. Takahashi918

newly exposed surfaces have to be moved for a certaindistance to meet with the fresh surface on the counter side.Therefore, the suppression of vibration amplitude willsuppress the probability of formation of the initial adhesion.The condition at which "zUmax is achieved will correspond tothe boundary of microslip and gross-sliding proposed byLum et al.,5) since the condition satisfies both the strongfrictional force and large frictional motion simultaneously.The bonding force has an optimum value which depends onthe ultrasonic power to be applied.

Figure 5 shows the dependence of "zF, "zU and "zE onthe ultrasonic power at a constant bonding force of 3.0N.It is clearly seen that "zF and "zE appear constant at everyultrasonic power investigated, whereas "zU increases byincreasing the ultrasonic power. It is easily understood that"zF is constant under a constant bonding force. The changeof "zU depending on the ultrasonic power, showing thetransition from microslip to gross-sliding at the ultrasonicpower between 0.5 and 1.0W, is also an expected behavior asexplained above. On the other hand, "zE was expected tochange depending on the work-hardening of the wire inducedby plastic deformation during ultrasonic vibration ("zU). This

result implies that the wire is not work-hardened by thedeformation of "zU.

It is revealed that "zU consists of only plastic componentwhich is not accompanied by work-hardening. The temper-ature change of the wire during ultrasonic bonding has to betaken into consideration to understand why the Al wire is notwork-hardened. The wire is heated by friction and plasticdeformation during applying the ultrasonic vibration. Anevidence of the temperature change is observed in Fig. 4. Theabsolute value of the gradient of "zF appears larger than thatof "zE. The gradients of "zF and "zE collaterally indicate theelastic modulus of the Al wire before and after applying theultrasonic vibration, respectively. Therefore, the difference inthese two gradients imply the change in the elastic modulusby heating16) during applying the ultrasonic vibration.Although the peak temperature of the wire in the vicinityof the interface may change depending on the bonding force,unloading the bonding force one second after finishing theapplication of ultrasonic vibration gives the Al wire anenough time to homogenize and to converge the temperatureinto a similar value slightly higher than the room temperature.Due to this temperature conversion, the dependence of "zEon the bonding force appears proportional.

Figure 6 shows the displacement behavior of the wedgetool during ultrasonic bonding of Al wire to SiO2 substrateunder the condition of 2.0W, 3.0N. The behavior appearsalmost the same with that shown in Fig. 3 except one point.The spring back of the wedge-tool by unloading the bondingforce ("zE) in the case of the SiO2 substrate is obviouslylower than that of the Si substrate. It should be noted that thesurface of Si substrate is covered by a native oxide layer,i.e., amorphous SiO2. Therefore, the chemical interaction ofthe substrates with the Al wire will be the same. In addition,the surface roughness of SiO2 substrates is similar to thatof Si substrates. Similarities in these points appear in thedeformation behavior of the Al wire as the similar valuesof "zF and "zU. On the other hand, the difference in "zEcorresponds to the elastic strain remaining in the Al wireafter application of ultrasonic vibration. Since the thermalconductivity of SiO2 is lower than that of Si by two orders ofmagnitude,17) the Al wire is heated to a higher temperature.

0 1 2 3 4 5 6 7 8

-100

-50

0

50

Bonding force, F / N

Dis

plac

emen

t of t

he w

edge

-too

l, Δ

z / μ

m

ΔzU

ΔzF

ΔzE

Al / Si, 1.0 W

Fig. 4 Change in "zF, "zU and "zE by the bonding force at a constantultrasonic power of 1.0W.

0 0.5 1 1.5 2 2.5 3 3.5-200

-150

-100

-50

0

50

Ultrasonic power, P / W

Dis

plac

emen

t of t

he w

edge

-too

l, Δ

z / μ

m

ΔzU

ΔzF

ΔzE

Al / Si, 3.0 N

Fig. 5 Change in "zF, "zU and "zE by the ultrasonic power at a constantbonding force of 3.0N.

0 500 1000 1500 2000-200

-150

-100

-50

0

50

Bonding time, t / ms

Pos

ition

of t

he w

edge

-too

l, z

/ μm

load bonding force

start ultrasonic vibration

stop ultrasonic vibration

unload bonding force

Al / SiO2, 2.0 W, 3.0 N

Fig. 6 Displacement behavior of wedge-tool during ultrasonic bonding ofAl wire to SiO2 substrate under the condition of 2.0W, 3.0N.

Deformation Behavior of Thick Aluminum Wire during Ultrasonic Bonding 919

Due to this effect, the value of "zE in the case of the SiO2

substrate is suppressed.Figure 7 shows the displacement behavior of the wedge

tool during ultrasonic bonding of Al wire to Al­Si substrateunder the same condition with that shown in Fig. 6. It isclearly observed that the behavior is similar also in the caseof Al­Si substrate. By changing the substrate from Si toAl­Si, two points appear different. One is that the wiredeformation during applying ultrasonic vibration is notsaturated within the time of 200ms. The other point is thatthe value of "zU is significantly low in the case of Al­Sisubstrate. These facts indicate that the deformation of theAl wire proceeds slowly when the Al­Si substrate is used.To understand the differences appeared in the deformationbehavior of the wire by changing the substrate material, thedifferences in surface state and deformability (hardness)

between Al­Si and Si substrates has to be considered. Thesurface of Al­Si substrates are rough compared to Si, andcovered by amorphous Al2O3 which is more chemicallystable than SiO2. These factors make the adhesion difficult.On the other hand, the Al­Si substrate is softer than Sisubstrate. This factor will have both positive and negativeeffects on the adhesion behavior. To break the bond-preventing layer covering the surface of the substrate willbe easy. However, the surface of Al­Si substrate will bedragged by the ultrasonic motion of the wire, i.e., the relativesliding amplitude between the wire and the substrate willbe suppressed. Although the easy breakage of the bond-preventing layer can compensate the difficulty of adhesioncaused by the surface state of Al­Si substrate, the suppressionof the relative sliding amplitude will play the key role inslowing down the deformation rate.

Figure 8 shows the expansion behavior of the adheredarea during ultrasonic bonding of Al wire to SiO2 substrateunder the condition of 1.0W, 3.0N. The interface at sevenrepresentative stages in the ultrasonic bonding processobserved from the back of the SiO2 substrate, as schemati-cally illustrated in Fig. 8(a), are depicted. The trapezoidsobserved on both sides of the wire are the bottom surfacesof the wedge-tool. The area in contact at the initial state isalmost negligible, as shown in Fig. 8(b). By loading thebonding force, small contact areas appear arrayed along theaxis line of the wire, as shown in Fig. 8(c). The wireappearing almost the same as the initial state indicates thatthe deformation of the wire by loading the bonding force isvery little. Figure 8(d) shows the interface after applying theultrasonic vibration for 10ms. The contact areas are mergedinto a lenticular area appearing in a monotonous contrast ofwhich width has expanded to 105 µm. This area is consideredas the adhered area. In addition, the wire surface appearsblurred, indicating that the wire is vibrating with a

0 500 1000 1500 2000-200

-150

-100

-50

0

50

Bonding time, t / ms

Pos

ition

of t

he w

edge

-too

l, z

/ μm

load bonding force

start ultrasonic vibration

stop ultrasonic vibration

unload bonding force

Al / Al-Si, 2.0 W, 3.0 N

Fig. 7 Displacement behavior of wedge-tool during ultrasonic bonding ofAl wire to Al­Si substrate under the condition of 2.0W, 3.0N.

Fig. 8 Change in the shape of the contact area during ultrasonic bonding of Al wire to SiO2 substrate under the condition of 1.0W, 3.0N.The interface was observed from the back of the SiO2 substrate as schematically illustrated in (a). The states at which (b) before bonding,(c) after the initial bonding force is applied, (d) the ultrasonic vibration is applied for 10ms, (e) 50ms, (f ) 150ms, (g) the bonding forceis still applied after finishing the application of the ultrasonic vibration and (h) the bonding force is removed are shown.

M. Maeda, Y. Yoneshima, H. Kitamura, K. Yamane and Y. Takahashi920

considerably large amplitude. On the other hand, theboundary line and interior of the contact area are sharplyobserved, indicating that the contact area of the wire is notsliding on the substrate with a large amplitude like the otherpart of the wire. Therefore, a high stress will be concentratedat the boundary of the adhered area to facilitate the plasticdeformation of the wire. Figures 8(e) and 8(f ) correspond tothe state of the interface after applying the ultrasonicvibration for 50 and 150ms, respectively. The length of theadhered area along the wire axis appears almost unchanged,whereas the width keeps expanding to 179 and 222 µm,respectively. Therefore, the expansion of the adhered areaproceeds to the direction perpendicular to the ultrasonicvibration. At these stages, the image of the wire surfacebecome sharp, indicating that the vibration of the wire isconstrained by the adhered area. Figures 8(g) and 8(h)correspond to the state of the interface after finishing theapplication of ultrasonic vibration and unloading the bondingforce, respectively. The adhered area does not change bythese operations of ultrasonic bonding. A decrease of theadhered area might be expected after unloading the bondingforce, since the displacement of the wedge-tool shows aspring-back for "zE at this stage. Therefore, this resultsuggests that the elastic deformation of the wire has littleeffect on the expansion of the adhered area.

Figure 9 shows the relationship among the amplitude atthe wedge-tool tip, the displacement of the wedge-tool, andthe expansion of the adhered area measured simultaneouslyduring ultrasonic bonding of Al wire with SiO2 substrateunder the condition of 1.0W, 3.0N. The adhered areaincreases rapidly by applying ultrasonic vibration at firstand then gradually slows down. This is very similar to thedisplacement behavior of the wedge-tool. Therefore, thedisplacement of the wedge-tool can represent the adheredarea formed by ultrasonic bonding. This knowledge isimportant, since the bond interfaces of practical electronicdevices cannot be observed from the back of the substrates.On the other hand, the amplitude evolution of the wedge-tool tip is difficult to correlate with the adhered areaexpansion. For example, the amplitude shows a significantdecrease at ultrasonic vibration time of 30ms. However, nosignificant change in the adhered area expansion behavioris observed.

5. Summary

The deformation behavior of Al wires during ultrasonicbonding to Al­Si, Si or SiO2 substrates were investigated indetail by laser displacement sensor, laser Doppler vibrometer,and high-speed video microscope. The following pointsbecame clear.(1) The deformation of the wire by applying the bonding

force is completed immediately. Further deformation isinduced by application of ultrasonic vibration.

(2) The deformation induced by applying the bonding forceconsists of only elastic component, whereas that byultrasonic vibration consists of only plastic component.The Al wire is not work-hardened by the plasticdeformation during application of ultrasonic vibration.

(3) The amount of deformation induced by ultrasonicvibration increases monotonously by increasing theultrasonic power. On the other hand, it increases at firstand then turn to decrease by increasing the bondingforce, showing an appropriate bonding force to facilitatethe wire deformation.

(4) The adhered area expands to the direction perpendicularto the ultrasonic vibration.

(5) The evolution of the wire deformation behavior andthe expansion of the adhered area show an intimatecorrelation with each other.

Acknowledgements

The present study has been supported from the JapanSociety for the Promotion of Science through Grants-in-Aidfor Scientific Research (project No. 18206076).

REFERENCES

1) G. G. Harman and J. Albers: IEEE Trans. Parts Hybrids Packag. 13(1977) 406­412.

2) V. H. Winchell, II and H. M. Berg: IEEE Trans. Compon. HybridsManuf. Technol. 1 (1978) 211­219.

3) J. E. Krzanowski and N. Murdeshwar: J. Electron. Mater. 19 (1990)919­928.

4) I. Lum, J. P. Jung and Y. Zhou: Metall. Mater. Trans. A 36 (2005)1279­1286.

5) I. Lum, M. Mayer and Y. Zhou: J. Electron. Mater. 35 (2006) 433­442.6) A. Kawahashi: Proc. 2004 Int. Symp. on Power Semiconductor

Devices and ICs (ISPSD), Kitakyushu, Japan (2004) pp. 23­29.7) S. W. Or, H. L. W. Chan, V. C. Lo and C. W. Yuen: Sens. Actuators A

65 (1998) 69­75.8) M. Mayer, O. Paul, D. Bolliger and H. Baltes: IEEE Trans. Compon.

Packag. Tech. 23 (2000) 393­398.9) A. Shah, H. Gaul, M. Schneider-Ramelow, H. Reichl, M. Mayer and Y.

Zhou: J. Appl. Phys. 106 (2009) 013503.10) M. Maeda, S. Kitamori and Y. Takahashi: Sci. Technol. Weld. Join. 18

(2013) 103­107.11) B. Derby and E. R. Wallach: Metal Sci. 16 (1982) 49­56.12) J. Pilling, D. W. Livesey, J. B. Hawkyard and N. Ridley: Metal Sci. 18

(1984) 117­122.13) Y. Takahashi and K. Inoue: Mater. Sci. Technol. 8 (1992) 953­964.14) Y. Takahashi, S. Shibamoto and K. Inoue: IEEE Trans. Compon.

Packag. Manuf. Tech. A 19 (1996) 213­223.15) Y. Takahashi and M. Inoue: J. Electron. Packag. 124 (2002) 27­36.16) H. H. Pham, M. E. Williams, P. Mahaffey, M. Radovic, R. Arroyave

and T. Cagin: Phys. Rev. B 84 (2011) 064101.17) National Astronomical Observatory of Japan, (ed.): Chronological Sci-

entific Tables 2012, (Maruzen, Tokyo, 2011) pp. 416­417 (in Japanese).

Fig. 9 Relationship among the amplitude at the wedge-tool tip, thedisplacement of the wedge-tool, and the expansion of the adhered areaduring ultrasonic bonding of Al wire with SiO2 substrate under thecondition of 1.0W, 3.0N.

Deformation Behavior of Thick Aluminum Wire during Ultrasonic Bonding 921