“Closed Loop” Study for Wire Bonding Process

1 1 closed Loop Study for wire Bonding ProcessJeremy KOH, GOH Kay S oon, Winston BAUTIS TA & Jimmy CAS TANEDASPT Asia Pte Ltd, 970 Toa Pay oh, #07-25/26, Sin gap ore :P resent ly, in st andard and some fine pit ch wirebonding applicat ions, the met hods forchecking the bond qualit y during volumeproduct ion st ill remains t he same. Theinspect ion met hods used t o det ermine thequalit y of t he bonds have proven t o beaccept able and reliable for many currentindust rial needs are ball shearing, measuringt he ball diamet er and wire pull measurement .These methods are accept able, but as thesemiconductor sector moves t owards ult ra finepitch wire Bonding , dat a obt ained by t heseconvent ional met hods may not be sufficient.

2 As t he bonded wires per device increases andt he size of t he bond decreases, the int eract ionbet ween machine, Bonding t ool, mat erial andmet hod has become much more sensit paper discusses t he use of a closed loop wire Bonding st udy to analyze t his trodu cti on :The ult ra fine pit ch applicat ions are commonlydone on high frequency wire bonder, whichproduces smaller displacement but at highercycle rate as compared t o the convent ionalbonder. Depending on the power sett ing ont he bonder, t he ult rasonic energy in the formof vibrat ion or displacement is t ransmitt eddirect ly from the transducer t hrough t hecapillary t o creat e the necessary balldeformat ion. Unlike Bonding on non-finepitch devices, which can accommodat e a widerange of ball size variat ion due to t he large padopening, Bonding on ult ra fine pit ch (UFP )devices required t he ball size t o be controlledwit hin a much t ight er t olerance.

3 In mostcases, a variat ion of +/-3 m of t he event ualball size is considered accept able. Given thet ight Bonding Process requirement, opt imumcapillary design and machine performance isnecessary to achieve a reliable bond processcont bje cti ve :The int ent of this paper is to est ablish t hefeasibilit y of using a closed loop wirebonding st udy mainly for fine pit ch and UFPbonding applicat ion. The st udy incorporatest he use of a laser vibromet er, wire bonder, thebond Process analyzer (BP A) and the physicalqualit y checking of the bonds. By using the closed loop st udy, verificat ion on designideas, Bonding set -up and condit ions can bedone at 3 different st ages (pre Bonding , duringbonding and post Bonding ).

4 T hese dat aobt ained at different st ages can used to just ifyt he act ual bond ci ple of close d l oop stu dies:The set -up of t he monit oring syst em used fort he st udies is illust rat ed in Figure 1. It consist sof a P C base laser vibromet er and a BP Aconnect ed t o the output of the ult rasonicgenerat or and the input to the t 1: closed loop m onitoring systemWire BonderLaser VibrometerBPAB onding Responses2 Act ual Bonding was performed on an ESEC3008 wire bonder using a 25 m Au wire . Fort he visual inspect ion and measurement , a highpower microscope of m resolut ion wasused for ball height , and ball sizemeasurement. Ball shear measurement wasperformed on t he ball shear t est er wit h a shearheight of 3 Vibromet erIn wire Bonding , the Bonding tool vibrat ionamplit ude is the most import ant paramet erwhich influences the energy delivered t o t hebond zone.

5 The amplit ude of such nano scalevibrat ion is measured wit h a laser vibrometer(Figure 2), which ut ilizes a 3mW helium-neonlaser and is aligned perpendicular to thebonding tool. Capillary movement duringbonding is feed back t o t he syst em, whichdisplays vibrat ion amplit ude direct ly as shownin Figure 2: Laser vibrometer set-upFigure 3: Ultrasonic displacem ent for a goodwaveformBond Process AnalyzerDuring Bonding , the impedance at the bondsurface changes cont inuous wit h t on the surface condit ion, thevibrat ion charact erist ics of t he capillary andt ransducer changes accordingly. The BP A canbe used t o read t he changes in the ultrasonicimpedance in real t ime during wire syst em analyzes the Bonding result andt he change in t he ult rasonic paramet ersst at ist ically to accumulat e t he changes, and todet ermine the bond qualit y by the ext ent of thechanges.

6 The set-up for the equipment isshown in Figure 4: Schem atic Diagram for BPA set-upCapillary O ptimizati on:During the format ion of t he ball bond, manyfact ors influence t he consist ency of the balldeformat ion. Factors such as machineparamet ers, capillary mat erial and design havea significant effect on t he desirable ball ionally, capillary select ion was based ondevice type and wire size. In most wirebonding applicat ions, a cone shape ceramiccapillary wit h a t aper angle of 20 or 30deg hasbeen ut ilized and deemed suit able for mostdevices. With t he increase demand for UFPbonding applicat ion, which requires a smalland consist ent ball deformat ion, currentcapillary mat erial and design configurat ion hast o be re-defined and opt ial st udies have indicat ed that t he maint aper angle (MTA) of the capillary has themost significant effect on the bond st abilityand consist ency.

7 Alt hough ot her fact ors, suchas t ip diamet er (T ) and bott leneck angle(BNA) can also influence the capillary3charact erist ic, t his is normally dict at ed by thebond pad pit ch (BPP ) of the 5: Profile of a Bottleneck CapillaryTo further underst and the characterist ics of t hecapillary used for UFP Bonding , int eract ionbet ween t he laser vibromet er and the wirebonder was set -up. In t his experiment , theult rasonic behavior of Cap A (20 MTA) andCap B (50MT A) was invest igat ed with realt ime Bonding . Bonding paramet ers was set at ,Force=140mN, T ime=10msec, P ower= paramet er sett ing was based on the act ualbonding characterizat ion on a 60 m BPP .plat 6: Ultrasonic behavior comparisonbetween Cap A and Cap BIn wire Bonding , it is necessary t o ensure t hatmaximum ult rasonic vibrat ion occurs at t he t ipof the Bonding tool for opt imum addit ion, a consist ent ult rasonicdisplacement with a minimum standarddeviat ion (SD) is necessary for consistent balldeformat ion.

8 In act ual fact , the int ent is have ast able Bonding response insensit ive t o a widerange of Process variables. From Figure 7, itcan be seen t hat lower ult rasonic displacementwas obt ained for Cap A design as compared toCap B. The higher SD observed was verifiedfrom t he act ual Bonding responses, whichindicat es a higher ball shear fluct uat ball shear reading was obt ained forunit s bonded wit h Cap A with higheroccurrence of ball lift ed failure mode. Thisobservat ion correlat es with the lowerult rasonic displacement seen earlier. Duringbonding, a 3% ball non-st ick was also not icedwit h smaller ball deformat ion as shown inFigure ACap BUltrasonicdisplacement, nmSD, size, mSD, height, mSD, ShearReading, gmfSD, mode:Ball ShearBall Lift58%42%73%27%Figure 7: Bonding behavior comparisonbetween Cap A and Cap BCap ACap BFigure 8: Ball size com parison4 Process O ptimization:In wire Bonding it is import ant to achieve arobust operat ing window.

9 This means t hatt here should not be any drast ic change in theresponse wit hin t he specified paramet er ly, the only way t o observe this is aft erbonding (during the ball size measurement s).However, the sit uat ion can be improved byincorporat ing t he laser vibromet er as amonitoring syst em t o st udy the displacementof the capillary during , t he st udy was based on thecharact erizat ion report done for the 60 m BPPby SP T. Only the response of t he ball bondwas monitored and the opt imized ball bondparamet ers are:Bond Power: - Force: 140 - 170mNCapillary Type : Cap BFigure 9: Contour plot for bond force andpowerFrom the opt imized paramet er range, t heresponses were monit ored wit h the result asshown in Figure 10.

10 Result s obt ained showst hat for all the possible combinat ions for bondforce and bond power, there is no drast icchange in t he displacement . This indicat est hat t he select ed Process window is sit uat ed int he stable Force : 140mNDisplacement (nm)Bond Power : eStd Power : Power : eStd Power : eStd 10: Displacem ent comparison fordifferent bond power and force combinationThis method of monitoring can be ext endedfor bonder to bonder port ability andcompat ibilit y. Theoret ically, by gett ing t hesame ult rasonic displacement as a funct ion ofbond power and bond force, the balldeformat ion should be similar even though themachine sett ings might be different .Figure 11: Relationship between bond power,bond force and ultrasonic displacem entIn Process Moni tori ng:In the indust ry, bond qualit y measurement hasall along been accessed through visualinspect ion and dest ruct ion t est and ball shear is5t he predominant met hod used for such t est todet ermine the bond st rength.