(direct current and pulse current) on changes of

Transcription

1 Influence of High Current Load on Electrical Properties of Adhesive Conductive Joints Ales Duraj, Pavel Mach, Radoslav Radev, Jan Matejec Czech Technical University, Departent of Electrotechnology Prague 6, , Czech Republic E-ail: Abstract Isotropic and anisotropic electrically conductive adhesives (ICAs and ACAs) are alternative aterials for substitution of coon used tin-lead solders in electronic assebly. Electrically conductive adhesives are potential aterials for the bonding ofsurface ounted coponents andfor soe special applications. In this research we are focused on an investigation of influence ofhigh current load on changes of quality of adhesive joints. Monitoredparaeters are electrical resistance and nonlinearity of V-C characteristics. Research is divided into two types ofhigh current load. a) direct current (DC), aplitude 1 A, b) current pulses.-], f = 5 Hz, aplitude 1.41 A. Introduction Isotropic and anisotropic electrically conductive adhesives (ICAs and ACAs) are alternative aterials for substitution of coon used tin-lead solders in electronic assebly. Electrically conductive adhesives (ECAs) were developed especially for joining in applications where the use of high teperature (necessary for soldering) is ipossible. Basic aterial of ICAs is usually epoxy or acrylic resin (insulating aterial) and electrically conductive filler (Ag flakes or balls). Electrically conductive adhesives, especially ACAs and ACFs are potential aterials for the bonding of surface ounted coponents and for soe special applications, e.g. flip-chip assebly, chip-on-fil (COF), chip-on-glass (COG) or for flexible circuit assebly applications (chip-on-flex). ACFs are also extensively used for connection between PCBs and LCD or liquid crystal odule (LCM). Joining with anisotropic conductive adhesives and fils is very fast and successfully spreading technology because of there is no requireent for applying additional underfill [1]. The basic aterials of ACAs are theroset or theroplastic (usually epoxy resin) with conductive filler- ainly balls with conductive layer (polyer + Au, Ni + Au). These particles can be defored between bonding electrodes during actual bonding and this deforation increases the contact area between the conducting particles and bonding electrodes. Thus it lowers the interconnection resistance [2]. The ain advantages of these types of conductive adhesives are anisotropic conductivity and suitability for fine-pitch connections where the space between leads is very sall (< 125,u). The disadvantage of ACAs and ACFs is a requireent of pressure during whole curing process. Pressure and heat (therocopression) are applied siultaneously so that the /6/$2. (D2OO6 IEEE conductive particles can ake contact with both bups and substrate tracks and to be electrically connected. Quality of these adhesive joints is evaluated according to their electrical and echanical properties. In this research we are focused on an investigation of influence of high current load (direct current and pulse current) on changes of quality of adhesive joints. Also investigation of influence of bonding pressure of tested ACA has been carried out. Curing pressure has been chosen in a range 3 MPa to 1 MPa and evaluated its effect on quality of adhesive joint. Monitored paraeters are electrical resistance and nonlinearity of V-C characteristics. Experients Investigation of influence of high current load on basic electrical properties of adhesive joints is significant especially for high power coponents because of electrical current and high teperature has radical effect on quality and reliability of bonds. Experient has been divided into two parts - study of effects of high direct current (DC) which is passing through testing adhesive joints and also testing of influence of high current pulses. Paraeters of loading conditions: a) DC: I= Is= A/5hours b) Current pulses : I 1,41 A (Irs= 1 A) / 5 hours, 1:1 (t1 = 1 s, T = 2 s), f = 5 Hz Both types of high current load were carried out for a 5 hours in abient conditions (2 C / 3 RH) without special cooling of testing boards. DC has been provided with power supply with stabilized output current. Generator of pulse current has been developed on departent of 1323

2 electrotechnology. Power supply for this generator has been the sae as for DC load. Effective values (Irs) of loading currents have been chosen the sae (1 A) for both types of loads because of identical theral effects (depending on resistance tested bonds). For DC load is I= Irs 1 A and for current pulses is axiu loading current I = 1,41 A - according to equations (1) and (2). Irs-= TJT/2 M rf (1) 2 <2 I Irs _ IM 1,41 = Experient has been divided into investigation of isotropic and also anisotropic electrically conductive adhesives. Tab.l. shows selected ECAs (3 types of ICAs and 1 type of ACA) for easureents. Table 1. Tested Electrically Conductive Adhesives. Type Properties Aepox bisphenol epoxy ICA (one-coponent resin adhesive) + forulated BF3 75 % Ag flakes Aepox ER 55 bisphenol epoxy ICA (one-coponent resin adhesive) + butyl glycidyl ether 75 % Ag flakes Aepox ER 656S bisphenol epoxy ICA (two-coponent resin adhesive) +forulated polyaine 8 Ag flakes LOCTITE 3447 Epoxy resin ACA Ni filler (Au coated) Tg= 15 C (2) Quality of these adhesives joints is evaluated according to their electrical paraeters: 1) Electrical resistance Rs of the single adhesive joint 2) Nonlinearity of C-V characteristics of 2 joints + SMD Preparation of testing saples Basic aterial of testing board (PCB) is coon used fibreglass lainate FR4 - shown in Figure 1. Material of conductive paths is plated 35,u Cu foil. Three types of surface etallization of conductive paths have been used: 1) Cu 2) Cu +- 5,t Sn (HASL) 3) Cu + 5,t Ni + 1 t Au Figure 1. Assebled testing board Tested joints have been prepared with assebly of SMD resistors (type 126 ORO, Microtech GbH) with special contacts etallization (CrNi-Cu-Ag) for adhesive connections - depicted in Figure 2. Figure 2. SMD resistor with lead-free contacts. Adhesives have been deposited on the PCB in defined aount by injection ethod, assebled with SMD resistors and then have been cured under defined conditions (approx C / 1-3 in.) All analyses of the ACA have been carried out in a range of bonding pressure 3 to 1 MPa (6-2 N / SMD) for investigation of influence of curing pressure on quality of joints. ACA was cured with hot air under defined teperature and various pressure conditions. Effective value Is of loading current (1 A) has been selected according to ax. power load of SMD resistors. Typical electrical resistance of single adhesive joint is around 2-3 Q and used SMD resistors have resistance around 1 Q. According to equation (3) the power dissipation in one assebled SMD resistor is about 5-7 W which is 1/4 of ax power Pax of SMD resistor (,25W). P=U.I=R12 (3) At these power conditions the SMDs and joints - teperature is up to 4 C (AT 2 C). These 1324

3 Table 3. Change ofjoint resistance Rs of ICAs after 5 hours of high DC and current pulses load. experients are ainly focused only on an effect of high current on changes of joint resistance and nonlinearity of C-V characteristics. So this teperature is a coproise between the highest current load and teperatures which have only very slight effects on easured paraeters of adhesive bonds. Applied current could be higher but influence of the teperature should be indispensable. of teperatures has been Measureent accoplished with accurate thero-vision ethod. For soe testing saples the joint resistance increase rapidly and power dissipation overpass tolerable value of Pax of SMD. These resistors have burned up and therefore these SMDs have been rejected fro evaluations. Exaple of easureent of teperature conditions on testing boards during high current load is shown in Figure 3. 45, X , a E z 3 25,- 2-15, e 1 e 5, - Initial value 6 E ER55MN 5 hrs. DC load El 5 hrs. current pulses load 1 5_ a4 F _ 3l t 2, Electrical resistance of the joints Resistance of the joints has been easured with the LCR eter HP The saples have been contacted with four-point probe ethod. The resistance of single resistors has been easured (Rr= 1 Q). Then the syste SMD resistor + 2 joints assebled on the board has been easured (RM) Single joint resistance Rs is evaluated according to RM -Rr 2 E Initial value joint resistanei KS - I 5 hrs. current pulses load bni paas~) 25MNER65 (4) S 5 hrs. DC load z 5 hrs. current pulses load Figure 4. Influence of high current loads and types of pads etallization on joint resistance Rs of ICAs. k I'J Table 2. Measured initial joint resistance Rs of ICA E 5 hrs. DC load I 1 n Initial value Tested Isotropic Conductive Adhesives: ei ER55MN 8 equation (4). R Ie~ Ied~ I Initial joint resistance Rs of tested ICAs is in the range 2-1 Q and depends on the type of pads etallization. Fro this easureents and also fro other experients [3] is clear that ost of ICAs is not very suitable for connection on pure Sn etallization. Differences between other types of surface finishes are not too high. Changes of Rs are not too different (< ± 2 o) for both DC and current pulses load (5 hrs.). Only for Sn pads is shift uch 1325

4 higher (increase of Rs up to 8-ties). Unexpected result is that on Ni/Au pads etallization the joint resistance has increased after DC load but after current pulses load Rs decreases for all three types of ICAs. For better understanding of this fact is probably necessary to accoplish rather longer high current loadings. Ea E t X ly Tested Anisotropic Conductive Adhesive: Table 4. Measured initial joint resistance Rs of ACA 3 25, 2 15, 1 5, Cu pads --Au pads A Sn pads Figure 5. Effect of bonding pressure and pads etallization on the joint resistance of ACA. Influence of bonding pressure on joint resistance is shown in Figure 5. Fro this easureents and also fro other experients [4] is clear that joint resistance Rs decreases with rising bonding pressure. Differences between tested Cu, Ni/Au and Sn etallization are not very strong. For bonding pressure p > 5 MPa is joint resistance Rs approx. between 2 and 25 Q and it is stabilized. Effect of DC or current pulses load on Rs has not linear tendency depending on bonding pressure. It is probably caused by rando generation and distribution of defects in interconnections between coponent leads, conductive particle in ACA and pad of PCB. These interconnections could be probably also iproved by high current which is passing through. Local heating, caused by high current density, in boundaries between conductive particles and substrate pads (coponent leads) could create better electrical connection and joint resistance decreases. Nevertheless bonding pressure p > 5 MPa has decreasing effect on change of Rs. Shifts of Rs above this bonding pressure are less than ± O 3 2 X r_ Change of joint resistance [/o% - DC load (5 hours) 2,99 3,54 4,43 ncu pads Au pads 1 D Sn pads Change of joint resistance [%] - current pulses (5 hours) 2,99 3,54 4, *Cu pads Au pads r Sn pads Figure 6. Influence of high current loads and 3 types of pads etallization on joint resistance Rs of ACA. Nonlinearity of the C-V characteristics There are a few of ethods of easureent of nonlinearity. In our case we used the ethod of easureent of the interodulation distortion. Interodulation is the result of two signals of different frequencies being ixed together, foring additional signals at frequencies that are not in general at haronic frequencies (integer ultiples) of either. The largest interodulation products appear at fl + f2 or fl - f2 (second-order interodulation), and less so at 2fl f2 or 2f2 - fl (third order interodulation). In principle, the interodulation products are described by following equation: f=n.f +.f2 The chosen frequencies: fi = 4,16 MHz, n = 1 f2= 15kHz, =2 Then easured frequency is f= 4,46 MHz Nonlinearity easureents have been perfored with equipent designed and constructed at the (5) 1326

5 departent of electrotechnology and carried over 2 joints and SMD resistor. Most frequently expression of nonlinearity of C-V characteristics is with third haronic index THI: THI = 2 log 3 Un E -1 *n. (6) [#V] In our case we used power easureent relative 1 to illiwatt - evaluation of nonlinearity in db. For evaluation of third haronics Uc = U3 in,u we used equation (7); U, = NI12 Joint nonlinearitv Uc - (Cu nkads) E Initial value ER 55MN Gi 5 hrs. DC load B 5 hrs. current pulses load F -8, 46 (7) D -4-2 NI = 2 log c U' [db] tow Q' O) E Initial value where NI is nonlinearity index (easured) [db] U1= 224 V - reference voltage (Zo= 5 Q) (if NI = -17,8 db then Uc = 1,u) E -8-6 E 5 hrs. DC load Joint -1 ER 55MN E 5 hrs. current pulses load nonlinearity Uc - (Sn pads) - ~,-4 Tested Isotropic Conductive Adhesives: -2 N-A Measureent of nonlinearity is in electrotechnics ER 55MN soeties used for evaluation of quality and R Initial value 5 hrs. DC load 5 hrs. current o El pulses load Fro other there reliability prediction. experients is clear that this ethod is very sensitive for sall Figure 7. Influence of high current loads and 3 types of shifts of joint quality. In our case the results are pads etallization on nonlinearity of ICAs joints. siilar - the lowest nonlinearity of adhesive joints has been achieved on coon used Cu pads while Table 5. Change of nonlinearity of tested ICAs after the highest nonlinearity is on Sn pads - this 5 hours of high DC and current pulses load. etallization is not suitable for adhesive bonding (Figure 7.). As was already said in [4], the easureent of nonlinearity is ore sensitive for low changes of adhesive joints quality than coon used evaluation of electrical resistance. This fact is sae for ICAs and also for ACAs joints. In Figure 7 is coparison of nonlinearity of adhesive joints on Cu pads and Ni/Au pas. Electrical resistance of these joints is siilar but nonlinearity is uch ore higher ( + 3 db) on Ni/Au etallization of pads. So an effect of additional layers (etallization) can be better registered by easureent of nonlinearity rather than with evaluation of electrical resistance. Change Tested Anisotropic Conductive Adhesive: of nonlinearity due to high current load is depicted Nonlinearity of C-V characteristics of ACA in Table 5. DC load has a little higher effect on joints has been also analyzed. Influence of bonding nonlinearity but there are no definable trends. pressure is shown in Figure 8. Differences between Cu and Sn etallization are inial and shapes of curves are also very siilar. So echanis of creation nonlinearity and also values of easured nonlinearity are very analogous for these two types of aterials. 1327

6 , ,-, , -15, * Cu pads * Au pads. Sn pads Figure 8. Effect of bonding pressure and pads etallization on nonlinearity of ACA. On the other hand the nonlinearity of ACA joints on Cu +H Ni/Au pads is over -H15 db higher for whole range of tested bonding pressure (3-1 Mpa). As was already said, this is probably caused by any boundary-lines between layers on PCB pads. These boundaries provide additional nonlinearity in interconnections between coponents and substrates and bonding pressure can not eliinate it. Change of ACA joint nonlinearity is shown in Figure 9. Changes of nonlinearity are ainly between ± 6 O for all bonding pressures and for all types of pads etallization. Nevertheless there are also soe unexpected extrees especially for high DC load of adhesive bonds on Sn pads sv X Chanae of ioint nonlinearitv ro/o - DC load (5 hours) 2,99 3,54 4, ccu pads EAu pads o Sn pads Change of joint nonlinearity [%/J] - current pulses (5 hours) 2,99 3,54 4,43 * Cu pads EAu pads 1 Sn pads Figure 9. Influence of high current loads and 3 types of pads etallization on joint resistance Rs of ACA. Conclusions This experient has been focused on investigation of high current load of ICAs and ACAs. Two types of load (DC and current pulses) have been applied on three types of pads etallization. Electrical resistance is typical paraeter for definition of quality of electrical connection. Measureent of nonlinearity of C-V characteristics shows big potential as a diagnostic tool for evaluation of sall changes in a quality of adhesive joints. Effect of current load and curing pressure of ACA has been also analyzed with Scanning Electron Microscopy (SEM).Nevertheless no visual defects have been found- Figure 1. Figure 1. ICA joint (top) after 1 A / 5 hrs. DC load, ACA joint (botto) after 1,41 A / 5 hrs. current pulse load Acknowledgents The authors would like to acknowledge for financial support provided by research grant MSM no Diagnostic of Materials, Czech Republic. References 1. Shyh-Ming Chang, Jwo-Huei Jou, Ada Hsieh, "Characteristic study of anisotropic-conductive fil for chip-on-fil packaging", Microelectronics Reliability, Vol. 41, Issue 12, Deceber 21, pp

7 2. Gauta S., S. Mridha, Tan Tin Chong, "Flip chip interconnect using anisotropic conductive adhesive", Journal ofmaterials Processing Technology, Vol. 89-9, May 1999, pp Duraj,A.,Mach,P.,"Influence of Difference Types of Surfaces on Properties of ECA's and Solders" Proc 15th European Microelectronics and Packaging Conference & Exhibition, Brugge, Belgiu, June. 25, pp Duraj,A.,Mach,P.,"Experient of Influence Curing Pressure of Anisotropically Conductive Adhesives on Paraeters of Joints" Proc 29th International Conference of International Microelectronics and Packaging Society - Poland Chapter, Koszalin- Darlowko, Poland, Septeber. 25, pp