127Journal of Integrated Circuits and Systems 2011; v.6 / n.1:127-130

Electrochemical Migration on Lead-FreeSoldering of PCBs

Luiz Tadeu Freire Mendes, Valtemar Fernandes Cardoso, Ana Neilde Rodrigues da Silva

Laboratory of Integrated System, LSI - Polytechnic School- University of Sao PauloAv. Prof. Luciano Gualberto, trav 3 n° 158, Cidade Universitária, Sao Paulo, Brazil

mendes@smatech.com.br;valtema@lsi.usp.br;neilde@lsi.usp.br

INTRODUCTION

In the last years the electronic components arebecoming more miniaturized and are integrated with ahigher density, so that electronic packages have smallpitches between the terminals and are more vulnerableto insulation failure. This terminal, when submitted tovariations in the environmental conditions such humid-ity, temperature and applied voltages became electro-chemically unstable. Under these conditions the metalsfrom terminals and or solder paste are ionized and aconducting filament forms between two adjacent termi-nal that behave as an anode and cathode. This filamentscan cause an short-circuit failure on the electronic com-ponent, which is known as electrochemical migration(ECM) [1-3]. The mechanism mainly responsible forthe ECM phenomena is the dendritic growth. Dendriticgrowth occurs as a result of metal ions traveling througha solution at the anode and depositing at the cathode,growing in a tree like formations [4-7]. The formationof dendrites is a significant failure mode in electrical andelectronic systems, particularly in microelectronic com-ponents on PCBs and electronic packages.

In system assembled in lead-free technologythe elements that are mainly achieved after ECM phe-nomenon are Sn, Ag and Cu [4,5,8]. In order to ana-lyze the effect of solder paste alloy and the PCB fin-ishing on ECM, we used a test structure and thewater-drop test. Deionized water is dropped betweenmetal terminals, and a constant DC voltage bias isapplied. The variation in the system resistance wasacquired in order to evaluate the ECM phenomena.

Experimental Procedure

The experiments were performed using a teststructure with a comb shape. The test structure wasmanufactured on a regular PCB print and etch tech-nology over FR4 type substrate. The substrate was 1.6mm thick with a copper layer of 0,5 oz. Two differentsurface finishing were used on the structures, tinapplied by hot air solder level (HASL) and electrolessnickel gold (ENIG). The fingers have 350 µm ofwidth and the distance between them is 102 or 254micron, in order to simulate a real distance betweenthe device terminals.

ABSTRACT

It is well known that in printed circuits boards assembled by SMT technology may occurElectrochemical Migration (ECM). This phenomenon appears mainly because the new packaginghas the terminals very close. Also the Electrochemical Migration may become a potential reliabilityproblem in electronic soldering when lead free technology is used in soldering electronic devices.Electrochemical Migration is an electrochemical process where metal on an insulating material, in ahumid environment and under an applied electric field, leaves its initial location in ionic form andredeposit. In a PCB two adjacent terminals may behave as electrodes so the dendrites grow fromcathode to anode. It can show different morphologies with the different migration elements involveddepending on the solder paste composition or PCB surface finishing. A structure with a comb shapeprinted on FR4 substrate was used in the experiments. The distance between the fingers in thestructure was 102 or 254 micron, in order to simulate a real distance between dispositive terminals.The factors considered during the experiments were surface finishing (ENIG or HASL), solder pastecomposition, distance between terminals (102 or 254 micron) and applied voltage (2 or 3 V). Al theexperiments were done two times. It was observed that the solder paste and the surface finishingdon’t influences the ECM process. Tin was the main metal that migrates. All the results obtained inthese study agrees with the literature.

07 (36)-AF:Modelo-AF 8/20/11 6:42 AM Page 127

Electrochemical Migration on Lead-Free Soldering of PCBsMendes, Cardoso & Silva

128 Journal of Integrated Circuits and Systems 2011; v.6 / n.1:127-130

Two different commercial solder pastes wereused in the experiments: solder paste A was obtainedfrom Almit, SRC Solder Paste type: HEE-LFM48W–TM–HP, Flux F12 (9) and the solder paste Bwas obtained from Heraeus type: F640. Both pastes Aand B are lead-free with the same alloy composition:Sn 96,5/Ag 3,0/Cu 0,5 (10). Until the end of thispaper we refer to the solder paste as A and B.

The solder paste was printed in a professionalprinter from Panasonic model KXF-1D3C. As a maskwas used an electroformed stencil 130 microns thick.

After printing the solder paste were sintered in aprofessional oven model XPM2-1030 from Vitronics.

In order to eliminate the solder flux from the sur-face between the pads the structures were ultrasonicallycleaned with acetone during 10 min followed by IPAduring 10 min and finally with DI water during 10 min.

Electrochemical migration (ECM) tests on solderalloys were conducted by applying constant DC voltagewith a power supply, as show in figure 1. The applied DCvoltage values were 2 and 3 V. Above 3 V the waterdecomposition were observed. The electrical current wasmeasured using an HP 34401A multimeter connectedto a microcomputer. The electrical current values weretransformed into electrical resistance. Before start themeasurements DI water was dropped over the structure.

The ECM tests were performed under an opti-cal microscope with a digital camera (Leica) in orderto observe the dendrite formation. After ECS test thestructures were analyzed by scanning electron micro-scope (Leica Stereoscan) in order to observe the den-drite morphology and by EDS (Oxford) to obtain thedendrite composition.

In the graph shown in figure 2 can be observedthat electrical resistance diminishes with the time indi-cating the metallic ions migration and the dendrite for-mation. However there are variations in these graphsthat can be attributed to the reaction between the metal-lic ions and the OH- ions from water decomposition.

In table I and II are organized the time to shortvalues obtained from the graph. Time to short is con-sidered when the dendrite reaches the opposite elec-trode.

Figure 1. Electrical arrangement of the structure used to performECM analyzes

RESULTS AND DISCUSSION

While performing the first experiments weobserved that the decomposition of water occurswhen the applied potential is greater than 4 V. So, toavoid the influence of water decomposition in theresults, we decided to conduct all experiments on 2and 3 V. In applied voltages lower than 2 V the obser-ved electrochemical migration is very slow.

Figure 2. Electrical resistance versus time for structures with sur-face finishing HASL and printed with solder paste a) A and b) B.

As expected the time to short is lower for lowerdistances between pads and applied voltages.

In some samples it was not observed the dendritegrowth. The cleaning process may leave some residuesof solder flux that can cause this result. The presence ofexcess of solder flux may interfere in the ECM process.

Table I. Time to short from structures with surface finish HASLand solder paste A and B.

Solder Distance Applied Time toVoltage short (seg)

A 2 248254 3 68

2 106102 3 10

B 2 no254 3 52

2 398, Spike102 3 16

R(O

hms)

R(O

hms)

Time (seg)

07 (36)-AF:Modelo-AF 8/20/11 6:42 AM Page 128

Electrochemical Migration on Lead-Free Soldering of PCBsMendes, Cardoso & Silva

129Journal of Integrated Circuits and Systems 2011; v.6 / n.1:127-130

In the SEM images it is not observed signifi-cant differences in the dendrite morphology. The EDSanalysis shown in figure 5 and 6 confirms that themain ion in the dendrite composition is tin. There isno significant peaks from the others expected ions likeCu, Ni, Au and Ag.

However, in the EDS analyzes from the struc-tures with ENIG finishing is observed a small peaksbelonging to Ni and Cu.

In some cases the solder don’t cover complete-ly the pad leaving some metal from the substrateuncovered. As a result these metals from pads alsomigrates through the space between the pads.

Figure 3. Electrical resistance versus time for structures with sur-face finishing ENIG and printed with solder paste a) A and b) B.

Table II. Time to short from structures with surface finish ENIGand solder paste A and B.

Solder Distance Applied Voltage Time toshort (seg)

A 254 2 —3 18

102 2 —3 8

B 254 2 —3 120

102 2 1763 6

Figure 4. SEM images of dendrites growth on a) solder paste A;b) solder paste B.

Figure 5. EDS results from structures printed with the solderpaste B and a) finishing of HASL and b) finishing ENIG.

07 (36)-AF:Modelo-AF 8/20/11 6:42 AM Page 129

Electrochemical Migration on Lead-Free Soldering of PCBsMendes, Cardoso & Silva

130 Journal of Integrated Circuits and Systems 2011; v.6 / n.1:127-130

CONCLUSION

As a conclusion the studies suggests that thesolder paste and the surface finishing don’t influencesthe ECM process. In this work was used two differentkind of solder paste from two different manufacturers.The results of the analyses performed suggest thatthere are no significant differences between the pastesbesides they has different quantities of solder flux. Tinwas the main metal that migrates. The others ionsfrom solder paste Ag and Cu don’t appears on EDS

analysis. All the results obtained in these study agreeswith the literature.

ACKNOWLEDGMENTS

The authors wish to tanks to (LCT) Laborató-rio de Caracterização Tecnológica from PolytechnicSchool for SEM and EDS analyzes.

REFERENCES

1. Dominant Migration Element in Electrochemical Migration ofEutectic SnPb Solder Alloy; Shin-Bok Lee, Ja-Young Jung,Young-Ran Yoo, Young-Bae Park, Young-Sik Kim and Young-Chang Jooa; 2006 Electronic Components and TechnologyConference, p. 621-24.

2. Electrochemical Migration Tests of Solder Alloys In PureWater; T. Takemoto, R. M. Latanisioni, T. W. Eagart, A.Matsunawa; Corrosion Science, Vol. 39, No. 8, pp. 1415-1430, 1997

3. Electrochemical migration characteristics of eutectic Sn-Pbsolder alloy in printed circuit board; Shin-Bok Lee, Young-RanYoo, Ja-Young Jung, Young-Bae Park, Young-Sik Kim, Young-Chang Joo; Thin Solid Films 504 (2006) 294 – 297

4. Electrochemical migration of lead free solder joints; D. Q.Yu,W. Jillek, E. Schmitt; J Mater Sci: Mater Electron (2006) 17:229–241.

5. Electrochemical migration of Sn-Pb and lead free solderalloys under distilled water; D. Q. Yu, W. Jillek, E. Schmitt; JMater Sci: Mater Electron (2006) 17: 219–227

6. Electromigration of Pb-free solder under a low level of currentdensity; J.S. Zhang, Y.C. Chana, Y.P. Wu, H.J. Xi, F.S. Wu;Journal of Alloys and Compounds 458 (2008) 492–499

7. Electrochemical Processes Resulting in Migrated ShortFailures in Microcircuits; Ghabor Harshnyi; IEEE TransactionsOn Components, Packaging, and Manufacturing Technology-Part A, vol. 18, N. 3, 1995

8. Electromigration degradation mechanism for Pb-free flip-chipmicro solder bumps; Toru Miyazaki, Tomoya Omata; Micro-electronics Reliability 46 (2006) 1898–1903

9. http://www.almit.com/dloads/Specs/ leadfree/ spec_lfm48w%20tmhp.pdf

10. http://heraeus-contactmaterials.com/en/products/solderpas-te/bleifrei/productpage_lead_free.aspx

Figure 6. EDS results from structures printed withthe solder paste A and a) finishing of HASL and b)finishing ENIG.

07 (36)-AF:Modelo-AF 8/20/11 6:42 AM Page 130