UNDERSTANDING WHISKER PHENOMENON - PART II Competitive Mechanisms

Transcription

1 UNDERSTANDING WHISKER PHENOMENON - PART II Competitive Mehanisms Chen Xu, Chonglun Fan, Anna Vysotskaya, Joseph A. Ays and Yun Zhang Luent Tehnologies, Eletroplating Chemials and Servies 236 Rihmond Valley Road, Staten Island, NY Leslie Hopkins Bell Laoratories, Luent Tehnologies Murray Hill, NJ Fred Stevie Luent Tehnologies Miroeletronis Orlando, FL It is well understood that the root ause for whisker formation is the internal stress. Many fators ontriute to internal stress. Three key fators are identified in our work "Understanding Whisker Phenomenon: Part I". In this paper, we will desrie our systemati, and in-depth investigation of these fators, and the dynami nature of the whisker growth phenomenon. We will also offer our understanding of the ompetitive nature of various whisker growth mehanisms and what set of parameters determine a partiular mehanism. This understanding provides the foundation for the reommendations that we offer to the eletroni industry. For more information, please ontat Chen Xu Eletroplating Chemials and Servies Luent Tehnologies 236 Rihmond Valley Road Staten Island, NY (718) xu@luent.om,

2 Introdution The eletroni industry is under extreme pressure to remove tin-lead solders from eletroni omponents [1-4]. Pure tin is one of the alternatives and may e the simplest system as a drop-in replaement. However, fear of whiskers has een a major onern inhiiting its implementation [5-11]. This is eause that tin whiskers, with a length from a few mirometers to several millimeters, may grow from eletroplated Sn and ause eletri shorts in eletroni devies, partiularly, in fine-pith high I/O (input/output) omponents. Tin whisker formation appears to e favored y the presene of organi and hydrogen inlusion, small size grains, intermetalli formation, internal stress and the appliation of external mehanial stress [5-12]. The attempt y the deposit to derease the internal energy through rerystallization and grain growth results in whisker growth, whih is further aelerated y the stress in the Sn oating. It has een also reognized that long range mass transport of Sn is neessary for the formation of very large whiskers oserved on eletroplated Sn. The high moility of Sn oupled with the desire to derease the internal energy in the eletroplated Sn leads to the whisker formation. Several remedies have een proposed to overome the prolem of whiskering: alloying with a small perentage of other metals (e.g. P, Bi), hot dipped tin, as well as reflow tin after plating are some ommon praties. Compared with eletroplated pure tin, these alternatives are less haraterized and have muh shorter reliaility histories. Though eah of them has signifiant merit in their own right, they have inherent limitations. For instane, solder joint emrittlement is a reliaility issue when preious metals are involved. Tin alloys suh as SnCu, SnBi and SnAg, on the other hand, have the following limitations: staility of plating proess (exessive Sn oxidation and Sn immersion), alloy omposition ontrol, metal ost, and availaility and ompatiility with existing tin-lead solders (e.g. Bi and Ag). Previously, we have reported our systemati kineti studies on the whisker growth of various Sn finishes and identified various fators affeting the kinetis of whisker growth (see part 1 y Zhang et al. in this proeeding). In this paper, we will desrie the results of our studies on the whisker growth mehanisms. Various tehniques were used to examine the loal struture and hemial omposition of the whisker. The residual stresses were also determined using x-ray diffration. Experimental results and disussion FIB results: To review the loal struture of whiskers, ross setions along the root of the whisker are made using foused ion eam (FIB). In a FIB experiment, an extremely small diameter eam of gallium ions is used to image the surfae and loate the whisker. The same foused ion eam is then used to remove materials from the surfae at high lateral resolution and ut through the whisker with an auray etter than 10 nm. Fig.1 shows FIB images of a whisker found on a matte Sn surfae, whih was plated diretly on a Cu sustrate. The sample was aged at room temperature for 13 months. The pitures in Fig.1 represent various stages of utting through the whisker. Fig. 1a is the FIB image of the surfae taken after a trenh is ut into the oating next to the whisker. FIB is then used to gradually ut through the whisker and FIB images were taken at various stage of utting. These results are shown in Figures 1, 1 and 1d. The grain struture around the whisker is niely revealed in these images. Three different layers an e identified in these images: Cu sustrate, Sn-Cu intermetalli layer and Sn layer. The intermetalli growth at the Sn and Cu interfae shows strong anisotropy, with some areas growing muh faster than others. There are also a very lear grain oundary etween the whisker and adjaent grains. The whiskers seem to originate from the middle of the Sn oating, rather than from the Sn-Cu interfae or the Sn surfae. Apparently, a

3 a whisker nuleate is formed within the Sn oating and then grows out of the Sn oating. Very similar results have een also otained for satin right Sn plated on Cu sustrate. Fig.2 shows a FIB image for a whisker found on the satin right Sn, whih was aged at room temperature for 18 months. Here again, intermetalli ompound formation is oserved at Sn and Cu interfae. Similar to the matte Sn, very lear grain oundary is oserved etween the whisker and adjaent grains and the whisker is originated within Sn film. It is also noteworthy that in oth ases the whisker is sitting on top of the intermetalli phase. Fig.2 FIB image of whisker found on satin right Sn d FIB experiments were also performed on right Sn, whih was plated over the Cu sustrate. The sample was aged at room temperature for 18 months. Fig.3 shows FIB images with inreasing magnifiation from a to. The length of this partiular whisker is aout 250 µm (see Fig.3a). As Fig.3 and 3 show, the long filament-type whisker originates from the nodule on the surfae. There is again a very lear grain oundary etween the filament whisker and nodule whisker. The filament is not in diret ontat with the Sn oating. The mass transport from Sn film to the filament whisker, neessary for the formation of this very long whisker, ours through the nodule whisker. The nodule whisker apparently Fig.1 FIB images of whisker found on matte Sn

4 a ats as a preursor state for the formation of the filament whisker. This is onsistent with the oservation that the nodule whisker is seen efore the filament whisker during the aging at room temperature as well as 50 o C. a Fig.4 SAM mapping of whisker found on matte Sn Fig.3 FIB images of whisker found on right Sn

5 Auger and SEM/EDS Mapping: a Sanning Auger mirosopy (SAM) and energydispersed spetrosopy (EDS) were used to determine the elemental omposition around and within the whisker. For the SAM measurements, the ross setion through the whisker was made using FIB. The sample is then transferred into the SAM hamer. Prior to the SAM mapping, the surfae was leaned y argon-ion sputter for 2 min. Fig.4 shows Auger mapping of a whisker found on the matte sample after 13 months aging at room temperature. The seondary eletron image is provided in Fig.4a, while the Cu and Sn mapping are shown in Fig.4 and, respetively. Consistent with the FIB results, a Sn-Cu intermetalli layer is seen etween Sn and Cu layer. Furthermore, large onentration of Cu is also seen within the whisker. SEM/EDS was also performed on whiskers found on right Sn. The right Sn was plated diretly on Cu and aged at room temperature for 18 months. The sample was ut using a diamond saw and finepolished to reveal the whisker. Fig. 5a, 5 and 5 show seondary eletron image, Cu mapping and Sn mapping around a nodule whisker. As the Sn and Cu mapping demonstrate, the whisker onsists mostly of Sn. However, there is also some Sn-Cu intermetalli formation within Sn oating near Sn- Cu interfae, even though not as muh as in the ase of matte and satin right Sn. The intermetalli growth shows again strong anisotropy. Fig.5 SEM/EDS mapping of whisker found on right Sn Stress Measurement: Sine it is generally elieved that the internal stress promotes whisker growth, we also performed in-situ stress measurement during the eletroplating of Sn as well as ex-situ measurement on plated parts. As disussed previously, the in-situ measurements shows a ompressive stress for the right Sn, a tensile stress for the matte Sn and an essentially zero stress for the satin right tin. This is onsistent with the oservation that right Sn shows the most whisker and satin right Sn shows the least whisker.

6 Internal stress in the right Sn oating was also measured using x-ray diffration (XRD). In a XRD experiment, the hange of the lattie onstant due to the stress is measured. The maro-stress an than e alulated from this measurement. XRD provides means for non-destrutive and loal stress measurement. The real-time analysis of stress evaluation during the aging an also e performed. The stress measurement results are summarized in Ta. 1. Measurements were performed oth along the rolling diretion as well as perpendiular to the rolling diretion. As Ta.1 demonstrate, oth diretions show virtually idential stress, indiating an isotropi iaxial stress in the Sn oating. No shear stress was oserved on any samples. The asplated right Sn oating on Cu shows ompressive stress of aout 5 MPa, onsistent with the in-situ stress measurement on this sample. The same sample after aging for 15 months at room temperature shows a muh higher ompressive stress. The inrease of the stress during the aging is most likely related to the intermetalli formation. The diffusion of Cu into the Sn oating oupies more spae and generates ompressive stress in the Sn-oating [12-13]. The uilt-up of ompressive stress during the aging explains the inuation time oserved for the whisker growth on right Sn. The presene of a Ni-underlayer minimizes or even eliminates the diffusion of Cu into Sn and Sn-Cu intermetalli formation. As a result of that, no ompressive stress is oserved on Sn/Ni/Cu. Ta. 1 Internal Stress measured y XRD Aging 0 o 90 o Time (M) Sn/Cu As plated -4 ± 1 MPa -5 ± 2 Mpa Sn/Cu 15 months -13 ± 2 MPa -14 ± 1 MPa Sn/Ni/Cu 3 months 14 ± 1 MPa 12 ± 1 Mpa Conlusion: Loal struture and hemial omposition of the whisker were studied using FIB, SAM and SEM/EDS. Residual maro-stress was also measured using XRD. Compressive stress (residual as well as generated y Sn-Cu intermetalli formation) has een identified to promote whisker formation. Referenes 1. M. Warwik, Implementating Lead Free Soldering European Consortium Researh, Journal of SMT, Volume 12, Issue 4, Otoer Fifth draft of EC Detive on Reyling of Waste Eletrial and Eletroni Equipment, June T. Baggio, The Panasoni Mini Disk Player. Turning A New Leaf in A lead-free Market, IPC Works 99, Mineapolis, MN, Ot., IPC Roadmap: A Guide for Lead-Free Eletronis Assemlies, Seond draft. APEX Meeting, Long Beah, CA, Marh 12 16, P. Harris, The Growth of Tin Whiskers, International Tin Researh Institute Report No. 734, S. C. Britton, Trans. Inst. Met. Fin., 52, p , S. Silverstein, Reasons for Failure Lost with Galaxy 4, Spae News, p3 and p 20, August 17-23, H.S. Park, Requirements to Prelude the Growth of Tin Whiskers, NASA Memo, NASA Parts Projet Offie-Code 310, Goddard Spae Flight Center to QR/Diretor, Reliaility, Maintainaility, and Quality Assurane Division, Feruary 14, Tin Whiskers Formation in Eletroni Components, Lessons Learned Notie y Lokheed Martin Astronautis, Notie #LLN , pp1-4, July G.W. Stupian, Tin Whiskers in Eletroni Ciruits, Aerospae Report No. TR-92 (2925-7, pp. 1-21, Deemer 20, B.D. Dunn, Whisker Formation on Eletroni Materials, ESA Sientifi and Tehnial Review, Vol. 2, No.1, pp.1-22, K.N. Tu, "Cu/Sn Interfaial Reations: Tin-Film Case Versus Bulk Case", Materials Chemistry and Physis, 46, (1996) B.-Z. Lee and D.N. Lee, Spontaneous Growth Mehanism of Tin Whiskers, Ata Mater., 46, (1998)