Estimation of Liquidus Temperature when SnAgCu BGAlCSP Components

Transcription

1 Estimation of Liquidus Temperature when SnAgCu BGAlCSP Components are Soldered with SnPb Paste Jianbiao Pan, Ph.D., Assistant Professor California Polytechnic State University Department of Industrial and Manufacturing Engineering San Luis Obispo, CA 93407, USA Phone: (805) Fax: (805) Abstract Recently, soldering of lead-free components with SnPb paste, or lead-free, is becoming a hot topic. One of major challenges in backward compatibility assembly is development of a right reflow profile for soldering of SnAgCu Ball Grid Array (BGA)/Chip Scale Package (CSP) components with SnPb paste. If SnAgCu reflow profile is used, reflow may be too high for or SnPb components in same board during assembly according to component rating per IPC/JEDEC J-STD-020C. In addition, flux in SnPb may not function properly in such a high reflow. On or hand, if SnPb reflow profile is used, SnAgCu solder ball may only partially melt. The incomplete of with BGA/CSP ball raises serious reliability concern. Therefore, it is important to know minimum reflow peak is able to achieve complete of SnPb paste with leadfree components. This paper presents a method to estimate of mixed compositions when SnAgCu BGA/CSP components are soldered with SnPb paste. The is minimum reflow peak able to achieve complete of SnPb paste with lead-free components. It will be shown depends on Pb ratio in mixed composition and is below 217 C, which is of SnAg3.0CuO.5 solder. The s of several experimental studies in literature are estimated and it is found estimated s are consistent with experimental results. A user interface is designed using Visual Basic for Application in Microsoft Excel environment to facilitate estimation of. It is expected estimation of mixed compositions will be able to guide process engineers to develop a right reflow profile in assembly. Introduction In response to Europe's Restriction of Hazardous Substances (RoHS) and or countries' lead-free directives electronics industry is moving toward lead-free soldering: However, some products, such as servers, are exempt from RoHS directive beyond 20. Additionally products such as medical equipment, and military and aerospace products are not required to be lead-free. These products will continue to be built with conventional tin-lead (SnPb) because reliability of lead-free solder joints for se high reliability applications is still unknown. Since many component manufacturers are migrating to lead-free production, components such as memory modules are no longer available in SnPb finish. Therefore, soldering of leadfree components with SnPb paste, which is called backward compatibility, must be studied. One of major challenges in assembly is development of right reflow profile for soldering of SnAgCu Ball Grid Array (BGA)/Chip Scale Package (CSP) components with SnPb paste. A schematic of BGA/CSP assembly is shown in Figure 1. If SnAgCu reflow profile (peak of 230 to 250 C) is used, full of SnPb paste and SnAgCu ball will be achieved as shown in Figure 2. But reflow may be too high for or SnPb components in same board during assembly according to component rating per IPC/JEDEC J-STD-020C. In addition, flux in SnPb may not function properly at such a high reflow. On or hand, if SnPb reflow profile is used, SnAgCu solder ball will only partially melt and won't be self-aligned as shown in Figure 3. The incomplete of solder and no self-alignment of BGA/CSP ball raise serious reliability concerns. Studies [1-4] show reliability of solder interconnections degraded significantly when SnAgCu ball is only partially mixed with SnPb paste. Gregorich & Holmes [1] reported reliability of backward compatibility assembly when mixed assembly was reflowed at peak of 200 C was much poorer than of control SnAgCu ball with SnAgCu paste in both accelerated cycling test from -40 C to +125 C and mechanical shock test. The poor reliability is believed to be due to inhomogeneous microstructure resulting from partial. The reliability of assembly improves as reflow increased to 225 C. Hillman, [2] evaluated reliability of a BGA package assembled using a peak reflow of 215 C with duration of time above 200 C at 40 seconds. They observed partial in joint microstructure. The reliability of solder joint was very poor as solder joint failed at only 137 cycles in cycling from -55 C to +125 C with dwell times of 11 minutes at each extreme and a ramp rate of C/minute maximum per IPC-97Ol guidelines. Hua, [3-4] reported similar results showing incomplete lead to unacceptable solder joints. Therefore, it is critical to achieve l x/06/$ IEEE th International Conference on Electronics Packaging Technology

2 complete mlxmg of SnPb paste with SnAgCu ball in BGA/CSP assembly. The degree of in assembly is expected to be a function of reflow peak and time above. Grossmann, [5] observed SnAgCu ball was only partially mixed with SnPb paste at reflow peak of 2 C; and SnAgCu ball was totally dissolved when reflow peak exceeded 217 C, which is of Sn3.0AgO.5Cu alloy. They also found a homogeneous reaction of SnPb paste with SnAgCu ball with a minimum formation of voids was achieved at peak reflow of about 230 C. Zbrzezny, [6] investigated various reflow profiles and concluded complete of solders was achieved when reflow peak reached 2l8-222 C. Most of se studies believed full is achieved only when reflow peak exceeds 217 C [4-6], however, a full of SnPb paste with SnAgCu ball can be achieved when peak reflow is below 217 C. For example, Nandagopal, [7-8] observed a full of SnPb paste and SnAgCu ball was accomplished at a peak reflow of2 C for about 15 to 25 seconds. They used Differential Scanning Calorimeter (DSC) to characterize time required to achieve full. Handwerker [9] indicated full occurred at 207 C for sufficient time for a Sn3.9AgO.6Cu solder ball constituting 75% of final solder. Snugovsky, [] described process using a SnPb phase diagram. From study, y concluded a complete mixture may be achieved at a lower than 217 C and depends on solder ball composition, ball/ ratio, dwell time, and component size. Although a considerable amount of work [1-12] has been done so far on assembly and its reliability, minimum able to achieve full is still unknown. The key in assembly is to develop a reflow profile with peak high enough to be able to achieve full of SnPb paste and SnAgCu ball, and peak low enough (prefer below 220 C) so SnPb components won't be damaged. Therefore, it is critical to know minimum reflow peak is capable of achieving a complete of SnPb paste with lead-free components. This paper presents an approach to estimate mixed composition when SnAgCu BGA/CSP components are soldered with SnPb paste. The mixed composition is minimum reflow peak able to achieve complete of SnPb paste with lead-free components. First, calculation of mixed composition is described. Then estimation of of mixed composition is presented. The s of different BGA/CSP component pitch levels are estimated. The s of several experimental studies in published literatures are estimated as well, and it is found estimated s are consistent with published experimental results. Finally, paper describes a user interface designed to facilitate estimation of. W Liquidus Temp = C Liquidus Temp =183C Figure 1. A schematic of BGA/CSP Backward Compatibility Figure 2. achieved when SnAgCu reflow profile is used Figure 3. when SnPb reflow profile is used Mixed Composition Calculation There are four alloying elements in mixed composition when SnAgCu BGA/CSP components are soldered with SnPb paste: Sn, Ag, Cu, and Pb. The percentage of each metal element in mixed composition can be calculated W - fpb X V Paste X fm x d SnPb (1) Pb - Vpaste X fm x d SnPb + V Eall X dsnagcu = fag X V Eall X d SnAgCu A g V Paste X fm x d SnPb + V Eall X d snagcu (2) (3) W Sn = loo-wpb-w Ag - W eu (4) where W pb, WAg, W Cu ' and Wsnare weight percentage, Ag, Cu, and Sn in mixed compositions, respectively; fpb is percentage in weight in SnPb

3 ; h g and feu are weight percentage of Ag and Cu in SnAgCu alloy; fm is volume percentage of metal content in SnPb ; d SnPb and dsnageu are density of SnPb and SnAgCu alloys. Vpaste is SnPb volume, which can be calculated L2 H(TR) Vpaste ={ for square aperature n(~jh(tr) for round aperture where L is stencil aperture length for square aperture, H is stencil thickness, D is stencil aperture diameter for round aperture, and TR is transfer ratio, which is defined as ratio of volume of deposited to volume of aperture. Vball is volume of a solder ball in BGA/CSP component. If ball diameter, D, is given, ball volume can be calculated Vball = ~ n( ~ J (6) (7) If sphere is reflowed and ball height, H, and radius, R, ::b;;~:m~ ;~r(~; RJj "" g;ven; For eutectic SnPb, fpb is 37 and typical value of f m is (or 50%). For Sn3.0AgO.5Cu solder alloy, fag =3.0 and feu =. The density of eutectic Sn37Pb, d SnPb ' is 8. 4g/cm3 and density of Sn4.0Ag05, dsnageu' 3 is 7.394g/cm [14]. For example, a 1 mm pitch Xilinx fine-pitch BGA (FG676) package with a 0.61 mm (24 mil) Sn3AgO.5Cu ball diameter is assembled with Sn37Pb paste. The is printed using a mm (5 mil) stencil with a mm (18 mil) diameter round aperture. The SnPb paste has 50% metal content in volume. Assume a 90% transfer ratio. Using Equation 6, we can calculate volume of Sn3AgO.5Cu ball is 0.118mm3 (7235 mi1 3). Using Equation 5, we can calculate volume of SnPb paste is mm3 (1125 mie). Using Equations 1 to 4, we can get final mixed alloy composition: 37 x x x 8.4 = x x x W = 3.0xO.118x7.394 =2.75 A g x x x W = x x = 046 Cu x x x W = pb WSn = =93.7 Therefore, mixed composition is 93.7% Sn, 3. 1% Pb, 2.8% Ag, and % Cu, all in weight. Estimation of Mixed Composition Liquidus Temperature After we know mixed compositions, next question becomes what is of mixed composition Sn3.lPb2.8AgO.5Cu. The phase diagram of common binary and ternary systems are relevant to solders can be obtained from National Institute of Standards and Technology (NIST) webpage (available at gov/phase/solder/solder.html). But phase diagram of complex quaternary SnPbAgCu is currently not available. The phase equilibria can be calculated from rmodynamic databases using CALPHAD method [15]. Thermodynamic calculation is a very useful tool in obtaining phase diagram information, but it requires reliable rmodynamic databases and specialized knowledge. In this paper, of ternary and quaternary systems is estimated using simple linearization of binary lines. The linearization method has been successfully used by National Institute of Standards and Technology (NIST) since National Center for Manufacturing Sciences (NCMS) lead-free study [16]. Thus, of quaternary SnPbAgCu system, a typical alloy system in assembly, can be calculated Tz = 232 C WAg -7.9Weu -1.3Wpb (8) with limits: WAg < 3.5; W Cu < 0.7; W Pb < 38; where Tz is of Sn-rich solder alloys; 232 is of Sn; WAg, W Cu, and WPb are percentage in weight of Ag, Cu, and Pb, respectively. The coefficients before se alloy elements are slope of binary lines. For example, 1.3 is slope of SnPb binary lines when Pb is less than 38% in weight; 7.9 is slope of SnCu binary lines when Cu is less than 0.7% in weight; and so on. It should be emphasized limitation of simple linearization is WAg < 3.5, Wcu < 0.7, and W Pb < 38. It should also be noted Equation 8 is an approximation. Based on Equation 8, of Sn3.0AgO.5Cu, Tz = 232 C x x -1.3 x 0 = 219 C If Ag content is over 3.5% and less than 4% wt, Ag 3 Sn is primary phase. In this case, Equation 8 is not valid. A simple Thus, fix is to add 5 C to Equation 8. of Sn3.8AgO.7Cu is ~ =232 C-3.1 x x x 0+ 5 = 220 C Currently reflow profiles in assembly are developed through costly trial-and-error method. It is expected estimation of mixed compositions will be able to guide process engineers to develop right reflow profile in assembly. Table 1 summarizes final joint compositions and with Sn3AgO.5Cu (SAC305) ball and Sn37Pb paste for typical BGA/CSP component pitch levels. The aperture size, shape, stencil thickness and ball diameter are based on Solectron's guideline for no-clean paste. The transfer ratio is assumed based on experiences. It shows final is lower than 2l7 C, of SAC305. The can be as low as 203 C. As pitch level decreases (except mm pitch), weight percentage increases and decreases. Equations 1 to 4 imply

4 depends depends on ratio of BGA ball volume and paste volume. Table Final Final Joint Compositions and Liquidus Temperature with Sn3.0AgO.5Cu Sn3.OAgO.5Cu Ball and Sn37Pb Paste Pitch Pitch (mm) (mm) size in mm mm size III shape shape Stencil Stencil thickness in thickness III mm mm Transfer Transfer ratio ratio (%) (%) Ball Ball diameter diameter %ofpb % %ofag % of Ag %ofcu % of Cu temp. temp. (0C) ( C) (21) (21) (18) (18) (16) (16) (14) (14) (11) (1 1) Round Round (6) (6) (4) (4) in Equation shows aa higher Pb percentage mixed mixed Equation 88 shows higher Pb percentage in compositions can reduce reduce mixed mixed composition compositions can composition The higher Pb percentage can be be achieved achieved by.. The higher Pb percentage can by more SnPb SnPb solder or reducing SnAgCu printing more paste or reducing SnAgCu printing in solder solder ball. ball. But But itit isis unclear unclear what what effect effect of of high Pb content content in high Pb mixed compositions on backward compositions on compatibility reliability. reliability. mixed al. [17] studied effect effect of Pb contamination contamination on on Zhu, Zhu, et [17] studied lead-free lead-free solder solder joint microstructure and and observed observed aa Pb-rich Pb-rich joint microstructure in formed in bulk bulk solder solder when when lead-free lead-free solder solder phase formed phase contains Pb Pb impurity. discussed influence influence of of contains impurity. Zeng Zeng [18] [18] discussed Pb-rich Pb-rich phase on solder solder joint The Pb-rich Pb-rich phase phase on joint reliability. reliability. The phase in be weakest weakest region bulk bulk solder, and crack crack may be region in solder, and may Pb-rich Pb-rich phase interface during may propagate propagate along along phase interface during may al. [11] found reliability testing testing [17, [17, 18]. 18]. Bath, Bath, et [11] found reliability in accelerated of backward accelerated reliability of compatibility assembly assembly in reliability 1 00 C with from O C to 0 C with cycling cycling (ATC) (ATC) from 0 C to minute aa cycle, even when when full full was achieved, was cycle, even was achieved, was minute than of of both both SnAgCu poorer than SnAgCu ball/snagcu ball/snagcu paste paste and and poorer SnPb ball/snpb But Bandagopal, al. [7] found ball/snpb paste. paste. But Bandagopal, et [7] found SnPb in both reliability of backward both reliability of compatibility assembly assembly in ATC from from O C to 0 C and -40 C to 125 C was better better than than 0 C to 0 C and -40 C to 125 C was ATC SnPb SnPb assembly when full full was achieved. achieved. assembly when was reliability data of of SnPb SnPb BGA BGA ball ball soldered soldered Furrmore, reliability data Furrmore, with SnAgCu forward compatibility), where high Pb SnAgCu paste paste (or (or forward compatibility), where high Pb with in content isis in mixed mixed compositions, was better better or or equal to compositions, was equal to content of of SnPb SnPb ball/snpb control assemblies assemblies [19]. ball/snpb paste paste control [19]. no conclusion conclusion can can be be drawn drawn yet what Therefore, no yet regarding regarding what Therefore, effect of Pb content has on on content has backward compatibility effect reliability. reliability. To assess assess method, we estimated estimated method, we To from published studies and s from published experimental experimental studies s and estimated estimated with published compared with published compared results. The The estimated estimated s and experimental results. s and experimental in Table If results are are summarized summarized in 2. If published experimental experimental results published Table 2. reflow peak used is higher than estimated estimated, full is expected. Orwise, partial is expected. Overall, Table 22 shows estimated s are consistent with reported experimental variances between experimental results. There are small variances studies in in estimated and reported results of studies references references 7 and 11. This could be due to inaccuracy transfer ratio assumptions. Since only a few studies have reflow peak s close to estimated, furr experimental study is needed to validate validate accuracy of estimation method. Temperature Table 2. Comparison of estimated Liquidus Temperature and Reported Reported Experimental Results Reference Reference Gregorich & Gregorich & Holmes [1] [1] Holmes Hillman, et et Hillman, al. [2] [2] al. Grossmann, Grossmann, al. [5] [5] et Nandagopal, Nandagopal, al. [7] [7] et Bath, et al. Bath, [11 [11]] 2090C 209 C 2190C 219 C Peak reflow reflow Peak used used 2000C 200 C 2250C 225 C 2150C 215 C 2160C 216 C 2120C 212 C 20C 2 C 2170C 217 C 20C 2 C 2180C 218 C 2050C 205 C 2270C 227 C 2140C 214 C Experimental Experimental results results User User Interface Interface Development Development To facilitate To facilitate estimation estimation of of mixed mixed composition composltlon,, aa user user interface interface was was developed developed using using Visual for Application Visual Basic Basic for Application in in Microsoft Microsoft Excel Excel environment. Focus environment. Focus has has been been on on designing designing an an easy-toeasy-tounderstand user understand user interface. interface. One One major major feature feature with with user user interface is is interface input input data, data, control control menu menu and and results results are are presented presented on on same same page. page. Anor Anor feature feature is is default data data from from typical default typical assembly assembly guidelines guidelines will will be be shown shown when when aa user user selects selects aa package package pitch pitch level. level. The The user user is is also also allowed allowed to to manipulate manipulate all all input input data. data. The The input input parameters parameters include package include package pitch, pitch, package package solder solder ball ball type, type, stencil stencil dimensions, and dimensions, and solder paste compositions. compositions. The The output output data data include mixed mixed compositions, compositions, solder include paste volume, volume, and and estimated estimated.. Figure Figure 44 shows shows interface interface for BGA/CSP BGA/CSP component form. for component user user form. Summary Summary A A method method to to estimate estimate mixed mixed composition composition when when lead-free lead-free components components are are soldered soldered with with tintinlead solder lead paste was was presented. presented. The The estimated estimated of published s of published experimental experimental studies studies are are consistent consistent s with reported reported experimental experimental results. with results. The The of of backward compatibility is lower than of assembly SnAgCu alloy. The The assembly is lower than of SnAgCu alloy. on depends on depends ratio ratio of Pb in in mixed mixed compositions. compositions. The The ratio of is influenced Pb is influenced by by BGA BGA ball ball volume, volume, ball ball solder solder ratio alloy alloy types, types, and and solder paste volume. volume. Generally Generally speaking, speaking, smaller of BGA/CSP BGA/CSP components, components, higher higher smaller pitch pitch level level of weight, Pb, thus thus lower lower weight percentage percentage of.

5 . But effect of high Pb content on reliability of assemblies is still unclear. A user interface for estimation was designed using Visual Basic for Application in Microsoft Excel environment. It is expected estimation of mixed compositions will be able to gmde process engineers to develop right reflow profile m assembly. Future experimental study is needed to confirm accuracy of estimation method for of mixed compositions. Acknowledgments The author would like to express his gratitude to Solectron Corporation, Milpitas, CA, for technical and financial support; to Kim Hyland, Dennis Willie, and Jasbir Bath without whom this work would not have been possible. The 'author would like to thank Nick Benigno for implementation of user interface. BGA Please input following data: - ['g] Package Dimensions Select Ball Type: Pitch(mm):~ Pitch (mils): rzo- I SAC305 ::oj r Solder Paste *Note: lmm equals roughly 40 mils Ball Diameter (mils): ~ Sn: ~ % I' Calculate Pb:.1 Stencil Dimensions -- I Shape: I ::oj " booooc },,-\- Size (mils): I 11 Stencil Thickness (mils): I 5 Transfer Ratio: I 70 % Pitch ~ % Back I Metal ~ Content in 50 % Paste: Help Clear I I Results: %Pb p %Agf2:0Z %Sn ~ %Cu ~ Calculated Liquidus Temperature: I C Solder Vol. (mii A 3): I Figure 4. BGA/CSP User Interface for Estlmatlon ofliqmdus Temperature References 1. Gregorich T. & Holmes P. Low-Temperature, High Reliability Assembly of Lead-free CSPS, IPCIJEDEC 4th International Conference on Lead-free Electronic Components andassemblies, Frankfurt, Germany, Hillman D., Wells M, and Cho K. The Impact of Reflowing a Pb free Solder Alloy Using a Tin/Lead Solder Alloy Reflow Profile on Solder Joint Integrity, International Conference on Lead-free Soldering, CMAP, Toronto, Canada, May 24-26, Hua F, Aspandiar R, Rothman T., Anderson C., Clemons G and Klier M Solder Joint Reliability of Sn-Ag-Cu BGA Components Attached with Eutectic Pb-Sn Solder Pate, Journal ofsurface Mount Technology, 2003, 16 (1) Hua F, Aspandiar R, Anderson C., Clemons G., Chung C., and Faizul, M. Solder Joint Reliability Assessment of Sn Ag-Cu BGA Components Attached with Eutectic Pb-Sn Solder, SMTA International, Chicago, IL, pp Grossmann G., Tharian 1., Jud P. and Sennhauser U. Microstructural investigation of lead-free BGAs soldered with tin-lead solder, Soldering & Surface Mount Technology, 2005: 17 (2), Zbrzezny AR, Snugovsky P, Lindsay T, and Lau R Reliability Investigation of Sn-Ag-Cu BGA Memory Modules Assembled with Sn-Pb Eutectic Paste Using Different Reflow Profiles, International Conference on Lead-free Soldering, CMAP, Toronto, Canada, May 24-26, Nandagopal B., Mei Z., and Teng S. Microstructure and Thermal Fatigue Life of BGAs with Eutectic Sn-Ag-Cu Balls Assembled at 2 C with Eutectic Sn-Pb Solder Paste, Proceedings of2006 IEEE Electronic Components and Technology Conference, San Diego, CA, 2006: Nandagopal B., Chiang D., Teng S., Thune P., Anderson 1., Jay R and Bath 1. Study on Assembly, Rework, Microstructures and Mechanical Strength of Backward Compatible Assembly, Proceedings of 2005 SMTA International, Chicago, IL, 2005: Handwerker C. Transitioning to Lead Free Assemblies, Printed Circuit Design and Manufacturing, March 2005:

6 . Snugovsky P., Zbrzezny AR, Kelly M, and Romansky M. Theory and Practice of Lead-free BGA Assembly using Sn-Pb Solder, International Conference on Lead-free Soldering, CMAP, Toronto, Canada, May 24-26, Bath J, Sethuraman S, Zhou X, Willie D, Hyland K, Newman K, Hu L, Love D, Reynolds H, Kochi K, Chiang D, Chin V, Teng S, Ahmed M, Henshall G, Schroeder V, Nguyen Q, Maheswari A, Lee MJ, Clech J-P, Cannis J, Lau J, Gibson C. Reliability Evaluation of Lead-free SnAgCu PBGA676 Components using Tin-Lead and Lead-free SnAgCu, Proceedings of 2005 SMTA International, Chicago, IL, 2005: Theuss H, Kilger T, and Ort T. Solder Joint Reliability of Lead-Free Solder Balls Assembled with SnPb Solder Paste, Proceedings of 2003 IEEE Electronic Components and Technology Conference, New Orleans, LA, 2003: Pan J and Bath 1. Lead Free Soldering Backward Compatibility, IPCIJEDEC 12th International Conference on Lead Free Electronic Components and Assemblies, Santa Clara, CA, March 7-9, Siewert T, Liu S, Smith DR, Madeni JC. Database for solder properties with emphasis on new lead-free solders. NIST & Colorado School of Mines, Release 4.0, Feb. 2002, gov/div853/lead free/solders.html 15. Kattner UR. and Handwerker CA. Calculation of Phase Equilibria in Candidate Solder Alloys, Z. Metallkd. 92 (2001): Kattner UR. and Handwerker CA. personal communication though s, l7.zhu Q, Sheng M, and Luo 1. The effect contamination on microstructure and mechanical properties of SnAg/Cu and SnSb/Cu solder joints in SMT, Soldering and Surface Mount Technology, 12 (3), 2000: Zeng X. Thermodynamic analysis of influence contamination on Pb-free solder joint reliability, Journal ofalloys and Compounds, 348 (2005): Handwerker CA, Bath J, Benedetto E, Bradley E, Gedney R, Siewert T, Snugovsky P, and Sohn 1. NEMI Lead-free Assembly Project: Comparision Between PbSn and SnAgCu Reliability and Microstructure, Proceedings of SMTA International, 2003.