A Viable Tin-Lead Solder Substitute: Sn-Ag-Cu

Transcription

1 Journal of Electronic Materials, Vol. 23, No. 7, 1994 Special Issue Paper A Viable Tin-Lead Solder Substitute: Sn-Ag-Cu CHAD M. MILLER, IVER E. ANDERSON, and JACK F. SMITH Ames Laboratory, Iowa State University, Ames, IA Rising concern over the use of lead in industry provides a driving force for the development of improved lead-free industrial materials. Therefore, a new leadfree base solder alloy Sn-4.7Agl.7Cu (wt.%) has been developed upon which a family of lead-free solders can be based. This solder alloy exhibits a ternary eutectic reaction at _+ I~ (L -~ q + O + [3-Sn; n = Cu6Sn~, 0 = Ag3Sn). Preliminary tests ofsolderability demonstrate intermetallic phase formation on model solder joint interfaces and good wettability in a fluxed condition suggest technological viability and motivates much more extensive study of this solder alloy. Key words: Differential thermal analysis, lead-free solder, scanning electron microscopy, tin-silver-copper eutectic, wavelength dispersive spectrometry, x-ray diffraction INTRODUCTION Lead metal and its compounds are known toxins when inhaled or ingested. Current congressional legislation is threatening to heavily tax or completely ban the use of lead in commercial industries. 1 This would further burden the U.S. electronics industry which is already reeling from overseas competition. In addition to the toxicity of lead, there are other problems concerning tin-lead solder. Current tin-lead solders lack shear strength, resistance to creep and resistance to thermal-mechanical fatigue? A solder which exhibits enhancements of these properties is crucial in avionics and automotive applications where the solder is subjected to many thermal cycles and sustained temperatures of up to 150~ Hence, there is a real need to develop new solders that have similar processing properties to tin-lead solders, but which are lead-free and have improved mechanical properties and microstructural stability. (Received December 20, 1993; revised March 7, 1994) This paper is a summary of a work in progress concerning the development of such a lead-free solder. The Sn-Ag-Cu alloy system has potential as a base alloy for a new family of lead-free solders. This system was chosen on the following criteria: i) reduced toxicity compared to tin-lead solders, ii) availability, iii) low cost, iv) reasonable wetting behavior, and v) moderately low melting temperature (from the published ternary phase diagram). 4 Published Sn-Ag-Cu phase diagram data 4 indicate a ternary reaction in the Sn-rich region at 225~ that is a class II four-phase equilibrium reaction (L + q --~ 0 + [~-Sn). ~ In the earlier study, 4 only one alloy sample was investigated in the Sn-rich region of the Sn-Ag- Cu system: Sn-5Ag-5Cu (wt.%). The present study included a larger set of Sn-rich ternary alloys and the characterization results were improved by utilizing modern microstructural, chemical and thermal analysis methods. The objective of this study was to determine whether or not a ternary eutectic reaction had been overlooked in this region of the phase diagram and, if identified, to investigate the alloy's behavior in 595

2 596 Miller, Anderson, and Smith Table I. Description and Composition of 25 g Chill Cast Alloy Samples Sample Number Composition of Samples Sample Description wt.% Sn wt.% Ag wt.% Cu 1 2a,2b Previously published four-phase equilibrium composition Four-phase equilibrium composition determined in this study Composition half way between samples 2 and 4 on ternary diagram Sn-Ag eutectic composition to verify binary phase diagram Pure Sn sample (99.99) for calibration of DTA instrument Table II. Type of Analysis Performed on Each 25 g Chill Cast Sample Composition of Optical Sple. Samples (wt.%) Micro- No. Sn Ag Cu scopy ~] q 2a ~] ~/ 2b ~] ~] q ~/ q *X-ray performed on bulk and atomized samples. DTA WDS X-Ray Cop~r wbms c~aned } ~h n~dc acid,.nsed ~h water. ~rex tube in Sy~em mmov~ ~rom fum~ and acelo~ and melhal~f, fumace at 250"C all~ed to eir co~. NO Flux ~, Fig. 1. Schematic of trial solder joint procedure. some solder processing applications. EXPERIMENTAL PROCEDURE Sample Preparation Several 25 g alloys in the Sn-rich region of the Sn- Ag-Cu system were prepared, see Table I. The constituent elements (>99.99% pure) of each sample were cleaned, weighed, and melted in 6 mm diameter Pyrex capsules under 1/3 atm He at 300~ for 1 h. Each sample was quenched into an ice water bath while remaining in the capsule. A large (4500 g) alloy sample of the new Sn-4.7Ag- 1.7Cu (wt.%) four-phase equilibrium composition was also cast by the Materials Preparation Center of Ames Laboratory. The alloying elements were heated, melted, and homogenized in an alumina crucible at 600~ under a 10 mtorr vacuum. After induction stirring for 15 rain at 600~ the melt was chill cast as an ingot in a Cu mold. To investigate the malleability of the solder, portions of this 10 cm diameter ingot were deformation processed by rod rolling, swaging, wire drawing, and rolling. The remaining 4000 g of the ingot was atomized with high pressure gas atomization (HPGA) 8 to prepare fine, spherical powder for x-ray diffraction analysis and for eventual formulation into fine and ultra-fine pitch solder pastesy The atomization process used Ar atomization gas at 7.6 MPa (1100 psi) and a pouring temperature of 300~ Analysis Microscopy Selected 25 g cast alloys were analyzed with optical microscopy, see Table II. Samples were sectioned with a diamond saw, mounted in epoxy, polished, and etched with 2 vol.% HC1 s in methanol. Each sample was then examined with an optical microscope. Atomized powder samples were analyzed with a scanning electron microscope (SEM). The SEM was used to determine the size and morphology of the HPGA solder powder. Wavelength Dispersive Spectrometry Wavelength dispersive spectrometry (WDS) provided quantitative composition information. Eight WDS runs were performed on a section of the 25 g cast Sn-3.6Ag-1.5Cu (wt.%) sample, the composition of the earlier reported L + q -~ 0 + ~-Sn reaction. 5 The eutectic regions in the sample were large enough (>30 ~tm) for WDS measurements to be made. The sample was mounted using the same procedure as the optical microscopy samples. The WDS analysis was performed with the beam current set at 30 na and accelerating voltage at 20 kv. Differential Thermal Analysis Differential thermal analysis (DTA) provided accurate determination of melting temperatures of the samples. The DTA was used to determine melting temperatures of mg samples of as cast alloys. The heating rate was 5~ with argon cover gas flowing through the DTA chamber at 50 cc/min. X-Ray Analysis X-ray diffraction was performed on both bulk and high pressure gas atomization powder samples using a diffractometer and CuK radiation. Phases present in the samples were determined by indexing the diffraction patterns. The scans were performed from 20 = ~ at l~ with a step size of 0.03 ~ This analysis verified the identity of phases that wavelength dispersive spectrometry and optical micros-

3 A Viable Sn-Pb Solder Substitute: Sn-Ag-Cu 597 a b c Fig. 2. Microstructure of the Sn-3.6Ag-1.5Cu (wt.%) chill cast 25 g sample: a fine eutectic (dark) penetrated by J3-Sn dendrites (light) (a, b), ~-Sn dendrites containing the acicular Cu6Sn s intermetallic phase (c). I o wt% Sn, Normalized I [ n wt% Ag, Normalized I -- e- - wt% Cu, Normalized i 100 I I I I I I I F ~ ~ 12 ave. wt% Sn 10 wt% Sn, 8 wt% Ag & Cu, Normalized go Normalized 85.~ ~ ave. wt% Ag -- ~ ~'~ 4 ~-" --~ ~ "~" ~, ave. wt%cu 80 I ~ L I o Run Number Fig. 3. Plot showing the results of each WDS run performed on different eutectic regions of the Sn-3.6Ag-l.5Cu (wt.%) sample. The error bars indicate standard deviation for each run for the respective element. copy indicated were present. Soldering Trials To explore the behavior of the new ternary eutectic alloy as a solder, trial solder joints were fabricated. Two twisted pairs of copper wire were dipped into individual one gram samples of liquid solder in air, Fig. 1. The twisted pairs were first cleaned in an aqueous solution of 30 vol.% nitric acid then rinsed with water, acetone, and methanol and allowed to air dry. One set of wires was fluxed with a zinc-chloride based flux immediately before soldering and the other set of wires was not fluxed. The solder was melted in open ended Pyrex tubes at 250~ Each set of wires was then dipped into molten solder and allowed to air cool in its Pyrex tube. Each solder joint was characterized in cross section to qualitatively examine the microstructure and wetting behavior of the solder. RESULTS The optical microscopic examination of the 25 g cast Sn-3.6 Ag-l.5 Cu (wt.%) sample revealed a eutectic microstructure penetrated by ~-Sn dendrites as seen in Figs. 2a and 2b. Inside some of the ~-Sn dendrites an acicular third phase was visible, Fig. 2c. The morphology of this phase matches that of the Cu6Sn ~ phase. 9 It was determined from the morphological characteristics and by WDS that this was indeed the intermetallic Cu6Sn 5 phase. It is unclear whether the Cu6Sn 5 phase resulted from insufficient dissolving of Cu during casting or whether it formed during solidification. Because the areas with eutectic microstructure were on the order of 30 pm in diameter (Fig. 2b), it was possible to analyze these areas with WDS to estimate the apparent eutectic composition. The compositions of eight individual WDS analyses of the eutectic region of the Sn-3.6Ag-l.5Cu (wt.%)

4 598 Miller, Anderson, and Smith sample are summarized in Fig. 3. The WDS composition analysis results were averaged yielding a grand average composition of Sn-4.7Ag-l.7Cu (wt.%). The standard deviation of the Ag and Cu compositions are 0.7 and 0.3, respectively. Based on this estimate of the ternary eutectic composition, two 25 g alloys of Sn- 4.7Ag-l.7Cu (wt.%) were cast (samples 2a and 2b of Table I). Optical microscopy performed on the 25 g Sn-4.7Ag- 1.7Cu (wt.%) samples revealed that the alloyswere composed almost completely of a eutectic microstruc- 100 i i J,, Sn-4.7 Ag-l.7 Cu (wt%) Powder Sample,, I.... I.... I, 80" c" ii, r e, degrees 100 Sn-4.7 Ag-l.7 Cu (wt%) Bulk Sample,,, ~ I, i L I i i i i I,, L, I i i ~ l l r l l l 80 84,J,' o~ 60 40" "~ I.... I.... i.... I.... i b b Fig. 4. Optical micrographs of the chill cast Sn-4.7Ag-l.7Cu (wt.%) Fig. 5. X-ray scans of the Sn-4.TAg-l.7Cu (wt.%) alloy in both showing a fine eutectic microstructure (a, b). atomized (a), and bulk form (b). 2 e, degrees Table III. Onset of Melting Temperatures for Various 25 g Chill Cast Samples Samples in Sn-Rich Composition of Samples Measured Onset Region of Sn-Ag-Cu System wt.% Sn wt.% Ag wt.% Cu of Melting, ~ a b Average onset of melting of Samples 1~ Reference Samples Published Melting Temperature, ~ ~

5 A Viable Sn-Pb Solder Substitute: Sn-Ag-Cu 599 ture (see Figs. 4a and 4b). The spacing of the eutectic was on the order of 0.5 ~m as shown in Fig. 4b. The eutectic microstructure of the Sn-4.7Ag-l.7Cu (wt.%) sample led to the hypothesis that this composition may correspond to the true ternary eutectic composition in the Sn-Ag-Cu system. To test this hypothesis, DTA measurements of 25 g cast samples in the tin-rich region of the Sn-Ag-Cu system (including Sn-3.6Ag-l.5Cu [sample 1, Table I], Sn-4. Ag-l.7Cu [samples 2a and 2b] and Sn-4.1 Ag- 0.9 Cu (wt.%) [sample 3]) were performed. These measurements showed an average onset of melting at 216.8~ and a variation of only _+0.4~ (see Table III). This temperature is lower than both the published binary eutectic temperature of Sn-Ag (221~ 1~ and the previously published four phase ternary equilibrium temperature (225~ 4 The accuracy of the DTA instrument was determined by measuring the melting temperature of a pure tin sample. The measured melting temperature of pure Sn (sample 5) was 231.2~ within 0.8~ of the accepted published value of ~ 1~ indicating that the accuracy of the DTA apparatus was within specified limits. The melting temperature of the Sn-Ag eutectic was also measured as 220.8~ within 0.2~ of the published value of 221oC. lo To further test the hypothesis, x-ray diffraction (see Figs. 5a and 5b) was performed on the Sn-4.7Ag- 1.7Cu (wt.%) alloy in atomized and as cast form. The Bragg reflections were indexed as belonging to [~-Sn, Ag~Sn, or Cu~Sn 5 phases. These phases constitute the eutectic microstructure seen in Fig. 4a and 4b. This supported the results of the WDS and optical microscopy and indicated that these three phases were present in the solidified sample. Two peaks labeled "x" were not indexed to these phases. Deformation processing of portions of the large 10 cm diameter cast Sn-4.7Ag-l.7Cu (wt.%) alloy produced useful sizes of solder sheet and wire. The alloy was rolled as thin as 100 ~m (4 mil) and still displayed ductile behavior; i.e. the sheet could repeatedly be bent back and forth without fracturing. The wire was drawn to a diameter of 760 ~m (30 mil) and appeared to have suitable ductility to be rolled into a coil for use in a hand soldering applications. Atomized powders of the Sn-4.7Ag-l.7Cu (wt.%) composition were observed with a scanning electron microscope (SEM), Fig. 6. The powders are spherical with an average particle diameter on the order of 10 ~tm in diameter. Examination of the fluxed solder joint with optical microscopy showed an intermetallic layer at the solder-copper interface Fig. 7a. There was also significant wetting by the solder on the copper wire which can be seen in Fig. 7b. In contrast to the fine eutectic microstructure of Figs. 4a and 4b, acicular and coarsened intermetallic phases were observed in the solder joint, Figs. 7a and 7b. The gray shaded, Fig. 6. Scanning electron micrograph of atomized Sn-4.7Ag-l.7Cu (wt.%) solder powders. b Fig. 7. Fluxed trial solder joint: lighter regions are Cu wires, darker regions are solder. Intermetallic layer formed atthe solder/cu interface demonstrating good bonding between the solder and Cu (a), gray, needle-like phase believed to be Ag3Sn, light, rounded phase believed to be Cu6Sn s (a), solder has low contact angle with the Cu wire indicating good wetting (b).

6 600 Miller, Anderson, and Smith needle-like phase is believed to be Ag3Sn while the light shaded, rounded phase is believed to be Cu6Sn 5. This coarsening is most likely a result of the slow cooling rate of the relatively large solder joint. As expected, the wetting Was greater for the fluxed joint than for the unfluxed joint. Experimental Results DISCUSSION The apparent eutectic melting temperature of samples 1-3 (Table III) is 216.8~ about 8~ lower than the published four-phase equilibrium temperature of 225~ Since 216.8~ is lower than the adjacent binary eutectic melting temperatures of the system, Sn-3.5Ag at 221~ and Sn-l.3Cu (wt.%) at 227~ this supports the conclusion that the Sn-Ag- Cu system contains a class I four-phase equilibrium (ternary eutectic) reaction at _+ I~ L ~ ~ [~-Sn(L = liquid, TI = Cu6Sn 5, 0 = Ag3Sn). The presence of these three phases in the as-solidified microstructure was also confirmed by x-ray analysis, Figs. 5a and 5b. The consistency of the melting temperatures of samples in the tin-rich region indicates that an isothermal equilibrium plane, which is characteristic of a class I four-phase equilibrium reaction, exists in the Sn-Ag-Cu system. This contradicts the previously published 4 class II four-phase equilibrium reaction at 225~ L + Ti -~ O + ~-Sn. The x-ray diffraction results shown in Figs. 5a and 5b show two "x" peaks which could not be identified. The overall difference in intensities of the two x-ray samples occurred because one sample was in bulk form and the other in powder form. Preferred orientation of the grains in bulk samples can result in x-ray diffraction intensities which do not accurately represent the volume percent of phases in the sample. Powder x-ray diffraction samples contain many randomly oriented grains and generally yield x-ray diffraction intensities which better represent the intrinsic volume fraction of the phases in the sample. The initial solder joint trials showed that the Sn- Ag-Cu solder can indeed function as a solder (Figs. 7a and 7b). A bonding layer that probably consists of the Cu3Sn and Cu6Sn 5 phases 9 at the solder/cu interface is visible. The low contact angle of the solder/cu interface indicates that the solder wets reasonably well, Fig. 7b. Factors which contributed to the enhanced thickness of the solder/cu bonding layer and the coarse intermetallic phases in the solder microstructure, Fig. 7a, are the time the joint was held at 250~ and the relatively slow cooling rate of the trial soldering process. One would expect that the solidification microstructure of the solder in an actual joint would more closely resemble the chill cast microstructure in Fig. 4b since a typical solder joint would be much lower in mass and would air cool much faster resulting in a more refined microstructure. 1~ Future Work The overall study of the Sn-Ag-Cu system includes both the experimental and theoretical determination of the ternary constitution relationships in this system By calculating the ternary phase diagram for this system, one may gain further insight into the ternary solidification reactions and predict the effects of quaternary alloying elements on the melting and solidification behavior of modified solders. Although not reported herein, initial work in this area included calculation of the respective binary phase diagrams for this system. Once completed, the binary data will be combined with experimental ternary alloy results to form the basis for calculating the ternary Sn-Ag-Cu diagram. Tests to determine the mechanical behavior of this alloy are underway. The microstructural effects of changes in cooling rate and annealing times and temperatures will also be investigated. In addition to these experiments, the effects of quaternary alloying additions on the microstructure and wetting behavior of the solder will be examined. The wetting behavior of this solder and a series of other lead-free solder alloy candidates is being extensively tested by Dr. Fred Yost at the Center for Solder Science and Technology at Sandia National Labs as part of a parallel collaborative study and will be reported at a later date. Preliminary results of an industrial trial performed at Allied Signal-Kansas City Division have indicated also that the Sn-Ag-Cu system performs favorably as a solder when used in wire form. Additional industrial testing and quantification will be conducted as part of technology transfer efforts related to this project. The solder powders that were produced in this study will be formulated into solder paste and subjected to a series of application specific tests in further work. SUMMARY A new lead-free base solder, Sn-4.7Ag- 1.7Cu (wt.%) has been developed which shows great promise as a lead-free solder system. This solder exhibits a ternary eutectic reaction; i.e., contains a class I four-phase equilibrium, L ~ ~ ~-Sn (11 = Cu6Sns, 0 = Ag3Sn) at a temperature of _+ I~ Initial trial solder joints made with this solder revealed that the Sn- 4.7Ag-l.7Cu (wt.%) alloy can function as a solder by wetting and forming a high quality bond with copper. The melting temperature of 216.8~ is slightly higher than currently available tin-lead solder pastes (Teu t ~ With further modifications, such as quaternary alloying additions, based on experimental and theoretical phase diagram relationships, the properties of the Sn-Ag-Cu solder may be enhanced to provide an alternative solder alloy for many conventional leaded solders used in electronic applications. ACKNOWLEDGMENTS The authors would like to acknowledge the support of Ames Laboratory, Iowa State University, funding from the U.S. Department of Energy (under contracts W-7405-Eng-82 [BES] and WM02 [OTD]) and the following people for their technical assistance: Alfred

7 A Viable Sn-Pb Solder Substitute: Sn-Ag-Cu 601 Kracher (Iowa State University, Geology Department), Kevin Dennis and Joel Harringa (Ames Lab), Austin Chang and Shuanglin Chen (University of Wisconsin-Madison), and Ursula Kattner (National Institute of Standards and Technology, Gaithersburg, Maryland). REFERENCES 1. L.E. Felton, C.H. Raeder and D.B. Knorr, JOM 45, 28 (1993). 2. Sungho Jin, JOM 45, 13 (1993). 3. W.L. Winterbottom, JOM 45, 20 (1993). 4. E. Gebhardt and G. Petzow, Z. Metallkde. 50, 597 (1950). 5. F.N. Rhines, Phase Diagrams in Metallurgy: Their DevelopmentandApplication, (NewYork: McGraw-Hill, 1956), p I.E. Anderson, R.S. Figliola and H. Morton, Mat. Sci. and Eng. A148, 101 (1991). 7. A.C. Chilton and K.W. Gaugler, Soldering and Surface Mount Technology Journal 10, 58 (1990). 8. Leco Corporation, Metallography Principles and Procedures, 1977, pp. 39, D.R. Frear, D. Grivas and J.W. Morris, Jr., J. Electron. Mater. 16, 181 (1987). 10. T.B. Massalski, Binary Phase Diagrams 2nd Ed., (Materials Park, OH: American Society of Metals, 1990). 11. S.L. Chen and Y.A. Chang, CALPHAD, 17, 113 (1993). 12. S L. Chen, K.C. Chou and Y.A. Chang, CALPHAD, 17, 287 (1993). 13. Ibid., p NOTES ADDED IN PROOF A recent preliminary calculation of the ternary phase diagram for Sn-Ag-Cu by Dr. Ursula Kattner indicated the occurrence of a ternary eutectic reaction at 217.4~ for a composition ofsn-3.8ag-2.3 Cu (wt.%) utilizing existing binary alloy thermodynamic and phase equilibria data without ternary interaction coefficients. Although further refinement of the phase diagram calculation is in progress, these preliminary results are in excellent agreement with the experimental results reported, herein. [Private communication on May 2, 1994 from U. Kattner (NIST, Gaithersburg, MD) to J.F. Smith.]