Pb-free Sn—Ag—Cu—Al or Sn—Cu—Al solder

Abstract

A solder alloy includes Sn, optional Ag, Cu, and Al wherein the alloy composition is controlled to provide a strong, impact-and thermal aging-resistant solder joint that has beneficial microstructural features and is substantially devoid of Ag.sub.3Sn blades.

Claims

1. A solder alloy that is solidified on an electrically conductive substrate comprising copper, said solder alloy consisting essentially of 0.95 to about 3.5 weight % Cu, from 0.15 to about 0.25 weight % Al, and balance consisting essentially of Sn, wherein the Al promotes nucleation of pro-eutectic Cu.sub.6Sn.sub.5 within the solder alloy which pro-eutectic Cu.sub.6Sn.sub.5 provides additional interface area for Sn nucleation, said solder alloy providing reduced undercooling that is less an 8 degrees C. upon solder solidification on the electrically conductive substrate as compared to a solder alloy free of Al.

2. A solder alloy that is solidified on an electrically conductive substrate comprising copper, said solder alloy consisting essentially of about 0.95 to about 3.05 weight % Cu, 0.15 to about 0.25 weight Al, and balance consisting essentially of Sn, wherein the Al promotes nucleation of pro-eutectic Cu.sub.6Sn.sub.5 within the solder alloy which nucleation, said solder alloy providing reduced undercooling that is less then 8 degrees C. upon solder solidification on the electrically conductive substrate as compared to a solder alloy free of Al.

3. A solder alloy consisting essentially of 0.95 to about 3.5 weight % Cu, from 0.20 to about 0.25 weight % Al, and balance consisting essentially of Sn, said solder alloy providing reduced undercooling upon solder solidification on an electrically conductive substrate comprising copper as compared to the solder alloy free of Al, wherein said reduced undercooling is less than 8 degrees C. upon solder solidification.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) FIG. 1 is bar graph illustrating the effect of aluminum additions (in weight %) to SAC 3595 solder alloy on undercooling values for DSC (differential scanning calorimetry) measurements at 0.17 C./second cooling rate.

(2) FIG. 2 shows a summary of DSC results for SAC3595+0.01% Al, SAC3595+0.025% Al, and SAC3595+0.05% Al solder joints (where % is weight %).

(3) FIGS. 3a and 3b are photomicrographs of the as-solidified microstructure of the SAC3595+0.01% Al solder joint and the SAC3595+0.025% Al solder joint, respectively, cooled at 0.17 C./s in selected DSC tests wherein the as-solidified microstructure comprises tin dendrites, interdendritic ternary eutectic, and pro-eutectic Cu.sub.6Sn.sub.5 particles adjacent and/or within the tin dendrites and the microstructure is devoid of Ag.sub.3Sn blades.

(4) FIG. 3c is an SEM of the typical as-solidified microstructure of the SAC3595+0.05% Al solder/Cu joint cooled at 0.17 C./s of a selected DSC test. FIG. 3d is an SEM (scanning electron micrograph) in back scattered electron mode of the microstructure of FIG. 3c. The Cu substrate is located at the top and bottom of FIGS. 3a-3d.

(5) FIG. 4a illustrates a microprobe image of the as-solidified microstructure of the SAC3595+0.05% Al solder joint and shows a metastable, intermediate Al-containing rejected solute phase region adjacent to the Cu.sub.6Sn.sub.5 interfacial layer, while FIG. 4b illustrates the profile of the Sn, Cu, Ag, and Al concentrations (wt %) across the joint interface (in m distance) of FIG. 4a, 5b, 7b (the interface between the Cu substrate and the solder is where the 100% Cu drops downwardly). In FIG. 4a, 5a, 7a, the Cu substrate is located on the left side of the microprobe image.

(6) FIG. 5a illustrates another microprobe image of the as-solidified microstructure of the SAC3595+0.05% Al solder joint, while FIG. 5b illustrates the profile of the Sn, Cu, Ag, and Al concentrations (wt %) across the solder joint of FIG. 5a.

(7) FIG. 6 illustrates nano-indentation hardness values of the Cu.sub.6Sn.sub.5 interfacial layer, Cu substrate metal, Al-containing rejected solute phase region, and the tin matrix measured within tin dendrites.

(8) FIG. 7a illustrates a microprobe image of the thermally-aged (for 1000 hr at 150 C.) microstructure of the SAC3595+0.05% Al solder joint, while FIG. 7b shows the profile of Sn, Cu, Ag, and Al content (wt %) across the thermally-aged joint interface (in m distance) of FIG. 7a (the interface being where the 100% Cu drops off).

(9) FIG. 8 is a bar graph showing undercooling values ( C.) for SAC 3595 and SAC 3595+0.05 wt % Al with multiple reflow cycles. For each alloy, five cycles were conducted wherein each cycle involved raising the temperature from 160 C. to 240 C. with a 30 second dwell followed by cooling from 240 C. to 160 C. at heating and cooling rates of 0.17 C./second.

(10) FIG. 9 is a bar graph showing Ag.sub.3Sn blades counts per 1000 m of interface for SAC+Al alloys for each solder alloy tested with different Al contents.

(11) FIG. 10 is a scanning electron micrograph (SEM) illustrating dark, dispersed phase seen on top surface of SAC+0.20Al.

(12) FIG. 11 is a SEM illustrating scratches and pullout from new IMC phase indicated by arrows.

(13) FIG. 12 is a bar graph illustrating prevalence of Cu.sub.33Al.sub.17 particles per 1000 m of interface for SAC+Al alloys that were tested.

(14) FIG. 13 is a bar graph illustrating nanohardness measurements showing hardness of Cu.sub.33Al.sub.17 phases and other solder joint solidification product phases taken in tin and in Ag.sub.3Sn blade phase regions. Sn matrix(lit.) and Ag.sub.3Sn(lit.) are published literature values.

(15) FIG. 14 is a bar graph summary of the undercooling results for the indicated alloys in one reflow cycle tests in DSC solder joints.

DESCRIPTION OF THE INVENTION

(16) The present invention involves reducing the unusually high undercooling of SAC (SnAgCu) solder joints described above, where there can be difficulty in nucleating Sn solidification as a pro-eutectic phase, especially during slow cooling, such as existing for ball grid array (BGA) joints. As mentioned above, increased undercooling of the solder joints can promote formation of undesirable pro-eutectic intermetallic phases, specifically Ag.sub.3Sn blades, that tend to coarsen radically, leading to embrittlement of as-solidified solder joints. To this end, the present invention provides a solder alloy comprising Sn, Ag, Cu, and Al having an alloy composition controlled to provide a strong, impact-and thermal aging-resistant solder joint having beneficial microstructural features described below and substantially devoid of Ag.sub.3Sn blades. The solder alloy has a relatively low liquidus temperature and a narrow liquid-solid mushy zone for solderability.

(17) In an illustrative embodiment of the invention, the solder alloy consists essentially of about 3 to about 4 weight % Ag, about 0.7 to about 1.7 weight % Cu, about 0.01 to about 0.25 weight % Al, and balance consisting essentially of Sn. The solder alloy preferably exhibits a relatively low solidus temperature of about 217 C.1 C. and narrow liquid solid mushy zone with a liquidus temperature not exceeding about 5 C., often less than 3 C., above the solidus temperature. Other alloying elements may be present in the solder alloy that do not substantially affect the melting temperature thereof.

(18) A preferred solder alloy pursuant to the invention consists essentially of about 3.4 to about 3.6 weight % Ag, about 0.8 to about 1.1 weight % Cu, about 0.03 to about 0.20 weight % Al, and balance consisting essentially of Sn.

(19) A still more preferred solder alloy consists essentially of about 3.45 to about 3.55 weight % Ag, about 0.9 to about 1.0 weight % Cu, about 0.04 to about 0.10 weight % Al, and balance consisting essentially of Sn.

(20) A still more preferred solder alloy consists essentially of about 3.45 to about 3.55 weight % Ag, about 0.75 to about 1.0 weight % Cu, about 0.04 to about 0.15 weight % Al, and balance consisting essentially of Sn.

(21) Another illustrative embodiment of the invention provides a Pb-free solder alloy consisting essentially of about 3 to about 4 weight % Ag, 0.95-y weight % Cu, and y weight % Al and balance consisting essentially of Sn wherein y is about 0.01 to about 0.25 weight %.

(22) Still another embodiment of the invention provides a still more preferred solder alloy consists essentially of about 3.45 to about 3.55 weight % Ag, about 0.80 to about 1.0 weight % Cu, about 0.10 to about 0.20 weight % Al, and balance consisting essentially of Sn, especially for BGA applications that involve thermal-mechanical fatigue environments, like avionics.

(23) The invention also envisions a modification of the alloy formulation to eliminate the Ag component for situations where higher solder melting alloys can be tolerated. Such modified solder alloy embodiments are described below.

(24) A still further illustrative embodiment of the invention provides a solder joint and solder process that embody a SnAgCuAl alloy of the type discussed above wherein the solder joint has a microstructure that comprises tin dendrites, interdendritic multiphase ternary eutectic (between the tin dendrites), and pro-eutectic Cu.sub.6Sn.sub.5 particles adjacent and/or within the tin dendrites and that is devoid of Ag.sub.3Sn blades. This microstructure is achievable at the relatively slow cooling rates employed for solder paste reflow and BGA solder processing.

(25) The as-solidified solder joint microstructure includes an interfacial layer comprising Cu.sub.6Sn.sub.5 and preferably an adjacent metastable, intermediate Al-containing rejected solute region as a zone of intermediate hardness between the hard, brittle interfacial layer and the softer tin matrix of the solder microstructure to provide a beneficial hardness gradient therebetween. The interfacial layer resides between the copper substrate and the solder of the solder joint.

(26) The solder joint is formed by the solder being solidified on an electrical wiring board and/or about copper electrical conductors in illustrative embodiments of the invention by various conventional soldering processes including, but not limited to, solder paste reflow and BGA.

(27) A thermally-aged solder joint (e.g. aged for 1000 hours at 150 C.) pursuant to the invention has an interfacial layer thickness that is about the same as the thickness as the interfacial layer thickness in the as-solidified condition (e.g. no more than 30% greater in thickness). As a result, the solder joint is resistant to thermal aging-induced embrittlement.

(28) For purposes of further illustrating the invention without limiting it, the present invention is described below with respect to modifying a near-eutectic alloy, SAC3595 solder alloy (Sn-3.5% Ag-0.95% Cu, in weight %) as a base by alloying with a fourth element, Al (aluminum) substituted for part of the Cu to reduce undercooling of solder joints. In modifying base SAC3595 solder alloy, Al was alloyed with the base solder alloy to promote nucleation of pro-eutectic Cu.sub.6Sn.sub.5 within the solder joint matrix (liquid alloy) in addition to its formation on the substrate interface, providing additional interfacial area for Sn nucleation. The Al addition also may strain the lattice of the Cu.sub.6Sn.sub.5 phase, in both pro-eutectic and interfacial layer phases, to make a more potent epitaxial nucleation catalyst for Sn, thus reducing the joint undercooling and the potential to form Ag.sub.3Sn blades, although applicants do not intend or wish to be bound by any theory in this regard.

(29) The bulk undercooling measurements for the solder joints made from the SAC3595+Al alloys that were selected (i.e. Al=0.01%, 0.025% and 0.05% by weight) are summarized in FIG. 1. That is, solder alloysSn-3.5% Ag-0.94% Cu-0.01% Al; Sn-3.5% Ag-0.925% Cu-0.025% Al, Sn-3.5% Ag-0.90% Cu-0.05% Al in weight %were tested. Each alloy was fabricated as a 100 g chill-cast ingot from component elements of 99.99% purity and drawn into solid wire of 1.7 mm diameter by the Materials Preparation Center of Ames Laboratory. For each Al concentration level tested, at least seven repeated trials were used.

(30) FIG. 1 shows that the Al addition is an active catalytic addition since these concentrations of Al have relatively lower undercooling values as compared to the average undercooling of unmodified SAC3595 base solder alloy. The range of undercooling values for unmodified SAC3595 is indicated in FIG. 1 by the left-hand bar at the zero concentration, with the data spread indicated by the bracket. Al can be seen to have a potent and consistent nucleation effect. Note that the nucleation temperature (T.sub.nuc) is defined as the onset point of the exothermic crystallization peak in each DSC thermogram, consistent with the literature in this field. Also, the bulk undercooling, T, is defined at the difference between the onset of melting at the solidus temperature (T.sub.sol) and the onset of nucleation, i.e., T=T.sub.solT.sub.nuc. Also note that observation on heating in a differential scanning calorimeter (DSC) of the solidus temperature for a eutectic alloy is also the singular eutectic melting point, T.sub.eut, and not just the start of a melting range between solidus and liquidus temperatures. The DSC apparatus used was a Pyris 1 power compensating DSC available from Perkin-Elmer wherein a pre-fluxed copper pan and copper lid accurately simulated a solder joint in the DSC test. The copper surfaces were cleaned with methanol and swabbed with flux (Johnson Mfg. No. 1 flux), and the fluxing action promoted on a hot plate at 180 C. Each pre-fluxed pan was loaded with a methonal cleaned thin disk of the selected solder alloy that weighed about 15 mg. After mild crimping, each pan sample was reflowed for one cycle in the DSC unit by heating at 10 C./min to a peak temperature of 240 C. for 30 seconds and cooling at 10 C./min (0.17 C./s) to ambient temperature to simulate BGA reflow cooling. At least seven separate (repeat) bulk undercooling (T) measurements (T.sub.eutT.sub.nuc=T) were performed for each solder alloy.

(31) Referring to FIG. 2, the SAC3595+0.01% Al alloy had a solidus temperature of 216.5 C. and a liquidus temperature of about 222 C. The SAC3595+0.025% Al alloy had a solidus temperature of 217.5 C. and a liquidus temperature of 226.5 C. The SAC3595+0.05% Al alloy had a solidus temperature of 217 C. and a liquidus temperature of 220 C. Note that the SAC3595+0.025% Al alloy appears to acquire an anomolous higher melting behavior and, as such, is a less desirable solder alloy choice in this series. However, subsequent analytical chemistry testing revealed that this particular alloy did not exhibit the desired composition and that later experiments showed a very consistent trend in liquidus temperature and undercooling with neighboring compositions.

(32) Referring to FIGS. 3a and 3b, the as-solidified microstructure of the SAC3595+0.01% Al solder joint and the SAC3595+0.025% Al solder joint, respectively, cooled at 0.17 C./s of a selected DSC test is comprised of tin dendrites, fine ternary eutectic between the tin dendrites, and pro-eutectic Cu.sub.6Sn.sub.5 particles adjacent and/or within the tin dendrites wherein the fine multiphase ternary eutectic includes a beta tin matrix with intermetallic phases, such as Cu.sub.6Sn.sub.5 and Ag.sub.3Sn, distributed in the tin matrix and wherein the microstructure is devoid of Ag.sub.3Sn blades. An interfacial layer comprising Cu.sub.6Sn.sub.5 resides between the copper substrate and the solder in the as-solidified solder joint. The Cu substrate is located at the top and bottom of the photomicrograph.

(33) Referring to FIGS. 3c and 3d, the typical as-solidified microstructure of the SAC3595+0.05% Al solder/Cu joint cooled at 0.17 C./s of a selected DSC test is comprised of the fine ternary eutectic between tin dendrites and pro-eutectic Cu.sub.6Sn.sub.5 particles adjacent and/or within the tin dendrites and is devoid of Ag.sub.3Sn blades.

(34) FIG. 4a illustrates a microprobe image of the as-solidified microstructure of the SAC3595+0.05% Al solder joint and shows an interfacial layer comprising Cu.sub.6Sn.sub.5 and an adjacent metastable, intermediate Al-containing rejected solute-rich region (SnAgCuAl solid solution phase) as a zone between the Cu.sub.6Sn.sub.5 interfacial layer and the ternary eutectic microstructure. The Cu.sub.6Sn.sub.5 interfacial layer resides between the copper substrate and the solder in the as-solidified solder joint. FIG. 4b illustrates the profile of the Sn, Cu, Ag, and Al concentrations across the solder joint of FIG. 4a showing that the rejected solute phase region contains SnCuAgAl.

(35) FIG. 5a illustrates another microprobe image of the as-solidified microstructure of the SAC3595+0.05% Al solder joint, while FIG. 5b illustrates the profile of the Sn, Cu, Ag, and Al concentrations across the solder joint of FIG. 5a. FIGS. 5a and 5b confirm that the rejected solute region is present as a zone between the Cu.sub.6Sn.sub.5 interfacial layer and the ternary eutectic microstructure. The rejected solute region is adjacent and separate from Cu.sub.6Sn.sub.5, the interfacial intermetallic layer in the as-solidified solder joint.

(36) Referring to FIG. 6, measured nano-indentation hardness values of the Cu.sub.6Sn.sub.5 interfacial layer, Cu metal substrate, rejected Al-containing solute region, and the tin matrix measured within tin dendrites are shown. It is apparent that the rejected solute region exhibits a hardness intermediate between the hardness of the hard, brittle Cu.sub.6Sn.sub.5 interfacial layer and the softer tin matrix. The rejected solute region is about 44% harder than the tin matrix. The solder joint thus exhibits a hardness gradient from the hard, brittle interfacial layer toward the softer tin matrix that improves impact resistance of the solder joint, consistent with the well known benefits of typical gradient microstructures in other alloy or composite systems. The nano-indentation measurements were made using a procedure similar to that used for microhardness measurement (see J. Mater. Res., Vol. 7, No. 6, June 1992) and used a diamond cube corner indent tip.

(37) FIG. 7a illustrates a microprobe image of the thermally aged (for 1000 hr at 15000) microstructure of the SAC3595+0.05% Al solder joint. In FIG. 7a, the Cu substrate is located on the left side of the microprobe image. The thermally-aged solder joint has an interfacial layer thickness that is about the same as the thickness as the interfacial thickness in the as-solidified condition (e.g. no more than 30% greater in thickness) so as to improve thermal aging embrittlement resistance of the solder joint. The interfacial layer of the thermally aged solder joint comprises an outer Cu.sub.3Sn layer (adjacent to the Cu substrate) and an inner Cu.sub.6Sn.sub.5 layer.

(38) FIG. 7b illustrates the profile of respective Sn, Cu, Ag, and Al concentrations across the thermally aged solder joint of FIG. 7a. The aluminum of the rejected solute phase region has become incorporated into the inner Cu.sub.6Sn.sub.5 interfacial layer

(39) The solder alloy pursuant to the invention is useful for joining electronic assemblies and electrical contacts and to substitute for Pb-containing solders in all surface mount solder assembly operations, including solder paste reflow and ball grid array joints.

(40) FIG. 8 is a bar graph showing undercooling values ( C.) for SAC 3595 and SAC 3595+0.05 wt % Al with multiple reflow cycles. For each alloy, five cycles were conducted wherein each cycle involved raising the temperature from 160 C. to 240 C. with a 30 second dwell followed by cooling from 240 C. to 160 C. at heating and cooling rates of 1 C./min and 0.17 C./second, respectively. The average undercooling value for the SAC 3595+0.05 wt % Al was 7.3 C. and generally was much smaller than the undercooling values for SAC 3595 after multiple reflow cycles.

(41) Shear strength also was measured by an asymmetric four-point bend (AFPB) method (see O. Unal, I. E. Anderson, J. L. Harringa, R. L. Terpstra, B. A. Cook, and J. C. Foley, J. Electron. Mater. 30, 1206 (2001) for larger solder joint specimens made with selected SAC 3595+Al alloys by hand-soldering with solid solder wire. The larger specimens (3 mm4 mm75 m gap) were reflowed at a peak temperature of 255 C. for 30 seconds and cooled at 1 C./s to 3 C./s to simulate typical surface mount (paste reflow) soldering processes. AFPB specimens (seven samples for each condition) were tested as-solidified and after thermal aging at 150 C. for up to 1,000 h. Microstructural analysis of the post-AFPB test specimens was performed with SEM on cross-sectioned metallographic specimens that were polished (ending with an aqueous slurry of 0.05-m SiO.sub.2) and ion milled to provide information on the failure mechanisms.

(42) Results on solder joints made from Sn-0.95 wt. % Cu (SC95) were used as a baseline. It should be noted that the large size of the Cu portion of the shear strength specimens for AFPB testing prevented them from having their undercooling measured in a calorimeter. However, the cooling rate for these specimens (quenched on a massive Cu block) is about nine times faster through the solidification temperature range, which has a tendency to promote higher undercooling but does not typically allow time for massive growth of any Ag.sub.3Sn blades that may nucleate. To cover this uncertainty in knowledge of the undercooling of these joints, the microstructure of selected specimens of each alloy was examined after testing and there was a confirmed absence of Ag.sub.3Sn blades.

(43) The most notable feature of the shear strength comparison is that the as-soldered (unaged) results for SAC3595+0.05% Al and SC95 were nearly identical at about 30 MPa and were lower than SAC3595 (about 41 MPa).

(44) Further, the thermally aged strength for SAC3595+0.05% Al was nearly constant at about 30 MPa out to 500 h of aging at 150 C. and only slightly less (29 MPa) at 1000 h. The shear strength of the of the SAC3595 and SAC3595+0.05% Al alloys seems to converge at about 30 MPa after 1,000 h of aging at 150 C. All joints show localized ductile shear failure at about 30 MPa after 1000 at 150 C.

(45) A comparison of SC95 and SAC3595+0.05% Al revealed that they start at about the same moderate shear strength, but the strength retention for SAC3595 modified by 0.05% Al is significantly better than the unalloyed SC95. The unaged and aged samples all exhibited localized ductile shear failures. Inspection of all the shear test stress-strain curves and microstructural examination of the weakest post-shear test joints (of seven repeat samples of each alloy) indicated that the Al additions effectively suppressed the nucleation and coalescence of pores that can embrittle SAC solder joints after prolonged high-temperature exposure. The relatively low initial shear strength and excellent strength retention results for the SAC 3595+0.05% Al solder appear to relate to the exceptional stability of the coarse Sn dendrites and fairly stable interdendritic ternary eutectic microstructure after thermal aging at 150 C. The SnAgCuAl solder alloy of the present invention should be useful for low temperature reflow of Pb-free solder paste and BGA balls (e.g. spheres) as well as other soldering applications. Another important advantage of the SnAgCuAl solder pursuant to the invention involves reduction or avoidance of the formation of Ag.sub.3Sn blades in the as-solidified solder joint microstructure. Analysis of all of the solder joint samples for the full range of Al additions revealed that a minimum of 0.05% by weight Al appears to completely suppress Ag.sub.3Sn blade phase formation, even at the slow cooling rate that is common for BGA assembly. This high level of control of the solder joint microstructure should produce superior results in board level impact conditions.

(46) More detailed Ag.sub.3Sn blade counting for the alloys was conducted on visible Ag.sub.3Sn blades seen protruding from either the top or bottom of each calorimetric joint interface. Blades of a length that were 50 m were recorded in FIG. 9. The total number of blades was counted per DSC joint, as well as the total interface length per DSC pan. From these values for Ag.sub.3Sn blades, the number of blades per 1000 m of interface was graphed for each alloy. Ag.sub.3Sn blades were prevalent in the baseline SAC3595 alloy, as well as for the 0.010% Al concentration. Blade suppression was seen for the alloys with additions of aluminum of 0.05 wt % or greater, but blades became visible again at the higher concentrations. Blades were seen slightly more frequently on the bottom interface than at the top interface at lower concentrations, but also had a greater deviation per sample. At aluminum concentrations of 0.15, 0.20, and 0.25, the Ag.sub.3Sn blade formation was more consistent per sample and was just as likely to occur on the top interface as the bottom. The need for an Ag.sub.3Sn blade suppressant can be seen clearly for the baseline SAC 3595 alloy where a wide variation in the number of blades can be seen.

(47) In addition, as aluminum concentration increased beyond 0.15A1, an increase in ternary eutectic phase fraction was seen, as well as the appearance of a new small equiaxed (<5 m) phase. In the SEM, EDS analysis indicated that the composition of the small equiaxed particles was slightly enriched in Cu, beyond a 2:1 ratio of Cu:Al. Comparison to the CuAl phase diagram and an extensive analysis of X-ray diffraction results revealed that the particles were probably Cu.sub.33Al.sub.17 phase. As shown in FIGS. 10 and 11, the particles can be seen either in direct contact with the Cu.sub.6Sn.sub.5 interfacial layer, attached to the copper substrate, or adjoined to pro-eutectic Cu.sub.6Sn.sub.5. The particle phase appears dark in an SEM when observed in backscattered electron mode. The size of this phase varies from 2-5 m with an average size near 3 m.

(48) WDS analysis of the particles determined the composition to be 62.2 at. % Cu-37.22 at. % Al-0.60 at. % Sn and it matches closely with the Cu.sub.33Al.sub.17phase from X-ray results and the initial EDS analysis. Further examination of the Cu.sub.33Al.sub.17 was needed, and was conducted on SAC 3595+0.20 Al, the alloy that contained the most Cu.sub.33Al.sub.17 particles of any of the given alloys (see FIG. 12). When examining Cu substrate/solder interfaces in the SEM at high magnification on backscattered electron mode, it can be seen that the new Cu.sub.33Al.sub.17 phase is faceted, hexagonal in shape and primarily on the top surface (lid side) of the calorimetric joints. In addition, the phase fraction of Cu.sub.33Al.sub.17 particles can also be seen in FIG. 12 as a function of Al content. The total number of particles were counted per DSC joint, as well as the total interface length per DSC pan. From these values for Cu.sub.33Al.sub.17 phase particles, the number of particles per 1000 m of interface was graphed for each solder alloy.

(49) Cu.sub.33Al.sub.17 particles show a different trend with composition than Ag.sub.3Sn blade formation. In other words, the phase fraction of particles increases with increasing aluminum until it reaches an apex at 0.25 wt %, then the Cu.sub.33Al.sub.17 content drops to a level comparable to 0.05 Al. Conversely, the suppression of Ag.sub.3Sn blades is only completely effective for intermediate levels of Al additions, 0.05 Al and 0.10 Al. To explain this behavior partially, one can observe that the formation of Cu.sub.33Al.sub.17 particles not only depletes the intentional Al addition, but also reduces the Cu concentration (about 2 faster, in at. %) in the molten solder alloy of the solder joint. Assuming that the formation of Cu.sub.33Al.sub.17 particles is as beneficial as the suppression of Ag.sub.3Sn blades, there appears to be a sweet spot in Al content that is centered between about 0.05 Al and 0.15 Al. However, with higher Al additions, there is increasingly less available Cu because of the substitutional alloying approach and because the Cu.sub.33Al.sub.17 particle formation depletes Cu rapidly. Therefore, less Cu.sub.33Al.sub.17 particles are formed at 0.25 Al. It should be noted that a minor extension of the sweet spot to higher Al could be realized if the Cu content was not reduced substitutionally with the Al addition, i.e., maintained at 0.95 Cu, without permitting a significant rise in the solder liquidus temperature.

(50) A significant piece of data (in FIG. 12) was that 98% of all particles seen in the SAC3595+Al were found on the top interface (lid side of the DSC pan sample). Such an observation suggests that the particles are affected by gravity and that a buoyancy effect was seen for the particles in the solder joint microstructures. This buoyancy explanation for segregation of the Cu.sub.33Al.sub.17 particles is consistent with their density, 6.45 g/cm.sup.3, which is less than the 6.99 g/cm.sup.3 for liquid Sn. Thus, this type of buoyancy driven segregation of the particles implies that the Cu.sub.33Al.sub.17 particles nucleate early in the Sn alloy liquid of the solder joint and float to the top of the joint before solidification of the Sn and other phases.

(51) As seen FIG. 11, gouges and unusual scratches were seen in micrographs of SAC3595+Al alloys. The location of scratches were often correlated with areas that exhibited pullout of the Cu.sub.33Al.sub.17 phase. With this idea in mind, nanohardness measurements were performed on the particles and the other microstructural phase constituents to *determine their relative hardness (see FIG. 13). This set of nanohardness measurements was made on a SAC3595+0.20 Al DSC joint, using the same cube corner indenter and load as the previous measurements. The hold time at maximum load was set at 10 seconds for this set of measurements. The largest effect of creep during nanoindentation is the initial penetration, so by increasing the hold time, time dependent effects like creep are lessened. These measurements of increased accuracy (shown in FIG. 13) were made in the tin matrix, on an Ag.sub.3Sn blade, of the Cu.sub.6Sn.sub.5 phase, and on the Cu.sub.33Al.sub.17 particles. Actually, two types of measurements were made for the Cu.sub.33Al.sub.17 particles; one was for particles engulfed by an Ag.sub.3Sn blade and the other type were made on particles supported apparently by the tin solder matrix. The hardness measurements made on the particles in the tin matrix were 31.45.8 GPa compared to particles in the Ag.sub.3Sn blade that were 49.12.5 GPa. The difference in apparent hardness can be attributed to the hardness of the phase surrounding the particles. The relatively soft tin matrix has more compliance and is more ductile than the harder Ag.sub.3Sn IMC. The true hardness of the particles is most likely closer to the hardness found of Cu.sub.33Al.sub.17 in Ag.sub.3Sn rather than Cu.sub.33Al.sub.17 in Sn. Because these Cu.sub.33Al.sub.17 particles do meet the definition of superhard particles, greater than 40-50 GPa, it is also useful to speculate that they might impart high wear resistance to a solder joint that is ground to expose the particles at the top of the joint or that the particles may be extracted and placed in another matrix for a cutting tool application.

(52) One worthwhile implication of the observations of hard particles that float to the top of a solder joint, doped with Al, is that certain types of thermal-mechanical fatigue (TMF) environments, especially in BGA joints, could benefit from the suppression of fatigue crack propagation by this type of microstructural feature. Hard Cu.sub.33Al.sub.17 particles of the observed size (about 3 m) could be very effective at reducing Sn grain boundary cracking, which is the normal TMF failure mechanism for Pb-free solder joints, particularly along the top of BGA joints. Thus, one of the sweet spot alloys, perhaps SAC3595+0.10 Al (or one with higher Cu and slightly higher Al) could be the optimum alloy for BGA joints that must resist high TMF conditions.

(53) Another implication of the work on these alloys is that the Ag content of the SAC3595+Al alloys does not seem to participate in the suppression of undercooling by heterogeneous nucleation or in the generation of the beneficial hard particles. Thus, the invention envisions to modify the alloy formulation to eliminate the Ag component for situations where higher melting solder alloys can be tolerated. For example, an embodiment of such modified SnCuAl solder alloy consists essentially of about 0.7 to about 3.5 weight % Cu, about 0.01 to about 0.25 weight % Al, and balance consisting essentially of Sn. A more preferred solder alloy consists essentially of about 0.8 to about 3.2 weight % Cu, about 0.03 to about 0.25 weight % Al, and balance consisting essentially of Sn. A still more preferred embodiment of this solder alloy consists essentially of about 0.95 to about 3.0 weight % Cu, about 0.15 to about 0.20 weight % Al, and balance consisting essentially of Sn.

(54) Still another such modified embodiment involves an alloy formulation to eliminate the Ag component for situations where higher solder melting alloys can be tolerated, where another illustrative embodiment of the invention provides a SnCuAl solder alloy consisting essentially of about 3.20-y weight % Cu, and y weight % Al and balance consisting essentially of Sn wherein y is about 0.15 to about 0.25 weight %.

(55) This embodiment of the invention has been tested and FIG. 14 shows the undercooling results in a comparison of the baseline SAC3595 and the baseline Sn-0.95 Cu with one Al level (0.20 Al) added to two different near-eutectic SnCu alloys, along with the undercooling for SAC3595+0.20 Al that was reported earlier. The suppression of undercooling in these SnCu solder alloys with the same Al addition is very similar to the SAC+Al alloy and this could produce the same type of microstructure control benefits in higher melting (solidus of 227 C.) solder joints.

(56) While the invention has been described in terms of specific embodiments thereof, those skilled in the art will appreciate that modifications and changes can be made thereto within the scope of the appended claims.

(57) REFERENCES, which are incorporated herein by reference: K. N. Tu, A. M. Gusak, and M. Li, J. Appl. Phys. 93, 1335 (2003). S. K. Kang, P. A. Lauro, D.-Y. Shih, D. W. Henderson, and K. J. Puttlitz, IBM J. Res. Dev. 49, 607 (2005). C. Anderson, Z. Lai, J. Liu, H. Jiang, and Y. Yu, Mater. Sci. Eng. A 394, 20 (2005). I. E. Anderson, J. Mater Sci. Mater. Electron. 18, 55 (2007). K.-W. Moon, W. J. Boettinger, U. R. Kattner, F. S. Biancaniello, and C. A. Handwerker, J. Electron. Mater. 29, 1122 (2000). D. Swenson, J. Mater. Sci. Mater. Electron. 18, 3954 (2007). I. E. Anderson, J. K. Walleser, and J. L. Harringa, JOM 59, 38 (2007). W. Liu and N.-C. Lee, JOM 59, 26 (2007). W. Liu, P. Bachorik, and N.-C. Lee, Proc. 58th Electron. Comp. Tech. Conf. (IEEE, 2008), pp. 627-635. S. K. Kang, M. G. Cho, P. Lauro, and D.-Y. Shih, J. Mater. Res. 22, 557 (2007). I. E. Anderson, J. C. Foley, B. A. Cook, J. L. Harringa, R. L. Terpstra, and O. Unal, J. Electron. Mater. 30, 1050 (2001). J. K. Walleser (M. S. thesis, Mat. Sci. & Eng. Dept., Iowa State University, 2008. I. E. Anderson and J. L. Harringa, J. Electron. Mater. 35, 113 (2006). A. Ohno and T. Motegi, J. Jpn Inst. Metals 37, 777 (1973). Materials Preparation Center, A. L., US DOE Basic Energy Sciences, Ames, Iowa, USA. Available from: www.mpc. ameslab.gov. O. Unal, I. E. Anderson, J. L. Harringa, R. L. Terpstra, B. A. Cook, and J. C. Foley, J. Electron. Mater. 30, 1206 (2001). I. E. Anderson, J. L. Harringa, and J. K. Walleser, Proc. 4th Int. Braz. Solder Conf (ASM Int./AWS, 2009), pp. 6873, ISBN: 978-0-87171-751-1. S. K. Kang, W. K. Choi, D.-Y. Shih, D. W. Henderson, T. Gosselin, A. Sarkhel, C. Goldsmith, and K. J. Puttlitz, JOM 55, 61 (2003). J. M. Song, J.-J. Lin, C.-F. Huang, and H.-Y. Chuang, Mater. Sci. Eng. A 466, 9 (2007). E. De Monlevade and W. Peng, J. Electron. Mater. 36, 783 (2007). K. S. Kim, S. H. Huh, and K. Suganuma, J. Alloys Comp. 352,226 (2003). S. K. Kang, JOM 55, 61 (2003). I. De Sousa, D. W. Henderson, L. Patry, S. K. Kang, and D-Y. Shih, Proc. 56th Electron. Comp. Tech. Conf. (IEEE, 2006), pp. 1454-1461, ISBN: 1-4244-0152-6.