intermatalilics of solder paste

Effect of the Intermetallics Particles Size on the Brittle to Ductile FractureTransition in a Bulk Sn-4wt%Ag-0.5wt% Cu SolderRatchev P., Vandevelde B., IMEC, Leuven, BelgiumVerlinden B., Katholieke Universiteit Leuven, Leuven, BelgiumAbstractThe fracture toughness of a bulk Sn-4wt.%Ag-0.5wt.%Cu lead-free solder was measured as function of the temperature by means of a pendulum impact (Charpy) test in as-cast and several annealed conditions. The casting was performed in a small mould in order to produce an initial microstructure consisting of fine intermetallic particles. Samples were annealed at different conditions in order to grow the SnAg3 particles to different average particles sizes. A ductile to brittle fracture transition was found at -28°C in as-cast condition, shifting downwards to -48°C upon annealing. The microstructure study shows that coarsening of the initial fine intermetallic particles during annealing is causing the shift of the transformation temperature downwards. Therefore a careful microstructure control is needed during the solder solidification after reflow in order to decrease the low temperature brittleness hazard. This is a significant reliability issue, which should be considered during the Sn-Ag-Cu alloys application.IntroductionBy July 2006 the use of lead in electronics in Europe will be forbidden and lead-free solders with at least comparable reliability should replace the Sn-Pb solder. Many Sn-based solders are under intensive study in the last years, but the mainstream are without any doubts the Sn-Ag-Cu alloys. Since the “soft” Pb is taken out of the solder, the Pb-free solders are stiffer, which can be translated in higher fracture toughness only if they have also a high ductility. It is well known that some metals loose their ductility at low temperature and show a brittle fracture mode. The ductile to brittle transition temperature is an important parameter for them. In a recent publication1, we have reported that bulk Sn, Sn-Cu, Sn-Ag and Sn-Ag-Cu lead-free solders show a sharp ductile to brittle transition, compared to no transition for the eutectic Sn-Pb. The transition temperature of Sn-0.5%Cu and Sn-0.5%Cu(Ni) alloys is -125°C, which seems to be an inherited property from the high purity Sn. The Ag-containing solders show a transition at higher temperatures: in the range of –78°to–45°C. The increase of the Ag content shifts the transformation temperature towards higher values. This was related to the higher volume fraction of SnAg3 particles in the solder volume1, which can serve as obstacles for the dislocation motion, causing dislocation pile-ups, voiding and crack nucleation2. We suggested that the failure mode in real solder joints could change around and below the brittle fracture temperature, i.e. the ductile “solder fatigue” failure mode to change to a catastrophic brittle fracture into the solder volume. The purpose of the present paper is a more systematic study of the effect of the intermetallic particles size on the brittle to ductile fracture transition of a near-eutectic Sn-4wt.%Ag-0.5%Cu solder. ExperimentalA near-eutectic Sn-4wt.%Ag-0.5%Cu solder was cast in a small mould in order to produce an initial microstructure consisting of fine intermetallic particles in the eutectic regions. Samples were annealed at different conditions in order to grow the SnAg3 particles to different average particle sizes. Since they come from the same casting, their volume fraction will be approximately the same. Bulk samples of 4 annealing conditions were tested, namely as–cast, 100h at 150°C, 1000h at 150°C, 1000h at 175°C. A pendulum impact test, well known as “Charpy test”, was performed in order to determine the amount of energy dissipated during fracture, which is a measure for the fracture toughness of the material, as function of the temperature. The tests were performed according to ASTM E23 standard with V-notch samples of size 10x10x55mm. The temperature of the sample was varied in the range of -195°C to 100°C, by means of an off-line heating/cooling system. More details about the test conditions can be found elsewhere1. The microstructure inspection was performed on cross-sections of annealed and fractured samples by means of optical microscopy and SEM in back-scattered electrons imaging mode (BEI).Results and DiscussionThe results of the impact test are given in Figure 1. The sharp ductile to brittle transition reported in 1 for the lead-free solders was completely confirmed.Figure 1 – Fracture toughness vs. test temperature as function of the annealing condition for the studied Sn-4wt.%Ag-0.5%Cu solder.Figure 2 – Fitting of the transition regions of the graph in Fig.1 in order to obtain the transitiontemperaturesA considerable effect of annealing on the ductile to brittle transition temperature was observed. In order to extract the transition temperatures, the transition region of the graphs was fitted as shown in Figure 2. The summary in Table 1 shows a change from -28°C for the as-cast condition to -48°C for the sample annealed for 1000h at 175°C. The significant change is between as-cast condition, from one side, and the annealed conditions, from other side. The difference between the annealed conditions is smaller. Although the considerable experimental scatter, a trend of decreasing the transition temperature with the annealing strength (higher time and/or temperature) was observed. Additionally, the fracture toughness in the ductile region increased with the annealing temperature/time at a fixed test temperature, see for instance the points at 100°C, see Figure 1. This is probably a softening effect related to the intermetallic particles coarsening during annealing (the later will be shown further in this paper). Since the fracture toughness is a product of the strength and the ductility of the material, the resulting gain in ductility should be larger than the loss of strength.Table 1 - Transition temperatures vs. annealingconditionsAnnealing Transitiontemperature,o C As-cast -28 ± 6100h at 150o C -42 ± 51000h at 150o C -40 ± 51000h at 175o C -48 ± 5The microstructure of the studied samples before testing is shown in Figures 3-6. It consists of areas of pure Sn and eutectic regions, where intermetalic particles (mainly SnAg3) are spread in a Sn matrix. As expected, the as-cast sample shows a fine SnAg3 intermetalic particle size, see Figure 3, which evolves during annealing towards coarser particles. It is not easy to compare quantitatively the particle size due to their complicated shape, which changes from lamellae-type, in as cast condition, to more elongated plate-like type, when annealed at 150°C, Figures 4 and 5, and to a more rounded type after annealing at 175°C, Figure 6. In order to correlate the SnAg3 intermetalic particles size with the transition temperature the average cross-sectional surface of the particles was measured for each annealing andconverted to an effective diameter, assuming a spherical particle shape. An average of 50 particles was measured per sample from several pictures by means of image analysis software. The results are shown in Figure 7.Figure 3 – Microstructure in as-cast condition.Figure 4 – Microstructure after 100h at 150o C.A clear trend is visible of shifting the ductile to brittle fracture transition temperature with the increase of the intermetallics particle size. On the other hand our previously reported results, see1, show a transition temperature of -58°C for an alloy with the same composition, but coming from another supplier in as-cast condition. The casting mould was much bigger than the one used in the present work, therefore the associated microstructure was much larger, as shown in Figure 8. The measured effective particle diameter for it was 1.7 µm and its position in the graph in Figure 7 is given as a purple star. It deviates considerably from the observed trend. Of course, this comparison is only possible if the particles volume fraction is the same in both cases. A slight difference in the composition (mainly the Ag content) can shift considerably the transition temperature as already shown elswere1. Since these 2 alloys come from different suppliers, some small composition differences are certainly possible. It has to be concluded that both composition and thermal history (cooling rate after casting, annealing etc.) are very important for the transition temperature. A problem of the present measurements is a considerable scatter of the data around the transition temperature of the order of ±5-6°C. This scatter is mostly related to the off-line system of temperature control, which can be a source of significant error. Additionally, the ductile to brittle transition as a physical phenomenon is also very often considered a source of considerable scatter, see 2.Figure 5 - Microstructure after 1000h at 150o C Figure 6 - Microstructure after 1000h at 175o CFigure 7 – Transition temperature as function of the effective particle diameter.Figure 8 - Microstructure of the 4wt.%Ag-0.5%Cu solder used in [1] coming from a largecasting (lower cooling rate).In order to go deeper in the fracture mechanism, cross sections were prepared of samples tested at the transition temperature and studied by means of SEM. The area just ahead of the cracked area in such a sample of the as-cast material is shown in Figure 9. The crack is clearly propagating by means of linking the fine voids, formed at the intermetallic particles. This voiding is shown in details at higher magnification in Figures 10 and 11 from the same area. The fact of the void formation at the particles interface is crucial for the understanding of their rolein the fracture process. Actually, this confirms that they serve as obstacles for the dislocation motion during plastic deformation. Consequently, dislocations are piling-up at their interface and form voids, as observed in Figures 9-11. These voids serveas crack nucleation sites and assist the crack propagation. Therefore, they are able to enhance the transition towards a brittle behavior of the material. As a consequence of that, the larger the particles number the higher will be the transition temperature.At a fixed volume fraction, as at the present experiment, the larger particles number will be related to a smaller particles size. Therefore a low cooling rate after reflow will be beneficial in order to produce a coarser intermetallic particles microstructure and to decrease the probability of brittle fracture in the solder to take place during the lifetime of the solder joints. Additionally, any annealing will be also beneficial for this, therefore the risk for the brittle fracture will tend to decrease with the lifetime of the solder joints.It is very important to realize that the low temperature brittleness of the Sn-Ag-Cu solder joints is a significant reliability issue. The present experiment showed that, using a high cooling rate, a microstructure can be produced that shows an rather high transition temperature of -28°C instead of the temperature of -58°C, that we reported previously for the commercial solder with the same composition but cast with a much lower cooling rate. This temperature was successfully shifted by means of annealing to -48°C, due to the microstructure modification. A ductile to brittle transition temperature of -28°C is completely unacceptable for a mainstream alloy as the near-eutectic Sn-4wt.%Ag-0.5%Cu. It enters the range of many applications and environmental testing conditions as thermal cycling and thermal shock, going to temperatures as low as –40 a or –60°C. Therefore, the fracture mode can change in that temperature range, i.e. the ductile “solder fatigue” failure mode to change to a catastrophic brittle fracture into the solder volume. In order to avoid that a careful microstructure control is needed during the solder solidification after reflow.Figure 9 – Microstructure of the as-cast sample fractured at transition temperature.Figure 10 – Enlargement of an area from Figure 9.Notice the voids at the IMC particles.Figure 11 – Enlargement of another area from Figure 9. Notice the voids at the IMC particles. ConclusionsA ductile to brittle fracture transition was found around -28°C for a bulk near-eutectic Sn-4wt.%Ag-0.5%Cu solder cast with a high cooling rate, compared to –58°C reported previously for a commercial solder with the same composition but cast with a much lower cooling rate. Our microstructural study present evidences that a very fine intermetallic particles size is responsible for the shift of the transformation temperature towards higher values. The SnAg3 particles serve as obstacles for the dislocation motion, causing dislocation pile-ups, voiding and crack nucleation. This temperature was successfully shifted by means of annealing to -48°C, due to the precipitates coarsening. Therefore a careful microstructure control is needed during the solder solidification after reflow in order to decrease the low temperature brittleness risk. Still this significant reliability issue that should be considered during the Sn-Ag-Cu alloys application. AcknowledgementsThe authors would like to acknowledge the financial support from the Flemish community in Belgium through the ALSHIRA project (IWT040373). Daniel Verkhoven and Bart Van De Lisdonk from Interflux Electronics, Belgium are acknowledged for performing the dedicated alloy casting. The author are grateful also to Mr. Frederic Duflos for his help with the Charpy tests. References:1.P. Ratchev et al, A Study of Brittle toDuctile Fracture Transition Temperatures inBulk Pb-Free Solders, Proceedings of EMPC2005 (IMAPS-Europe), June 12-15, 2005,Brugge, Belgium, pp.248-252.2.R. Reed-Hill, Physical MetallurgyPrinciples, Published by Van NostrandReinold Company, New York, USA, pp.541-547, 1970。