Transient liquid phase bonding of Sn-Pb solder with added Cu particles

In this paper, the 1%-20% Cu particles were uniformly distributed in SnPb solder by mechanical mixing method to prepare composite solder. The solder was welded by eutectic welding and the high temperature failure test was carried out. The melting point and internal structure of solder were observed before reflow, after welding and after aging, and the existence of high-temperature resistant skeleton was found. It could be seen from the DSC curve that the solder absorption peak changes in a step-like manner, and the proportion of the high melting point absorption peak gradually increases with the aging process. The micro-structure of solder joints showed that a thin layer of IMC was formed around the Cu particles in the solder joints after welding, and the IMC in the solder joints had grown continuously until they contacted each other and formed a skeleton structure with aging. The skeleton structure improved the high temperature resistance of the solder joint. EDS analysis showed that the distribution of Cu element in the solder was similar to that of Sn, and the Pb element was enriched. The distribution of metal elements indicates that the solder joint was mainly composed of Cu6Sn5 IMC skeleton and Pb rich phase, and the solder component phase was high melting point component transformation.


Introduction
Solder was one of the important interconnect materials in microelectronic packaging, which must maintain long-term electrical and mechanical reliability during subsequent operation and service of electronic components.SnPb solders were widely used in the interconnection process of electronic components and packaging substrates due to their excellent properties [1].Pb element had a soft texture and strong ductility, which could significantly improve the electromigration and tin whisker phenomenon during long-term service [2].Therefore, SnPb solder played a vital role in the interconnection process of high-reliability electronic components.With the increase of the interconnection level of electronic components and the increase of service temperature, the problem of remelting and high temperature resistance of SnPb solder had became increasingly prominent, which had become one of the important bottlenecks in the packaging of high-reliability electronic components.
The low temperature welding process could effectively reduce a series of problems such as substrate warpage, whisker growth, large residual stress, solder joint delamination and cracking which were caused by the welding process [3].The melting point of SnPb eutectic solder was 183℃, which could achieve good welding at low temperature, but its stable service temperature was only 146℃, which could not meet the demand of electronic components for high temperature resistant solder joints.Transient liquid phase welding (TLP) could effectively improve the melting point of solder joints in the welding process to meet the requirements of low temperature connection and high temperature service.TLP had the advantages of low material cost, no pollution, good thermal matching, and compatibility with existing processes.TLP welding melted low melting point metals at a lower temperature.The liquid low melting point metal and the solid high melting point metal reacted rapidly through the solid-liquid mutual diffusion to form the intermetallic compound phase with higher melting point, which could improve the high temperature resistance of the welding joint after welding [4].This method had been gradually applied in power device interconnection.
Since TLP welding technology was proposed In the 1980s [5], TLP had been widely studied in the welding process of layered helpless flux, such as Sn-Cu [6], Sn-Ag [7], Sn-Au [8], In-Ag [9], etc., and was suitable for chip bonding process.The reason why layered TLP welding was not widely used was that it took more than 30 minutes to form a layered full-IMC solder joint.The welding time was expensive, and it was not compatible with existing welding processes.To solve this problem, some researchers had further proposed TLP solder with core-shell structure.The main method was to increase the contact area and reaction rate by electroplating micron low melting point thin layer on the outside of high melting point metal particles [10].During welding, the melted surface low melting point metal layer interconnects with the surrounding solder or pad and rapidly reacts with the high melting point metal to form intermetallic compounds.However, due to the low melting point mass fraction during the soldering process, the solder had poor wettability and soldering performance [11].
In order to avoid the problems of long welding time of layered TLP solder joints and poor welding wettability of core-shell TLP solder, the method of adding high melting point Cu metal particles to the low melting point and high reliability 63Sn37Pb solder was adopted in this paper to make it have good wetting during the welding process.With the help of the high temperature screening process of high reliable electronic components, the high temperature resistant skeleton was formed to improve the high temperature resistance of solder joints while removing the early failed electronic components.Through DSC experiments, the formation of high temperature skeleton in solder joints and the improvement of high temperature resistance performance were indirectly proved, and the formation process of high temperature skeleton was explained by the microstructure of solder joints in different states.

Mechanism analysis 2.1 Phase Diagram Analysis
The equilibrium phase diagrams of Sn-Pb, Sn-Cu and Pb-Cu showed the changing law of phase and composition of Sn-Pb-Cu system.The PB-Cu phase diagram showed that the solution solubility of Pb in Cu was less than 0.09%, and the solution solubility of Cu in Pb was less than 0.023%, and no mesoporous metal was formed [12].Therefore, in the ternary system, the Pb element was an inert element, which only played the role of adjusting the properties of the solder joint and does not react with the Sn element.The Sn-Pb and Sn-Cu equilibrium phase diagrams showed that, near the Sn-Cu eutectic point, the increase of Cu mass fraction could cause its liquidus temperature to change.Using linear approximation to fit its liquidus [13], the relationship between liquidus temperature and Cu mass could be obtained as: In the formula,   was the liquidus temperature (°C) of the system.w was the mass fraction of Cu element in the system.It could be obtained from simple calculation that the liquid phase temperature will increase by 27℃ for every 1% increase in the mass fraction of Cu element on the left side of the eutectic point [13].For SnPb eutectic solder, the reaction between Cu particles and Sn element leaded to the gradual shift of the solder matrix component to the Pb-rich phase, and the overall liquidus of the solder rises, as shown in figure 1.However, the analysis of binary and ternary systems was based on the theory of the equilibrium phase diagram, which was far from the equilibrium state in the actual process of solder melting.The solder was prepared by mechanical mixing method and the composition was not uniform enough.Considering the above two reasons, the solder matrix reacted with the Cu additive particles during the welding process.However, the solder remained in the mixture state, and the melting process changed slightly.During the welding process, the solder matrix reacted with Cu particles to form local Sn-Cu, Pb-Cu eutectic phases and Cu3Sn, Cu6Sn5 and other IMC.Due to the short welding process, the elements inside the solder joint were far from uniform distribution.Some elements were concentrated and distributed in the solder joint, and the hybrid solder joint with multi-phase coexistence was formed.At the same time, the newly generated phase had good wettability and independent high melting point with the solder matrix, so the absorption peak of the solder joint after welding shown a step-like change, as shown in figure 2. In the process of high temperature screening of electronic components, the scattered Cu3Sn and Cu6Sn5 contacted each other through spherical growth and gradually formed a three-dimensional network structure, and finally generated a high temperature resistant skeleton structure, as shown in Figure 3. Cu particles with smaller particle size reacted completely with Sn to form Cu6Sn5 intermetallic.The Cu6Sn5 spherical shell was formed at the edge of the larger Cu particles, and the Cu3Sn spherical core was formed at the center.Pb elements in SnPb eutectic were further enriched because they do not participate in the reaction, forming Pb-rich phase at the gap of the skeleton structure.Due to the existence of IMC skeleton and the formation of PB-rich phase, the melting point of solder moved towards high temperature.The high-temperature resistant skeleton forms a solid interconnection and did not melt at high temperatures (below 400℃), so the solder had a form-preserving skeleton structure at high temperatures, and the high-temperature resistance was improved.

Analysis and Design of Solder Component
In the design of the solder composition ratio, only the reaction between the Cu particles and the solder matrix was considered.The reaction between the substrate and the solder was ignored because the surface area was much smaller than that of the metal particles.In order to determine the concentration of Cu required for the transformation of Sn phase into Cu6Sn5 high temperature resistant skeleton, the mass fractions of 1%, 5%, 10% and 20% Cu were studied.Considering that the relative atomic weights of Cu, Sn, and Pb atoms are 63.547, 118.710, and 207.2, respectively.Table 1 summarized the theoretically calculated values at different mass fractions, indicating the proportion of atoms of each phase formed after welding.Only Cu6Sn5 mesoporous metal was considered in the calculation.Although Cu3Sn was the phase with better high temperature resistance, a large amount of Cu6Sn5 was formed due to the lower energy state of Cu6Sn5.Cu3Sn was only the product of further reaction between Cu6Sn5 and Cu.Therefore, in the analysis, it was believed that Sn and Cu all reacted to form Cu6Sn5, and Cu3Sn would not affect the consumption of Sn in the solder by Cu.It could be seen from the calculation that the proportion of high temperature resistant phase generated by the solder joint after the addition of 1% and 5%Cu particles was less than 20%.It was difficult to form a stable skeleton structure in the solder joint.Only 10% and 20% Cu particles could achieve more than 30% mesoporous metal formation, forming a high temperature resistant skeleton..00After the Cu particles were added, the changes of the solder structure and the melting point of the solder at different stages were shown in Figure 4. Before soldering, the solder was mixed with Cu, SnPb and flux.The Cu particles did not react with SnPb and remain relatively independent.The melting point of the solder is 183°C.After soldering, the scattered Cu particles reacted with the solder.The small particles completely formed Cu6Sn5 IMC, and the large metal particles formed a Cu-Cu3Sn-Cu6Sn5 three-layer core-shell structure from the inside out.Therefore, there were absorption peaks corresponding to Cu6Sn5 (415°C) and Cu3Sn (676°C) in the DSC curve.Cu atoms diffused into the solder and form a small amount of SnCu eutectic and PbCu eutectic with SnPb solder.

Experimental design 3.1 Experimental design and experimental preparation
In order to maintain a larger contact surface area and increase the reaction rate between the solder and the Cu particles, No. 4 SnPb solder was selected for the preparation of the solder.The size of Cu particles was selected according to the reaction rate formula between solder and metal particles [16].
In the formula, Z was the thickness of the IMC layer, k0 was the initial reaction rate, t was the time, Q was the reaction activation energy, R was the gas molar constant, and T was the reaction temperature.The welding time of 5minute and welding temperature of 230℃ were submitted into the above formula, and the reaction thickness between Sn element and Cu element was about 1 micron.Although smaller metal particles helped the reaction to proceed fully, too small particle size would cause the metal particles to be removed along with the flux during the flux removal stage, resulting in poor wettability between the metal particles and the solder matrix.Based on the above factors, sub-micron metal particles were selected.The particle size statistics of Cu particles and SnPb particles were shown in figure .

Experimental results
The melting point of the solder at different stages represented the composition of the components present and the high temperature resistance.The melting points of different solders at different stages were shown in Figure 7.It was shown in Figure 7-a that the melting starting point of the solder before reflow, after reflow and after aging was in the range of 183°C-185°C, and the absorption peak was in the range of 189-191°C.The melting point basically did not occur within the experimental error range.The effect of Figure 7-b was basically the same as that of Figure 7a.Therefore, when the mass fraction of Cu was less than 5%, although the Cu particles reacted with the solder matrix, too few high-temperature phases were generated, which had no significant effect on the overall melting point of the solder.
When the mass fraction of Cu particles reached 10%, the melting point of the solder was consistent with that of the 63Sn37Pb solder before reflow.The solder started melting at 183℃, and the melting absorption peak still existed near 220℃.This phenomenon showed that the composition of the 63Sn37Pb matrix changed, forming part of the Sn-Pb-Cu eutectic structure, and the liquidus of the solder gradually increased.Since Pb was an inert element in the matrix, the solder liquidus lift came from the formation of Sn-Cu eutectic, and it was considered that the upper limit of the solder liquidus lift was the same as the Sn-Cu eutectic melting point of 227°C.In the DSC curve of Figure 7-c, another absorption peak at 330 °C appeared in the solder after reflow, which indicated that other high-temperature resistant phases were significantly generated in the solder.The ratio of heat absorption and weight of sample in unit time was measured by DSC tester, and the enthalpy value of sample was calculated.In the DSC curve after 250h aging, the absorption peak enthalpy of solder at 183℃ decreased from 31.47J/g to 29.43J/g, and the absorption peak enthalpy at 320℃ increased from 2.20J/g to 4.57J/g.This phenomenon showed that with the progress of the aging process, the proportion of the high temperature resistant phase in the whole solder gradually increased, and the high temperature resistant skeleton was formed in the solder matrix.

Experimental results and analysis
The solder melting process was an endothermic process that occurs when the solder structure changes from the solid phase to the liquid phase, which was determined by the properties of the solid phase itself.Therefore, the evolution law of the micro-structure of the solder could verify and explain the changing law of the melting point of the solder, as shown in Figure 9. Figure 9 showed that the solder reacted with the Sn element in the solder after reflow, such as Cu6Sn5 and Cu3Sn.The Cu element was distributed in a point-like aggregation type and remained relatively independent, and was not uniformly dispersed in the solder to form a eutectic structure.After reflow, the dispersed metal particles with the size less than 1 micron fully reacted with Sn element to form Cu6Sn5 intermetallic metal during the reflow process, which was consistent with the theoretically calculated reaction rate.The concentrated small metal particles and large metal particles could not be completely converted into Cu6Sn5 high temperatureresistant intermetallic due to the lack of surrounding Sn elements and the insufficient reaction time.The Cu core remained in the central part, forming a Cu3Sn inter-layer between the Cu core and Cu6Sn5 metal layer.It was shown in Figure 9a, b, c, d that as the content of Cu particles increased gradually, the proportion of intermediary metals in the solder increased, and the intermediary metals gradually contact from scattered points to form interconnections, forming a network structure.The solder matrix composition deviated more from the 63Sn37Pb eutectic with increasing Cu content.For 1% and 5% mass fraction Cu particles were added, the matrix composition only changed in a small range around the particles, and the overall eutectic state remains.When the mass fraction of Cu reached 10% or more, the overall composition of the solder changed significantly due to the reaction of a large amount of Sn element with Cu element, and an obvious Pb-rich phase was formed between the high temperature resistant skeleton.In addition to the improvement of high temperature resistance, the Pb-rich phase had a soft texture that could absorb stress during service.
The internal structure of solder joints of different components after aging at 200℃ for 250 hours was shown in Figure 9e,f,g,h.After screening, the residual Cu3Sn and Cu particles in the solder had been completely converted into Cu6Sn5, and their size had grown from the original submicron scale to more than 10 microns.Because the concentration of Cu6Sn5 was too small in the solder with Cu particles added at 1%, even after screening, the intermetallic metal was still scattered in the solder matrix.In the solder with 5% Cu particles added, due to the growth of the intermetallic metal, Cu6Sn5 began to gradually contact, and also due to the insufficient content, it only formed a local interconnection and did not form an overall skeleton structure.When the added mass fraction of Cu particles reached 10% or more, due to the growth of IMC and the increased probability of contact with each other, the overall relatively loose high temperature resistant skeleton began to form, and the enrichment of Pb elements began to appear.When the added mass fraction of Cu reached 20%, the high temperature resistant skeleton in the solder joint was further dense, and the enrichment distribution of Pb element between the skeletons was obviously manifested.Based on the above analysis, it was necessary to add Cu particles to form a high temperature resistant skeleton with an added content of 10% or more, which was basically consistent with the theoretical calculation.
Since the welding reaction process and the distribution of elements after welding were far away from the equilibrium state, an aggregated skeleton structure deviating from the equilibrium state was formed inside the solder joint.The DSC curve in the previous section showed that the absorption peak of the tissue was composed of an absorption peak of about 183℃ and an absorption peak of about 330℃.It could be seen from Figure 9-h that After 250h high temperature screening, Cu element and Sn element had fully reacted, and a small amount of SnPb eutectic phase still existed in the solder.Even though the eutectic phase deviated slightly from the eutectic component due to atomic diffusion, the solid phase line was still located at 183℃.Therefore, the absorption peak strength of Cu particles added solder varies at 183℃ under different conditions, but it always existed.The second absorption peak of solder was located at around 330℃, which was close to the melting point of Pb-Cu eutectic and Pb.Since Cu and Pb did not form IMC, the other absorption peak of the solder at around 330 ° C might be due to the formation of Pb-Cu eutectic.In the theoretical calculation, the enrichment of Pb element could reach 10% of the total element in the ideal state, so the enrichment of Pb element might be one of the reasons for the absorption peak.Although a large amount of Cu6Sn5 mesoporous metal was formed in the solder, there was no obvious absorption peak at 415℃.It could be seen from SEM that there was no new phase in the solder except Cu6Sn5, Cu, Sn and Pb, so the absorption peak at 330℃ was the absorption peak of the mixed high-temperature phase such as Pb-rich phase, Pb-Cu eutectic and Cu6Sn5 IMC.In addition, the porosity structure in Figure 9

Conclusion
In this paper, SnPb solders with different Cu mass fractions were used, and the good wettability of transient liquid phase soldering technology was used to achieve good soldering of solder joints.And with the help of the high temperature screening process of the components, the high temperature resistant skeleton was fully formed inside the solder joint to achieve good high temperature resistance.Its main conclusions were as follows: 1.The addition of Cu particles with a mass fraction of more than 10% could generate an overall high temperature resistant skeleton, and the skeleton became denser with the addition of Cu.
2. After reflow, the inside of the solder joint was SnPb eutectic and a mixture of Cu core and Cu6Sn5 shell.After aging screening, all Cu was converted into Cu6Sn5 and there was obvious enrichment of Pb element.
3. There were two absorption peaks around 183°C and around 330°C when the solder was melted after aging and screening, which were the melting points of SnPb eutectic phase and high temperature mixed phase respectively.

Figure 2 .
Figure 2. Solder melting point change diagram.In the process of high temperature screening of electronic components, the scattered Cu3Sn and Cu6Sn5 contacted each other through spherical growth and gradually formed a three-dimensional network structure, and finally generated a high temperature resistant skeleton structure, as shown in Figure3.Cu particles with smaller particle size reacted completely with Sn to form Cu6Sn5 intermetallic.The Cu6Sn5 spherical shell was formed at the edge of the larger Cu particles, and the Cu3Sn spherical core was formed at the center.Pb elements in SnPb eutectic were further enriched because they do not participate in the reaction, forming Pb-rich phase at the gap of the skeleton structure.Due to the existence of IMC skeleton and the formation of PB-rich phase, the melting point of solder moved towards high temperature.The high-temperature resistant skeleton forms a solid interconnection and did not melt at high temperatures (below 400℃), so the solder had a form-preserving skeleton structure at high temperatures, and the high-temperature resistance was improved.

Figure 4 .
Figure 4. DSC theory analysis and solder joint structure change diagram.
graph of SnPb particle size

Figure 5 .
Figure 5. Material particle size statistical analysis diagram.The SEM picture of the solder matrix after adding Cu particles with different mass fractions was shown in the figure6.The Cu particles were uniformly distributed in the solder and concentrated on the surface of the solder ball.With the increase of the added mass fraction, the density of Cu particles increased gradually.

Figure 9 .Figure 10 .
Figure 9. SEM images after reflow and after screening a)1% Cu after reflow, b)5% Cu after reflow, c)10% Cu after reflow, d)20% Cu after reflow, e)1% Cu after aging, f)5% Cu after aging, g)10% Cu after aging, h)20% Cu after aging.Solder element sketch scan could further reflect the distribution and enrichment of elements in solder.Figure10showed the distribution of elements inside the solder joint after adding 20% Cu particles after 250h screening.It could be seen from the figure10that the Cu element forms an obvious bulk skeleton structure, and the Pb element was distributed in the skeleton structure to form a complementary structure.The distribution of Sn element was relatively uniform, but it was enriched in the area of Cu element, indicating that Sn element has realized the diffusion into Cu particles and reacted with Cu particles.

Table 1 .
Mathematical calculation of sufficient amount of Cu particle additions to convert the whole Sn phase to Sn-Cu IMCS