Low Temperature Soldering Using SnBi Alloys

Low Temperature Soldering Using SnBi Alloys

Morgana Ribas, Ph.D., Anil Kumar, Divya Kosuri, Raghu R. Rangaraju, Pritha Choudhury, Ph.D.,

Suresh Telu, Ph.D., Siuli Sarkar, Ph.D.

Alpha Assembly Solutions, Alpha Assembly Solutions India R&D Centre

Bangalore, KA, India

[email protected]

ABSTRACT

Low temperature solder alloys are preferred for the assembly of temperature-sensitive components and substrates. The alloys in this category are required to reflow between 170 and 200o C soldering temperatures. Lower soldering temperatures result in lower thermal stresses and defects, such as warping during assembly, and permit use of lower cost substrates. Sn-Bi alloys have lower melting temperatures, but some of its performance drawbacks can be seen as deterrent for its use in electronics devices. Here we show that non-eutectic Sn-Bi alloys can be used to improve these properties and further align them with the electronics industry specific needs. The physical properties and drop shock performance of various alloys are evaluated, and their results are analysed in terms of the alloy composition, including Bi content and alloying additions.

Key words: lead-free, Sn-Bi solders, low temperature soldering, drop shock.

INTRODUCTION
The combined technical and economic benefits of low temperature soldering are expected to grow the use of low temperature solder pastes from <1% to 20% in the next ten years, according to the 2015 iNEMI roadmap [1]. Despite such opportunity for growth, there are limited low temperature alloy options that fulfil SMT soldering requirements. Sn-In and Sn-Bi systems would be obvious options due to their lower melting temperatures, but Sn-Bi lower cost and higher availability puts it at a competitive advantage. Since starting the transition to Pb-free solder, the electronics industry has considered Sn-58Bi as a possible replacement for eutectic Sn-Pb. However, due to its lower ductility and poor thermo-mechanical properties, Sn-58Bi was put aside and used only in limited applications that did not require much of thermal cycling or drop shock performance.

More recently, multi-step assembly processes, a surge in use of temperature sensitive components and industry desire of using cheaper substrates brought in additional challenges and low temperature solders came back under consideration. Furthermore, the miniaturization of portable electronics has also imposed requirements of increasingly thinner flip chip ball grid arrays (FCBGA). Recent investigations have showed that assembling these ultra-thin packages with standard lead-free solders, such as Sn-3Ag-0.5Cu (221oC liquidus temperature), results in dynamic warpage on package substrate and PCB [2-4]. This represents a severe limitation on using the latest generation of ultra-thin microprocessors. Use of low temperature solders can significantly reduce such warpage, but available Sn-Bi solders do not match Sn-Ag-Cu drop shock performance [5-6]. Besides these pressing technical requirements, finding a low temperature solder alloy that can replace alloys such as Sn-3Ag-0.5Cu solder can result in considerable hard dollar savings from reduced energy cost and noteworthy reduction in carbon emissions [7].

In previous works [8-11] we have showed how the use of micro-additives in eutectic Sn-Bi alloys results in significant improvement of its thermo-mechanical properties. Here we show how a non-eutectic Sn-Bi solder with 2 wt.% additives (referred here as X46) enables peak reflow temperatures as low as 190oC of mixed low temperature/Sn-Ag-Cu solder joints, resulting in significant fewer defects during reflow processing of electronic assemblies.

EXPERIMENTAL DESCRIPTION

Physical Properties
Differential Scanning Calorimetry (DSC) technique was used for assessing the melting characteristics of solder alloys. The experiments were performed using a Differential Scanning Calorimeter from TA Instruments. Samples with 10 mg in mass were heated under nitrogen atmosphere at 10 K/min heating rate. A Running Integral Analysis was performed on all DSC plots to determine the solid/liquid fraction at given temperatures.

Mechanical properties of the alloys were evaluated by tensile tests and elastic modulus measurements. The tensile tests were conducted in a universal testing machine such as the one showed in Figure 1(a). Ultimate tensile strength (UTS) and yield strength (YS), corresponding respectively to the maximum stress achieved during the test and the transition between elastic and plastic deformation, were evaluated as illustrated in Figure 1(b). Each alloy was machined into rounded specimens as per ASTM E8 tensile test standard, with 4 mm diameter and 16 mm gauge length, as showed in Figure 1(c). Stress-strain curves for at least five specimens of each alloy were recorded at room temperature, using a constant strain rate of 10-3mm/s. Average values of ultimate strength, yield strength and elongation are reported.

Elastic properties were measured using the ultrasonic pulse- echo technique. We use an ultrasonic thickness gage, and longitudinal and shear wave transducers to measure the sound velocities through the alloys, which are then used to compute their Young’s (i.e., elastic modulus) modulus. Through this technique the elastic modulus can be evaluated without the variability often observed when performing tensile tests. The elastic modulus is often used as an insight on the shock and vibration performance of alloys as it is a measure of the stiffness of the material.

Figure 1. (a) Instron universal testing machine with environmental chamber, (b) Stress-strain curve schematics and (c) Specimen geometry used for tensile testing.

Drop Shock

The ability of solder joints to withstand shock and vibration was evaluated using a board level drop shock test. We use the guidelines described in the JESD22-B111 (July 2003) standard and perform the test following service condition B (1500Gs, 0.5 msec duration and half-sine pulse) for setting up the shock intensity applied during the test. We use a Lansmont shock machine and data acquisition system for test vehicle evaluation and for monitoring the shock pulse curve, Figure 2. The data acquisition system enables detection of specific event conditions for each channel that can be used, for example, for configuring minimum event duration and corresponding voltage levels to detect intermittent electrical and permanent faults. An external power source (5V DC) is applied across the circuit and the generated acceleration is used as a trigger for the measurement of drop in potential gradient across the circuit. We use a test vehicle, showed in Figure 2(b), with 15 CTBGA84 that were assembled and tested till failure. Each failure is accounted once there is a drop of 1V or more inthe applied potential, for a duration of 0.5 msec or more. The characteristic life calculated from Weibull plots is given as the number of drops at 63.2% of cumulative failures.

Figure 2. (a) Lansmont drop shock testing machine, (b) Alent drop shock test vehicle and (c) half-sine shock pulse curve corresponding to JEDEC service condition B.

RESULTS AND DISCUSSION

As per their binary phase diagram [12], Sn and Bi form a eutectic composition around 138°C, with about 57 wt.% Bi and 43 wt.%Sn. As Bi content decreases, the liquidus temperature moves away from this eutectic composition, as showed in Table 1. Despite the fact that actual solder alloys have minor impurities, the solidus and liquidus temperatures showed agree well with the phase diagram. When used for SMT applications, the corresponding solder paste is generally reflowed at peak temperatures that are 25-30°C above the solder melting point. Thus, a Sn-Bi alloy with 35 wt.% Bi would need a reflow profile reaching around211°C, whereas decreasing the Bi content to 40 wt.% would require a peak reflow temperature around 200°C. Such temperatures would be higher than the desirable peak reflow temperature that leads to reduction in warpage and significant cost savings, as discussed earlier.

In the case the Sn-Bi solder is used to assemble an area array package with a higher melting point solder, it is also important to evaluate its transformation into liquid phase. Later we will describe the inter-diffusion between Sn-3Ag- 0.5Cu and Sn-Bi, in which these two alloys are joined at temperatures below the Sn-3Ag-0.5Cu melting point. Having the lower melting point alloy in liquid state facilitates the inter-diffusion between these two alloys for forming a solder joint. Table 1 also shows the liquid fractionof Sn-Bi alloys with varying Bi content. As per DSC analysis, when reducing the Bi content to 55 wt.%, the alloy is completely transformed into a liquid phase by 144°C Further reduction of Bi decreases the fraction of solder that is converted into liquid. For example, reducing Bi content from 50 to 45 wt.% results in a significant decrease in the liquid fraction, from 96% to only 78%.

Since early discussions on the possibility of adopting Sn-Bi alloy as a Pb-free solder, it was clear that their poor mechanical properties, notably excessive brittleness and lower elongation, would be a deterrent. Figure 3 shows the tensile properties of Sn-Bi alloys with decreasing Bi content. Reducing the Bi content in these alloys does not decrease much the ultimate tensile strength (UTS), but it does improve their elongation, as it was showed earlier [9,13]. Despite such encouraging results, reducing Bi content alone is not sufficient for improving its mechanical reliability in solder joints. For example, Figure 4 shows the effect of room temperature ageing on the Sn-45Bi microstructure. The Bi microstructure (white areas showed in Figure 4) visibly coarsened after only 24 hrs ageing. Hardness measurements confirmed this and showed that Sn- Bi tend to harden upon ageing, in this case increasing from 6.6 to 8.5 Hv2.5 after 24 hrs at room temperature. In addition to that, reducing Bi content alone does not improve drop shock performance to the same level of Sn-3Ag-0.5Cu, as showed in Figure 5. Reducing Bi content from 58 to 45 wt.% increases the drop shock characteristic life in 42%. Further reduction of Bi content to 40 wt.% results in another 77% increase in its drop shock characteristic life. However, despite such improvements in performance by reducing Bi content alone, the drop shock characteristic life of Sn-3Ag- 0.5Cu is still 60% higher than of Sn-40Bi, for example.

Another option for improving Sn-Bi alloys mechanical properties is the use of alloying additions. As we showed earlier, alloying additions can be used to improve eutectic Sn-Bi performance such as mechanical properties, drop shock and thermal cycling [8-11]. Table 2 shows the effect of additives such as Ag and In on the melting behaviour of Sn-Bi alloys. Ag addition of up to 1 wt.% does not have a significant change in solidus and liquidus temperature of the alloy. Addition of 1 wt.% In slightly reduces the solidus and liquidus temperature of Sn-58Bi. However, at 3 wt.% In the solidus temperature is reduced to 125°C and the DSC curve showed a peak around 100°C, which indicates presence of lower melting temperature compounds. Other ductile solder alloys can also have appropriate melting behaviour despite having multiple alloying additions. For example, alloy X46 has a solidus and liquidus temperatures of 138°C and 151°C, respectively, and 99% conversion to liquid phase at 144°C (Figure 6).

Alloy

Temperature,°C

Liquid Fraction, %
  Solidus Liquidus 139°C 140°C 142°C 144°C
Sn-55Bi 138 144 10 39 96 100
Sn-50Bi 138 155 6 26 90 96
Sn-45Bi 138 168 9 50 75 78
Sn-40Bi 138 178 16 66 75 77
Sn-35Bi 138 186 12 61 70 77

Figure 3. Tensile properties of Sn-Bi alloys. ©2013 IEEE. Reprinted, with permission, from Ref. [9].

Figure 4. : Effect of aging on the Sn-45Bi microstructure.

Figure 5. Drop shock results of Sn-Bi alloys.Drop shock results of Sn-Bi alloys.

Table 2. Effect of additives on Sn-Bi solder alloys melting temperature.

Alloy Solidus Temperature,°C Liquidus Temperature, °c
Sn-57.6Bi-0.4Ag 137 142
Sn-58Bi-1Ag 137 142
Sn-38Bi-1Ag 137 142
Sn-58Bi-1Ag-1In 133 137
Sn-58Bi-1Ag-3In 125 133*

*Peak at 100°C.

 

Figure 6. Liquid fractions of X46 alloy computed from DSC curve.

Void Reduction Strategy in Soldering
Void Reduction Strategy in Soldering