Steffen Wiese

Constitutive behaviour of lead-free solders vs. lead-containing solders-experiments on bulk specimens and flip-chip joints

Both high Pb-Sn and eutectic 63Sn-37Pb have been the suitable materials for the interconnection of electronic components because of their low cost and appropriate physical properties. Due to environmental awareness, and the health hazards caused by the lead in the solders, large efforts have been made to develop a lead-free soldering technology. Among the large variety of lead-free solders the Sn-Ag alloys are expected to be the best candidates. Furthermore, from a reliability point of view, there has been interest in the improved thermal fatigue resistance of solder interconnects. Therefore, in this study two lead-free solder alloys (Sn96.5Ag3.5, Sn95.5Ag3.8Cu0.7) and two lead-containing solder alloys(Sn63Pb37 Sn59Pb40Agl) were investigated and compared with each other in order to give an estimation of the reliability enhancement of the new lead-free soldering technology. These investigations were focused on mechanical and physical properties (coefficient of thermal expansion, stress-strain curves at different strain-rates, ultimate strength) as well as on the microstructural appearance of the solders.

Microstructural dependence of constitutive properties of eutectic SnAg and SnAgCu solders

The paper presents constitutive models for eutectic SnAg and SnAgCu solders. Experimental investigations were carried out on specimens of different microstructures. The three specimen types have been flip chip solder joints, pin trough hole solder joints and standard hulk solder specimens. The bulk solder specimen was a dog-hone type specimen (diameter = 3 mm, length = I17 mm). The pin trough hole solder joint consisted on a copper wire that was soldered into a hole of a double sided printed circuit board (thickness I .5 mm, solder gap 0.1 mm). The flip chip solder joint specimen consisted of two silicon chips which were connected by 4 flip chip joints (one on each comer). Flip chip bumps (footprint 200 pm x 200 pm, joint height 165 ... 200 pm) were created by printing solder paste. Constant-load creep tests were carried out on all three specimen types at temperatures between 5 “C and 70 “C. Creep data was taken for strain rates between IO-’’ SKI and IO” s-I. The specimens were tested in “as cast” condition and after thermal storage. Beside the creep behaviour of the solders, the time independent elastic plastic behaviour was determined. Strain-stress-curves were recorded from the “flip chip solder joint” specimens, using a micro shear tester. The microstructural properties of the bulk specimens and real solder joints were examined using metallographic sectioning, optical microscopy techniques, and SEMmicroprobe analysis. The results of the microstructural analysis were related to the investigated mechanical properties of the solders. Models of SnAg3.5 and SnAg4CuO.5, that can he used with the ANSYSTM FEM software package, will he presented.

The effect of downscaling the dimensions of solder interconnects on their creep properties

The creep behaviour of solders is an important input for accurate material models for FE-analysis of electronic assemblies. Usually the mechanical behaviour of solders, is been determined by tensile tests on bulk solder specimens. Although performing these tests is not complicated and the results are easy to interpret, one of the key problems lies in the fact that solder joints are very small and therefore cannot be represented by large tensile specimens. The paper describes the attempts to gain deformation data on ultra small solder joints. It compares creep data that was experimentally gained on bulky samples and on small solder joints.

Time-dependent material modeling for finite element analyses of flip chips

Finite element analysis (FEA) has been established as an effective method for reliability assessment of flip chip assemblies. The simulation results are significantly dependent on the selected material models. Regarding flip chip assemblies, this statement mainly applies to the tin lead solder of the flip chip joints and the encapsulant-the so-called underfill. Comprehensive material data of eutectic solder were determined on real flip chip joints. Based on these data, three modeling approaches were evaluated (target platform was the FEA code ANSYS): viscoplasticity (Anands model), power law creep (with 2 terms)+plasticity, and sinh law creep+plasticity (as user defined model in ANSYS). Underfills are often modeled as very simple elastic materials. Tensile tests were performed on underfill samples to study its real behavior. Two modeling approaches of a representative underfill were evaluated: linear elasticity and linear viscoelasticity. The properties of all above mentioned approaches are discussed in terms the simulated material behavior at various temperatures and deformation rates. For each combination of approaches, temperature cycling tests on a flip chip module were simulated by ANSYS. Different combinations of modeling approaches for solder and underfill led to different simulation results although each model was based on the same measurement data. The differences are discussed and conclusions are drawn about which modeling approach is preferable for typical applications.

Constitutive behaviour of copper ribbons used in solar cell assembly processes

One of the driving factors for a steady reduction in wafer and cell thickness is the present shortage of polysilicon feedstock combined with the need to reduce manufacturing costs in photovoltaic module production. Therefore materials and manufacturing processes must adapt to maintain acceptable mechanical yields and module reliability. The soldering of solar cell strings is a critical step in the production of photovoltaic modules. Mechanical load and temperature induced stresses can cause cracking in the cells. During the soldering operation, the cell and the wires heat up and expand and then later contract when the heat is removed below the melting point of the solder. The differential contraction between the Cu and the Si combined with thermal gradients, cause stress to build up in the system. Since the solder thickness (5 … 20 µm) is relatively small compared to thickness of the copper ribbon (100 … 200 µm) and the thickness of the silicon solar cell (160 … 200 µm), the constitutive behaviour of the copper ribbons is one of the key factors to reduce breakage after soldering of solar cells into strings.

Effect of Composition and Cooling Rate on the Microstructure of SnAgCu-Solder Joints

In this study the solder alloys SnAg3.5, SnAg3.0Cu0.5, SnAg3.8Cu0.7 and SnAg2.7Cu0.4Ni0.05 have been analysed in order to determine variations in microstructure caused by cooling rate, solder composition and ball diameter. Solder spheres with a diameter of approx. Oslash 1100 mum, Oslash 590 mum, Oslash 270 mum and Oslash 130 mum were solidified with cooling rates of 0.14 K/s, 1.1 K/s and 10.9 K/s. Cross sections of these specimens were analysed by optical light microscopy. Interpretations of the analysed microstructure allow a description of the solidification process, which takes place in a solder ball. It could be concluded that this process is divided in three stages: the formation of primary intermetallics, the formation of fine structured regions and the formation of coarse dendritic areas.

The Influence of Size and Composition on the Creep of SnAgCu Solder Joints

The paper presents creep data, that was gained on non eutectic SnAgCu-solder specimens with a variety of compositions. The non eutectic SnAgCu-alloys were tested in different specimen sizes: bulk specimens, FBGA solder balls, flip chip solder joints. The results of the creep experiments show that both solder alloy composition and solder joint size have a significant influence of the creep properties of the solder material. Bulk solder specimens have a rectangular cross section of 4mm times 3mm and contained the following alloys: Sn98Ag2, Sn97Ag3, Sn96Ag4, Sn97.5Ag2Cu0.5, Sn97.1Ag2Cu0.9, Sn98.8Ag2Cu1.2, Sn96.5Ag3Cu0.5, Sn96.1Ag3Cu0.9, Sn95.8Ag3Cul.2, Sn96.9Ag3Au0.1. FBGA solder balls contained three noneutectic alloys with a content of 0.5% < Ag < 4% and 0.2% < Cu < 0.8%. Flip chip solder joints contained eutectic Sn96.5Ag3.5 alloy and an non-eutectic SnAg alloy with Ag < 3%. Creep experiments have been carried out in a temperature range between T = 5degC ... 150degC. The microstructures of the various solder specimens have been analysed to understand their differences in creep behavior

Fatigue Life Prediction and Analysis of Wafer Level Packages with SnAgCu Solder Balls

Currently, various mainly tin-based lead free solders are in use or under evaluation. Many solder fatigue models have been developed to predict the fatigue life of solder joints under low-cycle fatigue conditions, however, because of the diversity of solders, more work is needed in this respect. The theoretical fatigue life prediction of solder joints, based upon non-linear finite element (FE-) calculation results, requires both a solder constitutive model and a Coffin-Manson type failure criterion, if an engineering type of evaluation is chosen. The material investigated was near-eutectic SnAgCu solder with lower silver content, i.e Sn98.5Ag1Cu0.5, which is one choice currently discussed. Properties like the lower creep resistance but a better brittle failure performance combined with lower cost can make the SnAg1Cu0. 5 solder a candidate material for a broader application. The study deals with the low-cycle fatigue performance of this material. A special wafer level test package (WLP) was used for the evaluation. The derived fatigue model is based on experimentally observed low-cycle fatigue data compared to FE results. In FEA effects of input quantities scatter, of different element types and mesh densities, and different analysis codes are studied and a general averaging strategy, to be used in Coffin-Manson type failure criterions for ball interconnects, is proposed

The Dependence of Composition, Cooling Rate and Size on the Solidification Behaviour of SnAgCu Solders

The scope of this study is to investigate the influences on the solidification of the micro structure of SnAgCu solders. It will be shown that the solidification process depends on solder composition, specimen size and manufacturing conditions. The influence of solder composition has been investigated on bulk solder ingots by varying the Ag content from 3.0 wt% to 3.8 wt% and the Cu content from 0.4 wt% to 1.5 wt%. The influence of an additional Au content was investigated on a SnAg3.0Cu0.5Au0.14 solder. Solidification experiments with different cooling rates from 0.006 K/s to 0.6 K/s have been carried on bulk solder ingots (length 23 mm; Oslash 7 mm) of the alloys SnAg3.0Cu0.5, SnAg3.8Cu0.7 and SnAg3.5Cu0.4. The results will point out the influence of cooling gradients on the microstructure. In order to compare the microstructure of these large specimen with real solder joints, solidification experiments with cooling rates from 0.33 K/s to 10.9 K/s have been carried out on SnAg3.0Cu0.5 solder balls with four different sizes (Oslash 130 mum, Oslash 270 mum, Oslash 590 mum and 0 1100 mum). Phase sizes and shapes as well as grain orientations have been investigated and compared. Experiments on directed solidification have been carried out in order to investigate the dendritic growth of beta-Sn dendrites. For that purpose a temperature gradient was generated inside the solder during solidification.

Creep and crack propagation in flip chip SnPb37 solder joints

The lifetime determining event in flip chip packages is the fracture of solder joints. Crack initiation and crack growth in micro solder joints, however, are supposed to differ very much from that of bulky samples of steel, nickel, copper etc. that are usually used in fracture mechanics tests. Hence, the commonly known fracture laws do not hold for FC joints while alternative laws have not been established yet because of a lack of experimental data. The paper presents the results of reversible shear tests on flip chip solder joints under isothermal conditions. Two micro shear testers have been designed and built for this task. One tester is optimized to achieve high precision. In contrast to similar setups, this tester is actively compensated for its finite stiffness. Therefore, it is able to record force displacement hysteresis with a resolution of better than 1 mN and 20 nm force and displacement measurements, respectively. The second tester works very similar but fits in a UHV chamber. In this way, it enables in-situ SEM observations during the test. The results of this study show that the deformation behavior of flip chip solder joints to be more alike to that of bulk samples with a comparable micro structure than it is commonly believed based on published data. The parameters of the determined creep equation indicating what deformation mechanism dominates at what strain rate. The results of the creep tests are compared with that of crack growth experiments. The influence of different deformation mechanisms on the crack growth rate is discussed.