Nanoindentation measurements on Cu–Sn and Ag–Sn intermetallics formed in Pb-free solder joints

Abstract

Nanoindentation testing has been used to measure the hardness and elastic modulus of Ag3Sn, Cu6Sn5, and Cu3Sn intermetallics, as well as Sn–Ag–Cu solder and pure Sn and Cu. The intermetallics were fabricated by solid-state annealing of diffusion couples prepared from a substrate (Cu or Ag) and a solder material (Sn or Sn–Ag–Cu solder), providing geometries and length scales as close as possible to a real solder joint. Nanoindentation results for the intermetallics, representing penetration depths of 20–220 nm and loads from 0.7 to 9.5 mN, reveal elastic/plastic deformation without evidence of fracture. Measured hardness values of Cu6Sn5 (6.5 ± 0.3 GPa) and Cu3Sn (6.2 ± 0.4 GPa) indicate a potential for brittle behavior, while Ag3Sn (2.9 ± 0.2 GPa) appears much softer and ductile. Using a bulk Cu6Sn5 sample, Vickers hardness testing revealed an indentation size effect for this compound, with a hardness of 4.3 GPa measured at a load of 9.8 N. An energy balance model is used to explain the dependence of hardness with load or depth, where the observation of an increasing amount of fracture with applied load is identified as the primary mechanism. This result explains discrepancies between nanoindentation and Vickers results previously reported.

Mechanical behavior of Ti cold spray coatings determined by a multi-scale indentation method

“…Vickers hardness, VH, was calculated by dividing the applied force (in kg) by the indent surface area (in mm 2 ) [5] in the standard manner. However, for a comparison between the nanohardness, the Vickers hardness were converted to the units of GPa and a hardness, H micro , based on a projected contact area of the Vickers indenter [5,17,34,35]:…”

“…A strain gradient plasticity model, commonly called the Nix-Gao model [8], has been used extensively in the literature, where the model allows one to determine a true material hardness independent of load or indent size. For ceramics and other brittle materials, such as intermetallic compounds, there is an indentation size effect due to the extent of fracture of the material [6,[15][16][17]. Depending on what one wishes to learn about the material, low load indentations can be used to explore the yield strength [17] or the high load limit can be applied to examine a hardness under conditions of extreme fracture [6,15,16].…”

Intermetallic Layer Growth Kinetics in Sn‐Ag‐Cu System using Diffusion Multiple and Reflow Techniques

“…[46][47][48]56,57] The interfaces identified by the SEM/EDS measurements between the Cu and IMC, and between IMC themselves at any stage of the annealing are linear because the interfacial processes take place in the solid state and the Cu 3 Sn grows out of the Cu 6 Sn 5 phase. [9,[15][16][17] The exception is the interface between initially liquid solder and a Cu 6 Sn 5 layer (scallop-like morphology) formed during the early stage of annealing. [54,58,59] The scallop-type interface is an effect of the fast diffusion of Cu into the solder during reflow process (a process called liquid grooving) as well as solidification after the reflow process.…”

“…The growth rate of the Sn/Cu IMCs follows a parabolic kinetic law as it is reported in the literature. [15,26,27,[44][45][46]48,52,54,56,57] The corresponding activation energies are reported in Table 3. The growth rate of the IMC layers follows the parabolic rate law (dX=dt ¼ k=X), the observed correlation was in the range from [31] Cu|Sn thin films 101.3 83.9 25-220 [64] Cu thin film|Sn 6 Cu|Sn-0.6Cu-0.05Ni -66.3 80-150 [47] 0.95 to 0.98 (R 2 linear least squares fitting).…”

“…Wang et al [5] studied the microstructural evolution of Cu/Sn-Ag($5 mm)/Cu Cu-bump-on-line (CuBOL) joints during isothermal annealing at 180°C for up to 72 h. According to their work, both IMCs are formed at the interface during the early stages of the Cu/Sn liquid phase reaction. The formation of Cu 3 Sn takes place at the expense of Cu 6 Sn 5 [9,[15][16][17] ; its standard Gibbs energy of formation is lower (À 8.25 kJ mol À 1 ) than that of Cu 6 Sn 5 (À 7.00 kJ mol À 1 ) at room temperature. [18,19] Sunwoo et al [12] postulated that absence of a Cu 3 Sn phase at the Cu/Sn interface is due to the limited nucleation rate of this phase.…”