Michael R. Notis

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

Nanoindentation testing has been used to measure the hardness and elastic modulus of Ag 3 Sn, Cu 6 Sn 5 , and Cu 3 Sn 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 Cu 6 Sn 5 (6.5 ′ 0.3 GPa) and Cu 3 Sn (6.2 ′ 0.4 GPa) indicate a potential for brittle behavior, while Ag 3 Sn (2.9 ′ 0.2 GPa) appears much softer and ductile. Using a bulk Cu 6 Sn 5 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.

A review: Constitution of the AlCrNi system

Abstract The existing literature on phase constitution in the AlCrNi system has been reviewed. A critical evaluation of experimental investigations confined to the AlNiCrNi region has been made. A reaction sequence describing phase equilibrium relationships has been developed based on the respective binary systems, microstructures observed in pack-aluminized samples and studies of equilibrated ternary alloys. A series of isothermal sections is presented for the solid state phase equilibria, particularly for the nickel-rich corner. Crystallographic and thermodynamic data are summarized briefly.

Microstructural evolution in lead-free solder alloys: Part I. Cast Sn–Ag–Cu eutectic

Coarsening of the ternary eutectic in cast Sn–Ag–Cu lead-free solder alloys was investigated. The process was found to follow r 3 ∝ t kinetics where r is the rod radius of the dispersed phase and t is time. The effective activation energy for the process is 69 ± 5 kJmol -1 . The two types of intermetallic rods, Cu 6 Sn 5 and Ag 3 Sn, in the eutectic structure coarsen at different rates, with each having a different rate-controlling mechanism. The overall coarsening kinetics for the Sn–Ag–Cu ternary eutectic is significantly slower than that found for the Pb-Sn eutectic, which has implications for long-term reliability of Sn–Ag–Cu solder joints.

Mechanical properties of intermetallic compounds in the Au-Sn system

The mechanical properties of intermetallic compounds in the Au–Sn system were investigated by nanoindentation. Measurements of hardness and elastic modulus were obtained for all of the confirmed room-temperature intermetallics in this system as well as the β phase (8 at.% Sn) and AuSn 4 . Overall, it was found that the Au–Sn compounds have lower hardness and stiffness than common Cu–Sn compounds found in solder joints. This finding is in contrast to common knowledge of “Au embrittlement” due to the formation of either AuSn 4 or (Au,Ni)Sn 4 intermetallic compounds. This difference in understanding of mechanical properties of these phases and the resulting joint strength is discussed in terms of reliability and possible failure mechanisms related to interface strength or microstructural effects. Indentation creep measurements performed on Au 5 Sn, Au–Sn eutectic (29 at.% Sn) and AuSn indicate that these alloys are significantly more creep resistant than common soft solders, in keeping with typical observations of actual joint performance.

Microstructural evolution in lead-free solder alloys: Part II. Directionally solidified Sn-Ag-Cu, Sn-Cu and Sn-Ag

The tin-silver-copper eutectic is a three-phase eutectic consisting of Ag 3 Sn plates and Cu 6 Sn 5 rods in a (Sn) matrix. It was thought that the two phases would coarsen independently. Directionally solidified ternary eutectic and binary eutectic samples were isothermally annealed. Coarsening of the Cu 6 Sn 5 rods in the binary and ternary eutectics had activation energies of 73 ′ 3 and 82 ′ 4 kJmol - 1 , respectively. This indicates volume copper diffusion is the rate controlling mechanism in both. The Ag 3 Sn plates break down and then coarsen. The activation energies for the plate breakdown process were 35 ′ 3 and 38 ′ 3 kJmol - 1 for the binary and ternary samples respectively. This indicates that tin diffusion along the Ag 3 Sn/(Sn) interfaces is the most likely the rate-controlling mechanism. The rate-controlling mechanisms for Cu 6 Sn 5 coarsening and Ag 3 Sn plate breakdown are the same in the ternary and binary systems, indicating that the phases evolve microstructurally independently of one another in the ternary eutectic.

Continuous cooling transformation kinetics versus isothermal transformation kinetics of steels: a phenomenological rationalization of experimental observations

Abstract The existing literature has been examined and rationalized to test the general validity of a number of generally accepted concepts concerning the overall transformation kinetics of ferrous alloys. Considerable confusion exists because of the mixup of the continuous cooling kinetics with the isothermal transformation kinetics. Therefore, these two topics are discussed separately. For the continuous cooling process, the following topics are examined: (1) the suppressibility of the martensite transformation at high cooling rates; (2) the cooling rate dependence of M s (martensite transformation-start temperature) and B s (bainite transformation-start temperature); (3) the formation conditions of lath martensite and twinned martensite; and (4) the various features of continuous cooling transformation (CCT) diagrams. For the isothermal transformation process, the following issues are examined: (1) the isothermal transformation kinetics of martensite; (2) the relationship between athermal transformation of martensite and isothermal transformation of martensite; (3) the general features of time-temperature-transformation (TTT) diagrams; (4) the validity of the “isothermal martensite” concept; and (5) the definition of M s and B s for isothermal transformations. Among the main conclusions are: (1) twinned martensite can be formed in all steels, even in pure iron and low-carbon and/or low-alloy steels; (2) isothermal transformation of martensite always follows C-curve kinetics; and (3) B s and M s for isothermal transformations are different from those obtained from cooling transformations. Comparison of literature results with the present assessment of isothermal B s and M s is made and good agreement is observed. The weakness of using TTT diagrams to analyze the continuous cooling kinetics is also discussed. Moreover, (metastable) product diagrams for austenite decomposition are established for both the continuous cooling process and the isothermal transformation process in order to develop a clearer paradigm for both processes.

Interface reactions between titanium thin films and (1¯1 2) sapphire substrates

We have studied the reactivity of Ti with the R-plane (1¯1 2) of sapphire from room temperature to 1250 °C by X-ray photoemission spectroscopy (XPS), transmission electron microscopy (TEM), X-ray diffraction (XRD), and Rutherford backscattering spectroscopy (RBS). The combination of these techniques allowed the interface reactions to be studied from the monolayer regime up to the bulk regime. XPS showed that at room temperature, monolayer coverages of Ti reduced the sapphire surface to form Ti-O and Ti-Al bonds. TEM, XRD, and RBS showed that annealing of room-temperature deposited Ti resulted in an interfacial region consisting of two layers, a Ti3Al[O] layer adjacent to the sapphire and a Ti0.67 [O0.33] layer at the free surface. The growth of the Ti3Al[O] layer had an activation energy of 103.4±25 kJ deg-mole. The nature of the interfacial reaction between Ti and sapphire was similar for Ti coverages from the monolayer to the bulk regime.

The microstructure of 55 w/o Al-Zn-Si (Galvalume) hot dip coatings

Light metallography, x-ray diffraction, scanning electron microscope (SEM), (EMPA), and scanning transmission electron microscope (STEM) were used to study the microstructural characteristics of 55 w/o Al—Zn-Si (Galvalume) hot-dip coatings on steel. Processing variables of coating thickness and dip time were studied and a thin foil sample preparation technique involving ion beam thinning for the STEM evaluation of coating cross sections was developed. The spangled coating surface was comprised of a fine dendritic network, and correlations between dendrite arm spacing and spangle size with coating thickness have been made. The overlay, or solidified bath, consists of grains of A1-rich dendrites along with Zn-rich interdendritic zones. EMPA in conjunction with TEM imaging showed that these Zn-rich interdendritic regions had a heavy concentration of precipitates. Selective etching of the overlay revealed a forest of Si particles which grew out from the alloy layer. The alloy layer was studied using the EMPA and STEM and it was found that up to five intermetallic phases were present after long dip times.

The effects of silicon on the reaction between solid iron and liquid 55 wt pct Al−Zn baths

The effect of various silicon levels on the reaction between iron panels and Al-Zn-Si liquid baths during hot dipping at 610°C was studied. Five different baths were used: 55Al−0.7Si−Zn, 55Al−1.7Si−Zn, 55Al−3.0Si−Zn, 55Al−5.0Si−Zn, and 55Al−6.88Si−Zn (in wt pct). The phases which formed as a result of this reaction were identified as Fe2Al5 and FeAl3 (binary Fe−Al phases with less than 2 wt pct Si and Zn in solution),T1, T2, T4, T8, andT5H (ternary Fe−Al−Si phases), andT5C (a quaternary Fe−Al−Si−Zn phase). Compositional variations through the reaction zone were determined. The phase sequence in the reaction zone of the panel dipped for 3600 seconds in the 1.7 wt pct Si bath was iron panel/(Fe2Al5+T1)/FeAl3/(T5H+T5C)/overlay. In the panel dipped for 1800 seconds in the 3.0 wt pct Si bath the reaction zone consisted of iron panel/Fe2Al5/(Fe2Al5+T1)/T1/FeAl3/(FeAl3+T2)/T5H/overlay. In the panel dipped for 3600 seconds in the 6.88 wt pct Si bath the phase sequence was iron panel/Fe2Al5/(Fe2Al5+T1)/(T1+FeAl3)/(T1+T2)/T2/T8/T4/overlay. The growth kinetics of the reaction zone were also studied. A minimum growth rate for the reaction zone which formed from a reaction between the iron panel and molten Al−Zn−Si bath was found in the 3.0 wt pct Si bath. The growth kinetics of the reaction layers were found to be diffusion controlled in the 0.7, 1.7, and 6.88 wt pct Si baths, and interface controlled in the 3.0 and 5.0 wt pct Si baths. The presence of the interface between theT2/T5H, Fe2Al5/T1, orT1/FeAl3 phases is believed responsible for the interface controlled growth kinetics exhibited in the 3.0 and 5.0 wt pct Si baths.

The reaction between solid iron and liquid Al-Zn baths

The reaction which occurred between iron panels and Al-Zn baths during hot dipping was investigated. Three baths were studied: 45Al-55Zn, 55Al-45Zn, and 75Al-25Zn (in wt pct) in the temperature range of 570 to 655 °C. The reaction between the iron panel and the Al-Zn bath was very severe and in all cases the iron panel was totally consumed by the bath in less than two minutes. The rapid attack of the iron panels by the Al-Zn baths was attributed to two separate causes depending on growth conditions. First, in some panels the intermediate layer which formed between the iron panel and the molten bath was nonadherent. This resulted in the direct contact of the molten bath with the iron panel at a nonequilibrium interface, which presented a large driving force and little inhibition for the reaction. Second, in panels containing an adherent alloy layer, the layer had channels of liquid Zn which extended from the molten bath to the iron panel. These channels allowed rapid transport of Zn and Al to the iron panel which resulted in a very high reaction rate. The controlling step in the reaction between the iron panel and molten Al-Zn bath was the diffusion rate of Al in the molten bath to the surface of the iron panel. The diffusion coefficient of Al in the molten bath was found to be in the range of 1 × 10-5 to 5 × 10-5 cm2/s. Microstructural, electron microprobe, and X-ray diffraction data are presented to support the above-mentioned mechanisms and conclusions.