Microstructure and Mechanical Properties of Pressureless Sintered Silver Die-attach Materials Open Access

Power electronics with wide-bandgap semiconductors are expected to be operated at temperatures greater than 200 °C to improve their energy efficiency. However, one of the challenges for high temperature operation is that die-attach materials, such as conventional solders like Pb-Sn and Sn-Ag-Cu become liquid and may remelt or become brittle. For this reason, it is expected that sintered silver materials will be used as die-attach materials in power electronics operated at these conditions, because silver has much higher melting point and favorable thermal and electrical conductivities than the conventional solders [1]–[4]. On the other hand, die-attach materials also need reliable mechanical properties to endure thermal stresses for long periods of time (several decades) for practical use in power electronic packages, which consist of many components that may have different constants of thermal expansion. Reliability tests in power electronics are usually conducted by thermal cycling tests and power cycling tests, where die-attach materials are subjected to thermal stresses generated by periodically application of high and low temperatures. Therefore, in developing new materials or fabrication procedures, it is important to measure the microstructural and mechanical properties of die-attach materials such as those based on sintered silver [5]–[7].

Sintered silver materials are generally classified into two types: those that are formed pressureless methods and those that are formed with pressure-assisted methods. Pressureless sintered materials are gaining much attention because they avoid damaging semiconductor dies from the application of pressures that can be as high as 30 MPa [1]–[4]. According to the size effect [8], silver nanoparticles can be sintered at temperatures less than 300 °C without external application of pressure. Nanoparticle-based sintered silver materials have also benefits in their mechanical properties such as low Young's modulus and high elongation, and they have high reliability at temperatures higher than 200 °C [9]. On the other hand, there are some reports that state sintered silver materials become brittle by continuous exposure to high temperatures (ca. 250 °C) after several hundred hours [10]–[12]. In this work, the microstructural and mechanical properties of a nanoparticle-based pressureless sintered silver material were examined to determine its reliability according to properties after exposure to high temperatures (ca. 250 °C).

Silver sinter pastes used in this work were prepared by adequately mixing silver nanoparticles with organic solvents. The pastes were printed on 20 mm square bare copper lead frames (thickness of 1 mm) with a stencil mask (thickness of 90 μm) hollowed out a square having a side of 11 mm. The printed pastes were then pre-dried at 70 °C for 30 min to remove most of the organic solvents. Immediately before printing, the bare copper lead frames were rinsed with dilute sulfuric acid to eliminate oxidized layers on the surface. Au/Ni-plated 10 mm square silicon dies (thickness of 0.3 mm) were mounted on the pre-dried pastes with a force of 0.2 MPa. Pressureless sintering was carried out in a nitrogen atmosphere with an electric oven. Oxygen concentration of every sintering process in this work was around 300 ppm according to actual measurements. The pressureless sintering process took 70 min, beginning from room temperature to 275 °C, that was followed by holding samples for 60 min at the maximum temperature.

The shape of the ISO 37-4 dumbbell specimen used in the tensile tests is given in Fig. 1. The silver pastes were printed on a glass plate with a stencil mask from which the dumbbell pattern was hollowed out. The upper surface of the printed paste was covered with another glass plate instead of the silicon die during the sintering to reduce the evaporation of organic solvents from the upper surface of the printed paste. All sintering procedure and conditions used the same general framework as just described.

High temperature exposure tests were conducted by placing samples in an electric oven at 250 °C in air up to 1000 h (about 42 d). Microstructure and porosity of the sintered silver layers were examined with a scanning electron microscope (SEM), Hitachi S4800. Cross sections of sintered silver layers were cut off with a blade and were treated by ion milling with argon. Porosity was calculated from cross-sectional SEM images with imaging software. Void and delamination were examined with a scanning acoustic microscope (SAM), Japan Krautkramer μ-SDS. All SAM images were taken from the side of copper lead frames. Tensile tests were conducted by a tensile testing machine, Instron 5969, and at a temperature of 23 °C with an initial strain rate of 10⁻³ s⁻¹.

Photos of a test piece prepared with the sintered silver material before and after the high temperature exposure test of 250 °C are given in Fig. 2. The surface color of the bare copper lead frame apparently changed from copper luster to black and dark brown with an increase in test time because of the oxidization of bare copper, whereas the surface color of the sintered silver that was outside of the die retained its silver luster even after 1000 h of heat treatment.

SAM images of the test piece before and after the high temperature exposure test are given in Fig. 3. White dotted lines in the SAM images show die position. White spots in Fig. 3d correspond to copper oxides on the surfaces of the copper lead frames. The copper oxides could not be removed by ultrasonic washing, so that these white spots are omitted from later discussion. Appearance of the central region of the die did not substantially change even after 1000 h although there was some grey areas that seemed to be macro voids in the sintered silver layer that were initially present before the heat treatment.

The white area seen along the white lines at the upper edge of the SAM images (Fig. 3) gradually enlarged with an increase in time and it apparently expanded after 1000 h heat treatment (white arrow in Fig. 3d). It is most likely that the white area was delamination at the edge region of the die where it had suffered the maximum thermal stress in the die. On the other hand, it is notable there were no white areas along any of the other three edges. It is probable that delamination can be reduced by improving the printing or mounting procedures.

Time dependence of cross-sectional SEM images of the test pieces are given in Fig. 4. Microstructure of the sintered silver consisted of a silver matrix and pores. Pores in the sintered silver were homogenously distributed in the silver matrix. Pore sizes were less than 1 μm and did not change much after the high temperature exposure test. The porosities were 8% at 0 h and 5% at 1000h, respectively, suggesting that there was no difference between before and after the test. It has reported that sintered silver materials can become brittle that is accompanied with an increase in pore size by continuous exposure to temperatures of 250 °C for long periods of times [10]–[12]. However, for the pressureless sintered silver materials formed in this work, no embrittlement or pore size changes were observed that can be attributed to the pressureless method and the use of nanoparticles.

The SEM images in the central regions showed that both interfaces of the sintered silver looked clean and had no cracks and delamination. On the other hand, the SEM image in the edge region at 1000 h showed a crack in the sintered silver layer and the copper lead frame itself (white arrow in Fig. 4d), which probably corresponds to the delamination seen in the SAM image (Fig. 2d).

Other similar type delaminations did not appear in a SAM image (not shown) of test pieces using a silver-plated copper lead frame instead of bare copper lead frames. It can be considered that the sintered silver material was not of the crack in the copper lead frame, but was most likely due to oxidation of the bare copper metal that caused a crack in the copper lead frame and thus a crack in the sintered silver layer.

Stress-strain curves of the sintered silver material before and after the high temperature exposure test are given in Fig. 5. Slopes at initial elastic regions of the curves were almost the same regardless of test time, meaning that Young's modulus did not depend on exposure time. Since Young's moduli were constant before and after the tests, it is most likely that the sintered silver material did not show embrittlement. Young's moduli of 14 GPa lower than conventional solders should provide for a high reliability in electronics packages operated at high temperatures [9].

The S-S curves showed that the material before the high temperature exposure test had brittle behavior, typical for sintered silver materials in which stress increases with an increase in strain and then the material suddenly breaks at a strain. It is most likely that the brittle behavior corresponds to the microstructure of sintered silver materials including a lot of pores, which can start to break as defects come under tensile force. Materials that undergo high temperature exposure became plastic and elongated at break points with increases from 1% to 3.7%. It is most likely that the increase in elongation was not caused by an increase in degree of sintering by the high temperature exposure, but by a decrease in the defects of microstructure, because tensile strength is generally a function of defects in a material. It can be considered that the high temperature exposure test had an annealing-like effect that lessened defects in the dumbbell specimen. In other words, the brittle behavior of the S-S curve at 0 h might not reflect the true nature of the sintered silver material itself, could be influenced by the shape of the dumbbell form and the dumbbell preparation. Further studies on the crystallography of the sintered silver material are underway.

Microstructural and mechanical properties of a sintered silver material originating in nanoparticles were measured. The sintered silver material could attach to bare copper by a pressureless sintering process. After an exposure test at a temperature of 250 °C and for 1000 h, the sintered silver material did not show any substantial change in microstructure, and it had the same Young's modulus of 14 GPa as that before the test. Pore size of the sintered silver material did not drastically change after the high temperature exposure test. The sintered silver material in this work showed neither embrittlement nor an increase in pore size, so that it can be considered to have reliable physical and mechanical properties at high temperatures.