Shear Strength and Thermomechanical Reliability of Sintered Ag Joints Containing low CTE Non-metal Additives for Die Attach Open Access

The demand for high power devices in application for automotive, renewable energy, and power industries have given rise to extensive attention to develop high temperature die-attach materials so that the final modules can be operated in high temperature hazardous environment. Increase of power density is a major trend in power electronics that will be used in inverters and converters, for examples, in EV and HEV, for variable speed drives, regenerative energies and other applications [1], [2]. Increasing power density means increase of current density in these devices, which results in high junction temperature or operation of these devices at high temperature. Today, removal of heat generated inside power chip with high current density from the chip into the interconnection layers on a time scale of some 10–100μs after the initial stress challenged, not only requires high thermal capacities of the silicon itself [1], [2], but also needs the interconnection layer with high electrical and thermal conductivity that could be survived in high operating temperature. Silver sintering with low temperature sintering process less than 275°C and the sintered Ag joints having high thermal and electrical conductivities have been mostly considered as an alternative to replacing high temperature lead-free soldering for die attach [3]–[5]. Another major advantage of the sintered silver joint over other high temperature eutectic soldering joints is the promised long-term reliability because of its high melting temperature (961°C). CTE mismatch of different materials used in different components in a power electronic package shown in Fig.1[3], however, commonly generates thermal stress into chips, substrate and the interconnection layer, especially at the interface between the joints and chip/substrate during thermal cycling test and/or high temperature operation, which would cause severe delamination of the metallization layers on die backside and at the interface on substrate side, or cracking of the die and substrate, or lift-off of wire-bonding[5]. To reduce thermo-mechanical stress induced by mismatch of CTE between different components in the power package, several options reported in literature have been tried as follows: 1) Reduction of chip thickness; 2)Use of direct bonded copper (DBC) and active metal brazing (AMB) ceramic substrates, instead of lead-frame; 3)Adjustment of CTE of the interconnection materials i.e. adding fillers with low CTE to silver sinter pastes as starting materials that can form low CTE sintered silver joints[6]; 4)Tune of elastic modulus <RXQJ¶V PRGXOXV of the sintered Ag joints[5], [8].

Formation of the sintering joint is normally based on metal inter-diffusion at an enhanced surface area of micrometer or nanometer scale particles and/or hybrid silver particles in a paste form with organic capping agent on the surface of the particle dispersed in organic solvent mixed with thinner and binder [5], [7]. Thus, it is ready method to add low CTE fillers to silver sinter paste to form a sintered Ag matrix joint containing low CTE particles that could play a role in reducing mismatch of CTE between interconnection and chip or substrate.

In our recent research work we have been developing several series of silver sinter pastes containing low CTE non-metals additives such as aluminum nitride (AlN), boron nitride (BN), silicon carbide (SiC), silicon dioxide (SiO₂) or silica, zirconium tungstate (ZrW₂O₈) particles to reduce and adjust CTE of the sintered Ag joints. Influence of the composition of the sinter silver pastes with a variety of the additives on shear strength, thermos-mechanical reliability and micro structural morphologies of the sintered Ag joint bonding Ag Si die on Ni/Au DBC substrate sintered by a pressureless sintering at 250°C in air for 30 to 60min have been investigated extensively.

In this present paper we will present a pressurelessly sintering process of silver sinter paste containing a low CTE non-metal particles; shear strength and effect of the composition on shear strength of the as-sintered joints; preliminary data for thermomechanical reliability based on thermal cycling and shock tests; characterization of voids and cracks by x-ray scanning as well as cross-section microstructure by SEM.

Silver nano particles was purchased from a supplier and used as received. Low CTE non-metal powder with microscale particle size were ordered from a company and used as received in the preparation of silver sinter pastes. Silicon wafers with titanium, nickel, and silver multiple layers of 2000Å thickness each were purchased from RVM Inc, OR and then diced to small dies with the size of 3×3 mm² by MPD, CA. DBC alumina substrate with Ni/Au surface finish having gold layer of 200nm and nickel layer of 4μm, respectively, was ordered from Remtec, Inc.

A typical sintering procedure in this research was conducted by pressure-less sintering at 250°C in the air for 60min after applying silver sinter paste onto Ni/Au DBC substrate using a stainless-steel stencil with 4mil thickness, followed by placing Ag Si die on the top of the printing pastes using pick-n-place machine. The detail sintering profile used in this study, including three steps of drying, sintering and cooling, is illustrated in Fig. 2 (a). Firstly, temperature rises to 120°C at a heating ramp of 9.5°C/min and then kept at 120°C for 10min; secondly, temperature further rise to 250°C with a heating ramp of 12°C/min and kept at 250°C for 60min; finally, the sintered joints with die and substrate cools down to ambient temperature.

X-ray scanning was used to check voids and cracks of the sintered joints synthesized by pressureless sintering of silver sinter paste and the sintered joints after thermal cycling, shock and thermal aging tests. Cross-sectional samples were prepared by cutting a specimen composed of die, Ag joint and Ni/Au DBC substrate, followed by imbedding into epoxy resin mold. Scanning electronic microscopy (SEM) with energy dispersive detector (EDS) was used to characterize micro structural morphologies. Thermal cycling test using a profile of −40 to 175°C and thermal shock test using a profile of −55 to 155°C with 10min of dwell time at low and high temperature zone were performed to estimate the thermo-mechanical reliability by measuring shear strength before and after thermal cycling test. Shear tests were performed at ambient and/or elevated temperature using XYZ Shear Tester Condor 250 with a customized hot stage at the shear speed of 100micron per second.

To understand thermal expansion behaviors of the metal composites, many researchers have induced different types of mathematic models to predicate the CTE of the composites. Typical models that were widely used among these models include rule-of-mixture (ROM) [9], Turner's model, Kerner's model [10] and Schapery's model [11].

The CTE of the composite according to ROM can be expressed as following equation (1):

where α is the CTE of the composite or each individual component, V is the volume fraction of individual component, and subscripts, c, m, p refer to the composite, matrix and particle respectively.

Kerner was first researcher to consider that the shape of additives would affect estimation of the CTE of the composites. The Kerner model assumes that the filler or additive is spherical and surrounded by a uniform layer of matrix. Therefore, the CTE of the composite is identical to that of a volume component composed of a spherical filler particle surrounded by a shell of matrix, both phase having the volume fractions present in the composite. Kerner model can be expressed by the following equation (2):

where α is the CTE of the composite or each individual component, V is the volume fraction of individual component, and subscripts, c, m, p refer to the composite, matrix and particle respectively, K is the bulk modulus, which is calculated by the standard relationship as is expressed according to equation (3):

where E and G is Young's modulus and shear modulus, respectively.

To estimate CTE of the sintered Ag joints with additives, the predicted CTE of the as-sintered Ag joints with low CTE non–metal additives is calculated based on ROM and Kerner models as is summarized in Fig. 3.

Shear strength of the sintered Ag joints with various content of additives were determined. Fig. 4 presents shear strength values as a function of the additive composition in the assintered joints. Shear strength values of the sintered joint containing additives are much lower than the sintered Ag joint without additives, and decrease with an increase of the additive content. It is worthy to notice that shear strength of the sintered joints with 2wt% of additives is ca 30MPa which meets requirement by DA5 and military standard and quite comparative to the sintered Ag joint without additive.

Therefore, high temperature shear, thermal cycling and shock tests on the sintered joints with 2.5wt% of additive were conducted to evaluate high temperature shear performance and thermomechanical reliability. High temperature shear strength of the sintered joints with and/or without low CTE non-metal additives is summarized in Fig.5. It can be seen that shear strength of the sintered Ag joint without additive gradually reduce from 40MPa to 20MPa from room temperature to 300°C whereas the sintered joints containing 2.5wt% of the additive remain the similar shear strength value to the as-sintered joint, 27MPa, from room temperature to 250°C, and then drop to 19.74MPa at 300°C. It seems that the additive particles would actually enforce shear strength of the sintered Ag joints at high temperature.

Fig. 6 shows shear strength of the sintered joints with and/or without low CTE additives before and after thermal shock test. It is obvious that shear strength values of both the sintered joints with and without additives increase with an increase of thermal shock cycles from time zero to 2000cycles, which means that there was no degradation in shear strength during thermal shock test using a profile of −55 to 155°C with a dwell time of 10min at the low and high temperature zone.

SEM cross-sectional morphologies of the sintered Ag joints with and without the additives shown Fig.7 indicate that there is no void existed in the as-sintered joint and/or evolving in the sintered joints during thermal shock test after 2515cycles. Neither was cracks observed at 2515cycles. Increase of shear strength of both the sintered joints with and without the additive is perhaps associated with further sintering of the sintered Ag during TST.

Shown in Fig.8 is shear strength values of the sintered Ag joints with and/or without low CTE additives as a function of thermal cycles using a profile of −40 to 175°C with dwell time of 10min at the high and low temperature. The sintered Ag joint without additive has the highest shear strength at 1000cycles which is even higher than the as-sintered joints at time zero. Shear strength remains the similar value, 34.7MPa, to the as-sintered Ag joints at time zero at 2000cycles.

SEM cross-sectional morphologies obtained indicate that there is no crack and void evolved in the sintered Ag joints without the additives during thermal cycling test as is shown in Fig.9(b).

For the sintered Ag joints containing the additives, shear strength gradually reduces from 27.35MPa to 20.99MPa at 1000cycles, but it is very interesting that it jumps up to 34.7MPa at 2000cycles and significantly higher than the assintered sample with shear strength of 27.4MPa at time 0. SEM cross-section morphologies of the sintered joints containing the additives reveal that there is no void, crack and delamination as is shown in Fig. 9(a) at 2000cycles of −40 to 175°C. It is indicative of good thermomechanical reliability of the sintered joints with the low CTE additives although the shear strength value is fairly lower than the sinter silver joint without the additives.

It was well known that formation of a dense layer and a depletion layer from microscale-based sintered Ag joints on the Au surface finish substrate or the region near Au metallization die backside due to the rapid surface diffusion of Ag atom into Au surface has been extensively reported [12], which normally lead to degradation of the sintered Ag joints in shear strength. It was found in our research that the sintered Ag joints with and without the additives did form a dense layer on Ni/Au DBC substrate after thermal aging at 250°C, respectively.

It can be seen from Fig.10 (a) that a 5μm thick dense Ag layer in the sintered joints without additive formed on the substrate side after 250hours of thermal aging, but no depletion layer observed. Thickness of the dense layer increases with increasing time of thermal aging at 250°C, but shear strength of the sintered joint without additive does not drop as is shown in Fig.11. The reason for the low shear value of the sintered Ag joints without the additive at 250hrs is that delamination occurs to the Ag metallization layer on die backside during shear test that is illustrated in Fig.12a.

After thermal aging at 250°C for 250hours, shear strength value of the sintered joint with the additives remains the similar shear value to the as-sintered joints. It is very interesting that there is no dense layer formed, and none of delamination of the Ag metallization layer occurred during shear test as is illustrated in Fig.10(b) and Fig.12(b), respectively. Shear strength of the sintered with the additive, however, dramatically drops to 12MPa after 500hours of thermal aging. SEM cross-section microstructural morphologies obtained reveal that there are numerous large voids and/or depletion layer formed near the end areas of the joints after thermal aging at 250°C for 500hours, which might be associated with deterioration of shear strength value.

Pressureless silver sinter pastes, with and without low CTE non-metal additive, can be sintered with pressureless sintering process at 250°C or even lower temperature in air to form strong bond joining between Ag Si die and Ni/Au DBC substrate. The content of low CTE non-metal additives has a significant influence on shear strength of the as-sintered joints and exhibits a shear strength values decrease with an increase of the additive content in silver sinter paste.

Thermal cycling (−40 to 175°C) and thermal shock (−55 to 155°C) test results reveal that both the sintered Ag joint with and/or without low CTE additive have good thermomechanical reliability.

HTS indicates that the sintered Ag joint without additive has excellent thermal stability and the sintered joint containing low CTE non-metal additive, with a benefit of reduction of thermal stress, can eliminate the delamination of the Ag metallization layer on die backside.

Low CTE non-metal additives have great influence on the structure of the sintered joint in which large voids and depletion layer at the end areas were observed after 500hour of thermal aging at 250°C, which is associated with the deterioration of shear strength.

In future, study on effect of the distribution of low CTE additive on property of the sintered joints is in progress.

Authors would like to thank all the coworkers in Indium for their kindness, support, collaboration and specially acknowledge for Co-op project program at RIT so that we could have a co-op student assisting with developing new product.