Gallium-based liquid metal alloys exhibit unrivaled thermal performance due to their high thermal conductivity and ability to achieve low bond-line thickness (BLT). However, pure liquid metal TIMs are difficult to use in high volume manufacturing flow, due to their low viscosity and high surface tension, which result in liquid metal squeezing out and causing short circuits. Liquid metal also suffers during reliability testing, most notably highly accelerated stress testing (HAST), which subjects the samples to high temperature and humidity for an extended period of time. HAST treatment results in the rapid oxidation of the liquid metal causing degradation in the thermal performance.1 In this study, we demonstrate that the oxidation of the liquid metal can be significantly reduced by encapsulating it in polymers to make liquid metal embedded elastomers (LMEE).2,3 LMEEs provide the mechanical properties of a traditional elastomer, achieving between 200% and 400% strain at break, along with the ease of dispensing of a polymer TIM. However, LMEEs are able to be compressed to stable BLT of less than 30 m and achieve low thermal resistance. Here, we show that in addition to the polymer matrix contributing to the mechanical strength, the matrix is also a protective layer for the liquid metal from the oxidative environment experienced within the HAST. This investigation compared thermal and mechanical performance of two types of LMEE TIM specimens within 10x10 mm2 and 25x25 mm2 silicon sandwich test specimens subjected to a HAST of 96 hrs at 130 and 85% relative humidity (HAST96). Mechanical properties were evaluated by stud pull to delaminate the test specimens in tension. Comparing T0 and post HAST96, resulting adhesion strength for the 10 mm test specimens was 529 186 kPa (T0) and 612 141 kPa (post-HAST), and 335 28 kPa (T0) and 414 86 kPa (Post-HAST) for 25 mm test specimens. The Si interfaces after stud pull test showed liquid metals remained in liquid state, as opposed to solid gallium oxide in a case of excessive oxidation of liquid metals. Additionally, thermal resistance also remained unchanged when evaluated in both ASTM D5470 and thermal test vehicle (TTV) test setups. However, after HAST96 we discovered an unexplained loss of signal in confocal scanning acoustic microscopy (CSAM) after HAST96. IPC-TM-650 dye-and-pry evaluation on test specimens show the dye unable to penetrate the LMEE TIM area, indicating no delamination occurred. Subsequent HAST96 treatment on tensile specimens of the LMEE base polymer reveal an unchanging elastic modulus and strain at break of 858 62 kPa and 260 90%, respectively, at T0 and 750 43 kPa and 288 58%, respectively, post HAST96. To provide a baseline, these results are compared against a liquid metal baseline, which we observed to oxidize and form porous structures which compromised the test specimens, making thermal characterization difficult.In this talk, we investigate all of these observations, illustrating representative use cases where each type of material and testing method are appropriate. This has implications for thermal management in a variety of applications and package types, ranging from central processing units (CPUs) for consumer devices, to large-die graphics processors (GPUs) and even power electronics for the automotive industry.
Mondal et al. (Tue,) studied this question.