Michael Zhang1
Lake Oswego High School1
Introduction<br/>Thermal dissipation surges as one of the top issues for computer processor and chip packaging. Thermal paste is a widely accepted solution [1,2]. Galinstan-based liquid metal thermal paste has attracted much attention and earned market share in recent years. Superior thermal conductivity gives them advantages over conventional thermal paste . But most of the suppliers do not disclose the ratios or even what kind of components they use. In this study, we explore the effects of different alloys and fillers on the thermal properties of Gallium and Galinstan.<br/>Experimental<br/>All materials in this study are bought through online retailers. A series of thermal pastes are designed by mixing Gallium-based liquid metals with fillers, such as In, Sn, Al, Cu, SiC, diamond. A propane gun was used to mix fillers with Ga.<br/>The apparatus for measuring thermal conductivity was made from scratch.<br/>The major parts are:<br/>1. Aluminium block is chosen as the heat block which is preheated using propane. A constant temperature of 200C is maintained by a heat source placed underneath the block.<br/>2. The solutions are injected into the PTFE tube via syringe. PTFE tubes are verified to sustain up to 250oC without any deformation. During experiments, up to 6 tubes, referred to as legs, transfer heat at the same time. Ideally, each leg is under the same conditions.<br/>3. Copper rods, 8mm diameter and 15mm length, are inserted into the ends of the PTFE tubes. One end is inserted into 8mm holes in the heat block, whereas the other end serves as a heat sink.<br/>4. Temperature is measured with a digital thermocouple thermometer with resolution 0.1oC.<br/>5. During the test, A cardboard box with reflective aluminum foil covering the interior wall covers the apparatus to reduce thermal dissipation due to radiation.<br/>At the very beginning of experiments, two legs of Galinstan were run at the same time to check test repeatability. When measuring temperature at the heat sink side vs time, temperature delta is within 20%.The thermocouple could detect differences in temperature between different substances accurately. The thermal performance collected by the homemade experimental setup can be trusted.<br/>Results and Discussion<br/>The first wave of the experiment compares Ga 100%, Ga80% : In 20%, Ga 68% : In 20% : Sn 12%. The mixture, a eutectic alloy, has a melting point below room temperature, and Sn and In cause this property. The viscosity allows the thermal paste to fill gaps between chip and heat sink.<br/>The 2nd wave of the experiment is to add Al power or Cu power into the homemade Galinstan (Ga 68%, In 20%, Sn 12%). For Al, the mixture looks like an amalgamation of liquid Galinstan and Al. In another solution, 325 mesh copper power was added into Galinstan, but is not alloyed. As a result, the thermal conductivity of Galinstan + Cu is relatively ~70% of pure Galinstan. An explanation for the 30% loss is that voids between liquid Galinstan and Cu power are not favorable for heat transfer. An innovative method is needed.<br/>The 3rd wave of the experiment introduced organic fillers, including SiC and diamond, into Galinstan. In theory, SiC and diamond have very high thermal conductivity[5]. When mixing with Galinstan, SiC had similar results as in Al and Cu power. Diamond mixes better with Galinstan probably thanks to its small particle size (50000 Grit) which is only a few micrometres. The thermal performance of Galinstan with filler diamonds was very poor. An explanation is that metal relies on free-flowing conduction electrons while diamond relies on phonon transfer. The metal-diamond interface produces phonon scattering [6].<br/>So far, adding metals as fillers into Ga or Galinsan is promising. Zinc (Zn) and Bismuth (Bi) are candidates in the next wave of tests.<br/>Later, we can test our best performing filler and add it to Thermal Grizzly Conductonaut, the best performing Liquid Metal thermal paste in the current market.