Design and evaluation of heat transfer fluids for direct immersion cooling of electronic systems.

摘要

The development of products that have desirable properties is an important goal of chemical product design. In general, new chemical products are identified empirically, based on the potential market for that product and the experience and insight of the designer. This approach is therefore limited by the designer's experience. By contrast, computer-aided molecular design (CAMD) is used in this work to conduct a systematic and exhaustive search of molecular structures and to generate a large number of feasible candidates for a specific application.;The application of interest in the present work is direct immersion phase change cooling of electronic systems. This interest stems from the need to find coolants that can meet the increasing thermal management demands that arise from miniaturization of electronics. Novel heat transfer fluids for direct immersion cooling of electronics were therefore identified using CAMD. In addition, the feasibility of improving existing heat transfer fluids by dispersing small amounts of high thermal conductivity solid nanoparticles in them was also investigated.;Since the CAMD approach requires reliable property estimation methods to screen candidates, group contribution (GC) methods for thermophysical properties relevant to heat transfer were critically evaluated using thermophysical property data for over 150 organic compounds. It was found that the predictive capabilities of the GC methods were inadequate for organosilicon compounds. Therefore, new GC values were developed for organosilicon compounds and are presented in this work.;The molecules generated by the CAMD algorithm were constrained by limiting their boiling points, enthalpy of vaporization, and thermal conductivity values. The candidates were screened further using a figure of merit (FOM) analysis for pool and flow boiling. A total of 52 compounds were identified after this analysis. From these 52 fluids, 9 fluids were selected for experimental evaluation based on commercial availability and the potential of their synthesis. Two of these fluids (1,1,1-trifluoro-3- methylpentane and 1,1,1-trifluoro-3-(2,2,2-trifluoroethoxy)propane) were synthesized in this work and the remaining 7 were purchased from commercial vendors.;The density, viscosity, and thermal conductivity of the 9 fluids were measured and these values were employed in the validation of the GC methods used in the CAMD approach. Pool boiling heat transfer studies demonstrated that the new fluids possess heat transfer properties that are superior to those of HFE 7200. As most of the new fluids contained fluorine, their environmental properties were also evaluated. A new GC method was developed for radiative forcing (RF) and validated with FT-IR based calculations. RF predictions were then used to calculate the global warming potentials (GWP) of the new fluids. The GWP of new fluids were found to be significantly lower than those of currently used coolants.;The second approach examined for the development of new coolants was the addition of dispersed nanoparticles to existing coolants to enhance thermal conductivity. Since there is considerable disagreement in the literature with respect to the mechanism of heat transfer in nanoparticle dispersions (or nanofluids), a critical review of experimental data and models for the thermal conductivity of nanofluids was conducted with particular emphasis on the effects of particle size. A modified geometric mean model was developed that takes into account the temperature dependence of the thermal conductivities of the individual phases, as well as the size dependence of the thermal conductivity of the dispersed phase. The rheological properties of nanofluids were also experimentally measured, and the effects of particle concentration, temperature, and shear rate on nanofluid viscosity were evaluated. The viscosity of nanofluids was found to increase by 2 orders of magnitude, while the thermal conductivity increase was found to be only 25 - 30 %. This increase in viscosity when particles are added to liquids suggests that this is not a feasible approach to undertake in order to improve existing coolants for electronics.