Abstract Preview

Here is the abstract you requested from the Thermal_2010 technical program page. This is the original abstract submitted by the author. Any changes to the technical content of the final manuscript published by IMAPS or the presentation that is given during the event is done by the author, not IMAPS.

CPU Spray Cooling Analysis
Keywords: spray cooling, CFD, analysis
Direct spray cooling on the inactive side of a CPU in a flip chip, lidless package was characterized through computational fluid dynamic (CFD) analysis and through test work. Commercial CFD software, Flow 3D, was used to carry out the analysis. A spray head containing multiple nozzles was attached to the backside of the CPU, and dielectric fluid, HFE-7100, was issued through the nozzles, impinging the exposed silicon. The size of the die was 20mm x 20mm. A high speed camera (6000 frames per second) was used to measure the spray droplet diameters, droplet speed, and the droplet surface tension via measuring the angle between the silicon surface and the region of the droplet that is in contact with the silicon surface. The spray coolant flow rate was measured, and this in addition to the measured droplet diameter and speed were used to establish the initial boundary conditions to conduct a design of experiments CFD analysis. The impact on the overall heat transfer convection coefficient and the critical heat flux by varying the coolant flow rate, the droplet diameter and speed, droplet distribution, droplet impact frequency, the droplet thermodynamic state at the nozzle, wall roughness, wall orientation, and heat flux rate applied to the active side of silicon. The above analysis was carried for out for the ideal “open condition” and for an “enclosed condition” in order to simulate spray head droplet formation and flow effects under different confinement conditions. The Flow-3D CFD analysis consisted of modeling a large cloud of vertically translating, individual spherical fluid droplets with given thermophysical properties whose diameter conformed to an optically measured size from test data (diameters in the multi-micron range) and computing the fluid impact dynamics with a heated die surface of unit area at given wall initial temperature and heat flux. The number of droplets at a given fluid density passing through a unit area at a given velocity then provided the mass flow rate to the die surface. The impacting droplets over time underwent various degrees of splashing, fragmentation, coalescence into a chaotic free surface fluid layer shape with evaporation at the free fluid surface (with the computed specific humidity affecting evaporation rate) while interacting with the heated silicon die surface. Heat transfer from the die surface into the rapidly changing shape of the fluid layer due to multiple spaced droplet impacts resulted from evaporative latent heat effects as well as conduction and convection effects. During high speed droplet impact with the transient surface fluid layer, the creation of “bare spots” occasionally opened up in the fluid layer- in this case, the CFD analysis allowed for conductive and convective heat transfer at the vapor/gas surface interface to the die surface. Fluid viscosity, surface tension and a k-ε turbulence model were assumed to compute the transient free surface fluid layer shape over a given time for various droplet sizes and mass flow rates to develop the statistics necessary to compute an effective heat transfer coefficient for the spray cooling process.
Charles Ortloff,
CFD Consultants International
Los Gatos, CA

  • Amkor
  • ASE
  • Canon
  • EMD Performance Materials
  • Honeywell
  • Indium
  • Kester
  • Kyocera America
  • Master Bond
  • Micro Systems Technologies
  • MRSI
  • Palomar
  • Plexus
  • Promex
  • Qualcomm
  • Quik-Pak
  • Raytheon
  • Specialty Coating Systems