Abstract Preview

Here is the abstract you requested from the Thermal_2017 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.

A Combined Thermal-LIF and PIV Approach for Getting to the Root of the Problem: How Much Vapor does an Immersed High Heat Flux Chip Generate?
Keywords: high heat flux chip, vapor, thermal-lif
Within the last decade, several advanced liquid cooling strategies have been explored, with emphasis on those taking advantage of the latent heat removed during liquid phase change and the vigorous convective heat transfer occurring at the liquid/vapor interface of the ebullition process. Pool boiling has the potential to reduce the form factor of many energy dense systems due to the substantial increase this thermal management technique offers over many more conventional cooling solutions [1]. Subcooling the ambient fluid around the boiling process [2], microstructured surfaces [3], and improved condensing surfaces with two-phase flow loops [4] have all been shown to improve the volumetric power density and lower the operating temperature of high heat flux components. What is often not examined is the fact that the improved heat transfer coefficients in pool boiling are artifacts of two equally important mechanisms, namely convective and latent heat transfer. The two mechanisms are interdependent as the amount of vapor generated at the surface drives the cyclical movement of fluid at the liquid/vapor interface, and the nature of the bubble departure (frequency, size, proximity to adjacent sites, contact angle) can harm or improve the heat transfer. This work shows that the amount of fluid entering the boiling process can be quantified using two-phase Particle Image Velocimetry (PIV), a 3D vector set of quenching coolant flow paths can be acquired, and the temperature profile of liquid surrounding the boiling process can be acquired through two-color thermal-LIF techniques. While this work focuses on ebullition on the macro (chip-scale) and micro (individual bubble-scale), the end goal is the same. Practically speaking, the more heat transfer that you can get with reduced quenching fluid requirements, the more efficient the overall system will be as it is much harder to condense vapor than to extract heat from the liquid phase. This work is limited to two-color thermal-LIF and fluorescent PIV on bare surfaces with PF-5060 as the working fluid, but net coolant requirements acquired from this work compared against that predicted with Csf parameters in the Rohsenow Correlation prove its applicability to any chip-scale surface/fluid combination. Calculating the hydrodynamic and thermal profiles around rising vapor will benefit predictive modelling of the performance and reliability of several commercially available high heat flux systems. Results show the expected condition that 10-20% of the heat transfer generated on a bare silicon surface comes from latent heat while the rest is transmitted through convective heat transfer at the microlayer vapor/surface interface. Vapor generation expectations for a given surface can improve performance/dryout modelling of terrestrial [5] and reduced-gravity [6] thermosyphons. Chip-scale experiments on microstructured surfaces will also be able to address dryout conditions on extremely high performance thermosyphons using microporous Boiling Enhancement Coatings (BEC’s) [5]. These microstructured thermosyphons have already shown an order of magnitude improvement in heat transfer over a conventional system [5]. Further assessment using this experimental technique is critical as previous work performed at both steady-state and transient conditions [7-8] has shown that the heat transfer coefficient can be increased up to 4 times and the global thermal resistance significantly reduced when using surfaces structured with micro-cavities, with fixed dimensions, by varying the distance between the cavities. Preliminary results, combining high-speed visualization with Particle Image Velocimetry (PIV) suggest that the distance between the cavities should be adjusted to balance the positive effects of increasing the number of active nucleation sites with the negative effects of bubble interaction mechanisms which tend to slow the departure frequency and form large vapour bubbles which isolate the surface. Passive Immersion Modules (PIMs) are largely condensation limited [9] meaning that understanding the amount and rate of vapor generation are the limiting factor in making direct two-phase cooling of CPU’s an even more attractive thermal management option. Previous work has shown that volumetric power densities of two-phase immersion cooled modules are 4 times that of air-cooled systems with vapor generation from a bare surface and seven times when microporous coatings are used [10]. For this previous work [10], the condensive area to heated area ratio was 12 and only half of the Critical Heat Flux (CHF) was reached before the condensation-limited burnout occurred. Optimization of fill ratios could have been used to increase the performance of this system even further, but where that point is directly dependent on the amount of vapor generated at the surface as well as the degree to which the rising vapor condensed before contact with the internal condenser. The latter could be solved with the thermal-LIF portion of this experimental technique, while the former could be solved with the fluorescent PIV technique presented in this work. Higher volumetric power densities in the data center will lead to less overhead and smaller server rooms.
Matt Spot Harrison, PhD Candidate Thermal Fluid Science
Oregon State University
, OR

  • 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