Search

The maximum heat load that a heat pipe can sustain at steady state is governed by the balance between the capillary pressure provided by the wick and the flow resistance to liquid resupply at the evaporator. At heat loads beyond the capillary limit, the wick will dry out at the evaporator. However, the nature of the imposed heat load may be highly transient in various applications ranging from consumer electronic devices to server processors, depending on the end user needs and workloads. For such scenarios, it becomes critical to assess the operation of heat pipes in response to brief transient heat loads which could be higher than the notional capillary limit that governs dryout at steady state. In the current study, experiments are performed to demonstrate that, under transient heating conditions, a heat pipe can sustain heat loads higher than the steady-state capillary limit for brief periods of time without experiencing dryout. The time-to-dryout during which the heat pipe can sustain a higher heat load is characterized for a step input higher than the steady-state capillary limit. The time-to-dryout is found to decrease as the magnitude of the step heat input increases.

Two-phase jet impingement is a compact cooling technology capable of dissipating the large heat fluxes required for thermal management of high-power electronics devices. It is important to understand the primary heat transfer mechanisms that occur during regimes of jet impingement for which boiling occurs, specifically in the confined impingement geometries common to electronics cooling applications. In this study, heat transfer from a surface is experimentally characterized in both confined jet impingement and pool boiling configurations. The dielectric liquid HFE-7100 is used as the working fluid. For the impingement configuration, the jet issues through a single orifice with a diameter of 2 mm, at exit velocities of 1 m/s and 3.33 m/s, into a confinement gap with an orifice-to-target spacing ratio of 3. Additional orifice-to-target spacings of 0.5 and 5 times the jet diameter are tested at the lower jet velocity. The heat flux applied at the surface was increased in steps to characterize the single-phase and two-phase heat transfer performance; all experiments were carried through to critical heat flux conditions. Over the range of velocities and orifice-to-target spacings tested, the jet impingement data in the fully boiling regime coincide with the pool boiling data. This result indicates, for the range of parameters considered in this study, that nucleate boiling is the dominant heat transfer mechanism in the fully boiling regime in confined jet impingement. The impinging jet velocity and orifice-to-target spacing only influence the single-phase heat transfer and critical heat flux.

Microchannel heat sinks allow removal of dense heat loads from high-power electronic devices at modest chip temperature rises. Such heat sinks are produced primarily using conventional subtractive machining techniques or anisotropic chemical etching, which restricts the geometric features that can be produced. Owing to their layer-by-layer and direct-write approaches, additive manufacturing (AM) technologies enable more design-driven fabrication and offer improved geometric freedom and complexity. There has been limited assessment of the capability of AM process to produce microchannel heat sinks that meet nominal geometry and performance targets, even for conventional designs. Direct metal laser sintering (DMLS) was used to demonstrate additive manufacture of an aluminum alloy (AlSi10Mg) microchannel heat sink with straight, parallel channels (D H = 500 μm). The thermal and hydraulic performance of the heat sink was characterized over a range of mass fluxes from 500 kg/m2s to 2000 kg/m2s using water as the working fluid. The straight microchannel design allows these experimental results to be directly compared against established predictions from the literature. The comparison facilitates an assessment of the role of geometric or property differences introduced by the AM process (relative to conventional fabrication approaches) in realizing the nominal design. The insights gained from this evaluation, which offers an improved description of the fabrication constraints and sets expectations on fidelity to the design targets, will be used in ongoing work that explores heat sink designs exploiting the full range of geometric complexity made possible with additive manufacturing.

Advances in the computational performance of electronic devices have created a clear need for improved methods of passive thermal management. This has led to renewed interest in the use of vapor chambers as heat spreaders in applications ranging from mobile devices to high-performance-computing and power electronics systems. While there has been significant effort to develop vapor chambers for these applications, their designs have largely relied on steady-state analyses and performance prediction. In many applications, however, the heat load is inherently transient in nature. Heat spreader design must consider transient performance in response to these use-case scenarios. While detailed numerical models of transient vapor chamber operation have been developed, a transient modeling approach with low computational cost is needed for parametric study and quick assessment of vapor chamber performance in system-level models. In the current work, a low-cost, transient vapor chamber model is developed targeting the geometries and operating conditions typical of thermal management applications. The model considers mass, momentum, and energy transport in the vapor chamber wall, wick, and vapor core as well as phase change at the wick-vapor interface. The governing equations are simplified to a system of first-order differential equations based on a scaling analysis and assuming a functional form for the temperature profile along the thickness dimension. For an example case, the model is demonstrated to have a computational cost reduction of three to four orders of magnitude as compared to using a finite-volume discretized numerical model, but with comparable levels of accuracy.

The rapid-bubble-growth instability at the onset of flow boiling was studied experimentally in a single, 500 pm-diameter borosilicate glass microchannel subjected to a uniform heat flux of 41.9 kW/m2. Fluid was driven through the microchannel in an open-loop test facility by maintaining a constant pressure difference between a pressurized upstream reservoir and the ambient. Simultaneous measurement of inlet and outlet fluid temperatures and pressures, mass flux, and wall temperature, synchronized with high-speed flow visualization, enabled the underlying mechanisms of the rapid-bubble-growth instability to be quantified. At the onset of boiling, a large liquid superheat (10.3 °C) relative to the outlet saturation temperature causes the nucleated vapor bubble to grow rapidly, expanding in both upstream and downstream directions. This explosive bubble growth results in significant flow reversal and generates a large pressure gradient within the microchannel immediately following boiling incipience. The onset of flow boiling significantly reduced the temperature of the fluid and of the heated wall relative to single-phase flow conditions.

Two-phase jet impingement is a compact cooling approach that provides high-heat-flux dissipation at manageable pressure drop. The heat transfer behavior of an impinging jet array is dependent on a set of geometrical parameters, operating conditions, and fluid properties. In the present study, a semi-empirical approach is developed to predict heat transfer from arrays of jets of liquid that undergoes phase change upon impingement. In the model, the jet array is divided into unit cells centered on each orifice that are assumed to behave identically. The impingement surface in each unit cell is divided into two distinctive regions: a single-phase heat transfer region directly under the jet, and a surrounding boiling heat transfer region along the periphery. Available correlations from the literature are used to estimate the heat transfer coefficient and surface temperature distribution in each region, and the mean surface temperature of the unit cell is estimated via area-averaging. The location of transition to boiling predicted by the model is consistent with prior experimental observations of an inward-creeping boiling front. The model results are first compared against existing experimental data in the literature, and the area-averaged thermal performance is found to be well-predicted. Additional experiments are also performed to evaluate the limits of applicability of the model. The semi-empirical modeling approach developed in this work successfully represents the different heat transfer modes and transitions that occur during two-phase jet impingement.

This work focuses on the fabrication and experimental characterization of a two-phase hierarchical manifold microchannel heat sink array for intrachip high-heat-flux dissipation. A test device with a 5 mm × 5 mm heated area and 9 × 9 array of heat sinks, each with 18 parallel channels (19 µm × 155 µm), is fabricated in silicon. A multi-layer hierarchical manifold fabricated in silicon is bonded to the heat sink array. Flow boiling experiments are conducted using HFE-7100 as the working fluid at mass fluxes of 200 kg/m2s and 300 kg/m2s. The test device is able to dissipate heat fluxes up to 445 W/cm2 at a chip temperature of less than 40 °C above the fluid and at a pressure drop less than 80 kPa. A maximum heat transfer coefficient of 31,900 W/m2K occurred at a mass flux of 300 kg/m2s and a heat flux of 301 W/cm2. The effects of heat flux on chip temperature, heat transfer coefficient, and pressure drop are investigated.

We formulate an analytical model to predict simultaneous condensation and melting on a vertical ice surface in air for a range of typical ambient conditions. The model solves the governing mass, momentum, and energy transport equations (including inertial terms) in the liquid film and vapor-gas domains. Transport of the water vapor species in noncondensable air is considered in the vapor-gas domain. The governing equations are simplified to a series of one-dimensional ordinary differential equations through the use of stream functions and similarity variables. A novel normalization of the liquid transport equations accounts for the unknown film thickness boundary condition. The equations are solved using a boundary value problem solver. The model is validated through qualitative comparisons to the behavior predicted by previous studies. Furthermore, an experimental validation is conducted using a laser displacement sensor to measure the liquid film thickness on an ice sheet. An interrogation of the spatial trends in the condensation and melting fluxes demonstrates that condensation is the predominant driving mechanism underlying the melting process, rather than free convection. A parametric study shows that the melting rate depends heavily on the water vapor content of the air.

A hierarchical manifold microchannel heat sink is used to dissipate heat from a small hotspot region superposed on a larger region of uniform background heat flux. A 5 mm × 5 mm overall chip footprint area is cooled using a 3 × 3 array of intrachip silicon microchannel heat sinks fed in parallel using a manifold distributor. Each heat sink consists of a bank of 25 high-aspect-ratio microchannels that are nominally 30 µm wide and 300 µm deep. The uniform background heat flux is generated with a 3 × 3 array of thin-film heaters fabricated on the chip; temperature sensors placed in each of these nine heating zones provide spatially resolved chip surface temperature measurements. An individually powered 200 µm × 200 µm hotspot heater is centered on the chip. The heat sink thermal and hydraulic performance is evaluated using HFE-7100 as the working fluid and for mass fluxes ranging from 600 kg/m2s to 2070 kg/m2s at a constant inlet temperature of 60 °C and outlet pressure of 122 kPa. Background heat fluxes up to 450 W/cm2 and hotspot fluxes of greater than 2500 W/cm2 are simultaneously dissipated. The chip temperature uniformity and maximum temperature rise during hotspot heating are assessed. For the case with the highest simultaneous background and hotspot heat fluxes, the measured heat sink pressure drop is ∼75 kPa and the average chip temperature is ∼30 °C above the fluid inlet temperature.

While academic research has made great advances over the last few decades in specific thermal management technologies, the transfer of ideas from the lab to products in industry has been less successful because of impedance at a variety of interfaces, real and virtual. To some extent, going from specific component-level advances to system-level implementation is challenged by the lack of more intimate knowledge among researchers of the product buildup, and where relevant, the supply chain as well. On the one hand, concerns related to intellectual property and confidentiality have hampered information-sharing and prevented clear, quantitative and comprehensive problem definition. But on the other, academics have often not invested the time and effort to imbed themselves into industry groups and learn about the packaging and manufacturing challenges, and marketing and competitive pressures. Much real engineering - and even science - must go into addressing these challenges, but the problems involved are not clean or well-defined and are certainly not demarcated by discipline. They must instead be teased out with patience. Arguably, this interface impedance between academic research and industrial practice has been less pronounced on the electrical engineering side of the house. Perhaps it is finally time for thermal management to not be treated as the perpetual afterthought by this industry - this will need inspired leadership in our leading companies, and also strategic stewardship by the thermal community.