Yoonjin Won

Phase purity and the thermoelectric properties of Ge2Sb2Te5 films down to 25 nm thickness

Thermoelectric phenomena strongly influence the behavior of chalcogenide materials in nanoelectronic devices including phase-change memory cells. This work uses a novel silicon-on-insulator experimental structure to measure the phase and temperature-dependent Seebeck and Thomson coefficients of Ge2Sb2Te5 films including the first data for films of thickness down to 25 nm. The Ge2Sb2Te5 films annealed at different temperatures contain varying fractions of the amorphous and crystalline phases which strongly influence the thermoelectric properties. The Seebeck coefficient reduces from 371 μV/K to 206 μV/K as the crystalline fraction increases by a factor of four as quantified using x-ray diffraction. The data are consistent with modeling based on effective medium theory and suggest that careful consideration of phase purity is needed to account for thermoelectric transport in phase change memory.

Cooling Limits for GaN HEMT Technology

The peak power density of GaN HEMT technology is limited by a hierarchy of thermal resistances from the junction to the ambient. Here we explore the ultimate or fundamental cooling limits made possible by advanced thermal management technologies including GaN-diamond composites and nanoengineered heat sinks. Through continued attention to near-junction resistances and extreme flux convection, power densities that may exceed 50 kW/cm2 - depending on gate width and hotspot dimension - are feasible within 5 years.

Zipping, entanglement, and the elastic modulus of aligned single-walled carbon nanotube films

Significance Aligned carbon nanotube films promise the unusual combination of high thermal conductivity and mechanical compliance. Here, the mechanical compliance of single-walled nanotube films has been measured and linked to their morphology and microscopic motions, including zipping, unzipping, and entanglement. The physical mechanisms governing the mechanical response include bending forces or van der Waals interactions, with the dominant mechanism depending on the nanotube density and alignment. The dependence of film morphology on mechanical modulus explored here provides the foundation for modeling of a variety of other properties including thermal and electrical conductivity. Reliably routing heat to and from conversion materials is a daunting challenge for a variety of innovative energy technologies––from thermal solar to automotive waste heat recovery systems––whose efficiencies degrade due to massive thermomechanical stresses at interfaces. This problem may soon be addressed by adhesives based on vertically aligned carbon nanotubes, which promise the revolutionary combination of high through-plane thermal conductivity and vanishing in-plane mechanical stiffness. Here, we report the data for the in-plane modulus of aligned single-walled carbon nanotube films using a microfabricated resonator method. Molecular simulations and electron microscopy identify the nanoscale mechanisms responsible for this property. The zipping and unzipping of adjacent nanotubes and the degree of alignment and entanglement are shown to govern the spatially varying local modulus, thereby providing the route to engineered materials with outstanding combinations of mechanical and thermal properties.

Phase and thickness dependent modulus of Ge2Sb2Te5 films down to 25 nm thickness

The mechanical properties of phase-change materials including Ge2Sb2Te5 (GST) are strongly influenced by the complex interaction of phase and imperfection distributions, especially at film thicknesses relevant for phase-change memory devices. This work uses a micromechanical resonator as a substrate to study the phase dependent modulus of GST films with thicknesses from 25 nm to 350 nm. The moduli of amorphous GST and crystalline GST films increase with decreasing thickness to 10 GPa and up to 60 GPa, respectively. The phase purity is studied using X-ray diffraction and energy dissipation data, which provide qualitative information about inelastic absorption.

Fundamental Cooling Limits for High Power Density Gallium Nitride Electronics

The peak power density of GaN high-electron-mobility transistor technology is limited by a hierarchy of thermal resistances from the junction to the ambient. Here, we explore the ultimate or fundamental cooling limits for junction-to fluid cooling, which are enabled by advanced thermal management technologies, including GaN-diamond composites and nanoengineered heat sinks. Through continued attention to near-junction resistances and extreme flux convection heat sinks, heat fluxes beyond 300 kW/cm2 from individual 2-μm gates and 10 kW/cm2 from the transistor footprint will be feasible. The cooling technologies under discussion here are also applicable to thermal management of 2.5-D and 3-D logic circuits at lower heat fluxes.

Thermal Modeling of Extreme Heat Flux Microchannel Coolers for GaN-on-SiC Semiconductor Devices

Gallium nitride (GaN) high-electron-mobility transistors (HEMTs) dissipate high power densities which generate hotspots and cause thermomechanical problems. Here, we propose and simulate GaN-based HEMT technologies that can remove power densities exceeding 30 kW/cm2 at relatively low mass flow rate and pressure drop. Thermal performance of the microcooler module is investigated by modeling both single- and two-phase flow conditions. A reduced-order modeling approach, based on an extensive literature review, is used to predict the appropriate range of heat transfer coefficients associated with the flow regimes for the flow conditions. Finite element simulations are performed to investigate the temperature distribution from GaN to parallel microchannels of the microcooler. Single- and two-phase conjugate computational fluid dynamics (CFD) simulations provide a lower bound of the total flow resistance in the microcooler as well as overall thermal resistance from GaN HEMT to working fluid. A parametric study is performed to optimize the thermal performance of the microcooler. The modeling results provide detailed flow conditions for the microcooler in order to investigate the required range of heat transfer coefficients for removal of heat fluxes up to 30 kW/cm2 and a junction temperature maintained below 250 °C. The detailed modeling results include local temperature and velocity fields in the microcooler module, which can help in identifying the approximate locations of the maximum velocity and recirculation regions that are susceptible to dryout conditions.

Numerical Simulation of Advanced Monolithic Microcooler Designs for High Heat Flux Microelectronics

This study considers CFD simulations with conjugate heat transfer performed in the framework of designing a complex micro-scale cooling geometry. The numerical investigation of the three-dimensional, laminar flow (Reynolds number smaller than 480) and the solid conduction is done on a reduced model of the heat sink micro-structure to enable exploring a variety of configurations at a limited computational cost. The reduced model represents a unit-cell, and uses periodic and symmetry boundary conditions to mimic the conditions in the entire cooling manifold. A simulation of the entire heat sink micro-structure was performed to verify the unit-cell set-up, and the comparison demonstrated that the unit-cell simulations allow reducing the computational cost by two orders of magnitude while retaining accurate results. The baseline design for the unit-cell represents a configuration used in traditional electronic heat sinks, i.e. a simple channel geometry with a rectangular cross section, with a diameter of 50 μm, where the fluid flows between two cooling fins. Subsequently three types of modified geometries with feature sizes of 50 μm were considered: baffled geometries that guide the flow towards the hotspot region, geometries where the fins are connected by crossbars, and a woodpile structure without cooling fins. Three different mass-flow rates were tested. Based on the medium mass-flow rate considered, the woodpile geometry showed the highest heat transfer coefficient with an increase of 70% compared to the baseline geometry, but at the cost of increasing the pressure drop by more than 300%. The crossbar geometries were shown to be promising configurations, with increases in the heat transfer coefficient of more than 20% for a 70% increase in pressure drop. The potential for further optimization of the crossbar configurations by adding or removing individual crossbars will be investigated in a follow up study. The results presented demonstrate the increase in performance that can be obtained by investigating a variety of designs for single phase cooling devices using unit-cell conjugate heat transfer simulations.Copyright

Reactive Metal Bonding of Carbon Nanotube Arrays for Thermal Interface Applications

Vertically aligned carbon nanotube (CNT) arrays can offer an attractive combination of high thermal conductance and mechanical compliance for thermal interface applications. These arrays require a reliable, thermally conductive bonding technique to enable integration into devices. This paper examines the use of a reactive metal bonding layer to attach and transfer CNT arrays to metal-coated substrates, and the thermal performance is compared with CNT arrays bonded with indium solder. Infrared microscopy is used to simultaneously measure the intrinsic thermal conductivity of the CNT array and the thermal boundary resistance of both the bonded and growth CNT interfaces over a range of applied compressive stresses. A coarse-grained molecular simulation is used to model the effects of compressive pressure on the CNT array thermal conductivity. Reactive metal bonding reduces the thermal boundary resistance to as low as 27 mm2 · K · W-1, which is more than an order of magnitude less than the nonbonded contact.

Inverse opals for fluid delivery in electronics cooling systems

We report the fabrication and fluid flow characterization of a class of open-cell copper foams known as copper inverse opals (CIOs). This material has finely controlled structure at the pore level, which may enable its use in microscale heat exchangers for microelectronics cooling. We fabricated CIOs by electrodepositing copper around a sacrificial template of packed polystyrene microspheres. We then removed the CIOs from their substrates and used electroetching to vary the pore structure and porosity. We characterized the geometry of the samples at various stages of fabrication with visual inspection and image analysis of scanning electron micrographs. We characterized the permeability with a through-plane flow rig and developed computational models for fluid flow in ideal face-centered cubic and hexagonally close-packed unit cells. Here we report the simulated and experimentally measured values of permeability. We also report experimental challenges that arise from the microscale dimensions of the samples.

Crust removal and effective modulus of aligned multi-walled carbon nanotube films

Carbon nanotubes (CNTs) have been attractive materials because of the unique combination of their small size and physical properties, such as high thermal conductivity and mechanical compliance. This paper extracts in-plane modulus of 100-220 μm-thick vertically aligned multi-walled carbon nanotube (VA-MWCNT) films. The films have low modulus in the range of 0.5-2 MPa, consistent with expectations due to the direction of CNT alignment. We study the effect of a top crust of entangled CNTs on the modulus by etching the surface of the VA-MWCNT films using O2 plasma. Scanning electron micrographs reveal that the surface etching removes the entanglements of the crust layer. The modulus values of the etched samples indicate that there is no significant effect of this crust on the modulus of thick films (>;100 μm). This is in contrast to previous work that showed that the crust layer had a very strong effect in thinner films.