Microchannels and micro pin fins have been employed for almost four decades in the cooling of microelectronic devices and research is still being done in this field to improve the thermal performance of these micro heat sinks. In this research, the constructal design and computational fluid dynamics code was used with a goal-driven optimisation tool to numerically investigate the thermal performance of a novel design of combining microchannels and micro pin fins for microelectronic cooling applications. Existing designs of microchannels were first optimised and thereafter, three to seven rows of micro pin fins were inserted into the microchannels to investigate whether there was further improvement in thermal performance. The microchannels and micro pin fins were both embedded in a highly conductive solid substrate. three-dimensional geometric structure of the combined micro heat sink was optimised to achieve the objective of maximised thermal conductance, which is also minimised thermal resistance under various design conditions. The micro heat sinks investigated in the study were the single microchannel, two-layered microchannels with parallel and counter flow configurations, three-layered microchannels with parallel and counter flow configurations, the single microchannel with circular-, square- and hexagonal-shaped micro pin-fin inserts and the two-layered microchannels with circular-shaped micro pin-fin inserts. A numerical computational fluid dynamics (CFD) package with a goal-driven optimisation tool, which employs the finite-volume method, was used to analyse the fluid flow and heat transfer in the micro heat sinks investigated in this work. The thermal performances of all the micro heat sinks were compared for different application scenarios. Furthermore, the temperature variation on the heated base of the solid substrate was studied for the different micro heat sinks to investigate which of the heat sink designs minimised the temperature rise on the heated base best. This is very important in microelectronic cooling applications because temperature rise affects the reliability of the device. The heat sink design that best maximised thermal conductance and minimised temperature rise on the heated base was chosen as the best for microelectronic cooling. For all the cases considered, fixed volume constraints and manufacturing constraints were applied to ensure real-life applicability. It was concluded that optimal heat sink design for different application scenarios could be obtained speedily when a CFD package which had an optimisation tool was used. / Thesis (PhD)--University of Pretoria, 2016. / Mechanical and Aeronautical Engineering / PhD / Unrestricted

The continual advance in manufacturing processes has resulted in significantly more compact, high performance, devices. Consequently, heat extraction has become the limiting factor, and of primary concern. Therefore, a substantial amount of research has been done regarding high efficiency micro heat exchangers, employing novel working fluids. This dissertation numerically investigated the thermal behaviour of microchannel elements cooled by Newtonian and non-Newtonian fluids, with the objective of maximising thermal conductance subject to constraints. This was done, firstly, for a two-dimensional simple microchannel, and secondly, for a three-dimensional complex microchannel. A numerical model was used to solve the governing equations relating to the flow and temperature fields for both cases. The geometric configuration of each cooling channel was optimised for Newtonian and non-Newtonian fluids, at a fixed inlet velocity and heat transfer rate. In addition, the effect of porosity on thermal conductance was investigated. Geometric optimisation was employed to the simple and complex microchannels, whereby an optimal geometric ratio (height versus length) was found to maximise thermal conductance. Moreover, analysis indicated that the bifurcation point of the complex microchannel could be manipulated to achieve a higher thermal conductance. In both cases, it was found that the non-Newtonian fluid characteristics resulted in a significant variation in thermal conductance as inlet velocity was increased. The ii characteristics of a dilatant fluid greatly reduced thermal conductance on account of shear-thickening on the boundary surface. In contrast, a pseudoplastic fluid showed increased thermal conductance. A comparison of the simple and complex microchannel showed an improved thermal conductance resulting from greater flow access to the conductive area, achieved by the complex microchannel. Therefore, it could be concluded that a complex microchannel, in combination with a pseudoplastic working fluid, substantially increased the thermal conductance and efficiency, as opposed to a conventional methodology. / Dissertation (MEng)--University of Pretoria, 2012. / gm2014 / Mechanical and Aeronautical Engineering / unrestricted

Controlling the temperature of the exhaust valves is paramount for proper functioning of engines and for the long lifespan of valve train components. The majority of the heat outflow from the valve takes place along the valve-seat-cylinder head-coolant thermal path which is significantly influenced by the thermal contact resistance (TCR) present at the valve/seat and seat/head interfaces. A test rig facility and experimental procedure were successfully developed to assess the effect of the valve/seat and seat/head interfaces on heat outflow from the valve, in particular the effects of the valve/seat interface geometry, seat insert assembly method, i.e. press or shrink fit, and seat insert metallic coating on the operating temperature of the valve. The results of tests have shown that the degree of the valve-seat geometric conformity is more significant than the thermal conductivity of the insert: for low conforming assemblies, the mean valve head temperature recorded during tests on copper-infiltrated insert seats was higher than that recorded during tests on noninfiltrated seats of higher conformance. The effect of the insert-cylinder head assembly method, i.e. shrink-fitted versus press-fitted inserts, has proved negligible: results have shown insignificant valve head temperature variations, for both tin-coated and uncoated inserts. On the other hand, coating the seat inserts with a layer of tin (20-22μm) reduced the mean valve head temperature by approximately 15°C as measured during tests on uncoated seats. The analysis of the valve/seat and seat/head interfaces has indicated that the surface asperities of the softer metal in contact would undergo plastic deformation. Suitable thermal contact conductance (TCC) models, available in the public domain, were used to evaluate the conductance for the valve/seat and seat/cylinder head interfaces. Finally, a FE thermal model of the test rig has been developed with a view to assess the quality of the calculated TCC values for the valve/seat and seat/head interfaces. The results of the thermal analysis have shown that predicted temperatures at chosen control points agree with those measured during tests on thermometric seats with an acceptable level of accuracy, proving the effectiveness of the used TCC models.

Controlling the temperature of the exhaust valves is paramount for proper functioning of engines and for the long lifespan of valve train components. The majority of the heat outflow from the valve takes place along the valve-seat-cylinder head-coolant thermal path which is significantly influenced by the thermal contact resistance (TCR) present at the valve/seat and seat/head interfaces. A test rig facility and experimental procedure were successfully developed to assess the effect of the valve/seat and seat/head interfaces on heat outflow from the valve, in particular the effects of the valve/seat interface geometry, seat insert assembly method, i.e. press or shrink fit, and seat insert metallic coating on the operating temperature of the valve. The results of tests have shown that the degree of the valve-seat geometric conformity is more significant than the thermal conductivity of the insert: for low conforming assemblies, the mean valve head temperature recorded during tests on copper-infiltrated insert seats was higher than that recorded during tests on noninfiltrated seats of higher conformance. The effect of the insert-cylinder head assembly method, i.e. shrink-fitted versus press-fitted inserts, has proved negligible: results have shown insignificant valve head temperature variations, for both tin-coated and uncoated inserts. On the other hand, coating the seat inserts with a layer of tin (20-22¿m) reduced the mean valve head temperature by approximately 15°C as measured during tests on uncoated seats. The analysis of the valve/seat and seat/head interfaces has indicated that the surface asperities of the softer metal in contact would undergo plastic deformation. Suitable thermal contact conductance (TCC) models, available in the public domain, were used to evaluate the conductance for the valve/seat and seat/cylinder head interfaces. Finally, a FE thermal model of the test rig has been developed with a view to assess the quality of the calculated TCC values for the valve/seat and seat/head interfaces. The results of the thermal analysis have shown that predicted temperatures at chosen control points agree with those measured during tests on thermometric seats with an acceptable level of accuracy, proving the effectiveness of the used TCC models.

As a result of manufacturing processes, real surfaces have roughness and surface curvature. The real contact occurs only over microscopic contacts, which are typically only a few percent of the apparent contact area. Because of the surface curvature of contacting bodies, the macrocontact area is formed, the area where microcontacts are distributed randomly. The heat flow must pass through the macrocontact and then microcontacts to transfer from one body to another. This phenomenon leads to a relatively high temperature drop across the interface. Thermal contact resistance (TCR) is a complex interdisciplinary problem, which includes geometrical, mechanical, and thermal analyses. Each part includes a micro and a macro scale sub-problem. Analytical, experimental, and numerical models have been developed to predict TCR since the 1930's. Through comparison with more than 400 experimental data points, it is shown that the existing models are applicable only to the limiting cases and none of them covers the general non-conforming rough contact. The objective of this study is to develop a compact analytical model for predicting TCR for the entire range of non-conforming contacts, i. e. , from conforming rough to smooth sphere-flat in a vacuum. The contact mechanics of the joint must be known prior to solving the thermal problem. A new mechanical model is developed for spherical rough contacts. The deformation modes of the surface asperities and the bulk material of contacting bodies are assumed to be plastic and elastic, respectively. A closed set of governing relationships is derived. An algorithm and a computer code are developed to solve the relationships numerically. Applying Buckingham Pi theorem, the independent non-dimensional parameters that describe the contact problem are specified. A general pressure distribution is proposed that covers the entire spherical rough contacts, including the Hertzian smooth contact. Simple correlations are proposed for the general pressure distribution and the radius of the macrocontact area, as functions of the non-dimensional parameters. These correlations are compared with experimental data collected by others and good agreement is observed. Also a criterion is proposed to identify the flat surface, where the influence of surface curvature on the contact pressure is negligible. Thermal contact resistance is considered as the superposition of macro and micro thermal components. The flux tube geometry is chosen as the basic element in the thermal analysis of microcontacts. Simple expressions for determining TCR of non-conforming rough joints are derived which cover the entire range of TCR by using the general pressure distribution and the flux tube solution. A complete parametric study is performed; it is seen that there is a value of surface roughness that minimizes TCR. The thermal model is verified with more than 600 data points, collected by many researchers during the last 40 years, and good agreement is observed. A new approach is taken to study the thermal joint resistance. A novel model is developed for predicting the TCR of conforming rough contacts employing scale analysis methods. It is shown that the microcontacts can be modeled as heat sources on a half-space for engineering applications. The scale analysis model is extended to predict TCR over the entire range of non-conforming rough contacts by using the general pressure distribution developed in the mechanical model. It is shown that the surface curvature and contact pressure distribution have no effect on the effective micro thermal resistance. A new non-dimensional parameter is introduced as a criterion to identify the three regions of TCR, i. e. , the conforming rough, the smooth spherical, and the transition regions. An experimental program is designed and data points are collected for spherical rough contacts in a vacuum. The radius of curvature of the tested specimens are relatively large (in the order of m) and can not be seen by the naked eye. However, even at relatively large applied loads the measured joint resistance (the macro thermal component) is still large which shows the importance of surface out-of-flatness/curvature. Collected data are compared with the scale analysis model and excellent agreement is observed. The maximum relative difference between the model and the collected data is 6. 8 percent and the relative RMS difference is approximately 4 percent. Additionally, the proposed scale analysis model is compared/verified with more than 880 TCR data points collected by many researchers. These data cover a wide range of materials, surface characteristics, thermal and mechanical properties, mean joint temperature, directional heat transfer effect, and contact between dissimilar metals. The RMS difference between the model and all data is less than 13. 8 percent.

Individual carbon nanotubes (CNTs) have superior thermal conductivity than conventional materials. The applications for CNTs range from heat sinks, thin films to thermal interface materials. However, when CNTs are grouped together in macroscopic quantities and embedded in different media their thermal conductivity changes. Therefore, it is important to determine the thermal conductance changes when CNTs are embedded in different media. In my research, CNTs were embedded in thin films and as aligned structures (fins) in water. Analytical and experimental methods were used to determine the thermal conductances of these aligned structures and thin films. The primary goals of this research were to develop novel analytical methods to determine thermal conductivity and also experimental techniques to determine effectiveness of the embedded CNTs as carriers of heat by thermal conductance evaluation. It is observed that CNTs fins are effective carriers of heat and result in up to 57% decrease in thermal resistance. In the case of CNTs embedded in thin films, it is important to consider non Fourier effects and neglecting non Fourier effects would lead to an underestimation of the thermal conductivity. In addition to the thermal conductivity value, the analysis also provides a way to determine the thermal relaxation time of thin films.

As a result of manufacturing processes, real surfaces have roughness and surface curvature. The real contact occurs only over microscopic contacts, which are typically only a few percent of the apparent contact area. Because of the surface curvature of contacting bodies, the macrocontact area is formed, the area where microcontacts are distributed randomly. The heat flow must pass through the macrocontact and then microcontacts to transfer from one body to another. This phenomenon leads to a relatively high temperature drop across the interface. Thermal contact resistance (TCR) is a complex interdisciplinary problem, which includes geometrical, mechanical, and thermal analyses. Each part includes a micro and a macro scale sub-problem. Analytical, experimental, and numerical models have been developed to predict TCR since the 1930's. Through comparison with more than 400 experimental data points, it is shown that the existing models are applicable only to the limiting cases and none of them covers the general non-conforming rough contact. The objective of this study is to develop a compact analytical model for predicting TCR for the entire range of non-conforming contacts, i. e. , from conforming rough to smooth sphere-flat in a vacuum. The contact mechanics of the joint must be known prior to solving the thermal problem. A new mechanical model is developed for spherical rough contacts. The deformation modes of the surface asperities and the bulk material of contacting bodies are assumed to be plastic and elastic, respectively. A closed set of governing relationships is derived. An algorithm and a computer code are developed to solve the relationships numerically. Applying Buckingham Pi theorem, the independent non-dimensional parameters that describe the contact problem are specified. A general pressure distribution is proposed that covers the entire spherical rough contacts, including the Hertzian smooth contact. Simple correlations are proposed for the general pressure distribution and the radius of the macrocontact area, as functions of the non-dimensional parameters. These correlations are compared with experimental data collected by others and good agreement is observed. Also a criterion is proposed to identify the flat surface, where the influence of surface curvature on the contact pressure is negligible. Thermal contact resistance is considered as the superposition of macro and micro thermal components. The flux tube geometry is chosen as the basic element in the thermal analysis of microcontacts. Simple expressions for determining TCR of non-conforming rough joints are derived which cover the entire range of TCR by using the general pressure distribution and the flux tube solution. A complete parametric study is performed; it is seen that there is a value of surface roughness that minimizes TCR. The thermal model is verified with more than 600 data points, collected by many researchers during the last 40 years, and good agreement is observed. A new approach is taken to study the thermal joint resistance. A novel model is developed for predicting the TCR of conforming rough contacts employing scale analysis methods. It is shown that the microcontacts can be modeled as heat sources on a half-space for engineering applications. The scale analysis model is extended to predict TCR over the entire range of non-conforming rough contacts by using the general pressure distribution developed in the mechanical model. It is shown that the surface curvature and contact pressure distribution have no effect on the effective micro thermal resistance. A new non-dimensional parameter is introduced as a criterion to identify the three regions of TCR, i. e. , the conforming rough, the smooth spherical, and the transition regions. An experimental program is designed and data points are collected for spherical rough contacts in a vacuum. The radius of curvature of the tested specimens are relatively large (in the order of m) and can not be seen by the naked eye. However, even at relatively large applied loads the measured joint resistance (the macro thermal component) is still large which shows the importance of surface out-of-flatness/curvature. Collected data are compared with the scale analysis model and excellent agreement is observed. The maximum relative difference between the model and the collected data is 6. 8 percent and the relative RMS difference is approximately 4 percent. Additionally, the proposed scale analysis model is compared/verified with more than 880 TCR data points collected by many researchers. These data cover a wide range of materials, surface characteristics, thermal and mechanical properties, mean joint temperature, directional heat transfer effect, and contact between dissimilar metals. The RMS difference between the model and all data is less than 13. 8 percent.

Gas phase biofilters are used for controlling odors from animal facilities. Some characteristics can affect their performance and moisture content is one very important. A methodology for controlling and measuring moisture content is required to optimize these systems. An experiment was conducted to determine the appropriate placement of a set of soaker hoses 1.2 m in length for water application. It was found that the soaker hose installed in the lower region of the biofilter coupled with appropriate and timely application of water was able to minimize drying of the compost. Thermal conductance proved to be a reliable indicator for measuring the moisture content. Biofilters using the soaker hoses together with the thermal conductance as a media moisture sensor were able to maintain moisture content above 30% w.b. which provided sufficient water for microbial activity and ammonia abatement. A characterization of the ammonia and nitrous oxide concentrations was performed in order to compare the behavior of the gases when water was applied versus no water addition. These analyses revealed that the overall performance was not significantly different between treatments. But a more detailed assessment inside the biofilter media is performed; it is possible to identify different processes taking place.

L’intérêt pour les propriétés thermiques de matériaux nanostructurés est croissant. Ces matériaux sont conçus pour être inclus dans les dispositifs micro-électroniques et les systèmes micro électromécaniques (MEMS) dont le comportement et la fiabilité dépendent fortement de l’évacuation de la chaleur générée. Les matériaux multicouches diélectrique/métal sont de bons candidats pour la conversion thermoélectrique et leur utilisation est envisagée pour diminuer les températures maximales dans les systèmes microélectroniques. La diminution de l’épaisseur des couches permet de diminuer la conductivité thermique, conduisant à un plus grand facteur de mérite de conversion thermoélectrique. Cette diminution est due à la diminution de la conductivité thermique intrinsèque de chaque couche lorsque leur épaisseur décroit jusqu’à des dimensions du même ordre de grandeur que le libre parcours moyen des porteurs de chaleur et à l’influence croissante de la conductance d’interface. La prédiction de la conductivité thermique de tels systèmes passe donc par une simulation fiable du transfert de chaleur aux interfaces. La dynamique moléculaire (DM) est un outil particulièrement bien adapté à ce type d’études. Cependant les résultats des simulations dépendent fortement des potentiels interatomiques utilisés. La comparaison des propriétés prédites à l’aide des différents potentiels interatomiques avec les valeurs expérimentales permet de valider les potentiels pour prédire les propriétés concernées. Dans le premier chapitre, les fonctions mathématiques et les paramètres utilisés dans les potentiels interatomiques sont explicités. Dans le deuxième chapitre, l’objectif est de proposer une méthodologie pour sélectionner les potentiels les plus appropriés pour les études de transfert de chaleur. Cette méthodologie est illustrée pour le Si qui est le semi-conducteur le plus utilisé au sein de dispositifs microélectroniques et MEMS ainsi que pour l’Au, l’Ag et le Cu qui sont les métaux les plus souvent considérés. La conductivité thermique du Si massif est calculée, en utilisant la dynamique moléculaire hors d’équilibre (DMNE) avec trois potentiels parmi les cinq évalués précédemment pour valider cette évaluation. Le système diélectrique/métal qui a été le plus étudié avec la dynamique moléculaire mais également de manière expérimentale jusqu’à présent est certainement le système Si/Au. Les films de Cu et Ag sur des substrats de Si orienté sont les principales combinaisons dans les circuits intégrés de grande échelle. Une paramétrisation du potentiel de type MEAM est développée pour calculer les interactions Si/Au, Si/Ag et Si/Cu dans la troisième partie de ce travail. Les potentiels croisés sont utilisés pour prédire la conductance d’interface et développer les courbes de densité d’états pour les interfaces Si/Au Si/Ag et Si/Cu. / Interest in thermal properties of nanostructuredmaterials is growing. These materials are designed to be included in microelectronic devices and micro electromechanical systems (MEMS) whose behavior and reliability depend strongly on the dissipation of generated heat. Multilayer materials dielectric/metal are good candidates for thermoelectric conversion and their use is considered to reduce the maximum temperatures in microelectronic systems. The decrease in the thickness of the layers reduces the thermal conductivity, leading to a larger figure of merit of thermoelectric conversion. This decreasing is due to the decrease of intrinsic thermal conductivity of each layer when the thickness decreases to the dimensions of the same order of magnitude as the mean free path of heat carriers and bigger influence of the interface conductance. Predicting the thermal conductivity of such systems therefore requires a reliable simulation of heat transfer at interfaces. Molecular dynamics is a tool particularly well suited to this type of study. However the simulation results depend strongly on interatomic potentials used. The comparison of properties predicted using different interatomic potentials with experimental results validates the potential for predicting the properties concerned. In the first chapter, the mathematical functions and parameters used in the interatomic potentials are explained. In the second chapter, the objective is to propose a methodology to select the most appropriate potential for studying heat transfer. This methodology is illustrated for Si, the semiconductor most used in microelectronic devices and MEMS as well as for Au, Ag and Cu which are the metals most often seen. The thermal conductivity of bulk Si is calculated using the nonequilibrium molecular dynamics with three potential among the five previously evaluated to confirm this assessment. The system dielectric/metal that has been most studied with molecular dynamics but also experimentally is certainly the system Si/Au. The Cu and Ag films on oriented Si substrates are in the main combinations of large-scale integrated circuits. A parametrisation of MEAM cross-potential is developed to calculate interactions Si/Au, Si/Ag and Si/Cu in the third part of this work. The cross-potentials are used to predict the interfacial thermal conductance and to predict the density of states curves for the interfaces Si/Au Si/Ag and Si/Cu.

<p dir="ltr">This Thesis is divided into four parts. Its main themes are the thermal transport characteristics of Surface Phonon-Polaritons (SPhPs) and Surface Plasmon Polaritons (SPPs).</p><p dir="ltr">Chapter 1 introduces the main problem at issue in this Thesis: the decline in thermal conductivity with decreasing thicknesses in electronic devices and the feasibility of optimizing polar semiconductors and metals to produce polaritons that augment heat dissipation at these length scales.</p><p dir="ltr">Chapter 2 discusses Surface Phonon-Polariton (SPhP)-mediated thermal conductivity, or radiation conduction, in polar semiconductors. It considers the propagation of SPhPs in the case of two semi-infinite planes consisting of air and a polar semiconductor with a dielectric function described by its transverse- and longitudinal-optical (TOLO) phonon energies. It characterizes twenty different polar semiconductors in terms of radiation conduction via SPhPs and proposes a Figure of Merit (FoM) that describes the effectiveness of polariton conductance using easily-measured TO and LO phonon energies and linewidths.</p><p dir="ltr">Chapter 3 considers the propagation of SPPs in the case of two semi-infinite planes consisting of air and a metal with a dielectric function described by the Lorentz-Drude (LD) model. This chapter characterizes the effectiveness of eleven different metals as radiation conductors via SPPs and relates polariton conductance to electrical resistivity. It proposes a FoM analogous to the Wiedemann-Franz law that relates the effectiveness of polariton conductance and thermal conductance to the material’s electron scattering or linewidth.</p><p dir="ltr">Chapter 4 chapter compares the relative effectiveness of SPhP- and SPP-mediated radiation conduction. It describes why SPPs demonstrate far higher polariton conductance values than SPhPs by highlighting the underlying mechanisms at work in both—that is, available modes of energy transmission and their respective mean free path lengths.</p>