Les générateurs thermoélectriques convertissent directement l'énergie thermique en énergie électrique. Ils pourraient devenir de plus en plus utiles à des fins de récupération d'énergie et font l'objet de recherches actives. Cependant, les meilleurs matériaux thermoélectriques sont rares et polluants.Le Silicium et le Germanium seraient des matériaux intéressants si leur efficacité thermoélectrique était améliorée. Pour ce faire, la nanostructuration est une voie possible, par exemple en introduisant des faces rugueuses ou de nouvelles interfaces semi-transparentes.Récemment, des nanofils polyphasés (composés d'une alternance de phases cubiques et hexagonales de Si et Ge) ont été fabriqués, mais la caractérisation expérimentale de nanostructures aussi complexes comprenant des matériaux exotiques peut être difficile. Dans cette thèse, nous étudions en détail le transport thermique dans des nanostructures avec des simulations numériques. Une méthode Monte Carlo originale a été développée, avec une description "full band" des matériaux. Elle inclut des modèles pour les faces rugueuses et les interfaces entre matériaux. Des simulations de Dynamique Moléculaire sont également effectuées pour caractériser les propriétés des interfaces.Nous confirmons que les phases hexagonales de Si et Ge ont une conductivité thermique inférieure à celle des phases cubiques correspondantes. Le modèle "full band" montre que le flux thermique est fortement anisotrope. Des modèles semi-analytiques habituels n'ont pas pu reproduire la conductivité thermique des nanostructures simulées avec des faces rugueuses.De plus, ces faces ont tendance à concentrer le flux de chaleur dans la direction principale de la nanostructure. Enfin, certaines interfaces polyphasées peuvent avoir une conductance thermique presque aussi faible que les interfaces Si/Ge, et pourrait ainsi améliorer significativement l'efficacité thermoélectrique des nanofils polyphasés. La méthode Monte Carlo présentée peut facilement être utilisée pour étudier une large gamme de matériaux, et elle est capable de modéliser des nanostructures arbitrairement complexes. A l'avenir, les simulations en Dynamique Moléculaire seront utilisées pour paramétrer un modèle plus réaliste d'interfaces. / Thermoelectric generators are able to directly convert heat into electrical energy. They could have a great potential in terms of energy harvesting, but unfortunately, the best thermoelectric materials are rare and pollutant.Silicon and Germanium would be attractive materials if their thermoelectric efficiency were improved. For this purpose, nanostructuring is a possible route, for instance via the introduction of rough boundaries or interfaces between materials.Recently, polytype nanowires (composed of a sequence of cubic and hexagonal phases of Si and Ge) have been fabricated, but the experimental characterization of such complex nanostructures with exotic materials is challenging.In this thesis, we study the details of thermal transport in nanostructures with numerical simulations. An original Monte Carlo method is developed, with a full band emph{ab initio} description of materials. It includes models for the rough boundaries and the solid-solid interfaces. Molecular Dynamics simulations are also performed to characterize the properties of interfaces.We confirm that the hexagonal phases of Si and Ge have lower thermal conductivity than their cubic counterparts. The full band model shows a strong anisotropy in the heat flux.Usual semi-analytical models failed to reproduce the thermal conductivity of simulated nanostructures with rough boundaries. Besides, those boundaries tend to focus the heat flux in the main direction of the nanostructure. Finally, some polytype interfaces can have an interfacial conductance almost as low as Si/Ge interfaces, and thus could improve significantly the thermoelectric efficiency of polytype nanowires. The presented Monte Carlo method could easily be used with a wide range of materials,and it can model arbitrarily complex nanostructures. In the future, the results from Molecular Dynamics simulation will be used to parametrize a more realistic model of solid-solid interfaces.

京都大学 / 0048 / 新制・課程博士 / 博士(理学) / 甲第20162号 / 理博第4247号 / 新制||理||1611(附属図書館) / 京都大学大学院理学研究科物理学・宇宙物理学専攻 / (主査)教授 川上 則雄, 教授 松田 祐司, 教授 前野 悦輝 / 学位規則第4条第1項該当 / Doctor of Science / Kyoto University / DFAM

京都大学 / 0048 / 新制・課程博士 / 博士(理学) / 甲第20166号 / 理博第4251号 / 新制||理||1611(附属図書館) / 京都大学大学院理学研究科物理学・宇宙物理学専攻 / (主査)教授 松田 祐司, 教授 川上 則雄, 教授 前野 悦輝 / 学位規則第4条第1項該当 / Doctor of Science / Kyoto University / DFAM

Over the past two decades, controlling thermal transport properties at the nanoscale has become more and more relevant. This is mostly motivated by the need of developing novel energy-harvesting techniques based on thermoelectricity and the necessity to control the heat dissipation in semiconductor devices. In this field, two major research lines can be identified: On one side 'phononics', which aims at developing devices such as thermal diodes, thermal transistors, and thermal logic gates, among others, and on the other side, phonon engineering aiming at controlling heat transport by producing or structurally modifying heterostructures made of novel nanomaterials (e.g., two-dimensional (2D) materials, nanotubes, organic systems). In order to gain insight into the factors controlling nanoscale heat flow and to be able to design highly-efficient thermal devices, the development of new computational approaches is crucial. The primary goal of the present thesis is the implementation of new methodologies addressing classical and quantum thermal transport at the nanoscale. We will focus on three major issues: (i) We will study thermal rectification effect in nanodevices made of novel 2D materials by means of nonequilibrium molecular dynamics simulations. The influence of structural asymmetry and substrate deposition on the thermal rectification will be investigated. (ii) To address quantum ballistic thermal transport in nanoscale systems, we will implement a nonequilibrium Green's functions (NEGF) treatment of transport combined with a density-functional based approach. Here, we will explore the dependence of the thermal transport properties of 2D materials and nanotubes on different intrinsic (structural anisotropy and grain boundaries) and external (molecular functionalization, strain engineering, and doping) factors. Finally, (iii) a time-dependent NEGF formalism will be developed and implemented to probe the transient and steady thermal transport in molecular junctions. In short, our results show that the mechanisms governing the thermal rectification effect in the 2D thermal rectifiers proposed in this work are shape asymmetries, interface material (planar stacking order), and changes in the degree of spatial localization of high-frequency modes (under nonequilibrium heat transport conditions). The rectification effect can be also controlled by substrate engineering. Moreover, we found that quantum ballistic thermal transport in 2D puckered materials displays an anisotropic behavior. The presence of structural disorder in the form of grain boundaries in graphene reduces overall its thermal transport efficiency. Dynamical disorder induced by coupling to a thermostat has however a weaker effect, suggesting that structural defects are playing a major role. External factors have a noticeable influence on the heat transport in new 2D materials and BNC heteronanotubes. On the other hand, we have also been able to characterize, from a quantum point of view, the phonon dynamics in carbon-based molecular junctions. We expect that the results obtained within this thesis will yield new insights into the thermal management of low-dimensional materials, and thus open new routes to the design of thermoelectric and phononic devices. / In den letzten zwei Jahrzehnten hat die Kontrolle der thermischen Transporteigenschaften im Nanobereich immer mehr an Bedeutung gewonnen. Dies ist vor allem auf die Notwendigkeit zurückzuführen, neue Energiegewinnungstechniken zu entwickeln, die auf Thermoelektrizität basieren, sowie auf die Problematik, die Wärmeabfuhr in Halbleiterbauelementen kontrollieren zu müssen. In diesem Bereich lassen sich zwei große Forschungslinien identifizieren: Auf der einen Seite 'Phononik', die unter anderem auf die Entwicklung von Bauelementen wie thermischen Dioden, Transistoren und Logikgattern abzielt, und auf der anderen Seite die Phononentechnik, die den Wärmetransport durch Herstellung oder strukturelle Modifikation von Heterostrukturen aus neuartigen Nanomaterialien (z.B. zweidimensionalen (2D) Materialien, Nanoröhren, organischen Systemen) steuert. Um einen Einblick in die Faktoren zu erhalten, die den Wärmefluss im Nanobereich steuern, und um hocheffiziente thermische Bauteile entwickeln zu können, ist die Entwicklung neuer Berechnungsansätze entscheidend. Das Hauptziel der vorliegenden Arbeit ist die Implementierung neuer Methoden, die sich mit dem klassischen und dem quantenthermischen Transport auf der Nanoskala befassen. Wir werden uns auf drei Hauptthemen konzentrieren: (i) Wir werden den thermischen Rektifikationseffekt in Nanobauteilen aus neuartigen 2D-Materialien mit Hilfe von Nichtgleichgewichts-Molekulardynamiksimulationen studieren. Der Einfluss von Strukturasymmetrie und Substratablagerung auf die thermische Rektifikation wird untersucht. (ii) Um den quantenballistischen Wärmetransport in nanoskaligen Systemen anzugehen, werden wir eine NEGF-Behandlung (Nichtgleichgewichts-Greensche Funktionen) des Transports in Kombination mit einem dichtefunktionalen Ansatz implementieren. Hier wird die Abhängigkeit der thermischen Transporteigenschaften von 2D-Materialien und Nanoröhrchen von verschiedenen intrinsischen (strukturelle Anisotropie und Korngrenzen) und externen (molekulare Funktionalisierung, Stammtechnik und Dotierung) Faktoren untersucht. Schließlich wird (iii) ein zeitabhängiger NEGF-Formalismus entwickelt und implementiert, um den transienten und stetigen Wärmetransport in molekularen Verbindungen zu untersuchen. Unsere Ergebnisse zeigen, dass die wesentlichen Mechanismen für die thermische Gleichrichtung in 2D thermischen Gleichrichtern durch Asymmetrien der Bauteilform, das Interface-Material (planare Stapelung Reihenfolge), und änderungen im Grad der räumlichen Lokalisierung von Hochfrequenz-Modi (unter Nicht-Gleichgewicht Wärmetransport-Bedingungen) gegeben sind. Der Gleichrichteffekt kann auch durch die Wahl des Substrats gesteuert werden. Darüber hinaus haben wir festgestellt, dass der quantenballistische Wärmetransport in 2D-Puckered-Materialien ein anisotropes Verhalten zeigt. Das Vorhandensein von strukturellen Störungen in Form von Korngrenzen in Graphen reduziert insgesamt die Effizienz des Wärmetransports. Dynamische Störungen, die durch die Ankopplung an einen Thermostaten hervorgerufen werden, haben jedoch eine schwächere Wirkung, was darauf hindeutet, dass strukturelle Defekte eine große Rolle spielen. Externe Faktoren haben einen nachweislichen Einfluss auf den Wärmetransport in neuen 2D-Materialien und BNC-Heteronanotubes. Weiterhin konnten wir auch die Phononendynamik in kohlenstoffbasierten molekularen Verbindungen quantitativ charakterisieren. Wir erwarten, dass die Ergebnisse dieser Arbeit neue Erkenntnisse über das Wärmemanagement von niedrigdimensionalen Materialien liefern und damit neue Wege für das Design von thermoelektrischen und phononischen Bauelementen eröffnen.

<p>  </p> <p>This dissertation focuses on predictive modeling of phonon-mediated thermal transport in thorium dioxide (ThO2) with defects. ThO2 has lately gained attention as it is a suitable model system for more complex nuclear reactor materials such as uranium dioxide and its mixed oxides. The reduction in thermal conductivity of the fuel as a result of irradiation-induced lattice defects is arguably the most important fuel performance metric in regard to reactor efficiency and safety. For this reason, the present work presents a theoretical investigation of thermal conductivity reduction seen in defect-bearing ThO2 and compares directly with experimental measurements. Thermal transport in irradiated ThO2 is first modeled here by a non-transport solution of the linearized Boltzmann transport equation (BTE) within the single-mode relaxation time approximation. Classic models for phonon-defect scattering rates are used to model point defects, voids, and dislocation loops in irradiated ThO2, and the resultant thermal conductivity is directly compared to experimental measurements of irradiated specimens. Our predicted conductivity values agree well with measured values near room temperature. However, discrepancy between our predictions and experimental values exist at lower temperatures where experimentally measured conductivity values seem to reach a saturation level while the model predicts further reduction in thermal conductivity. This discrepancy is most notable in higher irradiation dose samples where the thermal conductivity is almost completely controlled by the dislocation loop density. This hints at the conclusion that classic models for phonon-defect scattering rates which integrate out local variation of the defect strain field and replace this by a defect density may not be adequate to capture all physics of phonon-defect scattering, especially for dislocation loops at low temperatures. This motivated us to model defects through their spatially resolved lattice distortion fields and investigate phonon scattering in those fields in an explicit fashion. A transport solution of the phonon BTE is implemented based upon the Monte Carlo (MC) method, which explicitly tracks the phonon population as it evolves in space and time according to phonon group velocities and scattering rates. An expression for the scattering rate of phonons from an arbitrary strain field is derived from a generalized form of Grüneisen’s law of thermal expansion, and applied to the case of dislocations in ThO2. It is found that the localized strain in the material, resulting from the presence of a crystal defect, leads to a net heat flux into the strained region. This provides evidence for thermal fluxes in the absence of a temperature gradient, a phenomenon that cannot be captured via Fourier’s law. This evidence for material heating owing to the imposed strain of material defects would be immediately applicable to the field of thermoelectrics and defect engineering where large temperature gradients are desirable to improve the thermoelectric efficiency.  Although the model is applied specifically to the case of dislocations in ThO2, the derived phonon scattering rate expression is general and may be applied to any defect for which a strain field may be generated.</p>

<p>The need for sources of efficient and renewable energy has become an issue of great importance in recent years. Fossil fuels are diminishing in supply, and they not only pollute the environment, but have also proven to be inefficient in many cases, losing a large portion of the total energy generated to waste heat, instead of usable energy. </p><p>The first work this thesis addresses is the development of a Genetic Algorithm (GA) optimization method for the search and discovery of semiconductor materials for use in thermoelectric devices. The specific material in question is the Silicon Germanium superlattice. This structure made of alternating layers of Si and Ge is known to be one of the better materials for thermoelectric energy generation at elevated temperature, along with Bismuth-Telluride that targets room temperature. Previously, it has been shown that random multilayer (RML) structures can lower thermal conductivity as compared to periodic superlattices due to phonon localization. However, it was unknown which specific RML would yield the lowest thermal conductivity, due to the large design space from which these RML’s can be generated. Considering this, a global and non-smooth optimization method was employed to search for the best possible structure. Results not only showed that the thermal conductivity could be lowered even further, but that there was an optimal average period for the RML’s that produced the best results.</p><p>The second work discussed in this thesis concerns itself with the development of a Neural Network Potential (NNP) for use in Molecular Dynamics (MD) simulations. There are multiple methods for running MD including <i>ab-initio</i>methods such as Density Functional Theory (DFT) calculations and classical MD with the use of empirical potentials. Unfortunately, DFT is too time consuming for systems larger than a few hundred atoms, and empirical potentials can be inaccurate. Therefore, a NNP for bulk Silicon trained on DFT was developed, and it was shown that the phonon dispersion for Si could be accurately reproduced.</p>

Ash deposits in commercial coal-fired boilers frequently pose serious maintenance challenges and decrease thermal efficiency. A better understanding of fundamental thermal transport properties in ash deposits can help mitigate their negative effects. In order to characterize the thermal properties of boiler-side deposits, this work presents a thermal transport model and in-situ measurements of effective thermal conductivity in coal ash deposits. A simple model of the thermal transport through an ash deposit, with and with out slagging, was developed. The model approximates the deposit by dividing it into four regimes: particulate, sintered, solidified slag, and molten slag. The development of this model was auxiliary to the primary focus of this study: the in-situ measurement of effective thermal conductivity of ash deposits. Deposits of loosely-bound particulate ash were obtained experimentally using a down-fired drop tube reactor. Pulverized coal was fired and deposits were collected on an instrumented deposition probe. An approach is presented for making in-situ measurements of the temperature difference across the ash deposits, the thickness of the deposits, and the total heat transfer rate through the ash deposits. Using this approach, the effective thermal conductivity was determined for coal ash deposits formed under oxidizing and reducing conditions. Three coals were tested under oxidizing conditions: IL #6 Crown III coal, IL #6 Patiki coal and WY Corederro coal. The WY coal exhibited the lowest range of effective thermal conductivities (ke =0.05 to 0.175 W/mּK) while the IL #6 coals showed higher effective thermal conductivities (ke =0.2 to 0.5 W/mּK). The IL #6 Crown III coal and the WY Corederro coal were also tested under reducing conditions. A comparison of the ash deposits from these two coals, formed under oxidizing or reducing conditions, showed larger effective thermal conductivities in deposits formed under reducing conditions. The IL #6 Crown III coal exhibited the greatest increase (as high as 50%) in ke, under reducing conditions, over that measured in oxidizing conditions. For all of the experiments conducted, an increase in effective thermal conductivity with deposit thickness was observed, with sintering likely causing the increase in ke.

No description available.

<div>Metallic components and metal-dielectric interfaces appear widely in modern electronics and the thermal management is an important issue. A very important feature that has been overlooked in the conventional Fourier's equations analyses is the nonequilibrium thermal transport induced by selective electron-phonon (e-p) coupling and phonon-phonon (p-p) coupling. It signicantly affects many processes such as laser heating and ignoring this phenomenon can lead to wrong or misleading predictions. On the other hand, as devices shrink into nano-scale, heat generation and dissipation at the interfaces between different components start to dominate the thermal process and present a challenge for thermal mitigations. Many unresolved issues also arise from interfaces, such as the unexpected high interfacial thermal conductance (ITC) at metal-diamond interfaces. Both of these require a deep understanding of the physics at interfaces.</div><div><br></div><div><div>Therefore in this work, I present multi-scale simulations in metals/dielectrics and interfaces based on two-temperature model (TTM) and establish the new multitemperature model (MTM). The methods are combined with Fourier's Law, molecular dynamics (MD), Boltzmann transport equations (BTE) and implemented to predict the thermal transport in several materials and interfaces where e-p coupling and p-p coupling are important. First-principles studies based on density functional theory (DFT) are also presented as predictive approaches to acquire the properties, as well as investigating the new physical phenomenon of non-local e-p coupling in metals. This research seeks to provide general, sophisticated but also simple simulation approaches which can help people accurately predict the thermal transport process. It also seeks to explore new physics which cannot be captured and predicted by conventional analyses based on Fourier's Law and can advance our understanding as well as providing new insights in the current thermal analysis paradigm.</div></div><div><br></div><div><div>The rst part of this thesis focuses on the non-equilibrium thermal transport in metals and across metal-dielectric interfaces based on TTM. First of all, nonequilibrium thermal transport in metal matrix composites (MMC) is investigated. Metal particle is usually added to polymer matrix for enhanced thermal performance. Here we apply TTM calculations and manifest a \critical particle size" above which the thermal conductivity of the composite material can be enhanced. MD simulations are performed to predict the thermal properties. TTM-Fourier and TTM-BTE calculations are conducted as comparisons. The widely used Au-SAM (self-assemblymonolayers) material pair is chosen to demonstrate our models. For a 1-D SAMAu-SAM sandwich system, the two calculation approaches present almost identical results, and the critical particle size is 10.7 nm. A general interpretation of thermal transport in sandwiched metal thin lms between two dielectric materials is also presented. It is found that when the lm thickness is on the order of several nanometers, due to strong e-p non-equilibrium the thermal transport is dominated by phonons</div><div>and electrons hardly contribute.</div></div><div><br></div><div><div>Then the e-p non-equilibrium thermal transport across metal-dielectric interfaces is investigated using TTM-MD. One possible explanation to the unexpected ITC at metal-diamond interfaces is the cross-interface e-p coupling mechanism, which is based on the hypothesis that electrons can couple to phonons within a certain distance rather than just those at the same location. Therefore we extend TTM-MD by modifying its governing equation to a non-local integral form. Two models are proposed to describe the coupling distance: the \joint-phonon-modes" model and the \phonon-wavelength" model. A case study of thermal transport across Cu-Si interfaces is presented, and both models predict similar coupling distances of 0.5 nm in Cu and 1.4 nm in Si near the interfaces. The cross-interface e-p coupling can increase the ITC by 20% based on our models. Based on the results, we construct a new mixed series-parallel thermal circuit. It is shown that such a thermal circuit is essential for understanding metal-nonmetal interfacial transport, while calculating a single resistance without solving temperature proles as done in most previous studies is generally incomplete.</div></div><div><br></div><div><div>Inspired by the previous work, we investigate further into the physics of nonlocal e-p coupling. First-principles calculations based on DFT is used due to their predictive feature without assumptions or adjustable parameters. By calculating the e-p coupling in metal lms of different sizes, we nd that e-p coupling has size effect which can only be explained by a non-local coupling picture. Results show that in Al, electrons and phonons can couple to each other in a range of up to 2 lattice-constants, or 0.8 nm. The coupling strength between electrons and phonons in adjacent atomic layers still has 75% of that in the same layer. Comparative studies are also performed on Cu and Ag. Results show that their non-local e-p coupling is not as signicant as in Al, with coupling distances of 0.37 nm for Cu and 0.49 nm for Ag. Similar results in Cu and Ag also indicate that materials with similar electronic structures have similar non-local e-p coupling properties.</div></div><div><br></div><div><div>In TTM, it is assumed that phonons are in thermal equilibrium and have a common temperature. In the second part of this thesis we go beyond TTM to investigate the non-equilibrium between phonons as well. TTM is extended to a general MTM with e-p coupling strength for each phonon branch. An averaged scattering lattice reservoir is dened to represent p-p scattering. The thermal transport process in single-layer graphene under constant and pulse laser irradiation is investigated. Results show that the phonon branches are in strong non-equilibrium. A comparison with TTM reveals that MTM can increase the thermal conductivity prediction by 50% and the hot electron relaxation time by 60 times. We also perform MTM simulations on Si-Ge interfaces to investigate the effect of non-equilibrium thermal transport on ITC. Results show that thermal non-equilibrium between phonons will introduce additional resistance at the interfaces, which is similar with e-p non-equilibrium's impact on ITC at metal-dielectric interfaces.</div></div>

The ultra-short gate (LG < 20 nm) CMOS components (Complementary Metal-Oxide-Semiconductor) face thermal limitations due to significant local heating induced by phonon emission by hot carriers in active regions of reduced size. This phenomenon, called self-heating effect, is identified as one of the most critical for the continuous increase in the integration density of circuits. This is especially crucial in SOI technology (silicon on insulator), where the presence of the buried insulator hinders the dissipation of heat.At the nanoscale, the theoretical study of these heating phenomena, which cannot be led using the macroscopic models (heat diffusion coefficient), requires a detailed microscopic description of heat transfers that are locally non-equilibrium. It is therefore appropriate to model, not only the electron transport and the phonon generation, but also the phonon transport and the phonon-phonon and electron-phonon interactions. The formalism of the Boltzmann transport equation (BTE) is very suitable to study this problem. In fact, it is widely used for years to study the transport of charged particles in semiconductor components. This formalism is much less standard to study the transport of phonons. One of the problems of this work concerns the coupling of the phonon BTE with the electron transport.In this context, wse have developed an algorithm to calculate the transport of phonons by the direct solution of the phonon BTE. This algorithm of phonon transport was coupled with the electron transport simulated by the simulator "MONACO" based on a statistical (Monte Carlo) solution of the BTE. Finally, this new electro-thermal simulator was used to study the self-heating effects in nano-transistors. The main interest of this work is to provide an analysis of electro-thermal transport beyond a macroscopic approach (Fourier formalism for thermal transport and the drift-diffusion approach for electric current, respectively). Indeed, it provides access to the distributions of phonons in the device for each phonon mode. In particular, the simulator provides a better understanding of the hot electron effects at the hot spots and of the electron relaxation in the access.