Thermal Transport in Carbon-Based Nanomaterials
By Gang Zhang
()
About this ebook
Thermal Transport in Carbon-Based Nanomaterials describes the thermal properties of various carbon nanomaterials and then examines their applications in thermal management and renewable energy. Carbon nanomaterials include: one-dimensional (1D) structures, like nanotubes; two-dimensional (2D) crystal lattice with only one-atom-thick planar sheets, like graphenes; composites based on carbon nanotube or graphene, and diamond nanowires and thin films. In the past two decades, rapid developments in the synthesis and processing of carbon-based nanomaterials have created a great desire among scientists to gain a greater understanding of thermal transport in these materials.
Thermal properties in nanomaterials differ significantly from those in bulk materials because the characteristic length scales associated with the heat carriers, phonons, are comparable to the characteristic length. Carbon nanomaterials with high thermal conductivity can be applied in heat dissipation. This looks set to make a significant impact on human life and, with numerous commercial developments emerging, will become a major academic topic over the coming years. This authoritative and comprehensive book will be of great use to both the existing scientific community in this field, as well as for those who wish to enter it.
- Includes coverage of the most important and commonly adopted computational and experimental methods to analyze thermal properties in carbon nanomaterials
- Contains information about the growth of carbon nanomaterials, their thermal properties, and strategies to control thermal properties and applications, allowing readers to assess how to use each material most efficiently
- Offers a comprehensive overview of the theoretical background behind thermal transport in carbon nanomaterials
Related to Thermal Transport in Carbon-Based Nanomaterials
Titles in the series (97)
Microfluidics: Modeling, Mechanics and Mathematics Rating: 0 out of 5 stars0 ratingsNanotechnologies in Preventive and Regenerative Medicine: An Emerging Big Picture Rating: 0 out of 5 stars0 ratingsApplied Nanotechnology: The Conversion of Research Results to Products Rating: 0 out of 5 stars0 ratingsNano Optoelectronic Sensors and Devices: Nanophotonics from Design to Manufacturing Rating: 0 out of 5 stars0 ratingsNanobiomaterials in Clinical Dentistry Rating: 5 out of 5 stars5/5Nanotechnology Applications for Clean Water: Solutions for Improving Water Quality Rating: 0 out of 5 stars0 ratingsEmerging Nanotechnologies in Dentistry Rating: 0 out of 5 stars0 ratingsNanotechnology and Nanomaterials in the Treatment of Life-threatening Diseases Rating: 0 out of 5 stars0 ratingsEmerging Nanotechnologies for Diagnostics, Drug Delivery and Medical Devices Rating: 0 out of 5 stars0 ratingsNanoengineered Biomaterials for Regenerative Medicine Rating: 0 out of 5 stars0 ratingsPhysical Fundamentals of Nanomaterials Rating: 0 out of 5 stars0 ratingsNanotechnology: An Introduction Rating: 5 out of 5 stars5/5Nanomaterials and Devices Rating: 5 out of 5 stars5/5Nanotechnology Applications for Tissue Engineering Rating: 0 out of 5 stars0 ratingsSpace Microsystems and Micro/Nano Satellites Rating: 0 out of 5 stars0 ratingsHeat Transfer Enhancement Using Nanofluid Flow in Microchannels: Simulation of Heat and Mass Transfer Rating: 0 out of 5 stars0 ratingsGraphene and Related Nanomaterials: Properties and Applications Rating: 5 out of 5 stars5/5Advanced Supramolecular Nanoarchitectonics Rating: 0 out of 5 stars0 ratingsMicro-Drops and Digital Microfluidics Rating: 0 out of 5 stars0 ratingsMicromixers: Fundamentals, Design and Fabrication Rating: 0 out of 5 stars0 ratingsAssessing Nanoparticle Risks to Human Health Rating: 0 out of 5 stars0 ratingsApplications of Nanofluid for Heat Transfer Enhancement Rating: 0 out of 5 stars0 ratingsNew Approaches to Image Processing based Failure Analysis of Nano-Scale ULSI Devices Rating: 5 out of 5 stars5/5Lanthanide-Based Multifunctional Materials: From OLEDs to SIMs Rating: 0 out of 5 stars0 ratingsFabrication and Design of Resonant Microdevices Rating: 5 out of 5 stars5/5Thermoelectricity and Heat Transport in Graphene and Other 2D Nanomaterials Rating: 0 out of 5 stars0 ratingsNanomaterials for Biosensors: Fundamentals and Applications Rating: 0 out of 5 stars0 ratingsHybrid Nanostructures for Cancer Theranostics Rating: 0 out of 5 stars0 ratingsThermohydrodynamic Programming and Constructal Design in Microsystems Rating: 0 out of 5 stars0 ratingsBiomolecular Electronics: Bioelectronics and the Electrical Control of Biological Systems and Reactions Rating: 4 out of 5 stars4/5
Related ebooks
Nuclear Power Reactor Designs: From History to Advances Rating: 0 out of 5 stars0 ratingsCarbon Nanotubes and Graphene Rating: 0 out of 5 stars0 ratingsComprehensive Guide on Organic and Inorganic Solar Cells: Fundamental Concepts to Fabrication Methods Rating: 0 out of 5 stars0 ratingsMicro and Nano Thermal Transport: Characterization, Measurement, and Mechanism Rating: 0 out of 5 stars0 ratingsLaser-Induced Breakdown Spectroscopy Rating: 0 out of 5 stars0 ratingsOne-dimensional Nanostructures for PEM Fuel Cell Applications Rating: 0 out of 5 stars0 ratingsNanomaterials in Rocket Propulsion Systems Rating: 5 out of 5 stars5/5Biopolymer Composites in Electronics Rating: 5 out of 5 stars5/5Emerging Carbon Materials for Catalysis Rating: 0 out of 5 stars0 ratingsMicrosupercapacitors Rating: 0 out of 5 stars0 ratingsSpace Micropropulsion for Nanosatellites: Progress, Challenges and Future Rating: 0 out of 5 stars0 ratingsEmerging Nanomaterials for Recovery of Toxic and Radioactive Metal Ions from Environmental Media Rating: 0 out of 5 stars0 ratingsRadiation Technology for Advanced Materials:: From Basic to Modern Applications Rating: 0 out of 5 stars0 ratingsModelling of Flow and Transport in Fractal Porous Media Rating: 0 out of 5 stars0 ratingsNanostructured Materials Rating: 0 out of 5 stars0 ratingsCarbon Nanotube-Reinforced Polymers: From Nanoscale to Macroscale Rating: 0 out of 5 stars0 ratingsGlobal Progress on Molten Salt Reactors: A Companion to Dolan’s Molten Salt Reactors and Thorium Energy Rating: 0 out of 5 stars0 ratingsMaterials Science and Engineering of Carbon: Characterization Rating: 0 out of 5 stars0 ratingsPrecision Motion Systems: Modeling, Control, and Applications Rating: 0 out of 5 stars0 ratingsNanomaterials for Solar Cell Applications Rating: 0 out of 5 stars0 ratingsSupercapacitors: Materials, Design, and Commercialization Rating: 0 out of 5 stars0 ratings
Materials Science For You
1,001 Questions & Answers for the CWI Exam: Welding Metallurgy and Visual Inspection Study Guide Rating: 4 out of 5 stars4/5Geotechnical Problem Solving Rating: 0 out of 5 stars0 ratingsAdditive Friction Stir Deposition Rating: 0 out of 5 stars0 ratingsStoring Energy: with Special Reference to Renewable Energy Sources Rating: 0 out of 5 stars0 ratingsMad About Metal: More Than 50 Embossed Craft Projects for Your Home Rating: 0 out of 5 stars0 ratingsPhysical Metallurgy and Advanced Materials Rating: 5 out of 5 stars5/5Piping Materials Guide Rating: 4 out of 5 stars4/5Demystifying Explosives: Concepts in High Energy Materials Rating: 0 out of 5 stars0 ratingsApplied Welding Engineering: Processes, Codes, and Standards Rating: 0 out of 5 stars0 ratingsMetalworking: Tools, Materials, and Processes for the Handyman Rating: 5 out of 5 stars5/5Choosing & Using the Right Metal Shop Lathe Rating: 0 out of 5 stars0 ratingsManufacturing Technology for Aerospace Structural Materials Rating: 5 out of 5 stars5/5Handbook of Adhesion Rating: 0 out of 5 stars0 ratingsElectric Vehicle Battery Systems Rating: 0 out of 5 stars0 ratingsThe Art of Welding: Featuring Ryan Friedlinghaus of West Coast Customs Rating: 0 out of 5 stars0 ratingsHigh Pressure Pumps Rating: 4 out of 5 stars4/5Polymer Characterization: Laboratory Techniques and Analysis Rating: 0 out of 5 stars0 ratingsThe Foseco Foundryman's Handbook: Facts, Figures and Formulae Rating: 3 out of 5 stars3/5Surface Chemistry of Nanobiomaterials: Applications of Nanobiomaterials Rating: 0 out of 5 stars0 ratingsDistributed Renewable Energies for Off-Grid Communities: Empowering a Sustainable, Competitive, and Secure Twenty-First Century Rating: 0 out of 5 stars0 ratingsThe Dynamics of Nazism: Leadership, Ideology, and the Holocaust Rating: 0 out of 5 stars0 ratingsCrack Analysis in Structural Concrete: Theory and Applications Rating: 0 out of 5 stars0 ratingsNanobiomaterials in Dentistry: Applications of Nanobiomaterials Rating: 4 out of 5 stars4/5Skilletheads: <b>A Guide to Collecting and Restoring Cast-Iron Cookware</b> Rating: 0 out of 5 stars0 ratings
Reviews for Thermal Transport in Carbon-Based Nanomaterials
0 ratings0 reviews
Book preview
Thermal Transport in Carbon-Based Nanomaterials - Gang Zhang
journals.
Preface
Gang Zhang Singapore
fullerenes. Each of them has notably different electronic and mechanical properties. Diamond is a wide band gap insulator and one of the hardest natural materials. On the contrary, graphene has a semimetal electronic structure with linear band dispersion and extraordinarily high electron mobility. Carbon based materials have attracted a great deal of interest due to its unique combination of several fascinating properties, such as extreme hardness, chemical inertness and ultrahigh thermal conductivity, which are originated from its lattice structure and strong carbon–carbon bond. In the past two decades, rapid developments in synthesis and processing of carbon–based nanoscale materials have created a great demand for understanding of thermal transport. Carbon nano materials include: one-dimensional structures, like carbon nanotubes, diamond nanothread and diamond nanowires; two-dimensional crystal lattice with only one-atom-thick planar sheets, like graphene; composites based on carbon nanotube or graphene.
Usually, these carbon based nanomaterials have high thermal conductivity compared with other nanomaterials. The conduction of heat is one of the fundamental energy transport mechanisms in nature. Carbon nanomaterials with high thermal conductivity can be applied in heat dissipation. It looks set to make a significant impact on human life and, with numerous commercial developments emerging, will become a major academic topic over the coming years. More important, through studying the length dependent thermal conductivity and anomalous heat diffusion in carbon nanotubes, it was found that the thermal properties of nanostructures differ significantly from their bulk counterparts when the phonon characteristic lengths are comparable to the lengths of nanostructures. This provided the first rigorous proof for the breakdown of Fourier's law
and has been widely accepted as a basic law of heat conduction in nanoscale.
This book sets the subject into context by first of all describing the thermal properties of various carbon nanomaterials and then looking at the applications in thermal management and renewable energy. The aim of this book is to provide an introduction for both theorists and experimentalists to the thermal properties of carbon nanomaterials and then looking at the applications in thermal management and renewable energy. The book should also be useful for graduate-level students who want to explore this new field of research. With content relevant to both academic and commercial viewpoints, the book will interest researchers and postgraduates as well as consultants and industrial engineers.
The single chapters have been written by internationally recognized experts in material science and engineering and provide in-depth introductions to the directions of their research. Accordingly, the thermal properties of graphene, diamond nanostructures, graphene composite, diamond nanothread, carbon fibers are addressed. Moreover, one chapter outlines the phonon transport theory and simulation methods, one chapter features the experimental methodologies to explore nanoscale thermal transport, and one chapter introduces the lattice dynamics of graphene. This book provides the basis thermal transport theory, the different effects on thermal conductivity and other important details for better controlling thermal conductivity of carbon nanomaterials. I am sure that this book will be a useful reference not only for scientists and engineers exploring thermal transport at nanoscale, but also for graduate and postgraduate students specializing in physics and material science.
I am most grateful to Elsevier publisher for the invitation to edit this book, and for kind and efficient assistance in editing this book. I am also grateful to all book chapter authors for sharing their expertise in this multi-author monograph. Their strong efforts and enthusiasm for this project were indispensable for bringing it to success. I finally wish to thank my family, particularly my daughter Rhianna. She helped me to design the cover image.
2016 Winter
Chapter 1
Thermal Transport Theory
Hangbo Zhou; Gang Zhang Institute of High Performance Computing, A*STAR, Singapore
Abstract
In the chapter we review various theoretical methods for the evaluation of thermal conduction properties, carried by phonons. In detail, we first demonstrate the methods of calculating thermal conductivity from its near-equilibrium properties, including kinetic theory, Boltzmann transport equation method and the Green–Kubo approach together with equilibrium molecular dynamics. We then illustrate the non-equilibrium methods of calculating thermal currents, including the non-equilibrium molecular dynamics and the non-equilibrium Green's function formalism. For each method, we analyze its respective viewpoint of transport picture, the assumptions made, its strength and weakness in handling transport problem and its applicability in different situations.
Keywords
Thermal transport; heat conduction; Boltzmann transport equation; Green–Kubo formula; equilibrium molecular dynamics; non-equilibrium molecular dynamics; non-equilibrium Green's function
1.1 Introduction
The conduction of heat is one of the fundamental energy transport mechanisms in nature. The vibrational modes of lattices, also known as phonons, have critical impacts for heat control and management in the real world. Today, approximately 80% of the world's power is generated by heat engines that use fossil fuel combustion as a heat source, which is believed to be responsible for a large fraction of carbon dioxide (CO2) emissions worldwide. The heat engines used in most thermal power station typically operate at 30–40% efficiency. This means that roughly 10 TW of heat energy is lost to the environment. Thermoelectric modules can potentially convert part of the wasted heat directly into electricity, reduce the usage of fossil fuels and lowering carbon emission [1]. In addition to thermoelectric materials, thermal management is also significantly important for performance of electronic and optoelectronic devices. The application of nano devices has highlighted the need for greater quantitative understanding of materials at the nanoscale. Understanding the physics of thermal transport in nanoscale by computational study is particularly important, because their small size makes it challenging to apply standard experimental measurement methods. Over the past decades advances in computer science have spurred advances in fundamental theoretical techniques, mathematical modelling, and numerical simulation, giving rise to a revolution with extraordinary impact for nanoscience and nanoengineering.
The aim of this chapter is to provide an introduction for both theorists and experimentalists to the current thermal transport theory and computational technology. We address the current commonly used computational technologies, and their applications in study of nanoscale energy transport and conversion.
1.2 Near-Equilibrium Theory
1.2.1 Kinetic Theory
Heat flows unavoidably from a hot end to cool one, and an important question to this heat conduction is how fast it flows. The first qualitative answer to this question is attributed to Fourier two hundred years ago, which states that in solids the heat current density j, defined as the rate of heat flow per cross area, is proportional to the temperature gradient as
(1.1)
This equation is called the Fourier's law and it defines the thermal conductivity, κ. In principle κ is a tensor but in isotopic systems it is often treated as a number. The essence of this Fourier's law is that the thermal conductivity is believed to be an intrinsic quantity. In other words, it is only related to the material properties but gauges all the extensive variables like the size effect. The Fourier law has been widely used in many applications for bulk materials and it receives great success. However, the rigorous derivation of Fourier's law from microscopic picture meets great challenges, and the validity of Fourier's law is intensively discussed in recent literatures [2], especially in low-dimensional materials.
Fourier only answers the question partly in the sense that value of thermal conductivity is still undetermined. The thermal conductivity is closely related to the type of particles in materials that carry the energy and the transport of these energy-carrying particles. In carbon-based materials, the lattice vibrations, or the phonons, take the major responsibility for the heat conduction; hence here we focus on thermal transport carried by phonons. In the following, we will present various theoretical frameworks that evaluate the thermal transport properties under different viewpoints.
In the simplest manner, the heat conduction can be described as a diffusive process of energy-carrying particles, the phonons, attributed to Debye. Suppose the density of phonons is n, each phonon carries energy u, it travels at velocity v along all directions equally and they can travel with time τ before the next collision, then a simple derivation leads to the formula for the thermal conductivity [3]:
(1.2)
is the heat capacity of each phonon. In practice, the phonon velocity is often treated as sound velocity and τ .
This formalism is called the kinetic theory of phonons. Despite its simplicity, it reveals the major factors responsible for heat conduction: the phonon density, the velocity and the relaxation time. Actually these factors are also the most important quantities affecting thermal transport in more advanced theories, manifested as phonon spectral density (phonon distribution and density of state), the phonon dispersion relation and the phonons scatterings respectively.
The kinetic theory illustrates the clear picture of phonon transport, but it is not able to predict the thermal conductivity, because the relaxation time of phonon, or the mean free path of phonon, is unknown and the reason of phonon collision is not given. Actually the mean free path of phonons can differ in a large scale, from nanometers to centimeters for different materials. Hence the estimation of relaxation time becomes the key step in predicting the thermal conductivities. Importantly, the collisions of phonons may be caused by completely different reasons, such as phonon–phonon collision, boundary or interface effects, impurities and defects inside the materials and many others. As a result, the analysis of relaxation time becomes complicated and highly case-dependent. However, from experimental point of view, one can measure the thermal conductivity and hence estimate the relaxation time of phonons, which in turn helps the understanding of the fundamental transport mechanism of phonons.
There are few key assumptions implicitly made in the kinetic theory
(1) The phonons travel as independent particles, and equally distributed with respect to energy. However, the distribution of phonons strongly depends on the energy such that higher energy phonons are less occupied. The distribution is also strongly affected by temperature. Furthermore, not all the phonon energies are allowed and it is restricted by materials. And the allowed number of phonons at each energy interval is also material-dependent which is called density of state.
(2) All phonons travel at the same velocity. In fact the phonon velocity depends on the mode of lattice vibrations as well as its energy. In certain cases, some phonons are even localized without travelling.
(3) The relaxation time approximation. In kinetic theory the collisions are described as independent process. However, the rate of collision, and the outcome of collision are phonon distribution related. And the collision events are correlated. The actually phonon transport process is a many-body phenomenon while the relaxation time approximation are in the single particle scheme.
1.2.2 Boltzmann Transport Equation
Peierls derived the phonon Boltzmann transport equation (BTE) [4]. This BTE approach makes corrections to the assumption (1) and (2) made in the kinetic theory. In principle assumption (3) is also not necessary but in practice, the relaxation time assumption is often imposed in order to solve the BTE. BTE has been a standard theory in dealing phonon transport in bulk materials, which provides a clear picture of the transport process.
and time t. This form of distribution function determines that the BTE approach is a semi-classical theory in the single particle approximation, since one assumes that the position and momentum of phonons can be known simultaneously and the many-body effects are not taken into account.
is given. For instance, the local phonon density can be evaluated as
(1.3)
is completely due to the collision of the particles
(1.4)
By expanding the first term we get
(1.5)
is related to the external force, which is zero in most cases. The collision term depends on the detailed collision mechanisms and it is usually difficult to solve. However, one can skip the complicated collision process by imposing the relaxation time approximation.
after some time τ. Then the BTE becomes
(1.6)
is the inverse temperature and ℏω is the phonon energy. This distribution of phonon follows quantum theory, which avoids the BTE framework falling into a complete classical theory. Another possible entry of quantum mechanism into BTE is the relaxation time, which may be derived from quantum scattering theory. We address here that the energy of phonon and the velocity of phonon are now k-dependent quantities.
. This is where the temperature gradient comes in, in analog with the Fourier's law. In the condition of near-equilibrium. Under these approximations, the BTE can be solved as
(1.7)
The thermal conductivity can be obtained from the phonon distribution function, in the same spirit as the kinetic theory, except that now the phonon velocity and relaxation time are all k-dependent
(1.8)
for three-dimensional lattice.
We can find that in Equation (1.5) all the indigents are ready for the calculation of thermal conductivity, except for the relaxation time.
By comparing Equation (1.3) and Equation (1.5) we can find that now the phonon density is replaced by the integration of phonon distribution function in the k-space. Subsequently, the phonon-velocity and phonon energy become k-dependent. Hence BTE theory justifies the contribution of different phonons to the thermal conductivity. The kinetic theory result can be recovered if one treats all phonons in an equal basis.
In BTE theory under relaxation time approximation, the key quantity is the relaxation time. The evaluation of relaxation time relies on other methods, depending on the details of the scattering process. A typical way to calculate relaxation time is the Fermi–Golden rule, by regarding the collision as a perturbation to the normal phonon transport.
Phonon collision or phonon scattering can be caused by various reasons such as boundary scattering, impurity scattering, electron–phonon scattering or phonon–phonon scattering, just list a few. Furthermore, these scatterings can exist simultaneously, and affect each other in a complicated manner. However, to simplify the problem we may treat each scattering as an independent process. Different scattering mechanisms contribute a respective relaxation time. Then the total relaxation time can be evaluated according to Mattiessen's law
(1.9)
is the relaxation time due to scattering process i. One must note this is an empirical equation based on the assumption that all the scattering process are independent. In practice, this law gives great convenience to divide complicated situations into simple pieces of doable tasks, and allows us to analyze different scattering mechanisms individually.
For phonons in a single pure crystal scattering by impurities, Klemens derived the relaxation time can be estimated as [5]:
(1.10)
is the density of state and V is the volume of unit cell. It demonstrates that the impurity scattering has strong frequency dependence. The relation, propose by Callaway and Holland for some fitting constant A, is often used in simple analysis of phonon scattering due to imperfections.
The boundary scattering can be estimated via:
(1.11)
where v is the roughness of the boundary.
The phonons can collide within themselves. Two phonons may collide and combine into one phonon while a single phonon is possible to be splitted into two. These processes are called three phonon processes. The collision term in BTE has rigorous derivation from perturbation theory [8]:
(1.12)
where F is the three phonon interaction strength. The Umkalpp process of three phonon interaction can be approximated as
(1.13)
for some constant a and bis the Debye temperature.
1.2.3 Green–Kubo Formalism Approach
We have demonstrated the calculation of thermal conductivity in the framework of kinetic theory and Boltzmann transport equation. However, both methods are under the single particle approximation, by considering the relaxation time of the energy-carrying particles. In this section, we show that the correlation between the particles actually plays critical role to the thermal conductivity. The thermal conductivity, as well as many other quantities, can be directly evaluated once the correlation functions are known.
Solving the correlation function between the particles is a challenging task in non-equilibrium scenario. Hence the key idea of Green–Kubo formalism is that the non-equilibrium quantities, such as thermal conductivity, can be extracted from its equilibrium properties, under the condition that the system is not too far away from equilibrium. In such near-equilibrium cases, the correlation functions are evaluated according to equilibrium distribution, and we show that these equilibrium correlation functions have direct connection to transport quantities.
and a small perturbation B as
(1.14)
Suppose A is an observable operator as a response function of the perturbation B, then the expectation value of A up to the first order of B can be found to be
(1.15)
where the subscript I indicates the deviation from equilibrium expectation value, caused by the perturbation B. One may define a retarded Green's function as
(1.16)
implies the causality of time. Then the response function can be written as a convolution
(1.17)
By using the Kubo-identity for equilibrium density matrix ρ
(1.18)
we can find the response function can be re-written as
(1.19)
which is the so-called Kubo-formula.
In the above derivation, the perturbation is mechanical, in the sense that it can be written as a Hamiltonian. The question is whether we can apply this method to calculate thermal conductivity, by regarding temperature as a perturbation and thermal current as response function. Here the perturbation is statistical; it makes variation in density matrix, but not Hamiltonian. Hence in order to apply linear response theory, one needs to expand the density matrix and find a Hamiltonian-like perturbation due to temperature gradient. This is not a rigorous step but it is essential to find non-equilibrium quantities like thermal conductivity.
is the local temperature. Assuming that the temperature gradient is small, we can expand the temperature up to the first order of temperature gradient. By doing that we can find an effective perturbation Hamiltonian, which is
(1.20)
We then define the response function as the thermal currents J. By plugging this perturbation Hamiltonian and response operator into Equation (1.17), we can then find that
(1.21)
. We now observe that thermal conductivity is directly related to the current–current correlation function. This equation is called Green–Kubo formula for thermal conductivity.
The Green–Kubo formalism is in general and can be applied to both quantum system and classical systems. In the quantum systems, the current–current correlation function is often reduced to the Green's functions of position correlation functions. The challenges lay in the calculation of the Green's functions, where other methods such as path integral, or Monte Carlo are often used to evaluate the correlators. However, for classical system, we can regard it as a basis for the equilibrium molecular dynamics.
1.2.4 Equilibrium Molecular Dynamics
, Equation (1.19) becomes
(1.22)
This equation is the basis in equilibrium molecular dynamics simulation. The current operator can be written as the summation over local energy
(1.23)
. Integration of this correlation function up to infinity is the major task in the equilibrium molecular dynamics simulation. The correlation function characterize the memory effects of the system, hence it will decay to zero in the limit that time goes infinity. For a typical correlation function, it should decay at a rate comparable to an exponential function. Such behavior is necessary to get a convergent thermal conductance.
The integration of correlation function can be implemented directly using numerical methods to obtain the thermal conductivity. Alternatively, one may also use exponential function to fit the correlation function, so that the integration can be done analytically. This idea has been implemented by using either single exponential fitting or double exponential fitting. In double exponential fitting one distinguishes the relation time of the acoustic mode and optical mode in order to fit the correlation better.
Other than this time-domain calculation, the thermal conductivity can also be evaluated using frequency-domain approach, by using Fourier transformation, as
(1.24)
where J is now written as a vector. Then the heat current spectral density is
(1.25)
If the system is isotropic such that the thermal conductivity is directional independent, then the thermal conductivity is directly related to the spectral density as
(1.26)
where the factor of 1/3 is due to fact that κ only requires the correlation function of current operator at particular direction, which is only a component of vector J.
The accuracy of correlation function is determined by the size of ensembles in the simulation. However, due to finite time simulation, the correlation function becomes less accurate over time and it will be eventually dominated by the numerical errors. Hence we need to introduce a cut-off time to truncate the correlation functions in order to minimize the inaccuracy caused by the numerical noise. Conventionally a first dip method is used, where the correlation function is cut off at the point where it passes zero for the first time. However, Chen et al. found that there exists scenario that the correlation function maintains positive noisy values for a long time before it reaches zero [9]. In such cases the first dip method will overestimate the correlation function with a much larger cut-off time. Hence Chen et al. invented a so-called fist avalanche method to estimate the cut-off according to the fluctuations of the correlation function. Mathematically it can be achieved by defining a quantity
(1.27)
is larger than one for the first time. This method can avoid the accumulated error caused by non-zero noise over long time.
The choice of field potential is an important step in the molecular dynamics simulation and it will directly affect the accuracy of the results. Here we introduce some widely-used field potentials as example.
A commonly implemented potential is the Lennard–Jones (LJ) potential, it has the form of
(1.28)
This LJ potential is a pairwise potential with r describes the distance between two sites. The parameter σ . LJ potential is suitable for gases, simple fluids and importantly, the van der Waals forces in solids. Hence it is widely used for simulations of graphene-substrate interaction recently. However, due to the nature of pairwise interaction, it fails to capture the many-body effects. In material science, the many-body effects can be important in many semiconductors. For such cases, one needs to use many-body potentials.
as
(1.29)
denotes the distance between the ith atom and jth atom. ε and σ are two parameters characterize the energy and length scale. The pairwise and three-body forces are
(1.30)
. The h function is defined as
(1.31)
Such SW potential describes diamond-structured silicon quite well.
1.3 Non-Equilibrium Theory
1.3.1 Non-Equilibrium Green's Function
Non-equilibrium Green's function is a powerful tool to study quantum thermal transport developed in last decades [10]. The term quantum transport
means that the quantum effects become an important factor affecting the transport properties, applied to many scenarios such as low temperature, quantum confinement in one or more dimensions, system with strong quantum correlations and coherences, quantum tunneling or leakage. In such cases, the picture of single-particle transport fails to describe the full physics while the many-body effects start to play a critical role. Many nanostructures actually presents such quantum behavior like quantized thermal conductance. Hence a quantum many-body theory is demanding for the thermal transport as well.
The Landauer Equation
The picture of particle movement and collision is insufficient to describe many-body physics. In Landauer approach, the viewpoint of transport changes and it is the wave-function that travels and follows the Hamiltonian dynamics. When the travelling wave-function travelling from left side meets some barrier, it has some probability T being reflected (Fig. 1.1). The transmission probability is energy-dependent. The distribution of the states is temperature-dependent according to the energy of that state. Hence the total travelled energy can be written as:
(1.32)
is some constant fixing the units and f , describing the situation when the wave-function travels perfectly without any obstacles. Secondly, it guarantees the first law and second law of thermodynamics. Lastly but importantly, the length effect does not come into the picture, implying that the intrinsic quantity turns to be the thermal conductance, rather than thermal conductivity. If one gauges the size parameters in the same way, then the thermal conductivity may become length-dependent. The deep reason is that the Landauer approaches allows ballistic transport of phonons, in contrast to the diffusive transport used in kinetic theory, BTE theory or the Green–Kubo formalism. The information of length is implicitly hidden in the transmission function.
Figure 1.1 Landauer's scattering picture of heat flow. Phonons from left lead will either be transmitted to right lead or reflected back to the left lead.
The Landauer formula provides a new viewpoint for the thermal transport, but we can also find the assumptions it made. One major assumption is that the energy of state is conversed, indicating only the elastic scattering of the wave is allowed. Hence in the following we will rigorous formulate the many-body quantum transport problem from NEGF technique, using but beyond the Landauer's scattering picture.
We consider a set up with a left lead, a center, a right lead and their respective coupling as:
(1.33)
we can find that current can be expressed in terms of Green's functions as
(1.34)
. Detailed derivations can be found in the literature , are the lesser self-energy and advanced self-energy of the left lead, which takes account the effects of leads. We address that this formula, derived by Meir and Wingreen in 1992, is in general and applicable to quadratic or non-quadratic Hamiltonian in the center [11]. The information of temperatures of the leads is hidden inside the less Green's function and less self-energy. The equation reveals that the key quantities to evaluate the thermal current are nothing but the two-point correlation functions, of both leads and center.
There exist various types of Green's functions such as time-ordered, anti-time-ordered, lesser, greater, retarded and advanced Green's functions, with their own properties and applicability [12]. However, these Green's functions depend on each other. In equilibrium states, only one of them is independent. On the other side, in non-equilibrium states, we need two of them. Here for convenience we choose them to be retarded Green's function and lesser Green's function. The retarded Green's function carries the system dynamical properties and the lesser Green's function contains the information of statistics and distributions. These Green's functions are related to self-energies via Keldysh