This action might not be possible to undo. Are you sure you want to continue?

Chapter 21 MOLECULAR DYNAMICS METHOD FOR MICRO/NANO SYSTEMS

Shigeo Maruyama

Department of Mechanical Engineering, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8656, Japan Phone: +81-3-5841-6421, Fax: +81-3-5800-6983, E-mail: maruyama@photon.t.u-tokyo.ac.jp http://www.photon.t.u-tokyo.ac.jp/

21.1. INTRODUCTION

21.2. MOLECULAR DYNAMICS METHOD 21.2.1 Equations of Motion and Potential Functions 21.2.2 Examples of Potential Functions 21.2.3 Integration of the Newtonian Equation 21.2.4 Boundary Conditions: Spatial and Temporal Scales 21.2.5 Initial Condition and Control of Temperature and/or Pressure 21.2.6 Thermophysical and Dynamic Properties 21.2.7 Heat Conduction and Heat Transfer

21.3. MOLECULAR DYNAMICS OF PHASE-INTERFACE AND PHASE CHANGE 21.3.1 Liquid-Vapor Interface 21.3.2 Solid-Liquid-Vapor Interactions

2

21.3.3 Nucleation and Phase Change 21.3.4 Notes on Nucleation and Stability of Nano Bubbles in a Confined System 21.3.5 Crystallization of Amorphous Silicon 21.3.6 Formation of Clusters, Fullerene, and Carbon Nanotubes

21.4 HEAT TRANSFER ISSURES OF CARBON NANOTUBES 21.4.1 Introduction of Carbon Nanotubes 21.4.2 Nanotubes with Fluids 21.4.3 Heat Conduction of Carbon Nanotubes 21.4.4 Thermal Boundary Resistance between a Carbon Nanotube and Surrounding Materials

3

21.1. INTRODUCTION

Molecular dynamics simulations are becoming more important and more practical for microscale and nanoscale heat transfer problems. For example, studies of basic mechanisms of heat transfer such as phase change demand the understanding of microscopic liquid-solid contact phenomena. The efficient heat transfer at a three-phase interface (evaporation and condensation of liquid on a solid surface) becomes the singular problem in the macroscopic treatment. The nucleation theory of liquid droplets in vapor or of vapor bubbles in liquid sometimes needs to take account of nuclei of the size of molecular clusters. The effect of the surfactant on the heat and mass transfer through liquid-vapor interface is also an example of the direct effect of molecular scale phenomena on the macroscopic heat and mass transfer. Even though there has been much effort of extending our macroscopic analysis to extremely microscopic conditions in space (micrometer and nanometer scales), time (microseconds, nanoseconds and picoseconds), and rate (extremely high heat flux), there are certain limitations in the extrapolations. Hence, the bottom-up approach from molecular level is strongly anticipated. On the other hand, recent advances in microscale and nanoscale heat transfer and in nanotechnology require the detailed understandings of phase change and heat and mass transfer in nanometer and micrometer scale regimes. The chemical engineering processes to generate nanoscale structures such as carbon nanotubes or mesoporous silica structures are examples. The wetting of liquid or absorption is also important since the adhesive force is extremely important for micro/nano system and the creation of extremely large surface area is possible with nanoscale structures. The use of molecular dynamics simulations is straightforward for such a nanoscale system. Here, again, it is important to compare such nanoscale phenomena with macroscopic phenomena, because an analogy to the macroscopic system is often an important strategy in understanding a nanoscale phenomenon. Important physics intrinsic to a nanoscale system is usually found through the rational comparison

effective pair potential for water. 2]. However. various examples of potential functions such as the Lennard-Jones potential for rare gas. Because water molecules make hydrogen bonds. The molecular dynamics methods have long been used and are well developed as a tool in statistical mechanics and physical chemistry [1. Various techniques to obtain thermodynamic and thermo-physical properties by equilibrium and non-equilibrium simulations are explained. 8].4 with a macroscopic system. hydrogen-graphite (carbon nanotube). a simulation of liquid water droplets in contact with a platinum solid surface is introduced. suitable choices of potential functions are extremely important for realistic simulations. In addition to the simple Lennard-Jones system discussed in a previous review [3]. the classical molecular dynamics method. and embedded atom potential for solid metals are discussed. many-body potential for silicon and carbon. water-graphite (carbon nanotube) and metal-carbon. In addition to a brief description of the basic concept of molecular dynamics method. . it is a new challenge to extend the method to the spatial and temporal scales of macroscopic heat transfer phenomena [3-6]. solid-liquid-vapor interaction characteristics such as liquid-contact phenomena and absorption phenomena will be discussed from molecular stand point. After the brief introduction to the liquid-vapor interface characteristics such as the surface tension and condensation coefficient. alcohols and organic molecules. the thin film technology related to the semiconductor industry and recent nanotechnology demands the prediction of heat transfer characteristics of nanometer scale materials [7. one of the promising numerical techniques. is overviewed with a special emphasis on applications to inter-phase and heat transfer problems. In this chapter. Depending on the nature of electronic structure or simply the bonding characteristics. predictions of thermal conductivity and the thermal boundary resistance between solid materials and solid-liquid interfaces are discussed. Several examples of potential functions between heterogeneous systems are found such as between water-solid platinum. On the other hand. Finally. Examples of various applications of molecular dynamics simulations follow.

Especially for vapor bubble nucleation. For example. Topics in nucleation and phase change phenomena include homogeneous nucleation of liquid droplets for a Lennard-Jones fluid and water and heterogeneous nucleation of liquid droplets and vapor bubbles on a solid surface for a simple Lennard-Jones fluid.5 the contact phenomena are drastically different from the simple Lennard-Jones fluid. an example of a crystallization process of amorphous silicon is discussed. a direct simulation method using the Monte Carlo technique (DSMC) developed by Bird [10] has been widely used for the practical simulations of rarefied gas dynamics. MOLECULAR DYNAMICS METHOD Knowledge of statistical mechanical gas dynamics has been helpful to understand the relationship between molecular motion and macroscopic gas dynamics phenomena [9]. statistical mechanical treatment of solid-state matters has been well developed as solid state physics [e. 21.. Then.2. After the brief introduction to carbon nanotubes. The water droplet exhibits a finite contact angle on the absorbed monolayer of water film. The contact angle is determined by the surface energy between this monolayer water film and bulk liquid-water film. Recently.g. the nanoscale heat transfer problem related with carbon nanotubes is discussed. interaction of nanotubes with fluids such as the hydrogen-absorption problem and water-nanotube interactions are discussed. the compression of a confined liquid results a large modification of the free energy dependence on bubble radius. In addition. Finally. the molecular dynamics prediction of thermal conductivity along a nanotube and thermal boundary resistance with various environments are discussed. Here a critical discussion of the modification of classical nucleation theory for small confined system is addressed. The nucleation process of molecules such as fullerenes and singlewalled carbon nanotubes (SWNTs) are also discussed as the connection to nanotechnology. 11]. On the other hand. the direct simulation of the Boltzmann equation of phonon dynamics is being developed .

(21. In principal. it is straightforward to numerically solve Eq. saturation temperature and surface tension.2. and inter-phase phenomena can be solved without the knowledge of "thermo-physical properties" such as thermal conductivity. position vector and force vector of molecule i. Once the potential of a system is obtained. when we need to take care of liquid or inter-phase phenomenon. and solid and to interfaces of these three phases. viscosity. The most powerful tool for the investigation of the microscopic phenomena in heat transfer is the molecular dynamics method [e. In principal. solid states.g. (21. latent heat. the molecular dynamics method can be applied to all phases of gas. 21. (21.1). molecular dynamics simulations are expected to directly address phonon-phonon and phonon-boundary scattering models. liquid. This classical form of the equation of motion is known to be a good approximation of the Schrödinger equation when the mass of atom is not too small and the system temperature is high enough compared with the gap of quantum energy levels associated with the dynamics. Eq. For translational and rotational motions of molecules. some quantum corrections are sometimes essential for intra-molecule vibrational motion of small molecules even at room temperature. which is inevitable for phase-change heat transfer. respectively. Here.6 and applied to the heat conduction analysis of thin films [12. ri and Fi are the mass. 13] or nanowires [14].1) can be valid except for light molecules such as hydrogen and helium and except at cryogenetic temperatures. . On the other hand. 1-6]. liquid. any of gas. Furthermore. the equations of motion (Newton's equations) are solved for atoms as mi d 2 ri dt 2 = Fi = −∇ i Φ . Φ is the potential of the system. the statistical mechanics approach is not as developed as for the gas-dynamics or the solid-state statistics.1 Equations of Motion and Potential Functions In the classical molecular dynamics method.1) where mi.

It should be noted that the effective pair potential approximation in Eq.1. and embedded atom models for solid metals are briefly reviewed. this approximation is not used.2) where rij is the distance between molecules i and j. Lennard-Jones Potential The most widely used pair potential for molecular dynamics simulations is the Lennard-Jones (12-6) potential function expressed as ⎡⎛ σ ⎞12 ⎛ σ ⎞ 6 ⎤ φ (r ) = 4ε ⎢⎜ ⎟ − ⎜ ⎟ ⎥ . 21. The intermolecular potential of inert monatomic molecules such as Ne..3) where ε and σ are energy and length scales. potential forms for water and larger molecules.2) is convenient for simple potential systems but cannot be applied for complex potentials such as silicon and carbon.2 Examples of Potential Functions The determination of the suitable potential function is extremely important for a molecular dynamics simulation. r2 .7 The potential of a system Φ (r1 . and r is the intermolecular distance as shown in Fig. Kr and Xe is . respectively. Here. i j >i (21. In the following examples of the potential functions for covalent systems such as carbon and silicon and for the embedded atom model for solid metals. Ar..rN ) can often be reasonably assumed to be the sum of the effective pair potentials φ(rij) as Φ = ∑∑ φ (rij ) . 21. ⎝r⎠ ⎥ ⎢⎝ r ⎠ ⎣ ⎦ (21. the well-known Lennard-Jones potential for van der Waals interactions. (21..2. multi-body potentials for silicon and carbon.

21. or Tc = 163 K and Tt = 82 K with argon properties [17]. Since the modification of the potential makes the long-range correction too complex.3) is cutoff at the intermolecular distance rC = 2. 21. However.8 known to be reasonably expressed by this function with the parameters listed in Table 21. the simple cut-off strategy may be advantageous.2. SPC/E potentials employ the rigid water configuration in Fig. and argon parameters σ = 0. many studies in computational and statistical mechanics have been performed with this potential as the model potential.5σ to 5. respectively. The reduced formulas for typical physical properties are listed in Table 21. An example of the phase-diagram using the equations by Nicolas et al. a contribution to the properties from far away molecules with modified potentials can result in a considerable error as demonstrated for surface tension [19]. The simple form of SPC (Simple Point Charge) [20] and SPC/E (Extended SPC) [21] potentials were introduced by Berendsen et al. and τ = 2. the .2 × 10-12 s are often used to describe dimensional values in order to illustrate the physical meaning. for pressure or stress calculations.1. the equation of motion can be non-dimensionalized by choosing σ.2. Here. (21. several forms of smooth connection have been proposed (for instance [18]). In order to reduce the discontinuity at the cut-off point. ε = 1. Then.3. 16] is useful for designing a simulation. Here.34 nm. For the practical simulation. Effective Pair Potential for Water The effective pair potential form for water has been intensively studied.67 × 10-21 J. molecules are called Lennard-Jones molecules. energy and mass scale.35 and Tt* = 0.68. ε and m as length. The phase-diagram of a Lennard-Jones system [15. Moreover. Here. When a simulation system consists of only Lennard-Jones molecules.5σ. the Lennard-Jones function in Eq. the critical and triplet temperatures are Tc* = 1. the non-dimensional analysis has an advantage of generality over analysis made with the dimensional model. [15] is shown in Fig.

The Coulombic interaction is the sum of nine pairs of point charges.09572 nm and ∠HOH = 104. thermal conductivity. 21. Among these rigid water models. The assumed structure and the distribution of charges are the same as for TIP4P as shown in Fig. The positive point charges q were on hydrogen atoms.4) ⎜ R ⎟ ⎥ ⎢⎝ R12 ⎟ ⎠ ⎝ 12 ⎠ ⎦ i j 4πε 0 rij ⎣ where R12 represents the distance of oxygen atoms.3 with a different length rOM and amount of charge. as in Fig. . potential energy. This potential is regarded as one of the OPLS (optimized potential for liquid simulations) set covering liquid alcohols and other molecules with hydroxyl groups developed by Jorgensen [23].47°. The potential parameters of SPC/E and TIP4P listed in Table 21. SPC/E. Carravetta-Clementi (CC) potentials [24] were based on ab initio quantum molecular calculations of water dimer with the elaborate treatment of electron correlation energy.1 nm and the angle of HOH the tetrahedral angle θt = 2 cos −1 1 / 3 ≅ 109. and σOO and εOO are Lennard-Jones parameters. For CC potentials. specific heat. and the negative charge –2q was set at a distance rOM from the oxygen atom on the bisector of the HOH angle. the interaction of molecules is parameterized as a complex form [24]. φ12 = 4ε OO ⎢⎜ ⎜ ⎡⎛ σ ⎞12 ⎛ σ ⎞ 6 ⎤ q q e2 OO ⎟ − ⎜ OO ⎟ ⎥ + ∑∑ i j .9 negative point charge is on the oxygen atom with the distance of OH just 0.3 were optimized for thermodynamic data such as density. (21. TIP4P and CC potentials are well accepted in recent simulations of liquid water including the demonstration of the excellent agreement of surface tension with experimental results using the SPC/E potential [25].3. 21.52° based on the experimentally assigned value for the isolated molecule. [22] employed the structure of water molecule as rOH = 0. The effective pair potential of molecules at R1 and R2 are expressed as the superposition of Lennard-Jones functions of oxygen-oxygen ( ) interactions and the electrostatic potential by charges on oxygen and hydrogen as follows. evaporation energy. TIP4P potential proposed by Jorgensen et al. and structure data such as the radial distribution function and neutron diffraction results at 25 °C and 1 atm. self-diffusion coefficient.

which requires the iterative calculation of polarization depending on surrounding molecules.. hydrogen atoms are not explicitly modeled but methyl and methylene units are treated as pseudo-atoms.5) where qi and qj are the partial charges on united pseudo-atoms i and j. hydroxyl group..351 D for SPC/E to effectively model the induced dipole moment at liquid phase.4) for TIP4P water. All bond lengths are rigidly constrained to their equilibrium values. it must be always questioned if they are applicable to small clusters or liquid-vapor interfaces. i. TIP4P for water is one of the potentials of this collection. The polarizable potential forms derived from TIP4P [26. As in the case of water molecule. Even though the experimental permanent dipole moment of isolated water is 1. Methyl and methylene groups are labeled C1 through Cn from the tail group. . Potential for Larger Molecules in Liquid Phase (OPLS and AMBER) For alcohol [23] and other larger hydrocarbon molecules with hydroxyl groups [30]. i. the OPLS (optimized potential for liquid simulation) potential developed by Jorgensen is widely used. As mentioned in the previous section. MCY [28] and SPC [29] are used to simulate the structure of small clusters and transition of monomer to bulk properties. Because of all pairs of interaction of the Lennard-Jones term. rij is the distance between these atoms.10 Because all of these rigid water models are “effective” pair potentials optimized for liquid water.85 D. ⎡ ⎡⎛ σ ij = ∑∑ ⎢4ε ij ⎢⎜ ⎢⎜ rij i j ⎢ ⎣ ⎣⎝ A B 12 6 ⎞ ⎛ σ ⎞ ⎤ q q e2 ⎤ ⎟ − ⎜ ij ⎟ ⎥ + i j ⎥ ⎟ ⎜ rij ⎟ ⎥ 4πε 0 rij ⎥ ⎠ ⎝ ⎠ ⎦ ⎦ E AB (21. methyl group. most rigid models employ higher values such as 2. Equation (21. The interaction potential EAB between two molecules A and B is determined by the summation of Coulomb and Lennard-Jones interactions between all intermolecular pairs of sites as follows.e. The direct inclusion of the polarizability to the water models results in the many-body potential.e.5) is a little more complex than equation (21. 27]. to the head group.

and dihedral energies represented with a simple set of parameters. The molecular dynamics simulation code also called AMBER implementing these force fields is also developed by Kollman’s group [35]. the van der Waals interaction represented by a Lennard-Jones potential. Electrostatic and van der Waals interactions are only calculated between atoms in different molecules or for atoms in the same molecule separated by at least three bonds. (21. Cornell et al.7) This combination rule is often used for obtaining Lennard-Jones parameter of unknown combinations. Epair = ∑ K (r − r r bonds i< j eq )2 + ∑ Kθ (θ − θ angles eq )2 + 1 Vn [1 + cos(nφ − γ )] dihedrals 2 ∑ i< j ⎛ Aij ⎛ qi q j Bij ⎞ + ∑ ⎜ 12 − 6 ⎟ + ∑ ⎜ ⎜ Rij ⎟ electro static ⎜ 4πεRij van der Waals Rij ⎝ ⎝ ⎠ ⎞ ⎟ ⎟ ⎠ . the well known combination rule [31] is used as follows. The force field parameters obtained from molecular mechanics and accurate ab initio calculations are summarized as AMBER force field by Kollman et al. [32-34]. The potential field for larger organic molecules and biological molecules such as proteins and nucleic acids are much more complicated.11 In order to obtain the Lennard-Jones potential parameters σij and εij for variety of combinations of molecules. electrostatic interactions modeled by a Coulombic interaction of atom-centered point charges. σ ij = (σ ii + σ jj ) / 2 (21. Many-Body Potential for Carbon and Silicon The pair-potential approximation cannot be used for atoms with covalent chemical bonds such as silicon and carbon. [33] used the following additive potential form. SW potentials for silicon crystal proposed by Stillinger and Weber in 1985 .8) with the bonds and angles represented by a simple diagonal harmonic expression.6) ε ij = ε iiε jj (21.

The total potential energy of a system is expressed as the sum of every chemical bond as Φ=∑ i j ( i< j ) ∑f C (rij ) VR (rij ) − b*ijVA (rij ) . bij = ⎜1 + a n ⎨ ∑ f C (rik ) g (θ ijk )⎬ ⎟ ⎜ ⎟ ⎩k ( ≠i .12) Finally.11) The cutoff function fC(r) is a simple decaying function centered at r = R with the half width of D. respectively.9) expresses the modification of the attractive force VA(r) depending on θijk and the bond angle between bonds i-j and i-k. yield strength and thermal expansion coefficients. VR(r) and VA(r) are repulsive and attractive parts of the Morse type potential. j ) ⎭ ⎠ ⎝ −δ (21. carbon. A simplified form of Brenner potential removing rather complicated ‘conjugate terms’ is widely used for studies of fullerene [41] and carbon-nanotubes.(21. VR (r ) = f C (r ) VA (r ) = f C (r ) De exp − β 2S (r − Re ) S −1 { } } (21. phonon dispersion relations.12 [36] was made of a two-body term and a three-body term that stabilizes the diamond structure of silicon. Tersoff [37-39] proposed a many-body potential function for silicon. 1 ⎧ ⎪ 1 1 ⎡π ⎤ ⎪ f C (r ) = ⎨ − sin ⎢ (r − R) / D ⎥ 2 2 ⎣2 ⎦ ⎪ ⎪ 0 ⎩ (r < R − D ) (R − D < r < R + D ) (r > R + D ) (21.13) . { } (21.9) where the summation is for every chemical bond. the b*ij term of Eq. Brenner modified the Tersoff potential for carbon and extended it for a hydrocarbon system [40]. The Tersoff potential [38] is widely used for simulations of solid silicon. germanium and combinations of these atoms. b = * ij bij + b ji 2 n ⎛ ⎧ ⎫ ⎞ . melting point.10) De S exp − β 2 / S (r − Re ) S −1 { (21. This old potential is still widely used because this potential accurately describes elastic properties. Both the Tersoff potential and the simplified Brenner potential can be expressed as following in a unified form [3].

pair potential functions that are independent of coordination cannot reproduce the realistic potential field. Re and β are dissociation energy. respectively. it is well known that such pair potential functions are far from satisfactory when system is not the perfect crystal [44].15) where De. and force constants as well as elastic properties is achieved with a single parameter set. . Recently.10) and (21.13 g (θ ) = 1 + c2 c2 − 2 d 2 d + (h − cosθ ) 2 (21. the lattice constant and the compressibility for cubic metals (fcc and bcc) can predict reasonable elastic constants [43].4. The parameters calculated by Girifalco & Weizer [43] with energy of vaporization. See Table 5 of the previous review [4]. not as critical as in the covalent cases such as carbon and silicon because d-orbital electrons are a little more flexible. The angle dependence of potential is. The Morse potential is expressed as φ (r ) = De [exp{−2β (r − Re )} − 2 exp{− β (r − Re )}] (21. (21. Pair Potential and the Embedded Atom Method (EAM) for Solid Metal For solid metal crystals. lengths.14) Parameter constants for Tersoff potentials for silicon (improved elastic properties) [38] and carbon and Brenner potentials for carbon are listed in Table 21. This form of Eq. in general. several physical properties can be used to fit to the pair potential functions such as Morse potential and Johnson potentials. a modified ‘second generation empirical bond order potential’ for carbon was developed by Brenner et al. When some surface or interface defects are included.11). [42]. equilibrium bond length and a constant with dimensions of reciprocal distance. (21. See more discussions in the previous review [3].15) corresponds to the special case of S = 2 in the generalized Morse-type functions in Eq. For calculations involving solid metals and alloys. A significantly better description of bond energies.

Finally. the following ‘Embedded Atom Method’ (EAM) was suggested and developed by Daw and Baskes [45-48] as a way to overcome the main problem with two-body potentials. the bond-angle was explicitly handled and the possibility to extend the application range to covalent system such as silicon was explored. See the previous review [4] for a more detailed summary.17). positions are updated in Eq. Unlike the simulation of fluid dynamics. it should be noted that solid metals are always problematic for heat conduction problems due to the luck of the effective technique to handle free electrons in conjunction to molecular dynamics framework.16) ⎛ ∆t ⎞ ri (t + ∆t ) = ri (t ) + ∆tv i ⎜ t + ⎟ 2⎠ ⎝ (21. F (t ) ⎛ ∆t ⎞ ⎛ ∆t ⎞ v i ⎜ t + ⎟ = v i ⎜ t − ⎟ + ∆t i 2⎠ 2⎠ mi ⎝ ⎝ (21. ‘Modified Embedded Atom Method’ (MEAM) was also developed by Baskes group [49-52].3 Integration of the Newtonian Equation The integration of the equation of motion is straightforward.14 In contrast with the “bond order concept” for covalent atoms.2. Later. After the velocity of each molecule is calculated as Eq. they are classical potential functions and no quantum dynamics equation is solved when using the potential functions. 21.16).17) A typical time step ∆t is about 0. More elaborate integration schemes such as Gear’s predictor-corrector method [1] are sometimes employed . The basic concept for EAM is that the potential energy can be calculated by embedding each atom to the electric field made of all other atoms. a simpler integration scheme is usually preferred. It should be noted that ideas of electron density and density functional theory (DFT) are discussed in the formulation of the EAM and MEAM. Here. (21. (21.005 τ or 10 fs with argon property of Lennard-Jones potential. Verlet’s leap-frog method shown below is widely used in practical simulations [1]. however.

21. instability with large wavelength .15 depending on the complexity of the potential function and the demand for the accuracy of motion at each time step. This is not a serious problem for the short-range force such as Lennard-Jones potential. where the contribution from molecules in replica cells is approximated by a sophisticated manner. The implementation of the periodic boundary condition is very simple. which assumes that the system is simply homogeneous for an infinite length scale. On the contrary. Many molecular dynamics simulations have used the fully periodic boundary condition. The equation is the classical rotation equation of motion [1]. Any information beyond a boundary can be calculated with the replica of molecules. classical rotational equations have to be solved for rigid small molecules with such as rigid water potential. Although the principal equation to solve is Eq. On the other hand. the most important points of the design of a molecular dynamics simulation are the boundary conditions. the potential must be cutoff at a distance smaller than half the width of the base-cell scale. In order to avoid the calculation of potential between a molecule and its own replica.2. the efficient SHAKE algorism [1] is often used. which decays as r −6 . such as relatively large photon mean free path. The interactions of molecules are calculated beyond the periodic boundary with replica molecules. Usually. for large molecules expressed with such methods as OPLS and AMBER potentials need to be solved with the constraint dynamics. (21. For such constraint dynamics. the simple cutoff is not possible. the well-known Ewald sum method [1] is employed. if the bond length is fixed.4 Boundary Conditions: Spatial and Temporal Scales Since the spatial and temporal scales handled with the molecular dynamics method is extremely small compared to the scale of the normal microscopic heat transfer problems.1). since Coulombic potential decays only with r −1 . Many problems in heat transfer may include a phenomenon with a larger scale than the calculation domain.

55].16 or a large modulation of properties. the appearance of a new ‘interface’ leads to a significant change of the physics. The Nosé-Hoover thermostat method involves modifying of the equation of motion as . When it is impossible to use the periodic boundary condition. The equilibrium system can often be calculated for constant temperature and constant pressure conditions. The difficulty in the boundary condition is less for gas-phase molecules because the contribution of potential energy compared to kinetic energy is small.2. Remember that this control has to be applied in many steps because of the relaxation of potential energy. a solid wall or a vapor layer can be connected in the realistic fashion. If the potential contribution is ignored. some simple boundary condition such as mirror reflection boundaries can be used by simply changing the velocity component as if a molecule makes an inelastic reflection. 21. the velocity of a randomly selected molecule is changed to a value chosen from the Maxwell-Boltzmann distribution.5 Initial Conditions and Control of Temperature and/or Pressure The initial condition for each molecule is usually assigned by giving the velocity v = 3k BTC / m with random directions for all monatomic molecules. The choice of this interval has a significant influence on the result. The Maxwell-Boltzmann velocity distribution can be obtained after some equilibration calculations. At intervals. Then. Anderson method mimics random collisions with an imaginary heat bath particle. Several different levels of the solid boundary conditions are discussed in detail in the previous review [3]. The simple temperature control of the equilibrium system can be realized by just scaling the velocity of molecules as vi ' = vi TC / T with the current temperature T and the desired temperature TC. More elaborate techniques to realize the constant temperature system are known as the Anderson method [53] and the Nosé-Hoover method [54.

it should be stressed again that all of these techniques to realize constant temperature or constant pressure are artificial and are used to reproduce the statistical ensemble. The system pressure obeys dP / dt = ( PC − P) / t P (21. Ek is kinetic energy of the system. 21. Andersen [53] described a technique to realize the constant pressure simulation. but the physical meaning of the velocity re-scaling is not clear. r ' = χ 1 / 3r . The physical meaning of the modification of position or velocity of each molecule is not clear.20) where βT. 57] extended this technique to change the simulation box shape as well as size for solid crystal. All of these constant temperature techniques have been probed to give a statistically correct canonical ensemble.2. and Q is the thermal inertia parameter. Parrinello and Rahman [56. χ = 1 − βT ∆t (PC − P ) .19) by scaling the box size and position of molecules by a factor χ1/3 at each step. tp (21. The simulation box size V is altered similar to the action of a piston with a mass. Ek0 is kinetic energy corresponding to the desired temperature TC. respectively. tp are the isothermal compressibility and time constant parameter. Berendsen [58] proposed a very simple “pressure bath” technique for the constant pressure simulation.21) . Finally.18) where ζ is the friction coefficient. 2 dt dt dt Q ( 0 ) (21.6 Thermophysical and Dynamic Properties Molecular motion can be related to the thermodynamics properties through the entropy S defined as S = k B ln Ω (21.17 mi dζ 2 E k − E k d 2ri dr = = Fi − ζmi i .

24) There is no simple technique to measure entropy and free energy directly by the molecular dynamics method. The thermodynamics integration or test particle method is employed [2]. the thermodynamics temperature for monatomic molecules is simple expressed as the kinetic energy of molecules. U= 3 NkBT + 2 ∑∑φ (r ) i j >i ij (21. Even though the potential energy and pressure for a homogeneous fluid can be calculated for relatively small system. are fully described in Haile’s textbook [2]. 21. They cannot be defined as the time averages of phase-space trajectories. These properties are often obtained by the statistical Monte Carlo method. P= N 1 kBT − V 3V ∑∑ ∂r i j >i ∂φ ij ⋅ rij (21. there is an important consideration for practical calculation.23) The pressure is defined through the virial theorem as follows.7 Heat Conduction and Heat Transfer The thermal conductivity of homogeneous material can be calculated by three different .22) The internal energy is the combination of kinetic and potential energies. T= 1 3NkB ∑m v i =1 N 2 i i (21.2. The statistical mechanical discussions for the microcanonical ensemble system (NVE ensemble). Since potential energy and pressure are affected from quite long-range distances.18 where Ω is the phase-space volume. The ‘long-range correction’ techniques can be found in the previous review [4]. which is the ‘molecular dynamics’ ensemble. As the final result. but are related to the phase-space volume [2]. it is convenient to consider the potential cut-off corrections.

19 techniques of molecular dynamics: (1) equilibrium molecular dynamics with the Green-Kubo’s formula. The mechanism of heat conduction itself is also interesting [63. The heat flux through a volume is calculated as q= N N N N ⎤ 1 ⎡N 2 ⎢∑ mi vi v i + ∑∑ φij v i − ∑∑ (rij fij )v i ⎥ 2V ⎣ i i j ≠i i j ≠i ⎦ (21. 62]. Because of the third term. the heat conduction behavior of micro. and phonon scattering rates using the molecular dynamics simulations are anticipated [65-68]. all of these simulations should give the same thermal conductivity with adequate simulation conditions [61. For a homogeneous system. 13] or nanowires [14]. 59].and nano-devices are very much of importance. the calculation of heat flux is not trivial at all. It should be noted that sometime both methods (2) and (3) are called ‘non-equilibrium’ molecular dynamics. Simulations of thermal conductivity by the Boltzmann equation of phonon dynamics is being developed and applied to the heat conduction analysis of thin films [12. 64]. The validity of Fourier’s law in an extremely microscopic system such as thin films can only be examined by the direct non-equilibrium heat conduction calculation. phonon mean free paths. represents the energy transfer by the pressure work. The simulation of heat conduction does not always involve simple calculations using thermal conductivity. the direct calculations of phonon dispersion relations. the tensor product of vectors rij and fij. The third term. . Thermal Conductivity of Solids Since most direct heat transfer problems of micro/nano scale are currently related with thermophysical properties of solid materials.25) where the first and second terms related to summations of kinetic and potential energies carried by a molecule i. (2) non-equilibrium molecular dynamics (NEMD) developed by Evens [58. Here. and (3) non-equilibrium molecular dynamics with direct temperature differences.

one-dimensional steady heat conduction was realized using a pair of temperature controlling heat baths. is important only for such a small system. and (4) crystals with different size parameters.20 The non-Fourier heat conduction problems should also be addressed by molecular dynamics simulations [69]. Thermal Boundary Resistance The thermal boundary resistance between solid crystals or at a solid and liquid interface becomes important for small systems. [74] studied the thermal resistance between Lennard-Jones solid crystals by direct non-equilibrium molecular dynamics. Here. The thermal boundary resistance RT is determined from the temperature jump TJUMP and the heat flux qW as RT = TJUMP /qW. Matsumoto et al. it should be noted that the thermal boundary resistance does exist between epitaxially grown interfaces of solids or even solid-liquid interfaces [70-72]. (2) crystals with different interaction parameters. When a constant heat flux is applied across the liquid-solid interface. and hence. The direct molecular dynamics simulation of the thermal boundary resistance and the temperature jump across liquid-solid interfaces can be demonstrated in a simple simulation system [70]. A good review of simulation of heat conduction problems in micro/nano scale solid can be found in reference [8]. Direct experimental measurements of thermal boundary resistance and relevant molecular dynamics simulations are discussed for carbon nanotube-fluid system in a later section. Except for the case (4). This thermal boundary resistance of that investigation was equivalent to a 5~20 nm thickness layer of liquid. The reduction of thermal conductivity by the super-lattice [73] is an excellent example of the importance of thermal boundary resistance. a distinctive temperature jump near the solid-liquid interface can be observed. the observed temperature profiles had . Four types of atomically continuous solid-solid interfaces were investigated: (1) crystals with different masses. (3) crystals with various surface interaction parameters. Inside an fcc crystal.

the density profile. pressure tensor. Fig. the thermal boundary resistance was modeled through the energy reflection rate [75]. (21. When the liquid layer is thick enough. More. Quite accurate predictions of surface tension have been demonstrated for . Ohara et al. and two liquid-vapor interfaces can be realized. the tangential pressure exhibits a sharp drop near the liquid-vapor interface region. from which the temperature jump was measured. practically.1 Liquid-Vapor Interface Prediction of surface tension is one of the benchmarks to examine the applicability of the molecular dynamics method and the potential function model to the liquid-vapor interface.21 a discontinuity at the interface. surface tension in the interface can be calculated as γ LG = ∫ zG zL [PN ( z ) − PT ( z )]dz . Detailed discussions are in the previous review [3]. the bulk property of liquid can be obtained at the central region. Though the normal pressure keeps the constant value.3. With similar molecular dynamics simulations of Lennard-Jones crystals. By taking a time average. The typical simulation system is the liquid slab between vapor regions. and surface tension can be reasonably predicted.3. The reduction of heat flux due to the temperature jump was larger than that predicted with a simple acoustic model. MOLECULAR DYNAMICS OF PHASE-INTERFACE AND PHASE CHANGE 21.26) where z is the coordinate perpendicular to the interface and PN and PT are normal and tangential components of pressure. 21.4 shows an example of such interface calculated with 8000 Lennard-Jones molecules. have analyzed the possible explanation of the thermal boundary resistance [71] based on an “intermolecular energy transfer” concept developed by the detailed studies of heat conduction in liquid phase water [63] and a Lennard-Jones fluid [64]. 21. The integrated value of this tangential pressure drop is the surface tension.

the bulk property of liquid is expected at the central region. 80] pointed out that this macroscopic concept could not be directly converted to the molecular scale concept and that the ‘molecular exchange’ process must be considered. The definition of the radius of a droplet is not straightforward. since the size of the droplet is normally very small and the liquid-vapor interface has a certain width. Tsuruta et al. should require a completely different approach because such small clusters do not have the well-defined central liquid part assumed in the statistical mechanical discussions. Matsumoto et al. Thompson et al. At the other extreme. Through the detailed studies of the liquid-vapor interphase phenomena of argon. The determination of the condensation coefficient by molecular dynamics simulations is also a very fascinating task. The well-known Young-Laplace equation relates the curvature of a liquid-vapor interface and surface tension to the pressure difference.22 Lennard-Jones fluid [75] and water [76] by integrating the difference of normal and tangential components of pressure tensor across the surface. The condensation coefficient has been simply defined as the ratio of rates of the number of condensation molecules to incident molecules. Tsuruta et al. Another typical benchmark configuration is the liquid droplet surrounded by its vapor [3]. water. Molecular dynamics simulations of the effect of surfactant in liquid-vapor interface are also . which may be important in the nucleation theory. the surface tension for very small clusters. [79. [78] used the spherical extension of Irving-Kirkwood's formula to calculate the normal pressure profile. Recently. [82. Roughly a thousand molecules are enough to calculate the reasonable value of the bulk surface tension for argon. When the size of the droplet is large enough. On the other hand. 83] successfully modeled the simulated condensation coefficient by a ‘transition state theory’ similar to chemical reaction dynamics. [81] have reported a significant dependence of the trapping rate on the normal velocity of incident molecules. and methanol. The equimolar dividing radius is one of the choices but more elaborate discussions based on the statistical mechanics are found in the literature (see references in [3]).

23

being developed. An example is the simulation of the effect of n-alcohols in the liquid-vapor interface of water and of LiBr solution by Daiguji [84].

21.3.2 Solid-Liquid-Vapor Interactions

Lennard-Jones Model System

Solid-liquid-vapor interaction phenomena have played a very important role in phase-change heat transfer. The importance of the liquid wetting to the surface is apparent in dropwise condensation, high-heat-flux boiling heat transfer, and capillary liquid-film evaporators. There are good reviews of the connection between microscopic and macroscopic views of the wetting phenomena by Dussan [85], and from a slightly more microscopic point of view by Koplik and Banavar [86]. Furthermore, molecular dynamics treatment of simple Lennard-Jones liquid droplet on the solid surface and Lennard-Jones vapor bubble on the solid surface are discussed in the review [3]. In brief, except for the two or three liquid layers near the surface, the averaged shape of the Lennard-Jones liquid droplet is close to semi-spherical [87-89]. A snapshot of a droplet on the solid surface and the 2D density distribution are compared in Fig. 21.5 [89]. By fitting a circle to the density contour disregarding a few layers of liquid near the solid surface, “contact angle” can be measured. The effect of the interaction potential on the shape of the liquid droplet is summarized in Fig. 21.6 [87-90]. The cosine of measured contact angle is linearly dependent on the strength of the surface potential. The contact angles measured from the droplet and the bubble are almost consistent as several images of 2-D density profiles in the insets of Fig. 21.6. Here, as the measure of the strength of the interaction, the depth of the integrated potential

ε SURF = (4 3π / 5)(σ INT 2 / R0 2 )ε INT

(21.26)

was employed [3]. The controversial discussions whether the Young’s equation can hold or not in such

24

system has been historically discussed (see references in [3]). However, at least the concept of surface energy balance should be valid. For the macroscopic contact, the well-known Young’s equation relates the contact angle to the balance of surface energies as cosθ =

γ SG − γ SL γ LG

(21.27)

where γSG, γSL and γLG are surface energies between solid-gas, solid-liquid, and liquid-gas, respectively. With a change in the energy parameter between liquid and solid molecules, γLG is kept constant and γSG does not change much compared with γSL because the vapor density is much lower compared with liquid density. With increase in εSURF, γSL should decrease almost proportionally because of the gain in potential energy in liquid-solid interface. Then, (γSG - γSL) should increase and the Young’s equation predicts the increase in cosθ. Hence, the linear relation in Fig. 21.6, roughly speaking, is consistent with Young’s equation.

Water Droplets on a Platinum Solid Surface

As a more practical simulation of solid-liquid-vapor interaction, a water droplet on a platinum solid surface is shown in Fig. 21.7 [90]. Water molecules were modeled with the SPC/E potential. The interaction potential between water and platinum molecules was expressed by the potential function developed by Zhu and Philpott [91], based on the extended Hückel calculations. The potential function consists of a water-molecule-conduction electron potential, an anisotropic short-range potential, and an isotropic short-range r-10 potential as follows.

**φH O−surf = φH2O−cond + φan (O; rO ) + φisr (O; rO ) + ∑ [φan (H; rH ) + φisr (H; rH )]
**

2

(21.28)

H

φH2O−cond = ∑

l ,k

ql qk (21.29) 2rlk

25

6 3 ⎡⎛ ⎞ ⎛ ⎞ ⎤ σ p − Pt 2 σ p − Pt 2 ⎟ −⎜ ⎟ ⎥ ⎢⎜ φan ( p; rp ) = 4ε p − Pt ∑ ⎜ αρ pj )2 + z pj 2 ⎟ ⎜ (ρ pj α )2 + z pj 2 ⎟ ⎥ j ⎢⎝ ( ⎠ ⎝ ⎠ ⎦ ⎣

(21.30)

φisr ( p; rp ) = −4ε p − Pt ∑

j

c p − Ptσ p − Pt rpj

10

10

(21.31)

where α = 0.8, σO-Pt = 0.270 nm, εO-Pt = 6.44×10-21 J, cO-Pt = 1.28, σH-Pt = 0.255 nm, εH-Pt = 3.91×10-21 J, cH-Pt = 1.2. Here, Eq. (21.29) represents Coulomb potential between the point charge of a water molecule and its image charge located at symmetrical position below the image plane. This potential takes the minimum value when a water molecule sits on top of a platinum atom heading hydrogen atoms upward. The contact structure of the water droplet with a finite contact angle on the absorbed monolayer water film in Fig. 21.7 was quite a surprise. It was the first demonstration of such a structure with equilibrium molecular dynamics simulation. Even though the existence of a thin liquid film is sometimes assumed in the liquid-solid contact theories, the simple energy consideration using the Young’s equation should conclude that the water droplet should be very wettable to a water film. However, the contact angle is determined by the surface energy between this monolayer water film and bulk liquid water. The monolayer film does not have the characteristics of bulk water. It has quite high density due to the strong interaction to the platinum surface atoms, and the interaction of this film with normal water strongly depends on the density of the film. The higher density of the water monolayer results the lower interaction. With this assumption, the clear dependence of contact angle on the platinum crystal lattice structure can also be explained [91].

21.3.3 Nucleation and Phase Change

Homogeneous Nucleation

Homogeneous nucleation is one of the typical macroscopic phenomena directly affected by

and the energy scale parameter was changed to reproduce various wettabilities. Their similar simulation [94] for water of TIP4P potential at 350 K resulted the nucleation rate two orders of magnitude smaller than the classical nucleation theory. have demonstrated the direct molecular dynamics simulations of the nucleation process for Lennard-Jones fluids [93] and for water [94]. On the other hand. 21. This cooling method mimicked the carrier gas of supersonic jet experiments. After the equilibrium condition at 160 K was obtained. the condensation latent heat must be removed for the successive condensation. The nucleation rate was seven orders of magnitude larger than the prediction of classical nucleation theory. Yasuoka et al. [93] used 5000 Lennard-Jones molecules for the simulation mixed with 5000 soft-core carrier gas molecules connected to a Nosé-Hoover thermostat for the cooling agent. After quenching to the supersaturation condition. Then larger clusters grew preferentially near the surface for wettable cases. just in good agreement with the “pulse expansion chamber” experimental results. The phantom molecules modeled the infinitely wide bulk solid kept at a constant temperature with proper heat conduction characteristics. Through the detailed study of growth and delay of nuclei size distribution. Yasuoka et al.26 the molecular scale dynamics. Argon vapor consisting of 5760 Lennard-Jones molecules was suddenly cooled by the solid wall controlled with the phantom technique [3]. for the less wettable condition. Fig. small clusters were appeared and disappeared randomly in space. The potential between argon and solid molecule was also represented by the Lennard-Jones potential function. relatively large clusters grew without the help of surface just like the . temperature of the solid surface was suddenly set to 100 K or 80 K by the phantom molecule method. Heterogeneous Nucleation Heterogeneous nucleation is also important in many heat transfer problems. Initially.8 shows an example of the heterogeneous nucleation of liquid droplet on a solid surface [95]. they have estimated the nucleation rate and the critical size of nucleus. Recently.

the stability of a bubble in a confined space is compared with usual classical nucleation theory.32) 4 3 πr . A is surface area = 4πr2. For r larger than the critical value rcrit. However. the stability of a vapor bubble depends on the critical radius of the bubble.4 Notes on Nucleation and Stability of Nano Bubbles in a Confined System Direct molecular dynamics simulation of nucleation of bubbles in liquid can be possible and a nanoscale bubble can be stably realized in a confined system. the net change in Gibbs free energy is expressed as 4 ∆G∞ = γA + ∆gVgas = (4πγ )r 2 + ( π∆g )r 3 . According to the classical nucleation theory. Vgas is volume difference between liquid and vapor. A molecular dynamics simulation of homogeneous [96. Figure 21. Appearance of a vapor bubble with radius r in a supersaturated liquid results the reduction of free energy of ∆gVgas and the increase of surface energy as γA . 3 (21. and ∆g is free energy 3 where γ is surface tension. ∆G∞ increases with r2 through the surface energy term. the stability of a nanoscale bubble confined in small space is quite different from that in an open space. (21. ∆G∞ . When r is small. Even though the formation of a vapor bubble can be reproduced by expanding the liquid to the negative pressure. The negative value of ∆g at supersaturated condition is the driving force of the nucleation of a bubble.27 homogeneous nucleation. The nucleation rate and free energy needed for cluster formation were not much different from the prediction of the classical heterogeneous nucleation theory in case of lower cooling rates. Hence. 21. 97] or heterogeneous [90] nucleation of a vapor bubble is much more difficult compared to the nucleation of a liquid droplet. qualitative comparison of nucleation rate or critical radius is not easy.9 shows ∆G∞ in Eq. Here.32).3.

4 3 ∆G = ∆G∞ + ∆GL = γA + ∆gVgas + ∆g LVliquid = (4πγ )r 2 + ( π∆g )r 3 + ∆g L ( L3 − πr 3 ) 3 4 (21. Here. as in the case of usual molecular dynamics simulations.9 shows the expected approximate shape of ∆GL and ∆G = ∆G∞ + ∆GL . a bubble smaller than rcrit is unstable and should shrink. Further increase in bubble size corresponds to the considerable compression of liquid. the schematic shape ∆GL in Fig. Fig. The free energy change can be approximately express by a function of liquid density from ρ ∆g L = f ( L ) = ρ sat ⎛V f ⎜ sat ⎜V ⎝ L ⎞ ⎟= ⎟ ⎠ ⎛ 3 4 3 ⎞ ⎜ L − π rsat ⎟ 3 ⎟ f⎜ ⎜ L3 − 4 πr 3 ⎟ ⎜ ⎟ 3 ⎝ ⎠ (21. The increase in density of liquid causes a large increase in free energy ∆GL . When the bubble size becomes such that the saturated liquid and vapor co-exist with radius rsat. Hence. 21. It is observed that ∆G = ∆G∞ + ∆GL becomes minimum value when the radius is slightly .34) The density ratio ρ L / ρ sat can roughly related to the ratio of intermolecular distance (ρ L / ρ sat ) . A nano-bubble confined in a constant small volume. 1/ 3 Then. the free energy change in the confined system should contain this ∆GL term as follows.33) This additional term ∆GL should be negligible when liquid volume is large enough. ∆GL should become minimum. it is assumed that ∆GL should gradually decrease with increase in bubble size since the liquid density increases from the negative pressure condition. The free energy term from the confined liquid should be quite important.9 was drawn by assuming that the free energy should be proportional to the shape of Lennard-Jones potential energy.28 decreases with r3. but a bubble grown larger than rcrit by the statistical fluctuation should stably keep growing. hence a rapid increase in ∆GL is expected. Hence. 21. (21. leading to the increase in liquid density. The volume occupied by the liquid should decrease with the growth of a vapor bubble. should be controlled by another free energy term in addition to that of Eq.32).

21.3. Then. With the solid phase epitaxy (SPE) configuration [e. the newly developed crystallization system that consisted of a Si cluster on an amorphous surface with a small number of seed atoms was also simulated. ∆G∞ + ∆GL ' should have no minimum value.6 Formation of Clusters.. By using the Tersoff (C) potential [38]. 21.g. the crystal growth in (001) and (111) directions were compared. 21. Fullerene. and Carbon Nanotubes . it is still possible to approximately simulate the nucleation process beyond the critical radius and critical free energy. With further smaller system. However.3. It was observed that growth in (111) direction was dominant in Si crystallization at the high temperature region possibly because of instability of (001) amorphous/crystal (a/c) interface.5 Crystallization of Amorphous Silicon An example of a crystal growth process from amorphous silicon is shown in Fig. the crystallization process was calculated. Simulated bubbles in a confined system should be stable with this minimum point. 21.29 larger than rsat. When the confined system size is large enough. However. the artifacts induced by the periodic boundary condition could be completely neglected. An example with smaller rsat is shown as ∆GL ' and ∆G∞ + ∆GL ' in Fig.9.10 [98]. The same argument is also applicable to a system with a liquid droplet surrounded in its vapor. 99]. since the change of free energy of vapor with density is not as large as for liquid. With this configuration. the confinement effect should be much stronger. The simulated stable bubble should be much smaller than the critical size for free space and the simulated free energy barrier is much smaller. From the time profile of the Si crystal nuclei size. a reasonable simulation of the nucleation process is possible with relatively small system. when the system size is smaller or the degree of supersaturation is less. the classical critical nuclei size was estimated to be around 110 atoms.

The molecular dynamics simulation [102] of laser-oven or arc-discharge processes is very much similar to the simulation of endohedral metallofullerene formation. In catalyst CVD process. alumina. Figure 21. the completely random vapor mixture of 500 carbon atoms and one of the Ni clusters obtained above were allocated in 20 nm cubic fully-periodic simulation cell. As the initial condition of clustering process. After saturation within about 2 ns. all carbon atoms attached to the exposed surface are absorbed in the metal cluster. 100] and endohedral metallofullerene formations [101]. Using the simplified Brenner [41] potential for carbon-carbon interaction and the empirical metal-carbon potential functions [101]. the nucleation process of SWNT by successive collisions of hydrocarbon molecules to a metal cluster was simulated. 21. SWNTs formation mechanism for CVD techniques [103-105] is getting more attention because they are expected to supply large amount of SWNTs at a lower cost. Some of the carbon networks separate from the surface of the particle and in specific cases some cap structures appear. Recently.11 (a)). magnesia or zeolite. The formation mechanism of SWNTs is theoretically interesting and also extremely important for the development of large scale and controlled generation techniques. hexagonal carbon networks are formed inside the metal-carbon binary cluster (Fig. as if the sphere surface of the metal . hydrocarbon or alcohol [107] will be decomposed on the catalytic metal cluster supported by silica. At first. In both process. carbon atoms are supplied to the metal cluster at a high temperature condition. the initial Ni cluster structures of various sizes were obtained by annealing a fcc crystal at 2 ns in 2000 K. In the first stage. The interaction between isolated carbons was represented by the Lennard-Jones potential to avoid the generation of carbon clusters before the collision to the metal catalyst.30 The intermediate cluster structure are extremely important when some of them can be stably observed as the end products such as the simulation of fullerene formations [41. CO [106]. because the actual experimental condition is almost the same except for the small amount of metal doped in the graphite material.11 shows the nucleation process of a SWNT in a catalytic CVD process [103].

1 Introduction of Carbon Nanotubes Since their discovery by Iijima [108] in 1991. 21. 21.11 (f)). Furthermore. The geometry of an SWNT can be described as a single layer of a graphite sheet that is rolled up into a cylinder. Many applications such as field effect transistors.11 (b)). 111]. single electron transistors. When separation of the carbon network occurs. (21. 21. 21. and mechanical properties.11 (c)). and composite materials for high mechanical and thermal properties have been examined.11 (g. optical. After the cap coalesces into annular graphitic protrusion (Fig.11 (e)). electron field emission sources. single-walled carbon nanotubes (SWNTs) [109] show great potential for use in a wide variety of applications [110.35) . 21. 21. Continuous supply of carbon atoms leads to the formation of annular graphitic protrusion (Fig. m) . The diameter of the tube varies from 0. optical switches. 21. f)). carbon nanotubes shown in Fig.4 HEAT TRANSFER ISSUES OF CARBON NANOTUBES 21. The geometry can be specified by the chiral vector C h = na1 + ma 2 ≡ (n. which can be regarded as an initial stage of the growth process of SWNT (Fig. the larger carbon network covers the surface with a certain curvature (Fig. chemical sensors.” Due to their novel electronic.11 (d)). 21. nano-wires.4.7 to about 3 nm and the length is about a few microns to even a few centimeters. supersaturated carbon atoms inside the Ni cluster gradually lift up the carbon-shell surface leading to a half-cap structure that appears after 40 ns (Fig.12 have been regarded as one of the most important materials in “nanotechnology. This lifting-up of the carbon cap results in formation of its stem. the area of bare metal surface decreases but some of the open areas are preserved especially in the crystalline part where additional carbon atoms can still flow.31 cluster played as a template for cap formation (Fig.

Recently. The electronic density of states shows divergences called Van Hove singularity. By rolling the carbon sheet up so that two atoms at the origin and at Ch in the honeycomb lattice will meet. Even though absorption of other molecules is quite interesting. One example is the hydrogen absorption with SWNTs. Most of studies of SWNTs have started after the discoveries of macroscopic generation techniques. Details of properties of carbon nanotubes are describes in monographs [110. (1997) [112]. the laser-furnace method in 1996 and the arc-discharge method in 1997. 21. Many of electronic and optical applications are based on these molecular-like DOS structures. An alternative catalytic CVD method using alcohol as the carbon source was introduced by the author’s group [107]. 111]. Hydrogen Absorption with Single-Walled Carbon Nanotubes Since the suggestion of high efficiency storage of hydrogen with SWNTs by Dillon et al. the bulk generation of SWNTs by the catalytic CVD process called HiPco [106] was developed. the geometry of a SWNT is uniquely determined by the chiral index (n. experimental determinations of the storage capacity and mechanism of storage have .32 where a1 and a2 are unit vectors of the honeycomb lattice of the graphite sheet.2 Nanotubes with Fluids Contact of fluid with a nanotube is an important problem to be solved with molecular dynamics simulations.4. The electronic property of SWNT is most exciting that it may be metallic or semiconductor depending on the chirality. Contact of water with SWNTs is also quite interesting. Hence. m). the hydrogen case has been the focus of many simulations and experiments because of the technically important problem of hydrogen storage for fuel cell hydrogen source. Ch will be the circumference of the SWNT.

The potential energy has minimum at the interstitial sites (between nanotube). An example of molecular dynamics simulation of hydrogen absorption is shown in Fig. with parameters: εHH = 3.and vapor. the liquid. or 4.2 wt % at 10 MPa at room temperature [114].14 shows a snapshot of a water cluster in a SWNT with quite a large diameter of 2. σHH = 2.77 nm.179 Å.762 meV and σHC = 3.13 [118]. Hydrogen storage capacity of high purity SWNT was reported to be more than 8 wt % at 12 MPa and 80 K [113]. with a surface tension of 73 mNm-1. Therefore. should wet carbon nanotube. it is expected that water. Even though the contents of this aqueous solution are unclear. 21.interface structure and phase-change in the completely closed space of about 50 nm are truly exciting. The interaction potential is too weak for the efficient hydrogen storage at room temperature for the use in fuel-cell vehicles. Most of molecular dynamics simulations have been consistently predicting the very low hydrogen storage capacity at room temperature [116.33 been extensively studied. potential forms between H2-H2 and C-H2 were both expressed by Lennard-Jones (12-6) functions.180 meV. Another experimental work [121] suggests that liquids with a surface tension lower than 100-200 mNm-1 can wet a carbon nanotube. The interaction potential between carbon and water is simply . and has a low value at the endohedral sites (inside a nanotube). 120] shows the trapped aqueous solution in MWNTs generated with a hydrothermal method. 117]. Recent experimental studies are concluding that the hydrogen storage capacity of SWNTs at room temperature is less than 1 wt % [115]. Water in Carbon Nanotubes Some recent experimental works [119. Each nanotube was regarded as rigid molecule and the van der Waals potential between nanotubes was derived as a Lennard-Jones (8-4) function by integrating the potential function between carbon atoms. Figure 21.928 Å. The solution was quite wetting to the walls of carbon nanotubes. εHC = 2. Assuming the physical adsorption of hydrogen to the surface of carbon nanotubes.

Koga et al.316 nm. the wall of nanotube is strongly hydrophobic. and no periodic boundary condition was applied in order to simulate the finite length effect. phonon density of states and phonon dispersion relations were extracted from the simulated .3 Heat Conduction of Carbon Nanotubes A heat conduction simulation along a carbon nanotube is shown in Fig. 123] studied the wetting phenomena of SWNTs more in detail by molecular dynamics simulations. The thermal conductivity of nanotubes. In addition to the Lennard-Jones potential between oxygen and carbon. Maniwa et al.16 and the heat flux obtained by the integration of the additional force by the phantom molecules. 21. Walther et al. [125] showed consistent experimental results later.08×10-21 J and σOC = 0.15 [126. The power-law behavior of thermal conductivity suggests the divergence for infinitely long system as discussed by Lepri et al. Just like graphite. [129. which was speculated to be higher than any other material along the cylindrical axis [128] was measured from the simulations for nanotubes with several different chiralities and lengths. 127]. [122. 21. The interaction parameters are εOC = 1. The preliminary result showed that the thermal conductivity was about 200 ~ 600 W/mK but was strongly dependent on the nanotube length.4. It is clearly seen that the water cluster is not very wetting but the potential energy is lower than in vacuum. As the first step to connect the molecular dynamics simulation to the phonon concept approach. [124] found an ice-nanotube structure in an SWNT by molecular dynamics simulations. The simulation results show that pure water does not wet a carbon nanotube. The temperature at each end of a nanotube was controlled by the phantom technique [3]. 130]. 21.34 Lennard-Jones function between carbon and oxygen atom. The thermal conductivity was calculated from the measured temperature gradient as in Fig. The potential function employed was the simplified Brenner bond order potential [41]. a quadrupole interaction between the carbon and the partial charges on hydrogen and oxygen atoms was also considered.

21.38) Rα ' (k . t) of atoms in a line along the z-direction separated by a = 3ac−c (ac-c: bond length) are used. 128 data points in the z direction multiplied by 8192 temporal points were used. The calculated density of states is shown in Fig.4. t ) exp(ikz − iωt ) Here.38) of the displacement vector r’(z. and the heat transfer from an SWNT to water are introduced here for examples [132]. ω ) = ∫ dt rα ' ( z .4 Thermal Boundary Resistance between a Carbon Nanotube and Surrounding Materials In addition to the thermal conductivity along an SWNT. In this chapter. t) from the equilibrium position re(z). z for each velocity component in the cylindrical coordinate. Position vectors r(z. 21. The phonon dispersion relation obtained from eigenvalues and eigenvectors of the dynamical matrix were made with the force-constant tensor scaled from 2 D graphite [131] as shown in Fig. t ) − re ( z ) . The heat transfer rate can be well expressed by the thermal boundary resistance. α takes r. The phonon density of states was measured as the power spectra of velocity fluctuations in Eq. (21. the heat transfer from an SWNT to other SWNTs.36) Here.17(d) as the comparison.37) (21. r ' ( z .35 trajectories. 21. heat transfer from a nanotube to the surrounding material is an important issue for practical applications using carbon nanotubes as electrical devices and composite materials. The photon dispersion relations were also directly measured as the time-space 2-D Fourier transforms in Eq. or the thermal conductance through the boundary. In the case of Fig. k is the wave vector along z-axis. For this calculation. Dα (ω ) = ∫ dt exp(− iωt ) vα (t )vα (0) (21.36). 21. 8192 temporal data points saved every 5 fs were used. θ.17 as the right hand side inserts of phonon dispersion relations. t ) = r ( z . (21. (21.17. The thermal conductance of a nanotube junction. interfacial thermal .

In this case a (12. At the beginning of the computation. This temperature jump can be modeled by assuming that there is a virtual boundary between two nanotubes with different structures.693 nm diameter. The geometrical structure of the SWNT was the armchair type (5. After that. 21. 6) armchair nanotube were smoothly connected using 5-membered and 7-membered rings at the junction. Thermal Boundary Resistance of SWNT Bundles As the initial condition. Thermal Conductance at an SWNT Junction A junction of nanotubes with different chiralities is shown in Fig. the heat transfer from the central tube to surrounding tubes is clearly .18. Fig. The thermal boundary resistance RT is simply calculated as RT = 1/K. The thermal conductance at the virtual interface is defined as the heat transfer through the boundary. the temperature distribution was measured as in the bottom panel of Fig.18. the whole system was kept at 300 K for 100 ps. the temperature of the central SWNT was suddenly increased to 400 K using the velocity scaling method for 10 ps.19(a) shows a time history of the temperature of the hot (central) tube and cold (surrounding) tubes. where K is the thermal conductance. By applying different temperatures at each end. the cross-sectional area was defined as πbd [132]. 21.36 conductance K is used.39) The thermal conductance at the junction is calculated as 1. K ≡h= Q . A∆T (21. a bundle of 7 SWNTs arrange as in Fig. all the temperature controls were turned off. Here. hence. 21. The temperature jump at the junction is clearly observed. 21.4×104 MW/m2K. Here. 0) zigzag nanotube in the left-hand side and a (6.12 (b) with the length of 5 nm were placed in a 5×6×6 nm simulation cell. Then. 5) with 0.

⎡ ⎛ 1 1 Thot − Tcold = T0 exp ⎢− ⎜ ⎜ρ c V + ρ c V ⎢ ⎝ hot hot hot cold cold cold ⎣ Bi = hL ⎤ ⎞ ⎟ KSt ⎥ .41) can be adopted since the characteristic length of an SWNT is extremely small.37 observed. Then. The diameter of SWNT is about 1 nm and the Biot number in Eq.41) λ (21. 10) SWNT with length 20. water molecules and the SWNT were equilibrated at temperature of 300 K. If an SWNT is considered to be a solid material and heat transfer from the central tube to surrounding tubes is expressed by heat transfer coefficient or thermal conductance at the boundary.19 is understood by the concept mentioned above. ⎟ ⎥ ⎠ ⎦ (21.118×10×10 nm fully-periodic simulation cell in the similar configuration as in Fig. (21. The thermal conductance between SWNTs in a bundle [132] was calculated to be about 15 MW/m2K. (21.14. Thermal Boundary Resistance between an SWNT and Water One (10. 21. (21. (21.42) The excellent agreement to an exponential fit by Eq. 21. only the temperature of the SWNT was suddenly heated up to 400K.19(b). The monotonic decay of the temperature difference in Fig.42) becomes extremely small. the thermal conductance K was estimated. (21. ⎛ t Thot − Tcold = T0 exp⎜ − ⎝ τ ⎞ ⎟. 21. In order to examine this heat transfer.41).40) in Fig. the lumped capacity method in Eq. Comparing Eq. 21.40) with Eq.7 [ps]. At the initial stage of simulation. temperature differences between the central and surrounding tubes are drawn in Fig.40) where T0 = 875 [K] and τ = 29. After applying the heat .19(b) was well approximated by an exponential function. ⎠ (21.118 nm containing 192 water molecules were prepared in the 20.

38 for 1 ps. [133] reported nanotube-SDS interfacial thermal conductance as K = 12 MW/m2K by picosecond transient absorption experiments. physical mechanism of thermal conductance is being explored through molecular dynamics simulations [135]. all temperature control was turned off. The interfacial thermal conductance is calculated to be 8 MW/m2K using the lumped capacity method similarly to the case of SWNT bundle simulation [132]. Shenogin et al. Furthermore. [134] obtained K = 20 MW/m2K for octane-nanotube interface by their molecular dynamics simulations for long enough nanotubes. Huxtable et al. Very recently. .

2. J. Phys. 1998. Handbook of Heat Exchanger Update. New York.13. Molecular Dynamics Simulation.. Haile. no. W.. pp. New York. G.. Thermophys. Eng.. Curtiss and R. Eng. John Wiley & Sons. 2000. pp. New York. Molecular Dynamics Methods in Microscale Heat Transfer. C. G. P. B. Quantitative Validation of the Boltzmann Transport . Begell House. R. Taylor & Francis. J. 2000. Rev. A. D. Goodson. B. Molecular Theory of Gases and Liquids. Washington D. Kaviany. vol. vol. Kittel. [6] D. pp.39 References [1] M. Poulikakos. 4. W. Molecular Dynamics Study in Microscale Thermophysical Engineering: Current Activities and Future in Japan. Molecular Dynamics Simulations in Nanoscale Heat Transfer: A Review. S. no.. 1954. vol. [13] A. Bird. Majumdar. ed. no. J. [10] G. Maruyama. Thermophys. Advances in Numerical Heat Transfer. Molecular Dynamics Method for Microscale Heat Transfer. New York.. Thermophys. Ohara. 7th ed.13. Introduction to Solid State Physics. 189-226. 2. pp. no. McGaughey and M. 3. Tildesley. [4] S. Bird. 1996.7-1-2. J. 2003. F. Oxford University Press. New York. [2] [3] J. no. John Wiley & Sons. [7] A. Majumdar. [9] J. pp. C. M. 2000. Mahan. Sparrow. H. Ford. [8] D. G. New York. Appl. Nanoscale Thermal Transport. 57. [11] [12] C. [5] T. 213-221. Minkowycz and E. S. 1987. 4. Arcidiacono and S. Hirschfelder. Micro. 1992. Maruyama. Computer Simulation of Liquids. J. K. 23. E. 77-82. Chen. pp. H. Thermal Conductivity and Ballistic-Phonon Transport in the Cross-Plane Direction of Superlattices. Phys. Eng. Maris. 2. Micro. 793-818. A. Micro. Merlin and S. pp. R. M. Nanoengineering: Promises and Challenges. John Wiley & Sons. 181-206.7-33. 7. 4. Phillpot. O. Oxford University Press. 2003. 14958-14973. 93. Molecular Gas Dynamics and the Direct Simulation of Gas Flow. vol. Maruyama. Cahill. 1994. vol. vol. K. 2002. Allen and D. 2.

Predicting the Thermal Conductivity of Si and Ge Nanowires. Pullmann). 926-935. Klein. Streett and D. van Gunsteren and J. 3. Optimized Intermolecular Potential Functions for Liquid Alcohols. vol. Jorgensen. D. vol. [18] S. Chem. Hansen and L. [20] H. P. R. vol. Numerical Experiments on the Stochastic Behavior of a Lennard-Jones Gas System. in Intermolecular Forces (ed. R. J. W. pp. pp. Bruin and J. 1969. A. F. 1980. 8. 1429-1454. 12. W. Nanoletters. 89. Tildesley.. Reidel. Gubbins. Berendsen. Chem. Impey and M. Phys. Nicolas . W. Rev. 094303-1-094303-12. Chem. Jorgensen. H. 1979. J. 69. 10. Rev. L. Stoddard and J. 1987. vol.. J. pp. Phase Transitions of the Lennard-Jones System. C. Phys. Majumdar. [19] M. 79. 1981.6269-6271. L. Madura. C. 73. Sikkenk. 1988. Carravetta and E. no. J. Grigera and T. Verlet. [23] W. Water-Water Interaction Potential: An Approximation of the . 1276-1284. 6. J. A Molecular Dynamics Simulation of the Lennard-Jones Liquid-Vapor Interface. Comparison of Simple Potential Functions for Simulating Liquid Water. Clementi. Postma. 37-5. Phys. no. vol. Ford. [16] F. J. 1986. pp. Phys. Phys. 3789-3792. pp. E. pp. Chem. Berendsen. J.. no. 5401-5403. D. [15] J.. Analytic Representation of Thermodynamics Data for the Lennard-Jones Fluid. p. B. 24. J. Rev. K. Phys. Mingo. J. L. D.. pp. Nijmeijer. 1983. Equation of State for the Lennard-Jones Fluid. no. Molecular Physics. 1973. B. vol. 184. 1713-1716. 2003. A. B. [14] N. P.40 Equation Phonon Thermal Conductivity Model Under the Single-mode Relaxation Time Approximation. M. [17] J. J. C. Chandrasekhar. J. pp. [24] V. no. [22] W. Bakker. P. L. H. J. [21] H. 1504-1512. J. Li. Lee. Dordrecht. Phys. Hermans. F. 151-161. 90. 2004. 2. Yang. vol. vol. and A. Straatsma. vol. J. vol. Phys. P. pp. Chem. pp.. 331. 91. The Missing Term in Effective Pair Potentials.

Cornell. Soc. E. P. vol. Weiner.. Theory of Simple Liquids. no. Ghio. D. [30] W. [29] D. and Liquid-Vapor Interface of Water with Many-Body Potentials. G. Chem. J. M. pp. Tildesley and G. Academic Press. Lett. vol. Ferguson. Kollman. Am. no. 6. 2646-2651. 1990. A New Force Field for Molecular Mechanical Simulation of Nucleic Acids and Proteins. C. pp. 20. no. Jr. Phys. 372. Merz. Case. J. 7949-7956. G. An Anisotropic Polarizable Water Model: Incorporation of All-Atom Polarizabilities into Molecular Mechanics Force Fields. Swenson. 2003. D. 1994. vol. [25] J. D. P. 1984. I. Hansen and I. Optimized Intermolecular Potential Functions for Liquid Hydrocarbons. 81. Jr. L. pp. K-Jespersen and R. Madura and C. [27] C.. R. T. [28] U. J. Fox. Dang and T-M. Caldwell and P. no. J. Chem. 102. X.41 Electron Correlation Contribution by a Function of the SCF Density Matrix. Gould. C. Am. Molecular Dynamics Simulation of the Orthobaric Densities and Surface Tension of Water. Phys.. 98. 6638 (1984). N. 842-847. Corongiu. J. Alagona. Chem. [26] L. 19. D. R. Chem. K. Molecular Dynamics Study of Water Clusters. W. Alejandre. Cieplak. pp. 4574-4583. Flexible TIP4P Model for Molecular Dynamics Simulation of Liquid Water. 8149-8159. Niesar. A. S. Levy. Chapela. J. 1986. pp. Bhattacharya. Chang. [32] S. Y. A. Phys. [31] J.. Weiner. K. Soc. Profeta. A Second Generation Force Field for the . J. Lawrence and J. C.. Chem.. Clementi. Skinner. Chem. 106... Kollman. D. D. 106. 1997. Chem. vol. Molecular Dynamics Simulations of Liquid Water Using the NCC ab initio Potential. 765-784. Kneller and D. M. Phys. J. 1984. G. vol. Bayly. Phys. J. and P. I. A. Bernardo. L. J. vol. 2nd ed. Jorgensen. J. Ding.. London. [33] W. K. Phys. R. C. Liquid. U. pp. 106. pp. A. P. McDonald. 94. 1995. 11. vol. Spellmeyer. Singh. J. 4180-4187. M. P. Chem.

Modeling Solid-State Chemistry: Interatomic Potentials for Multicomponent Systems. H. Shenderova. Tsui. no. San Francisco. S.9458-9471. H. [39] J. Stillinger and T. B. Harrison. Rev. B. 14. Simmerling.42 Simulation of Proteins. J. no.A. Rev. Stuart. vol. Cheatham III. J.E.A. Lett. [37] J. Radmer. A Molecular Dynamics Simulation of the Fullerene Formation Process.L. [35] D. A. Tersoff. vol. O. 8. 6991-7000. J. no.S. 3-4. pp. and Organic Molecules. A. Merz. Pitera. W. vol. G. Chem. Molecular Mechanical Models for Organic and Biological Systems Going Beyond the Atom Centered Two Body Additive Approximation: Aqueous Solution Free Energies of Methanol and N-Methyl Acetamide. Cheng. W. pp. Rev. no. B. 31. 286.. 1989. Empirical Interatomic Potential for Silicon with Improved Elastic Properties. Singh. Vincent. Am. Caldwell and P.C. [36] F. New Empirical Approach for the Structure and Energy of Covalent Systems. Maruyama. Empirical Potential for Hydrocarbons for Use in Simulating the Chemical Vapor Deposition of Diamond Films. Soc. Rev. Phys. Computat. 1048-1057. 1990. 1998. 1995. pp.M. A. Wang. Crowley. Yamaguchi and S. 12. [42] D. pp. Chem. vol. University of California. 2001. P. 117. K. R. Weiner and P. 8. Caldwell. pp. V. Nucleic Acid Base.K. [40] D. D. pp. 42. Brenner. Pearlman. 5262-5271. 1985. T. J. vol. 22. vol.J. 39. Weber. Kollman. T. Phys. A Second-Generation Reactive Empirical Bond Order (REBO) Potential Energy Expression for . C..A. J. Tersoff. 336-342.A. Y. AMBER 7.V. [34] P. pp. Darden. and Amide Hydrogen Bonding and Chloroform/Water Partition Coefficients of the Nucleic Acid Bases.L. J.. M. 1988. 5566-5568. Massova. J. Phys. 9902-9905. Kollman (2002). Stanton. Ni and S. Seibel. Phys. Sinnott. no. 37. W. [41] Y. B. Case. Gohlke. J.W. Phys. Computer Simulation of Local Order in Condensed Phase of Silicon. Rev. A. Ross. B. 38. R.L. B. Chem. vol. Tersoff. Duan. 1988. [38] J. Cieplak. I. Nucleic Acids. pp. Brenner. Phys. U. J.J. vol. 5179-5197. B.

S. and Their Alloys. [48] M. Baskes. J. 2666-2669. 1983. vol. Atomistic Calculations of Composite Interfaces. pp. Rev. Phys. pp. 14. Mater. F. Matter. Anderson. 6085-6100. B. M. Phys. B. 33. 1984. [54] S. and M. 511-519. Modelling Simul. Semiempirical Modified Embedded-Atom Potentials for Silicon and Germanium. pp. 81. vol. Molecular Dynamics Simulations at Constant Pressure and/or Temperature. 1989. 2727-2742. 295-321. A. Surfaces. pp. Wright. Application of the Embedded-Atom Method to Covalent Materials: A Semiempirical Potential for Silicon. 505-518. 72. 114. Phys. I. Phys. Rev. 1973.. J. [50] M. [51] M. Baskes.. vol. [44] R. no. Modified Embedded Atom Potentials for HCP Metals. pp.. Chem. pp. pp. Rev. Lett.. and Other Defects in Metals. 4. Johnson. 2384-2393. Rev. Baskes. Daw. Pt. Modified Embedded-Atom Potentials for Cubic Materials and Impurities. S. Johnson. I. 2. S. I. 46. Eng. 7983-7991. 2002. S. Daw and M. 1987. pp. Rev. 1992. I. . M. pp. 1994. 1984. 59. A. vol. 1. pp. vol. 783-802. Nelson and A. Baskes.: Condens. 40. I. [45] M. Quantum Mechanical Calculation of Hydrogen Embrittlement in Metals. Phys. 1980. 1986. Phys. no. 29. vol. Phys. pp. vol. Foiles.. Phys. vol. I. B. [49] M. Daw and M. Rev. [47] S. J. J. Embedded-Atom Method: Derivation and Application to Impurities. vol. pp. Lett. Mater. 1959. Rev. B. 2. 3. I. Modelling Simul. 147-163. vol. Baskes. vol. Ni. vol. pp. 6443-6453. [52] M. Eng. Weizer.. Girifalco and V. [53] H.. Pd. Angelo and C. J. Ag. C. Phys. Baskes. A. Chem. Sci. Empirical Potentials and Their Use in the Calculation of Energies of Point Defects in Metals. J. G. vol.. Baskes. E. F: Metal Phys. Au. [43] L. 1994. Phys. Sci. Application of the Morse Potential Function to Cubic Metals. L. Baskes and R. A Unified Formulation of the Constant Temperature Molecular Dynamics Methods. Embedded-Atom-Method Functions for the fcc Metals Cu. 50. [46] M. 1285-1288. 687-690. Bisson. Phys. Nosé. Semiempirical.43 Hydrocarbons. I.

Micro. R. Phys. London 1990. 37th Japan Heat Transfer Conf. 111.. P. J. 52.. Phys. vol. Chem. 86. Eng. J. Phillpot and P. A. Ohara. Haak. 6492-6500. [58] H. . [56] M. 9667-9672. [64] T. vol. Proc. Chem. Phys.. MD Simulation and Phonon Analysis for Heat Conduction in Solids. 1982. 457-460. Schelling. J. Parrinello and A. Contribution of Intermolecular Energy Transfer to Heat Conduction in a Simple Liquid. J.-C. 1985. [63] T. vol. Rahman. Intermolecular Energy Transfer in Liquid Water and Its Contribution to Heat Conduction: An Molecular Dynamics Study. Thermal Conductivity of the Lennard-Jones Liquid by Molecular Dynamics Calculations. Tien. 31. no. vol. 7182-7190. Rev. Parrinello and A. Crystal Structure and Pair Potentials: a Molecular Dynamics Study. Phys. J. Hoover. Ciccotti.. [59] D. Statistical Mechanics of Non Equilibrium Liquids. Evans. Thermophys. Wakabayashi. no. pp. Phys. Chem. T. Postma. pp. B. J. Rahman. 1987. 1984. R. 111. 133-137. pp. 1999. M. Phys. pp. 1981. Appl. Comparison of Atomic-Level Simulation Methods for Computing Thermal Conductivity. [60] D. van Gunsteren. Lett.44 [55] W. C. 91A... 1999. 144306/1-144306/12. pp. Vogelsang. Ohara. [61] R. F. L. pp. 45. Academic Press. J. Berendsen. Phys. J. pp. Phys. Homogeneous NEMD Algorithm for Thermal Conductivity: Application of Non-Canonical Linear Response Theory. Matsumoto. K. pp. Lukes and F. [65] C. Rev. Komiyama. vol. M. vol. 81. [57] M. pp. vol. pp. C. and J. 1196-1199. pp. Chem. 3. Lett.. vol. J. 1980. Polymorphic Transitions in Single Crystal: a New Molecular Dynamics Method. Phys. DiNola. Evans and G. Hoheisel and G. W. S. 65. J. 8. 1998. 3. 2.. Morriss. 1695-1697. Keblinski. Molecular Dynamics Simulation of Thermal Transport in Solids. Canonical Dynamics: Equilibrium Phase-Space Distributions.Chou. G. vol. R. vol. A. [66] M. 2002. Molecular Dynamics with Coupling to an External Bath. P. vol. [62] P. Rev.. Makino and H. 3684-3690. 6371-6375.

[73] B. vol. Xue. Phys. J. 1988.1. vol. Eng. J. Wakabayashi and T.. F. no. Evaluation of the Phonon Mean Free Path in Thin Films by using Classical Molecular Dynamics. U. Phys. vol. 155-167. Kor. 247-249. Lee. Kor. Makino. Joulain. . A Study on Thermal Resistance over a Solid-Liquid Interface by the Molecular Dynamics Method. Thermophys. pp. A. 2. 2000. J. J. 2004.-H. Thermal Sci. no. S. Lee. 1919-1925. 6. 118. 316-317. Ser. 1. 2003. no. Nijmeijer. vol. Micro. 2003. P.. 189-196. 2. Soc. Phys. vol. no. J. 43. no.. Daly and H. 1999. 44. Bakker. H. 671. 2004. [71] T. 2002. Thermal Resistance of Crystal Interface: Molecular Dynamics Simulation. 317-325. Eng. P. S. Chem. [68] S. 3789-3792. pp.. Maruyama. Kor. 2000. no. R. pp. 68. 2004. Choi. S. vol. Suzuki. Calculation of the Thermal Conductivity of Superlattices by Molecular Dynamics Simulation. Phys.. [76] M. Volz. Intermolecular Energy Transfer at a Solid-Liquid Interface.. vol. Trans. Phys. pp. [67] S. 747-753. Matsumoto. 4.. Soc. K. A.-K Kim and J. J. Eng. Carminati and K. Chem. H.45 975-976. Choi and S.-H. [69] S. Soc. 3. [72] L. Choi. J. Sikkenk. Maruyama. [70] S. Kimura. B. K. Maris. [75] S. C.-K Kim and J. Eastman. pp. vol. no. 5. R. JSME. Keblinski. C. Bruin and J. Physica B.-H.-H. Variations in the Thermal Conductivity of Insulating Thin Films with Temperature and Pressure. Micro. Phillpot. Maruyama and T. 89. pp. pp. S.-H. Thermal Response of Silicon Crystal to Pico-Femtosecond Heat Pulse by Molecular Dynamics. 2002. Thermophys. S. 8.. [74] M. no. Ohara and D. pp. Feasibility Study of a New Model for the Thermal Boundary Resistance at Thin Film Interfaces. Two Regimes of Thermal Resistance at a Liquid-Solid Interface. A Molecular Dynamics Simulation of the Lennard-Jones Liquid-Vapor Interface. 897-906. no. 4. 45. pp. Choi and J. vol. Maruyama. 7. 63-68. vol.

. K. Chantry and J. [86] J. pp. 101. New York. R... pp. [79] K.. J. Phys. Maruyama and H. pp. no. B. Tsuruta. 11. A Molecular Dynamics Simulation of a . S. Chem. A Generalized Expression for the Condensation Coefficient Based on Transition State Theory and Molecular Dynamics Simulation. Yasuoka and Y. P. Tanaka. 3.46 [77] J. International Center Heat Mass Transfer Symposium. 1392-1399. Tildesley and G. On the Spreading of Liquids on Solid Surfaces: Static and Dynamic Contact Lines. no.. 11. 1. pp. [82] G. Thompson. A Molecular Dynamics Study of Liquid Drops. Phys. no. 1995. K. Methanol. Evaporation and Condensation at a Liquid Surface. vol. 257-292. vol. Ann. II. M. vol. 371-400. Chem. Kataoka. 530-542. E. vol. vol. 9. Ann. Rowlinson. [78] S. 229-240. Matsumoto and Y. pp. Koplik and J. 1994. J. Gubbins. Chem. Saruwatari. R. Dussan. Nishio). 1736-1743. [87] S. 102.Chapela. Phys. pp. I. Yasuoka. 108.. pp. Evaporation and Condensation at a Liquid Surface. 1538-1549. 118. Tsuruta.1984. M. Molecular Dynamics Studies on the Condensation Coefficient of Water. Fluid Mech. [81] T. 81. no. 1997. vol. no. 1994. Chem. Phys. [85] V. 9. Yokohama. K. Begell House. pp. Nagayama and T. Banavar. Kataoka. 2001. Nagayama. [84] H. Rev. J. Phys. J. 2003. B. J. no. Matsumoto. J. 2004.. 1979. Chem. A. Condensation Coefficient and Interphase Mass Transfer. Walton. 7904-7911. vol. H. J. Rev. 101. Continuum Deductions from Molecular Hydrodynamics. Tamashima and T. Matsumoto. Fluid Mech. vol. International Symposium on Molecular and Microscale Heat Transfer in Materials Processing and Other Applications (ed. Phys. Tanasawa and S. Alejandre. Molecular Dynamics Simulation of the Orthobaric Densities and Surface Tension of Water. Molecular Dynamics Study of n-Alcohols Adsorbed on an Aqueous Electrolyte Solution. R.. pp. Phys. S. [80] M. vol. Argon. A. R. Daiguji. J. 115. 4. Chem. 7911-7917. 4574-4583. B. 1995. J. Chem. Masuoka. pp. I. Tsuruta and G. E. [83] T. D. 27.

T. Maruyama. Maruyama and T. 12th Int. no. pp. pp. 1. Thermophys. 2. Chem. Matsumoto. Phys. Wu and C.1..47 Liquid Droplet on a Solid Surface. Micro. Kimura and M. J. Phys. Joint Conf. II. Maruyama. vol. Vol. Eng. [91] T. pp.. no. Matsumoto. 109. 2000. Kinjo and M. pp. 7. 19. pp. 2002. no. 1995. 10. [97] Y. Y. [88] S. 49-62. Maruyama and T.-B. Grenoble. S. vol. Maruyama. Proc. Maruyama.. 6. Kimura.. 38th . Cavitation Processes and Negative Pressure. 109. Eng.. vol.. A Molecular Dynamics Simulation of Bubble Nucleation in Homogeneous Liquid under Heating with Constant Mean Negative Pressure. J. 1998.S. pp. Fluid Phase Equilibria.A. pp.Zhu and M.Yasuoka and M. 3-13. 1. no. no. vol. A Molecular Dynamics Simulation of a Bubble Nucleation on Solid Surface. Water. Philpott. T. W. 23-29. 1995. A Molecular Dynamics Simulation of Heterogeneous Nucleation of a Liquid Droplet on Solid Surface.. Matsumoto. [93] K. Kimura. Micro. Chem. 2002. vol. pp. no. Proc.-C. 69-74. 137-151. Sci. Kimura and S. 1994. 1998. no. 9. MD Simulation of Crystallization Process of Silicon. 6961-6968. Micro. 8451-8462. pp. Int. 8. Heat Transfer Conf. I. Chem. Thermophys. Molecular Scale Aspects of Liquid Contact on a Solid Surface. J. [95] T. Molecular Dynamics of Homogeneous Nucleation in the Vapor Phase. Kurashige. Maui. [94] K. 537-542. Interaction of Water with Metal Surfaces. 19. Proc. 8463-8470. Eng.6. Lennard-Jones Fluid. vol. [90] S. vol. pp.. 2003. Matsumoto. 100. Pan. [89] S. March 19-24. R. pp. Molecular Dynamics Simulation of Water Droplet in Contact with Platinum Surface. Yamaguchi and T. Thermophys. 557-562. 1998. 144. Kimura and S. [92] S. Heat & Technology. vol. Liquid Droplet in Contact with a Solid Surface. Inoue... J. Therm. 1998. vol. Yasuoka and M. [96] T. 2002. Phys. U. Lu. 343-350. Molecular Dynamics of Homogeneous Nucleation in the Vapor Phase. [98] S. Eng. 2. ASME-JSME Thermal Eng.

Helical Microtubules of Graphitic Carbon. Phys. 108. 2001. Smith and R. Phys. Chem. 3. Phys. Lett. H. A. 187-189. Kojima. 2004. 1800-1805. Bolton. The Role of the Catalytic Particle Temperature Gradient for SWNT Growth from Small Particles. Gilmer. 229-234. 385-388. pp. vol. Y. [99] G. Lett. Willis. Maruyama. 2004. J. [107] S. [108] [109] S. 393. Colbert. vol. no. 1-4. Lett. Phys. A. R. Nature. J. pp. 17369-17377. pp. Miyauchi. J. 1998. 56-58. A. 286. 2002. H. 343-349. no. 2001.. Phys. K. 1998. pp. 3-4. vol. E. [105] F. Bolton and A. Technol. Smalley. Thin Film Deposition: Fundamentals and Modeling. Low-Temperature Synthesis of High-Purity Single-Walled Carbon Nanotubes from Alcohol. S. pp. . Physica B. vol. Sci. Chem.4. Ichihashi. no. Chem. Ding. [103] Y. Maruyama. [102] Y. vol. Shibuta and S. pp. Chem. 354. Ding. pp. Huang and C. Rosen. 343-344. 19. J. Yamaguchi and S. Maruyama and Y. Sci. 12. 309-313. Nucleation and Growth of Single-Walled Carbon Nanotubes: A Molecular Dynamics Study. vol. 382. D. 603-605. Single-Shell Carbon Nanotubes of 1-nm Diameter. pp. pp.. T. B. Molecular Dynamics Simulation of Generation Process of SWNTs. vol. A Molecular Dynamics Demonstration of Annealing to a Perfect C60 Structure. P. 2002. 1999. Comp. 9. pp. 1991. [100] S. Euro. Yamaguchi. 1993. Lett.. 363. Shibuta and S. Vac. vol.48 Japan Heat Transfer Conf. Maruyama. no. Molecular Dynamics Simulation of Formation Process of Single-Walled Carbon Nanotubes by CCVD Method. 3-4. A. vol.. Kohno. Iijima and T. Mat. [106] M. Bronikowski.. A Molecular Dynamics Study on the Formation of Metallofullerene. 360. Gas-Phase Production of Carbon Single-Walled Nanotubes from Carbon Monoxide via the HiPco Process: A Parametric Study. Chem. vol. Maruyama. Iijima. Nature. Chiashi and M.. S. 354-380. [101] Y. Rosen and K. 2003. pp. K. D. 323-1-4. 381-386. Phys. Roland. [104] F.

P. Smalley. 405-409. Lett. Johnson. J. London. Gototsi. J. A. Molecular Dynamics Simulation of Hydrogen Storage in Single-Walled Carbon Nanotubes. Bekkedahl. T. ASME Heat Transfer Division 2000. Science of Fullerenes and Carbon Nanotubes. K. Carbon. no. D. G. 352-358. M. G. Dresselhaus. 1996. 39. C.79. 377-379. Phys. Liu. Imperial College Press. Liu. vol.. vol. 2001. [117] K. 320. Appl. A. G. Hydrogen Storage Capacity of Carbon Nanotubes. Proc. 577-586. Bethune and M. Chem. vol. Lett. Kimura. Academic Press.. Cong. [111] R. Storage of Hydrogen in Single-Walled Carbon Nanotubes. S. Monte Carlo Simulations of H2 Physisorption in Finite-Diameter Carbon Nanotube Ropes. J.. Dresselhaus and M. M. Phys. Colbert.A.A. J. K. S. C. Eklund.49 [110] M. pp. 2000. Maruyama and T. Orlando. Fan. Meisner and C. Fultz. and Vapor-Grown Fibers. Heben. T. Y. [118] S. 2. 2291-2301. Chem. vol. Molecular Simulation of Hydrogen Adsorption in Single-Walled Carbon Nanotubes and Idealized Carbon Slit Pores. Libera. Dresselhaus and G. 1997. 2000. 74. 1998. Cheng and M. New York. pp. C. G. vol. 16. M. pp. H. [112] A. Filaments. Lett. Appl. pp. Ahn. [119] Y. A. 1999. 1999. . Y. Dresselhaus. Smith and R. Saito. B. 1. C. Hydrogen Storage in Single-Walled Carbon Nanotubes at Room Temperature. Physical Properties of Carbon Nanotubes. Megaridis. H. Williams and P. no. no 7.M. Jones. A. Ye. Science. Dresselhaus.C. [114] C. 1127-1129. Güvenç-Yazicioglu and C. 286. Rinzler. no. Witham. 2307-2309. pp. vol. Kiang. C. H. H. vol. Tibbetts . Nature. pp. Dillon. D. 110. S. Phys. vol. 386. E. 1021-1023. [115] G. no. [116] Q. pp. S. Olk.G. Liu. pp. Hydrogen Adsorption and Cohesive Energy of Single-Walled Carbon Nanotubes. In Situ Multiphase Fluid Experiments in Hydrothermal Carbon Nanotubes. 1999. [113] Y. Wang and J. 5.. 27. Phys. K.

vol.50 2001 [120] C. Phys. Walther. 2000. H. vol. Structural Characteristics and Energetics. 2001. 1850-1852. Unusually High Thermal Conductivity of Carbon Nanotubes. J. [130] R. Livi and S. vol. Molecular Dynamics Simulation of Contact Angles of Water Droplets in Carbon Nanotubes. J. Y. [129] S. pp. Carbon Nanotubes in Water. 84. Tanaka and X.14. Kwon and D. Tanigaki. [128] S. Phys. Lepri. vol. no. Lett. G. Phys. Attoliter Fluid Experiments in Individual Closed-End Carbon Nanotubes: Liquid Film and Fluid Interface Dynamics. Gototsi. Abe. [124] K. pp. Nature. A. vol. J. 1. 9980-9987. T. Nature. 327. 193-195. 4613-4616. R. [126] S. Berber. H. Chem. pp. H.105. 697-702. Eng. vol. Suzuki. 2002. 1-4. Fluids. A.. Hiura and K. T. 2001. B. Kataura. 2002. Phys. Walther. 265. Science. Maniwa. Memory Effects and Heat Transport in One-Dimensional Insulators. pp. no. Heat in One Dimension. Thermophys. Jpn. 2002. [122] J.L. 71. Koumoutsakos. Phys. M. and Y. H.. Jaffe. Halcioglou and P. vol. M. B. Koga. 2863-2866. S. 1994. Libera. Y. Achiba. 41-50. Güvenç-Yazicioglu. A Molecular Dynamics Simulation of Heat Conduction of Finite Length SWNTs. Formation of Ordered Ice Nanotubes inside Carbon Nanotubes. Koumoutsakos. pp. Ebbesen. Micro. Gao. [123] T. 18. pp. [125] Y. J. Tománek. Soc. A Molecular Dynamics Simulation of Heat Conduction of a Finite Length Single-Walled Carbon Nanotube. W.. vol. T. F. 2002. 421.. [127] S.-K. Megaridis. pp. Capillary and Wetting of Carbon Nanotubes. no 12. T. Noca and P. 323. no. no 2. Dujardin. Phase Transition in Confined Water inside Carbon Nanotubes. [121] E. 2001. pp. pp.412. K. Maruyama. Jaffe. . Halicioglu. Matsuda. 802-805. pp. vol. R . Eur. vol. pp. G. Physica B. J. 2000. 441-446. Werder. 12. Maruyama. Nanoletters. H. H. Kiba and K. 1. 7. Rev. C. Zeng. Lepri. 2003. vol. L5-L8.

G. pp. pp. Keblinski and D. Chem. S. Taniguchi and J. Shiomi. Molecular Dynamics Simulations of Heat Transfer of Carbon Nanotubes. Xue. pp. [135] K. Igarashi. G. 95. R. vol. M. Bolton and S. Phys. to be submitted. Cahill. vol. Siddons. Maruyama.. Strano. Phys. pp. G. M. Keblinski. R. S. Y. Usrey. 2. Saito. Dresselhaus. Y. Ozisik. Shim and P. Role of Thermal Boundary Resistance on the Heat Flow in Carbon-Nanotube Composites. S. 1998. 57. Interfacial Heat Flow in Carbon Nanotube Suspensions. Shenogin. P. vol. Cahill. 731-734. T.51 [131] R. 2003. vol. [133] S. Ozisik. M. Takeya. Appl. Shenogin. [134] S. Rev. 8136-8144. 161-170. T. P. Phys. L. . J. T. D. Xue. Eng. Raman Intensity of Single-Walled Carbon Nanotubes. Gustavsson.. Huxtable. 291. Thermophys. [132] S. Micro. Barone.. L. 4145-4153. G. 2004. Dresselhaus and M. Energy Transfer Mechanisms in Gas-Carbon Nanotube Collision. B. Kimura. 2003.. Nature Mat.

50×10-21 1.52 Table 21.274 0.67×10-21 2.365 0.340 0.398 ε [J] 0.2 121 163 232 .20×10-21 ε/kB [K] 36.1 Parameters for Lennard-Jones potential for inert gas molecules. σ [nm] Ne Ar Kr Xe 0.25×10-21 3.

Property Length Time Temperature Force Energy Pressure Number density Density Surface tension r* = t* = T* = f* = φ* = P* = N* = ρ* = γ* = Reduced Form r/σ t/τ = t(ε/mσ2)1/2 kBT/ε fσ/ε φ/ε Pσ3/ε Nσ3 3 σ ρ/m 2 γσ /ε .2 Reduced properties for Lennard-Jones system.53 Table 21.

316 6 1.847 6 e rOM qHa qM a Charge of electron e = 1.423 8 e -0.54 Table 21.52 0.315 4 1.47 0.015 0.04 e [nm] [°] [nm] [J] [nm] [C] [C] 0.3 Potential parameters for water.077 2 0.100 109.095 72 104.079 7 0 0.60219×10-19 C .52 e -1. SPC/E rOH ∠HOH σOO -21 εOO ×10 TIP4P 0.

5724×10-7 7.5 1.1315 1.384 -5.55 Table 21.9825×10-1 0.4 Parameters for Tersoff potential and Brenner potential.1447 1.2295 1.2751×10-1 1/(2n) 3.80469 19 2.285 0.185 0.7058×10-1 0.325 0.8734×10-1 1/(2n) 1.1000×10-6 7. Tersoff (Si) De [eV] Re [nm] S -1 β [nm ] Tersoff (C) 5.195 0.640 1.29 1.4316 14.015 Brenner (C) 6.656 1.6660 0.5769 19.0039×105 1.8049×104 4.6217×101 -5.5 -1 0.1644 0.015 2.015 A N δ c d h R [nm] D [nm] .1304×10-2 1 0.

86 2.848 1.758 0.009 101 69 23 - .112 135 110 79 49 - 0.43 2.29 1.527 0.5 Calculation conditions of contact of Lennard-Jones fluid.404 0.56 0.99 3.228 0.56 Table 21.688 0.581 0. Droplet Bubble ε * SURF εINT [10-21J] θ [deg] εINT [10-21J] θ [deg] E0 E1 E2 E3 E4 E5 0.935 1.73 1.

Lennard-Jones (12-6) potential.5σ Intermolecular Distance. .1. r Figure 21. φ(r) ε r 0 –ε 6 2σ 2σ σ 1.57 2ε Potential Energy.

2 1 0.6 3 0.05 0 –0.1 –0.1 0.4 0.2 0.3 –0.4 0.2. Nicolas et al.2 0.15 Temperature T* = kT/ε 0.4 1.5 0.2 –0.8 0.3 0.4 –0.6 0 0.8 Number density N* = Nσ Figure 21.58 1. Phase diagram of Lennard-Jones fluid based on the equation of state by .5 1.6 0.

CC. . SPC/E. Classical rigid models of water for 4-sites and 3-sites models. TIP4P.59 +qH O -qM rOH rOM +qH ∠HOH Figure 21.3.

4.5 K in 6×6×30 nm box).60 Number Density [1/nm ] 3 0 40 20 –20 0 0 10 z [nm] 20 30 Figure 21. Tangential Pressure [MPa] . Density and tangential pressure distribution of a flat liquid-vapor interface (8000 argon molecules saturated at 99.

.000 40 20 0 0 20 40 60 Radius (Å) 80 100 120 (b) Density profile Figure 21.5.61 (a) snapshot 80 0.025 60 Height (Å) 0. A snapshot of a liquid droplet on solid surface compared with the two-dimensional density profile: 32000 Lennard-Jones molecules on a surface.

e). .62 50 50 40 Height [Å] 30 20 10 Droplet Height [Å] (d) E4 40 30 20 10 0 0 (a) E4 E4 Contact angle Hc/R1/2 (=cos θ) 1 Bubble E5 0 0 50 40 Height [Å] 30 20 10 0 0 10 20 30 Radius [Å] 40 (b) E2 E2 E4 0 10 20 30 Radius [Å] 40 50 (e) E2 E2 40 Height [Å] 10 20 30 Radius [Å] 40 E3 30 20 10 0 0 50 40 Height [Å] 30 20 10 (c) E1 E1 E1 10 20 30 Radius [Å] 40 0 0 10 20 30 Radius [Å] 40 –1 1 2 ε*SURF=εSURF/εAR 3 4 Figure 21.6. Dependence of contact angle on the integrated depth of surface potential. Inserted are two dimensional density distributions for droplets (a-c) and for bubbles (d.

63 30 0. SPC/E water molecules were interacting with the harmonic platinum surface represented by 3 layers of (111) surface through the water-platinum potential proposed by Zhu-Philpott.7.060 [Å-3] Height [Å] 20 10 0 0 0. A snapshot and 2-dimensional density profile of an equilibrium water droplet on a platinum surface at 350K. .000 [Å-3] 10 20 30 40 Radius [Å] 50 60 Figure 21.

64 (a) 500 ps (b) 1000 ps (c) 1500 ps Figure 21. Only clusters larger than 5 atoms are displayed for clarity.8. Nucleation of liquid droplet on a solid surface. .

.9.65 ∆G∞+∆GL' ∆GL 0 Free Energy ∆GL' ∆G∞+∆GL ∆G∞ 0 1 Bubble Radius r/r crit 2 Figure 21. Modification of free energy for a confined nano-bubble.

66 Figure 21. . Snapshots of Solid Phase Epitaxy (SPE) growth of silicon crystal in (111) at 2100K.10.

Carbon atoms are not shown for clarity. Gray circles represent nickel atoms. Snapshots of the catalytic CVD growth process of the cap structure of an single-walled carbon nanotube. . Gray arrows show a typical supply route of carbon atoms from expose metal surface.67 Figure 21.11.

. Geometrical structure of carbon nanotubes. (a) Single-walled carbon nanotube (SWNT) with (10. 10) chirality. (d) Peapod: nanotube filled with fullerenes.68 Figure 21. (c) Multi-walled carbon nanotube (MWNT).12. (b) A bundle of SWNTs.

. Physisorption of hydrogen in a bundle of single-walled carbon nanotubes at 77K. (a) A snapshot.69 (a) (b) (c) -3 -2 -1 0 1 2 x coordinate (nm) 3 0 –50 –100 Energy (meV) (a) (b) Figure 21. (b)(c) potential energy distribution.13.

20) nanotube with 2. (512 water molecules in (20. Structure of a water cluster in a carbon nanotube. .70 Figure 21.14.77 nm diameter).

Molecular dynamics simulation of heat conduction along a single-walled carbon nanotube.71 290K z 6.3 ∼ 404 nm 310K 0.7 nm 310 phantom (310K) Temperature (K) 300 q = 25.0 GW /m 64. 5) chirality and 202 nm in length. Temperature distribution along a nanotube with (5.5 K/µm phantom (290K) 290 –100 0 Position z (nm) 100 2 Figure 21.15. .

5) (8.10) λ ∝L 0. Dependence of thermal conductivity on length of nanotubes for 300K.15 λ ∝L λ ∝L 0.11 200 10 100 Length of nanotube L (nm) Figure 21. .72 700 Thermal conductivity λ (W/mK) 600 500 400 (5.8) (10.27 300 0.16.

The phonon density of states calculated as power spectra of vr. Phonon dispersion relation and photon density of states for 101 nm long (5. vz are shown as the right hand side of each dispersion relation. 5) SWNT. θ and z components of displacement are shown in (a).73 (a) r 1500 Energy ν/c (cm-1) (b) θ (c) z (d) Saito et al. Dispersion relations from r. (b) and (c) respectively. (d) The dispersion relations solved from the dynamical matrix using the force-constant tensor scaled from those for 2D graphite.17. vθ. . 1000 500 0 0 5 10 3 a Wave vector k (1/nm) π c−c 0 5 10 3 a Wave vector k (1/nm) π c−c 0 5 10 3πa Wave vector k (1/nm) c−c 0 5 10 π 3a c − c Wave vector k (1/nm) Figure 21.

83 nm 310 (6.6) d = 8.3 A Temperature (K) 300 (12.5 A 290 –200 0 Position (nm) 200 Figure 21. .18 Temperature jump and thermal boundary resistance at a junction of single-walled carbon nanotubes.0) d = 0.0) d = 9.74 (12.95 nm (6.6) d = 0.

(a) Time histories of the temperature of hot (central) SWNT and cold (surrounding) SWNTs. (b) Time history of the temperature difference. .19 Measurement of the thermal boundary resistance between single-walled carbon nanotubes in a bundle.75 400 Central SWNT 100 temperature[K] 300 temperature[K] Surrounding SWNTs 0 –200 0 time[ps] 200 –200 0 time[ps] 200 (a) (b) Figure 21.

Sign up to vote on this title

UsefulNot useful- Automated Force Field Parameterization for Nonpolarizable and Polarizable Atomic Models Based on Ab Initio Target Databy Rodrigo Kato
- Gemma C. Solomon et al- The Chameleonic Nature of Electron Transport Through Pi-Stacked Systemsby KomodoDS
- Metropolis Overviewby Elijah Norman
- R. González-Férez and P. Schmelcher- Electric-field-induced adiabaticity in the rovibrational motion of heteronuclear diatomic moleculesby Cvgdnnny

- thomas2016.pdfby Djanny Lorduy
- Evgeni Y. Poliakov, Vladimir Chernyak, Sergei Tretiak and Shaul Mukamel- Exciton-scaling and optical excitations of self-similar phenylacetylene dendrimersby Glade680
- What is Temperatureby Ciprian Alex Sky
- Yong Li, C. Bruder and C. P. Sun- Generalized Stern-Gerlach Effect for Chiral Moleculesby Cvgdnnny

- 12592765
- 3Dmoldyn07.pdf
- 33
- FK7012 - Home Exercises - Quantum Chemistry
- Automated Force Field Parameterization for Nonpolarizable and Polarizable Atomic Models Based on Ab Initio Target Data
- Gemma C. Solomon et al- The Chameleonic Nature of Electron Transport Through Pi-Stacked Systems
- Metropolis Overview
- R. González-Férez and P. Schmelcher- Electric-field-induced adiabaticity in the rovibrational motion of heteronuclear diatomic molecules
- thomas2016.pdf
- Evgeni Y. Poliakov, Vladimir Chernyak, Sergei Tretiak and Shaul Mukamel- Exciton-scaling and optical excitations of self-similar phenylacetylene dendrimers
- What is Temperature
- Yong Li, C. Bruder and C. P. Sun- Generalized Stern-Gerlach Effect for Chiral Molecules
- MOLDY USER MANUAL
- Theoretical Study of the Effect of Hydroxy Subgroup on the Electronic and Spectroscopic Properties of Azulene Molecule
- KTOM Notes
- Gemma C. Solomon et al- Understanding quantum interference in coherent molecular conduction
- Intro 2 MD Simulation
- HOMO-LUMO MIT Lecture Excellent)
- Spartan08Manual_New.pdf
- (CM2213&2223) Organic Chemistry Principles in Context - Green.pdf
- nrtl-sac paper
- EP- Molecular Spectra-End Sem -Student Handout-2014
- Gromacs Manual
- 0953-8984_22_13_133001
- Adhesion of Photon-driven Molecular Motors to Surfaces via 1,3-Dipolar Cycloaddition Reactions
- 3D QSAR
- Chemistry 104 Lecture 1 -- IMFs
- Demarco 2015
- Estrucctura Del Agua
- 1517.full
- NHHandbook

Are you sure?

This action might not be possible to undo. Are you sure you want to continue?

We've moved you to where you read on your other device.

Get the full title to continue

Get the full title to continue reading from where you left off, or restart the preview.

scribd