† Corresponding author. E-mail:
With the size reduction of nanoscale electronic devices, the heat generated by the unit area in integrated circuits will be increasing exponentially, and consequently the thermal management in these devices is a very important issue. In addition, the heat generated by the electronic devices mostly diffuses to the air in the form of waste heat, which makes the thermoelectric energy conversion also an important issue for nowadays. In recent years, the thermal transport properties in nanoscale systems have attracted increasing attention in both experiments and theoretical calculations. In this review, we will discuss various theoretical simulation methods for investigating thermal transport properties and take a glance at several interesting thermal transport phenomena in nanoscale systems. Our emphasizes will lie on the advantage and limitation of calculational method, and the application of nanoscale thermal transport and thermoelectric property.
The size of electronic devices is being reduced to as small as sub-ten nanometer magnitude, and is approaching the feature wavelength of many particle or quasi-particle, such as electron, phonon, and so on. Such small nano-structures exhibit significantly different electronic and thermal properties compared to that of macro ones, and bring us numerous opportunities and challenges. For example, the decreasing device size enables the significant increase of integration degree in integrated circuits, but on the other hand, the enhancement of integration requests better thermal management, including both heat dissipation, heat energy conversion, and so on. These facts make the thermal management an important issue.
Recently, breakthroughs in thermal transport field, such as high-performance thermoelectric materials,[1,2] phonon diode,[3] phonon triode,[4] topological effect of phonon,[5–9] and so on, are emerging endlessly. More excitingly, the rapid development of the synthesis and processing, such as molecular beam epitaxy (MBE)[10] and chemical-vapor deposition (CVD),[11] are enabling the production of devices with well-defined structure and smaller size, and are facilitating the commercial application of these phonon nano-devices.
However, there are still many problems unsolved in this field. For instance, the further application of large-scale power generating by thermoelectric materials is in a dilemma.[12] It is known that most of the energy is dissipated in the form of waste heat. However, the figure of merit (ZT), which describes the thermoelectric energy conversion efficiency, is difficult to improve. Over the past decade, the maximum ZT observed in experiment is still no more than 3,[13] which is far away from mass business application for power generating. On the other hand, the thermal management[14] and near-field heat transport problem[15] have always been plaguing us. The issues of controlling of heat flow, heat dissipation, and heat energy conversion in nanoscale systems have become a specific science field nowadays.[16]
Compared with experiments, the computer simulation has the inherent advantages on researching small size structures. With decreasing further size of devices and emergence of low-dimensional materials, such as graphene,[17] graphene-like structure (MXene,[18] phosphorene,[19] and so on), single molecule junction,[20] and so on, the thermal transport mechanism is different from the macroscopic ones. Fourierʼs law was proved to be no longer applicable to the anomalous heat transport[21,22] in some nanoscale structures. Fortunately, many new theory models are set up as the powerful tools to study the underlying physics mechanism of nanoscale thermal transport. It is expected to provide a guideline for us to design excellent performance nanoscale devices. The comparison and application range of different theoretical methods is discussed in Section
Describing the thermal conductivity accurately is an important issue for the design of nano devices. However, the experiment measurement cannot completely eliminate the impact of the external environment, and the temperature is so difficult to define clearly under the sub-nano size. In recent years, a serial of theoretical methods were proposed, such as the dielectric continuum model,[23] scattering matrix method,[24] molecular dynamics simulation,[25] non-equilibrium Green function method,[26] and Boltzmann transport equation,[27] to predict the thermal conductivity in nanoscale or larger scale. In this section, we mainly introduce different theoretical methods for thermal transport property researches.
Experimental research has been confirmed that the thermal transport at low temperature and small size is mainly determined by the ballistic process.[28,29] The dielectric continuum model and scattering matrix method were widely used in the study of ballistic thermal transport in low temperature region. Recently, the thermal conductivity of various low-dimensional systems, such as abrupt structures,[30,31] rough surfaces,[32] defects,[33] and stub structure,[34,35] have been widely studied by these methods.
In long wavelength limit, the dielectric continuum model is a reasonable approximation to study the thermal transport property on the condition that the structure is continuous and compact, and if the wavelength is larger than the size of materials, the structure of inner system can be ignored. In this model, the stress, strain, and displacement are functions of continuous coordinate so that the conception of continuous and limitation applied in mathematic deduction is so convenient. In past few years, the dielectric continuum model was widely used to study the thermal transport property in nanostructure at low temperature, such as supperlattice[36,37] and wave-guide structure.[38] At low temperature, only several acoustic phonons were excited in system so that the dielectric continuum model provides sufficient accuracy for the research of thermal transport. In this model, the wave equation of the entire system can be defined as
The scattering matrix is a processing method on the basis of the dielectric continuum model. In general, the transfer matrix and scattering matrix method are both in common use to solve the transport problem of periodic or quasi-periodic quantum structures. The probability distribution of particle in any position can be easily obtained by transfer matrix. However, because the retard mode inevitably increases with the exponent, and result in singularity of transfer matrix, so that the transfer matrix method cannot describe the complex shaped and larger size difference structure. Compared with the transfer matrix method, the scattering matrix method not only have the benefit of the transfer matrix method, but also overcome the disadvantage of transfer matrix method. Therefore, it is more suitable to deal with the particle transport problem in complex structure.
An example was given here, the structure scheme is show in Fig.
The non-equilibrium Green function (NEGF) method originates mainly from the quantum field theory. In general, this method were used to study the ballistic transport of electron and phonon in nanoscale, also the phonon–phonon interaction[39,40] and electron–phonon coupling[41] can be considered. However, the amount of computing is so huge that it cannot be done well with the current computing power. Therefore, we only discuss the linear part of Greenʼs function method in this review. Considering a two-probe model that is compose by three parts, i.e., the left semi-infinite region (L), the central scattering region (C), and the right semi-infinite region (R), the Hamiltonians of entire system can be written as
In small size and low temperature limitation, only the acoustic modes are excited in system so that the dielectric continuum model is an efficient method to calculate thermal conductance. However, there are some non-physics caused by dielectric continuum approximation, such as infinite mean-free-path of phonon, different phonon thermal conductivity, zero temperature gradient, etc. In fact, if the size of materials close to the mean free path of phonon, the heat transport process is not only ballistic but also diffuse, it is the combination of both ballistic and diffuse,[48] and the phonon–phonon interaction is objective existed in a real system so that the expansion of potential energy must be considered to higher order terms in thermal transport research.
The molecular dynamics simulation, which can take the anharmonic effect into account, has been risen up as a powerful tool in recent years for the researching of thermal conductivity and frequency dependent thermal property. Compared with the previous method as is discussed in above, MD method based on force field (FF) which can get more information without any initial physical environment settings. Because of that, the molecular dynamics simulation is an ideal method for investigating the fundamental thermal property for complex structure[49–53] and various structure dependent property.[54–59] In addition, the MD simulation can also be used in liquids,[60] surfaces,[61,62] clusters,[63–65] topological insulators,[66] and so on.
There are two different methods to research the heat transport properties in molecular dynamics simulation, i.e., the non-equilibrium (NEMD) and the equilibrium molecular dynamics (EMD), respectively. The NEMD method is also known as the direct method, which is based on the Fourierʼs law to calculate thermal conductivity. For the non-equilibrium method, by introduced a temperature gradient into the both ends of the entire system artificially, the heat flux will be constructed when the entire system tends to the statistic non-equilibrium steady state. The temperature of relative columns atom along the longitudinal of materials is defined by energy equipartition theorem, and the thermal conductivity can be simply obtained by Fourierʼs laws
To calculate intrinsic thermal conductivity of materials, the EMD method will be more suitable. Because there is not any artificial heat bath introduced into the entire simulation system. In short, the EMD method is based on Green–Kubo formula[70,71] originating from fluctuation-dissipation theory. It can be defined as follows:
For the larger size materials, according to dynamics theory, the lattice vibration can be approximate to phonon gas, which satisfies with the Bose–Einstein distribution. The advantage of this method is the various nonlinear transport processes, such as the Umklapp process, electron–phonon coupling, can be considered in calculation. If the phonon distribution function in non-equilibrium state is little difference with the the equilibrium state so that we can simply regard the phonon scattering process as the phonon relaxation time. In general, we can use Fermiʼs golden rule to obtain all phonon scattering rate.[75,76] However, the ultra-high computing cost makes the scattering rate of different phonon modes so difficult to obtain. Hence, we can simplify the solution by relaxation time approximation. Usually, there are four different kinds of phonon scattering processes:[77–80] three-phonon scattering, impurity scattering, boundary scattering, and the electron–phonon scattering. Based on Matthiessenʼs rule and without considering the coupling relations between them, the total relaxation time can be written as
With the increasing ICs chips integration level and the device size shrinks, the thermal management has become an extremely urgent issue. It is important to find the ultra-high thermal conductivity materials to serve as the heat dissipation or conduct device. But on the other hand, the ultra-low thermal conductivity materials, which can be used in thermoelectric cooling or energy conversion, are also a ways to solve these problems. All these problem requires an efficient way to control heat flow in nanoscale. Because of that, the in-depth understanding of thermal property of various materials is directly relates to the lifetime of electronic device and energy conversion efficiency. In this section, we would like to give a brief review on the thermal property of various nanostructure and novel physical mechanisms from the theoretical simulation standpoint.
Unlike the bulk materials, the low dimensional structure, such as graphene, will be hugely influenced by quantum confinement effect. It has drawn great attention in recent years due to the unique physical properties, such as the robust negative differential resistance phenomena,[81] excellent mechanical property,[82] ultra-high thermal conductivity,[83–85] and quantum anomalous Hall effect,[86] and so on. Moreover, other graphene-like materials, such as MoS2,[87] graphyne,[88] borophene,[89] silicene,[90] etc., were emerging endlessly. These discoveries make possible to understand the thermal transport mechanism in nanoscale.
For 2D materials, there are various factors can influencing the thermal conductivity significantly,[91] such as size,[92] defect and doping,[93,94] edge,[95] substrate,[96] and so on. In general, the high-frequency phonon will be greatly influenced by defects and impurities,[97,98] while the low-frequency phonons are insensitive to these kinds of effect.[99] However, the previous works does not take the different phonon modes into account. Utilizing the NEGF method, Peng et al.[100] described the influence of different types defect for thermal conductivity of graphene nanoribbon. It is found that different vibration modes will show different behaviors on different defect sizes and types, as shown in Fig.
Interestingly, the two kinds of phonon modes show different characteristics in cut-off frequency after we separate the phonon transmission spectrum into out-of-plane and in-plane vibration. The further result shows that the flexural phonon modes (FPMs), especially the high-frequency region, is more influenced by cavity defect, which agrees well with previous works. More interestingly, for the IPMs (in-plane phonon modes) the low-frequency phonon is no longer sensitive to the cavity defect, and even the high-frequency phonon has also almost not affected by this defect type. From the perspective of the phonon density of state, the difference of FPMs and IPMs will be more obvious. The FPMs is totally localized at the frequency 25 cm−1. This indicates that the low-frequency phonon influenced by vacancy defect on graphene and other 2D materials may be also strongly associated with the shape of the defect rather than only the high frequency phonon will be influenced by vacancy defect.
With the dimension further decreasing, the quantum effect will affect the physical property of materials more strongly. Before the discovering of graphene, the one/zero dimensional nano structure have been observed by many experimental researches. The thermal property of nanowire,[101,102] nanotube,[103–106] quantum dot[107] and another shaped quantum structure[108] has been attracted increasing attention in recent years. Many interesting physical phenomena associate with these structures were observed by theoretical simulation and experiment, such as the anomalous heat conduction,[109] phonon-assisted heat generation,[110] splitting behavior of acoustic phonon,[111] reversal of thermal rectification,[112] and so on. For these structures, due to the size of device is much smaller than mean free path of phonon so that the scattering matrix method and nonequilibrium Greenʼs function method can be more convenient to describe the transmission coefficient of different phonon modes, which consist well with the experimental result in low temperature. Like the spectroscope, the theoretical simulation can easily separate the different phonon modes contribution of thermal conductance so that to study the underlying physical mechanism of thermal transport, and this is usually unattainable in experiment measurements.
Figure
From the prospective of informatics, the heat is destructive for information memory and transport. However, the emergence of phononics make the people understand that the heat can be regarded as the information or signal. Since then, there are many novel mechanisms of thermal transport were observed in experiment and theoretical calculation, such as the thermal rectification, phonon resonance, negative thermal differential resistance, thermoelectric, and so on. The analysis of these different mechanism will be conducive to further improve the performance of nanoscale function devices.
The electron diode and spin filter device has been widely researched by theory and experiment,[115–120] as same as the electron diode, by coupling two different materials we can construct a heterojunction to form the asymmetry of heat current when the temperature difference is inverted. The first model to open the possibility of building the thermal rectifier[121] is proposed by Terraneo in 2002. Since then, the application of thermal rectification effects, such as thermal logic gates,[4] thermal transistor,[122] and thermal memory,[123] have been widely studied in theory. The first solid-state thermal rectifier experimentally also realized by using the asymmetrically mass-loaded carbon and boron nitride nanotubes.[124] In short, finding the way to control the heat transport and thermal rectification has been become an important issue for the designing of thermal devices. However, the origin of thermal rectification effect is not one-fold. Such as the interface,[125] grain boundary,[126] defect and doping,[127] mass and structure graded,[128–132] each of these can influence the thermal rectification ratio.
Recently, an important mechanism to better control the heat flux in nanoscale structure was proposed by Liu et al.[133] They demonstrated that the core-shell nanowires have great potential to serve as a heat cable to control the channel of thermal transport. As shown in Fig.
Accordingly, a more efficient mechanism to enhance thermal rectification ratio[135] was further examined by MD simulations. As shown in Fig.
The success of thermal rectifier and heat cable suggests that the heat flux can be manipulated as the signals. Most applications of thermal devices are relevant to heat conduction in the nonlinear regime. Controlling the heat flux plays a vital role in the operation of these devices. As shown in Fig.
Besides the management and manipulation of thermal energy, with the growing energy shortages, looking for a new energy of environment protection and pollution-free has been received considerable attention in recent years. Thermoelectric materials can generate electricity directly from waste heat, and thus provide a solution to these problems. It plays an important role in new energy resources field. The energy conversion efficiency of thermoelectric materials is determined by the thermoelectric figure of merit (ZT). It is defined as follows:
On the one hand, there are various ways to enhance thermoelectric energy conversion efficient of inorganic crystals, such as defect[143–145] and doping,[146–148] core-shell structure,[149–151] superlattice,[152,153] branch structure,[154,155] strain[156,157] and electric field engineering,[158] phonon-drag effect,[159] and so on, which are widely studied in experiment and theoretical simulation.
For the thermoelectric performance regulation, the defect and doping plays vital roles in reduction of phonon thermal conductivity, which can be obviously suppress the phonon vibration density of states in the entire frequency range, As is shown in Fig.
Another efficient way to obtain high ZT is superlattice structure, as shown in Fig.
While as for the strain engineering, we can see that, the phonon frequency will redshift with the strain loaded, as show in Fig.
Besides the bulk materials, with the progressing of molecular synthetic technique, the organic molecular system[161–168] has also been paid more attention in recent years. The molecular junction[169–173] and molecular crystals,[174] and so on were both expected to serve as the powerful thermoelectric materials. The electrode and molecule are coupling with each other,[175] which is dynamics coupling between transmitted electrons and molecular vibration that can strongly affect the transmission process of electron so that we have to consider the electron–phonon coupling effect in molecule devices. Without considering electron–phonon coupling effect, based on nonequilibrium Greenʼs function method, Cao et al.[176] confirmed that compared with the perfect graphene nanoribbon, the single molecular will dramatically enhance the scattering rate of phonon so as to greatly reduces the thermal conductivity of both acoustic and optical phonon, as shown in Figs.
Furthermore, an interesting phenomenon will be founded when we take the electron–phonon interaction into account in molecule system, that is, the electric current will obviously enhanced by electron–phonon coupling,[177–179] as shown in Figs.
On the other hand, the parameters coupling relationship of ZT has always plagued us so that the energy conversion efficiency is difficult to improve. Except the traditional thermoelectric effect, the piezoelectric effect and triboelectric effect also have great potential to convert disorder energy into usable energy. Since the realization of piezoelectric in Zinc Oxide nanowires[186] in 2006, piezoelectricity have been a real hotspot in recent years. Various nanogenertors based on different materials have been developed,[187–189] and the application of these nanogenertors has covered electronics powering, signal sensing and so on. However, the energy source of piezoelectricity is usually irregular ambient natural energy like wind or manually provided mechanical energies, and consequently, either the output electric potential is aperiodic or the energy source is not the basic energy like heat in our nature. The advantage of thermoelectric can directly convert heat energy into electricity, but the output is limited to direct current and the difficulty in increasing the efficiency is widely known.
Encouragingly, utilizing the molecular dynamics simulation, Liu et al.[190] predicted a novel mechanism for thermoelectric energy conversion based on the combination of lattice dynamic and piezoelectric effect. As is known, the lattice wave can be approximately regarded as the harmonic vibration so that the amplitude of the vibration can be easily estimated by the following classical equation:
We present a brief overview of the different theoretical model in nanoscale thermal transport field, and a serial of novelty phenomena were predicted by these model. At first, we introduced several kinds of theoretical method and the application range of different theories from the ballistic to diffusion transport. At present, the theoretical research of heat transport still lacking a standard model for different systems so that the choice of method is so important, which has a direct impact on the correctness of the results. The introduction and comparison of different theoretical methods aim to provide better choice and guidance to study the heat transport properties in different system.
Based on these theoretical models, we present several efficient ways to manipulate the heat flux. And the various novel thermal transport mechanism in nanoscale structure were predict by theoretical simulation, such as the various defect effect, standing waves, negative differential thermal resistance, the destructive of spin filter by phonon. These studies are mainly focus on the regulation of electron and phonon properties. Moreover, we designed a series of approaches to remarkable enhance the thermoelectric performance in nanoscale structure. For the follow-up researches, these discovers could be provide a guideline for design and theory research on microelectronic device so that to further promote the application of thermal and thermoelectric energy conversion devices.
[1] | |
[2] | |
[3] | |
[4] | |
[5] | |
[6] | |
[7] | |
[8] | |
[9] | |
[10] | |
[11] | |
[12] | |
[13] | |
[14] | |
[15] | |
[16] | |
[17] | |
[18] | |
[19] | |
[20] | |
[21] | |
[22] | |
[23] | |
[24] | |
[25] | |
[26] | |
[27] | |
[28] | |
[29] | |
[30] | |
[31] | |
[32] | |
[33] | |
[34] | |
[35] | |
[36] | |
[37] | |
[38] | |
[39] | |
[40] | |
[41] | |
[42] | |
[43] | |
[44] | |
[45] | |
[46] | |
[47] | |
[48] | |
[49] | |
[50] | |
[51] | |
[52] | |
[53] | |
[54] | |
[55] | |
[56] | |
[57] | |
[58] | |
[59] | |
[60] | |
[61] | |
[62] | |
[63] | |
[64] | |
[65] | |
[66] | |
[67] | |
[68] | |
[69] | |
[70] | |
[71] | |
[72] | |
[73] | |
[74] | |
[75] | |
[76] | |
[77] | |
[78] | |
[79] | |
[80] | |
[81] | |
[82] | |
[83] | |
[84] | |
[85] | |
[86] | |
[87] | |
[88] | |
[89] | |
[90] | |
[91] | |
[92] | |
[93] | |
[94] | |
[95] | |
[96] | |
[97] | |
[98] | |
[99] | |
[100] | |
[101] | |
[102] | |
[103] | |
[104] | |
[105] | |
[106] | |
[107] | |
[108] | |
[109] | |
[110] | |
[111] | |
[112] | |
[113] | |
[114] | |
[115] | |
[116] | |
[117] | |
[118] | |
[119] | |
[120] | |
[121] | |
[122] | |
[123] | |
[124] | |
[125] | |
[126] | |
[127] | |
[128] | |
[129] | |
[130] | |
[131] | |
[132] | |
[133] | |
[134] | |
[135] | |
[136] | |
[137] | |
[138] | |
[139] | |
[140] | |
[141] | |
[142] | |
[143] | |
[144] | |
[145] | |
[146] | |
[147] | |
[148] | |
[149] | |
[150] | |
[151] | |
[152] | |
[153] | |
[154] | |
[155] | |
[156] | |
[157] | |
[158] | |
[159] | |
[160] | |
[161] | |
[162] | |
[163] | |
[164] | |
[165] | |
[166] | |
[167] | |
[168] | |
[169] | |
[170] | |
[171] | |
[172] | |
[173] | |
[174] | |
[175] | |
[176] | |
[177] | |
[178] | |
[179] | |
[180] | |
[181] | |
[182] | |
[183] | |
[184] | |
[185] | |
[186] | |
[187] | |
[188] | |
[189] | |
[190] |