Chapter I Introduction
1.2 Motivation of this research
SWNTs have recently been considered to be a promising candidate material for use in field emitters, nano-electronic devices (such as single electron transistors) [Collins-2001-706] [Bachtold-2001-1317], and others. In such applications, one of the key issues is to effectively manipulate the nanostructures at lower deposition temperatures in order to be compatible with the IC processes. Recently, low temperature processes to synthesize the catalyst-assisted SWNTs by CVD methods have been widely studied [Qin-1998-3437] [Kurt-2001-1723] [Tsai-2001-NCTU].
However, most of these deposited SWNTs show morphologies in randomly entangled fashions [Delzeit-2001-368]. The methods to grow the well-aligned SWNTs on the desired locations are not quite successful so far.
The treatment of the catalyst to minimize the catalyst particle sizes before CNTs growth represents another technological challenge in SWNTs growth.
Physical vapor deposition (PVD) is the most popular approach for depositing catalytic materials, because it is highly compatible with the IC process. The catalyst films are typically treated with H-plasma to become well-distributed nano-particles, and CNT are subsequently grown from these pre-treated catalystic nano-particles. However, the agglomeration of nano-particles is unavoidable during the heating process, which does not particularly favor the fabrication of SWNTs. One way to minimize the catalyst particle size is to use the ultra-thin catalytic film, but the agglomeration effect during H-plasma pretreatment makes it difficultly to form SWNTs[Delzeit-2001-368]. Although Zhong et al. [Zhong-2005-1558] recently reported using Al2O3/Fe/Al2O3 with a sandwich-like structure to fabricate vertically aligned SWNTs, in which the Al2O3 buffer layer can inhibit the coarsening of Fe catalyst particles, the presence of the buffer layer on the top or bottom of catalyst layer raises other problems when SWNTs are used in electrical devices, such as adhesion, electrical conductivity and impurities. Accordingly, the purpose of this work was to develop a process to
fabricate the well-aligned SWNTs at low temperatures without buffer layer application. The idea is to use the catalyst precursor (CoCrPtOx) to promote the miniaturization of nanoparticles and to prevent the agglomeration effect during pretreatment and the initial stage of CNTs growth.
Chapter Ⅱ Literature review
2.1 Structures and properties of CNTs
Since discovery of CNTs in 1991 [Iijima-1991-56] , many superior properties of CNTs have attracted much attention of the scientists. The excellent properties of CNTs must be closely related to its unique structure. It was proposed that a graphene sheet of (0001) plane can be rolled to become various forms of CNTs structures: armchair, zigzag and chiral CNTs. The CNTs can be pictured as the fullerene-related structures with the end caps containing pentagon and hexagon rings. As shown in Fig.2.1 [Dresselhaus-1996-p756], if a C60 structure is bisected normal to a five-fold axis, the “armchair” tubule can be formed. ‘Zigzag’ tubule is formed in the way of bisecting C60 structure normal to a three-fold axis. The other ways of bisecting C60 structure can form the chiral CNTs. The caps of C60, C70 and C80 [Fig. 2.1(a)] are corresponding to the CNTs structures in armchair, zigzag and chiral, respectively [Figs. 2.1(c) and 2.1(d)]. In mathematics, scientists proposed a vector to define CNTs [Saito-92-2204]:
Ch = na1+ma2 ≣ (n,m) (2-1)
The Ch called chiral vector, and the angle between Ch and a1 is chiral angle θ,
while a1 and a2 denoted the unit vectors of graphene sheet. As shown in Fig. 2.2, the structures of CNTs in zigzag, armchair or chiral form are classified by the θangle or range, i.e. 0o, 30o or 0o<θ<30o, respectively. The chiral vector is expressed as a pair of integers (n, m) for mapping planar graphene sheet. The zigzag, armchair and chiral CNTs are corresponding to the chiral vectors of (n, 0), (n, n) and (n, m), respectively.
In 1992, Hamada and Saito [Hamada-1992-1579] [Saito-1992-2204] proposed theoretically that the CNTs can be a conductor or a semiconductor depending on its chirality. Their theories were proved experimentally by Wildoer and Odom
[Wildoer-1998-59] [Odom-1998-62] in 1998, using STM. They indicated that the two parameters, helicity and diameter, can be adopted to distinguish the metallic from semi-conducting properties of CNTs, i.e. the differences in the band gap and the Fermi energy shift. Among them, the armchair CNTs have two integers n and m equal to each other, thus have bands that cross the Fermi level and therefore are truly metallic. The chiral and zigzag nanotubes had two possibilities: (a) If n-m = 3k, where k is a integer except zero, then it was metallic with an energy gap of about 1.7 - 2.0 eV; (b) If n-m ≠ 3k, then it was semiconducting with an energy gap of about 0.5 - 0.6 eV [Fig. 2.3]. These show the nanotubes electrical properties are very sensitive to the wrapping angle and the tube diameter.Ballistic transport in the CNT channel was assumed.
Fig. 2.1CNT structures of armchair, chiral and zigzag tubules.[Dresselhaus-1996-p756]
Fig. 2.2 The construction of CNT from a single graphite sheet[Saito-1992-2204]
Fig. 2.3 The relation between properties and structure of CNTs[Saito-1992-2204]
In addition to the special features on electrical properties of CNTs, another key point of what scientists concerned is their field emission properties. The field emission is an electron emission phenomenon through tunneling effect when an electric field is applied on the surface of a material with negative potential. Due to suitable geometric contours, high thermal stability, good mechanical strength and high chemical stability, CNTs becomes a good field emission material. The field emission is much more fascinating than thermal emission since it just needs to apply a low electrical field (~V/μm) at room temperature. Usually, the field emission properties can be expressed by Fowler-Nordheim equation [Spindt-1976-5248]：
(2-2) voltage (V), a work function of material (eV), the effective emission area (m2), and the field enhancement factor, respectively. Base on Eq. 2-2, one can obtain：
1 curve is called F-N plot. A material with good field
emission properties often shows a negative slope on the F-N plot. Base on Eq.
(2-2), the ways to improve the field emission properties of a material can be achieved by increasing the effective emission area, α, and the field emission factor, β, which is related to the aspect ratio or geometric factor of emitter.
Furthermore, the too-close distance between emitters can deteriorate their emission properties, which is often called the screen-effect. [Gröning-2000-665] This signifies that the manipulation of the tube number density is also an important issue in field emission studies.
( )
In addition to special marvelous electrical properties, the mechanical properties, e.g. the reported Young’s modules may reach over 1 TPa [Eric-1997-1971]
which makes CNTs the stiffest material in the world. CNTs also behave elongation to failure of 20-30% [Pan-1999-3152] with a high tensile strength about 60 GPa[Yu-2000-637]. These experiment values enable the CNTs to become the highest strength/weight ratio material on earth, and to be used for potential applications in reinforcement of the composites.
Furthermore, thermal manage is now a very hot research in the world and the thermal conduction of CNTs are also excellent. Che et al. [Che-2000-65]
evaluated thermal conductivity of 10-nm-long CNTs are great than 2800 W/mK, almost equal to diamond. Berber et al. [Berber-2000-4613] predicted the thermal conductivity of (10,10) SWNT is about 6600 W/mK at room temperature. This is a very important support for their electronic and thermal applications.
2.2 Synthetic methods of SWNTs
There are many methods being developed to synthesize SWNTs, where arc-discharge, laser ablation and chemical vapor deposition (CVD) are the most popular methods [Lee-2001-245]. In those methods, carbon sources can be in gas or solid phases. The morphology and properties of SWNTs are often controlled by manipulating the following process parameters: substrate temperature, precursor
gases and gas ratio, catalyst, pretreatment conditions, applied bias, etc. However, the proposed methods still suffer the following problems: low yielding, low uniformities in structure and property, etc.
(a) Arc-discharge method[Saito-1995-33]
Arc-discharge method is believed to be the earliest way to synthesize CNTs.
When CNTs were first identified by Iijima[Iijima-1991-56], it was produced by this method. Figure 2.4 shows the schematic diagram of the arc-discharge system
[Saito-1995-33]. There are two graphitic rods as anode and cathode. Between these two electrodes, arcing occurs when DC voltage is applied. In the situation of anode containing small amount of catalyst such as Fe, Co, Ni and Y, the SWNTs can be synthesized[Bethune-1993-605], and MWNTs can be fabricated by using pure graphite as two electrodes. Generally, the discharge is operated at a voltage range between 20 and 40 V with current from 40 to 100 A under He or Ar atmosphere of 10-500 Torr. Carbon clusters collided out from the anodic graphite rod by electron bombardment are deposited on the cathode surface. The production on cathode may include amorphous carbon, fullerenes, carbon cluster, carbon nanotubes and varieties of other carbon structures. Therefore, purification of the nanostructures is generally an important issue for applications.
Another drawback of this process is its low yielding in producing CNTs.
Fig. 2.4 Schematic diagram of arc-discharge system [Saito-1995-33]
(b) Laser ablation[Guo-1995-49]
Laser ablation method was first reported by Guo’s group in 1995[Guo-1995-49], as shown schematically in Fig. 2.5. There is an incident laser beam for vaporizing graphite target under helium or argon gas atmosphere at pressure of 500 Torr. The productions are swept out by the flowing gas and to be deposited on the water cooled collector. Therefore, it is also called laser vaporization method. The graphite target containing Co, Ni, Fe, or Y is a more favor condition to form SWNTs.
Fig. 2.5 Schematic diagram of laser ablation system [Guo-1995-49]
(C) Chemical vapor deposition (CVD)
The CVD method is a mature technique in thin film processes. Most of films can be fabricated by CVD method, including metals, semiconductors and insulators. It is essential for CVD process to introduce some forms of the energy to decompose precursor gases and deposit the reaction product on the substrate surface. The introduced energy may include thermal, microwave, RF, or others.
Thus, it gives rise to different process names, depending on the source of the applied energy.
(1) Thermal chemical vapor deposition: [Lee-2001-245]
Figure 2.6 shows a schematic diagram of thermal CVD apparatus in the synthesis of carbon nanotubes. The method was used pyrolysis of hydrocarbon source to synthesize the SWNTs. This method is also catalyst assisted SWNTs growth method and the quality of SWNTs is sensitive to the pyrolysis temperature. The specimen is placed in a quartz boat with coated transition metals or theirs alloy on a substrate, and then the boat is positioned in a CVD reaction furnace, and nano-size fine catalytic metal particles are formed after an additional etching of the catalytic metal film with NH3 gas at a temperature in 750 to 1050°C. Reaction gas is supplied in one end of the apparatus, and gas outlet in the other. The merit of this method is easily to deposit large area, uniform and good quality of SWNTs. However the drawback is not compatible
with IC (integrated circuit) process due to working temperature over 600°C.
Fig. 2.6 Schematic drawings of thermal CVD system [Lee-2001-245]
(2) Plasma enhanced CVD (PECVD): [Qin-1998-3437]
PECVD system was employed to deposit the SWNTs with many merits comparing with other methods, e.g. compatible process with IC process, cheaper, less contamination, high yielding and controlled alignment of SWNTs, thus current attention has focused on developing new techniques for the preparation of vertically aligned SWNTs by using CVD methods.
According to the methods of plasma excitation can be classified the different plasma system such as microwave plasma CVD (MPCVD), RF or DC bias excited plasma CVD (PECVD), microwave plasma assisted hot filament CVD (MP-HFCVD) and electron cyclotron resonance CVD (ECR-CVD) and so
fourth. Generally, the power supplies for discharge of plasma CVD are DC bias;
radio frequency (RF) (13.56 MHz) and microwave (2.47 GHz) are typical of high frequency power supply. Using the plasma CVD process to produce SWNTs can be prepared by applying decomposition of hydrocarbon (such as CH4, C2H2, C2H4 and C6H6) or monoxide and even decomposed of metal complex on various substrates that coated transition-metal film.
The common used of microwave plasma CVD, such as MP-CVD, PE-HF-CVD, and ECR-CVD, to synthesize SWNTs can be ranked in terms of their working pressure, where MP-CVD or PE-HF-CVD and ECR-CVD were operated with the pressure range of P < 10-3 Torr and 10-1 < P < 100 Torr, respectively. The MP-CVD system [Qin-1998-3437] as shown in Fig. 2.7, with the high density of plasma ball permits a contamination-free and a modification of plasma shape through tuning of the cavity. The PE-HF-CVD system applied the current on the tungsten filament to efficiently increase temperature in the chamber [Kurt-2001-1723] as shown in Fig. 2.8 The ECR-CVD as shown in Fig. 2.9 is known for its own advantages of high dissociation percentage of the precursor gas, high uniformity of plasma energy distribution and large area of CNTs deposition [Tsai-2001-NCTU].
Fig. 2.7 Schematic drawing of MPCVD apparatus [Qin-1998-3437]
Fig. 2.8 Schematic drawing of PE-HF-CVD apparatus [Kurt-2001-1723]
Fig. 2.9 Schematic drawing of MPCVD apparatus [Tsai-2001-NCTU]
(d) Buffer layer-assisted growth of SWNTs
Buffer materials are most effectively to promote SWNTs formation. Wang et al. [Wang-2006-in press] demonstrate that AlON buffer layer is the best material for SWNTs fabrication. Moreover, the results also indicate that the favorable conditions for synthesizing SWNTs networks are a high substrate temperature, low CH4/H2 ratio and thin catalyst thickness with AlON as the buffer layer material. The roughness of AlON film is presenting a rough (~1 nm) surface [Fig. 2.10], whereas Si wafer surface is quite neat. These results suggest the protrusion on rough buffer layer (e.g., AlON) can cause instabilities on Co
Plasma
catalyst surface and form nuclei sites for tubes growth afterwards. They also have measured the roughness of other buffer materials, such as AlN, TiN, and TiO2. The results show their surface are quite smooth and roughness (Rrms) are on the order of angstroms at most, show hardly effect to enhance SWNTs.
Figure 2.11 shows the growth processes of buffer layer assisted growth of network SWNTs.
Fig. 2.10 AFM image of AlON buffer layer of 10 nm where shows rough surface. The rms value of surface roughness is about 1.0 nm. [Wang-2006-in press]
Fig. 2.11 Schematic diagram of the growth processes of SWNTs networks assisted by buffer layer. [Wang-2006-in press]
(e) Alloy catalyst-assisted growth of SWNTs
Alloy catalysts were recently employed to growth CNTs like CoMo, FeNi, CoPt, FePt [Kuo-2003-799] and so on. But no extremely dense vertically aligned SWNTs were grown by alloy catalysts in the past. PtOx has been recentlyused in optical storage media and nonvolatile memory technology because of its explosive effect of the decomposition of PtOx at an appropriate process temperature. Kim et al. [Kim-2003-1701] proposed that PtOx film will decompose to Pt nano-particles and O2 by laser thermal pretreatment. The decomposition temperature is approximately 500°C (1 atmosphere in air). Figures 2-12(a) and 2-12(b) show that PtOx film was self-assembled to nano-particles by laser pretreatment and then transmittance is decreased (reflectance is increased) by
the decomposition of PtOx film to Pt nano-particles and O2 [Fig. 2.13]. Thus, it is so interesting to adopt PtOx with other metals, like Co and Cr, to be alloy-catalyst for synthesis of SWNTs.
Fig. 2.12 TEM images of (a) PtOx thin film and (b) self-assembled Pt nanoparticles after laser pretreatment. [Kim-2003-1701]
Fig. 2.13 Thermo-optical properties of PtOx single layer. [Kim-2003-1701]
(n: refractive index, k: extinction coefficient).
2.3 Growth mechanisms of SWNTs
Regarding the use of CVD methods for SWNTs growth, many growth mechanisms have been proposed[]. However, most of the mechanisms are based on the original model of carbon nanofibres proposed in 1970s by Backer.
[Backer-1978-14] It is believed that nanotubes grow as carbon precipitates from a supersaturated metal catalyst that resides at either the base or the tip of a growing nanotube. Catalyst/substrate interactions and temperature gradients across the catalyst particle are considered to be important factors that determine the growth mechanism. Catalyst size was believed to determine the CNT diameter, and SWNT was synthesized when catalyst size is about 1~2 nm.
However, most of these models were proposed without sufficient and systematic supporting experimental evidence, and they often lacked details about the physical mechanisms and the effects of various process parameters. Thus, the kinetics of nanotube nucleation and growth are not well known yet. Other growth mechanisms are described as follows:
(a) Ball-and-stick catalyst scooting model [kuo-2004-p.9-10]
Birkett et al. [Birkett-1997-111] proposed that transition metals show a high propensity for decoration fullerene surfaces. A carbon fragments bind to the metal clad fullerene and they may self-assemble as a surrounding circular hexagonal chicken-wire-like fence. Once formed as a belt, the network could
propagate as a cylinder, so called open edge growth. This model predicts the SWNT diameter will be d(C60) + 2*d(Interplanar distance), i.e. 0.7 nm + 2(0.34) nm = 1.38 nm ,which is in excellent agreement with observation. Another possible model was scooter mechanism[Thess-1996-483] which considered that a few metal atoms chemisorbed and scooted around the open edge of the sheet and kept tube open and grow. When metal atoms aggregated and lost its kinetic energy for scooting, SWNT growth will stop. Figure 2.14 shows the schematic diagram of ball-and-stick scooting model.
Fig. 2.14 Schematic diagram of ball-and-stick scooting model [Birkett-1997-111]
(b) Root growth mechanism
When numerous SWNTs grow by a single catalyst particle and their
diameter are much smaller than catalyst, this is called root growth mechanism.
Saito et al.[saito-1994-L526] proposed that when catalyst metal is evaporated together with carbon by arc discharge, carbon-metal alloy particles are formed on the cathode surface. Since the carbon-metal compound soot was produced in a carbon-rich at atmosphere, the initial alloy particles in a liquid phase contained more carbon than solubility limit in a solid state. Therefore, with the decrease of temperature of the cathode, the liquid alloy particles begin to segregate excess carbon on their surfaces. When the cooling of particles proceeded at a moderate rate and the supersaturation of carbon in metal particles was not so high, the carbon gradually segregated on the surface and formed graphitic layers one by one. When cooling was rapid and the initial content of carbon in a particle was high compared with that in the process mentioned above, the supersaturation of carbon became rather high. According to the classical nucleation theory, high supersaturation would bring about nucleation of graphite at numerous sites on the surface of a particle. Since the formation of flakes of graphitic occurs suddenly, a large number of tiny graphitic flakes are formed. These flakes curl and then close their open ends in order to saturate dangling bonds at their periphery. Among this random assembly of graphitic flakes, seeds of SWNTs may be formed [Figs. 2.15 and 2.16].
Fig. 2.15 TEM image of radiate sea-urchin-like SWNTs [saito-1994-L526]
Fig. 2.16 Schematic diagram of root growth mechanism -1 [saito-1994-L526]
Zhou et al.[zhou-1994-1593] proposed another model that SWNTs was separated from catalyst surface by graphite layers, as shown in Fig. 2.17. Furthermore,
Gavillet et al[Gavillet-2001-275504] suggest a common growth mechanism based on a vapor-liquid-solid model. The first step of the process is the formation of a liquid nanoparticle of metal supersaturated with carbon [Fig. 2.18(a)]. And then there is a competition between the formation of a graphitic sheet [Fig. 2.18(b)]
and the nucleation of single-wall nanotubes [Fig. 2.18(c)]. In order to obtain long nanotubes [Fig. 2.18(d)], the root-growth process should continue for a sufficiently long time, until local temperatures are too low, leading to the solidification of the nanoparticles. Figures 2.18(e) and 2.18(f) show that nucleation did occur but growth did not take place so that carbon has partly condensed into amorphous carbon flakes or into a few graphitic layers.
Fig. 2.17 TEM image of the interface between the multilayered cage and the single-walled tubes [zhou-1994-1593]
Fig. 2.18 Schematic diagram of root growth mechanism -2 [Gavillet-2001-275504]
(c) Yarmulke mechanism
This model was proposed by Dai et al.[Dai-1996-471] that SWNTs are formed on molybdenum nanoparticles in the size range of 1~4 nm by the disproportionation of CO at 1200°C, indicates that there can be a slight variation
This model was proposed by Dai et al.[Dai-1996-471] that SWNTs are formed on molybdenum nanoparticles in the size range of 1~4 nm by the disproportionation of CO at 1200°C, indicates that there can be a slight variation