Carbon Nanotubes for Tomorrow’s Electronics: Review and Analysis of Vapor Grown Synthesis and Characterization of Their Electrical Impedance
Submitted on December 6th, 2001
CNT background | CNT Synthesis by Vapor Deposition
Growth Mechanisms in Vapor Grown Carbon Nanotubes
Electron Transport in CNT | Nanowiring of CNT
Integrating SWNT into Electronics | Directed assembly of CNT Circuits
Influence of CVD Synthesis Variations on Electrical Impedance | References
Technology is defined as the application of science, especially to industrial or commercial objectives. Microelectronics (especially applied to IT) has proven to be possibly the dominating technology over the past twenty years. Technology is constantly pushing the limits of microelectronics, allowing circuits to become twice as small and twice as fast approximately every 18 months. But, as current fabrication methods and materials become pushed to their limits, the size of microelectronics will reach some theoretical minimum. Carbon nanotubes can provide alternatives to current materials used in electronics. But, they require new fabrication techniques, and they have different properties than conventional materials. This paper reviews some synthesis techniques of vapor grown carbon nanotubes (specifically chemical vapor deposition) that will be used in tomorrow’s electronic circuits. Some of their electrical properties will be studies, especially impedance characteristics (vector sum of resistance, capacitance, and inductance). But first, some background will be given on vapor grown nanotubes: synthesis and growth mechanisms.
Since Iijima’s discovery (Iijima 1991) of carbon nanotubes (CNT) in 1991, research on the growth characterization and application development has received significant attention. A CNT is a cylindrical form of carbon, configurationally equivalent to a two-dimensional graphene sheet rolled into a seamless cylinder. The nanotubes can be single walled (SWNT) or multi-walled (MWNT) with a concentric cylindrical arrangement. The separation between walls in multiwalled tubes is a constant 3.4 Å. Several individual SWNTs can pack together to form nanotube ropes. The nanotubes are characterized by a chiral vector c = na + mb where a and b are vectors defining a unit cell in the planar graphene sheet and n and m are integers. Depending on chirality (i.e., the values of n and m), CNT can be either metallic or semiconducting. If (n-m)/3 is an integer, the nanotubes is metallic; otherwise it is a semiconductor (Dresselhaus, 1996). In addition to such unique electrical properties, CNT exhibits extraordinary mechanical properties. The Young’s modulus of SWNT has been estimated to be over 1 Tera Pascal and the tensile strength is about 200 GPa. The thermal conductivity of CNT has been estimated to be 3000 W/mK which is second only to epitaxial diamond. Though CNTs are quite unreactive, they can be functionalized with functional groups either at the tip or on the sidewall of the nanotubes. Given the combination of electronic and mechanical properties and possibilities for functionalization, CNTs offer significant promise for applications in nanodevices, composites, sensors and several other fields.
Since Iijima’s discovery (Iijima 1991), the field of carbon nanotubes has become a separate subject in material science. These materials attracted more and more attention from physicists and chemists. Their surprisingly new and unique physical and chemical properties generated tremendous interest. At the beginning, the quantity of carbon nanotubes available for experiments was very low, since the early arc-discharge method produced only tiny amounts. The increasingly demand on having more and more nanotubes brought the fast development in the ways of production, including not only improvements of the arc-discharge technique but also new methods like the laser evaporation and the catalytic vapor deposition of gaseous hydrocarbons.
The first successful experiments on the preparation of MWNTs by Yacaman (Yacaman, 1993) and Ivanov (Ivanov, 1994) were followed by hundreds of modified or new procedures. On the basis of these experiments, several fundamental principles were concluded. In most cases the syntheses are carried out at high temperature applying a stream of carbon source diluted with inert compounds such as nitrogen, argon, etc. The experimental set-up consists of a high temperature oven in which the catalysts are placed onto a highly resistant ceramic or metal plate. The nature and yield of the deposit obtained in the reaction are controlled by varying different parameters such as the nature of the metals and the supports, the hydrocarbon sources, the gas flows, the reaction temperature, the reaction time, etc. By selecting the proper conditions, both the physical (e.g. length, shape, diameter) and chemical properties (e.g. number of defects, graphitization) of MWNTs can be designed in advance. For the catalyst preparation, various methods are applied, such as impregnation, ion-exchange and mechanical grinding of the support and the metal components. Activity of the catalyst samples prepared in different ways shows dependence on the nature of the catalyst support (silica or different pore diameter, zeolites of various structure) and on the pH of the solution during the synthesis of catalyst samples. The supported Co (Ivanov, 1994), Ni (Yudasaka, 1995), and Fe (Hernadi, 1996) catalysts were found to be the most active in the CVD. When other elements (e.g. Cu, Cr, Mn) are used, only a negligible amount of carbon nanotubes is formed. As far as the catalyst support is concerned, its quality is of considerable importance. While Si-, Mg-, and Al-containing materials (silica, alumina, zeolites, magnesia) prove to be applicable as supports, the different forms of carbon are not suitable in NT production. These results suggest that, contrary to received wisdom, the support has a peculiar role in the reaction. The hydrocarbons most frequently used are acetylene, ethylene and benzene because of their high carbon content. It was also shown that generally, increasing reaction time results in the formation of larger amounts of amorphous carbon due to the deactivation of the catalytic sites. Determining the optimal reaction temperature is a very complicated task since several requirements come into conflict: while a high temperature favors proper graphitization, the homogeneous pyrolysis of the hydrocarbons becomes excessive above a certain temperature depending on the carbon source used. Moreover, the metals and supports demand different temperature ranges.
In 1996, Dai et al (Dai, 1996) and showed that although small quantities were obtained, the modified CVD was a practicable plan for this purpose. In these studies Fe, Mo, Co, Ni were the catalytically active particles particularly on Si- and Al-based supports in the 800-1200 oC temperature range. Dai’s work suggested that the small size of the catalytic metal particles is crucial for the growth of SWNTs. In a later work Dai et al showed (Kong, 1998) that high quality SWNTs could be produced by chemical vapor decomposition of methane on supported transition metal oxide catalysts. The experimental set-up was similar to apparatus generally producing multiwalled nanotubes, however some factors including the catalyst composition, the support, and the hydrocarbon were different. Methane was used instead of the generally applied acetylene or ethylene because of its kinetic stability at high temperatures. Since there is no pyrolytic decomposition, the carbon atoms needed for nanotubes growth are produced by catalytic reaction from methane on the metal surfaces. Moreover, decreasing the reaction time to 10 minutes from the usually applied hour(s) prevents the outer surface of nanotubes from being coated with amorphous carbon. Finally, Fe2O3 was found to be significantly more efficient in the SWNTs production that CoO or NiO.
Colomer et al (Colomer, 1999) published an improved method for SWNTs production. In his work the synthesis of SWNTs is carried out by ethylene decomposition on supported metal catalysts prepared by impregnation of various supports. Co, Ni, and Fe and their binary and ternary mixtures are used as the metallic components, while the supports are silica and alumina. The role of the support is to disperse metal particles. Catalysts are prepared by combining sonication and impregnation. The reaction is carried out in a fixed bed quartz reactor at 1080 oC for one hour. It was shown by Transmission Electron Microscopy that depending on the metal composition either individual SWNTs or SWNTs bundles are formed. In fact, the best yield of SWNTs is obtained by using Fe-Co binary mixture supported by alumina. It was shown that the synthesis of SWNTs is generally more efficient on alumina than on silica. Moreover, in the case of silica the SWNTs are coated by a thick layer of amorphous carbon. The bundles formed by this technique seem to be very similar to the bundles synthesized by arc-discharge or laser evaporation techniques and their length is up to a few microns.
Many possible applications of CNTs require mass production of high-quality CNTs. To date, various synthetic methods, e.g. arc discharge, laser vaporization, pyrolysis, and plasma-enhanced and thermal chemical vapor deposition have been developed. Recently, the synthesis of CNTs using CVD method has been extensively studied because of relatively lower growth temperature (below 1000 oC) compared to arc-discharge or laser vaporization methods. Moreover, the CVD method offers a route to control over the diameter, length, and alignment of CNTs.
Understanding of the growth mechanisms of CNTs would help in controlled growth and development of new synthesis methods. However, the complexity of the growth process makes it difficult to investigate at the molecular level. Given the different synthetic techniques, a variety of growth mechanisms were suggested. In the synthesis using plasma-enhanced CVD, the catalytic particles were usually encapsulated at the tip of CNTs (Ren, 1998; Murakami, 2000), which was explained by adopting a tip growth mechanism. Wang and co-workers (Li, 2000) reported the capped Fe catalytic particles in the bamboo-shaped CNTs synthesized by the pyrolysis method and proposed a tip growth mechanism involving two sized catalytic particles. In contrast, Dai and co-workers (Fan, 1999) suggested a base growth mechanism that the CNTs grow upward from the metal particles attached to the substrate.
It has been postulated that the formation and growth of nanotubes are an extension of other known processes in which graphitic structures are formed over metal surfaces at temperatures below 1000 oC from carbon. It is also very obvious that the shape of graphitic carbon produced depends on the chemical properties and the physical dimensions of the metal particles. The most efficient metals were found to be Fe, Co, and Ni. The peculiar ability of the mentioned metals to form graphitic carbon layers is thought to be related to the combination of different factors. These include their catalytic activity towards the decomposition of carbon compounds, the possibility to form unstable carbides and finally, the speed of carbon diffusion in the metal particle, which must be extremely rapid. Derbyshire et al (Derbyshire, 1975) showed that over thin metal foils carbon dissolves to form a solid solution. After cooling, the carbon precipitates onto the surface, forming a continuous thin film of highly crystalline graphite parallel to the metal surface. The high crystallinity obtained in a very short time shows that carbon atoms are extremely mobile and able to move easily over and through metal.
There are a vast number of ways in which graphite sheet can be rolled up to form a seamless cylinder and hence a wide variety of nanotubes exist. Nanotubes are characterized by their chiral (or wrapping) vector, c (Dresselhaus, 1996) such that c = na1 + ma2 where a1 and a2 are the basis vectors of the graphene lattice and m , n are integers (see figure 1). The chiral vector spans the circumference of the tube formed by joining the dotted lines shown in figure 1. Those tubes with chiral vectors of the form (n, 0) are termed zigzag tubes (figure 2(a)) whereas when n = m a so-called ‘armchair’ tube results (figure 2(b)). All other values of n and m produce a chiral tube (figure 2(c)). Closely following the synthesis of nanotubes came the remarkable theoretical prediction that their electronic properties could be changed between metallic or semiconducting simply by varying the tube diameter or its helicity, i.e. by changing the values of n and m (Hamada, 1992; Saito, 1992). A key theoretical result is that armchair nanotubes are metallic, whereas for all other SWNTs, when n – m = 3l (l an integer) the tubes are metallic; otherwise they are semiconducting.
The electronic conductivity of CNTs has been predicted (Hamada, 1992; Saito, 1992) to depend sensitively on tube diameters and wrapping angle, with only slight differences in these parameters causing a shift from a metallic to semiconducting state. In other words, similarly shaped molecules consisting of only one element (carbon) may have very different electronic behavior. Although the electronic properties of multi-walled and single-walled nanotubes have been probed experimentally, it is difficult to relate these observations to the corresponding structure. It has been observed (Odom, 1998) in both metallic and semiconducting carbon nanotubes that the electronic properties do depend sensitively on the wrapping angle. The bandgaps of both tube types have been shown to be consistent with theoretical predictions (Wildoer, 1998).
Although a number of groups carried out conductance measurements on multi-walled tubes using both scanning tunneling spectroscopy (Dresselhaus, 1996) and two- and four-probe geometries (Ebbesen, 1996), conclusive experimental verification of the theoretical prediction outlined above appeared only recently (Odom, 1998). Figure 3 shows atomically resolved STM images of a range of nanotubes: armchair, zigzag, and chiral tubes with different degrees of chirality. By acquiring scanning tunneling spectra for each of these types of tube it was possible to correlate their structural and electronic properties.
Figure 4(a) shows the tunneling spectra, acquired at 4 K, for various tubes. Their derivatives (dI/dV) are shown in figure 4(b). These, apart from the contribution of a bias-dependent transmission term (Feenstra, 1993), provide a good measure of the local density of states of the tubes. Wildoer et al (Wildoer, 1998) observe two distinct ‘families’ of dI/dV spectra for chiral tubes: those with a gap of ~0.5 – 0.6V (spectra 1 – 4 in figure 3(a)) and those with a much larger apparent gap, 1.7 – 2.0V (spectra 5 – 7). The gap values for the first family of curves are plotted versus the tube diameter in figure 4(c).
The diverse electronic properties of CNTs open a possibility of developing nanoelectronic devices by combining metallic and semiconducting nanotubes (Chico, 1996). These nanoelectronic devices would be formed by combining many nanotubes components with different electronic properties, similarly to microelectronic devices. Even though it has been theoretically shown that nanotubes heterojunctions can have the desired properties as nanoelectronic components (Chico, 1996; Yao, 1999), it would be very difficult to form many junctions in a precisely controlled manner. Another promising approach is the modification of different parts of a single nanotubes to have different electronic properties using controlled mechanical or chemical processes.
Recently, a molecular-scale device element based on a suspended, crossed nanotubes geometry that leads to bi-stable, electrostatically switchable ON/OFF states was introduced (Rueckes, 2000). The device elements are naturally addressable in large arrays by the carbon nanotube molecular wires making up the devices. These reversible, bi-stable device elements could be used to construct non-volatile random access memory and logic function tables at an integration level approaching 1012 elements per square centimeter and an element operation frequency in excess of 100 GHz. Also, various possibilities of CNTs for electronic applications such as field-effect transistors and resonant tunnel transistors have been reported (Antonov, 1999; Trans, 2000). However, these require an ability to synthesize, isolate, manipulate, and connect individual nanotubes. Dai and co-workers (Kong, 1998; Zhou, 2000) showed a novel strategy for making high-quality individual nanotubes, grown at 1000 oC, bridging two metallic islands on silicon wafers patterned on the micrometer scale. Their functional applications for electronic devices were also demonstrated. Also, Papadopoulos et al introduced a Y-junction formation technique using branched nanochannel alumina templates (Papadopoulos, 2000). For electronic applications the grown CNTs were removed from the templates and dispersed onto pre-patterned electrodes. Finally, and electronic filed was applied to form a low-resistance contact to the electrodes.
A paper by Li et al (Li, 1999) presents direct nanowiring of carbon nanotubes using growth barrier technology utilizing chemical vapor deposition (CVD) at the relatively low process temperature of 650-750 oC. Using this method, CNTs bridging two parallel patterned structures with a perfect “Y-junction” or “straight line” were formed. The most important aim of their work is to demonstrate that good selectivity is achieved using a low-temperature CVD process as well as conventional photolithography, which is the cheapest method and easily combinable with the conventional Si-based process.
First, in their experiment (Li, 1999) the layer structure was prepared by stacking SiO2-Ni-Nb layers on a doped Si substrate, where Nb metal is introduced as a barrier layer for vertical growth, covering the top of the Ni catalytic layer. A key technique of this method is to employ the thermally stable Nb layer for the CNT growth barrier as well as the electrode terminals after growth. In most cases, not only the high growth temperatures but also the chemical properties of process gases limit the thermal stability of the barrier material laid on top of the catalytic Ni during CNT growth.
Because the selectively grown CNTs from Li’s experiments formed highly robust contacts, measurements of their electrical properties were feasible and the obtained results are reliable. The typical resistance versus temperature characteristics showed that the resistance of the wire decreased from about several tens of kilo-ohms at 295 K to 80 Mega-ohms at 90 K. The origin of the rapid increase of the resistance could be attributed to the weak localization interaction and effect of magnetic material of Ni pads existing at both ends of the CNT (Li, 1999).
The results of Li’s work show a promising technique for forming CNT bridges between electrodes for highly integrated electronic devices using a growth barrier technique. These “straight line” and “Y-shape” CNT bridges act as active components as well as nanowires for the achievement of molecular-scale electronic devices with high integration density and are a key building block for ultra-miniaturized nanomachines (Papadopoulos, 2000).
While there has been considerable progress in understanding the electrical properties of nanotubes, as well as exploring their possible applications (Bockrath, 1997; Tans, 1998), what has been lacking is an efficient strategy for integrating nanotubes into electronic structures. In particular, it would be desirable from both a scientific and technological point of view to control not only the diameter and chirality of a SWNT, but also its length, position, and orientation. It is equally desirable to develop a means of making robust, low-resistance electrical contacts between nanotubes and metallic electrodes. These goals pose challenges to nanotubes synthesis, processing, and assembly strategies.
In 1998 Kong et al (Kong, 1998-2) developed a CVD method to synthesize high quality SWNTs on catalytically patterned surfaces. This technique readily yields large numbers of SWNTs at specified locations, and opens up new possibilities in addressing the issues raised above. A paper by Soh et al (Soh, 1999) describes the combination of this synthesis method with microfabrication techniques to obtain many nanotubes-based electrical circuits on a single substrate with controllable positions and length (ranging from 300 nm to 10 microns), and to connect the tubes to macroscopic electrodes. They find that the contacts formed in this way often have very low resistance, of order of the quantum of resistance (tens of kilohms) even at low temperatures.
The synthesis procedure (Soh, 1999) involves methane CVD on SiO2 substrates that contain patterned catalyst islands. First, polymethylmethacrylate (PMMA) resist is spun onto a substrate. Electron-beam lithography is used to expose the PMMA film followed by developing the exposed resist to obtain an array of square wells in the PMMA film. The catalyst material consists of 15 mg of alumina nanoparticles, 0.05 mmol Fe(NO3)3 × 9H2O and 0.015 mmol of Mo2(acac)2 in 15 ml of methanol. The mixture is stirred for 24 h and sonicated for 1 h before being deposited onto the substrate containing the patterned PMMA film. After solvent evaporation, the PMMA film is lifted off to obtain an array of catalyst islands on the substrate. CVD synthesis is carried out for 10 min at 900 oC in a 1-inch-diameter tube furnace using a methane flow rate of 5000 ccm/min. Under these conditions, nanotubes grown from the catalyst islands are predominately individual SWNTs with few structural defects (Kong, 1998). The diameter distribution of the SWNTs is in the range of 0.7 – 5 nm with a peak at 1.5 nm. The synthesized nanotubes emanate from the catalyst islands and are often found forming bridges between adjacent islands. SWNT bridging two islands are frequently observed (Kong, 1998).
After the CVD synthesis of a single-walled nanotubes chip, electrical contact pads are placed over the catalyst islands (15 nm of Ti followed by 60 nm of Au) by electron-beam lithography and metal evaporation (Soh 1999). This step utilized predefined alignment marks so that the metal contacts fully cover the catalyst islands and extended over their edges by 0.5 microns. Thus, the bridging SWNTs are contacted by the Ti metal film at both ends and sides. The interactions between tube-island and tube-substrate are mechanically strong (Kong, 1998), allowing the nanotubes to withstand mechanical forces in the lithographic step for metal contacts. By setting the separation between the catalytic islands and varying the spacing between the edges of the metal pads, they (Soh, 1999) were able to control length of the nanotubes used for transport measurements. The lengths of the individual SWNTs that were measured were in the range of 0.3 – 10 microns. When the pads are closely spaced, multiple tubes tend to bridge the gap.
Circuits formed by nanotubes bridges between metal electrodes were first characterized at room temperature by measuring electrical resistance. Resistances ranging from 15 kilohms to several Megaohms were observed. Soh et al (Soh, 1999) found that the lower resistance tubes tend to remain good conductors at low temperatures.
The studies published by Soh and Kong (Soh, 1999; Kong, 1998; Kong 1998-2) present a chemical vapor deposition synthetic approach to high-quality single-walled carbon nanotubes. By patterning substrates with regular arrays of catalyst islands, SWNTs were grown at well-defined locations on surfaces, which allow for controlled integration of nanotubes into electrical architectures by combing microfabrication techniques. The approach represents a new chemical route to novel nanotubes materials that can be readily addressed individually. This synthetic approach opens up new window in systematically studying the physics in one-dimensional systems and exploring molecularly scale electrical devices at a large scale.
The publications also presented the results of electron transport studies of several representative SWNTs. The resistance of individual SWNTs was found to be about 100 times smaller than anything reported previous to these studies. It was postulated that a good electrical contact is established between the metal pads and the ends of the nanotubes. In particular, at the end where the nanotube originates from, chemical bonds between nanotube and the metal catalyst particle should be accounted for the excellent electrical contact (Kong, 1998). Future work in nanotubes synthesis will focus on controlling the orientation of the SWNTs and the diameter and length of the nanotubes. Transport studies will focus on elucidating the intrinsic properties of various nanotubes as well as the nature of metal-tube contacts (Soh, 1999)
Electronic transport measurements have already revealed a wide range of effects that are potentially useful for electronic device implementation. Presently, one of the most difficult problems in fabrication of electronic devices is contacting individual CNTs with macroscopic contacts that are necessary for communication with the external world. The usual method for contacting CNTs in research devices relies on adaptation of convectional microfabrication techniques for either positioning CNTs on top of prefabricated leads, or first localizing randomly deposited CNTs on a substrate, and then fabricating electrodes over them (Trans, 1997; Pablo, 1999). These methods are clearly not feasible for mass production of CNT based devices. The contacts realized by these approaches are typically side-contacts (Anantram, 2000), meaning that the leads are electrically contacted only with the outside wall of the CNT. Such contacts result in weak interaction between the carbon and metal atoms, and are characterized by high contact resistance.
It has been demonstrated that chemical vapor deposition can be utilized to grow selectively single wall carbon nanotubes on prepatterned micrometer-scale catalyst islands. These experiments showed that SWNTs could spontaneously interconnect the catalyst islands, forming metal lead/CNT/metal lead structures (Kong, 1998). In a study by Wei et al (Wei, 2000), they describe an alternative method for direct fabrication of CNT circuits. In comparison with the previous work, they simplified patterning such that it involves only a single e-beam lithography step, and instead of a specially prepared catalyst nanoparticle matrix they are using a continuous thin film of Fe as catalyst. The CVD technique is also different. Instead of hot wall CVD, typically used for CNT CVD growth, they used a molecular jet CVD technique. Electronic transport measurements were used to study the properties on CNTs and CNT-metal contacts.
Wei’s study (Wei, 2000) involved electrode structures fabricated on 500-nm-thick films of SiO2 that was thermally grown on Si(100). Schematic diagram of the sample fabrication sequence is shown in Fig. 5. Conventional electron-beam lithography was used to first define an electrode pattern with a desired gap size from 1 to 3 microns (see Figs. 5 (a) and (b)). Following pattern definition, 10 nm of Ti followed by 5 nm of Fe were deposited by electron-gun evaporation. The Fe serves as the catalyst for CNT growth, and the Ti prevents delamination of the Fe layer at elevated temperatures. The schematic of the electrode structure after lift-off is shown in Fig. 5 (c). CNTs were grown on these samples by CVD using acetylene at a chamber pressure of 100 mTorr and a substrate temperature of 660 °C. Under these conditions CNTs selectively grow only on catalyst leads and eventually form a bridge between nearby electrodes as illustrated on the sketch in Fig. 5 (d).
Wei et al (Wei, 2000) used a standard multimeter was used to measure the resistance of the CNTs through the catalyst leads at room temperature (RT). Most of the samples have a resistance below 100 kV, which agrees with values reported by others for MWCNT-metal contacts (Pablo, 1999). Electronic transport measurements as a function of temperature were conducted to characterize the properties of CNT circuits. At RT, a linear I – V was observed indicative of metallic behavior. With decreasing temperature, the I – V gradually became nonlinear, and a low conductance region emerged around V=0. Nonlinear behavior was dominant at low temperatures. The nonlinear region was not affected by a gate voltage that was applied to the Si substrate. Variation of the gate voltage in the range from 210 to 10 V failed to confirm either a field-effect or Coulomb blockade. Further increasing the gate voltage caused current leakage. Similar absence of a gate effect with MWCNTs has been reported by others (Martel, 1998). One sample was found to exhibit a decreasing current with increasing source-drain voltage - a behavior known as negative differential resistance. A weak dependence on gate voltage was observed for these I – V curves. This observation confirms the presence of a functional gate and indicates that the absence of gate voltage modulation of the transport properties is a characteristic of the CNT circuits.
Wei’s studies (Wei, 2000) describe a method for directed assembly of CNT electronic circuits. The CNTs selectively grow on predeposited Fe catalyst electrodes which also serve as electrical leads. The processing parameters can be optimized such that a single CNT bridges two adjacent electrodes. These connections were robust and routinely survived repeated temperature cycling from room temperature to 2 K during electronic transport measurements. The room temperature resistance of the carbon nanotube circuits measured at the electrode leads was typically less than 100 kV. At low temperature a gap in the conductance of the CNTs appears at V=0, but it was not affected by changes in the gate voltage. The origin of the nonlinear response at low temperature is unclear at the present time. Fabrication of CNT circuits with more complex functionality can be envisioned by utilizing appropriately designed catalyst patterns (Wei, 2000).
Influence of CVD Synthesis Variations on Electrical Impedance (R, C, and L)
So far, the nanotube-based electronic devices that have been investigated have relied on the DC response of the nanotubes. It is, however, well known that the AC response of devices is at least equally, and often even more, important for understanding and designing high speed applications. AC transport measurements are of great interest because they provide information about the electrochemical capacitance, the nonequilibrium charge distribution, the dynamic coupling of devices, and the transport dynamics of a conductor. Moreover, the induced charges within a conductor, on a nearby gate, or on the surface of an electron reservoir can all play an important role in determining AC transport. While quantum transport under DC condition is well understood, the AC response of systems is complicated by the presence of time-dependent fields that can take the system out of equilibrium. Under AC conditions, electro-dynamics shows that displacement currents are induced, which can substantially alter the transport properties of the system (Buttiker, 1993). In order to predict the dynamic conductance of nanoscale conductors such as the carbon nanotubes, the inclusion of these displacement currents into the theoretical formalism is absolutely necessary for two fundamental reasons. First, they are needed if the total current in the system is to be conserved. Second, quantum transport must be gauge invariant, which implies that the physics will depend only on the voltage differences (Buttiker, 1993).
In 1998 Liu et al (Liu, 1998) investigated the inductance and capacitance of CNT. Chemical vapor deposition (CVD) of hydrocarbons on a metal-sensitized surface is a convenient method to control the shape and properties of carbon nanotubes. It also provides a new class of carbon nanotube/inorganic composite materials. With conventional technology, capacitors can be obtained in integrated circuits by utilizing the transition capacitance of a reverse-biased p-n junction or a thin-film technique. No practical inductance values have been obtained on silicon substrates, and the use of inductors is avoided in circuit designs wherever possible. Control of the formation conditions of carbon nanotubes on sensitized surfaces provides a new vision of carbon electronics.
Carbon nanotubes were prepared by CVD of hydrocarbons (Liu, 1998) such as methane, propane, and LPG (liquefied petroleum gas, a mixture of propane and butane). Carbon was deposited on alumina granules (g-alumina with low soda, 1/8 in. x 1/8 in. cylindrical form, purity 99+%, lot # 129307-s, Strem Chemicals, Inc.) from hydrocarbon/hydrogen mixtures under atmospheric conditions. The alumina surface was sensitized by nickel (Ni/alumina). Alumina granules were added to a 0.1 M aqueous solution of nickel nitrate, filtered, and dried at 150 oC. The Ni/alumna was calcined at 900 oC to form the nickel aluminate phase, followed by an in situ reduction to form highly dispersed metallic nickel. For the low temperature CVD, the Ni/alumina was calcined at 330 oC. The Ni/alumina granules were placed in a quartz tubular reactor (14 mm internal diameter) in an electric furnace. The LPG/hydrogen mixture was introduced at 550 oC, propane/hydrogen at 900 oC, and methane/hydrogen at 1030 oC.
Sample 1, prepared by carbonization at the high temperature of 1030 oC for 6 h (Table 1, no. 1), was a good conductor with a low resistance of 3.1 ohms. To understand properties of colligated carbon nanotubes on an alumina surface, a high-frequency (max. 35 MHz) rectangular wave (Fig. 6, input pulse) as applied to the carbon nanotube/Ni/alumina bead and the electronic behavior monitored. Sample 1 generated multiple damping waves (Fig. 6A, no. 1), representing the characteristic inductive effect. The large number of damping waves obviously indicates the presence of inductance originating from the carbon nanotube/alumina sample. The same behavior with fewer damping waves was observed for samples 2 and 3, which were prepared under the same preparation conditions except for shorter carbonization times. The appearance of the damping waves in the output confirmed the formation of an RL circuit in a single bead of carbon nanotube/alumina. An individual carbon nanotube of graphite-like structure is known to behave like a wire. Interweaving wires should lead to the current lagging behind the voltage, generating the inductive character.
For increasing carbonization time from 2 to 6 h, the carbon content increased from 0.121 to 0.292 gcarbon/galumina (samples 3-1). It is important to notice the increment of inductance from 1.0 to 1.4 nH along with the decrease in resistance from 10.0 to 3.1 ohms. The inductance increased as the carbon content increased (Liu, 1998). Interestingly, with low carbon contents of 0.086 and 0.045 gcarbon/galumina, the carbon nanotube/alumina samples behave as capacitors of 1.6 and 2.9 pF, respectively (samples 4 and 5, see Table 1 and Fig. 6B, no. 5). The electronic properties of the carbon nanotube/alumina sample prepared at 1030 oC changed from those of a capacitor to those of an inductor as the carbonization time increased. These results indicate that carbon nanotubes grown on the Ni/alumina surface at 1030 oC were inductors and capacitors generating RLC circuits.
On the other hand, beads with carbon nanotubes grown at low temperatures of 550-580 oC showed different behavior when the high-frequency wave was applied. Sample 6, which was prepared at 550 oC, had a carbon content of 0.047 gcarbon/galumina (similar to sample 5) and was a poor conductor with 45 kilohms resistance (Liu, 1998). The shape of the output with a rectangular wave applied was indicative of the presence of a capacitor. Sample 6 showed a long rise time (tr) of 150 ns (Fig. 6, no. 6), exhibiting the characteristics of a differentiating circuit consisting of a resistance and a capacitor in a single bead. The output pulse indicated that the bead functioned as a capacitor and that an RC circuit had been formed from the carbon nanotubes on Ni/alumina. The capacitance was expected to originate from individual carbon nanotubes. Direct measurements showed sample 6 to have a capacitance of 0.4 pF. The rise time of an RC circuit, which depends on both resistance and capacitance, is an important factor in the design of microcircuits (Ketchum, 1965). Sample 7, with the higher carbon content of 0.176 gcarbon/galumina, produced at 580 oC for 3 h carbonization time, showed a rise time of 130 ns due to the lower resistance of the bead, although it had the same capacitance as sample 6.
It is interesting to note that sample 7 did not show inductance but only capacitance, although its carbon content was between those of samples 1 and 2, which had inductive character. The high crystallinity of the carbon nanotubes formed on the alumina surface was an important factor for the inductive properties of the system. Capacitance became higher for lower carbon contents with increased crystallinity of the carbon nanotubes. The crystallinity of the carbon nanostructures on the alumina surface might be attributable to recrystallized alumina. At the high temperature of 1030 oC, recrystallized alumina provided growth points for carbon (Liu, 1998). Sample 9, prepared at 900 oC with 0.083 gcarbon/galumina, displayed a short rise time of 35 ns; the shape of the output of the resulting circuit was a rectangular wave almost the same as the input pulse. Such a rectangular output was unexpected and seems to be attributable to the RL and RC characters counterbalancing each other.
The above results show that the electronic character of carbon nanotubes on an alumina surface depends on the preparation conditions, such as sensitization, pretreatment, and carbonization conditions. The development of the hydrocarbon CVD method for inorganic substrates allows the control of the R, L, and C characters of the carbon nanotube/substrate combinations and enables the design of RLC circuits on substrates. The carbon nanotube microcircuit differs from conventional ones, where separate capacitors, inductors, and resistors are linked two-dimensionally to perform electronic functions. Carbon nanotubes on an alumina surface have the mixed properties of a resistor, a capacitor, and an inductor in a single bead. Moreover, it could be a three-dimensional electronic circuit if the CVD conditions are controlled as a function of time. The characteristics of the carbon nanotube microcircuits can be designed and controlled effectively by optimizing the CVD conditions. If a micropatterning technique is employed to sensitize the surface of substrates, different kinds of microcircuits can be fabricated and a series of electronic functions can be arranged on the surface of substrates. The results described above provide a new vision of carbon electronics. The formation of three-dimensional carbon nanotube RLC circuits can open up a new avenue for the application of carbon nanotubes. The development of nanoscale circuits would be a key technology for next-generation nanoscale electronics.
Anantram M.P., Datta S., and Xue Y. 1999. http://xxx.lanl.gov/abs/cond-mat/9907357
Antonov R.D. and Johnson A.T. 1999. Phys Rev Lett. 83, 3274
Bockrath M., Cobden D.H., McEuen P.L., Chopra N.G., Zettl A., Thess A., and Smalley R.E. 1997. Science 275, 1922.
Buttiker M., Pretre A., and Thomas H. 1993. Phys. Rev. Lett. 70, 4114
Chico L. 1996. Phys. Rev. Lett. 76, 971
Colomer J.F., Bister G., Willems I., Konya Z., Fonseca A., Van Tendeloo G., and Nagy J.B. 1999. J. Chem. Soc., Chem. Commun. 1343-1344.
Dai H.J., Rinzler A.G., Nikolaev P., Thess A., Colbert D.T., and Smalley R.E. 1996. Chem Phys. Lett. 260, 471-475.
Derbyshire F.J., Presland A.E.B., and Trimm D.L. 1975. Carbon, 13, 111-113.
Dresselhaus M.S., Dresselhaus G., and Eklund P.C. 1996. Science of Fullerenes and Carbon Nanotubes (Academic Press, San Diego, CA, 1996).
Ebbesen T.W., Lezec H.J., Hiura H., Bennet J.W., Ghaemi H.F. and Thio T. 1996. Nature 382, 54
Fan S., Chapline M.G., Franklin N.R., Tombler T.W., Cassell A.M., and Dai H. 1999. Science 283, 512.
Feenstra R.M. 1993. Scanning Tunneling Microscopy (Boston, MA: Academic)
Hamada N., Sawada S-I, and Oshiyama A. 1992. Phys. Rev. Lett. 68, 1579
Hernadi K., Fonseca A., Nagy J.B., Bernaerts D., and Lucas A. 1996. Carbon 34, 1249-1257
Iijima S., 1991. Nature 354, 56-58.
Ivanov V., Nagy J.B., Lambin Ph., Lucas A., Zhang X.B., Zhang X.F., Bernaerts D., Van Tendeloo G., Amelinckx S., Van Landuyt J. 1994. Chem. Phys. Lett. 223, 329-335.
Kong J., Cassell, A.M., and Dai H.J. 1998. Chem. Phys. Lett. 292, 567-574.
Kong J., Soh H., Cassell A., Quate C.F., Dai H. 1998. Nature 395, 49
Ketchum D.J. and Alvarez E.C. 1965. Pulse and Switching Circuits. McGraw-Hill, New York
Li W.Z., Xie S.S, Qian L.X., Chang B.H., Zou B.S., Zhou W.Y., Zhao R.A., and Wang G. 1996. Science 274, 1701-1703.
Li J., Papadopoulos C., Xu J.M., 1999. Nature 402, 253
Li D.C., Dai L, Huang S., Mau A.W., Wang Z.L. 2000. Chem. Phys. Lett. 316, 349.
Liu K., Roth S., Duesberg G.S., Kim G.-T., Schmid M. 1998. AIP Conf. Proc. (USA). 442, 61-64
Martel T., Schmidt T., Shea H.R., Hertekl T., and Avouris Ph. 1998. Appl. Phys. Lett. 73, 2447.
Murakami H., Hirakawa M., Tanaka C., Yanakawa H. 2000. Appl. Phys. Lett. 76, 1776
Odom T.W., Huang J.L., Kim P., and Lieber C.M. 1998. Nature 391, 62-64.
Pablo P.J., Graugnard E., Walsh B, Andres R.P., Datta S., and Reifenberger R. 1999. Appl. Phys. Lett. 74, 323.
Papadopoulos C., Rakitin A., Li J., Vedeneev A.S., and Xu J.M. 2000 Phys. Rev Lett. 85, 3476
Ren Z.F., Huang Z.P, Xu J.W., Wang J.H., Bush P., Siegal M.P., Provencia P.N. 1998. Science 282, 1105
Rueckes T., Kim T., Joselevich E., Tsewng G.Y., Cheung C-L, Lieber C.M. 2000. Science 289, 94
Saito R., Fujita M., Dresselhaus G., and Dresselhaus M.S. 1992. Appl. Phys. Lett. 60, 204
Soh H.T., Quate C.F., Morpurgo A.F., Marcus C.M., Kong J., and Dai H. 1999 Appl. Phys. Lett. 75, 627-629
Sun L.F., Mao J.M., Pan Z.W., Chang B.H., Zhou Y, Wang G., Qian L.X., and Xie S.S. 1999. Appl. Phys. Lett. 74, 644.
Trans S.J. and Dekker C. 2000. Nature 404, 834.
Tans S.J., Devoret M.H., Dai H., Thess A., Smalley R.E., Geerligs L.J., and Dekker C. 1997. Nature 386, 474
Wei Y.Y., and Eres G. 2000. Nanotechnology 11, 61-64
Wildoer J.W.G., Venema L.C., Rinzler A.G., Smalley R.E., Dekker C. 1998. Nature 391, 59-62.
Yacaman M.J., Yoshida M.M, Rendon L., and Santiesteban J.G. 1993. Appl. Phys. Lett. 62, 202-204.
Yao Z. 1999. Nature 402, 971
Yudasaka M., Kituchi R., Matsui T., Ohki Y., and Yoshimura. 1995. Appl. Phys. Lett. 67, 2477-2479.
Zhou C., Kong J., and Dai H. 2000. Phys. Rev. Lett. 84, 5604.
TOP OF PAGE | BACK to engineering credentials | BACK to engineering projects | JAMES' HOME
last modified 8 january 2002 send problems and queries to webmaster at: jamessolberg@hotmail.com