Precision Engineering 28 (2004) 16–30

Nanotechnology and nanostructured materials:trends in carbon nanotubes

A.G. Mamalisa,∗, L.O.G. Vogtländerb, A. Markopoulosaa Department of Mechanical Engineering, Manufacturing Technology Division, National Technical University of Athens,

9 Iroon Polytechniou Avenue, Athens 15780, Greeceb Department of Mechanical Engineering, Production Technology and Organisation Division,

Delft University of Technology, Delft, The Netherlands

Received 13 June 2002; received in revised form 25 October 2002; accepted 20 November 2002

Abstract

Carbon nanotubes have attracted the attention of many researchers since their discovery last decade. These carbon molecules are tinytubes with diameters down to 0.4 nm, while their lengths can grow up to a million times their diameter. Using their remarkable electricalproperties, simple electronic logic circuits have been built. These structures are promising for the semiconductor industry which is leadingthe search for miniaturisation. They are not only very good conductors, but they also appear to be the yet found material with the biggestspecific stiffness, having half the density of aluminium. This paper is written to give a consolidated view of the synthesis, the propertiesand applications of carbon nanotubes, with the aim of drawing attention to useful available information and to enhancing interest in thisnew highly advanced technological field for the researcher and the manufacturing engineer.© 2003 Elsevier Inc. All rights reserved.

Keywords:Nanotechnology; Nanostructured materials; Carbon nanotubes

1. Introduction

Nanotechnology is considered to be the technology of thefuture, it is perhaps today’s most advanced manufacturingtechnology and has been called “extreme technology”, be-cause it reaches the theoretical limit of accuracy which is thesize of a molecule or atom. In manufacturing industry, twointerrelated trends are clearly seen: the trend towards minia-turisation and the trend towards ultra-precision processing.Both trends are moving in the direction of nanotechnology,because both are tending to dimensions which lie in the rangeof several nanometres.

Nanotechnology deals with materials and systems havingthe following key properties[1]:

• they have at least one dimension of about 1–100 nm;• they are designed through processes that exhibit funda-

mental control over the physical and chemical attributes ofmolecular-scale structures;

• they can be combined to form larger structures.

∗ Corresponding author. Tel.:+30-1-772-3688; fax:+30-1-772-3689.E-mail address:mamalis@central.ntua.gr (A.G. Mamalis).

Taniguchi introduced the term ‘nanotechnology’ in 1974to describe the manufacturing of products with tolerancesless than 1�m [2]. However, Feynman, who won the No-bel prize in 1965, claimed in his talk “There’s plenty ofroom at the bottom” in 1959, that “. . . almost any chemi-cally stable structure, that can be specified, can in fact bebuilt . . . ” [3]. He introduced the concept of building withmolecules, “bottom-up” manufacturing, in contrast with the“top-down” manufacturing, we are familiar with. The pi-oneering work of Drexler in molecular nanotechnology isimportant here; in his popular books “Engines of creation”(1986) and “Nanosystems” (1992) he described nanoscale“assembler”-robots which build structures molecule bymolecule and even replicate themselves[4,5].

To image these tiny structures, special microscopes areneeded. Scanning electron microscopes image structures byanalysing the scattered electrons on a substrate by computer.Scanning probe microscopes use extremely sharp probes withtips of radius about 10 nm to scan the surface. The scanningtunnelling microscope measures a tunnelling current whichoccurs when the tip is about 1 nm above the surface and a volt-age is applied; the current is held constant by moving the tipvertically while scanning the surface. The restriction that thesubstrate has to be an electrical conductor, led to the inven-

0141-6359/$ – see front matter © 2003 Elsevier Inc. All rights reserved.doi:10.1016/j.precisioneng.2002.11.002

A.G. Mamalis et al. / Precision Engineering 28 (2004) 16–30 17

tion of the atomic force microscope. This device also uses aprobe, but this one is attached to a flexible (in vertical depic-tion) cantilever, which is pressed into light contact with thesurface while scanning. The vertical movement is followedby detecting the reflections of a laser beam on it; this is onetype of AFM, however, several short range very high resolu-tion displacement transducers are used in different types ofAFM. Computers construct the final images. These micro-scopes can also be used to move nanoscale objects (and evensingle atoms) on a surface, being an important device to han-dle nanotubes.

To produce nanoscale shapes or lines, special processes areneeded. The most important ones all deal with some kind ofenergy beam, which reduces the material by ablation (instantvaporisation). These techniques are (in order of increasingpower): photolithography, X-ray lithography, electron beammachining, focused ion beam machining, laser beam machin-ing (femtosecond lasers and excimer lasers). A detailed ac-count on engineering nanotechnology can be found in[6,7].

2. Nanostructured materials—nanotubes

Of increasing interest in nanotechnology are the nanos-tructured materials, with dimensions, i.e. grain size, layerthickness or shapes, below 100 nm. This wide group of ma-terials enables access to new ranges of electronic, magnetic,mechanical or optical properties. Polycrystalline materialswith grain sizes less than a few nanometres exhibit differentproperties because they are relatively highly affected by thegrain boundaries. They appear to be very strong; for exam-ple, highly wear resistant coatings are being developed outof these materials.

Researchers at IBM used magnetic nanoparticles withinseveral ultra-thin layers to develop advanced data storage de-vices. Sensors for disk-drives have been developed with manytimes the sensitivity of previous devices, allowing more bitsto be packed on the surface of each disk. Nanoscale struc-tures can potentially store trillions of bits of data per squareinch, giving them a capacity 10–100 times greater than thatof present memory devices.

Nanotubesbelong to a promising group of nanostruc-tured materials. Although other nanotubes based on boronnitride and molybdenum have been reported, currently car-bon nanotubes are by far the most important group. Thesetubes contain one or several concentric graphite like layerswith diameters in the range of 0.4 nm up to tens of nanome-tres. Their main structural characteristics, processing andapplications are reviewed below.

2.1. Nanotubes structure

The discovery of the “buckyball”, i.e. a football shaped C60molecule, reported by Kroto et al. in 1985[8], had a strongimpact and marked the beginning of a new era in carbon ma-terial science. Iijima in 1991 discovered the carbon nanotube

Fig. 1. High resolution TEM image of the end of a typical nanotube showingseveral concentric layers with caps at the end[10].

[9], in the soot at the negative electrode of an arc discharge lit-tle tubes mixed with a large amount of other forms of carbonwere found. Such multi-walled carbon nanotubes (MWNT)contained 2–50 concentric graphite cylinders with a diameterof 3–10 nm and a length of up to 1�m. This initial work ledmany groups throughout the world to produce and purify nan-otubes. Soon it became clear that nanotubes have unique elec-tronic and mechanical properties that are expected to lead tobreakthrough industrial applications. Later on, single-walledcarbon nanotubes (SWNT) were developed. Because of ad-hesive forces nanotubes often bunch to form ropes. The tubescan either be open-ended or have caps formed from half aC60 molecule at either end (seeFig. 1).

To explain the carbon structure in nanotubes, the differencebetween diamond structure and that of graphite should beconsidered (seeFigs. 2 and 3). In diamond, each carbon atomis attached to four others in a three-dimensional lattice, whichgives diamond its strength. On the other hand, in graphite,each carbon atom is attached to three others in a plane andform a hexagonal lattice, whilst the remaining bond is usedto hold the planes above and below. The bonds in the planeare stronger than in diamond, but the interplanar bonds arerelatively weak and enable the planes to slide. Therefore,whereas diamond is isotropic, graphite is anisotropic.

The structure of a nanotube is similar to that of graphite,with the difference that the sheets are closed to form a tube.In the ideal case, a carbon nanotube consists of either onecylindrical graphite sheet (single-walled nanotube) or severalnested cylinders (multi-walled nanotube) with an interlayerspacing of 0.34–0.36 nm, that is close to the typical atomicspacing of graphite. The C–C bonds have a length of 0.14 nm,which is shorter than the bonds in diamond, indicating thatthe material is stronger than diamond[12].

The rolling-up of the hexagonal lattice can be performedin different ways. The sheet can be rolled-up along one ofthe symmetry axes, producing, either a zigzag or an armchairtube (seeFig. 4). It is also possible to roll-up the sheet ina direction that differs from a symmetry axis, therefore achiral nanotube can be obtained. Besides the chiral angle, thecircumference of the cylinder can also vary. By consideringthe rolling-up of the sheet as the “placement” of the atom at (0,

18 A.G. Mamalis et al. / Precision Engineering 28 (2004) 16–30

Fig. 2. Ball-stick model of a nanotube; the balls represent the carbon atomsand the sticks their bonds[11].

Fig. 3. Differences between graphite (top) and diamond lattice (bottom)[11].

Fig. 4. (a) The carbon lattice and the ways it can be rolled-up to form azigzag, an armchair or a chiral tube, depicted with its chiral angle; the atomat position (11, 7) is projected on (0, 0) like all the other atoms on the dottedline to form a tube. (b) STM image of the (11, 7) chiral tube[14].

0) on the atom at (n, m), tubes can be classified using this pairof integers (seeFig. 4); the roll-up vector (n, m) specifies theoriented width, recording the number of steps along theaandb directions. The thinnest superconductive carbon nanotubefabricated was just 0.24 nm in diameter[13], however mostof the tubes produced in normal ways have larger diameters.

2.2. Fabrication of nanotubes

Split by heat, carbon atoms recombine in many ways insoot, some in amorphous blobs, but others in football-shapedspheres (buckyballs) or in long cylindrical capsules (nan-otubes). Notable progress has been made in the synthesis ofthese carbon nanotubes. In general, there are three ways tomake soot that contains a reasonably high yield of nanotubes;electric arc discharge (EAD), laser ablation (LA) and chem-ical vapour deposition (CVD).

2.2.1. Electric arc discharge methodThe first identified nanotubes were fabricated by a di-

rect current EAD between carbon electrodes within a no-ble gas, like argon or helium[8]. Macroscopic quantities ofmulti-walled nanotubes were produced with an improved arcdischarge method[15]. In this process, the carbon electrodesare placed a few millimetres apart and the current of approx-imately 100 A vaporises the carbon into a hot plasma, someof which recondenses in the form of nanotubes. Note that,

A.G. Mamalis et al. / Precision Engineering 28 (2004) 16–30 19

the nanotubes form only where the current flows, i.e. on thelarger negative electrode. The voltage of about 20 V, main-tains a high temperature of 2000–3000◦C [16].

The typical yield of nanotubes is up to 30% by weight. Thetubes have diameters between 2 and 20 nm and tend to beshort (50�m or less), deposited in random sizes and direc-tions; the typical rate of deposit is about 1 mm/min. It is to benoted that, an addition of a small amount of transition metalpowder, like cobalt, nickel or iron to the rods, favours thegrowth of single-walled nanotubes. The metal clearly servesas a catalyst, preventing the growing tubular structures fromwrapping around and closing into a smaller fullerene cage.The presence of a catalyst also allows the reduction of tem-perature. Without such cooling, the arc is too hot, and thenanotubes coalesce and merge rapidly into disorder; to min-imise this effect a water-cooled cathode is used[17].

2.2.2. Laser ablationSWNT can be efficiently produced by LA of a graphite

rod, according to Thess et al.[18]. These highly uniformtubes have a greater tendency to form aligned bundles thanthose prepared using arc-evaporation. In 1996, that samegroup produced what was considered to be among the bestsingle-walled nanotube material generated; this consists ofnanotubes, mostly of the armchair type (over 70% of thevolume of material) bundled together into crystalline ropesof metallic character[18]. These ordered nanotubes wereprepared by laser vaporisation of a carbon target in a fur-nace at 1100–1200◦C, at a much lower temperature thanwas previously thought necessary for fabricating nanotubes.A cobalt–nickel catalyst assists the growth of the nanotubes,presumably because it prevents the ends from being “capped”during synthesis. By using two laser pulses, growth condi-tions can be maintained over a larger volume and for a longertime. This scheme provides more uniform vaporisation andbetter control of the growth conditions. The diameter rangeof the tubes can be controlled by varying the reaction temper-ature. A flow of argon or nitrogen gas sweeps the nanotubesfrom the furnace to a water-cooled copper collector placedjust outside of the furnace. A disadvantage of this method isthat it requires expensive lasers.

2.2.3. Chemical vapour depositionDespite the described progress of synthetic techniques for

nanotubes, there still remained two major problems in theirsynthesis, i.e. large scale and ordered synthesis. But, in 1996a CVD method emerged as a new candidate for nanotubesynthesis; in that year a group from the Chinese Academyof Science used CVD to produce a 50�m thick film of nan-otubes that were highly aligned perpendicular to the surface[19]. This method is capable of controlling growth direc-tion on a substrate and synthesising a large quantity of nan-otubes. In this process a mixture of hydrocarbon gas, acety-lene, methane or ethylene and nitrogen is introduced into thereaction chamber. During the reaction, nanotubes are formedon the substrate by the decomposition of the hydrocarbon

at temperatures 700–900◦C and atmospheric pressure[20].The process has two main advantages: the nanotubes are ob-tained at much lower temperature, although this is at the costof lower quality, and the catalyst can be deposited on a sub-strate, which allows for the formation of novel structures.

2.2.3.1. The substrate.The preparation of the substrate andthe use of the catalyst deserve special attention, because theydetermine the structure of the tubes. The substrate is usuallysilicon, but also, glass and alumina are used. The catalystsare metal nanoparticles, like Fe, Co and Ni, which can bedeposited on silicon substrates either from solution, electronbeam evaporation or by physical sputtering. The nanotubediameter depends on the catalyst article size, therefore, thecatalyst deposition technique, in particular the ability to con-trol the particle size, is critical to develop nanodevices.

Porous silicon is an ideal substrate for growing self-orientednanotubes on large surfaces. It is proved that nanotubesgrow at a higher ratio (length per minute), and they are bet-ter aligned than on plain silicon (seeFig. 5 and[21]). Thenanotubes grow parallel to each other and perpendicular tothe substrate surface, because of catalyst–surface interactionand the van der Waals forces developed between the tubes.The growth direction of the tubes is determined by the poresbecause of their template effect. The pores in the substratehave to be aligned and be small enough to grow nanotubesin an ordered way. If, for example, silicon powder is pressedinto a porous pallet, the nanotubes will grow randomlyon it.

Two new methods have been developed to prepare sub-strates with aligned pores in which catalyst particles are em-bedded:

• Thesol–gel methoduses a dried silicon gel, which has un-dergone several chemical processes, to grow highly alignednanotubes (seeFig. 6). The substrate can be re-used afterdepositing new catalyst particles on the surface. The lengthof the nanotube arrays increases with the growth time, andreaches about 2 mm after 48-h growth[20].

• Electrochemical etchingof phosphor-doped n+-type sili-con is the other method to create porous silicon (seeFig.5). A thin nanoporous layer, with pore diameter 3 nm, iscreated on top of a macroporous layer. After electron beamevaporation of Fe to create the catalyst-nanospots, this an-odised silicon is used to synthesise very well aligned bun-dles of nanotubes growing perpendicular to the surface.After removing the grown tubes, the catalyst remains onthe substrate, which makes re-use of the substrate very easy[21].

Hexagonal close-packed nanochannel alumina tem-plates with a Co-Ni catalyst is another substrate, wherewell-graphitised free-standing nanotube arrays can be grown[22]. Like silicon, the alumina is anodised and, subsequently,cobalt is deposited on the alumina, which is used as a catalyst.An important advantage of the template method is that the

20 A.G. Mamalis et al. / Precision Engineering 28 (2004) 16–30

Fig. 5. Carbon nanotube arrays grown on patterned silicon, which form neatly shaped blocks; (a) n+-type porous silicon substrate, (b) p-type plain siliconsubstrate[21].

Fig. 6. SEM images of aligned tube roots obtained by removing the silicasubstrate: (A) low magnification, (B) high magnification[20].

nanotubes prepared in this way can be diameter-controlledand well defined.

2.2.3.2. Gas phase metal catalyst.In the methods, de-scribed above, the metal catalysts are deposited or embeddedon the substrate before the deposition of the carbon is started.A new method is to use a gas phase for introducing the cata-lyst, in which both the catalyst and the hydrocarbon gas arefed into a furnace followed by catalytic reaction in the gasphase. The latter method is suitable for large-scale synthesis,because the nanotubes are free from catalytic supports andthe reaction can be operated continuously.

A high pressure carbon monoxide (CO) reaction method,in which CO gas reacts with iron pentacarbonyl (Fe(CO)5)to form SWNTs, has been developed[18]. SWNTs have alsobeen synthesised from a mixture of benzene and ferrocene(Fe(C5H5)2) in a hydrogen gas flow[13]. In both methods,catalyst nanoparticles are formed through thermal decompo-sition of organo metallic compounds, such as iron pentacar-bonyl and ferrocene.

The reverse micelle method is promising, which containscatalyst nanoparticles (Mo and Co) with a relatively homo-geneous size distribution in a solution. The presence of sur-factant makes the nanoparticle soluble in an organic solvent,such as toluene and benzene. The colloidal solution can besprayed into a furnace, at a temperature of 1200◦C; it va-porises simultaneously with the injection and a reaction oc-curs to form a carbon product. The toluene vapour and metalnanoparticles act as carbon source and catalyst, respectively.The carbon product is removed from the hot zone of the fur-nace by a gas stream (hydrogen) and collected at the bottomof the chamber[23].

2.2.3.3. Plasma enhanced hot filament CVD.The temper-atures, which are required for all these methods, are too high

A.G. Mamalis et al. / Precision Engineering 28 (2004) 16–30 21

and not suitable for the fabrication of electronic devices,since most electrical connections are made of aluminiumwith melting point below 700◦C. When using cold cathodeflat panel displays, see below, the carbon nanotube emittershave to be grown perpendicular to the surface of the dis-play glass, while the strain point of the best display glass isonly 660◦C. Therefore, low-temperature synthesis methodssuch as plasma-enhanced hot filament CVD, are suggested[24]. The plasma is created by an electric field and heat pro-duced by a tungsten filament coil. The growth rate of the nan-otubes is a few times higher than in the normal CVD process(about 120�m/h). In 1998 Ren et al. succeeded in growingwell-aligned carbon nanotubes on nickel-coated glass belowthe temperature of 660◦C using this method and a gas mix-ture of ammonia and acetylene[24].

Very recently, a new method, which required even lowertemperatures (below 520◦C) has been reported; in thisprocess a microwave plasma is used[25]. Due to thenon-equilibrium state between electrons and other heavyparticles in the plasma, the temperature for synthesisingcarbon nanostructures could be greatly decreased. The lowtemperatures are promising, but the method needs some im-provements, to create better controlled nanotube structures.

2.3. Treatment of nanotubes

2.3.1. PurificationThe three different methods of the production of nanotubes

suffer some serious limitations; all produce mixtures of nan-otubes and nanoparticles sticking together in larger lumps.The tubes have a wide range of lengths, many defects and avariety of twists to them. Therefore, the main concern is howto separate them from the worthless soot and how to purifythe tubes. Various post-growth treatments have been devel-oped to purify the tubes and also to eliminate the defects inthe tubes. The material can be treated in an ultrasonic bath tofree many tubes from the particles that are originally stucktogether[15].

The larger contaminants can be easily removed due to theirrelatively high weight, for example by dispersing the pow-der in a solvent and subsequent centrifugation. The smallerparticles are more difficult to eliminate. One possibility forMWNTs is to perform an oxidative treatment, either by heat-ing the powder in air at 650◦C or by a liquid phase treat-ment in acidic environment. For SWNTs, standard methodsto eliminate catalyst particles and amorphous carbon involvere-fluxing the raw material in acid followed by centrifugationor cross-flow filtration.

Another possibility for purification is to employ physi-cal methods that do not damage the tubes, but separate theobjects as a function of their size. For MWNTs, a purifica-tion method that uses the properties of colloidal suspensionshas been developed. Smaller objects remain dispersed whilelarger particles form aggregates that are deposited as a sedi-ment after a few hours. A related method, the size-exclusionchromatography, was successfully used for the purification

and size selection for MWNTs. Purification procedures forSWNTs without any acidic treatment have also been reportedand involve microfiltration or size-exclusion chromatography[12].

A method to eliminate the defects in carbon nanotubes isby annealing at high temperatures, up to 3000 K; during thisprocess, impurities and defects in the tubes are eliminated.

To distinguish the different chiral types of nanotubes iseven more complex. A solution for separating nanotubes fortheir electrical properties was found recently. By applying acurrent between metal electrodes, conducting nanotubes canbe burned away until only the semiconducting ones are left[26].

2.3.2. StabilityAlthough nanotubes exhibit ideal characteristics when they

are in an ultrahigh vacuum environment, samples in moreordinary conditions, where they are exposed to air or wa-ter vapour, show properties that are different. Nanotubes are,therefore, very sensitive to contaminants, like oxygen, attach-ing to them. They severely affect the electrical properties,which is a big problem for devices made from nanotubes. Thesoot of fabricated nanotubes is often dispersed in ethanol,in which it can be preserved without damage to the tubes.Note, that the tubes themselves are stable, maintaining theirshapes, regardless of the contamination mentioned above, upto 2800◦C in vacuum, and till 750◦C in air. For comparison,metal wires in microchips melt between 600 and 1000◦C[17].

2.4. Properties of nanotubes

2.4.1. Mechanical propertiesIn order to measure the mechanical properties of the nan-

otubes and because of their small dimensions, novel methodsare used. The Young’s modulus of elasticity was estimatedafter measuring the thermal vibrations of nanotubes; a veryhigh average value of 1.8 TPa was found[27]. Wong et al.[28] used a scanning force microscope to bend nanotubesthat were mechanically fixed at one end. By measuring vibra-tions of nanotubes in an electrical field, Poncharal et al.[29]found a value below 1 TPa. This is true both for multi-walledand single-wall nanotubes because the modulus is mainly de-termined by the carbon–carbon bonds within the individuallayers.

Salvetat et al.[30] found that multi-walled nanotubesgrown by the arc discharge method had a modulus of about1 TPa, whereas those grown by the catalytic decompositionof hydrocarbons had a modulus that was smaller by one totwo orders of magnitude. The nanotubes were placed across“nanopores” and an atomic force microscope was used tobend them in the middle. These results demonstrate thatonly highly ordered and well-graphitised nanotubes havea stiffness comparable to graphite, whereas those grownby catalytic decomposition are weaker because of defects(seeFig. 7aand[30]). Further insights into the mechanical

22 A.G. Mamalis et al. / Precision Engineering 28 (2004) 16–30

Fig. 7. (a) An AFM microscope of a multi-walled nanotube across a pore to measure its Young’s modulus by bending it with an AFM tip[30] (b) and (c) SEMimages of a multi-walled nanotube held between two AFM tips to measure its tensile strength[31].

properties of multi-walled nanotubes are reported in[31].The ends of a multi-walled nanotube were attached to a pairof AFM tips and stretched until it broke (seeFig. 7b andc). The tips of the tubes were attached on the AFM tips byelectron beam deposition of carbonaceous material. A ten-sile strength of the nanotubes, ranging from 11 to 63 GPawas obtained; for comparison, high strength steel alloysbreak at about 2 GPa. The Young’s modulus ranged from270 to 950 GPa[31]. In this process, the AFM tips onlymake contact with the outside of the nanotube, so the out-ermost layer carries most of the load. The outermost tuberuptures at the tensile limit and slides over the inner tubes(“sword-in-sheath failure”). There are relatively weak vander Waals interactions between the layers whilst the shearstrength between the layers is small. This property is veryinteresting for applications like nanobearings.

When nanotubes are compressed, they show remarkableproperties. They bend over to surprisingly large angles, beforethey start to ripple and buckle, and then, finally kinks aredeveloped. Note that all these deformations of the carbonnanotubes are elastic, all disappearing completely when theload is removed[31].

The density of bundled nanotubes is 1.33–1.40 g/cm3. Thisis very low, as compared with aluminium, possessing a den-sity of 2.7 g/cm3 [17].

2.4.2. Electrical propertiesGraphite carbon nanotubes emerge as interesting conduct-

ing materials. Because electron waves can reinforce or cancelone another, an electron spreading around the circumference

of a nanotube can completely cancel itself out; therefore,only electrons with the right wavelength remain. From all thepossible electron wavelengths, or quantum states, availablein a flat graphite sheet, only a tiny subset is allowed whenthat sheet is rolled into a nanotube. That subset depends onthe circumference of the nanotube, as well as on the chirality(twist) of the nanotube.

In a graphite sheet, one particular electron state, designatedas the Fermi point, provides graphite with almost its wholeconductivity, none of the electrons in other states are free tomove about. All armchair tubes and one out of three zigzagand chiral tubes combine the right diameter and degree oftwist to include this special Fermi point in their subset of al-lowed states. These nanotubes are truly conducting metallicnanowires. The remaining two-thirds of nanotubes are semi-conductors. For example, ifn − m is the roll-up factor (seeFig. 4) is three times an integer, the carbon nanotube has anextremely small gap, and at room temperature, it shows ametallic behaviour. Forn = m, the tubes are metallic whilstfor other values ofn−m, the tubes behave as semiconductorswith a band gap[14], indicating that, like silicon, they do notpass current easily without an additional amount of energy.

Carbon nanotubes do not possess the same band gap, be-cause for every circumference there is a unique set of allowedvalences and conducting states. The smallest-diameter nan-otubes have very few states that are spaced very far apart inenergy. Band gaps of 0.4–1 eV can be expected for SWNTs,corresponding to diameters between 0.6 and 1.6 nm. Asnanotube diameters increase, more and more states are al-lowed and the spacing between them reduces. In this way,

A.G. Mamalis et al. / Precision Engineering 28 (2004) 16–30 23

different-size nanotubes can have band gaps as low as zero(like metal), as high as the band gap of silicon, and almostanywhere in between[12]. No other known material can beso readily tuned. Note, however that, the growth of nanotubescurrently provides a wide range of different geometries, andresearchers are seeking improvements so that the specifictypes of nanotubes can be guaranteed. Because of their geom-etry, nanotubes are sensitive to defects and impurities. Thissensitivity gives rise to unusual properties. For two metalliccarbon nanotubes, depending on the helicity, a junction canstop the current because of the mismatch of wavefunctionsymmetry [32]. Note, also, that in defect-free nanotubeselectrons travel without any of the scatter that gives metalwires their resistance. A lack of scattering may also explainwhy nanotubes appear to preserve the “spin” state (quantumproperty) of electrons as they travel along. This unusual be-haviour is now under consideration for the possible construc-tion of “spintronic” devices that switch on or off in responseto electrons’ spin, rather than merely to their charge.

Thick multi-walled nanotubes may display an even morecomplex behaviour, as each layer of the tube has a slightly dif-ferent geometry. By tailoring their composition, multi-walledtubes that are self-insulating or able to carry multiple signalsat once, like nanoscopic coaxial cables, may be fabricated.However, the understanding and control of nanotube growthstill falls far short of these goals.

The fact that metallic nanotubes are really good conduc-tors is underlined by comparing them with copper. A bun-dle of nanotubes, could conduct about 1 billion A/cm2 whilst,copper wires instead saturate at about 1 million A/cm2 [17].The electronic properties of single-walled carbon nanotubesare shown to be extremely sensitive to the chemical environ-ment[33]. Exposure to air or oxygen dramatically affects thenanotubes resistance and other electronic properties. Theseparameters can be reversibly “tuned” by surprisingly smallconcentrations of adsorbed gases, and an apparently semi-conducting nanotube can be converted into an apparent metalthrough such exposure. Hence, the electronic properties ofa given nanotube are not specified only by the diameter andchirality of the nanotube, but also depend critically on the gasexposure history.

Another very interesting electronic property of nanotubes,is their field emission; they emit electrons from their tips,when they are placed in an electrical field (seeFig. 8 and

Fig. 8. A schematic diagram of field emission of a nanotube[34].

[35]). Because they are so sharp, the nanotubes emit elec-trons at lower voltages than electrodes made from most othermaterials, and their strong carbon bonds allow nanotubes tooperate for longer periods without damage. The nanotubecan activate phosphors at 1–3 V if the electrodes are spaced1�m apart. Molybdenum tips, by comparison, require fieldsof 50–100 V/�m and have very limited lifetimes[17].

The thermal properties of nanotubes are also very impres-sive. Nanotubes are stable in vacuum up to 2800◦C, and inair up to 750◦C. The heat transmission is predicted to be ashigh as 6000 W/mK at room temperature. This can be com-pared with nearly pure diamond, which is a very good heatconductor and transmits 3320 W/mK[17].

3. Applications of nanotubes

3.1. Electronics and microchip manufacturing

The fact that nanotubes are very good conductors, and thatthey also appear as semiconductors or even insulators, makesthem very useful for minuscule electronic devices like logiccircuits composed of several transistors. The making of tinycircuits might be promising for the semiconductor industry.This industry is focused to make computer chips smaller ev-ery year, but also has to make them as cheaply as possible.

The conventional method ofphotolithographyuses maskswhich have to be patterned once, which is time consumingand expensive, but they can be used many times. To createa normal chip, several masks are required. In comparisonto the making of the mask, the manufacturing of the chipsthemselves is very rapid. Therefore, only large productionquantities are profitable. The problem is that, this techniqueis reaching the limit of the smallest possible width of thelines (about 150 nm). Smaller wavelengths have to be used toprevent the blurring which occurs when the features get toosmall.

One leading contender iselectron beam lithography. Anelectron beam does not diffract at atomic scales, so it doesnot cause blurring of the edges of features. The techniquehas been used to write lines with widths of only 50 nm ina layer of photoresist on a silicon substrate. The electronbeam instruments currently available are very expensive. Theprocess is also very slow, because the beam must write everyline on the chip, instead of making use of a patterned mask.This makes them impractical for large-scale manufacturing.

Another technique employed islithography using X-rayswith wavelengths between 0.1 and 10 nm or extreme ultravi-olet light with wavelengths between 10 and 70 nm. Becausethese forms of radiation have much shorter wavelengths thanthe ultraviolet light currently used in photo lithography, theyminimise the blurring caused by diffraction. At a wavelengthof 13 nm, extreme ultraviolet lithography (EUV) will even-tually have the ability to print a transistor-element of only40 atoms in width, but some difficulties arise. The mask andlenses, transparent at longer wavelengths, absorb this radi-

24 A.G. Mamalis et al. / Precision Engineering 28 (2004) 16–30

Fig. 9. A single-molecule transistor that operates at room temperature, con-sisting of an individual semiconducting nanotube on two metal nanoelec-trodes with the substrate as a gate electrode[37].

ation, therefore, EUV uses mirrors for both the mask andthe lenses. The entire circuit-printing process has to be per-formed in a vacuum because air itself absorbs radiation atthis wavelength. The masks are very sensitive to impuritiesand will distort the image if it contains more than a handfulof defects measuring even 50 nm. Also, the technology of us-ing extreme ultraviolet light or X-rays is still far away fromcommercial use in the microelectronics industry.

The search for completely other concepts for microchipshas already begun, and nanotubes seem to offer one solu-tion. Several research groups have successfully built workingelectronic devices using carbon nanotubes.

Essential devices like field-effect transistors (FET) havebeen developed. They use a single semiconducting nanotubebetween two metal electrodes as the channel through whichelectrons flow. The current in this channel can be switched onor off by applying voltages to a nearby third “gate” electrode.It was found that this electrode can change the conductivityof the nanotube channel by a factor of 1 million or more,compared to silicon FETs. Because of its tiny size the carbonnanotube FET (CNT-FET) should switch reliably using muchless power than a silicon-based device. It is predicted that,such a nanoscale device could run at clock speeds of oneterahertz or more.

The fabrication of a CNT-FET starts with placing a tubeon the insulating silicon dioxide (SiO2) layer by spincoatingof a suspension, on a substrate with pre-arranged conductingpads. These pads are connected to the tube by metal leads,lithographically deposited across the tube. This technique isdesignated as thefour-probe technique[36]. The silicon layerbelow the silicon dioxide is used as the back gate. The firstnanotube-based devices operated at very low temperatures,but in 1998 the first transistor was reported that worked atroom temperature, with electrical characteristics remarkablysimilar to silicon devices (seeFig. 9and[34]).

Fig. 10. Model of a nanotube kink junction, with 5- and 7-rings; the tubesegment at the top can be metallic whereas the bottom segment is semicon-ducting[32].

To avoid this labour intensive process, experiments havebeen performed to grow single-walled carbon nanotubes bychemical vapour deposition of methane at controlled loca-tions on an insulating SiO2 substrate using patterned catalyticislands. Some of the tubes bridge two metallic islands, creat-ing simple electrical circuits[38]. The control of the chiralityof the nanotubes constitutes a shortcoming of the process.

Introducing a pentagon and a heptagon into the hexagonalcarbon lattice of a nanotube creates a kink in the tube. Inthis way, two tube segments with different atomic and elec-tronic structures can be fused together to create intramolec-ular metal–metal, metal–semiconductor, or semiconductor–semiconductor junctions that are only a few atoms in cross-section (seeFig. 10). Note that a metal–semiconductorkink junction behaves like a molecular diode; electricityis permitted to flow only in one direction (seeFig. 11 and[32]).

Metal–metal junctions lead to an improved version of a sin-gle electron transistor (SET), which is another promising de-

Fig. 11. A perspective view of a carbon nanotube kink junction between twoelectrodes on an insulating substrate SiO2 [32].

A.G. Mamalis et al. / Precision Engineering 28 (2004) 16–30 25

vice. It has been proposed even as the future alternative to con-ventional silicon electronic components, but their practicaluse has been limited by the fact that they only operate at verylow temperatures. However, recently the first single electrontransistor operating at room temperature was reported[39].The device is similar to the FET described above, but with ashort nanotube section of about 20 nm, that was manipulatedby an atomic force microscope to create a Coulomb island.SETs consist of such a conducting island connected by tun-nel barriers to two metallic leads. Strong bends (“buckles”)within the metallic carbon nanotubes are constructed usingan AFM tip, and act as nanometre sized tunnel barriers forelectron transport. The created resistance is in the order of0.5 M�. For temperatures and bias voltages, that are low, rel-ative to a characteristic energy required to add an electron tothe island, electric transport through the device is blocked.Conduction or ‘Coulomb charging’ is observed at room tem-perature, with an additional energy of 120 meV, by tuning avoltage on a close-by gate.

The next step in assessing the suitability of these devicesfor computer electronics involves the integration of individ-ual CNT-FETs to formlogic gates. To build such logic cir-cuits, nanotube devices, that use electrons (n-type FET) andholes (p-type) as the carriers of electricity, are needed. Theproblem was that all CNT-FETs showed p-type characteris-tics, that means that, they were ON for negative gate bias[37].Therefore, the first n-type CNT-FBT had a great impact; itwas made by direct doping of the tube with an electropositiveelement such as potassium. Potassium atoms (K atoms evap-orated from an alkaline metal dispenser) are adsorbed ontothe surface of the nanotube, donating electrons to convert thenanotube from p- to n-type. By covering half of the nanotubewith PMMA, which is also used as the photoresist in pho-tolithography, p–n junctions were produced[40]. Note thatnot only by doping, but also by annealing in a vacuum, theelectrical character can be changed from p-type into n-type.Using vacuum annealing or doping it is possible to maken-type CNT-FETs, p- and n-CNT-FETs on the same sub-strate can be fabricated. These complementary CNT-FETsare assembled to form the first intermolecular logic gates.At first, a “NOT” gate or voltage inverter was demonstrated(seeFig. 12and[41]). To use a logical gate as part of a morecomplicated computing system, a gain (output/input ratio) of1 is required. The intramolecular inverter had a gain of 1.6and, therefore, it can be used to drive another gate or a morecomplicated logic circuit.

Recently, otherlogic circuitswere constructed from indi-vidual carbon nanotubes; a logic NOR, a static random-accessmemory cell (RAM), an ac ring oscillator, and another in-verter were reported[42]. In these circuits, one, two, or threetransistors were used. They all operated at room temperature,had a high gain (>10), and a large on–off ratio (>10). To con-struct nanotube circuits, electron beam lithography was usedto pattern local aluminium gate contacts and exposed themto air to form very thin insulating layers on the aluminiumleads. The silicon dioxide insulator in former FETs was

Fig. 12. An AFM showing the design of an intramolecular logic gate con-sisting of a single nanotube bundle, positioned over the gold electrodes toproduce two p-type CNT-FETs in series. The device is covered by PMMA,a window is opened by electron beam lithography to expose part of the nan-otube, and subsequently, potassium is evaporated through this window toproduce an n-CNT-FET, while the other CNT-FET remains p-type[41].

relatively thick, preventing sufficient capacitive couplingbetween the gate contact and the nanotube. Because of the re-duced insulator thickness, the new nanotube transistors wereable to operate with a highly increased gain ratio. Formernanotube devices placed on a chip switched simultaneously,because the controlling gate was the entire supporting siliconchip. But now each transistor can be controlled by its ownlocal “gate” contact. These developments contributed to abetter prospect for the use of nanotube logic circuits in chips.

Despite recent progress, a number of technical challengesmust be overcome to make a robust, commercially viablecomputer integrated on the molecular scale. Circuits mustbe produced that are entirely on the molecular scale, notjust incorporating molecular-scale components. The heatproduced by all these tiny densely spaced devices may alsobecome a problem because of the limited heat dissipation,which occurs when devices are reduced to the nanoscale.Furthermore, the molecular devices are very sensitive tothe noise caused by electrical, thermal and chemical fluc-tuations. The addition of contaminants (oxygen) changesthe electrical properties of the nanotube, so they must beprotected against these influences. The circuits, which arebuilt using nanotubes, are all built one at a time with greateffort. The attachment of a nanotube requires combinationallithography for the electrodes and high-resolution tools suchas atomic force microscopes to locate and even positionthe nanotubes. It is not only the handling of the tube whilecreating the circuit a problem, but also the synthesis of theright type of tubes remains a problem. Ways to grow nan-otubes in specific locations, orientations, shapes and sizesand to merge nanotubes with silicon nanowires, thus makingconnections to circuits fabricated by conventional means areunder consideration. Therefore, the possibility of using car-bon nanotubes as both the transistors and the interconnectionwires in microchip circuits is becoming possible. It may bestated that, this is a long way from the massively parallel,

26 A.G. Mamalis et al. / Precision Engineering 28 (2004) 16–30

complex and automated production of microchips from sil-icon on which the computer industry is built. Until now themajor challenge was to construct logic circuits of individualmolecular-scale structures. This has been achieved, provingthat this technology is still developing, and it is only a matterof time to overcome the above-described problems.

Field emission(seeFig. 8), has long been seen as a po-tential technology for replacing bulky, inefficient televisionsand computer monitors with equally bright but thinner andmore power-efficient flat-panel displays. It was shown thatnanotubes emit electrons very efficiently when immersed inan electric field[35]. Efficient field emission was also ob-tained from carefully aligned nanotubes, whereas randomlyoriented nanotubes have been also used with similar results[33]. Clusters of upright nanotubes in neat little grids havebeen grown. At optimum density, such clusters can emit morethan 1 A/cm2, which is more than sufficient to light up thephosphors on a screen and is even powerful enough to drivemicrowave relays and high frequency switches in cellularbase stations.

Ise Electronics in Japan has used nanotube composites tomake prototype vacuum-tube lamps in six colours that aretwice as bright as conventional lightbulbs, longer-lived and atleast 10 times more energy-efficient. Engineers at Samsungin Korea created flat panel displays by spreading nanotubeson a thin film over control electronics with phosphor-coatedglass on top. Now, the company has built some improveddevices, which are as bright as a cathode-ray tube but willconsume one-tenth as much power. Further improvement ofthis technology may lead to easy-to-make and low-cost flatpanel displays.

Considering the small sizes of carbon nanotubes, high per-formance may be expected for them due to sunlight trapping.Large surface area and high thermal conductivity enable arapid heat transfer from carbon nanotubes to the surroundingenvironment. Furthermore, they are physically and chemi-cally stable. Scientists have constructed a solar cell by syn-thesising aligned carbon nanotubes on an Au film[43].

Although carbon nanotubes are known to be metallic orsemiconducting, depending on shell helicity and diameter,boron-nitride nanotubes are expected to always be insulating.This property results in the construction of a nano-insulatingshield for any conducting material, like a gold wire or a carbonnanotube, encapsulated within. Boron-nitride nanotubes aremore stable to oxidation than carbon, and as tough as carbonnanotubes[44].

3.2. Automotive and aerospace industry

Carbon fibre is already used to strengthen a wide rangeof materials, and the special properties of carbon nanotubesmean that they could be the ultimate high-strength fibre,which has a tensile strength 20 times, and a strength-to-weight ratio of 100 times that of steel. Nanotubes havealready been used for reinforcement of nanostructural com-posite materials, polymers and concrete. Being stronger

Fig. 13. Artificial impression of a space elevator using a carbon nanotuberope[45].

than diamond, it is the strongest material known. Becauseof the low density these materials are very useful for cars,aeroplanes and space vehicles[7].

Whereas a crack in a normal plate will grow because of theconcentrated tensions at the tip of the crack, a broken nan-otube produces almost no effect on the others. The tiny crackis blocked, and the chain reaction of fracture is terminated.There is good reason to expect a macroscopic 1 in. thick rope,where 1014 parallel buckywires are holding together, to bealmost as strong as theory predicts, about 40 GPa[16]. Thisstrength combined with the low density may be crucial forthe developing of a space elevator, which uses a rope to reachthe stationary orbits of satellites (seeFig. 13and[45,46]). Itcan be calculated that the cable made of nanotubes must beas strong as 63 GPa. This is almost achievable whilst no othermaterial known nowadays comes even close to the requiredstrength.

The fact that deformations of nanotubes are extremelyelastic, especially in the case of single-walled nanotubes, isanother very useful property. When used in a car, after a crash,all the buckles and kinks unfold and the material will showno damage at all. Other, less futuristic applications mightinclude: lightweight bullet-proof vests, earthquake-resistantbuildings and elements for bridges. Therefore, the highstrength of carbon nanotubes makes them promising candi-dates in reinforcement applications, but there are many bigproblems that must be overcome. First, the properties of theindividual tubes must be optimised. Second, the tubes mustbe efficiently bonded to the material they are reinforcing, sothat they actually carry the loads and third, the load mustbe distributed within the nanotube itself to ensure that theoutermost layer does not shear off.

A.G. Mamalis et al. / Precision Engineering 28 (2004) 16–30 27

Incorporation of conducting carbon nanotubes in con-struction materials, such as concrete or structural plastics,provides opportunities for real time monitoring of materialintegrity and quality. Some General Motors cars alreadyinclude plastic parts to which nanotubes were added. Suchplastic can be electrified during painting allowing for a bettersurface quality.

One of the major research objectives is to develop materialsor structures with excessive storage capacity per unit volumeand weight for gases such as H2 or CH4 to power fuel cells forultralow-emission vehicles or for electric power generation.The biggest problem with these promising fuel cells is howto store hydrogen without using extreme pressure or largetanks. Carbon nanotubes could be the solution, because theycan adsorb hydrogen molecules within their tubes.

The hydrogen can be released gradually to supply efficientand inexpensive fuel cells. But so far the best reports indicate6.5% hydrogen uptake, which is not dense enough to make afuel cell economical[47].

3.3. Biomedicine and nanotechnology

Open-ended nanostraws could penetrate into a cellularstructure for chemical probing or could be used as ultrasmallpipettes to inject molecules into living cells. Using computersimulations, It has been shown[48] that water moleculeswill quickly enter and flow through a carbon nanotube of8 nm in diameter. A separate set of simulations shows thatcertain organic molecules also will flow through such nan-otubes. The nanotubes conduct water at a rate similar to thatof certain channels in the kidneys. These unusual transportproperties of carbon nanotubes might be used in biomedicalapplications, such ashighly targeted drug delivery.

The small size and sensitivity of nanotubes make theassembly of extremely powerfulsensorspossible. Semi-conducting nanotubes change their electrical resistancedramatically when exposed to alkalis, halogens and othergases at room temperature. For example, semiconductingcarbon nanotubes have been used to detect gas molecules[49], whilst semiconductor nanowires have been used asdetectors for a wide range of biological compounds. Also,nanowired field-effect transistors have been converted intosensors by modifying their surfaces with molecular recep-tors. This technology has the potential of detecting singlemolecules using only a voltmeter from a hardware store. Butthe problem remains, that nanotubes are very sensitive tomany chemicals, including oxygen and water, and they maynot be able to distinguish one chemical or gas from another.

Scanning probe tips (SPT) are very important for the devel-opments in nanotechnology, because they reveal the surfacetopography of various materials. Therefore, the tips have to beextremely sharp and must not show wear after several scans.Individual carbon nanotubes have been attached to siliconcantilevers in conventional AFM instruments (seeFig. 14and[49]); this was an important advancement in proximal probemicroscopy because such tips can not only conduct (needed

Fig. 14. (A) A SEM image of a bundle of nanotubes attached to an AFM tipat the end of a protruding single nanotube, (B) detail to (A), (C) detail to (B)[49].

for STM tips), but they must also be resistant to mechanicaldamage, which is providing, at the same time, superior imag-ing capabilities due to their relative sharpness. For example,a nanotube-tipped atomic force microscope (seeFig. 15), cantrace a strand of DNA and identify chemical markers, reveal-

Fig. 15. A SEM image of two multi-walled carbon nanotubes (MWNT)attached to a blunted silicon AFM tip[50].

28 A.G. Mamalis et al. / Precision Engineering 28 (2004) 16–30

ing which of several possible variants of a gene is present inthe strand. This is the only method yet invented for imagingthe chemistry of a surface, but it is not yet used widely; so far,it has been used only on relatively short pieces of DNA[50].

To use a nanotube as a robust probe, it can be bonded onthe side of the tip on a conventional silicon cantilever us-ing a soft acrylic adhesive[49]. Electron beam deposition ofcarbonatious material is also a method to attach nanotubesto probe tips. Scientists used this for example in the previ-ously described method of testing the strength of nanotubes[31]. However, the present method of mechanically attach-ing nanotube bundles for tip fabrication is difficult and timeconsuming. A technique for growing individual carbon nan-otube probe tips directly, with control over the orientation,by CVD from the end of silicon tips, based on the anodisedsilicon method has been developed[51].

Kim and Lieber [52] fabricated nanotubenanotweez-ers, by depositing free-standing electrically independentelectrodes onto tapered glass micropipettes, which can beroutinely made with end diameters of 100 nm[52]. Carbonnanotubes were attached to the independent Au electrodesusing an approach similar to that used for the fabrication ofsingle-nanotube SPM tips[49]. This approach was used toattach multi-walled or single-walled nanotube bundles of20–50 nm in diameter to the two electrodes under the directview of an optical microscope. The arms of the tweezerswere about 4�m long. The size of the nanotweezers waslimited only by the optical microscope resolution used tomonitor the attachment process. If required, it is possibleto make substantially smaller nanotweezers by carrying outnanotube attachment within a SEM or by directly grow-ing the tweezers’ arms by chemical vapour deposition[51](Fig. 16).

The devices work, because the application of an appropri-ate bias voltage across two tweezer arms can effect a me-chanical change in the two arms that results in closing andopening[53]. The operating power depends on the size. Thedescribed nanotweezers operated at 8 V and if the devices be-come smaller, the needed voltages will also decrease. Notethat, although, the tweezers can pick up objects that are largecompared with their width, nanotubes are so sticky that most

Fig. 16. Nanotweezer[52].

Fig. 17. Computer images of nanotube-bearings[54].

objects cannot be released, therefore, they are not yet veryuseful in practice.

As mentioned above, during the breaking mechanism, theintershell interaction between the shells in a multi-walledcarbon nanotube is predominantly that of van der Waalsforces[31]. So, these shells can slide or rotate easily andform ideal linear and rotationalnanohearings(seeFig. 17)reproduced from[54]. Cumings and Zettl were the first whoshowed controlled and reversible telescoping of multi-wallednanotubes[55]. Robust ultralow-friction linear nanobear-ings and constant-force nanosprings were demonstrated.Repeated extension and retraction of telescoping nan-otube segments revealed no wear or fatigue on the atomicscale. Hence, these nanotubes may constitute near perfectwear-free surfaces. The results demonstrate that nanotubesmay be used fornanomechanicalor nano-electromechanialsystems(NEMS). Low-friction/low-wear nanobearingsandnanospringsare essential devices in NEMS technologies.The transit time for complete nanotube core retraction (onorder of 1–10 ns) implies the possibility of exceptionally fastelectromechanical switches.

4. Conclusions

Macroscopic amounts of good quality nanotubes canpresently be fabricated by several groups worldwide. Thethree different methods to synthesise nanotubes, namely arcdischarge, laser ablation, and CVD, are still in use today andimproved to create better-aligned nanotubes and less amor-phous carbon material. For large-scale synthesis, the CVDmethod is most promising. The substrates can be recycledand the process works at lower temperatures (500–1000◦C),which makes the process suitable for direct growth on sev-eral devices. Nanoparticles of a metal catalyst are applied tocreate nanotubes of better quality. Also the structure of thesubstrate is very important. It is shown that aligned poresin silicon and alumina are most suitable to create neatlyaligned nanotubes. The deposition of the metal catalyst onthe substrate deserves special attention, because it enablesmanipulated growth. A problem is the fact that, it is not pos-sible yet to create one specified nanotube with determineddiameter and structure.

A.G. Mamalis et al. / Precision Engineering 28 (2004) 16–30 29

The theoretical understanding of the electronic structureand related properties of nanotubes has reached a very goodlevel. However, despite to that many potential applicationsare mentioned in the literature, only the outstanding fieldemission properties of nanotubes have already been appliedin practical devices. The semiconductor industry will be ruledby lithographic processes in the near future. Logic circuitsmade out of nanotubes have shown very good performance,but they are still “handmade” and the prospects of massivelyparallel, complex and automated production are not yet clear.

The ultimate strength (around 40 GPa) and stiffness (elas-tic modulus around 1 TPa) of nanotubes makes them suitablefor reinforcing materials and ropes, which can bear high de-formations, and extremely high tensile forces. These mate-rials are not on the market yet, but their development is ex-pected in the near future.

Applications of nanotubes for nanotechnology devices areon hand. Scanning probe tips are improved by attaching anextremely sharp nanotube. Nanotweezers, to grab nanopar-ticles, are also constructed, and even nanobearings (linearand rotational) and nanosprings have been fabricated. Thefact that the tubes can conduct molecules like little strawsis very important for future medical applications. Extremelysensitive sensors can be made and the storage of gasses, likemethane and hydrogen, in nanotubes (with high density) isfeasible by developing fuel devices like the fuel cell.

Nanotubes are commonly found in laboratories today, andresearch is stimulated by large amounts of money invested init. Some companies are already specialising in the productionof carbon nanotubes, and give research a boost. It is not knownexactly when and where nanotubes will find their applicationson a massive scale, but it seems highly likely. That’s thetrigger that makes the research on them so exciting.

References

[1] Rocco MC. National Science Foundation, Official who oversees thenanotechnology initiative, vol. 285. Scientific American; 2001. p. 3.

[2] Taniguchi N. On the basic concept of nanotechnology. Proc. ICPBTokyo; 1974.

[3] Feynman R. There’s plenty of room at the bottom: an invitation to entera new field of physics. Engineering & Science Magazine, CaliforniaInstitute of Technology, USA; 1960.

[4] Drexler KE. Engines of creation: the coming era of nanotechnology.Doubleday; 1986 (ISBN: 0-385-19973-2).

[5] Drexler KE. Nanosystems: molecular machinery, manufacturing andcomputation. New York: Wiley; 1992 (ISBN: 0-471-57518-6).

[6] Corbett J, McKeown PA, Peggs GN, Whatmore R. Nanotechnol-ogy: international developments and emerging products. Ann CIRP2000;49(2):523–45.

[7] Mamalis AG. New trends in nanotechnology. In: Proceedings of theSymposium on the 50th Anniversary of Department of ManufacturingEnginering of Budapest University of Technology and Economics, Bu-dapest, Hungary; 2001.

[8] Kroto HW, Heath JR, O’Brien SC, Curl RL, Smalley RE. Nature1985;318:162.

[9] Iijima S. Helical microtubules of graphitic carbon. Nature 1991;354:56–8.

[10] Harris PJF.http://www.rdg.ac.uk/∼scsharip/tubes.htm. Carbon nan-otube site. Cambridge University Press; 1999.

[11] Cohen ML. Nanotubes, nanoscience, and nanotechnology. Mater SciEng: C 2001;15(1/2):1–11.

[12] Bonard JM, Kind H, Stöckli T, Nilsson LO. Field emission from carbonnanotubes: the first five years. Solid-State Electronics 2001;45(6):893–914.

[13] Tang ZK, Zhang L, Wang N, Zhang XX, Wen GH, Li GD, et al. Super-conductivity in 4 Angstrom single walled carbon nanotubes. Science2001;292:2462–5.

[14] Wildöer JWG, Venema LC, Rinzler AG, Smalley RE, Dekker C. Nature1998;391:59.

[15] Ebbesen TW, Ajayan PM. Large scale synthesis of carbon nanotubes.Nature 1992;358:220–2.

[16] Yakobson BI, Smalley RE. Fullerene nanotubes: C1.000.000 and be-yond. Am Sci 1997;85:324–37.

[17] Collins PG, Avouris P. Nanotubes for electronics. Scientific American;December 2000. p. 38–45.

[18] Thess A, Lee R, Nikolaev P, Dai H, Petit P, Robert J, et al. Crystallineropes of metallic carbon nanotubes. Science 1996;273:483–7.

[19] Li WZ, Xie SS, Qian LX, Chang BH, Zou BS, Zhou WY,et al. Large-scale synthesis of aligned carbon nanotubes. Science1996;274:1701–3.

[20] Xie S, Li W, Pan Z, Chang B, Sun L. Carbon nanotube arrays. MaterSci Eng A 2000;286(1):11–5.

[21] Fan S, Liang W, Dang H, Franklin N, Tombler T, Chapline M, et al. Car-bon nanotube arrays on silicon substrates and their possible application.Phys E: Low-Dimensional Syst Nanostructures 2000;8(2):179–83.

[22] Zhang XY, Zhang LD, Li GH, Zhao LX. Template synthesisof well-graphitized carbon nanotube arrays. Mater Sci Eng A2001;308(1/2):9–12.

[23] Ago H, Obshima S, Uchida K, Yumura M. Gas-phase synthesis ofsingle-wall carbon nanotubes from colloidal solution of metal nanopar-ticles. J Phys Chem B 2001;105(43):10453–6.

[24] Ren ZF, Huang ZP, Xu JW, Wang JH, Bush P, Siegal MP, et al. Synthe-sis of large arrays of well-aligned carbon nanotubes on glass. Science1998;282:1105–7.

[25] Wang X, Hu Z, Chen X, Chen Y. Preparation of carbon nanotubes andnano-particles by microwave plasma-enhanced chemical vapor depo-sition. Scripta Mater 2001;44(8/9):1567–70.

[26] Avouris P, et al. Thomas 3. New York: Watson Research Centre, IBMResearch Division; 2001.

[27] Treacy MMJ, Ebbesen TW, Gibson JM. Exceptionally highYoung’s modulus observed for individual carbon nanotubes. Nature1996;381:678–80.

[28] Wong EW, Sheehan PE, Lieber CM. Nanobeam mechanics: elas-ticity, strength, and toughness of nanorods and nanotubes. Science1997;277:1971–4.

[29] Poncharal P, Wang ZL, Ugarte D, De Heer WA. Electrostatic deflec-tions and electromechanical resonances of carbon nanotubes. Science1999;283:1513–6.

[30] Salvetat JP, Andrew G, Briggs D, Bonard TM, Basca RR, Kulik AJ.Elastic and shear moduli of single-walled carbon nanotube ropes. PhysRev Lett 1999;82:944.

[31] Yu MF, Lourie O, Dyer MJ, Moloni K, Kelly TF, Ruoff RS. Strengthand breaking mechanism of multiwalled carbon nanotubes under tensileload. Science 2000;287:637–40.

[32] Yao Z, Postma HWC, Barents L, Dekker C. Carbon nanotube in-tramolecular junctions. Nature 1999;402:273.

[33] Collins PG, Bradley K, Ishigami M, Zettl A. Extreme oxygen sensitivityof electronic properties of carbon anotubes. Science 2000;287:1801–4.

[34] http://ipewww.epfl.ch/grbuttet/manips/nanotubes/NTfieldemission1.htm. Ecole Polytechnique Federale de Lausanne, Institut de Physiquedes Nanostructures (IPN).

[35] Rinzler AG, Hafner JH, Nicolaev P, Lou L, Kim SG, Tomanek D, etal. Unraveling nanotubes: field emission from an atomic wire. Science1995;269:1550–3.

30 A.G. Mamalis et al. / Precision Engineering 28 (2004) 16–30

[36] Ebbesen TW, Lezec HJ, Hiura H, Bennett JW, Ghaemi HF, ThioT. Electrical conductivity of individual carbon nanotubes. Nature1996;382:54–6.

[37] Tans SJ, Verschueren ARM, Dekker C. Room-temperature tran-sistor based on a single carbon nanotube. Nature 1998;393:49–52.

[38] Kong J, Soh HT, Cassell AM, Quate CF, Dai H. Synthesis of individualsingle-walled carbon nanotubes on patterned silicon wafers. Nature1998;395:878–81.

[39] Postma HWC, Teepen T, Yao Z, Grifoni M, Dekker C. Carbon nanotubesingle-electron transistors at room temperature. Science 2001;293:76–9.

[40] Zhou C, Kong J, Yenilmez E, Dai H. Modulated chemical doping ofindividual carbon nanotubes. Science 2000;290:1552–5.

[41] Derycke V, Martel R, Appenzeller J, Avouris P. Carbon nanotube inter-and intramolecular logic gates. Nano Letters 1; 2001. p. 453. Webreleasehttp://pubs.acs.org/journals/nalefd/index.html.

[42] Bachtold A, Hadley P, Nakanishi T, Dekker C. Logic circuits withcarbon nanotube transistors. Science 2001;293:1317–20.

[43] Cao A, Zhang X, Xu C, Wei B, Wu D. Tandem structure of alignedcarbon nanotubes on Au and its solar thermal absorption. Solar EnergyMater & Solar cells 2002;70:481–6.

[44] Golberg D, Bando Y, Kurashima K, Sato T. Synthesis and characteri-zation of ropes made of BN multiwalled nanotubes. Scripta Materialia2001;44(8/9):1561–5.

[45] http://flightprojects.msfc.nasa.gov/fd02elev.html. NASA MarshallSpace Flight Center, Flight Projects Directorate, FD02 AdvancedProjects Office, Huntsville, AL 35812.

[46] Edwards BC. Design and deployment of a space elevator. Acta Astro-nautica 2000;47(10):735–44.

[47] Dillon AC, Jones KM, Bekkedahl TA, Kiang CH, Bethune DS, HebenMJ. Storage of hydrogen in single-walled carbon nanotubes. Nature1997;386:377–9.

[48] Hummer, et al. Carbon nanotubes turn on water flow. Technology re-view; 2001. p. 29.

[49] Dai H, Hafner JH, Rinzler AG, Colbert DT, Smalley RE. Nanotube asnanoprobes in scanning probe microscopy. Nature 1996;384:147–50.

[50] Hebard AF. http://www.phys.ufl.edu/∼argus/imagegallery/twotipssem.htm. Department of Physics, University of Florida.

[51] Hafner JH, Cheung CL, Lieber CM. Growth of nanotubes for probemicroscopy tips. Nature 1999;398:761–2.

[52] Kim P, Lieber CM. Nanotube nanotweezers. Science 1999;286:2148–50.

[53] Baughman RH, Cui C, Zakhidov AA, Iqbal Z, Barisci JN, Spinks GM,et al. Carbon nanotube actuators. Science 1999;284:1340–4.

[54] ZettI A.http://Physics.berkeley.edu/research/zettl/welcome.html. Con-densed matter physics. Department of Physics, University of Californiaat Berkeley.

[55] Cumings J, Zettl A. Low-friction nanoscale linear bearing realized frommultiwall carbon nanotubes. Science 2000;289:602–4.