《复合材料 Composites》课程教学资源（学习资料）第五章 陶瓷基复合材料_fibrous monoliths-24-CARBON NANOTUBE

COMPOSITES SCIENCE AND TECHNOLOGY ELSEⅤIER Composites Science and Technology 61(2001)1899-1912 www.elsevier.com/locate/compscitech Advances in the science and technology of carbon nanotubes and their composites: a review Erik t. Thostensona, Zhifeng Ren, Tsu-Wei Chou* Department of Mechanical Engineering and Center for Composite Materials, University of Delaware, Newark, DE 19716, US.A Department of Physics, Boston College, Chestnut Hill, M. 02167, USA Received I May 2001; received in revised form 19 June 2001; accepted 21 June 2001 Abstract Since their first observation nearly a decade ago by lijima (lijima S. Helical microtubules of graphitic carbon Nature. 1991 354: 56-8), carbon nanotubes have been the focus of considerable research. Numerous investigators have since reported remarkable physical and mechanical properties for this new form of carbon. From unique electronic properties and a thermal conductivity higher than diamond to mechanical properties where the stiffness, strength and resilience exceeds any current material, carbon nanotubes offer tremendous opportunities for the development of fundamentally new material systems. In particular, the excep- tional mechanical properties of carbon nanotubes, combined with their low density, offer scope for the development of nanotube reinforced composite materials. The potential for nanocomposites reinforced with carbon tubes having extraordinary specific stiff- ness and strength represent tremendous opportunity for application in the 21st century. This paper provides a concise review of recent advances in carbon nanotubes and their composites. We examine the research work reported in the literature on the structure d processing of carbon nanotubes, as well as characterization and property modeling of carbon nanotubes and their composites C 2001 Elsevier Science Ltd. All rights reserved 1. Introduction consequence of their symmetric structure. Many researchers have reported mechanical properties of car- In the mid 1980s, Smalley and co-workers at Rice bon nanotubes that exceed those of any previously niversity developed the chemistry of fullerenes [2]. existing materials. Although there are varying reports in Fullerenes are geometric cage-like structures of carbon the literature on the exact properties of carbon nano- atoms that are composed of hexagonal and pentagonal tubes, theoretical and experimental results have shown faces. The first closed, convex structure formed was extremely high elastic modulus, greater than I TPa(the the C6o molecule. Named after the architect known for elastic modulus of diamond is 1.2 TPa)and reported designing geodesic domes, R. Buckminster Fuller, strengths 10-100 times higher than the strongest steel buckminsterfullerene closed cage of 60 carbon at a fraction of the weight. Indeed, if the reported atoms where each side of a pentagon is the adjacent side mechanical properties are accurate, carbon nanotubes of a hexagon similar to a soccer ball( the Coo molecule is may result in an entire new class of advanced materials often referred to as a bucky ball)[]. A few years later, To unlock the potential of carbon nanotubes for appli- their discovery led to the synthesis of carbon nanotubes. cation in polymer nanocomposites, one must fully Nanotubes are long, slender fullerenes where the walls understand the elastic and fracture properties of carbon of the tubes are hexagonal carbon (graphite structure) nanotubes as well as the interactions at the nanotube and often capped at each end matrix interface. Although this requirement is no dif- These cage-like forms of carbon have been shown ferent from that for conventional fiber-reinforced com- to exhibit exceptional material properties that are a posites [3], the scale of the reinforcement phase diameter has changed from micrometers(e.g. glass and carbon 4 Corresponding author. Tel: +1-302-831-2421; fax: +1-302-831 fibers)to nanometers In addition to the exceptional mechanical properties E-mail address: chou(@ me. udeledu(T.w. Chou) associated with carbon nanotubes, they also posses 0266-3538/01/ S.see front matter C 2001 Elsevier Science Ltd. All rights reserved. PII:S0266-3538(01)00094-X

Advances in the science and technology of carbon nanotubes and their composites: a review Erik T. Thostensona , Zhifeng Renb, Tsu-Wei Choua,* a Department of Mechanical Engineering and Center for Composite Materials, University of Delaware, Newark, DE 19716, USA bDepartment of Physics, Boston College, Chestnut Hill, MA 02167, USA Received 1 May 2001; received in revised form 19 June 2001; accepted 21 June 2001 Abstract Since their first observation nearly a decade ago by Iijima (Iijima S. Helical microtubules of graphitic carbon Nature. 1991; 354:56–8), carbon nanotubes have been the focus of considerable research. Numerous investigators have since reported remarkable physical and mechanical properties for this new form of carbon. From unique electronic properties and a thermal conductivity higher than diamond to mechanical properties where the stiffness, strength and resilience exceeds any current material, carbon nanotubes offer tremendous opportunities for the development of fundamentally new material systems. In particular, the excep￾tional mechanical properties of carbon nanotubes, combined with their low density, offer scope for the development of nanotube￾reinforced composite materials. The potential for nanocomposites reinforced with carbon tubes having extraordinary specific stiff- ness and strength represent tremendous opportunity for application in the 21st century. This paper provides a concise review of recent advances in carbon nanotubes and their composites. We examine the research work reported in the literature on the structure and processing of carbon nanotubes, as well as characterization and property modeling of carbon nanotubes and their composites. # 2001 Elsevier Science Ltd. All rights reserved. 1. Introduction In the mid 1980s, Smalley and co-workers at Rice University developed the chemistry of fullerenes [2]. Fullerenes are geometric cage-like structures of carbon atoms that are composed of hexagonal and pentagonal faces. The first closed, convex structure formed was the C60 molecule. Named after the architect known for designing geodesic domes, R. Buckminster Fuller, buckminsterfullerene is a closed cage of 60 carbon atoms where each side of a pentagon is the adjacent side of a hexagon similar to a soccer ball (the C60 molecule is often referred to as a bucky ball) [2]. A few years later, their discovery led to the synthesis of carbon nanotubes. Nanotubes are long, slender fullerenes where the walls of the tubes are hexagonal carbon (graphite structure) and often capped at each end. These cage-like forms of carbon have been shown to exhibit exceptional material properties that are a consequence of their symmetric structure. Many researchers have reported mechanical properties of car￾bon nanotubes that exceed those of any previously existing materials. Although there are varying reports in the literature on the exact properties of carbon nano￾tubes, theoretical and experimental results have shown extremely high elastic modulus, greater than 1 TPa (the elastic modulus of diamond is 1.2 TPa) and reported strengths 10–100 times higher than the strongest steel at a fraction of the weight. Indeed, if the reported mechanical properties are accurate, carbon nanotubes may result in an entire new class of advanced materials. To unlock the potential of carbon nanotubes for appli￾cation in polymer nanocomposites, one must fully understand the elastic and fracture properties of carbon nanotubes as well as the interactions at the nanotube/ matrix interface. Although this requirement is no dif￾ferent from that for conventional fiber-reinforced com￾posites [3], the scale of the reinforcement phase diameter has changed from micrometers (e.g. glass and carbon fibers) to nanometers. In addition to the exceptional mechanical properties associated with carbon nanotubes, they also posses 0266-3538/01/$ - see front matter # 2001 Elsevier Science Ltd. All rights reserved. PII: S0266-3538(01)00094-X Composites Science and Technology 61 (2001) 1899–1912 www.elsevier.com/locate/compscitech * Corresponding author. Tel.: +1-302-831-2421; fax: +1-302-831- 3619. E-mail address: chou@me.udel.edu (T.-W. Chou)

1900 E.T. Thostenson ef al/ Composites Science and Technology 61(2001) 1899-1912 uperior thermal and electric properties: thermally stable up to 2800C in vacuum, thermal conductivity about twice as high as diamond, electric-current-carrying capacity 1000 times higher than copper wires [4. These exceptional properties of carbon nanotubes have been investigated for devices such as field-emission display [5]. scanning probe microscopy tips [6], and micro- electronic devices [7, 8]. In this paper we provide an overview of the recent advances in processing, character- ization, and modeling of carbon nanotubes and their 0, Chiral Angle composites. This review is not intended to be compre hensive, as our focus is on exploiting the exceptional mechanical properties of carbon nanotubes toward the development of f macroscopic structural materials Fig. I. Schematic diagram showing how a hexagonal sheet of grap Indeed, the exceptional physical properties of carbon is rolled to form a carbon nanotube tifunctionalea a present the opportunity to develop mul- nanotubes al notube composites with tailored physical two limiting cases exist where the chiral angle is at 0o nd mechanical properties and 30. These limiting cases are referred to as ziz-zag (0)and armchair(30%) based on the geometry of the carbon bonds around the circumference of the nanotube 2. Atomic structure and morphology of carbon nanotubes The difference in armchair and zig-zag nanotube struc- tures is shown in Fig. 2. In terms of the roll-up vector, the Carbon nanotubes can be visualized as a sheet of ziz-zag nanotube is (n, 0)and the armchair nanotube graphite that has been rolled into a tube. Unlike dia-(n, n). The roll-up vector of the nanotube also defines the mond, where a 3-D diamond cubic crystal structure is nanotube diameter since the inter-atomic spacing of the formed with each carbon atom having four nearest carbon atoms is known neighbors arranged in a tetrahedron, graphite is formed The chirality of the carbon nanotube has significant as a 2-D sheet of carbon atoms arranged in a hexagonal implications on the material properties. In particular, array. In this case, each carbon atom has three nearest tube chirality is known to have a strong impact on the neighbors. 'Rolling sheets of graphite into cylinders electronic properties of carbon nanotubes. Graphite is forms carbon nanotubes. The properties of nanotubes considered to be a semi-metal, but it has been shown depend on atomic arrangement(how the sheets of gra- that nanotubes can be either metallic or semiconduct phite are 'rolled), the diameter and length of the tubes, ing, depending on tube chirality [9] and the morphology, or nano structure. Nanotubes Investigations on the influence of chirality on the exist as either single-walled or multi-walled structures, mechanical properties have also been reported. The ind multi-walled carbon nanotubes (MWCNTs) are analytical work of Yakobson et al. [10, 11] examined the simply composed of concentric single-walled carbon nanotubes(SWCNTs 2. Nanotube structure oO The atomic structure of nanotubes is described in terms of the tube chirality, or helicity, which is defined ooo ao op- ocp by the chiral vector, Ch, and the chiral angle, 6. In Fig. I we can visualize cutting the graphite sheet along the dotted lines and rolling the tube so that the tip of the chiral vector touches its tail. the chiral vector often known as the roll-up vector, can be described by the oooO following equation eo9φ8°。。8 here the integers (n, m) are the number of steps along the ziz-zag carbon bonds of the hexagonal lattice and al and a are unit vectors, shown in Fig. 1. The chiral strations of the atomic structure of (a) an armd angle determines the amount of 'twist in the tube. the a ziz-zag nanotube

superior thermal and electric properties: thermally stable up to 2800
C in vacuum, thermal conductivity about twice as high as diamond, electric-current-carrying capacity 1000 times higher than copper wires [4]. These exceptional properties of carbon nanotubes have been investigated for devices such as field-emission displays [5], scanning probe microscopy tips [6], and micro￾electronic devices [7,8]. In this paper we provide an overview of the recent advances in processing, character￾ization, and modeling of carbon nanotubes and their composites. This review is not intended to be compre￾hensive, as our focus is on exploiting the exceptional mechanical properties of carbon nanotubes toward the development of macroscopic structural materials. Indeed, the exceptional physical properties of carbon nanotubes also present the opportunity to develop mul￾tifunctional nanotube composites with tailored physical and mechanical properties. 2. Atomic structure and morphology of carbon nanotubes Carbon nanotubes can be visualized as a sheet of graphite that has been rolled into a tube. Unlike dia￾mond, where a 3-D diamond cubic crystal structure is formed with each carbon atom having four nearest neighbors arranged in a tetrahedron, graphite is formed as a 2-D sheet of carbon atoms arranged in a hexagonal array. In this case, each carbon atom has three nearest neighbors. ‘Rolling’ sheets of graphite into cylinders forms carbon nanotubes. The properties of nanotubes depend on atomic arrangement (how the sheets of gra￾phite are ‘rolled’), the diameter and length of the tubes, and the morphology, or nano structure. Nanotubes exist as either single-walled or multi-walled structures, and multi-walled carbon nanotubes (MWCNTs) are simply composed of concentric single-walled carbon nanotubes (SWCNTs). 2.1. Nanotube structure The atomic structure of nanotubes is described in terms of the tube chirality, or helicity, which is defined by the chiral vector, C~h, and the chiral angle, . In Fig. 1, we can visualize cutting the graphite sheet along the dotted lines and rolling the tube so that the tip of the chiral vector touches its tail. The chiral vector, often known as the roll-up vector, can be described by the following equation: C~h ¼ na~1 þ ma~2 ð1Þ where the integers (n, m) are the number of steps along the ziz-zag carbon bonds of the hexagonal lattice and a~1 and a~2 are unit vectors, shown in Fig. 1. The chiral angle determines the amount of ‘twist’ in the tube. The two limiting cases exist where the chiral angle is at 0
and 30
. These limiting cases are referred to as ziz-zag (0
) and armchair (30
) based on the geometry of the carbon bonds around the circumference of the nanotube. The difference in armchair and zig-zag nanotube struc￾tures is shown in Fig. 2. In terms of the roll-up vector, the ziz-zag nanotube is (n, 0) and the armchair nanotube is (n, n). The roll-up vector of the nanotube also defines the nanotube diameter since the inter-atomic spacing of the carbon atoms is known. The chirality of the carbon nanotube has significant implications on the material properties. In particular, tube chirality is known to have a strong impact on the electronic properties of carbon nanotubes. Graphite is considered to be a semi-metal, but it has been shown that nanotubes can be either metallic or semiconduct￾ing, depending on tube chirality [9]. Investigations on the influence of chirality on the mechanical properties have also been reported. The analytical work of Yakobson et al. [10,11] examined the Fig. 2. Illustrations of the atomic structure of (a) an armchair and (b) a ziz-zag nanotube. Fig. 1. Schematic diagram showing how a hexagonal sheet of graphite is ‘rolled’ to form a carbon nanotube. 1900 E.T. Thostenson et al. / Composites Science and Technology 61 (2001) 1899–1912

E.T. Thostenson et al. Composites Science and Technology 61(2001)1899-1912 1901 instability of carbon nanotubes beyond linear response Their simulations show that carbon nanotubes are remarkably resilient, sustaining extreme strain with no signs of brittleness or plasticity. Although the chirality has a relatively small influence on the elastic stiffness, they concluded that the Stone-Wales transformation, a reversible diatomic interchange where the resulting structure is two pentagons and two heptagons in pairs snm plays a key role in the nanotube plastic deformation under tension The Stone- Wales transformation shown in Fig 3, occurs when an armchair nanotube is stressed Fig 4. TEM micrograph showing the layered structure of a multi- in the axial direction. Nardelli et al. 12] theorized that walled carbon nanotube the Stone- Wales transformation results in ductile frac- ture for armchair nanotubes are held together by secondary, van der Waals bonding Single-walled nanotubes are most desired for fundamental 2. 2. Nanotube tions of the carbon nanotubes. since the intra-tube interactions fur As mentioned before, fullerenes are closed, convex ther complicate the properties of carbon nanotubes cages that are composed of pentagons and exagons Indeed. both single and multi-walled nanotubes show The Stone-Wales transformation introduces a new unique properties that can be exploited for use in com defect in the nanotube structure, the heptagon. Hepta gons allow for concave areas within the nanotube. Thus, the heptagonal defects in nanotubes can result in many possible equilibrium shapes. Indeed, most nanotubes are 3. Pr ocess ssing of carbon nanotubes for composite not straight cylinders with hemispherical caps materials In addition to different tube morphologies resulting from defects, carbon nanotubes can be single walled or Since carbon nanotubes were discovered nearly a dec multi-walled structures. Fig. 4 shows a transmission ade ago, there have been a variety of techniques devel electron microscope (TEM) image showing the nano- oped for producing them. lijima [1] first observed multi structure of a multi-walled carbon nanotube where sev- walled nanotubes, and lijima et al. 13] and Bethune et al eral layers of graphitic carbon and a hollow core are evi- [14] independently reported the synthesis of single-walled dent. Multi-walled carbon nanotubes are essentially nanotubes a few years later. Primary synthesis method concentric single walled tubes, where each individual tube for single and multi-walled carbon nanotubes include can have different chirality. These concentric nanotubes arc-discharge [1, 15, laser ablation [16), gas-phase cata lytic growth from carbon monoxide [17], and chemical 食 vapor deposition(CVD)from hydrocarbons [18-201 methods. For application of carbon nanotubes in com posites, large quantities of nanotubes are required, and the scale-up limitations of the arc discharge and laser ablation techniques would make the cost of nanotube- 5 based composites prohibitive. During nanotube synthesis impurities in the form of catalyst particles, amorphous carbon, and non-tubular fullerenes are also produced T rate the tubes. The gas-phase processes tend to produce nanotubes with fewer impurities and are more amenable to large-scale processing. It is the authors' belief that gas 5 phase techniques, such as CVD, for nanotube growth offer the greatest potential for the scaling-up of nano- ube production for the processing of composites. In his section, we briefly review the primary techniques for producing carbon nanotubes and some of the benefits and draw backs of each technique lijima [1] first observed nanotubes synthesized from Fig. 3. The Stone- Wales transformation occurring in an armcha the electric-arc discharge technique. Shown schemati nanotu be under axial tension cally in Fig. 5, the arc discharge technique generally

instability of carbon nanotubes beyond linear response. Their simulations show that carbon nanotubes are remarkably resilient, sustaining extreme strain with no signs of brittleness or plasticity. Although the chirality has a relatively small influence on the elastic stiffness, they concluded that the Stone-Wales transformation, a reversible diatomic interchange where the resulting structure is two pentagons and two heptagons in pairs, plays a key role in the nanotube plastic deformation under tension. The Stone-Wales transformation, shown in Fig. 3, occurs when an armchair nanotube is stressed in the axial direction. Nardelli et al. [12] theorized that the Stone-Wales transformation results in ductile frac￾ture for armchair nanotubes. 2.2. Nanotube morphology As mentioned before, fullerenes are closed, convex cages that are composed of pentagons and hexagons. The Stone-Wales transformation introduces a new defect in the nanotube structure, the heptagon. Hepta￾gons allow for concave areas within the nanotube. Thus, the heptagonal defects in nanotubes can result in many possible equilibrium shapes. Indeed, most nanotubes are not straight cylinders with hemispherical caps. In addition to different tube morphologies resulting from defects, carbon nanotubes can be single walled or multi-walled structures. Fig. 4 shows a transmission electron microscope (TEM) image showing the nano￾structure of a multi-walled carbon nanotube where sev￾eral layers of graphitic carbon and a hollow core are evi￾dent. Multi-walled carbon nanotubes are essentially concentric single walled tubes, where each individual tube can have different chirality. These concentric nanotubes are held together by secondary, van der Waals bonding. Single-walled nanotubes are most desired for fundamental investigations of the structure/property relationships in carbon nanotubes, since the intra-tube interactions fur￾ther complicate the properties of carbon nanotubes. Indeed, both single and multi-walled nanotubes show unique properties that can be exploited for use in com￾posite materials. 3. Processing of carbon nanotubes for composite materials Since carbon nanotubes were discovered nearly a dec￾ade ago, there have been a variety of techniques devel￾oped for producing them. Iijima [1] first observed multi￾walled nanotubes, and Iijima et al. [13] and Bethune et al. [14] independently reported the synthesis of single-walled nanotubes a few years later. Primary synthesis methods for single and multi-walled carbon nanotubes include arc-discharge [1,15], laser ablation [16], gas-phase cata￾lytic growth from carbon monoxide [17], and chemical vapor deposition (CVD) from hydrocarbons [18–20] methods. For application of carbon nanotubes in com￾posites, large quantities of nanotubes are required, and the scale-up limitations of the arc discharge and laser ablation techniques would make the cost of nanotube￾based composites prohibitive. During nanotube synthesis, impurities in the form of catalyst particles, amorphous carbon, and non-tubular fullerenes are also produced. Thus, subsequent purification steps are required to sepa￾rate the tubes. The gas-phase processes tend to produce nanotubes with fewer impurities and are more amenable to large-scale processing. It is the authors’ belief that gas￾phase techniques, such as CVD, for nanotube growth offer the greatest potential for the scaling-up of nano￾tube production for the processing of composites. In this section, we briefly review the primary techniques for producing carbon nanotubes and some of the benefits and drawbacks of each technique. Iijima [1] first observed nanotubes synthesized from the electric-arc discharge technique. Shown schemati￾cally in Fig. 5, the arc discharge technique generally Fig. 3. The Stone-Wales transformation occurring in an armchair nanotube under axial tension. Fig. 4. TEM micrograph showing the layered structure of a multi￾walled carbon nanotube. E.T. Thostenson et al. / Composites Science and Technology 61 (2001) 1899–1912 1901

E.T. Thostenson et al. Composites Science and Technology 61(2001)1899-1912 growing nanotubes feedthrough connection connection Fig. 5. Schematic illustration of the arc-discharge technique (after Fig. 6. Schematic of the laser ablation process(after Ref (4D Ref.[22) involves the use of two high-purity graphite rods as the Smalley and his co-workers at Rice University have anode and cathode. The rods are brought together refined the process to produce large quantities of single- under a helium atmosphere and a voltage is applied walled carbon nanotubes with remarkable purity. The until a stable arc is achieved. The exact process variables So-called HiPco nanotubes(high-pressure conversion of depend on the size of the graphite rods. As the anode is carbon monoxide) have received considerable attention consumed, a constant gap between the anode and cath- as the technology has been commercialized by Carbon ode is maintained by adjusting the position of the Nanotechnologies Inc(Houston, TX) for large-scale anode. The material then deposits on the cathode to production of high-purity single-walled carbon nano- form a build-up consisting of an outside shell of fused tubes material and a softer fibrous core containing nanotubes Other gas-phase techniques utilize hydrocarbon gases and other carbon particles. To achieve si as the carbon source for production of both single and nanotubes, the electrodes are doped with a small multi-walled carbon nanotubes via CVD [25-28. Niko amount of metallic catalyst particles [13-15, 21, 22 laev and co-workers [17 point out that hydrocarbons first used for the initial synthes of fullerenes. Over the years, the technique has been As a consequence, nanotubes grown from hydrocarbons improved to allow the production of single-walled can have substantial amorphous carbon deposits on the nanotubes [16, 23, 24]. In this technique, a laser is used to surface of the tubes and will require further purification vaporize a graphite target held in a controlled atme steps. Although the disassociation of hydrocarbons at sphere oven at temperatures near 1200C. The general low temperatures affects the purity of the as-processed set-up for laser ablation is shown in Fig. 6. To produce nanotubes, the lower processing temperature enables single-walled nanotubes, the graphite target was doped the growth of carbon nanotubes on a wide variety of with cobalt and nickel catalyst [16]. The condensed substrates, including glass material is then collected on a water-cooled targel in- synthesize aligned arrays of carbon nanotubes with One unique aspect of CVD techniques is its ability to Both the arc-discharge and the laser-ablation ques are limited in the volume of sample they can pro- controlled diameter and length. The synthesis of well- duce in relation to the size of the carbon source(the aligned straight carbon nanotubes on a variety of sub- anode in arc-discharge and the target in laser ablation). strates has been accomplished by the use of plasma In addition, subsequent purification steps are necessary enhanced chemical vapor deposition(PeCvd)where the to separate the tubes from undesirable by-products. plasma is excited by a DC source[18-20]or a microwave These limitations have motivated the development of source [29-33]. Fig 7a and b shows the ability to grow gas-phase techniques, such as chemical vapor deposition straight carbon nanotubes over a large area with (CVD), where nanotubes are formed by the decomposi- excellent uniformity in diameter, length, straightness, tion of a carbon-containing gas. The gas-phase techni- and site density. Adjusting the thickness of the catalyst ques are amenable to continuous processes since the layer controls the diameter of the tubes, shown in carbon source is continually replaced by flowing gas In Fig. &a and b addition, the final purity of the as-produced nanotubes In CVd growth of straight carbon nanotube arrays, can be quite high, minimizing subsequent purification described by Ren et al. [19]a substrate is first coated a layer of nickel catalyst. High-purity ammonia is then Nikolaev et al. [17 describe the gas-phase growth of used as the catalytic gas and acetylene as the carbon single-walled carbon nanotubes with carbon monoxide source. a direct-current power generates the required as the carbon source. They reported the highest yields of plasma, and a deeply carbonized tungsten filament single walled nanotubes occurred at the highest acces- assists the dissociation of the reactive gases and supplies ble temperature and pressure(1200C, 10 atm). heat to the substrate. Control over the nanotube length

involves the use of two high-purity graphite rods as the anode and cathode. The rods are brought together under a helium atmosphere and a voltage is applied until a stable arc is achieved. The exact process variables depend on the size of the graphite rods. As the anode is consumed, a constant gap between the anode and cath￾ode is maintained by adjusting the position of the anode. The material then deposits on the cathode to form a build-up consisting of an outside shell of fused material and a softer fibrous core containing nanotubes and other carbon particles. To achieve single walled nanotubes, the electrodes are doped with a small amount of metallic catalyst particles [13–15,21,22]. Laser ablation was first used for the initial synthesis of fullerenes. Over the years, the technique has been improved to allow the production of single-walled nanotubes [16,23,24]. In this technique, a laser is used to vaporize a graphite target held in a controlled atmo￾sphere oven at temperatures near 1200
C. The general set-up for laser ablation is shown in Fig. 6. To produce single-walled nanotubes, the graphite target was doped with cobalt and nickel catalyst [16]. The condensed material is then collected on a water-cooled target. Both the arc-discharge and the laser-ablation techni￾ques are limited in the volume of sample they can pro￾duce in relation to the size of the carbon source (the anode in arc-discharge and the target in laser ablation). In addition, subsequent purification steps are necessary to separate the tubes from undesirable by-products. These limitations have motivated the development of gas-phase techniques, such as chemical vapor deposition (CVD), where nanotubes are formed by the decomposi￾tion of a carbon-containing gas. The gas-phase techni￾ques are amenable to continuous processes since the carbon source is continually replaced by flowing gas. In addition, the final purity of the as-produced nanotubes can be quite high, minimizing subsequent purification steps. Nikolaev et al. [17] describe the gas-phase growth of single-walled carbon nanotubes with carbon monoxide as the carbon source. They reported the highest yields of single walled nanotubes occurred at the highest acces￾sible temperature and pressure (1200
C, 10 atm). Smalley and his co-workers at Rice University have refined the process to produce large quantities of single￾walled carbon nanotubes with remarkable purity. The so-called HiPco nanotubes (high-pressure conversion of carbon monoxide) have received considerable attention as the technology has been commercialized by Carbon Nanotechnologies Inc (Houston, TX) for large-scale production of high-purity single-walled carbon nano￾tubes. Other gas-phase techniques utilize hydrocarbon gases as the carbon source for production of both single and multi-walled carbon nanotubes via CVD [25–28]. Niko￾laev and co-workers [17] point out that hydrocarbons pyrolize readily on surfaces heated above 600–700
C. As a consequence, nanotubes grown from hydrocarbons can have substantial amorphous carbon deposits on the surface of the tubes and will require further purification steps. Although the disassociation of hydrocarbons at low temperatures affects the purity of the as-processed nanotubes, the lower processing temperature enables the growth of carbon nanotubes on a wide variety of substrates, including glass. One unique aspect of CVD techniques is its ability to synthesize aligned arrays of carbon nanotubes with controlled diameter and length. The synthesis of well￾aligned, straight carbon nanotubes on a variety of sub￾strates has been accomplished by the use of plasma￾enhanced chemical vapor deposition (PECVD) where the plasma is excited by a DC source [18–20] or a microwave source [29–33]. Fig. 7a and b shows the ability to grow straight carbon nanotubes over a large area with excellent uniformity in diameter, length, straightness, and site density. Adjusting the thickness of the catalyst layer controls the diameter of the tubes, shown in Fig. 8a and b. In CVD growth of straight carbon nanotube arrays, described by Ren et al. [19] a substrate is first coated with a layer of nickel catalyst. High-purity ammonia is then used as the catalytic gas and acetylene as the carbon source. A direct-current power generates the required plasma, and a deeply carbonized tungsten filament assists the dissociation of the reactive gases and supplies heat to the substrate. Control over the nanotube length Fig. 5. Schematic illustration of the arc-discharge technique (after Ref. [22]). Fig. 6. Schematic of the laser ablation process (after Ref. [4]). 1902 E.T. Thostenson et al. / Composites Science and Technology 61 (2001) 1899–1912

E T. Thostenson et al/ Composites Science and Technology 61(2001)1899-1912 Fig. 7. Micrographs showing the straightness of MWCNTs grown via PECVD [191 Fig 8. Micrographs showing control over the nanotube diameter:(a)40-50 nm and(b)200-300 nm aligned carbon nanotubes [19]- and graphitization is accomplished by changing the tube furnace. Fig. 10 is a SEM micrograph of the fur growth time and temperature, respectively, and applica- nace -grown carbon nanotubes showing the same ran- tion of the dC plasma results in tube growth in the dom, curled structure associated with thermal CVD direction of the plasma. The use of an alternating micro-(Shown in Fig 9). The outer diameters of these tubes wave frequency source to excite the plasma results in the range from 10-50 nm. These tangled, spaghetti-like growth of carbon nanotubes that occur directly normal to nanotubes can be produced at a larger quantity and lower the surface of the substrate. Bower et al. [29] showed that cost than PECVD tubes but there is less control over in microwave plasma-enhanced CVD (MPECV length, diameter, and structure alignment of the carbon nanotubes results from the self- bias that is imposed on the surface of the substrate from the microwave plasma. Fig 9a shows the alignment of 4. Characterization of carbon nanotubes carbon nanotubes grown normal to the surface of an optical glass fiber. To gain further insight into the Significant challenges exist in both the micromechanical mechanism for tube alignment, the tubes were grown for characterization of nanotubes and the modeling of the two minutes under the microwave- induced plasma fol- elastic and fracture behavior at the nano-scale Challenges lowed by 70 min with the plasma off. Fig 9b shows the in characterization of nanotubes and their composites results of this experiment. The upper portion of the include (a) complete lack of micromechanical character nanotubes are straight, indicating alignment in the ization techniques for direct property measurement, (b) plasma, and the base shows a random, curled structure tremendous limitations on specimen size, (c)uncertainty associated with thermal CVD. In addition, the growth in data obtained from indirect measurements, and (d) rate under the plasma enhancement was 40 times faster inadequacy in test specimen preparation techniques and than the thermal CVD lack of control in nanotube alignment and distribution In addition to highly aligned arrays of carbon nano- In order better to understand the mechanical proper tubes, large quantities of carbon nanotubes can be pro- ties of carbon nanotubes, a number of investigators have essed by conventional CVd techniques. Unlike attempted to characterize carbon nanotubes directly PECVD, which requires the use of specialized plasma Treacy et al. [34] first investigated the elastic modulus of equipment, tangled carbon nanotubes are grown in nanotu bes by measuring, in the

and graphitization is accomplished by changing the growth time and temperature, respectively, and applica￾tion of the DC plasma results in tube growth in the direction of the plasma. The use of an alternating micro￾wave frequency source to excite the plasma results in the growth of carbon nanotubes that occur directly normal to the surface of the substrate. Bower et al. [29] showed that in microwave plasma-enhanced CVD (MPECVD) alignment of the carbon nanotubes results from the self￾bias that is imposed on the surface of the substrate from the microwave plasma. Fig. 9a shows the alignment of carbon nanotubes grown normal to the surface of an optical glass fiber. To gain further insight into the mechanism for tube alignment, the tubes were grown for two minutes under the microwave-induced plasma fol￾lowed by 70 min with the plasma off. Fig. 9b shows the results of this experiment. The upper portion of the nanotubes are straight, indicating alignment in the plasma, and the base shows a random, curled structure associated with thermal CVD. In addition, the growth rate under the plasma enhancement was 40 times faster than the thermal CVD. In addition to highly aligned arrays of carbon nano￾tubes, large quantities of carbon nanotubes can be pro￾cessed by conventional CVD techniques. Unlike PECVD, which requires the use of specialized plasma equipment, tangled carbon nanotubes are grown in a tube furnace. Fig. 10 is a SEM micrograph of the fur￾nace-grown carbon nanotubes showing the same ran￾dom, curled structure associated with thermal CVD (shown in Fig. 9). The outer diameters of these tubes range from 10–50 nm. These tangled, spaghetti-like nanotubes can be produced at a larger quantity and lower cost than PECVD tubes, but there is less control over length, diameter, and structure. 4. Characterization of carbon nanotubes Significant challenges exist in both the micromechanical characterization of nanotubes and the modeling of the elastic and fracture behavior at the nano-scale. Challenges in characterization of nanotubes and their composites include (a) complete lack of micromechanical character￾ization techniques for direct property measurement, (b) tremendous limitations on specimen size, (c) uncertainty in data obtained from indirect measurements, and (d) inadequacy in test specimen preparation techniques and lack of control in nanotube alignment and distribution. In order better to understand the mechanical proper￾ties of carbon nanotubes, a number of investigators have attempted to characterize carbon nanotubes directly. Treacy et al. [34] first investigated the elastic modulus of isolated multi-walled nanotubes by measuring, in the Fig. 7. Micrographs showing the straightness of MWCNTs grown via PECVD [19]. Fig. 8. Micrographs showing control over the nanotube diameter: (a) 40–50 nm and (b) 200–300 nm aligned carbon nanotubes [19]. E.T. Thostenson et al. / Composites Science and Technology 61 (2001) 1899–1912 1903

E.T. Thostenson et al. Composites Science and Technology 61(2001)1899-1912 Thermal CVD MPECVD (b) Fig. 9. Micrographs showing(a)nanotubes aligned normal to the surface of a glass fiber and(b) the influence of MPECVD on the structure of the Single-walled nanotubes tend to assemble in'ropes'of nanotubes. Salvetat and co-workers [35 measured the properties of these nanotube bundles with the AFM. As he diameter of the tube bundles increases the axial and shear moduli decrease significantly. This suggests slip- ping of the nanotubes within the bundle. Walters et al [37 further investigated the elastic strain of nanotube bundles with the AFM. On the basis of their experi- mental strain measurements and an assumed elastic modulus of 1.25 TPa, they calculated a yield strength of 45+7 GPa for the nanotube ropes. Indeed, their calcu lated value for strength would be much lower if the elastic modulus of the nanotube bundle is decreased as a Fig 10 Micrograph showing tangled, spaghetti-like carbon nano- consequence of slipping within the bundle, suggested by tubes grown with conventional CVD techniques. Salvetat et al. [ 36] Yu and co-workers [38, 39] have investigated the ten sile loading of multi-walled nanotubes and single-walled transmission electron microscope, the amplitude of their nanotube ropes. In their work, the nanotubes were intrinsic thermal vibration. The average value obtained attached between two opposing AFM tips and loaded over 1l samples was 1.8 TPa. Direct measurement of under tension. Their experimental set-up is shown in the stiffness and strength of individual, structurally iso- Fig. 11. For multi-walled carbon nanotubes [38] the lated multi-wall carbon nanotubes has been made with failure of the outermost tube occurred followed by pull- an atomic-force microscope(AFM). Wong and co- ut of the inner nanotubes This 'sword and sheath' workers [35] were the first to perform direct measure- telescoping failure mechanism of multi-walled carbon ment of the stiffness and strength of individual structu- nanotubes in tension is also shown in Fig. 11. The rally isolated multi-wall carbon nanotubes using atomic experimentally calculated tensile strengths of the outer force microscopy. The nanotube was pinned at one end most layer ranged from 11 to 63 GPa and the elastic to molybdenum disulfide surfaces and load was applied modulus ranged from 270 to 950 GPa. In their sub to the tube by means of the AFM tip. The bending force sequent investigation of single-walled nanotube ropes was measured as a function of displacement along the [39], they assumed that only the outermost tubes assem- unpinned length, and a value of 1. 26 TPa was obtained bled in the rope carried the load during the experiment, for the elastic modulus. The average bending strength and they calculated tensile strengths of 13 to 52 GPa measured was 14.2+8 GPa and average elastic moduli of 320 to 1470 GPa. Xie et

transmission electron microscope, the amplitude of their intrinsic thermal vibration. The average value obtained over 11 samples was 1.8 TPa. Direct measurement of the stiffness and strength of individual, structurally iso￾lated multi-wall carbon nanotubes has been made with an atomic-force microscope (AFM). Wong and co￾workers [35] were the first to perform direct measure￾ment of the stiffness and strength of individual, structu￾rally isolated multi-wall carbon nanotubes using atomic force microscopy. The nanotube was pinned at one end to molybdenum disulfide surfaces and load was applied to the tube by means of the AFM tip. The bending force was measured as a function of displacement along the unpinned length, and a value of 1.26 TPa was obtained for the elastic modulus. The average bending strength measured was 14.28 GPa. Single-walled nanotubes tend to assemble in ‘ropes’ of nanotubes. Salvetat and co-workers [35] measured the properties of these nanotube bundles with the AFM. As the diameter of the tube bundles increases, the axial and shear moduli decrease significantly. This suggests slip￾ping of the nanotubes within the bundle. Walters et al. [37] further investigated the elastic strain of nanotube bundles with the AFM. On the basis of their experi￾mental strain measurements and an assumed elastic modulus of 1.25 TPa, they calculated a yield strength of 457 GPa for the nanotube ropes. Indeed, their calcu￾lated value for strength would be much lower if the elastic modulus of the nanotube bundle is decreased as a consequence of slipping within the bundle, suggested by Salvetat et al. [36]. Yu and co-workers [38,39] have investigated the ten￾sile loading of multi-walled nanotubes and single-walled nanotube ropes. In their work, the nanotubes were attached between two opposing AFM tips and loaded under tension. Their experimental set-up is shown in Fig. 11. For multi-walled carbon nanotubes [38] the failure of the outermost tube occurred followed by pull￾out of the inner nanotubes. This ‘sword and sheath’ telescoping failure mechanism of multi-walled carbon nanotubes in tension is also shown in Fig. 11. The experimentally calculated tensile strengths of the outer￾most layer ranged from 11 to 63 GPa and the elastic modulus ranged from 270 to 950 GPa. In their sub￾sequent investigation of single-walled nanotube ropes [39], they assumed that only the outermost tubes assem￾bled in the rope carried the load during the experiment, and they calculated tensile strengths of 13 to 52 GPa and average elastic moduli of 320 to 1470 GPa. Xie et Fig. 9. Micrographs showing (a) nanotubes aligned normal to the surface of a glass fiber and (b) the influence of MPECVD on the structure of the nanotubes [29]. Fig. 10. Micrograph showing tangled, spaghetti-like carbon nano￾tubes grown with conventional CVD techniques. 1904 E.T. Thostenson et al. / Composites Science and Technology 61 (2001) 1899–1912

E.T. Thostenson et al. Composites Science and Technology 61(2001)1899-1912 Fig. 11. Micrograph phs showing(a) the apparatus for tensile loading of MWCNTs and(b) the telescoping, ""sword and sheath"fracture behavior of the MWCNT [38] al. [40] also tested ropes of multi-walled nanotubes in with parameters determined from first principles. A tension. In their experiments, the obtained tensile comparison of the bending stiffnesses of single-walled strength and modulus were 3.6 and 450 GPa, respec- nanotubes and an iridium beam was presented. The tively. It was suggested that the lower values for bending stiffness of the iridium beam was deduced by strength and stiffness may be a consequence of defects using the continuum Bernoulli-Euler theory of beam in the CVD-grown nanotubes bending. Overney and co-workers concluded that the beam bending rigidity of a nanotube exceeds the highest values found in any other presently available materials 5. Mechanics of carbon nanotubes Besides their experimental observations, Iijima et al [47] examined response of nanotubes under compression As discussed in the previous section, nanotube defor- using molecular dynamics simulations. They simulated mation has been examined experimentally. Recent the deformation properties of single- and multi-walled investigations have shown that carbon nanotubes pos- nanotubes bent to large angles. Their experimental and sess remarkable mechanical properties, such as excep- theoretical results show that nanotubes are remarkably tionally high elastic modulus [34,35, large elastic strain flexible. The bending is completely reversible up to and fracture strain sustaining capability [41, 42]. Similar angles in excess of 110, despite the formation of com conclusions have also been reached through some theo- plex kink shapes. Fig. 12 shows their numerical and retical studies [43-46], although very few correlations experimental results, demonstrating the exceptional between theoretical predictions and experimental studies resilience of carbon nanotubes at large strain. have been made In this section we examine the mechan. Ru [48] noticed that actual bending stiffness of single- ics of both single walled and multi-walled nanotubes walled nanotubes is much lower than that given by the elastic-continuum shell model if the commonly defined 5. Single-walled carbon nanotubes representative thickness is used. Ru proposed the use of an effective nanotube bending stifness as a material Theoretical studies concerning the mechanical prop parameter not related to the representative thickness rties of single-walled nanotubes have been pursued With the aid of this concept, the elastic shell equations extensively. Overney et al. [43] studied the low-frequency can be readily modified and then applied to single-wal- vibrational modes and structural rigidity of long nano- led nanotubes. The computational results based on this tubes consisting of 100, 200 and 400 atoms. The calcula- concept show a good agreement with the results from tions were based on an empirical Keating Hamiltonian molecular dynamics simulations

al. [40] also tested ropes of multi-walled nanotubes in tension. In their experiments, the obtained tensile strength and modulus were 3.6 and 450 GPa, respec￾tively. It was suggested that the lower values for strength and stiffness may be a consequence of defects in the CVD-grown nanotubes. 5. Mechanics of carbon nanotubes As discussed in the previous section, nanotube defor￾mation has been examined experimentally. Recent investigations have shown that carbon nanotubes pos￾sess remarkable mechanical properties, such as excep￾tionally high elastic modulus [34,35], large elastic strain and fracture strain sustaining capability [41,42]. Similar conclusions have also been reached through some theo￾retical studies [43–46], although very few correlations between theoretical predictions and experimental studies have been made. In this section we examine the mechan￾ics of both single walled and multi-walled nanotubes. 5.1. Single-walled carbon nanotubes Theoretical studies concerning the mechanical prop￾erties of single-walled nanotubes have been pursued extensively. Overney et al. [43] studied the low-frequency vibrational modes and structural rigidity of long nano￾tubes consisting of 100, 200 and 400 atoms. The calcula￾tions were based on an empirical Keating Hamiltonian with parameters determined from first principles. A comparison of the bending stiffnesses of single-walled nanotubes and an iridium beam was presented. The bending stiffness of the iridium beam was deduced by using the continuum Bernoulli-Euler theory of beam bending. Overney and co-workers concluded that the beam bending rigidity of a nanotube exceeds the highest values found in any other presently available materials. Besides their experimental observations, Iijima et al. [47] examined response of nanotubes under compression using molecular dynamics simulations. They simulated the deformation properties of single- and multi-walled nanotubes bent to large angles. Their experimental and theoretical results show that nanotubes are remarkably flexible. The bending is completely reversible up to angles in excess of 110
, despite the formation of com￾plex kink shapes. Fig. 12 shows their numerical and experimental results, demonstrating the exceptional resilience of carbon nanotubes at large strain. Ru [48] noticed that actual bending stiffness of single￾walled nanotubes is much lower than that given by the elastic-continuum shell model if the commonly defined representative thickness is used. Ru proposed the use of an effective nanotube bending stiffness as a material parameter not related to the representative thickness. With the aid of this concept, the elastic shell equations can be readily modified and then applied to single-wal￾led nanotubes. The computational results based on this concept show a good agreement with the results from molecular dynamics simulations. Fig. 11. Micrographs showing (a) the apparatus for tensile loading of MWCNTs and (b) the telescoping, ‘‘sword and sheath’’ fracture behavior of the MWCNT [38]. E.T. Thostenson et al. / Composites Science and Technology 61 (2001) 1899–1912 1905

E.T. Thostenson et al. Composites Science and Technology 61(2001)1899-1912 potential and molecular dynamics simulations. Their molecular-dynamics simulations show that carbon nano- tubes, when subjected to large deformations, reversibly switch into different morphological patterns. Each shape change corresponds to an abrupt release of energy and singularity in the stress/strain curve. These transforma- tions are explained well by a continuum shell model. With properly chosen parameters, their model provided a very accurate'roadmap'of nanotube behavior beyond the lin ear elastic regime. They also made molecular dynamics simulations to single- and double-walled nanotubes of different chirality and at different temperatures [45]. Their emulations show that nanotubes have an extremely large breaking strain (in the range 30-40%)and the breaking train decreases with temperature. Yakobson [11] also applied dislocation theory to carbon nanotubes for Fig. 12. TEM micrograph and computer simulation of nanotube describing their main routes of mechanical relaxation buckling [47]. under tension. It was concluded that the yield strength of a nanotube depends on its symmetry and it was believed Vaccarini et al. [49] investigated the influence of that there exists an intra-molecular plastic flow. Under anotube structure and chirality on the elastic proper- high stress, this plastic flow corresponds to a motion of ties in tension, bending, and torsion. They found that dislocations along helical paths within the nanotube wall the chirality played a small influence on the nanotube and causes a stepwise necking, a well-defined new sym tensile modulus. However, the chiral tubes exhibit metry, as the domains of different chiral symmetry are asymmetric torsional behavior with respect to left and formed. As a result, both the mechanical and electronic right twist, whereas the armchair and ziz-zag tubes do properties of carbon nanotubes are changed not exhibit this asymmetric torsional behavior. The single walled nanotubes produced by laser abla- A relatively comprehensive study of the elastic prop- tion and arc-discharge techniques have a greater ten- erties of single-walled nanotubes was reported by Lu dency to form ropes'or aligned bundles [15, 23]. Thus, [44]. In this study, Lu adopted an empirical lattice- theoretical studies have been made to investigate the dynamics model [50], which has been successfully adopted mechanical properties of these nanotube bundles. Ru in calculating the phonon spectrum and elastic properties [52] presented a modified elastic-honeycomb model to of graphite. In this lattice-dynamics model, atomic inter tudy elastic buckling of nanotube ropes under high actions in a single carbon layer are approximated by a pressure. Ru gave a simple formula for the critical pres sum of pair-wise harmonic potentials between atoms. sure as a function of nanotube Youngs modulus and The local structure of a nanotube layer is constructed wall thickness-to-radius ratio. It was concluded that sin from conformal mapping of a graphite sheet on to a gle-walled ropes are susceptible to elastic buckling under cylindrical surface. Lu's work attempted to answer such high pressure and elastic buckling is responsible for the nanotubes depend on the structural details, such as size electrical resistivity of single walled nanotubes odes and basic questions as: (a)how do elastic properties of pressure-induced abnormalities of vibration mo and chirality? and(b) how do elastic properties of Popov et al. [53] studied the elastic properties of tri- nanotubes compare with those of graphite and diamond? angular crystal lattices formed by single-walled nano- Lu concluded that the elastic properties of nanotubes are tubes by using analytical expressions based on a force- insensitive to size and chirality. The predicted Youngs constant lattice dynamics model [54]. They calculated modulus (I TPa), shear modulus(0.45 TPa), and various elastic constants of nanotube crystals for nano- bulk modulus(0.74 TPa)are comparable to those of tube types, such as armchair and zigzag. It was shown diamond. Hernandez and co-workers [51] performed cal- that the elastic modulus, Poissons ratio and bulk mod culations similar to those of Lu and found slightly higher ulus clearly exhibit strong dependence on the tube values( 1. 24 TPa)for the Youngs moduli of tubes. But radius. The bulk modulus was found to have a max unlike Lu, they found that elastic moduli are sensitive to imum value of 38 GPa for crystals composed of single- both tube diameter and structure walled nanotubes with 0.6 nm radius Besides their unique elastic properties behavior of nanotubes has also received considerable 5.2. Multi-walled carbon nanotubes attention. Yakobson and co-workers [10, 46] examined the instability behavior of carbon nanotubes beyond linear Multi-walled nanotubes are composed of a number of response by using a realistic many-body Tersoff-Brenner concentric single walled nanotubes held together with

Vaccarini et al. [49] investigated the influence of nanotube structure and chirality on the elastic proper￾ties in tension, bending, and torsion. They found that the chirality played a small influence on the nanotube tensile modulus. However, the chiral tubes exhibit asymmetric torsional behavior with respect to left and right twist, whereas the armchair and ziz-zag tubes do not exhibit this asymmetric torsional behavior. A relatively comprehensive study of the elastic prop￾erties of single-walled nanotubes was reported by Lu [44]. In this study, Lu adopted an empirical lattice￾dynamics model [50], which has been successfully adopted in calculating the phonon spectrum and elastic properties of graphite. In this lattice-dynamics model, atomic inter￾actions in a single carbon layer are approximated by a sum of pair-wise harmonic potentials between atoms. The local structure of a nanotube layer is constructed from conformal mapping of a graphite sheet on to a cylindrical surface. Lu’s work attempted to answer such basic questions as: (a) how do elastic properties of nanotubes depend on the structural details, such as size and chirality? and (b) how do elastic properties of nanotubes compare with those of graphite and diamond? Lu concluded that the elastic properties of nanotubes are insensitive to size and chirality. The predicted Young’s modulus (1 TPa), shear modulus (0.45 TPa), and bulk modulus (0.74 TPa) are comparable to those of diamond. Hernandez and co-workers [51] performed cal￾culations similar to those of Lu and found slightly higher values (1.24 TPa) for the Young’s moduli of tubes. But unlike Lu, they found that elastic moduli are sensitive to both tube diameter and structure. Besides their unique elastic properties, the inelastic behavior of nanotubes has also received considerable attention. Yakobson and co-workers [10,46] examined the instability behavior of carbon nanotubes beyond linear response by using a realistic many-body Tersoff-Brenner potential and molecular dynamics simulations. Their molecular-dynamics simulations show that carbon nano￾tubes, when subjected to large deformations, reversibly switch into different morphological patterns. Each shape change corresponds to an abrupt release of energy and a singularity in the stress/strain curve. These transforma￾tions are explained well by a continuum shell model. With properly chosen parameters, their model provided a very accurate ‘roadmap’ of nanotube behavior beyond the lin￾ear elastic regime. They also made molecular dynamics simulations to single- and double-walled nanotubes of different chirality and at different temperatures [45]. Their simulations show that nanotubes have an extremely large breaking strain (in the range 30–40%) and the breaking strain decreases with temperature. Yakobson [11] also applied dislocation theory to carbon nanotubes for describing their main routes of mechanical relaxation under tension. It was concluded that the yield strength of a nanotube depends on its symmetry and it was believed that there exists an intra-molecular plastic flow. Under high stress, this plastic flow corresponds to a motion of dislocations along helical paths within the nanotube wall and causes a stepwise necking, a well-defined new sym￾metry, as the domains of different chiral symmetry are formed. As a result, both the mechanical and electronic properties of carbon nanotubes are changed. The single walled nanotubes produced by laser abla￾tion and arc-discharge techniques have a greater ten￾dency to form ‘ropes’ or aligned bundles [15,23]. Thus, theoretical studies have been made to investigate the mechanical properties of these nanotube bundles. Ru [52] presented a modified elastic-honeycomb model to study elastic buckling of nanotube ropes under high pressure. Ru gave a simple formula for the critical pres￾sure as a function of nanotube Young’s modulus and wall thickness-to-radius ratio. It was concluded that sin￾gle-walled ropes are susceptible to elastic buckling under high pressure and elastic buckling is responsible for the pressure-induced abnormalities of vibration modes and electrical resistivity of single walled nanotubes. Popov et al. [53] studied the elastic properties of tri￾angular crystal lattices formed by single-walled nano￾tubes by using analytical expressions based on a force￾constant lattice dynamics model [54]. They calculated various elastic constants of nanotube crystals for nano￾tube types, such as armchair and zigzag. It was shown that the elastic modulus, Poisson’s ratio and bulk mod￾ulus clearly exhibit strong dependence on the tube radius. The bulk modulus was found to have a max￾imum value of 38 GPa for crystals composed of single￾walled nanotubes with 0.6 nm radius. 5.2. Multi-walled carbon nanotubes Multi-walled nanotubes are composed of a number of concentric single walled nanotubes held together with Fig. 12. TEM micrograph and computer simulation of nanotube buckling [47]. 1906 E.T. Thostenson et al. / Composites Science and Technology 61 (2001) 1899–1912

E.T. Thostenson et al / Composites Science and Technology 61(2001)1899-1912 relatively weak van der Waals forces. The multi-layered 6. Nanotube-based composites structure of these nanotubes further complicates the modeling of their properties Although there is experimental variability in the direct Lorents [55 derived the tensile and bend haracterization of carbon nanotubes theoretical and ing stiffness constants of ideal multi-walled nanotubes in experimental observations reveal their exceptional terms of the known elastic properties of graphite. It is properties. As a consequence, there has been recent suggested that unlike the strongly anisotropic thermal interest in the development of nanotube-based compo- expansion in conventional carbon fibers and graphite, sites. Although most research has focused on the devel- he thermal expansion of carbon nanotubes is essen- opment of nanotube-based polymer composites, tially isotropic. However, the thermal conductivity of attempts have also been made to develop metal and nanotubes is believed to be highly anisotropic and its ceramic-matrix composites with nanotubes as reinfor- magnitude along the axial direction is perhaps higher cement. Here we review the recent work in this area and than that of any other material. discuss the challenges that are associated with the Lu [44] also calculated the elastic properties of many development of these new composites multi-walled nanotubes formed by single-layer tubes by means of the empirical-lattice dynamics model. It was 6.. Processing and characterization of nanotube-based found that elastic properties are insensitive to different polymer composites combinations of parameters, such as chirality, tube dius and numbers of layers, and the elastic properties The reported exceptional properties of nanotubes are the same for all nanotubes with a radius larger than have motivated others to investigate experimentally the one nm. Interlayer van der Waals interaction has a mechanics of nanotube-based composite films. Uniform negligible contribution to both the tensile and shear dispersion within the polymer matrix and improved nanotube/matrix wetting and adhesion are critical issue Govindjee and Sackman [56] were the first to examine in the processing of these nanocomposites the use of continuum mechanics to estimate the proper The issue of nanotube dispersion is critical to efficient es of multi-walled nanotubes. They investigated the reinforcement. In the work of Salvetat et al. [36] dis validity of the continuum approach by using Bernoulli- cussed earlier, slipping of nanotubes when they are Euler bending to infer the Youngs modulus. They used assembled in ropes significantly affects the elastic prop- a simple elastic sheet model and showed that at the erties. In addition to slipping of tubes that are not bon nanotube scale the assumptions of continuum mechan- ded to the matrix in a composite the aggregates of ics must be carefully respected in order to obtain rea- nanotube ropes effectively reduce the aspect ratio sonable results. They showed the explicit dependence of(length/diameter) of the reinforcement. It is, however, material properties'on system size when a continuum difficult to obtain a uniform dispersion of carbon nano- cross-section assumption was used tubes in the polymer matrix. Shaffer and Windle [61] Ru [57] used the elastic-shell model to study the effect were able to process carbon nanotube/polyvinyl-alcohol of van der Waals forces on the axial buckling of a dou- composite films for mechanical characterization. The ble-walled carbon nanotube. The analysis showed that tensile elastic modulus and damping properties of the the van der Waals forces do not increase the critical composite films were assessed in a dynamic mechanical axial buckling strain of a double-walled nanotube. Ru thermal analyzer(DMTA)as a function of nanotube [58, 59] thereafter also proposed a multiple column loading and temperature. From the theory developed model that considers the interlayer radial displacements for short-fiber composites, a nanotube elastic modulus coupled through the van der Waals forces. This model of 150 MPa was obtained from the experimental data was used to study the effect of interlayer displacements This value in a microscopic composite is well below the on column buckling. It was concluded that the effect of values reported for isolated nanotubes. It is not clear interlayer displacements could not be neglected unless whether this result is a consequence of imperfections in the van der Waals forces are extremely strong the graphite layers of catalytically own na Kolmogorov and Crespi [60]investigated the interlayer used for the investigation or whether it relates to a fun interaction in two-walled nanotubes. A registry-depen- damental difficulty in stress transfer dent two-body graphite potential was developed. It was Qian et al. [62] characterized carbon-nanotube/poly demonstrated that the tightly constrained geometry of a styrene composites. With only the addition of 1% by multi-walled nanotube could produce an extremely weight(about 0.5% by volume) they achieved between smooth solid-solid interface wherein the corrugation 36-42% increase in the elastic stiffness and a 25% against sliding does not grow with system size. The ener- increase in the tensile strength. Fig. 13 is a TEM micro- getic barrier to interlayer sliding in defect-free nanotubes graph of their nanotube film showing the mechanisms of containing thousands of atoms can be comparable to fracture. As for conventional fiber composites, we see that for a single unit cell of crystalline graphite. nanotube pull-out, nanotube fracture, as well as crack

relatively weak van der Waals forces. The multi-layered structure of these nanotubes further complicates the modeling of their properties. Ruoff and Lorents [55] derived the tensile and bend￾ing stiffness constants of ideal multi-walled nanotubes in terms of the known elastic properties of graphite. It is suggested that unlike the strongly anisotropic thermal expansion in conventional carbon fibers and graphite, the thermal expansion of carbon nanotubes is essen￾tially isotropic. However, the thermal conductivity of nanotubes is believed to be highly anisotropic and its magnitude along the axial direction is perhaps higher than that of any other material. Lu [44] also calculated the elastic properties of many multi-walled nanotubes formed by single-layer tubes by means of the empirical-lattice dynamics model. It was found that elastic properties are insensitive to different combinations of parameters, such as chirality, tube radius and numbers of layers, and the elastic properties are the same for all nanotubes with a radius larger than one nm. Interlayer van der Waals interaction has a negligible contribution to both the tensile and shear stiffness. Govindjee and Sackman [56] were the first to examine the use of continuum mechanics to estimate the proper￾ties of multi-walled nanotubes. They investigated the validity of the continuum approach by using Bernoulli￾Euler bending to infer the Young’s modulus. They used a simple elastic sheet model and showed that at the nanotube scale the assumptions of continuum mechan￾ics must be carefully respected in order to obtain rea￾sonable results. They showed the explicit dependence of ‘material properties’ on system size when a continuum cross-section assumption was used. Ru [57] used the elastic-shell model to study the effect of van der Waals forces on the axial buckling of a dou￾ble-walled carbon nanotube. The analysis showed that the van der Waals forces do not increase the critical axial buckling strain of a double-walled nanotube. Ru [58,59] thereafter also proposed a multiple column model that considers the interlayer radial displacements coupled through the van der Waals forces. This model was used to study the effect of interlayer displacements on column buckling. It was concluded that the effect of interlayer displacements could not be neglected unless the van der Waals forces are extremely strong. Kolmogorov and Crespi [60] investigated the interlayer interaction in two-walled nanotubes. A registry-depen￾dent two-body graphite potential was developed. It was demonstrated that the tightly constrained geometry of a multi-walled nanotube could produce an extremely smooth solid-solid interface wherein the corrugation against sliding does not grow with system size. The ener￾getic barrier to interlayer sliding in defect-free nanotubes containing thousands of atoms can be comparable to that for a single unit cell of crystalline graphite. 6. Nanotube-based composites Although there is experimental variability in the direct characterization of carbon nanotubes, theoretical and experimental observations reveal their exceptional properties. As a consequence, there has been recent interest in the development of nanotube-based compo￾sites. Although most research has focused on the devel￾opment of nanotube-based polymer composites, attempts have also been made to develop metal and ceramic-matrix composites with nanotubes as reinfor￾cement. Here we review the recent work in this area and discuss the challenges that are associated with the development of these new composites. 6.1. Processing and characterization of nanotube-based polymer composites The reported exceptional properties of nanotubes have motivated others to investigate experimentally the mechanics of nanotube-based composite films. Uniform dispersion within the polymer matrix and improved nanotube/matrix wetting and adhesion are critical issues in the processing of these nanocomposites. The issue of nanotube dispersion is critical to efficient reinforcement. In the work of Salvetat et al. [36] dis￾cussed earlier, slipping of nanotubes when they are assembled in ropes significantly affects the elastic prop￾erties. In addition to slipping of tubes that are not bon￾ded to the matrix in a composite, the aggregates of nanotube ropes effectively reduce the aspect ratio (length/diameter) of the reinforcement. It is, however, difficult to obtain a uniform dispersion of carbon nano￾tubes in the polymer matrix. Shaffer and Windle [61] were able to process carbon nanotube/polyvinyl-alcohol composite films for mechanical characterization. The tensile elastic modulus and damping properties of the composite films were assessed in a dynamic mechanical thermal analyzer (DMTA) as a function of nanotube loading and temperature. From the theory developed for short-fiber composites, a nanotube elastic modulus of 150 MPa was obtained from the experimental data. This value in a microscopic composite is well below the values reported for isolated nanotubes. It is not clear whether this result is a consequence of imperfections in the graphite layers of catalytically grown nanotubes used for the investigation or whether it relates to a fun￾damental difficulty in stress transfer. Qian et al. [62] characterized carbon-nanotube/poly￾styrene composites. With only the addition of 1% by weight (about 0.5% by volume) they achieved between 36–42% increase in the elastic stiffness and a 25% increase in the tensile strength. Fig. 13 is a TEM micro￾graph of their nanotube film showing the mechanisms of fracture. As for conventional fiber composites, we see nanotube pull-out, nanotube fracture, as well as crack E.T. Thostenson et al. / Composites Science and Technology 61 (2001) 1899–1912 1907

E.T. Thostenson et al. Composites Science and Technology 61(2001)1899-1912 Because the interaction at the nanotube/ matrix inter face is critical to understanding the mechanical behavior of nanotube-based composites, a number of researchers have investigated the efficiency of interfacial stress transfer. Wagner et al. [67 examined stress-induced fragmentation of multi-walled carbon nanotubes in polymer films. Their nanotube-containing film had a thickness of approximately 200 um. The observed frag mentation phenomenon was attributed to either pro- cess-induced stress resulting from curing of the polymer or tensile stress generated by polymer deformation and transmitted to the nanotube. from estimated values of nanotube axial normal stress and elastic modulus 500nm Wagner and co-workers concluded that the nanotube, polymer interfacial shear stress is on the order of 500 Fig 13. Fracture mechanisms in nanotube-based composites[] MPa and higher. This value, if reliable, is an order of magnitude higher than the stress-transfer ability of cur- rent advanced composites and, therefore, such inter- faces are more able than either the matrix or the bridging by the nanotubes. They also used short-fiber nanotubes themselves to sustain shear. In further work composite theory to demonstrate that 10% by weight of Lourie and Wagner [68-70 investigated tensile and carbon fibers(about 5% by volume) in the research of compressive fracture in nanotube-based composites Tibbetts and McHugh [63] would be required to achieve Stress transfer has also been investigated by raman he same increase in elastic modulus with 1%(by spectroscopy. Cooper and co-workers [71] prepared composite, but to take full advantage of the exceptional Raman band(2610 cm -)to a lower 28<99 weight) of carbon nanotubes composite specimens by applying an epoxy-resin/nano Fig. 13 shows significant pull-out of the nanotubes tube mixture to the surface of an epoxy beam. After the rom the matrix. Clearly there is reinforcement as evi- specimens were cured, stress transfer between the poly denced by improvements in stiffness and strength of the mer and the nanotubes was detected by a shift in the g avenu ber. The stifness, strength, and resilience of carbon nanotubes, shift in the G Raman band corresponds to strain in the strong interfacial bonding is critical. Jia et al. [64 graphite structure, and the shift indicates that there is showed that the nanotubes can be initiated by a free- stress transfer, and hence reinforcement, by the nano- radical initiator, AiBN (2, 2'-azobisisobutyronitrile), to tubes. It was also concluded that the effective modulus open their bonds In their study of carbon-nanotube/ of single-walled nanotubes dispersed in a composite poly(methyl methacrylate)(PMMA) composites, the could be over 1 TPa and that of multi-walled nanotubes possibility exists to form a C-c bond between the was about 0.3 TPa In their investigation of single-wal nanotube and the matrix. Gong et al. [65] investigated led nanotube/epoxy composites, Ajayan et al. [72]sug- surfactant-assisted processing of nanotube composites gest that their nearly constant value of the Raman peak with a nonionic surfactant. Improved dispersion and in tension is related to tube sliding within the nanotube interfacial bonding of the nanotubes in an epoxy matrix bundles and, hence, poor interfacial load transfer resulted in a 30% increase in elastic modulus with between the nanotubes. Similar results were obtained by addition of I wt% nanotubes Schadler et al. [73. Multi-walled nanotube/epoxy com Lordi and Yao [66] looked at the molecular mechan- posites were tested in both tension and compression ics of binding in nanotube-based composites. In their The compressive modulus was found to be higher than work, they used force-field-based molecular-mechanics the tensile modulus of the composites, and the raman calculations to determine the binding energies and slid peak was found to shift only in compression, indicating ing frictional stresses between pristine carbon nanotubes poor interfacial load transfer in tension and different polymeric matrix materials. The binding Even with improved dispersion and adhesion, micro energies and frictional forces were found to play only a mechanical characterization of these composites is diffi minor role in determing the strength of the interface. cult because the distribution of the nanotubes is The key factor in forming a strong bond at the interface random. Thus, attempts have been made to align nano- is having a helical conformation of the polymer around tubes in order better to elucidate the reinforcement the nanotube. They suggested that the strength of the mechanisms. Jin et al. [74] showed that aligned nano- interface may result from molecular-level entanglement tube composites could be obtained by mechanical of the two phases and forced long-range ordering of the stretching of the composite. X-ray diffraction was used to determine the orientation and degree of alignment

bridging by the nanotubes. They also used short-fiber composite theory to demonstrate that 10% by weight of carbon fibers (about 5% by volume) in the research of Tibbetts and McHugh [63] would be required to achieve the same increase in elastic modulus with 1% (by weight) of carbon nanotubes. Fig. 13 shows significant pull-out of the nanotubes from the matrix. Clearly there is reinforcement as evi￾denced by improvements in stiffness and strength of the composite, but to take full advantage of the exceptional stiffness, strength, and resilience of carbon nanotubes, strong interfacial bonding is critical. Jia et al. [64] showed that the nanotubes can be initiated by a free￾radical initiator, AIBN (2,20 -azobisisobutyronitrile), to open their p bonds. In their study of carbon-nanotube/ poly(methyl methacrylate) (PMMA) composites, the possibility exists to form a C–C bond between the nanotube and the matrix. Gong et al. [65] investigated surfactant-assisted processing of nanotube composites with a nonionic surfactant. Improved dispersion and interfacial bonding of the nanotubes in an epoxy matrix resulted in a 30% increase in elastic modulus with addition of 1 wt.% nanotubes. Lordi and Yao [66] looked at the molecular mechan￾ics of binding in nanotube-based composites. In their work, they used force-field-based molecular-mechanics calculations to determine the binding energies and slid￾ing frictional stresses between pristine carbon nanotubes and different polymeric matrix materials. The binding energies and frictional forces were found to play only a minor role in determing the strength of the interface. The key factor in forming a strong bond at the interface is having a helical conformation of the polymer around the nanotube. They suggested that the strength of the interface may result from molecular-level entanglement of the two phases and forced long-range ordering of the polymer. Because the interaction at the nanotube/matrix inter￾face is critical to understanding the mechanical behavior of nanotube-based composites, a number of researchers have investigated the efficiency of interfacial stress transfer. Wagner et al. [67] examined stress-induced fragmentation of multi-walled carbon nanotubes in polymer films. Their nanotube-containing film had a thickness of approximately 200 mm. The observed frag￾mentation phenomenon was attributed to either pro￾cess-induced stress resulting from curing of the polymer or tensile stress generated by polymer deformation and transmitted to the nanotube. From estimated values of nanotube axial normal stress and elastic modulus, Wagner and co-workers concluded that the nanotube/ polymer interfacial shear stress is on the order of 500 MPa and higher. This value, if reliable, is an order of magnitude higher than the stress-transfer ability of cur￾rent advanced composites and, therefore, such inter￾faces are more able than either the matrix or the nanotubes themselves to sustain shear. In further work, Lourie and Wagner [68–70] investigated tensile and compressive fracture in nanotube-based composites. Stress transfer has also been investigated by Raman spectroscopy. Cooper and co-workers [71] prepared composite specimens by applying an epoxy-resin/nano￾tube mixture to the surface of an epoxy beam. After the specimens were cured, stress transfer between the poly￾mer and the nanotubes was detected by a shift in the G0 Raman band (2610 cm 1 ) to a lower wavenumber. The shift in the G0 Raman band corresponds to strain in the graphite structure, and the shift indicates that there is stress transfer, and hence reinforcement, by the nano￾tubes. It was also concluded that the effective modulus of single-walled nanotubes dispersed in a composite could be over 1 TPa and that of multi-walled nanotubes was about 0.3 TPa. In their investigation of single-wal￾led nanotube/epoxy composites, Ajayan et al. [72] sug￾gest that their nearly constant value of the Raman peak in tension is related to tube sliding within the nanotube bundles and, hence, poor interfacial load transfer between the nanotubes. Similar results were obtained by Schadler et al. [73]. Multi-walled nanotube/epoxy com￾posites were tested in both tension and compression. The compressive modulus was found to be higher than the tensile modulus of the composites, and the Raman peak was found to shift only in compression, indicating poor interfacial load transfer in tension. Even with improved dispersion and adhesion, micro￾mechanical characterization of these composites is diffi- cult because the distribution of the nanotubes is random. Thus, attempts have been made to align nano￾tubes in order better to elucidate the reinforcement mechanisms. Jin et al. [74] showed that aligned nano￾tube composites could be obtained by mechanical stretching of the composite. X-ray diffraction was used to determine the orientation and degree of alignment. Fig. 13. Fracture mechanisms in nanotube-based composites [62]. 1908 E.T. Thostenson et al. / Composites Science and Technology 61 (2001) 1899–1912