Fundamental Electronic Properties And Applications Of

Acc. Chem. Res. 2002, 35, 1018-1025Fundamental ElectronicProperties and Applications ofSingle-Walled Carbon NanotubesMIN OUYANG,† JIN-LIN HUANG,† ANDCHARLES M. LIEBER*,†,‡Department of Chemistry and Chemical Biology and Divisionof Engineering and Applied Science, Harvard University,Cambridge, Massachusetts 02138Received February 13, 2002ABSTRACTRecent scanning tunneling microscopy studies of the intrinsicelectronic properties of single-walled carbon nanotubes (SWNTs)are overviewed in this Account. A brief theoretical treatment of theelectronic properties of SWNTs is developed, and then the effectsof finite curvature and broken symmetry on electronic properties,the unique one-dimensional energy dispersion in nanotubes, theinteraction between local spins and carriers in metallic nanotubessystems, and the atomic structure and electronic properties ofintramolecular junctions are described. The implications of thesestudies for understanding fundamental one-dimensional physicsand future nanotube device applications are also discussed.I. IntroductionSingle-walled carbon nanotubes (SWNTs) represent a newclass of materials for investigating fundamental onedimensional (1D) physics and for exploring nanoelectronics and molecular electronics.1-4 Among the many interesting properties exhibited by nanotubes,1 it is theelectronic properties of SWNTs that are arguably the mostsignificant characteristic of this new material. A singleSWNT can be either metallic or semiconducting,5-11depending only on diameter and chirality, while the localcarbon-carbon bonding remains constant. The ability toyield both metallic and semiconducting SWNTs withoutMin Ouyang was born in Fujian, China. He received B.S. and M.S. degrees inelectronics from Peking University, and a Ph.D. in physical chemistry under thesupervision of Charles Lieber from Harvard University in 2001. Ouyang’s researchinterests focus on the fundamental properties and device applications of lowdimensional materials. He is currently pursuing postdoctoral research in physicswith David Awschalom at the University of California, Santa Barbara.Jin-Lin Huang received his undergraduate and doctoral degrees in physics fromFudan University. He joined Charles Lieber’s group at Columbia University in 1989as a postdoctoral fellow and is currently a research associate in the group atHarvard. Huang is interested in the basic physics of low-dimensional electronicsystems and nanostructures, with specific emphasis on the development andapplication of low-temperature ultrahigh vacuum scanning tunneling microscopyto these systems.Charles M. Lieber attended Franklin and Marshall College for his undergraduateeducation, and after doctoral studies at Stanford University and postdoctoralresearch at the California Institute of Technology, started his independent careerin 1987 at Columbia University. Lieber moved to Harvard University in 1991, andnow holds a joint appointment in the Department of Chemistry and ChemicalBiology, where he is the Mark Hyman Professor of Chemistry, and the Divisionof Engineering and Applied Sciences. Lieber’s research is focused broadly onnanoscale science and technology, including the rational synthesis, fundamentalphysical properties, and hierarchical organization of nanoscale materials, as wellas the development of functional nanoelectronic and photonic systems.1018ACCOUNTS OF CHEMICAL RESEARCH / VOL. 35, NO. 12, 2002doping is unique among solid-state materials and has ledto speculation that SWNTs might thus serve as a keybuilding block for carbon-based electronics.4Scanning tunneling microscopy (STM) is a powerfultool for probing the intrinsic electronic properties ofSWNTs since both the atomic and electronic structurescan be determined simultaneously for individualSWNTs.9-11 Herein, we review recent STM investigationsof the fundamental electronic properties of SWNTs andrelated nanoelectronic devices. First, a brief descriptionof the basic relationship between the structure andelectronic properties of SWNTs is presented. Second, theroles of finite curvature and broken symmetry in perturbing the electronic properties of SWNTs are discussed.Third, studies probing the unique one-dimensional energydispersion of SWNTs are presented. Fourth, the interactions between local spins of external magnetic impuritiesand conduction electrons in both extended and finitemetallic SWNTs are described. Last, characterization ofseveral types of intramolecular SWNT junctions are presented. In concluding, the implications of these studiesfor understanding fundamental 1D physics and futuredevice applications are considered.II. Theoretical Background and Initial STMStudiesA SWNT can be viewed as a seamless cylinder obtainedby rolling-up a section of a two-dimensional (2D) graphenesheet (Figure 1A). The structure of a SWNT is uniquelycharacterized by the roll-up vector, Ch ) na1 ma2 (n,m), where a1 and a2 are the graphene primitive vectorsand n,m are integers (Figure 1B). The translation vector,T, is directed along the SWNT axis and perpendicular toCh; the magnitude of T corresponds to the length of the(n,m) SWNT unit cell. Once (n,m) is specified, otherstructural properties, such as diameter (dt) and chiralangle (θ), can be determined: dt ) (31/2/π)acc(m2 mn n2)1/2 and θ ) tan-1[31/2m/(2n m)], where acc is thenearest-neighbor carbon atom distance of 0.142 nm.Among the large number of possible Ch vectors, there aretwo inequivalent high-symmetry directions. These aretermed “zigzag” and “armchair” and are designated by(n,0) and (n,n), respectively.The basic electronic band structure of SWNTs can bederived from a graphene sheet while neglecting hybridization effects due to the finite curvature of the tubestructure. Graphene is a semimetal with valence andconduction bands degenerate at only six corners (KB) ofthe hexagonal first Brillouin zone. The Fermi surface ofthe graphene sheet is thus reduced to these six points(Figure 2A,B).12 In SWNTs, the wavevector k is quantizedalong the circumferential direction due to periodic bound* To whom correspondence should be addressed at the Departmentof Chemistry and Chemical Biology, Harvard University, 12 Oxford St.,Cambridge, MA 02138. Phone: 617-496-3169. Fax: 617-496-5442/6731.E-mail: cml@cmliris.harvard.edu.† Department of Chemistry and Chemical Biology.‡ Division of Engineering and Applied Science.10.1021/ar0101685 CCC: 22.00 2002 American Chemical SocietyPublished on Web 08/15/2002

Electronic Properties and Applications of SWNTs Ouyang et al.FIGURE 1. (A) Schematic of the roll-up of a graphene sheet to forma SWNT structure. (B) OO′ defines the chiral vector Ch ) na1 ma2 (n,m). Translation vector, T, is along the nanotube axis andperpendicular to Ch. The shaded, area represents the unrolled unitcell formed by T and Ch. The chiral angle, θ, is defined as the anglebetween the Ch and the (n,0) zigzag direction. (n,0) zigzag and (n,n)armchair SWNTs are indicated in blue and red, respectively.and otherwise it should be semiconducting. From thecriteria KB‚Ch ) 2πq, we thus expect that SWNTs aremetals when (n - m)/3 is an integer, and otherwise theyare semiconductors.The 1D band structure of SWNTs can be furtherconstructed by zone-folding the 2D graphene band structure into the 1D Brillouin zone of an (n,m) SWNT, andthe electronic density of states (DOS) can be computedfrom the band structure by summing the number of statesat every energy level.1,13 Several important characteristicsof the electronic properties of SWNTs can be immediatelyobtained from this π-only tight binding model.3,14-16 First,SWNTs exhibit well-defined spike-like features in the DOS,that is, van Hove singularities (VHS).17 Second, the DOSat EF is zero for semiconducting SWNTs (n - m * 3q) butnonzero (albeit small) for metallic SWNTs (n - m ) 3q).Third, the VHS spacing has a characteristic “1-2-4”pattern relative to EF (with spacing 1ξ-2ξ-4ξ) for semiconducting SWNTs, and “1-2-3” from EF (with spacing3ξ-6ξ-9ξ) for metallic SWNTs, where ξ ) 2π/3 Ch .Fourth, the first VHS band gaps for semiconducting andmetallic SWNTs are EgS ) 2γ0acc/dt and EgM ) 6γ0acc/dt,respectively, and are independent of chiral angle θ to firstorder.The first experiments that addressed directly thesebasic theoretical predictions were carried out by Odomet al.9 and Wildöer et al.10 using low-temperature STM.These initial STM studies characterized the atomic structures and electronic DOS of SWNTs, and thereby confirmed the existence of both semiconducting and metallicSWNTs for a wide range of structures. Subsequently, Kimet al.13 and Odom et al.3 reported the first detailedcomparisons of experimentally determined SWNT VHSwith tight binding calculations for metallic and semiconducting tubes. The good agreement between theory andthese experiments showed that the essential physics ofSWNT band structure is captured by the π-only model.However, other important issues, such as the effects offinite curvature and broken rotational symmetry, whichare essential to a complete understanding of SWNTs’electronic properties and potential device applications,were not addressed. We examine these and other fascinating questions below.III. Finite Curvature Effect of SWNTsFIGURE 2. (A) Three-dimensional plot of the π and π* grapheneenergy bands and (B) a 2D projection. (C) Allowed 1D wavevectorsfor a metallic (9,0) SWNT. (D) Allowed 1D wavevectors for asemiconducting (10,0) tube. The black hexagons define the firstBrillouin zone of a graphene sheet, and the black dots in the cornersare the KB points. Blue lines represent allowed wavevectors, k,within the first Brillouin zone.ary condition: k‚Ch ) 2πq, where q is an integer. Therefore, only a particular set of states, which are parallel tothe corresponding tube axis with a spacing of 2/dt, areallowed (Figure 2C,D). On the basis of this simple scheme,if one of the allowed wavevectors passes through a FermiKB of the graphene sheet, the SWNT should be metallic,For a SWNT with sufficiently small diameter, the hybridization of σ, σ*, π, and π* orbitals can be quite large.18Full-valence tight binding calculations19 and analyticalcalculations for a Hamiltonian on a curved surface20 havesuggested that the finite curvature of SWNTs will stronglymodify the electronic behavior of SWNTs and open upsmall energy gaps at EF. Recently, Ouyang et al.11,21 andKleiner et al.22 have independently developed a Fermipoint shifting model to provide a direct understanding offinite curvature effects. Briefly, in this model, the finitecurvature is found to induce shifts of the Fermi points ofSWNTs from original KB (Figure 3). For example, the Fermipoints of “metallic” zigzag SWNTs are found to shift awayVOL. 35, NO. 12, 2002 / ACCOUNTS OF CHEMICAL RESEARCH 1019