K w. Kolasinski/ Current Opinion in Solid State and Materials Science 10(2006)182-191 growth and axial growth can exhibit different temperature the rate of adsorption, the Gibbs-Thomson correction for dependencies and an observed decrease in axial growth rate the finite curvature of the surface, diffusion-induced Nw for both Gap above 500C and GaAs above 440C is growth, and the nucleation-mediated growth at the related to increased sidewall growth. Sidewall growth is liquid-solid interface with allowances for tapering of the also involved in producing tapered nanowires. The sub- nanowires. This enables them to make very specific predic strate can play an active role in the growth kinetics. Here tions. For example, the tapering of GaasNw grown by the Sioz substrate suppresses epitaxial growth, whereas MBE on the GaAs(1 l l)B surface at T=590C can be epitaxial growth is competitive with nanowire growth on explained by the reduction in the Ga flux to the catalyst III-V substrates caused by the high rate of desorption of Ga atoms from Extensive studies of growth kinetics and the develop- the sidewalls. Tapering due to the nucleation and growth ment of a mass-transport model have also been performed of islands on the gaAs(1 10) sidewalls is not thought to in Lund [41, 44, * 46,70]. Seifert and co-workers [46] have be significant since the activation barrier for nucleation is developed a model of a nanowire growing from a metal too high. They also are able to reproduce well growth rates particle which need not be liquid. They make the following as a function of nanowire diameter so that they can fit data assumptions:(i) the metal particle is assumed to be hemi- regarding the dependence of nanowire diameter on nano- spherical, (ii) there is steady state adatom diffusion on the wire length and the nanowire length(for a given diameter) substrate and nanowire sides toward the metal particle, as a function of growth temperature 11) the processes within the metal particle(diffusion) as well as at the metal-semiconductor interface (nucleation) 4. Conclusion need not be considered in detail, and (iv) the interwire sep- aration is fairly large. The authors note that assumption Characterization studies of catalytic growth are demon- (iii) limits the generality of the model because there are rating an ever larger array of systems that exhibit the cases in which these processes are rate determining. None- growth of nanowires and nanotubes. Our understanding theless, for the specific case of lll/V nanowire growth, of catalytic growth, regardless of whether it is labelled experimental growth conditions can easily be set to validate VLS, VSS, SLS, SCLS or otherwise, is becoming increas- these assumptions ingly sophisticated. However, may questions still remain Their model describes a mass-transport-limited system unanswered particularly regarding the size and position in which deposition occurs on the substrate as well as of the catalyst relative to the nanowire, in other words, on the nanowire walls and the metal particle. Growth is regarding aspects of float versus root growth and single favoured at the metal/ semiconductor interface, which acts prong versus multiprong growth and why the particle can as a sink, and is kinetically hindered on the substrate and be sometimes larger and sometimes smaller than the nano- walls. Transport via diffusion across the sub- wire. It is now clear that growth can occur by autocatalysis strate and up the walls plays a vital role. The microscopic as well as by catalysis due to a foreign metal/alloy particle details of this model are sketchy and the explanation for The catalytic particle can be either liquid or solid but this why incorporation does not occur on the substrate and does not really change the dynamics of growth. The result- walls is vague, nonetheless its phenomenological descrip- ing nanowires can be unitary, binary, ternary or even qua- tion of length versus radius for various temperatures is ternary compounds and they have been produced either pure or in doped forms. In some cases, more than one type Tersoff and co-workers have investigated the Au cata- of nanowire can be grown simultaneously. Catalytic lyzed VLS growth of SiNWs [14, 16]. They use disilane growth represents a powerful method for producing ID (Si2H6) as the source gas at 10-10 Torr. A thin film nanostructures and there certainly are many more surprises Au(2-3 nm) is deposited on a Si(ll 1)substrate, which to be discovered in this area of research is then heated to 500-650C. No detectable dependence of growth rate on the wire diameter is observed. They ascribe this rather unexpected result to the irreversible character of References dissociative adsorption on the catalyst surface, which is the rate determining step for these conditions. They have also The papers of particular interest have been highlighted observed periodic sawtooth faceting of SiNW walls during growth. Growth occurs at the(111)facet at the end of the of special interest. wire, however, the size and shape of this facet oscillates [ u Wagner RS, Ellis WC. Vapor-liquid-solid mechanism of single crystal growth. Appl Phys Lett 1964; 4: 89-90 forces generated by the surface energies of the wire and [2] De Jong KP, Geus JW. Carbon nanofibers: catalytic synthesis and the droplet. They also predict that this will be observed applications. Catal Rev 2000: 42: 481-510 in any system in which the orientations parallel to the [] Anantram MP, Leonard F. Physics of carbon nanotube electronic growth direction are not stable devices. Rep Prog Phys 2006: 69: 507-61 4 Ajayan PM. Nanotubes from carbon Chem Rev 1999: 99: 1787-99 Perhaps the most general scheme to explain nanowire [5] Dai H. Carbon nanotubes: opportunities and challenges. Surf Sci growth is that of Dubrovskii et al. [15]. They incorporategrowth and axial growth can exhibit different temperature dependencies and an observed decrease in axial growth rate for both GaP above 500 C and GaAs above 440 C is related to increased sidewall growth. Sidewall growth is also involved in producing tapered nanowires. The sub￾strate can play an active role in the growth kinetics. Here the SiO2 substrate suppresses epitaxial growth, whereas epitaxial growth is competitive with nanowire growth on III–V substrates. Extensive studies of growth kinetics and the develop￾ment of a mass-transport model have also been performed in Lund [*41,44,*46,70]. Seifert and co-workers [*46] have developed a model of a nanowire growing from a metal particle which need not be liquid. They make the following assumptions: (i) the metal particle is assumed to be hemi￾spherical, (ii) there is steady state adatom diffusion on the substrate and nanowire sides toward the metal particle, (iii) the processes within the metal particle (diffusion) as well as at the metal-semiconductor interface (nucleation) need not be considered in detail, and (iv) the interwire sep￾aration is fairly large. The authors note that assumption (iii) limits the generality of the model because there are cases in which these processes are rate determining. None￾theless, for the specific case of III/V nanowire growth, experimental growth conditions can easily be set to validate these assumptions. Their model describes a mass-transport-limited system in which deposition occurs on the substrate as well as on the nanowire walls and the metal particle. Growth is favoured at the metal/semiconductor interface, which acts as a sink, and is kinetically hindered on the substrate and nanowire walls. Transport via diffusion across the sub￾strate and up the walls plays a vital role. The microscopic details of this model are sketchy and the explanation for why incorporation does not occur on the substrate and walls is vague, nonetheless its phenomenological descrip￾tion of length versus radius for various temperatures is quite good. Tersoff and co-workers have investigated the Au cata￾lyzed VLS growth of SiNWs [14,16]. They use disilane (Si2H6) as the source gas at 108 –105 Torr. A thin film of Au (2–3 nm) is deposited on a Si(1 1 1) substrate, which is then heated to 500–650 C. No detectable dependence of growth rate on the wire diameter is observed. They ascribe this rather unexpected result to the irreversible character of dissociative adsorption on the catalyst surface, which is the rate determining step for these conditions. They have also observed periodic sawtooth faceting of SiNW walls during growth. Growth occurs at the (1 1 1) facet at the end of the wire; however, the size and shape of this facet oscillates periodically. They explain this based on the balance of forces generated by the surface energies of the wire and the droplet. They also predict that this will be observed in any system in which the orientations parallel to the growth direction are not stable. Perhaps the most general scheme to explain nanowire growth is that of Dubrovskii et al. [*15]. They incorporate the rate of adsorption, the Gibbs–Thomson correction for the finite curvature of the surface, diffusion-induced NW growth, and the nucleation-mediated growth at the liquid–solid interface with allowances for tapering of the nanowires. This enables them to make very specific predic￾tions. For example, the tapering of GaAsNW grown by MBE on the GaAs(1 1 1)B surface at T = 590 C can be explained by the reduction in the Ga flux to the catalyst caused by the high rate of desorption of Ga atoms from the sidewalls. Tapering due to the nucleation and growth of islands on the GaAs(1 1 0) sidewalls is not thought to be significant since the activation barrier for nucleation is too high. They also are able to reproduce well growth rates as a function of nanowire diameter so that they can fit data regarding the dependence of nanowire diameter on nano￾wire length and the nanowire length (for a given diameter) as a function of growth temperature. 4. Conclusion Characterization studies of catalytic growth are demon￾strating an ever larger array of systems that exhibit the growth of nanowires and nanotubes. Our understanding of catalytic growth, regardless of whether it is labelled VLS, VSS, SLS, SCLS or otherwise, is becoming increas￾ingly sophisticated. However, may questions still remain unanswered particularly regarding the size and position of the catalyst relative to the nanowire, in other words, regarding aspects of float versus root growth and single￾prong versus multiprong growth and why the particle can be sometimes larger and sometimes smaller than the nano￾wire. It is now clear that growth can occur by autocatalysis as well as by catalysis due to a foreign metal/alloy particle. The catalytic particle can be either liquid or solid but this does not really change the dynamics of growth. The result￾ing nanowires can be unitary, binary, ternary or even qua￾ternary compounds and they have been produced either pure or in doped forms. In some cases, more than one type of nanowire can be grown simultaneously. Catalytic growth represents a powerful method for producing 1D nanostructures and there certainly are many more surprises to be discovered in this area of research. References The papers of particular interest have been highlighted as: * of special interest. [1] Wagner RS, Ellis WC. Vapor–liquid–solid mechanism of single crystal growth. Appl Phys Lett 1964;4:89–90. [2] De Jong KP, Geus JW. Carbon nanofibers: catalytic synthesis and applications. Catal Rev 2000;42:481–510. [3] Anantram MP, Le´onard F. Physics of carbon nanotube electronic devices. Rep Prog Phys 2006;69:507–61. [4] Ajayan PM. Nanotubes from carbon. Chem Rev 1999;99:1787–99. [5] Dai H. Carbon nanotubes: opportunities and challenges. Surf Sci 2002;500:218. K.W. Kolasinski / Current Opinion in Solid State and Materials Science 10 (2006) 182–191 189