How Evaporating Carbon Nanotubes Retain Their Perfection? Feng Ding, Kun Jiao, Yu Lin, and Boris I. Yakobson* Department of Mechanical Engineering and Materials Science and Department of Chemistry, Rice UniVersity, Houston, Texas 77005 Received November 24, 2006; Revised Manuscript Received January 27, 2007 ABSTRACT We present a mechanism of high-temperature sublimation of carbon nanotubes that does not destroy their ordered makeup even upon significant loss of mass. The atoms depart to the gas phase from the pentagon-heptagon dislocation cores, while the bond disruption is immediately repaired, and the 5|7 seamlessly propagate through the lattice. This explains a broad class of unsettled phenomena when at high temperature or under radiation the nanotubes do not become amorphous but rather shrink in size nearly flawlessly. Carbon nanotubes owe their amazing and important proper- ties (electronic, optical, or mechanical) to the unique structure, a honeycomb net seamlessly wrapped into a cylinder. Essentially it renders them crystals with no facets or surface, where each atom has exactly the same environ- ment. 1 How carbon atoms assemble into such tubules through catalytic growth turned out to be a daunting problem for theory. Here we discuss a process opposite to growth, a decomposition of nanotube into carbon atoms as it occurs at high near-sublimation temperature 2-4 or under electron radiation. 5-7 As the analysis below demonstrates, such seemingly random loss of atoms can proceed in a rather organized fashion with the peculiar self-repair mechanism. It shows that sublimation is inherently coupled with the dislocation dynamics in two-dimensional (2D) crystals: ejection of atoms can create edge dislocation dipoles, and it governs the dislocation climb-glide movements, while the dislocation cores serve as scavengers for possibly emerging point defects. This way, a membrane crystal can lose substantial fraction of its mass while maintaining almost perfect structure: it simply shrinks down in scale without much disorder or amorphization. This also well explains the recent remarkable observation of superplastic nanotubes, 4 where their great elongation in spite of multifold mass reduction must proceed through the sublimation-plasticity relaxation process described below. Although the discussion can broadly apply to 2D crystals and tubes (single graphene layer, 8 cylindrical micelles, 9 micro- tubules 10 ), carbon nanotubes alone offer an abundant record of experimental study. For example, almost 4-fold reduction of diameter (1.4-0.4 nm) under an electron beam did not disrupt the tubular structure, 5 or a thin single-wall neck formed from multiwall tube irradiated at 600 °C. 6,7 Similarly, irradiation was able to transform a carbon cluster into a well- structured carbon onion. 11,12 Most striking is the recent observation of perfectly preserved structure (besides the few mobile kinks) in the course of plastic stretching of a tube, although most of its body (80% mass loss) expired into gas. 4 This accumulating evidence poses a compelling question of how a graphitic layer can avoid defect buildup and especially larger holes while losing a great fraction of its atoms. High temperature of the atomic lattice brings significant thermal agitation (often enhanced by an electron beam of the microscope). Carbon atoms can more frequently exchange their positions, e.g., via Stone-Wales (SW) bond rotations, 1 by forming Schottky vacancies, interstitial-vacancy pairs, etc. Furthermore, some of the atoms are increasingly likely to entirely abandon the lattice for the benefit of a more entropic gas state, the essence of sublimation, which results in total mass reduction. Intuitively, the exodus of atoms from the lattice should randomly create a growing number of vacancies, inducing disorder and perhaps amorphization, or aggregating in holes (Figure 1a-c). This notion contrasts with the very clean cylindrical shape preserved in experi- ments in spite of the big reduction in size. 3-5 How can the vast loss in mass be reconciled with the smooth morphing of an ever-sealed tubule (Figure 1a f d) from its initial to final geometry? Natural propensity of covalent bonds to restore connectivity, to self-heal in a network is important but not sufficient, as a trial simulation illustrates: in the course of random removal of atoms, many dangling bonds do pair nicely, but the degree of disorder and size of holes progressively increase (Figure 1e-g). Apparently, nature has a more clever mechanism preserving a nearly perfect 2D lattice in spite of its evaporation. To understand it, consider more thoroughly possible C breakout from different locations of a tube containing generic * Corresponding author. E-mail: biy@rice.edu. NANO LETTERS 2007 Vol. 7, No. 3 681-684 10.1021/nl0627543 CCC: $37.00 © 2007 American Chemical Society Published on Web 02/16/2007