First-Principles-Based Multiscale, Multiparadigm Molecular Mechanics and Dynamics Methods for Describing Complex Chemical Processes - Multiscale Molecular Methods in Applied Chemistry - page 18

Recently, we have demonstrated the use of reaxFF to the challenging problem of
elucidating the growth process of carbon nanotubes (CNTs). Understanding this
process is critical for determining the control variables that lead to chiral-specific
(with semiconducting or metallic electrical conductivity behavior) mass production
of CNTs. These results are summarized in the following section.
Application Example: Dynamics of CNT Growth
Since their discovery in 1991 [ 66 ], CNTs have been widely studied. Researchers
have proposed CNT applications to an ample set of technologies [ 67 ] including
interconnects, transistors, and diodes for microelectronics [ 68 ], as well as electro-
chemical transducers [ 69 ], sensor components [ 70 ], field emission devices [ 71 ], and
even gas sensors [ 72 ]. The mass production of uniform, well-characterized CNTs is
crucial for realizing many of these applications. However, while CNT synthesis has
been demonstrated for numerous catalysts, and a wide range of reaction conditions,
complete product control has remained elusive [ 73 ]. Thus, multiple investigations
aimed at elucidating the key mechanism or mechanisms of CNT growth are still
being carried out, in the hope that a more fundamental understanding of the growth
process will result in better synthetic control [ 74 ]. Experimental observations have
shed some light on CNT growth mechanisms. Atomic force microscopy (AFM),
scanning electron microscopy (SEM), and tunneling electron microscopy (TEM)
have been used to support instances of tip and base growth mechanism in different
synthesis procedures [ 75 - 77 ]. More recently time-resolved, high-resolution in situ
TEM studies have highlighted the role of catalyst deformation in SWNT growth
and provided direct experimental validation for a Yarmulke mechanism for nucle-
ation [ 78 , 79 ]. Nevertheless, these cutting edge techniques provide overarching,
general descriptions rather than detailed, atomistic mechanisms for each stage of
CNT synthesis.
To fill in these experimentally inaccessible details, mechanistic studies often
appeal to atomistic simulations. DFT is now widely used to explore catalytic
systems, and has been applied to simplistic models of CNT growth [ 74 , 80 , 81 ].
Nevertheless, the usefulness of DFT is hampered by stringent limitations on the
number of atoms and especially the number of structural iterations that it is feasible
to consider with current computer technology [ 82 ]. Tight binding (TB) methods,
which use approximations (i.e., simplified integrals) to reduce the computational
cost of handling electron-electron interactions explicitly, have been used in con-
junction with MD simulations to study this problem [ 83 ]; however, the timescales
necessary for observing the growth process are still beyond the reach of this approach -
even though TB calculations are typically a couple orders of magnitude faster than
DFT [ 83 ]. Monte Carlo methods have provided another popular means of “simulat-
ing” CNT growth [ 82 , 84 ]. At best, however, Monte Carlo methods show a
succession of possible snapshots from the growth process, leaving the mechanistic
details hidden.