We explain the evolution and termination of vertically aligned carbon nanotube (CNT) “forests” by a collective mechanism, which is verified by temporal measurements of forest mass and height, as well as quantitative spatial mapping of CNT alignment by synchrotron X-ray scattering. We propose that forest growth consists of four stages: (I) self-organization; (II) steady growth with a constant CNT number density; (III) decay with a decreasing number density; and (IV) abrupt self-termination, which is coincident with a loss of alignment at the base of the forest. The abrupt loss of CNT alignment has been observed experimentally in many systems, yet termination of forest growth has previously been explained using models for individual CNTs, which do not consider the evolution of the CNT population. We propose that abrupt termination of CNT forest growth is caused by loss of the self-supporting structure, which is essential for formation of a CNT forest in the first place, and that this event is triggered by accumulating growth termination of individual CNTs. A finite element model accurately predicts the critical CNT number density at which forest growth terminates and demonstrates the essential role of mechanical contact in maintaining growth of self-assembled films of filamentary nanostructures.

Understanding the population growth behavior of filamentary nanostructures, such as carbon nanotubes (CNTs), is hampered by the lack of characterization techniques capable of probing statistical variations with high spatial resolution. We present a comprehensive methodology for studying the population growth dynamics of vertically aligned CNT forests, utilizing high-resolution spatial mapping of synchrotron X-ray scattering and attenuation, along with real-time height kinetics. We map the CNT alignment and dimensions within CNT forests, revealing broadening and focusing of size distributions during different stages of the process. Then, we calculate the number density and mass density of the CNT population versus time, which are true measures of the reaction kinetics. We find that the mass-based kinetics of a CNT population is accurately represented by the S-shaped Gompertz model of population growth, although the forest height and CNT length kinetics are essentially linear. Competition between catalyst activation and deactivation govern the rapid initial acceleration and slow decay of the CNT number density. The maximum CNT density (i.e., the overall catalyst activity) is limited by gas-phase reactions and catalyst-surface interactions, which collectively exhibit autocatalytic behavior. Thus, we propose a comprehensive picture of CNT population growth which combines both chemical and mechanical cooperation. Our findings are relevant to both bulk and substrate-based CNT synthesis methods and provide general insights into the self-assembly and collective growth of filamentary nanostructures.

Aligned carbon nanotube (CNT) structures are promising for many applications; however, as-grown CNT "forests" synthesized by chemical vapor deposition (CVD) are typically low-density and mostly comprise tortuous defective CNTs. Here, we present evidence that the density and alignment of self-organized CNT growth is limited by mechanical coupling among CNTs in contact, in combination with their diameter-dependent growth rates. This study is enabled by comprehensive X-ray characterization of the spatially and temporally-varying internal morphology of CNT forests. Based on this data, we model the time evolution and diameter-dependent scaling of the ensuing mechanical forces on catalyst nanoparticles during CNT growth, which arise from the mismatch between the collective lengthening rate of the forest and the diameter-dependent growth rates of individual CNTs. In addition to enabling self-organization of CNTs into forests, time-varying forces between CNTs in contact dictate the hierarchical tortuous morphology of CNT forests, and may be sufficient to influence the structural quality of CNTs. These forces reach a maximum that is coincident with the maximum density observed in our growth process, and are proportional to CNT diameter. Therefore, we propose that improved manufacturing strategies for self-organized CNTs should consider both chemical and mechanical effects. This may be especially necessary to achieve high density CNT forests with low defect density, such as for improved thermal interfaces and high-permeability membranes.

Understanding the collective growth of carbon nanotube (CNT) populations is key to tailoring their properties for many applications. During the initial stages of CNT growth by chemical vapor deposition, catalyst nanoparticle formation by thin-film dewetting and the subsequent CNT nucleation processes dictate the CNT diameter distribution, areal density, and alignment. Herein, we use in situ environmental transmission electron microscopy (E-TEM) to observe the catalyst annealing, growth, and deactivation stages for a population of CNTs grown from a thin-film catalyst. Complementary in situ electron diffraction and TEM imaging show that, during the annealing step in hydrogen, reduction of the iron oxide catalyst is concomitant with changes in the thin-film morphology; complete dewetting and the formation of a population of nanoparticles is only achieved upon the introduction of the carbon source, acetylene. The dewetting kinetics, i.e., the appearance of distinct nanoparticles, exhibits a sigmoidal (autocatalytic) curve with 95% of all nanoparticles appearing within 6 s. After nanoparticles form, they either nucleate CNTs or remain inactive, with incubation times measured to be as small as 3.5 s. Via E-TEM we also directly observe the crowding and self-alignment of CNTs after dewetting and nucleation. In addition, in situ electron energy loss spectroscopy reveals that the collective rate of carbon accumulation decays exponentially. We conclude that the kinetics of catalyst formation and CNT nucleation must both be addressed in order to achieve uniform and high CNT density, and their transient behavior may be a primary cause of the well-known nonuniform density of CNT forests.