Enhanced CNT nucleation through ion bombardment: reactive MD simulations E. C. Neyts, A. Bogaerts Research group PLASMANT, Department of Chemistry, University of Antwerp, Antwerp, Belgium Abstract: Reactive molecular dynamics simulations demonstrate that carbon nanotube cap nucleation can be enhanced by ion bombardment in a limited energy window. The observed enhancement is explained by an ion induced carbon network restructuring mechanism. These results are of interest for plasma-assisted growth of carbon nanotubes. Keywords: molecular dynamics, carbon nanotubes, nucleation 1. Introduction Carbon nanotubes (CNTs) are hollow cylindrical structures consisting of a hexagonal carbon network. Their unique properties offer perspective on a plethora of applications [1]. They can for example be used as interconnects in silicon IC fabrication because of their high current carrying capacity (>107 A cm-2) or as heat sinks to dissipate heat from computer chips due to their very high thermal conductivity (> 3500 W m-1 K-1). Especially single walled carbon nanotubes (SWCNTs) offer perspective, since they can be either metallic or semiconducting (with a bandgap ranging from 0 eV to about 2 eV), depending on their chirality. They allow 1dimensional conduction and are envisaged for use in nanoscale electronics such as single electron transistors, as electron field emitters, for hydrogen storage, as actuators, chemical sensors, in super strong polymeric composite materials, etc. Furthermore, many properties of SWNTs are tunable to the required application, such as their photoelectrochemical activity by controlling their length. This offers opportunities in e.g. fabrication of efficient opto-electronic devices, nanotube optical detectors or emitters [2]. Currently, however, the applicability of CNTs is limited by a lack of control over their fundamental properties such as the diameter, length and chirality. Especially for electronic applications control over these properties is of crucial importance. Traditionally (begin ’90s), CNTs were produced by means of arc discharges and laser ablation [3]. Subsequently, formation of CNTs by thermal chemical vapor deposition (CVD) was developed, allowing for a lower production cost and possible large scale production. Since about 10 years, CNTs are also grown by means of plasma-enhanced chemical vapor deposition (PECVD) [4]. PECVD allows the formation process to occur at lower temperatures, which is beneficial for e.g. deposition on temperature sensitive polymeric substrates. Very recently, it has also become possible to generate SWCNTs in a PECVD process. Besides the lower growth temperature, PECVD for SWCNT growth has additional advantages over thermal CVD such as alignment of the SWCNTs during the growth (instead of forming spaghetti as in thermal CVD). Furthermore, using PECVD, freestanding SWCNTs can be produced, and most importantly, it offers perspective on controlling the chirality of the growing SWCNTs [5]. Because these are very recent developments, until now there has been very little basic research on the precise PECVD-based SWCNT growth process. To allow the effective application of SWCNTs in the micro-electronics industry, control over the growth process is very desirable. Especially precise control over the length, diameter and finally also the chirality is required to produce SWCNTs on a large scale and in a cost-effective manner, with specific properties for specific applications. This is, however, not possible with the current knowledge of, and control over, the exact growth process. By means of PECVD it is already possible (to a limited extent) to obtain specific tube lengths, as well as distributions in diameter and chirality. However, real control over the chirality has not yet been achieved. Furthermore, the underlying fundamental growth mechanisms for PECVD-based SWCNT growth are currently largely unknown. As shown in Figure 1, PECVD introduces a number of species not involved in thermal CVD growth, all of which have their specific influence on the growth process [6]. Fig. 1 Various species in PECVD of SWCNTs affect the growth process, which are not present in thermal CVD. Reproduced with permission from [6]. Indeed, various aspects of PECVD are crucial in the growth process. One of these is the electric field [7]. It is known that the electric field can align SWCNTs [4, 6, 7]. However, the exact role of the field on the growth mechanism is unknown. Furthermore, it is possible to grow SWCNTs in a PECVD-setup at lower temperatures than in thermal CVD. Most probably, the electric field plays a crucial role in this phenomenon, which is not yet elucidated [4, 6]. Another important factor is ion bombardment [8]. In contrast to thermal CVD, in PECVD plasma ions can bombard the substrate on which the SWCNTs are growing. This can lead to sputtering. Unwanted structures such as amorphous carbon can hence be sputtered. However, ion bombardment most probably also affects the growth of the SWCNTs. In this contribution, we investigate the influence of ion bombardment on the nucleation process of SWCNTs. 2. Methodology In a MD simulation, all atoms in the system are followed through space and time by integration of the equations of motion. Forces acting on the atoms are calculated as the negative gradient of the interatomic potential that describes all interatomic interactions. The interatomic potential that we used is the so-called ReaxFF potential [9], which is based on the bond distance / bond order relation on the one hand, and the bond order / bond energy relation on the other hand (Abell formalism). In contrast to nearly all other reactive potentials, ReaxFF also includes non-bonded interactions, including Van Der Waals and Coulomb interactions, and hydrogen bonds. This potential allows the simulation of chemical reactions with an accuracy that is comparable to or even better than PM3, while ReaxFF is about 100 times faster. In turn, PM3 is about 100x faster than quantum chemistry methods [9]. Note that while DFT calculations are even more accurate, they currently require prohibitively long calculation times to simulate SWNT growth within a reasonable time. The initial structure of the simulations was a previously simulated SWCNT cap structure, on the surface of a Ni40 nanocatalyst particle. The carboncarbon, carbon-nickel and nickel-nickel Reax parameters are those developed by Mueller et al. [10]. In previous simulations, we demonstrated that these parameters are sufficiently accurate to capture all the essential processes taking place during SWCNT growth, leading to the formation of SWCNTs with definable chirality [11, 12]. The initial structure is then bombarded by 200 sequential Ar ions in the energy range 5 – 50 eV. The argon-carbon and argon-nickel interactions are modeled through the Molière potential [13]. The temperature of the system was set to 1000 K, maintained by the Berendsen heat bath. The input structure is shown in Figure 2. Fig. 2 Input structure, consisting of a partial SWCNT cap on a Ni40 nanocluster, to be bombarded by Ar ions in the energy range 5 – 50 eV. 3. Results and discussion In Figure 3, we show the global effect of the ion bombardment of the SWCNT cap for three different energies: 10 eV, 20 eV and 50 eV. The top part of the figure shows the input structure in top and side views, whereas the bottom of the figure shows the structures obtained after the 200 consecutive impacts. Fig. 3 Resulting structures (bottom) as emerging from 200 consecutive Ar impacts with the indicated energies, starting from the structures shown at the top of the figure. It is clear from Figure 3, that ion bombardment at low energy, here exemplified for the 10 eV case, at least visually does not result in much noticeable result, i.e., the structure is not destructed, nor does it seem to be enlarged. At 50 eV, on the other hand, it is clear that the carbon network is entirely destructed, and no SWCNT cap remains. At this energy, also a number of Ni-atoms are actually sputtered. At an intermediate energy, however, as here exemplified for the 20 eV case, the carbon network is found to be enlarged due to the ion bombardment. It should be noted that in this simulation, no carbon atoms were added. Thus, the ion bombardment effectively enhances the SWCNT cap nucleation. This result is indeed validated by dedicated experiments [8]. A more quantifiable measure is the evolution of the number of rings, or equivalently, the change in the number of rings relative to the original input structure. This is shown in Figure 4. It can be seen in the figure that while visually the structure does not change very much at low ion impact energies, the number of graphitic rings does increase, even at only 5 eV. Thus, even at such low ion energies, ion bombardment seems to be beneficial for the nucleation of the SWCNT cap. The figure shows that at 5 eV, the increase in the number of rings is smooth up to about 120 impacts (corresponding to an increase of about 15-20%), after which there is no further change in the number of rings relative to the input structure. % change 60 in num 40 ber of 20 ring s 0 0 -20 50 100 150 Impacts is limited. In this stage, the network is damaged by the ion impacts, albeit without destroying it. Only after these first 70 impacts, there is a strong decrease in the number of rings. The destruction of the network at these energies is directly related to the carbon displacement energy. Indeed, the carbon displacement energy in graphene and CNTs is estimated in the range 14-32 eV [14-17]. Given that the kinetic mass transfer factor for Ar-C collisions is 71%, an impact energy of 40 eV always has the capability of sputtering a carbon atom from the network. Note that also lower ion energy can suffice to remove carbon atoms from the network, as many carbon atoms are locally not coordinated in a perfect graphene network, and thus have a reduced displacement energy. From these observations, we can conclude that both at low and intermediate ion energies, the nucleation process is beneficially influenced by the ion bombardment. As an example of how the network grows (i.e., how the number of graphitic rings increases) due to the ion bombardment at intermediate energies, we show in Figure 5 the evolution of the network. 200 -40 -60 Fig. 4 Evolution of the procentual change in the number of graphitic rings as a function of the number of impacts for ion impact energies of 5 eV (dotted line), 15 eV (solid line) and 40 eV (dashed line). Fig. 5 Evolution of the carbon network due to the ion bombardment. The total number of rings in the patch is indicated, as well as the number of graphitic (i.e., pentagons, hexagons and heptagons) rings (values between brackets). The nickel atoms are not shown for clarity. Increasing the ion impact energy to 15 eV, however, clearly has a much more profound effect. Indeed, while at 5 eV, the procentual increase in the number of rings is about 10-15%, this increase amounts to almost 50% at 15 eV. This results in the enlargement of the carbon network as shown in Figure 1 at intermediate energies. It can also be seen that at this energy, the increase in the number of rings occurs in two steps: a first steep increase during the first 30 impacts (up to a relative change of 20%), and then a second steep increase between 60 and 100 impacts (up to a relative increase of 50%). After about 100 impacts, the relative increase in the number of rings does not further change. Increasing the energy further to 40 eV, it is clear from Figure 4 that the extent of the carbon network is now decreasing, with a clear decrease in the number of graphitic rings. At 40 eV, this decrease amounts to 50%. Up to about 70 impacts, the change in the number of rings Figure 5 shows that in a consecutive manner, the carbon network grows. Recall that in this simulation, no carbon atoms are added to the structure. Thus, the network enlargement is entirely due to the ion impacts. It should be noted that the effect of ion impacts (i.e., adding energy to the structure through momentum transfer) is not similar to simple thermal heating. Indeed, while in thermal heating energy is delivered to the entire structure (i.e., nickel cluster and carbon network), the ion bombardment delivers most of its energy to the directly targeted atoms (i.e., the carbon network). This energy is dissipated to all atoms, but the bonds that are most affected are those that are directly connected to the targeted atoms. Note, however, that the underlying bond switching dynamics in Ni and Ni/C clusters are very fast in any case, certainly at elevated temperatures [18]. Second, since essentially the carbon network is heated most by the ion bombardment, it is not impossible that the ion bombardment delivers the energy needed to enlarge the network and enhance the nucleation, while the nickel nanoparticle remains at fairly low temperature. 4. Conclusion Reactive molecular dynamics simulations were used to investigate the effect of Ar ion bombardment on the nucleation of SWCNTs. It was found that the ion bombardment does not have much of an influence at low impact energies (below 10 eV), but is beneficial for the nucleation in a medium energy range (10 – 25 eV). In this energy range, the ion bombardment leads to a carbon network restructuring, forming new graphitic rings. At higher energies (above 25 eV), carbon atoms can be sputtered from the network, and the ion bombardment becomes detrimental for the nucleation and growth. These results demonstrate that ion bombardment is not necessarily detrimental (as was previously suggested), and possibly open up the road to a more controlled growth process, e.g., by PE-CVD. 5. Acknowledgments The authors would like to thank Ken Ostrikov for the many interesting discussions and Adri van Duin for providing the ReaxFF model. The authors also gratefully acknowledge financial support from the Prime Minister’s Office through IAP VI. This work was carried out in part using the Turing HPC infrastructure at the CalcUA core facility of the Universiteit Antwerpen, a division of the Flemish Supercomputer Center VSC, funded by the Hercules Foundation, the Flemish government (department EWI), and the Universiteit Antwerpen. 6. References [1] R. H. Baughman, A. A. Zahkidov, and W. A. de Heer, Science, 297, 787 (2002). [2] N. M. Gabor, Z. Zhong, K. Bosnick, J. Park, P. L. McEuen, Science, 325, 1367 (2009). [3] A. Thess, et al., Science, 273, 483 (1996). [4] M. Meyyappan, J. Phys. D: Appl. Phys., 42, 213001 (2009). [5] T. Kato, S. Kuroda, R. Hatakeyama, J. Nanomater., 2011, 490529 (2011). [6] E. C. Neyts, J. Vac. Sci. Technol. B, 30, 030803 (2012). [7] E. C. Neyts, A. C. T. van Duin, A. Bogaerts, J. Am. Chem. Soc., 134, 1256 (2012). [8] E. C. Neyts, K. Ostrikov, Z. J. Han, S. Kumar, A. C. T. van Duin, A. Bogaerts, Phys. Rev. Lett., 110, 065501 (2013). [9] A. C. T. van Duin, S. Dasgupta, F. Lorant, and W. A. Goddard III, J. Phys. Chem. A, 105, 9396 (2001). [10] J. E. Mueller, A. C. T. van Duin, W. A. Goddard III., J. Phys. Chem. C, 114, 5675 (2010). [11] E. C. Neyts, Y. Shibuta, A. C. T. van Duin, A. Bogaerts, ACS Nano, 4, 6665 (2010). [12] E. C. Neyts, A. C. T. van Duin, A. Bogaerts, J. Am. Chem. Soc., 133, 17225 (2011). [13] R. Smith, in Atomic and Ion Collisions in Solids and at surfaces (Cambridge University Press, Cambridge, England, (2005), p. 46 [14] O. Lehtinen, J. Kotakoski, A.V. Krasheninnikov, A. Tolvanen, K. Nordlund, J. Keinonen, Phys. Rev. B, 81, 153401 (2010). [15] G. L. Montet and G. E. Myers, Carbon 9, 179 (1971). [16] F. Banhart, Rep. Prog. Phys., 62, 1181 (1999). [17] A.V. Krasheninnikov, F. Banhart, J. X. Li, A. S. Foster, R. M. Nieminen, Phys. Rev. B, 72, 125428 (2005). [18] E. Neyts, Y. Shibuta, A. Bogaerts, Chem. Phys. Lett., 488, 202 (2010).
© Copyright 2026 Paperzz