The tangled beginnings of carbon nanotube growth

Vertically aligned carbon nanotubes (CNTs), grown by plasma-enhanced chemical vapor deposition (PECVD), are used for many applications such as flat displays or portable x-ray sources. They were first reported in 1998,¹ and extensive studies have been carried out since then.^2–4 These have shown that plasma has a dual effect on CNT growth. It facilitates growth by decomposing acetylene on the catalyst particles and tailors the CNTs by plasma etching. Recently, while studying plasma etching, we discovered that the CNTs evolve through three consecutive stages:⁵ randomly entangled I-CNTs at the beginning, followed by partially aligned II-CNTs, and then fully aligned III-CNTs. By examining the CNTs with electron microscopes, at various stages during PECVD growth, we saw that the randomly entangled I-CNTs were etched away and replaced by partially aligned II-CNTs. These could only survive for several minutes and eventually were completely replaced by fully aligned III-CNTs as the final product.

Usually, PECVD produces fully aligned CNTs.^1,2 In order to observe the two early-stage CNT growth patterns, we found that a growth time of less than 10min and a nickel catalyst thickness of about 5nm were necessary. Moreover, the partially aligned CNTs were grown on a 4in wafer. The macroscopic wafer is already very large in comparison with the CNTs, and so this suggests that the growth of the II-CNTs might be scalable. If so, II-CNTs in heat exchangers could facilitate the cooling of integrated circuits. Carbon has a high specific heat, and that, combined with a graphite-like lattice structure, gives carbon nanotubes thermal conductivities as high as 1000W/m/K. The smaller size, higher area density, and more regular lattice of the II-CNTs suggest that they could transfer heat better than III-CNTs.

In our experiment, we coated cleaned silicon substrates with a 10nm buffer layer of titanium and a 5nm nickel catalyst layer by sputtering. We placed the coated substrate on a heating stage, which functions as the cathode of the plasma. Above the sample, we fixed an anode (which was either roughly 6mm or 50mm in diameter in different trials), followed by standard CNT growth through PECVD.⁵ We then employed a scanning electron microscope and a transmission electron microscope, working with 5kV and 200kV sources, respectively, to characterize the CNTs.

Figure 1. Scanning electron microscope (SEM) images (a)–(c) and the corresponding transmission electron microscope images (d)–(f) showing the morphological evolution of CNTs during their growth from randomly entangled I-CNTs (a, d) to partially aligned II-CNTs (b, e) to fully aligned III-CNTs (c, f).

Figure 2. Digital camera images showing the partially aligned II-CNTs grown on a 4in silicon wafer: (a) Plasma during the CNT growth, in the plasma-enhanced chemical vapour deposition chamber, and (b) the CNTs grown on the whole surface of the wafer (inset SEM image showing the II-CNT length, at τ= 7min).

With growth time τ = 1min, Figure 1(a) and (d) shows the many rapidly generated and randomly entangled I-CNTs. However, these CNTs cannot grow further because the plasma etches them away. At this first stage (τ = 0–4min), the I-CNTs experienced fast growth for the first minute and were etched for the next three. By τ = 4min, most of the I-CNTs were removed. At the second stage, τ = 4–10min, Figure 1(b) and (e) shows that the partially aligned II-CNTs are very uniform in diameter (about 20nm). The inset displays the good graphitization of the multiwalled II-CNTs. The II-CNTs are also shown on the silicon wafer in Figure 2, after 7min of growth. However, starting from τ = 10min, plasma etching removed most of the II-CNTs. While they were etched away, the fully aligned III-CNTs, which have been extensively reported,^1–4 started to nucleate and grow. As shown in Figure 1(c) and (f), the III-CNTs align perpendicularly to the substrate with their catalyst particles facing the plasma ion bombardment. In this case, the nickel catalyst particles act as ‘safety helmets’ that protect the main body of the III-CNTs from plasma etching. Accordingly, the III-CNTs grow only along their axis lines. After more than 15min of growth, the II-CNTs are completely replaced with III-CNTs on the whole surface of the 4in silicon wafer.

In summary, multiwalled CNT growth proceeds through three stages due to the dual action of the plasma. It begins with the growth of randomly entangled I-CNTs , which are then etched away and replaced by partially aligned II-CNTs. Finally, fully aligned III-CNTs, protected by catalyst particles, grow vertically from the substrate while the II-CNTs are etched away. If II-CNTs can be mass-produced, they may be useful for cooling integrated circuits. In the future, we will study the PECVD mechanism further with the aim of growing even narrower, longer, higher area density, and more graphitized carbon nanotubes for better thermal conductivity.