Introduction

Since the first demonstration of producing graphene layers by mechanical exfoliation of bulk graphite1, many growth or synthesis approaches for producing few-layer graphene have been investigated, including chemical vapour deposition (CVD) on metal catalysts2,3,4,5,6,7,8, reduced graphene oxides9, thermal decomposition of SiC10,11 and molecular beam epitaxy12,13,14. However, in comparison with Si or III–V semiconductors, it is extremely difficult to control the growth orientation of the graphene layer, and hence its assembly in an electronic device is believed to be problematic. Direct visualization of how the graphene network evolves during growth is highly desired in an effort to precisely understand the atomic processes of the growth mechanism. Furthermore, to realize well-controlled graphene nanodevices, identifying the orientation relationship between the seed crystal and the growing layer or detecting the influence of the catalytic atoms during growth would be of immense benefit. The extremely high CVD growth rate (on the order of micrometres per minute) makes atomic resolution analysis during this method unfeasible.

The edge structures of graphene ribbons are of fundamental importance during growth and preparation processes because the electronic, optical, magnetic, mechanical and electrochemical properties depend not only on the ribbon width but also on the edge structures15,16,17,18,19,20,21,22,23,24,25. Graphene nanoribbons generally have edge structures that are a combination of armchair (AC) and zigzag (ZZ) geometries26,27, and recent theoretical calculations and experimental studies have also demonstrated the existence of an alternate pentagon–heptagon edge structure (ZZ (57)) reconstructed from a periodic hexagon ZZ structure20,28,29,30,31,32,33. Graphene nanoribbons with AC and ZZ edges have distinct electronic transport properties34, but atomic-scale edge defects can significantly affect the electronic properties35. Pristine graphene edge structures are of particular importance as they govern the quality of the subsequent synthesized material. Low-energy electron microscopy, photoemission electron microscopy and low-energy electron diffraction can give unique insight into the fundamental growth mechanisms of graphene on metal substrates36,37. Invaluable insight into the atomic structure of graphene including defects and edge stabilities has been provided by scanning tunnelling microscopy38,39 and transmission electron microscopy (TEM), especially aberration-corrected high-resolution TEM33,40,41,42,43,44,45,

Figure 5: Step-edge growth rate.
figure 5

(a) Comparison between high-speed growth (CVD) and low-speed growth in this study. In this study, between 500 and 700 °C, graphene grows from step-edge. Three different growth rates at different residual hydrocarbon gas pressures are shown. The growth rates at 2.9 × 10−5 Pa are shown by circles and the growth rates at 1.2 × 10−5 Pa are shown by rectangles. When T<500 °C, amorphous carbons appear as contaminations. If T>700 °C, carbon atoms at step-edges are etched and no more graphene growth is found. (b) The growth speed measurements on different selected areas repeated five times at each residual hydrocarbon gas pressure for 1.2 × 10−5 and 2.9 × 10−5 Pa.

Full size image

We know that the step-edge in-plane growth seen here is not just simple physical absorption of C atoms because single Si catalyst atoms are present at the bilayer edges. The electron beam also plays an important role in the growth by accumulating hydrocarbon gas at the scanned area; there is no growth in areas far from the scanned area as shown in Supplementary Fig. 7. It is well known that graphene growth is strongly dependent on the number of available C atoms that are able to bond within a certain time, that is, if a large number of C atoms encounter graphene edges and do not have enough time to locate an energetically optimal location before forming covalent bonds, then the graphene will just be contaminated with amorphous C. This is the case when graphene is observed in the STEM at RT and low-vacuum conditions. However, if the substrate graphene is heated to 500–700 °C, then the incident C atoms are more likely to form sp2 bonds with the step-edge atoms, and in-plane graphene growth will occur. At temperatures above 700 °C, the etching effect is dominant, and no growth was observed (Fig. 5). As shown in Fig. 5, the growth rates found in this study were very low (several angstroms per minute) in comparison with that of CVD20 (several micrometres per minute). Even though the vacuum conditions and gas source pressure were different from those in CVD, we were still able to successfully induce graphene growth and observe the growth at atomic resolutions.

Such in situ growth could be used to precisely design the orientation of graphene or other heterobilayer materials. When the pristine seed edges are terminated with ZZ or AC edges, the orientation of new growth area will remain the same; when the seed edges are terminated with some 5–7 defects, the new growth area will rotate by 30° with respect to the original layer. If the seed edges can be modified to terminate with other kinds of defects, rotation angle between 0° and 30° will be achieved. Since the growth speeds are dependent on the residual hydrocarbon gas pressures, observations by using environmental electron microscope with controlled carbon source are expected to offer more important information and attract attention to techniques controlling the growth of graphene or similar heterobilayer materials.