Powerful computer simulations have resolved the mechanism for the nanoscale assembly of the ‘hut’-like clusters that form after a few layers of atoms have been deposited on certain solid surfaces.
From the beautiful snowflakes that form from a random aggregation of water molecules, to the creation of a living organism, nature has found such efficient means of self-assembly that, in contrast, human techniques often seem crude. Even in our most impressive technologies for fabricating microstructures on surfaces — such as the lithographic techniques used to create integrated circuits — human efforts still seem like chiselling patterns out of stone. At the nanometre scale, the resolution that can be achieved using lithography is reaching its limit, and a new set of tools is needed. By better understanding nature's methods for assembly on solid surfaces, involving diffusion, nucleation and growth, it might be possible to orchestrate these phenomena such that a complete computer chip consisting of several billion transistors could assemble itself, like a complex biological organism. Indeed, for nanotechnology to become affordable, nanostructures will have to build themselves; normal manufacturing methods will be useless. The laws of physics do not preclude this possibility, but our present understanding of surface physics is still too shallow to achieve such complex self-organization and assembly.
Zhu and colleagues1, writing in Physical Review Letters, have quantified one of nature's mechanisms for creating nanometre-scale objects when atoms are deposited and grow into thin films on a solid surface of the same material (called thin-film homoepitaxy). Using computer simulations and a first-principles approach2, these authors model how atoms ‘rain’ down on to a solid surface and diffuse across it, by hopping between surface binding sites. The atoms may aggregate to form nuclei, which then grow into islands. Whether an island retains a two- or three-dimensional shape depends, in the conventional perspective, on how easily atoms can ‘step down’ from the top of the island to the lowest unfilled layer of the developing film.
The first atomic-level insight into this process came in 1966 from the elegant field-ion-microscopy studies led by Ehrlich3 and concurrent studies led by Schwoebel4. Ehrlich observed individual tungsten atoms hopping between binding sites on a tungsten surface and noted that when atoms approached downward step edges, they were often reflected away. This apparent repulsion was explained with bond-counting arguments: on flat crystallographic terraces, atoms have more bonds to other substrate atoms than they do if they are near the top of a step edge; there, the loss of substrate-atom bonds makes them less secure and hence they can be reflected from the edge. The energetic ‘unwillingness’ of atoms to descend step edges is now quantified by the Ehrlich–Schwoebel barrier.
On the basis of the same bond-counting arguments, homoepitaxial atoms residing at the bottom of a step edge of a close-packed surface should be the most secure, because they have a full complement of neighbouring atoms in the substrate below them and additional neighbours in the step edge beside them. It seems unlikely that these highly secure atoms would move upwards onto the tops of islands. However, evidence has been uncovered5 for such upward motion in experimental studies of aluminium homoepitaxy on its less close-packed (110) surface, using atomic force microscopy and low-energy electron diffraction. After about ten layers of atoms had been deposited, at a temperature of about 400 K, nanocrystalline ‘huts’ were seen to emerge rapidly from the substrate, growing to a height of about 50 nm (equivalent to more than 200 atomic layers) after the deposition of only another 20 layers (Fig. 1a). The emergence of these structures indicates significant upward motion of atoms.
To investigate this unusual occurrence, Zhu et al.1 used density-functional theory, which allows the quantification, with the best accuracy available, of the many-electron nature of chemical bonding at surfaces and how it dictates energy barriers, mechanisms and timescales for atomic motion. They show that on the (110) surfaces of a variety of metals (including aluminium), atoms ascend steps by incorporating themselves into the step edge and pushing a step-edge atom onto the top of the step (Fig. 1b). Interestingly, the energy barriers for this upward motion can be lower than barriers for downward motion.
If atoms prefer to move upwards, why do they wait until ten layers have been deposited before beginning to form these monumental structures? To understand the relationships between the atomic processes that lead to assembly, Zhu et al. incorporated their results using density-functional theory into a simulation of the statistical mechanics of the system, called an ‘ab initio kinetic Monte Carlo’ simulation2,6,7. The process of thin-film growth is simulated as a series of discrete events by choosing and actuating various atomic-scale processes (such as atomic deposition, atom hopping on a terrace, down a step, up a step, and so on) with probabilities based on their timescales.
Their simulations verify experimental observations: a few rough layers act as necessary precursors from which nanocrystals arise. The islands develop into small mounds whose sides form ‘minifacets’, or small surfaces with a different crystallographic structure from that of the (110) substrate. In this case, atoms race up the sides of the minifacets even faster than they step up single steps (such as those shown in Fig. 1b) and actuate the rapid rise of the huts.
Zhu et al.1 have unravelled one of nature's secrets for self-assembly. But there is still a long way to go to achieve true mastery of the art, such that magnetic memory devices, catalysts or integrated circuits can be fabricated simply by throwing atoms onto a surface and letting them organize themselves.
The reference citation in the second sentence of the figure legend is incorrect as published in print (reference 2 should be reference 5) but has been changed here.
Zhu, W., Buatier de Mongeot, F., Valbusa, U., Wang, E. G. & Zhang, Z. Phys. Rev. Lett. 92, 106102 (2004).
Ruggerone, P., Ratsch, C. & Scheffler, M. in Growth and Properties of Ultrathin Epitaxial Layers. The Chemical Physics of Solid Surfaces Vol. 8 (eds King, D. A. & Woodruff, D. P.) 490–544 (Elsevier, Amsterdam, 1997).
Ehrlich, G. & Hudda, F. G. J. Chem. Phys. 44, 1039–1049 (1966).
Schwoebel, R. & Shipsey, E. J. J. Appl. Phys. 37, 3682–3686 (1966).
Buatier de Mongeot, F. et al. Phys. Rev. Lett. 91, 016102 (2003).
Fichthorn, K. A., Merrick, M. L. & Scheffler, M. Appl. Phys. A 75, 17–23 (2002).
Kratzer, P. & Scheffler, M. Comput. Sci. Eng. 3, 16–25 (2001).
About this article
Hetero-diffusion of Au epitaxy on stepped Ag(110) surface: Study of the jump rate and diffusion coefficient
The European Physical Journal Plus (2017)
Frontiers of Physics (2015)
Computational Materials Science (2014)
Physical Review E (2013)