Mechanochemistry at the Single Bond Limit: Towards 'Deterministic Epitaxy'

Lead Research Organisation: King's College London
Department Name: Physics


Can we manipulate atoms just like we control bits of information in a computer? Could we ever build a matter compiler - a device that positions atoms, one by one, to construct a macroscopic product like a table, a computer, or even a building? In other words, could we ultimately push 3D printing all the way down to the atomic level?

This is the essence of the highly controversial "molecular manufacturing" concept put forward by Eric Drexler in the eighties, originally inspired by Richard Feynman's thoughts on the ultimate limits of miniaturisation back in the late fifties. Drexler's ideas were, and continue to be, widely critiqued and criticised by many (including the authors of this proposal) but at the core of his molecular manufacturing scheme is a demonstrably valid process: computer-controlled and atomically precise chemistry driven purely by mechanical force. This type of mechanochemistry is now implemented in the lab (and studied theoretically) by a small number of research groups across the world, including those involved in this proposal.

Our core objective is a little less grandiose than the fabrication of a macroscopic or, indeed, microscopic object using single atom manipulation. Nonetheless, it is an exceptionally challenging goal: the fabrication of a 3D object -- a nanoparticle -- an atom at a time. Although there are now many impressive examples of single atom control being used to form a variety of artificial structures at surfaces -- with IBM's recent "A Boy And His Atom" video, which has now amassed over 5M views, being a particularly elegant demonstration -- to date a 3D object has not been constructed. There are very good reasons for this; extending atomic manipulation and positioning to the third spatial dimension will involve a very different approach to interacting with atoms and molecules. Developing those protocols forms the core of our proposal.

It was the invention and subsequent application of a radically different type of microscope called the atomic force microscope (AFM) which enabled computer-controlled single atom mechanochemistry (of the type envisaged by Drexler) to be realised. The AFM is a microscope like no other -- it doesn't use lenses, mirrors, or any type of optical element to generate an image. Instead, an atomically sharp tip is brought close (within a few atomic diameters) to a surface. At this distance a number of important forces and interactions kick in, including, at the smallest separations, the formation of a chemical bond between the atom at the end of the tip and an atom directly underneath the probe. By scanning the tip back and forth across the surface whilst monitoring how the chemical force changes it's possible to build up an image of a surface with not only atomic, but single bond, resolution.

AFM is capable of a lot more than 'just' ultrahigh resolution imaging, however. The tip-sample force field can be mapped, the strength of single bonds measured, and, of key importance to this proposal, single atoms can be manipulated via chemomechanical force alone. Unlike its predecessor, the scanning tunnelling microscope, the AFM -- particularly the variant we use in our research, dynamic force microscopy (DFM) -- does not rely on the flow of an electrical current between tip and sample. With DFM, atoms can be moved through chemical force alone and this, along with the much higher sensitivity of DFM to the orientation and strength of single chemical bonds, has the potential to provide the exceptionally high levels of atomic-level control required to fabricate 3D nanostructures.


10 25 50
Description We have attempted a very detailed theoretical study of an ability of an AFM tip to vertically manipulate an atom on a crystal surface. Specifically, and probably for the first time, we attempted to consider a complete manipulation cycle comprising of two steps: (i) pick up an atom from the surface, and then (ii) place this atom back on the surface at a different location fully restoring the original tip ready to repeat this kind of manipulation cycle again.

First if all, we have developed special computational tools to perform this kind of study. These tools enable us to consider, semi-automatically, the potential energy surface of a tip at different lateral positions on the surface and at different heights. The peculiarity of our method is that we are not primarily interested in the 3D tip force calculations which became routine, but in understanding the ability of the tip to pick up or deposit an atom. This means that in our method, at each lateral position, we have to lower the tip to a particular height above the surface and then retract it recording any substantial changes in the surface (or tip) geometry. Then the tip is lowered more, and the retraction process is repeated. The whole process would include many such approach-retraction cycles, to different tip heights above the surface, with the lowest tip position, even possibly corresponding to a significant indentation of the tip into the surface. This way one can try to test various tip structures upon their ability to perform a complete detachment/attachment cycle so that the tip would perform as a 'crane'.

First of all, we found that detaching an atom from a terrace on this surface is energetically too costly and hence cannot be considered as a possible experimental strategy.

Then a very large number of tips have been considered consisting of Ga, As, and Au atoms and their combinations at the tip apecies, interacting with the GaAs (110) surface. The goal was to understand whether these tips can deposit one or more of their atoms on the surface. These calculations enabled us to construct a lateral map for each tip tried (more than 10 tip structures) in their ability to: (i) deposit their atom(s), and (ii) leave the surface structure minimally destroyed. Au tips were found to be too soft for the job, easily destroyable by the GaAs surface. However, for one of the tips we found that an exchange mechanism is in place when at a certain lateral position of the tip the apex Au atom is replaced by the surface As atom. GaAs based tips are stiffer and can deposit between 1 to 3 their apex atoms, and, depending on the level of indentation, may only slightly or considerably destroy the initial surface structure.

Then, we come across a set of simulations with one particular pyramid GaAs tip that is able to deposit just one Ga apex atom on the surface leaving it as an adatom. The rest of the blunt tip was found to be able to pick up that atom again at a slightly different lateral position fully restoring the original tip. These calculations essentially mean that this blunt tip can indeed act as a crane by picking up and then depositing the Ga adatom performing the complete vertical manipulation cycle with Ga adatoms, any number of times, repeatedly. Then, we find that the same tip can also pick up and deposit, repeatedly, As adatoms as well at different lateral positions above the surface. Hence, this work demonstrated, as a proof of principle, that it is possible to have a tip such that would be able to work as a crane on the GaAs (110) surface by picking up and depositing anywhere on the surface both Ga and As atoms.

Attempts have also been made to verify if this blunt tip is able of performing such repetitive deposition/puck-up cycles for other atoms, such as Au, Sb, In, Al and P. Surprisingly, we have found that, apart from P, Sb and In, it can!

Inspired by this breakthrough, we then considered, using the Nudged Elastic Band (NEB) method, the energy barriers for both parts of the cycle in each case (retraction/deposition) at different tip heights. These calculations should open up a new avenue in our future work related to studying kinetics of the vertical manipulation with an oscillating AFM tip that could be done using kinetic Monte Carlo (kMC) approach.

Encouraged by these exciting results, we started searching for other possible tips that would be able to perform similar job. We have found a GaAs tip with several Au atoms at its apex that are capable to pick up and then deposit (without changing the tip structure) a Au atom on the GaAs surface, also repeatedly. Even a deposition of two Au atoms next to each other was demonstrated.

Hence, our simulations proved that tips that can pick up and deposit a Ga, As, Al and Au atom on the GaAs surface do indeed exist, and hence experimental effort should be directed towards finding them. Our Nottingham partners are aware of that work and have started relevant experiments (they experienced some technical difficulties before, but now the situation has changed to the better).

This work was reported in an oral talk at NC AFM 2018 conference in Finland (17-21 September). We have prepared the first publication on these results that is theory only. However, it was decided to refrain from submitting this paper and wait for experimental results to support our findings as this way a higher impact of this work would be expected.

Unfortunately, the funding is now stoped and it is not longer possible to continue this work by Dr. Abbasi. Our plans for future work would be to see if immersing tips into clusters of atoms would give useful clues as to the "best" tip structures. Our further direction would also involve (i) simulating building of nano-structures, (ii) finding more tips, (iii) continue NEB simulations for picking up and depositing go atoms, and (iv), when this is done, performing kMC simulations of the vertical manipulation kinetics.
Exploitation Route We believe that our work may shed light on what is important in searching for the "best" AFM tips that are able to perform the complete retraction/deposition cycle. Our methodology can be easily adopted to other systems (tips and surfaces).
Sectors Aerospace, Defence and Marine,Chemicals,Education,Electronics,Energy,Environment,Pharmaceuticals and Medical Biotechnology