I recently posted on a surprising atomically precise fabrication process, and in a comment, Tom Craver remarked that
If a method like this could be combined with Single Layer Deposition , it seems like it’d be getting awfully close to allowing building 3D structures — embedded in a solid, so the next step would be figuring out how to “free” the structure.
This is a good idea, good enough, in fact, to be the objective of a serious corporate research project. Zyvex Labs aims to combine methods for atomically precise surface modification and for depositing single atomic layers, with the ultimate aim of building embedded, 3D objects. The embedded objects could indeed be exposed after fabrication, by selective etching.
The process they aim to develop starts with a hydrogen-passivated silicon surface, then removes selected H atoms to activate specific atomic sites. The following steps would perform atomic layer deposition of a different material on the activated sites, then complete the layer with silicon, repassivate the surface with hydrogen (on two materials, now), and repeat. There are a bunch of challenging problems in this, of course, since the process requires both going beyond standard homogeneous-surface atomic-layer deposition processes and achieving adequate performance in the atomically precise H desorption steps.
H-atom removal has been demonstrated, together with deposition of a different material on the on the activated site; other proof-of-concept experiments may have also have been done at this point. A modest but potentially useful objective would be to bury a precise pattern of dopant atoms for an electronic device, or perhaps phosphorous atoms for a quantum computing device [pdf]. There’s been some progress on this:
We demonstrate that it is possible to fabricate an atomically-precise linear array of single phosphorus bearing molecules on a silicon surface with the required dimensions for the fabrication of a silicon- based quantum computer.
Minimizing machinery for mechanosynthesis
The Zyvex approach is, by the way, a nice example of the generality of the principle of atomically precise patterning through mechanosynthesis. It achieves precise control through mechanically directed reactions, but the mechanical control is achieved by positioning what is, in effect, a catalyst (the tip), and not by directly guiding the motion of any of the reactants. The transport of the reactant molecules themselves occurs by diffusion, which is very convenient at this stage of the game.
By contrast, the process described in the post does position reactants directly, but the tip is recharged by diffusion (in this case, across the surface of the tip itself, a process that has some very nice characteristics). In other variations, molecules reach a tip by diffusion through a solvent, which also avoids the complexity of using mechanical means to load a tip. Positional control then occurs only where it is required: the surface of the workpiece.
Fully directed processes of the sort I describe in Nanosystems have advantages that will be worth pursing only when implementing nanomechanical systems becomes easy. Fully directed processes can be more efficient, and can increase the range of feasible reactants and product materials, but these advantages are secondary. The primary value is in atomically precise control of fabrication operations and the resulting ability to make intricate, atomically precise nanostructures and nanosystems.
I think that the most attractive pathway to these objectives passes through self assembly and progressive upgrades in soft-machine based mechanosynthesis; these are very different from atom-by-atom processes in UHV, both in implementation and in throughput.
{ 1 trackback }
{ 5 comments… read them below or add one }
Scott Jensen 03.24.09 at 4:19 pm UTC
Has Zyvex given a timeline when they think they’ll achieve their objective?
Andrew 03.25.09 at 5:15 pm UTC
Just came across this: a way to controll the AFM probe’s position in three dimensions to better than 40 picometers over 100 seconds:
controlled the probe’s position in three dimensions to better than 40 picometers (1 nanometer = 1000 picometers) over 100 seconds.
Andrew 03.25.09 at 5:15 pm UTC
Sorry, the link should be: http://esciencenews.com/articles/2009/03/25/making.a.point.picoscale.stability.a.room.temperature.afm
Eric Drexler 03.27.09 at 6:03 pm UTC
@ Scott Jensen — I haven’t heard of a timeline, and research of this sort has enough constraints and scientific unknowns to make predictions unwise, not only regarding timelines, but regarding feasibility.
Although timelines are always hard to predict, an attractive feature of self-assembly approaches is that they involve less discovery and more design. Zyvex is gambling — and it may prove to be a very good gamble — on the discovery of new chemical processes involving small molecules. Processes of this sort may be similar to one another, but each is substantially unique. Although experimentation and discovery can be guided by well-informed guesses there is no systematic way to design one.
In macromolecular engineering, by contrast, structures are made by combining many pieces in a systematic way, and this process creates a vast, combinatorial design space. This makes design possible, though supplemented with a lot of trial-and-error in the early days. If the operating principles and general structure of an intended product are sufficiently similar to those of known systems, there can be good reason to be confident that the design space contains many structures that would satisfy the product’s functional specifications. Confidence in feasibility still doesn’t provide a timeline, of course.
Eric Drexler 03.27.09 at 6:08 pm UTC
@ Andrew — That’s a nice piece of work, and a substantial advance in the state of the art.