Hm, now that I think of it, you might use an atom probe to shape the tip of the tip into a nice smooth hemisphere. That’s if the tip is conductive, of course.

Is it possible that one of the electron microscope detector technologies could be adapted for this purpose? Wikipedia has one sentence that indicates this may be the case. An integrated electron microscope / tip characterizer in a single tool might be useful.

Chris

If possible, would it be easier to place an AFM within a TEM? The idea being to attempt functionalize (somehow) an AFM tip, then lay the AFM tip down and use the SEM to see if the moeity is indeed on the AFM tip. This would obviate the need for an STM for this portion of the experiment.

The key idea is that IETS is used for *tip* characterisation. I have in mind a diamond and metal surface which are placed alongside each other: the metal surface is used as an appropriate substrate for IETS characterisation. Yes, this involves registration problems but, in principle, these are surmountable via closed loop technology.

IEST is painfully slow, however, and – prompted by discussions with other members of the sand pit – I’m thinking about other “pure force” mechanisms for chemical identification.

Chris, I hadn’t seen the paper to which you refer. I’ll try to find time to read it this evening (sandpit discussions/debate permitting!).

Bye for now,

Philip

http://www.newscientisttech.com/article/dn10911-nanoscopic-coaxial-cable-transmits-light.html

Put a buckytube in a coax sheath made of aluminum oxide and Cr or Al, and leave the tube sticking out the end, and it’ll gather light and transmit it up to 50 microns.

They talk about building an array of these things that gathers light at one end and puts electricity out the other. And speculating on quantum computer applications.

Chris

I thought IETS was only for metal surface experiments. It wouldn’t be a tool useful for hydrogen-passivated diamond experiments, would it? Maybe if the diamond was doped? I’d envisioned a mechanosynthesis experiment on diamond required an AFM enclosed within an electron microscope (with the unknown chemistry part of the mechanochemistry being the current showstopper).

Of course, there are other problems that anything nanoscale will face. Upper bounds do exist for machine performance. But in many cases, biology should be viewed as setting lower not upper bounds.

The below problem dimensions are mostly independent, implying a lot of different directions for possible improvement.

* Biological molecular machines rely on slow thermodynamic relaxation and diffusion processes for efficient operation. An engineered machine, especially if built with strong materials, could transfer stronger forces between machines and could use mechanical rather than entropic springs, balancing energy faster and in a smaller volume.
* Biological molecular machines work in the drag of water. It’s well established that some enzymes do not need water. It should be possible to engineer chemical reaction mediators that work in vacuum.
* All biological systems are limited to design spaces that are accessible to incremental blind improvement. Obviously, we don’t need to stick with that limitation.
* Biological organisms must metabolize complex and varying chemical inputs. We can provide refined chemicals.
* Organisms maintain internal state in the face of external perturbations via a series of complex feedback loops. Machines may maintain their state by being designed to be insensitive to perturbations. This may waste some matter or some energy, but that is OK for many engineering applications.
* Organisms must self-repair. Machines don’t have to. If they’re cheap enough to rebuild, they can just be replaced when they break. (This does not preclude recycling.) If they’re numerous and redundant/fault-tolerant and break only rarely, then breakages can be ignored for a useful product lifetime.
* Organisms must resist predators and parasites. Some machine-structure materials would be susceptible to attack by oxygen, water, or organisms, but in general machines can be protected by simple barriers.
* Organisms must grow from smaller to larger instances, from the inside out. Machines can be built externally in their finished “adult” form.
* Organisms must, with few exceptions, maintain the processes of life at all times. Machines, being simpler, can usually be frozen in place and restarted.

One thing I am thinking about ahead of next week is a set of grand challenges that if possible could change the world if realised. If we could make a perfect ‘matter-compiler’ then, for instance, we should be able to make a better version of photosystem 2 (or similar biological light harvester) that would fix CO2 and water, or produce electricity at a great efficiency. Can we precisely engineer the stiffness of materials down to the atomic level – maybe that way we could make highly efficient heat pumps. How about making self-cooling materials that work due to brownian motors that rotate only in one direction and have a dipole fixed to the centre of rotation. Line that up with a conducting nanowire can could that be a nano-battery. Nanoscale computers presumably would best be powered by smaller batteries? Could theory help us direct the software so we can really design materials with (semi-)predictable properties? (Would that be a bit like video killing the radio star though?)

Can we engineer self-assembling / dynamic and even evolving entities (using a combintorial library) and also interface this with some top down directed assembly – taking top down and bottom up assembly paradigms and puting them together is something I often here but we are short on routes. One of the exciting things about next week is that the collection of people coming seem to perfectly bridge that divide.

In my own work I self assemble architectures in solution on multiple length scales from molecules that are 1 nm in size to mm using the same building blocks (these range from pure organic to pure inorganic, metal oxide and metallic building blocks). I can change the stickyness of the building blocks to build in error correction and even deposite them on surfaces, assemble them in 3d, and increasingly am trying to direct their assembly at the inteface and measure their physical properties and spatial organisation. But there are big gaps here, I still suffer from the illusion of control – supramolecular chemistry is great at allowing the constructing of large molecular architectures but they often do not follow my design! There are also many new physical properties that we can introduce by changing the local environment, often by encapsulation. It is possible to end up with a Russian-Doll assembly of molecules within molecules – maybe this is one route to produce amazing new optical, magnetic, conducting devices etc.

The idea of coupling directed self assembly with pattern formation and trying to use self aggregating nanoparticles or clusters to build structures that can go from the nano to the mm is interesting. For instance can we interaface self-assembling / dynamic and even evolving entities (using a combintorial library) with some top down directed assembly – taking top down and bottom up assembly paradigms and puting them together is something I often here but we are short on routes.

One of the exciting things about next week is that the collection of people coming seem to perfectly bridge that divide. Cannot wait….

I suspect that familiarity with this kind of nanoscale system, in which “simple computational models (e.g. the Ising model) can give rise to complicated and rich dynamic behaviour” is what makes some scientists say the molecular manufacturing approach is “too mechanistic” or “too simplistic,” and then start looking for the complexity they assume we’re trying to sweep under the rug.

But there may be a deeper point. If the system’s behavior is accurately described by analog values and continuous equations, doesn’t that imply that it must involve relatively large numbers of atoms? And if there are far more atoms (more precisely, far more degrees of freedom) than can be directly specified by the precision of the input variables, doesn’t that imply that the output probably isn’t atom-precise? An exception could be if the atoms self-organize according to strong forces like ionic crystallization (e.g. salt crystals), but can that be compatible with rich dynamic behavior in self-assembly?

To relate this to the Ideas Factory’s purpose: Are there classes of analog-behaving system that, while they form interesting nanoscale structures, can be excluded from consideration as goals and/or methods because their analog nature must preclude atom-precision?

Chris

Granted you have to focus on things you can accomplish in 3-5 years, it wasn’t clear to me whether you intended to make nanoscale devices directly, or whether you might choose to spend the time making tools (perhaps including theoretical or computational tools) with which to make devices 6-10 years from now. Your “matter compiler” description–”output a macroscopic product in which every atom is precisely placed”–seemed to imply the latter. So I’ve been trying to answer the question, “Where are we going–What kinds of things should the matter compiler compile?”

In terms of what you can do in the next 1-2 years–The DNA structure people are doing amazing things, and could probably be assisted by a number of cross-disciplinary tools. For example, I’m told they can build 3D structures more easily than they can verify what they’ve built. Can LEAP be used to make 3D images of flash-frozen structures? Or, what about a microfluidic DNA synthesizer that could produce molecules by the thousands instead of billions and reduce materials costs? Perhaps a scanning-probe nanopipette (borrowed from cellular surgery) might help to reduce diffusion times for building meta-structures out of large self-assembled building blocks. I don’t know–I’m not a DNA person–but my strong impression is that giving them better tools will pay big dividends.

I am very much in favor of Philip Moriarty’s suggestion to study 3D diamond synthesis.

It might be useful to ask the question: How much information can be delivered to the nanoscale and embodied in a product, and how rapidly? DNA synthesis is around a kilobyte per hour; I guess temperature and pH change are about the same; scanning probe chemistry with cm-scale microscopes might be similar (Philip?). Being able to select one of a thousand materials is very good. But to those who are interested in function emerging from physical structure (including, I think, a lot of DNA people), a kilobyte of structure is quite small and limited, compared to what they’d really like to play with.

So I’d like to propose the goal of a kilobyte per minute inserted into atom-precise structures:

What modalities can deliver a kilobyte per minute to the nanoscale? I can think of four: Scanning beam (FIB, electron microscope), microfluidic injection of pre-made DNA strands, direct electrodes, and pulsed light (photons are hard to catch and even harder to store, but color-specific actuators feeding simple computational apparatus–e.g. nanoscale stepping motors–might be useful).

What nanoscale phenomena can latch or store a kilobyte per minute? A photoresist can do it. DNA binding *might* be able to do it in a microfluidic environment. With development, fast redox actuators might be able to, again if they were coupled to some kind of computational structure.

What nanoscale processes can be used for atom-precise fabrication? Can any of them be driven by incoming or stored information? Which of them have operating frequencies under fifty milliseconds? Displaced electrons or deformed molecules may be used to modulate chemistry. Bonus question: What externally-driven atom-precise fabrication processes can create 3D structures?

The set of techniques that passes these tests is small, but probably is not zero. Let me add two forward-looking questions that it may be too early to ask:

Could any nanoscale structures, produced by the previously selected technologies, be useful in enhancing the performance or convenience of high-information-rate fabrication? Do we have the luxury of making this a criterion in selecting which technologies to develop?

Is there any point in thinking about kilobyte per second fabrication?

Chris