Working face aggregation
From Wise Nano
Working Face Aggregation is a way of assembling large products from small blocks. It turns out that stacking small blocks, even sub-micron ones, to make a product can be a pretty fast way of assembling large products.
Several ways have been proposed to build a product from small blocks, including Convergent assembly. Working-face aggregation, as shown in Drexler's nanofactory animation/model (previewed at Foresight conference, not yet available), appears to be one of the simplest and fastest mechanisms.
Contents |
Block placement pattern
The blocks in the product would be turned at an angle relative to the working face. The working face would appear to be an array of pyramids--formed by previously-placed blocks--between pyramid-shaped holes. This means that new blocks can slide into their sockets (formed by previous blocks) with very coarse alignment.
The working face must be held while new blocks are placed. This means that not evey hole can be filled at once, since the blocks to fill the layer of holes consitute a solid layer joined edge-to-edge with no space for the face-holding arms. The simple solution is to place half the blocks in a layer, maintain the grip on the new blocks to hold the layer, release the original grips (on the sub-layer) and retract them, and place the second half of the new layer. With this method, at least half of all blocks at the face are held at any given time.
Starting a new object requires just a bit of planning, because the first layer of blocks has to have the correct faces gripped. Basically, the grippers should concentrate in alternate holes, leaving the other holes clear for insertion of second-layer blocks. One these second-layer blocks are placed and remain gripped, all the first-layer grippers can release, allowing the second layer to be completed. From there, construction can precede normally.
Speed and Scaling
As is usual in nanoscale engineering, what matters for throughput is not the absolute size of the thing being handled, but its size relative to the system handling it. In general, scaling laws say that operation speed increases linearly as size decreases. To move a 2-micron block from point A to point B takes twice as long as moving a 1-micron block from point A to point B (assuming the size of the handling machine is proportional to the size of the block moved).
So a column of 1-micron blocks could be built at the same linear rate as a column of 2- or 10-micron blocks. And of course, as the size of the operation shrinks, more of them can fit in the working face area. Between area and operation speed, the overall speed of assembly (meters or kilograms per minute of extruded product) should not change much over a wide range of block sizes. As long as the blocks don't shrink so small that the robot handling them is bigger than they are, then the blocks can be any size that's convenient for the designer. And with 1-nm features, it's certainly possible to build a handling robot that's smaller than a 200-nm cube--which is a convenient size for modular nanotech manufacturing.
Given a reasonable operation speed of 1 meter per second, it looks like a 200-nm block can be moved into place (over a distance of five times its length) in about a microsecond. If the fastening uses something fast and simple like ridge joints or direct chemical reactions, that won't add to the assembly time, since those will form automatically as the block is placed. (Even if a fastening operation is needed, its speed will presumably be able to scale along with the block size.)
If this is true, then a linear meter of these blocks could be assembled/extruded in about 5 seconds--across the entire area of the working face. Of course this is back-of-the-envelope, but 5 seconds is at least three orders of magnitude faster than it needs to be.
Reliability
Whenever massively parallel small machinery is used, reliability must be considered: even a tiny fraction of errors guarantees a large number. If the working face machinery occupies a one-micron by one-square-meter volume, it will contain up to 10^14 nanoblocks. We can expect a failure rate (from earth-surface background radiation) of 5x10^-10 per minute per nanoblock, so several nanoblocks will fail every second.
There are several ways to deal with this. One is to swap out the placement mechanisms when they fail. Build the systems modular, and include a second mechanism that physically removes the module and replaces it when failure is detected. This requires a large mechanism that may take some time to operate and needs space to be reserved just behind the working face for access to the placement mechanism modules.
A second way is to design redundancy into each placement mechanism, so that each component is redundant. This requires analyzing lots of failure modes, and might be hard to squeeze in; there won't be all that much room at the working face.
A third way is to work around the failure. Build the mechanisms with a little extra reach, so they can reach into their neighbor's spaces and perform their neighbor's functions. Then when a mechanism fails, its neighbors can do its work. One awkward situation is if it fails with a block gripped and partway extended. But the grippers can be made to pop loose (using Van der Waals force), so adjacent mechanisms could remove the block and place it, then collapse and stow the failed arm. Another awkward situation is if it fails in a way that causes malfunction rather than freezing, in which case it could tangle with its neighbors. This would have to be prevented by design.
The chance of three adjacent manipulators failing is well above 10^-21 (though not as high as 10^30), so all we have to worry about is two adjacent manipulators failing. This would put an extra 1/3 load on each of three neighbors for each failed manipulator, slowing down the system by 1/3.
Advantages
Assembly area is external to the nanofactory: the nanofactory can be quite thin/compact.
Long, skinny pieces can be built, because they only need to be supported on the growing end (unless they're so big and so unsupported that gravity causes a problem). This is a major advantage over convergent assembly.
Disadvantages
The working face will presumably be porous, so will have to be shielded from all contamination. This could be done by building a thin box with the product inside, with the cross-section of the box completely filling the working face. Before the box was removed, the top of the subsequent box would be assembled, keeping the working face capped. The box would have to be filled with argon gas to avoid collapse while maintaining cleanliness. For efficiency, the mechanosynthetic fabricators (if that's what's used to build the blocks) should work in vacuum rather than argon; air resistance costs a surprising amount of energy. But the pressure doesn't have to be zero, and squeezing the blocks through form-fitting tubes on their way to the working face should allow a pretty high vacuum to be maintained with not too much pumping.
References
Nanosystems 14.3.1a describes the general concept of working-face aggregation.
