Planar vs Convergent Assembly
From Wise Nano
Title: Planar Assembly: Building Large Products From Tiny Modules
Contents |
Introduction
Molecular manufacturing promises to build large quantities of product with a desktop nanomanufacturing appliance. This raises the possibility of building large products from very small components. (Sub-micron modules with nanoscale features and roughly similar shapes are called "nanoblocks.") Several methods have been proposed. This paper surveys them, compares two of the leading candidates in detail, and concludes that the winner--planar assembly--is quite simple, effective, and high-performance.
Background and History
Exponential Molecular Manufacturing
The ability to build nanoscale actuators, sensors, and signal processing suggests the ability to directly manipulate molecules and provoke chemical reactions as desired. This would allow construction of programmable molecular "parts" for engineered "machines." And if the machines could be directed to build duplicates of themselves, then productivity could increase rapidly. (A machine building a duplicate is not self-replication of the kind that can run amok, if it requires a stream of specialized inputs such as instructions or complex purified chemicals, or if the machine is too big to be mobile.)
The ribosome implements some of these functions; its actuation is quite intricate, but its control is rather simple: it is controlled entirely by a stream of instructions in the form of a strand of messenger RNA. The only sensing and processing it does is to detect when a reaction has happened and allow the process to step forward. Even this function is more in the form of a mechanical interlock than an explicit computation. A ribosome includes RNA as well as protein, and so cannot build a full ribosome.
Several concepts and architectures have been proposed for engineered nanoscale molecule-based manufacturing systems. Aside from the idea of reengineering bacteria (Bradbury), these have tended to use machine/engineering rather than biological/complex approaches.
To make useful quantities of product, any nanoscale fabrication system must be able to build more fabrication systems (possibly via a network of cooperating systems), and the systems must be able to work under automated control. This raises the question of throughput. In today's factories, the question is throughput vs. cost: will the factory take so much time to produce each piece of product that it adds too much to the product's cost? In nanoscale fabrication, the question is throughput vs. size.
The time for a fabrication system to process its own mass will be a function of two factors: the size of the system, and the rate of processing individual molecules. Carbonic anhydrase is an enzyme that can process 600,000 molecules per second. For machinery, scaling laws indicate that frequency of operation increases linearly with decreasing size, and the operation speed of a 100-nm mechanism should be about 1,000,000 (1E6) operations per second. A 200-nm cube contains approximately 1E9 atoms, so could in theory be built in less than an hour by a single fabrication mechanism. Since ribosomes are only 20-30 nm in diameter, and since engineered molecular features can be on the order of 1 nm in some chemistries, 8,000,000 cubic nanometers will probably be ample for a set of engineered fabrication machines that make up a desktop nanomanufacturing applicance (with external control).
It will require approximately 50 doublings to get from sub-micron to gram scale. If each doubling takes an hour, then gram scale fabrication can be achieved in a few days. Megagram (ton) scale fabrication would require only one more day.
Origins of the idea
When Feynman spoke in 1959 about "plenty of room at the bottom," he developed the concept of nanoscale manufacturing with atomic precision, and proposed running a billion tiny factories in parallel. However, he did not describe even a concept for combining the output of those factories. That would have to wait for a more concrete manufacturing proposal.
Starting in 1981, Drexler proposed nanoscale manufacturing systems in sufficient detail to intrigue several other designers. In particular, he tied together the concepts of atomically precise manufacturing via nanoscale machinery; nanoscale fabrication building additional nanoscale manufacturing systems, with exponential growth; and large heterogeneous products being made from the combined output of these systems. Several different architectures were proposed to build large products.
Scaffold Filling
In [Engines of Creation], Drexler proposed that small, self-contained fabrication systems ("assemblers"), with onboard computers and the ability to move and navigate, would join to form a large-scale scaffold. Once in position, they would each build a small piece of the product. During construction, the product would be intepenetrated by the scaffold of assemblers. When finished, the assemblers would exit the product. A feedstock-rich fluid would provide power and materials for the assemblers. Instructions would be transmitted between them once they linked up into the scaffold.
This method uses needlessly complex and inefficient devices. It would be difficult to fit all the required functionality into sufficiently small assemblers; if the assemblers are too big, then each one has to build too much of the product, and overall productivity drops.
Modular Robotics
Small robots, even sub-micron robots, might work together to implement larger structures and behaviors. Robots could be built by individual nanoscale fabrication systems and then combined after construction.
Hall has proposed "utility fog," with long-legged robots that combine to form trusses. Large shapes could deform and/or flow to reach new configurations. The mechanical performance of the system is not spectacular--the strength per volume and per density is comparable to wood or soft plastic.
Bishop and Michaels independently developed the concept of cubes that would fasten at their faces and slide past each other to form large-scale shapes. However, Michaels' cubes were built in multiple sizes, so were intended for top-down construction rather than nano-fabricator construction.
Mark Yim and others have built physical instantiations of modular robots with varying degrees of success. The biggest problem is that despite their apparent simplicity, they are still so complex that the MTBF is small (hint: something that needs to be fixed by making the mechanisms very robust).
The NIAC-funded work on Kinematic Cellular Automata examined the application of this paradigm for self-replication and made some compelling discoveries, most notably that the system could be less complex than a Pentium.
Convergent Assembly
In [Nanosystems] (1992), Drexler published a suggestion of Merkle's: that small parts and blocks could be fastened together more or less permanently within a factory. In convergent assembly the manufacturing system surrounds the product, just as in today's factories. (In contrast, in the scaffold method, the product grows around the manufacturing system.)
Drexler suggested that parts could be fastened together using alignment pegs and adhesive surfaces (Nanosystems 9.7). Even passivated surfaces will attach due to surface forces, and unterminated surfaces will bond; Drexler suggested that two unterminated diamond (110) surfaces might be made to bond seamlessly to each other. A variety of mechanical fastening strategies could also be used. Drexler provided projections of factory mass and performance for handling parts at the whole range of scales from nanometer to centimeter, and concluded that a 1-kg factory could manufacture 1 kg per hour of product.
Merkle wrote several papers in the 1990's further developing the convergent assembly concept and simplifying the factory layout. Phoenix published a lengthy paper in 2003 developing a detailed architecture for a convergent-assembly system: physical layout, performance, reliability, control, power, and mechanical and functional fastening of blocks.
Planar assembly
A few paragraphs of Nanosystems (Sec. 14.3.1a) suggested adding small modules to a planar working face. This approach was not pursued further until Drexler and Burch began to collaborate on an animated illustration of nanofactory function (2004). This approach has several advantages over convergent assembly, and appears to have few disadvantages. The next two sections will analyze the two methods.
Convergent Assembly
Convergent assembly is basically an extension of the assembly-line concept. Small parts are manufactured, then joined together to make larger parts. But whereas a modern factory might have four levels of assembly line from washers to washing machines, a nanofactory might have twenty levels--all completely automated.
Method
There is a range of approaches to part design. At one extreme, parts are made in unique shapes and proportions, and then fastened together. At the other extreme, modules are all made in the form of cubes or blocks with uniform shape and size, each composed of smaller cubical modules down to a level where a block can be built efficiently by a single fabrication system.
Advantages
If the fastening system does not require much manipulation, if scaling holds, and if part manipulation at the macro level doesn't take a long time, then convergent assembly can be quite fast. Each part may have to travel only a meter or two through the factory, and twenty stages of joining are enough to go from sub-micron to meter-scale. If the parts can simply be pushed together in order to join them (using ridge joints, a strong mechanical zero-insertion-force fastening system proposed by Phoenix), then the entire convergent assembly process may require only a few seconds.
Conversely, if robotics are provided to perform intricate manipulations, then a wide range of parts can be joined; this should still be fast, in most reasonable cases.
Constraints and Problems
Convergent assembly requires robotics in a wide range of scales. It also needs a large volume of space for the growing parts to move through. For example, in the Phoenix nanofactory architecture, the fabricators and computers occupy about 2% of the total volume, and 97% is empty space. <check this.>
Not only the robots, but their actions, must be designed at multiple scales. If the action consists only of pushing cubes together, this does not appear too difficult. But if more complicated operations are needed, such as joining odd-shaped pieces using small manipulated fasteners, then the problem may quickly become many times more difficult than designing a macro-scale fully automated ("lights out") factory. Each product--and each part of each product--would require its own sequence of operations, delaying the development of truly general-purpose manufacturing.
In the modular or block approach, the assembly robotics are simple but the design of the cubes is constrained. There are two problems. The first is that every large-scale component, mechanism, or device must be decomposable into cubes. If the component must be placed at a certain location relative to block boundaries, that places significant constraints on the product design. So unless its position within the product is constrained to fit neatly on a multiscale grid, the boundaries of large-scale cubes may cut the component at any positionl; this would mean that the design must be insensitive to being cleaved at any point.
The problem gets worse if large-scale blocks require large-scale fasteners, because these fasteners would also appear at semi-random places in the component volume. Another problem is that the faces of the blocks must be fairly stiff and reasonably complete (self-supporting) so that all blocks can be handled and joined into larger blocks. This implies that products, as manufactured, will be quite dense, unless voids can be filled with collabsible fine-grained scaffold. Dense products are not a good fit for the high strength promised by some chemistries (it would be wasteful of material); sparse products are not a good fit for a system where the factory encloses the product. In order to fulfill this constraint, products would probably need to inflate or unfold after manufacture. This requires additional design.
Planar assembly
The idea of planar assembly is to take small modules, all roughly the same size, and attach them to a planar work surface, adding to the working plane of the product under construction. In some ways, this is similar to the concept of 3D inkjet-style prototyping, except that there are billions of inkjets, and instead of ink droplets, each particle is molecularly precise and can be full of intricate machinery. Also, instead of being sprayed, they would be transported to the workpiece in precise and controlled trajectories, and the workpiece (including any subpieces) can be gripped at the growing face.
Method
Small modules supplied by any of a variety of fabrication technologies would be delivered to the assembly plane. The modules would all be of a size to be handled by a single scale of robotic placement machinery. This machinery would attach them to the face of a product being extruded from the assembly plane. The newly attached modules would be held in place until yet newer modules were attached. Thus, the entire face under construction serves as a "handle" for the growing product. If blocks are placed face-first, they will form tight parallel-walled holes, making it hard to place additional blocks; but if the blocks are placed corner-first, they will form pyramid-shaped holes for subsequent blocks to be placed into. Depending on fastening method, this may increase tolerance of imprecision and positional variance in placement.
The speed of this method is counterintuitive; one would expect that the speed of extrusion would decrease as the module size decreased. But in fact, the speed remains constant. For every factor of module size decrease, the number of placement mechanisms that can fit in an area increases as the square of that factor, and the operation speed increases by the same factor. These balance the factor-cubed increase in number of modules to be placed. This analysis breaks down if the modules are made small enough that the placement mechanism cannot scale down along with the modules. However, sub-micron kinematic systems are already being built via both MEMS and biochemistry, and robotics built by molecular manufacturing should be better. This indicates that sub-micron modules can be handled. See Nanoscale actuators for more information.
Advantages
This approach requires only one level of modularity from nanosystems to human-scale products, so it is simpler to design. Blocks (modules) built by a single fabrication system can be as complex as that system can be programmed to produce. Whether the feedstock producing system uses direct covalent deposition or guided self-assembly to build the nanoblocks, the programmable feature size will be sub-nanometer to a few nanometers. Since a single fabrication system can produce blocks larger than 100 nanometers, a fair amount of complexity (several motors and linkages, a sensor array, or a small CPU) could be included in a single module.
Programmable, or at least parameterized, (or at worst case, limited-type)modules would then be aggregated into large systems and "smart materials." Because of the molecular precision of the nanoblocks, and because of the inter-nanoblock connection, these large-scale and multi-scale components could be designed without having to worry about large-scale divisions and fasteners, which are a significant issue in the convergent assembly approach (and also in contemporary manufacturing).
Support of large structures will be much easier in planar assembly than in convergent assembly. In simplistic block-based convergent assembly, each structure (or cleaved subpart thereof) must be embedded in a block. This makes it impossible to build a long thin structure that is not supported along each segment of its length, at least by scaffolding.
In planar assembly, such a structure can be extruded and held at the base even if it is not held anywhere else along its length. The only constraint is the strength of the holding mechanism vs. the forces (vibration and gravity) acting on the system; these forces are proportional to the cube of size, and rapidly become negligible at smaller scales. In addition, the part that must be positioned most precisely--the assembly plane--is also the part that is held. Positional variance at the end of floppy structures usually will not matter, since nothing is being done there; in the rare cases where it is a problem, collapsible scaffords or guy wires can be used. (The temporary scaffolds used in 3D prototyping have to be removed after manufacture, so are not the best design for a fully automated system.)
This indicates that large open-work structures can be built with this method. Unfolding becomes much less of an issue when the product is allowed to have major gaps and dangling structures. The only limit on this is that extrusion speed is not improved by sparse structures, so low-density structures will take longer to build than if built using convergent assembly.
Surface assembly of sub-micron blocks places a major stage of product assembly in a very convenient realm of physics. Mass is not high enough to make inertia, gravity, or vibration a serious problem. (The mass of a one-micron cube is about a picogram, and under 100 G acceleration would experience a nanoNewton of force. This is comparable to the force required to detach 1 square nanometer of van der Waals adhesion (tensile strength 1 GPa, Nanosystems 9.7.1). Resonant frequencies will be on the order of MHz, which is easy to isolate/damp.) Stiffness, which scales adversely with size, is significantly better than at the nanoscale. Surface forces are also not a problem: large enough to be convenient for handling--instead of grippers, just put things in place and they will stick--but small enough that surfaces can easily be separated by machinery.
<insert picture>
(The problem posed by surface forces in MEMS manipulation is greatly exacerbated by the crudity of surfaces and actuation in current technology. Nanometer-scale actuators can easily modulate or supplement surface forces to allow convenient attachment and release.)
Sub-micron blocks are large enough to contain thousands or even millions of features: dozens to thousands of moving parts. But they are small enough to be built directly out of molecules, benefiting from the inherent precision of this approach as well as nanoscale properties including superlubricity. If blocks can be assembled from smaller parts, then block fabrication speed can improve.
Centimeter-scale products can benefit from the ability to directly build large-scale structures, as well as the fine-grained nature of the building blocks (note that a typical human cell is 10,000-20,000 nm wide). For most purposes, the building blocks can be thought of as a continuous smooth material. Partial blocks can be placed to make the surfaces smoother--molecularly smooth, except perhaps for joints and crystal atomic layer steps.
Constraints and Problems
Extrusion speed
The speed the product can be extruded at is a function of three factors: the time to deliver the block to the surface; the time to fasten the block to the surface; and the fractional area of the working surface where blocks can be installed in parallel.
The block delivery rate can be estimated from the time required to maneuver large blocks in conventional handling operations. A meter-scale palette can be moved by forklift at a rate of multiple centimeters per second, including lifting and placing. Conveyor belts handling uniform products can move at centimeters or even meters per second. Assuming the blocks are delivered by conveyor to a point within ten block widths of the product surface, flexible robotic handling at that point may work at a rate of 1 cm/s or better. Thus 1-cm blocks could be positioned in 10 seconds, 1-micron blocks could be positioned in 1 msec, and so on.
The time to fasten a block to a surface depends on the method of fastening. Any joint that requires only pressure, or is actuated via built-in mechanisms, may fasten in a very small fraction of the block placement time. A joint that requires external robotic manipulation may take many times the block placement time. Since several joints of the former class exist and appear useful at sub-micron scales, it is reasonable to assume that block fastening will not be a significant fraction of assembly time in a good design.
The surface must be held while blocks are being attached. This requires that blocks be maneuvered past the surface-holding mechanism. But the mechanism need not be as wide as a block, so if blocks form a grid pattern, then every other location can be held while blocks are placed into alternate locations. Thus it appears that the overall extrusion rate could be almost half the rate of placing single blocks. For all block sizes down to sub-micron, then, depositing a meter of blocks might take about an hour.
Modular design constraints
Although there is room for some variability in the size and shape of blocks, they will be constrained by the need to handle them with single-sized machinery. A multi-micron monolithic subsystem would not be buildable with this manufacturing system: it would have to be built in pieces and assembled by simple manipulation, preferably mere placement. Phoenix's "expanding ridge joint" system appears to work for both strong mechanical joints and a variety of functional joints.
Human-scale product features will be far too large to be bothered by sub-micron grain boundaries. Functions that benefit from miniaturization (due to scaling laws) can be built within a single block. Even at the micron scale, where these constraints may be most troublesome, the remaining design space is a vast improvement over what we can achieve today or through existing technology roadmaps.
Sliding motion over a curved unlubricated surface will not work well if the surface is composed of blocks with 90 degree corners, no matter how small they are. However, there are several approaches that can mitigate this problem. First, there is no requirement that all blocks be complete; the only requirement is that they contain enough surface to be handled by assembly robotics and joined to other blocks. Thus an approximation of a smooth curved surface with no projecting points can be assembled from prismatic partial-cubes, and a better approximation (marred only by joint lines and crystal steps) can be achieved if the fabrication method allows curves to be built. Hydrodynamic or molecular lubrication can be added after assembly; some lubricant molecules might be built into the block faces during fabrication, though this would probably have limited service life. Finally, in clean joints, nanoscale machinery attached to one large surface can serve as a standoff or actuator for another large surface, roughly equivalent to a forest of traction drives.
The grain scale may be large enough to affect some optical systems. In this case, joints like those between blocks can be built at regular intervals within the blocks, decreasing the lattice spacing and rendering it invisible to wave propagation.
Factory architecture
The factory would be in the form of a flat plate composed of several levels. The intake level would take in feedstock chemicals and manipulate them to form molecular parts. This would be in intimate contact with a feedstock/cooling fluid. This will be the most energy-intensive part of the operation, since it manipulates individual molecules; the rest of the layers can probably be cooled by conduction. The parts would then be formed into modular blocks, either in the fabrication layer or after transport. The blocks would be moved to a switching layer where they could be shuffled sideways; then they would be delivered to the output layer, which would grip the assembly face of the product and place new blocks.
<insert picture here>
The factory's fabrication speed will be limited by two things: first, the rate of placement of blocks, roughly a meter per hour; second, the rate of fabrication of blocks, in which a fabrication station might make its mass every hour. This means that to extrude a meter per hour of solid product, the fabrication layer of the factory would have to be a meter thick. Several considerations, including cooling and reliability, cast doubt on the workability of this.
However, a square-meter factory producing a kg per hour would require only a mm of fabrication layer containing just a few thousand sub-layers of fabrication stations. As blocks were produced, they would be passed to the location that needed them, so that the kilogram of blocks (>1E16 blocks, a large fraction of a liter) could be placed anywhere within a cubic-meter volume.
Feedstock and cooling fluid would flow through micron-scale channels between banks of fabrication stations. (Exact layout to be added; it's not a difficult engineering constraint.)
The extruded product would be protected by a thin covering or caul that would grow along with the product, to exclude contaminants from the assembly surface and the interior of the nanofactory. A virtually unlimited length of product could be extruded, making it possible for a small factory to make larger factories by growing them edgewise.
(Design for reliability to be addressed later.)
Conclusion and Further Work
It appears that surface assembly is a powerful approach to constructing meter-scale products from sub-micron blocks, which can themselves be built by individual fabrication systems implementing molecular manufacturing or directed self-assembly. Surface assembly appears to be competitive with, and in many cases preferable to, all previously explored systems for general-purpose manufacture of large products. It is hard to find an example of a useful device that could not be built with the technique, and the expected meter-per-hour extrusion rate means that even large products could be built in their final configuration (as opposed to folded).
The ability to build large sparse products that do not require unfolding is a significant advantage. Phoenix's 2003 "Nanofactory" architecure required unfolding. His paper handwaved about the ability to unfold a nanofactory, but even that was an especially easy case because the interior structure of his nanofactory was almost entirely rectilinear. The ability to build products with more complex internal structure was not well justified, though Phoenix had some preliminary ideas that he hoped could handle curves, elongated shapes, and changes in cross section (density). Surface assembly bypasses the folding problem entirely.
This leaves only the problems of how to build sub-micron placement robotics, how best to utilize large volumes of sub-micron blocks to implement high-performance human-scale products, and how to fabricate molecularly precise sub-micron functional block modules. The first problem--sub-micron robotics--should be fairly straightforward once sub-micron functional blocks are available with nanoscale features. The second problem--building large products out of programmable materials--should be accessible to today's engineering practice, given that the block size is smaller than today's engineering tolerances. (Reliability in the face of damage from ambient radiation is a new and substantial problem, but can be dealt with by fairly simple redundancy approaches, as explained elsewhere.)
The third problem, fabrication of nanoblocks, is beginning to be a focus of research. Blocks built by pure self-assembly may not be sufficiently configurable to make it easy to engineer a nanofactory or complex product. However, even limited actuation should allow the self-assembly process to be guided, programmed, or restricted so as to make a variety of blocks from one set of inputs. Once configurable functional blocks are available, it appears that manufacture of large products may require only the porting of today's design skills to a new volume-filling manufacturing system. The growing number of architectures and demonstrated nanoscale capabilities indicates that a solution may not be far off, at least for simple functions.
