for the technologist is to find just the right variation of process conditions—for example, the changes in temperature or the addition of impurities—that result in the desired material properties. Therefore, molecular self-assembly is certainly feasible for the manufacture of materials.
Simple devices such as sensors for medical diagnostics are built every day with the aid of processes that exemplify molecular self-assembly. More complex structures can be generated by more sophisticated self-assembly processes. Processes requiring dynamic steering of process variables are often called “directed” self-assembly. “Templated” self-assembly describes processes requiring control of spatial boundaries such as container material and geometry. Thus, molecular self-assembly is also feasible for the manufacture of devices.
As spatial and temporal variations of boundary conditions and process variables become more complex, the emphasis shifts from self-assembly to the flow of information in the control system. However, the committee could not identify a “bright line” distinction between self-assembly and more complex integrated manufacturing processes. For instance, the above-mentioned example of the self-assembly of the bacterial ribosome from its constituent proteins is an elegant biological phenomenon, but it is only one part of the complex process that has evolved to build the ribosome. The various constituent proteins are themselves the product of RNA-driven amino acid catalysis called RNA translation in other functioning ribosomes, and RNA molecules are, in turn, the product of another catalytic process called DNA transcription. This complex assembly process, proceeding in every living cell, involves more than just self-assembly.
Manufacturing processes that can build very complex objects with high yield and repeatability will generally include processes more complex than simple self-assembly. This statement follows primarily from the fact that simple self-assembly does not include a mechanism for error correction.3 The error rate for assembly of any two constituent parts can often be arranged to be very low, but the total probability of any error will tend to be the sum of the error rates for assembly of all the individual parts. Thus, the probability of a critical error occurring at some point in the assembly process will increase with the complexity of the system and the number of parts that must interoperate. At some level of complexity, the yield of a simple self-assembly process will become negligible.
Practical manufacturing systems solve this problem in a number of ways.