Some Thoughts On Molecular Engineering 25 Apr 94

This essay is designed to explore current trends which may have a bearing on the future development of a new discipline which I term Molecular Engineering. I also suggest that the time may be ripe for the creation of an inter disciplinary research centre to explore the emerging techniques which may be applicable.

A considerable amount has been written on the possible emergence of Nano Technology - much of it by Eric Drexler (e.g. 'Engines of Creation'). Most practising scientists and engineers probably regard this as science fiction rather than a serious exploration of future trends. I must confess that I found it fascinating but rather far fetched when I first came across it. However, in the last few years I have done a lot of fairly wide ranging reading (I am not working because I suffer from M.E. - or C.F.S), and have been struck by how many new techniques and developments might be pertinent in the development of a new discipline which would design, develop and construct artefacts at the molecular scale.

There are several emerging disciplines which focus on the use of evolutionary methods to develop or 'design' new artefacts.:

By seeding, cloning and selecting DNA strands 'in the test tube', base sequences can be 'designed' to attach to a given target molecule. The momentum behind the Human Genome project has led to accelerated development of machinery for this kind of genetic material manipulation, and huge numbers of candidate sequences can be generated and selected in a short time frame (hours). As of 1992, this new technique was limited to designing genetic molecules, but it seems possible that it could be extended to design more general molecules.

(This could be viewed as both an 'Automated Design Tool' and a 'Volume Production Tool' for the Molecular Engineer).

The study of Artificial Life, (which has been largely driven by the Centre for Non Linear Studies at Los Alamos, with John Holland and Chris Langton being prime movers), is teaching us a lot about how to control evolutionary processes. It has a lot of potential in utilising computer based populations to construct entities which are successful according to specified criteria. There are several approaches to artificial life being explored, with differing goals; one of the most interesting recently has been the work of Thomas S. Ray currently at ATR Human Information Processing Research Laboratories in Japan (see 'Zen and the Art of Creating Life' posted in the alife section of the Cyber forum on CompuServe). Ray has succeeded in evolving a wide variety of software 'creatures' in a virtual machine environment with little or no outside 'help'.

The work on a-life has spawned a new discipline in genetic algorithms, which has already had some notable successes in constructing algorithms for traditionally difficult problems (e.g. The travelling salesman). Along with the general work on alife, this has led to a better understanding of evolutionary selection; for instance how to 'lock in' gains that have been achieved, yet still allow further mutation and selection to take place for further improvement. This work has still a long way to go (after all, natural evolution has several billion years lead on us!), and in my opinion is one of the most interesting avenues of current development in any discipline. The scope for utilising controlled evolutionary processes is almost limitless, and in particular, there is a good chance that it will lead to a whole new approach to the development of Artificial Intelligence, possibly along with the interesting work now showing promise in Neural Networks - and maybe even in 'bio chips' - growing 'neurones' onto silicon.

(Genetic Algorithms could lead to an 'Automated Design Tool' for the Molecular Engineer)

Once again, I suspect most readers will be suspicious of this topic, which often tends to appear in computer games show on TV. But VR is a new development of real importance (see 'Virtual Reality' by Howard Rheingold) and has many applications. I am particularly aware of the potential benefits because I worked for many years in Computer Aided Design, which has as one of its problems the modelling and display of physical engineering constructs. According to Rheingold, VR is being used (from memory at the University of North Carolina ?) to design molecules - especially targeted at the drug industry. By modelling molecules on the computer, and allowing users to see them in 3D, they can be manipulated to 'fit' onto other molecules - with the Van Der Waal forces being computed and fed back to the user via suitable mechanical hardware.

(This could be viewed as a 'Manual Design Tool' for the Molecular Engineer).

Over the last few years, great strides have been made in Tomography, mainly in health care. Software techniques have been improved, but also gains have come from the improved cost performance of today's micro processors. There still seems no slow down in the 'Moore's Law' speed/size gains driven by Intel, AMD, Cyrix, Motorola etc. (I remember being lectured to by a visiting expert from a US think tank when I was at Elliot Automation's Computing Research Lab on the limits imposed on computing by fundamental physics - I don't think he had it right then, and its still a very difficult call now - especially with people like David Deutsch talking about 'quantum computers'!). It seems inevitable that tomographic scanning will continue to improve in resolution; of course its a long way to being able to scan down to the molecular level - but I don't see it as conceptually impossible. Its a technology problem, and its being driven by the semiconductor boys, and they have a lot of money. (This could be viewed as an 'Analysis Tool' for the Molecular Engineer)

The current generation of scanning tunnelling microscopes, and atomic force microscopes allow us for the first time to 'see' matter at the atomic and molecular level. In addition, the techniques used in these devices are also applicable to atomic level manipulation, raising the possibility of 'constructing' molecules by manipulating atoms. I saw some of these devices in use at Cambridge University last year, and it is clear that the techniques are still in the early days of development, but with great potential none the less.

(This could be viewed as both a 'Prototype Constructing Tool' and a 'Quality Control Tool' for the Molecular Engineer)

There are other relevant developments that stem from the Semiconductor industry, which has been steadily reducing the dimensions of its chip fabrication lines (now sub micron), and finding other ways to define and construct very small structures:

Utilising individual quantum 'wells' and 'barriers', semiconductor technology can now produce structures which can isolate as few electrons as desired, producing 'designer atoms'. These fabrication techniques are still essentially planar, and so not suitable for general purpose molecular engineering. However, they are yet another example of how we are developing techniques for atomic level artefacts.

Differing developments in semiconductor technology have led to the ability to construct micron level beams, levers, bearings, and even electric motors. A technique known as LIGA utilises high energy x-rays to build micro structures several hundred microns thick. This method was developed in Germany, where Government funding is already going into an area termed 'Microsystems', combining micro mechanics and electronics. Japan is also putting funding into 'building small machines', with a micromachine technology project at MITI.

This is a general 'catch all' category, but in particular, I would mention the emergence of 'designer materials', the best example probably being carbon fibre. Work is also under way to try to produce much stronger materials by utilising carbon fibres in helical structures (see 'The New Science of Strong Materials' by J.E. Gordon). Other examples of new materials are ceramic 'whiskers', Kevlar and Memory metals. There has also been a trend towards producing new kinds of molecular structure, examples being Aerogels, and Carbon 60 and its relatives.

There are always potential new technologies clamouring for attention, so why should Molecular Engineering be singled out for special treatment? Indeed, it is true that, for example Bio Technology, Genetic Engineering and High Temperature Superconductors are all very promising avenues (and maybe even Cold Fusion?? - what a pity that term has been used, since what ever is happening in these cells, it probably is not 'classical' fusion. If it had been called 'super heat' instead, it might not arouse such fury among the physicists). However, it seems to me clear that the greatest potential of all is in engineering the very small - the sheer scope of application is immense - fabrication, construction on small and large scale, electronics and computing and hence also telecommunications, health, possibly transport, all could be revolutionised by successful Molecular Engineering. The difficulties are not small, but neither are they insuperable, and the benefits are incalculable.

According to Professor Colin Humphreys (Head of Department of Materials Science and Metallurgy at Cambridge), Japan is now putting more funding into materials science than any other area of technology. It is a discipline who's time has come, and will be a major influence on human socio economic systems in the next century; as we learn more about the molecular structure of materials it seems clear that a major part of materials science will be the study and control of materials at these very small scales.

Yes, these centres are now fashionable, but there are good reasons why they are. As science and technology become more specialised, with more and more niches, individuals who work in these niches have to spend longer and longer becoming expert in their fields. We are in danger of being swamped by the 'tall thin men', who have forgotten what is happening in the world outside of their narrow domain. Of course, this is a generalisation, but no less true for all that. It is therefore no accident that numerous interdisciplinary centres have sprung up in the last decade. I personally have long been convinced that many of the most interesting developments come at the boundaries of traditional domains. For sure, it is necessary that the inhabitants of those domains continue with their valuable work in pushing forward the boundaries of their specialities. But to break into new ground, it is often best to combine ideas and techniques from different areas. This is a technique well known to plant breeders, F1 hybrids are produced by cross fertilisation from isolated populations. When evolving populations are put under environmental stress, the genes capable of thriving under the new conditions are most likely to come from the 'genetic fringes' of the population.

Some of the emerging methods I have outlined above come from very different areas; it is likely that many of the experts in those areas are unaware of all the other ideas I have talked about. Yet many of these developments seem to be reaching in similar directions, and in particular they all seem to be potentially applicable to what I call Molecular Engineering. I believe the time is ripe for an initial exploration of this new field, carefully seeded with knowledge from these differing areas. I do not expect any sudden progress, I know that this is a long road to start on, but the rewards are commensurate.

I do not suppose that I am the only person to be thinking along these lines. Indeed in the Lent 94 issue of CAM Professor David King (Head of Chemistry dept at Cambridge) suggested that 'a (interdisciplinary) centre for materials synthesis would be very timely'; but possibly I have been able to gather together threads from quite a wide area to broaden the scope of such a development. Molecular Engineering is not going to be easy, and we need all the help we can get.