University of Nottingham physicist Philip Moriarty is one of the few scientists who has been able to do extensive research into molecular mechanosynthesis. In 2004 Moriarty engaged in a debate with Chris Phoenix over the feasibility of molecular manufacturing. In 2008 Moriarty received a grant from the British Government to examine the viability of mechanosynthesis. In this Next Big Future interview with Sander Olson, Moriarty discusses the progress that has been made during the past decade, the challenges of working with diamond, and the prospects for building components out of silicon and diamond.
Question: You began the project for experimental work on molecular mechanosynthesis about five years ago. How is the project going?
Answer: The mechanosynthesis project has actually only been running for about 2.5 years http://gow.epsrc.ac.uk/ViewGrant.aspx?GrantRef=EP/G007837/1 now and the initial goal was to explore the possibility of atom-by-atom assembly on diamond surfaces , i.e. to test the viability of Drexler’s original vision of making components out of diamond. But as Drexler himself recently pointed out diamond is a very difficult material to work with. As a result, in Nottingham we have a parallel effort focused on silicon, which is much, much easier to work with than diamond. For example, we only very recently achieved atomic resolution using non-contact atomic force microscopy on a hydrogen-passivated diamond surface. Moving beyond imaging to atomic manipulation of the diamond surface is going to be much more challenging than for silicon.
Question: What makes diamond so difficult to use for mechanosynthesis?
Answer: It’s very easy to make clean silicon surfaces with wide atomically flat ‘terraces’ spanning tens (sometimes hundreds) of nanometers simply by heating up the sample in a vacuum. The preparation of diamond surfaces requires significantly more effort and results, even for the best samples, in rather narrow terrace widths and a relatively high density of adsorbates which are difficult to remove and can lead to the tip of the scanning probe microscope changing during a scan. Moreover, diamond is hard and the tip can easily get worn and blunted during scanning and manipulation. (We use tungsten tips which are generally coated with silicon for our experiments).
Question: So silicon could provide a proof of concept for diamond?
Answer: To a certain extent. About seven years ago I had a debate with http://www.softmachines.org/wordpress/?p=70 Chris Phoenix, who is a proponent of Drexlerian nanotech. Although the debate got rather heated (which I regret), it led directly to the mechanosynthesis grant proposal mentioned above. A key challenge I put to Chris was to use a scanning probe microscope to ‘grow’ atom-by-atom (or dimer-by-dimer) a row of dimers on either a silicon or a diamond surface. This is an objective we’re keenly focused on (for silicon) at the moment. We’ve recently shown that dimers on Si(100) can be manipulated, at the single bond level, using chemical force http://periodicvideos.blogspot.com/2011/03/joy-of-breakthrough-real-one.html alone. Our next set of experiments will focus on using the type of exchange reactions pioneered by Custance, Morita et al. to attempt the atom-by-atom growth of a layer of silicon on Si(100).
15 minute video of moving dimers
The videos show dimers being moved with mechanical force alone.
Question: If you succeed, will that prove the viability of molecular manufacturing?
Answer: It will prove the viability of key aspects of Drexler’s mechanosynthesis ‘machine language’ – i.e. single atom chemistry driven solely by mechanical force and employing pre-determined tip structures/tip tools. By the time I retire (~ 2040!), I’d really hope that we are at the point where we could simply instruct a computer to build nanostructures, and let the computer handle all the details – no human operator involvement required. But this is very far from proving the viability of molecular manufacturing. It’s also very far from the scenario that, for example, Ray Kurzweil put forward back in 2000: “Around 2030, we should be able to flood our brains with nanobots that can be turned off and on and which would function as “experience beamers” allowing us to experience the full range of other people’s sensory experiences…
Nanobots will also expand human intelligence by factors of thousands or millions. By 2030, nonbiological thinking will be trillions of times more powerful than biological thinking.”
Question: So you are still a skeptic of the concept of molecular manufacturing?
Answer: I am a skeptic. I believe that the concept of molecular manufacturing – of creating macroscopic objects atom by atom for any material, is flawed. I do not believe that this technique can be scaled-up to manufacture macrosized objects for arbitrary materials. In “Nanosystems” Drexler makes a careful and clever choice of the type of system required for mechanosynthesis/molecular manufacturing, taking into account the key surface science issues. I’ve never been able to see why it is then claimed that these schemes are extendable to all other materials (or practically all elements in the periodic table), for the reasons I discussed at considerable length in my debate with Chris Phoenix.
But I want to take this opportunity to give credit to Drexler. He has been the subject of a lot of criticism – some of it rather non-scientific and ad hominem– from what might be described as the ‘traditional’ (i.e. non-molecular manufacturing) nanoscience community. Drexler deserves significant kudos for the concept at the heart of the molecular manufacturing scheme; single atom chemistry driven purely by (chemo)mechanical forces is demonstrably valid. Richard Smalley, despite raising other important criticisms of the molecular manufacturing concept, misunderstood key aspects of mechanosynthesis and put forward flawed objections to the physical chemistry underlying Drexler’s proposals.
My misgivings arise, however, when a universal molecular manufacturing technology is extrapolated from mechanosynthesis, leading to claims that assemblers will be able to synthesize “virtually anything”. Again, I covered my objections to this in the debate with Chris.
Question: But can’t this process be massively scaled up, using large-numbers of computer controlled tooltips?
Answer: In principle, yes. But regardless of the degree of scale-up there are fundamental surface science issues associated with moving from the nanoscopic to even the microscopic scale. IBM invested a lot of time and effort into the concept of parallel probes with their ground-breaking millipede technology. But this involved surface manipulation via indentation into a polymer – it was very far from single atom manipulation.
It is also a massive challenge to obtain uniform tip properties and shapes – getting a tip that gives the ‘right type’ of atomic resolution is still very much a black art amongst scanning probe microscopists. We use a variety of ‘ad hoc’ methods (including crashing the tip into the surface) to coerce the tip into the state we require.
Automated tip optimization, including auto-recovery of a particular tip structure, is something that we’re pursuing in Nottingham at the moment via a collaboration with the computer science department here (Prof. Natalio Krasnogor’s group). Dr. Richard Woolley, a postdoctoral researcher in the Nottingham Nanoscience group, and Julian Stirling, a PhD student, are developing metrics and methods for real-time scanning probe image optimization via, for example, evolutionary algorithms. Our ultimate goal is to automate scanning probe microscopy to the point where a computer can image and manipulate atoms to a pre-defined blueprint, with no human operator involvement required.
Question: Are you using Atomic Force Microscopes (AFMs) or Scanning Tunneling Microscopes (STMs)?
Answer: We use two commercial scanning probe instruments for our mechanosynthesis-related work (from Omicron Nanotechnology and Createc). These are both what are called http://nanotechweb.org/cws/article/indepth/44336 qPlus non-contact atomic force microscopes. qPlus AFM, introduced by Franz Giessibl, uses stiff tuning fork sensors (virtually identical to those used as the timing element in quartz watches and clocks) to detect the force between a tip and a sample. The great advantage of qPlus AFM is that STM can be carried out in parallel (leaving aside some technical issues related to cross-talk of the force and tunnel current signals).
We generally use STM to ensure that the tip is giving stable atomic resolution before transitioning to qPlus AFM. All of our mechanosynthesis work, however, involves only qPlus AFM at 0 bias. This means that the atomic manipulation events are purely driven by chemical force and are not affected by tunneling electrons.
Question: How much have AFMs improved over the last decade?
Answer: There have been substantial improvements in AFMs over the past ten years, consisting mostly of numerous incremental steps forward in the sensing and manipulation technologies (e.g. atom tracking methods are becoming increasingly important in state-of-the-art NC-AFM work) . But for us, the adoption of tuning forks has been of huge importance. This reduces the oscillation of the tip down to the angstrom (or sub-angstrom) level, resulting in much higher sensitivity to short range chemical forces arising from covalent bond formation, and facilitating controlled atomic manipulation.
Question: What advances do you anticipate for microscopy for the next five years?
Answer: The key is the tip – scanning probe microscopists need to get a handle on creating tips that have well-defined structures. The IBM Zurich and Almaden groups routinely pick up single atoms or molecules to get a well-defined tip structure – and have done stunning work using CO-tips (e.g. Gross et al.’s paper on high resolution single molecule imaging in Science in 2009) but the problem with SPM is that the tip structure can spontaneously change, particularly following an atomic/molecular manipulation process. Automated optimization via autonomous computer control is of key importance in this regard – we need to remove the bottleneck in atomic-scale engineering due to uncertainties in the probe structure.
As you mentioned before, the other key issue is scale-up, getting thousands of tips working in parallel. I don’t plan on directly researching that because it’s a very difficult engineering problem and I would prefer to focus on the physics and chemistry of atom manipulation.
Question: Is your funding adequate?
Answer: Funding for the mechanosynthesis work is currently adequate, but I am frustrated and dismayed by the increasing emphasis that the UK research councils are placing on near-term commercialization. The research councils are now placing academics under a great deal of pressure to eschew long-term research in favor of research with short-term, application-driven (and near-market) goals. This is immensely frustrating since I and most researchers are in academia specifically to do long-term, curiosity-driven research.
Question: Apart from your lab, are any other Universities engaged in mechanosynthesis research?
Answer: Yes, but they probably wouldn’t call what they are doing mechanosynthesis, I guess. There has been beautiful recent work by Sugimoto and co-workers (including Custance, Morita, Perez, and Jelinek) where they have constructed atomic patterns using the interchange of atoms between a tip and a silicon surface. This strikes me as being very close to the spirit of mechanosynthesis. The IBM Zurich group (Leo Gross, Gerhard Meyer and co-workers) has made very important advances in terms of elucidating the role that the tip plays in scanning probe microscope image formation. There are also of course very many other SPM groups involved in atomic and molecular manipulation – far too many to list here. But nobody has yet gone beyond 2D or in-plane atomic manipulation using scanning probes. Our goal is to actually build a 3D nanostructure on an atom-by-atom basis.
Question: Zyvex is one of the few corporations actively working on mechanosynthesis. Are you collaborating with them?
Answer: John Randall from Zyvex visited Nottingham last year and we have a loose collaboration with Zyvex via John. He is very interested in the automatic probe optimization work we are pursuing. They are more bullish on massively scaling up the probes than I am. They are doing excellent work and I am interested in engaging in a more active collaboration with them.
Question: Chris Phoenix has argued that molecular manufacturing is probable by 2030. What is your take on such predictions?
Answer: Although I credit Chris with initiating a valuable debate with me, and for engaging in useful and productive research on his own, I see those claims as being, let’s say, rather optimistic! Our, and very many other scanning probe groups’, atomic manipulation work is currently carried out in conditions only a couple of a degrees above absolute zero, in a vacuum comparable to that found at the surface of the moon, and it is difficult, to put it mildly, to imagine a system that could work reliably and quickly at room temperature, with trillions of tips, being operational by 2030.
Question: How much progress do you anticipate in nanotechnology during the next decade?
Answer: In terms of our work here in Nottingham, by 2021, I’d like to be able to grow a 3D object using single atom manipulation, layer by layer. This might not seem a grandiose goal, but it is necessary in order to produce even nanoscale 3-D objects . I hope that the IBM millipede concept is further developed (if not by IBM then by another group) towards atomic manipulation using arrays of tips. That would be an exciting breakthrough. Outside of the atomic manipulation arena, I think that there will be significant advances in combining scanning probe microscopy with high temporal resolution measurements, along the lines of the work by Loth et al. late last year.
by Adam Sweetman, Sam Jarvis, Rosanna Danza, Joseph Bamidele, Subhashis Gangopadhyay, Gordon A. Shaw, Lev Kantorovich, and Philip Moriarty
Accepted Tuesday Feb 22, 2011
We reversibly switch the state of a bistable atom by direct mechanical manipulation of bond angle using a dynamic force microscope. Individual buckled dimers at the Si(100) surface are flipped via the formation of a single covalent bond, actuating the smallest conceivable in-plane toggle switch (two atoms) via chemical force alone. The response of a given dimer to a flip event depends critically on both the local and non-local environment of the target atom – an important consideration for future atomic scale fabrication strategies.