Towards routine manufacture at the nanoscale
NEMS must overcome a final important hurdle before nanoscale machines, sensors and electronics emerge from industrial production lines. Put simply, when we combine state-of-the-art processes from two disparate fields - nanolithography and MEMS micromachining - we increase the chances that something will go awry during manufacturing. Fortunately, sustained and careful work is beginning to solve these problems and is revealing the way to build robust, reliable NEMS. Given the remarkable success of microelectronics, it seems clear that such current troubles will ultimately become only of historical significance.
But there is a special class of difficulties unique to NEMS that cannot be so easily dismissed. NEMS can respond to masses approaching the level of single atoms or molecules. However, this sensitivity is a double-edged sword. On the one hand it offers major advances in mass spectrometry; but it can also make device reproducibility troublesome, even elusive. For example, at Caltech we have found that it places extremely stringent requirements on the cleanliness and precision of nanofabrication techniques.
Some applications of NEMS
Ultimately, NEMS could be used across a broad range of applications. At Caltech we have used NEMS for metrology and fundamental science, detecting charges by mechanical methods and in thermal transport studies on the nanoscale (see Schwab et al. in further reading). In addition, a number of NEMS applications are being pursued that might hold immense technological promise.
In my opinion, most prominent among these is magnetic resonance force microscopy (MRFM). Nuclear magnetic resonance was first observed in 1946 by Edward Purcell, Felix Bloch and their collaborators, and is now routinely used for medical imaging. The technique exploits the fact that most nuclei have an intrinsic magnetic moment or "spin" that can interact with an applied magnetic field. However, it takes about 1014-1016 nuclei to generate a measurable signal. This limits the resolution that can be attained in state-of-the-art magnetic resonance imaging (MRI) research laboratories to about 10 µm. Meanwhile, the typical resolution achievable in hospitals is about 1 mm.
One would assume then that the detection of individual atoms using MRI is only a distant dream. However, in 1991 John Sidles of the University of Washington at Seattle proposed that mechanical detection methods could lead to nuclear magnetic resonance spectrometry that would be sensitive to the spin of a single proton. Achieving this degree of sensitivity would be a truly revolutionary advance, allowing, for example, individual biomolecules to be imaged with atomic-scale resolution in three dimensions.
Magnetic resonance force microscopy (MRFM) could thus have an enormous impact on many fields, ranging from molecular biology to materials science. The technique was first demonstrated in 1992 by Dan Rugar and co-workers at IBM's Almaden Research Center, and was later confirmed by Chris Hammel at the Los Alamos National Lab in collaboration with my group at Caltech, and others.
Like conventional magnetic resonance, MRFM uses a uniform radio-frequency field to excite the spins into resonance. A nanomagnet provides a magnetic field that varies so strongly in space that the nuclear-resonance condition is satisfied only within a small volume, which is about the size of atom. This magnet also interacts with the resonant nuclear spins to generate a tiny "back action" force that causes the cantilever on which the nanomagnet is mounted to vibrate. For a single resonant nucleus, the size of this force is a few attonewtons (10-18 N) at the most. Nonetheless, Thomas Kenny's group at Stanford, in collaboration with Rugar's group at IBM, has demonstrated that such minute forces are measurable.
By scanning the tip over a surface, a 3-D map of the relative positions of resonating atoms can be created. Although Rugar and co-workers detected a signal from some 1013 protons in their early experiments, the sensitivity still exceeded that of conventional MRI methods.
In another area of research, Clark Nguyen and co-workers at the University of Michigan are beginning to demonstrate completely mechanical components for processing radio-frequency signals.
With the advent of NEMS, several groups are investigating fast logic gates, switches and even computers that are entirely mechanical. The idea is not new. Charles Babbage designed the first mechanical computer in the 1820s, which is viewed as the forerunner to the modern computer. His ideas were abandoned in the 1960s when the speed of nanosecond electronic logic gates and integrated circuits vastly outperformed moving elements. But now that NEMS can move on timescales of a nanosecond or less, the established dogma of the digital electronic age needs careful re-examination.
To the quantum limit - and beyond
The ultimate limit for nanomechanical devices is operation at, or even beyond, the quantum limit. One of the most intriguing aspects of current nanomechanical devices is that they are already on the verge of this limit. The key to determining whether NEMS are in this domain is the relationship between the thermal energy, kBT, and the quantity hf0, where kB is the Boltzmann constant, h is the Planck constant, f0 is the fundamental frequency of the mechanical resonator and T is its temperature.
When the temperature of the device is low and its frequency is sufficiently high that hf0 greatly exceeds kBT, then any thermal fluctuations will be smaller than the intrinsic quantum noise that affects the lowest vibrational mode. In this limit, the mean square amplitude of the vibration can be quantized and can only assume values that are integral multiples of hf0Q/2keff. A full exploration of this quantum domain must wait for crucial technological advances in ultrasensitive transducers for NEMS that will enable us to measure tiny displacements at microwave frequencies.
In spite of this significant challenge, we should begin to see signs of quantum phenomena in nanomechanical systems in the near future. Even the first NEMS resonators produced back in 1994 operated at sufficiently high frequencies that, if cooled to 100 mK, only about 20 vibrational quanta would be excited in the lowest fundamental mode. Such temperatures are readily reached using a helium dilution refrigerator. So the question that comes to mind is whether quantized amplitude jumps can be observed in a nanoscale resonating device? If so, one should be able to observe discrete transitions as the system exchanges quanta with the outside world. At this point, the answer to the question seems to be that such jumps should be observable if two important criteria can be met.
The first is that the resonator must be in a state with a definite quantum number. In general, transducers measure the position of the resonator, rather than the position squared. The continual interaction between such a "linear transducer" and the quantum system prevents the resonator from being in a state characterized by a discrete number of quanta. Transducers that measure the position squared were discussed in 1980 by Carlton Caves, now at the University of New Mexico, and co-workers at Caltech in a pioneering paper on quantum measurements with mechanical systems (see Caves in further reading). And it now seems possible to transfer their ideas to NEMS.
The second criterion is more problematic. The transducer must be sensitive enough to resolve a single quantum jump. Again, ultrahigh sensitivity to displacements is the key needed to unlock the door to this quantum domain. A simple estimate shows that we must detect changes in the mean square displacement as small as 10-27 m2 to observe such quantum phenomena. Is it possible to achieve this level of sensitivity? My group at Caltech has recently made significant progress towards new ultra-sensitive transducers for high-frequency NEMS - and we are currently only a factor of 100 or so away from such sensitivity.
In related work, last year Keith Schwab, Eric Henriksen, John Worlock and I investigated the quantum limit, where hf0 >> kT, for the first time in thermal-transport experiments using nanoscale beams fabricated from silicon nitride (figure 6). As the smallest features on the devices are scaled down in size, the energy spacing between the phonons - the quanta of vibrational energy - increases. When the temperature is lowered, fewer and fewer of these modes of vibration (or phonons) remain energetically accessible. Effectively, this means that most of them cannot participate in thermal transport. Indeed, in a beam that is small enough, only four phonon modes can transport energy between the system and its surroundings.
We found that the thermal conductance in this regime becomes quantized. In other words, each phonon mode that transports energy can only provide a maximum thermal conductance given by ¼k2T/6h. Quantum mechanics thus places an upper limit on the rate at which energy can be dissipated in small devices by vibrations.
In spite of the complications encountered at the quantum level, the rewards in terms of intriguing physics will be truly significant. Force and displacement measurements at this limit will open new horizons in science at the molecular level, new devices for quantum computation, and the possibility of being able to control the thermal transport by individual phonons between nanomechanical systems or between a system and its environment.
Once we have passed into this realm of quantum mechanics, the division between quantum optics and solid-state physics becomes increasingly blurred. Many of the same physical principles governing the manipulation of light at the level of individual photons will come into play for both the mechanical and thermal properties of nanoscale systems.
Future outlook
NEMS offer unprecedented and intriguing opportunities for sensing and fundamental measurements. Both novel applications and fascinating physics will undoubtedly emerge from this new field, including single-spin magnetic resonance and phonon counting using mechanical devices. To take full advantage of these systems we will have to stretch our imaginations, as well as our current methods and "mindsets" in micro- and nanoscale science and technology.
But there remains a gap between today's NEMS devices that are sculpted from bulk materials and those that will ultimately be built atom by atom. In the future, complex molecular-scale mechanical devices will be mass-produced by placing millions of atoms with exquisite precision or by some form of controlled self-assembly. This will be true nanotechnology. Nature has already mastered such remarkable feats of atomic assembly, forming molecular motors and machinery that can transport biochemicals within cells or move entire cells.
Clearly, for us to attain such levels of control and replication will take sustained effort, involving a host of laboratories. Meanwhile, in the shorter term, NEMS are clearly destined to provide much of the crucial scientific and engineering foundation that will underlie future nanotechnology.