What's next for nanotech?
12/01/2004
Solid State Technology asked nanotechnology experts to put forth their vision of the challenges ahead.
Thermal management: A hot topic at the nanoscale level
 Theodorian Borca-Tasciuc, Rensselaer Polytechnic Institute, Troy, New York |
Thermal management is an increasingly serious problem that could hinder the successful development and operation of next-generation electronic devices. With continued scaling, heat-transport problems will aggravate at all levels, including the individual junctions on the transistors, chip, and overall system. Emerging devices are still in their infancy and far from integration; thermal issues have been virtually unexplored. The key to addressing them is a good understanding of energy and charge transport across space scales (from nanometers to centimeters) and time scales (from picoseconds to seconds).
At the larger scales - chip, package, and overall system - heat transport is relatively well understood based on continuum theories for conduction, convection, radiation, and phase-change phenomena. Engineers can consider the optimum thermal-management strategy based on simplicity (cost), as well as the ability to dissipate the chip power load under the allowable temperature differences between the chip and the ambient. As power dissipated/unit area increases, however, more complex systems need to be designed. For example, natural or forced air convection cooling is replaced with single-phase or phase-change refrigerant systems.
The nanoscale challenge. When in the nanoscale range, device dimensions approach characteristic length scales of the heat and charge carriers, while future time scales may approach values comparable to some of the relaxation times between carriers.
In semiconductor materials, heat is carried mostly by atomic lattice vibrations, or phonons. Both particle and wave models are used to describe the phonon thermal transport. As phonons scatter with each other or with electrons, impurities, and crystal boundaries, they give rise to a thermal resistance. The average distance between collisions is on the order of 1-100nm at room temperature. If at least one of the device dimensions is the same as the phonon mean-free path or the phonon wavelength, the classical heat diffusion law does not provide valid predictions of the device temperature. When that is the case, one must look at the microscopic picture of the phenomena. Studies over the last decade identified several major mechanisms believed to affect thermal transport near device junctions: size, interface, and nanoscale hot-spot effects.
Mechanisms at the nanoscale level. Size and interface effects are relevant for devices smaller than the mean-free paths of phonons carrying the heat. In size effects, the increased scattering of phonons on boundaries and interfaces leads to a reduction in the effective thermal conductivity. For example, this effect is believed to explain the 50% drop in the effective thermal conductivity along 100nm-thick silicon films. The effect is even greater for thermal transport perpendicular to the film thickness and also in systems with higher densities of interfaces. In these cases, reductions can be as large as two orders of magnitude, such as for thermal transport across Si/Ge superlattices. Properties affecting the phonon boundary scattering, such as the interface structural quality and acoustic properties, may strongly affect the thermal conductivity value. As device dimensions approach the wavelength of the phonons carrying the heat, interference phenomena may change the phonon spectrum in the structure. This effect may reduce the group velocity and increase the scattering rates of the phonons carrying the heat, further reducing the effective thermal conductivity.
The nanoscale hot-spot effects are due to heat generation from electrons accelerated under high electric fields. Simulations of the electron transport between source and drain show that electrons tend to release their energy in a region a few nanometers wide near the drain side of the device. Because the heat-generation region is smaller than the mean-free path of the heat carriers, phonons from the surrounding material cannot efficiently dissipate the heat from the hot spot. Simulations show that the local temperature of the lattice rises above predictions based on classical diffusion theory by a factor comparable to the ratio between the phonon mean-free path and the size of the hot spot. Furthermore, if theoretical models reflect how the phonons are generated, the excess temperature rise is proportional to the square of the above ratio. This is due to the fact that the generated phonons have low group velocities, in addition to being confined in a small region.
More complex theoretical predictions, including simultaneous electron and phonon transport and electron-phonon interactions, may be necessary to model heat generation and thermal transport in future nanoscale devices. These models may have to reflect the quantization of electron, and possibly phonon, energies. Experiments also will be needed to validate our understanding of thermal transport at the nanoscale and to develop and test effective thermal-management solutions.
Theodorian Borca-Tasciuc is director of the Nanoscale Thermophysics and Energy Conversion Laboratory at Rensselaer Polytechnic Institute, 110 8th St., Troy, NY 12180-3590; e-mail [email protected].
Focusing nanotechnology on the human scale
 Tony Edwards, FEI Co., Hillsboro, Oregon |
The semiconductor industry has pioneered the development and application of nanoscale technologies, but it has learned the hard way that the most elegant technology is worthless without a killer application. Most of the industry’s propositions have long since been reduced to dollars and cents, the universally applicable measure of value in our economic system. Examples of nanotechnology applications illustrate the point.
Tens of thousands of dollars may be invested in a single wafer before it completes the manufacturing process, so finding process excursions early is vital. While electron microscopes can find defects caused by contamination or other extraneous events, focused ion beams (FIB) give process engineers the ability to cut into the wafer to analyze subsurface defects and 3D structures. Combined ion- and electron-beam instruments can expose subsurface features and acquire high-resolution images on a single platform.
Other applications include trimming thin-film heads with a FIB to final dimensions for hard-disk drives, creating smaller heads than otherwise possible and permitting greater storage density. When developing a new circuit design, engineers may use an ion beam to rewire the nanoscale structures of an IC to cut through existing lines and deposit new ones, allowing them to test the new design without repeating the lengthy and expensive manufacturing process at the wafer level.
Engineers also use electron beams to fabricate nanoscale structures. One nanotechnology company, Nantero, is using e-beam lithography in a scanning electron microscope to create prototypes for an entirely new type of memory. The device is essentially a nanoscale electromechanical switch based on the behavior of carbon nanotubes in an applied electrical field. The device promises nonvolatile storage with the speed and longevity of dynamic RAM and a 10× improvement in storage density. While elegant in concept, the success of the product will depend on human-scale values - economic cost and performance - relative to competing technologies.
The problem confronting promoters of new applications is that the economic benefits are not always easy to define. Consider the life sciences and the exploding field of molecular biology. The recent decoding of the entire human genome has presented researchers with a vast treasure trove of nanoscale information that they have barely begun to exploit.
While it is too early to reduce the benefits that may result to simple economic terms, an intuitive analysis of the potential benefits staggers the imagination. The genome encodes the molecular sequence of every protein produced by our bodies. These proteins, nanoscale chemi-mechanical machines, are the very stuff of life. Although we know their molecular sequence, we do not yet know their ultimate shapes and functions.
Conventional methods for determining shape and function require weeks or months for a single protein, and there are tens of thousands of them. Newly developed techniques for high-resolution imaging of frozen specimens in a transmission electron microscope can provide higher-level structural analysis in a day. Determining the structure and function of these proteins (the human proteome) will be the great project of the next decade.
Although it is impossible to conceive all the human-scale benefits that may result from nanotechnology, it is equally impossible to doubt their ultimate value. The danger to promoters of nanotechnology lies in becoming so enamored of operating at the same spatial scale as the fundamental building blocks of our world that we forget that most of the things we truly care about affect us at the human scale. Working at the nanoscale to serve the human scale is the heart of the matter.
The greatest discoveries in nanotechnology are surely yet to be made. As with any truly revolutionary technology, it is impossible to know in advance where most paths will lead. In some cases, as in the semiconductor industry, the coming of nanotech is more of an evolution than a revolution. New advances align closely with existing processes, and economic models give relatively predictable results. In other cases, as in molecular biology, unknowns dominate the equations and outcomes are uncertain.
Those of us attempting to advance this technology must evaluate every new idea ultimately in dollars and cents. When this is not possible, we must still find a way to evaluate the benefits of the research on a human scale. In so doing, we may hope to generate the greatest good for the greatest number of people, but perhaps more importantly, we may build the public’s trust and support for a revolution where the best is yet to come.
Tony Edwards is vice president, SEM and DualBeam Business Line, at FEI Co., 5350 NE Dawson Creek Dr., Hillsboro, OR 97124; ph 503/726-7500, e-mail [email protected].