The term "adaptive quantum design" denotes a methodology for systematically seeking robust, manufacturable designs of semiconductor devices — especially semiconductor optoelectronic devices having nanoscale or even atomic- scale features. This methodology has been developed to complement advances in fabrication capabilities that make it possible to tailor semiconductor devices ever more precisely, such that it likely will soon be possible to routinely control the positions of features as small as atoms and molecules within devices. Because the number of atom configurations that could, potentially, be fabricated is almost unimaginably large and quantum fluctuations and collective quantum phenomena become important at molecular and atomic scales, traditional design methods and traditional models of device physics based on classical physics and semiclassical approximations of quantum phenomena are not adequate for exploration of the vast space of design options.
In adaptive quantum design, to meet the challenge of identifying the best (or nearly the best) among many possible designs, one utilizes a combination of modern computing power, adaptive algorithms, and realistic computational models of device physics to solve optimaldesign problems. For a given case, the optimal-design problem can be summarized as the problem of identifying the best device configuration(s) to obtain a measure of device response that most closely matches an objective function specified by the designer. Typically, the optimal-design problem involves searching for a broken-symmetry spatial configuration of a semiconductor. (A brokensymmetry configuration is one in which the chemical composition and/or the crystalline structure of the material is altered from its natural crystalline form that exhibits a symmetry. A prime example of symmetry is translational symmetry, in which composition and crystalline structure are the same at any position along one of the natural crystalline axes.)
Accordingly, in adaptive quantum design, the optimal-design problem is solved by means of computational searches to numerically identify the best broken-symmetry spatial configuration of atoms and molecules that produces the best approximation of the objective function. The two major ingredients of adaptive quantum design are a model representing the quantum device physics and an algorithm (e.g. a genetic algorithm) that searches the space of parameters representing the various configurations.
In a study, this methodology was applied in the design of InxGa1-AsyP1-y devices containing broken-symmetry quantum wells (see figure) optimized for excitonic absorption. [In excitonic absorption, excitons (hole/electron pairs somewhat analogous to hydrogen atoms) are formed by absorption of photons of less than the bandgap energy in a direct-bandgap semiconductor. The binding energy of an exciton can be increased by confining the electron and the hole within a two-dimensional quantum well, causing the absorption spectrum to include a strong peak just below the bandgap energy; this fact has been exploited for use in such modern optoelectronic devices as modulators and detectors.] The study included design and fabrication of an electro-absorption modulator. In a test at a wavelength of 1,545 nm, the device was found to perform comparably to the best in a commercial product line of modulators. Further improvements in device performance could be achieved through attention to radio-frequency aspects of the design, which were omitted from the study to enable concentration on demonstrating the basic nature of adaptive quantum design.
This work was done by A. F. J. Levi of the University of Southern California for the Defense Advanced Research Projects Agency (DARPA). For more information, download the Technical Support Package (free white paper) at www.defensetechbriefs.com/tsp under the Electronics/Computers category. DARPA-0002
This Brief includes a Technical Support Package (TSP).
Adaptive Quantum Design of Semiconductor Devices
(reference DARPA-0002) is currently available for download from the TSP library.
Don't have an account? Sign up here.