Research

Research in this group focuses on the control of defect behavior in semiconducting materials to make nanoscale devices of interest in energy, environmental, and microelectronics applications. Despite the harmful sound of "defect," such species can actually be beneficial for semiconductor properties. For example, controlled substitution of dopant atoms for host atoms in a semiconductor (as "substitutional defects") is absolutely essential for the operation of microelectronic devices. Our work aims primarily at controlling the behavior of substitutional, interstitial, and vacancy defects within semiconducting solids and on their surfaces. Indeed, "defect engineering" seeks to control the primary kinds of defects in semiconductors as well as their concentrations, spatial distribution, and mobility. We have discovered several new physical mechanisms to accomplish this control that work particularly well at the nanoscale. The mechanisms include saturation of dangling bonds at a surface and photostimulation. Our work employs both experiments and computations to develop this fundamental science base while simultaneously applying the findings to practical applications.

New Mechanisms for Defect Engineering In the same way that gases react with surfaces from above, point defects within a solid can react from below. Little attention has been paid to this form of surface chemistry. Our basic idea is that saturating dangling bonds at a semiconductor surface controllably interferes with both the annihilation and generation of defects. Consider the example of interstitial atoms. A surface having many dangling bonds can annihilate interstitial atoms by adding them to those bonds. However, if the same surface becomes saturated with a strongly bonded adsorbate, annihilation requires the insertion of interstitials into existing bonds. Such insertion should have a higher activation barrier and a correspondingly reduced probability of occurrence. By analogy, surfaces with many dangling bonds should be capable of generating interstitials more rapidly than saturated surfaces. The atoms on atomically clean surfaces have lower coordination than atoms on saturated surfaces, so that interstitial creation from a clean surface should require less bond breaking and exhibit a lower activation energy. Higher generation rates should ensue. To better understand these phenomena, we employ measurements of solid-state diffusion together with electron microscopy of large defects. Most applications of defect engineering focus on regions in close proximity to surfaces, so this approach has particular practical relevance. Defect engineering by photostimulation works by changing the average charge state of defects. The average charge state in turn affects the concentration of these atoms, their diffusion rate, or both.

The principles outlined above were first discovered in silicon, with applications in microelectronic devices. However, the same principles operate in other semiconductors such as metal oxides used in catalysts for energy applications. In fuel cell applications, for example, metals supported on semiconducting oxides such as Pt/Co₂O₃ and Pd/V₂O₅ are known to be among the best catalysts for cathodes and anodes, respectively. In a similar way, noble metals on semiconducting oxides can be used for generating hydrogen from water using sunlight. In these cases, the support does not stand by passively but participates actively in controlling the reactivity of the metal. Active metal particles are often very small - sometimes only a few nanometers in diameter. The activity can be strongly influenced by the surface and near-surface electronic properties of the underlying semiconductor support. Those semiconductor properties in turn depend upon the concentration and spatial distribution of substitutional and vacancy defects that serve as dopants. Thus, we view these systems as nanoscale semiconductor devices whose behavior can be tailored through control of the defects in the semiconductor. This approach provides a new way to investigate the properties of metal oxide catalysts, and to create novel catalyst structures with improved activity and selectivity in practical applications.

Integrated circuit performance has been improving for several decades according to Moore's Law: that is, speed doubles roughly every 18 months. Much of this improved performance comes from shrinking the constituent transistors. At the heart of semiconductor transistors lie structures called "pn junctions," which are narrow regions where the semiconductor changes from being electron-rich to electron-poor (n-type to p-type). By 2010, the International Technology Roadmap for Semiconductors (produced by the Semiconductor Industry Association) indicates that pn junction depth will need to decrease to 10 nm. Such junctions are created by "ion implantation and annealing," in which ions of the dopant are accelerated and literally shot into a silicon wafer the way bullets can be embedded into a pile of stones. Heating (i.e., "annealing") the wafer after implantation gives the atoms in the structure enough energy to move around so that damage left over from implantation can be healed, and so that more dopant atoms can move into useful sites. The annealing is done with very powerful lamps. However, the annealing also makes the dopant sink deeper and adversely affect device performance. Since nearby surfaces as well as photostimulation are involved, we are employing our two methods of defect engineering to devise new ways solve this problem.