Research

The technologically useful properties of a solid often depend upon the defects it contains. Despite the harmful sound of "defects," their types and concentrations can be precisely tuned to enhance device performance. For example, silicon-based integrated circuits rely upon defects such as vacancies and interstitials to mediate the diffusion of dopant atoms that are critical to device performance. Defects in TiO₂ and ZnO affect the performance of photoactive devices, effectiveness of catalysts and photocatalysts, sensitivity of solid-state electrolyte sensors, and efficiency of devices for converting sunlight to electrical power.

Since defects affect many aspects of semiconductor behavior, the ability to control the type, concentration, spatial distribution, and mobility of such defects is important for practical applications. The practice of such control is termed "defect engineering." Considerable progress has been made in developing this ability for silicon, but the methods still need considerable development for other semiconductors such as metal oxides.

Research in this group focuses on developing new methods for defect engineering in semiconductors to make nanoscale devices of interest for energy, environmental, and microelectronics applications. We have discovered several new physical mechanisms to accomplish this control that work well at small length scales below about one micrometer. The mechanisms include photostimulation and reactions of defects at surfaces. Our work employs both experiments and computations to develop a fundamental science base while simultaneously applying the findings to practical applications. Several specific projects are described below.

In the same way that gases react with surfaces from above, solid defects can react from below. Little attention has been paid to this form of surface chemistry up to now. Yet as devices shrink and surface-to-volume ratios increase, surface phenomena become increasingly important and surface-based defect engineering strategies become more viable. Our basic idea is that making surfaces more chemically active by removing passivation can controllably increase the annihilation rate (or generation rate, depending upon the application) of defects at that surface.

Consider the example of interstitial atoms. A chemically passive surface whose bonds are fully saturated has little capacity to absorb interstitials from the underlying bulk. But a chemically active surface having many dangling bonds can easily absorb interstitials by adding them to those bonds. Exposing the surface to controlled amounts of a passivating agent yields an absorbing capacity that is intermediate between the two extremes. Analogous arguments can be formulated in cases where the surface generates interstitials rather than absorbs them. Consequently, higher generation rates should ensue.

To better understand these phenomena, we make indirect measurements of defect concentrations by monitoring solid-state diffusion rates in the vicinity of surfaces. Atomic defects themselves are very difficult to see directly, but the macroscopic diffusion rate of an isotope or dopant generally scales with defect concentration. Typical experiments involve exposing single crystal surfaces of semiconductors such as TiO₂ or ZnO to isotopically labeled oxygen gas at several hundred degrees Celsius. The gas atoms exchange through the surface with oxygen atoms in the bulk, and we measure the rate of this diffusional exchange with secondary ion mass spectrometry. To interpret the results in terms of defect behavior, we have developed several distinct mathematical models for the diffusion in metal oxides. These models not only help us to understand the interaction of defects with surface dangling bonds, but also the interaction of electrically charged defects with near-surface electric fields.

We focus on TiO₂ and ZnO because these materials are used in photocatalysis for environmental cleanup and solar hydrogen production (TiO₂), as well as in nanoscale devices for sensing (ZnO, TiO₂) and nanophotonics (ZnO). Oxygen defects play an especially important role in determining the behavior of these oxides.

In integrated circuits based on silicon, widely practiced methods exist to precisely control the concentration of charge carriers in the semiconductor. Dopant atoms such as boron or arsenic can be introduced by various methods, and essentially every dopant atoms becomes ionized and releases charge that contributes to the electrical conductivity. In metal oxides, however, the situation is much more complicated. Dopants typically do not ionize fully, and native defects within the host such as oxygen vacancies can act as unintentional dopants themselves. Worse yet, many of the methods commonly used in silicon to measure the concentration of charge carriers do not work very well for metal oxides.

We are developing methods to make charge concentration measurements in metal oxides such as TiO₂ more reliably, as well as methods to minimize charged native defects and introduce fully ionized dopants more controllably. For this work, we synthesize films of TiO₂ (in its anatase form) that are only 30 to 300 nm thick. The synthesis method we employ is atomic layer deposition, in which a solid substrate is exposed to alternating flows of source gases to build up a thin film one atomic layer at a time. For TiO₂, we use Ti(OCH(CH₃)₂)₄ to provide titanium and H₂O to provide oxygen. Dopant atoms such as Mn, Nb, and Cr are also introduced in to the film from the gas phase during growth.

Once the material is grown, we subject it to a powerful battery of physical, chemical, and electrical characterization techniques to understand the nature of the oxide that has formed. Methods include x-ray photoelectron spectroscopy (XPS), x-ray diffraction (XRD), x-ray reflectivity (XRR), and scanning electron microscopy (SEM) to understand film composition, structure, and morphology. We determine electrical properties by measuring film capacitance as a function of applied voltage in a special diode test structure. The method is complicated but provides rigorous and accurate determination of TiO₂ carrier concentration, which we correlate to the performance of a catalyst or sensor.

The discussion above outlines how surface chemistry affects the behavior of defects in the underlying semiconductor. Yet the reverse is also true - the defect-induced electronic properties of the solid can influence surface reactivity. Examples include surface-active "devices" such as certain catalysts and sensors. We seek to exploit such effects in metal oxide heterostructures, in which one semiconducting oxide is supported on another. Such supported oxides are used widely used in industry for production of chemical products (oxidation of alcohols, oxidative dehydrogenation of alkanes) and for environmental cleanup (selective catalytic reduction of NO_x in large-scale combustion flues). Through the synthesis methods described above, we seek to tune the activity and selectivity of supported catalysts: in particular, vanadia (V₂O₅) supported on titania (TiO₂). We believe the reactivity of the V₂O₅ surface can be tuned by changing the charge carrier concentration in the underlying TiO₂ via controllable doping. We therefore grow thin films of vanadia by chemical vapor deposition from the volatilized precursors vanadium oxide triisopropoxide and water. The growth takes place on titania thin films that have been doped in a way that we predict will improve activity or selectivity of the overlayer. Then we make reaction rate measurements in a small low-vacuum reactor using a quadrupole mass spectrometer for the analysis of the gas phase composition. The partial oxidation of methanol is used as a test reaction.