Research

The technologically useful properties of a solid often depend upon the atomic-scale defects it contains. Despite the harmful sound of "defects", their presence is often essential to the operation of nanoscale devices based on semiconductors. The types and concentrations can sometimes be tuned precisely 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 is necessary for device fabrication. Specially chosen dopant atoms that replace silicon host atoms in the crystal lattice represent substitutional defects, and microelectronic devices could not function without them. Defects in TiO₂ and/or ZnO affect the performance of photoactive devices such as blue-green emitters, the effectiveness of catalysts and photocatalysts for environmental and energy applications, the sensitivity of solid-state electrolyte sensors, the efficiency of devices for converting sunlight to electrical power, the electrical transmission of transparent conductors for solar cells, and the magnetic properties of spintronic devices.

Because 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," with doping representing an especially vital dimension. Research in this group focuses on developing new methods for defect engineering and doping in semiconductors to make nanoscale devices of interest for energy, environmental, and nanoelectronics applications. We have discovered several new physical mechanisms to accomplish this control that work especially well near room temperature. Our work employs both experiments and computations.

In the same way that gases and liquids react with surfaces from above, solid defects can react from below. Figure 1 shows an example of water molecules breaking apart on a metal oxide surface, which in turn injects oxygen interstitials (O_i) into the bulk to react with oxygen vacancies (V_o) and hydrogen interstitial atoms (H_i) that reside there. 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 injection/creation rate, depending upon the application) of defects at that surface.

Fig.1

We have discovered through self-diffusion measurements and first-principles computations that poison-free oxide surfaces inject interstitial oxygen atoms into the crystalline solid when simply contacted with liquid water near room temperature. These interstitials diffuse quickly to depths of 20-2000 nm and eliminate prominent classes of unwanted defects or neutralize their action. The mild conditions of operation access a regime for oxide fabrication that relaxes important thermodynamic constraints hampering defect regulation by conventional methods at higher temperatures. The surface-based approach appears well suited for use with nanoparticles, porous oxides and thin films for applications in advanced electronics, renewable energy storage, photocatalysis and photoelectrochemistry.

As an example of the power of this approach, we have demonstrated isotopic fractionation of ¹⁸O by a factor of three below natural abundance in a near-surface region up to 90 nm wide. Figure 2 shows example depth profiles of ¹⁸O, with the isotopic depletion appearing in the form of a "valley" in concentration near the surface. The shape of the valley depends upon whether the water itself or the O₂ dissolved in it is enriched in ¹⁸O. We have discovered that the mechanism of diffusional isotopic purification is governed by the statistics of defect hopping combined with steep chemical gradients of oxygen interstitials originating from the submerged surface. Slightly acidic and slightly basic liquid solutions both enhance the fractionation, which represents an example of using liquid-surface chemistry (via pH) to control the defect-surface chemistry. There is no reason to believe this fundamental physical picture is restricted only to oxygen or TiO₂.

Fig.2

Since defects can be electrically charged, electric fields that often exist near surfaces can also be manipulated (Fig. 3) to attract or repel charged defects nearby. We employ optical methods to measure these electric fields.

To better understand all 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 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. Fig. 4 depicts a typical experimental apparatus, which employs an electrochemical cell with reference, counter and working (oxide) electrodes by green, blue and red arrows, respectively (done in collaboration with Prof. Xiao Su). Oxygen and other atoms inject from the surface into the bulk, and we measure the rate of this injection with secondary ion mass spectrometry. We are performing analogous experiments with metal ions dissolved in the liquid to understand how the metal-related defects behave.

Fig.3

Fig.4

To interpret the diffusion results, we have developed several mathematical models for the diffusion in metal oxides. These models involve the solution of differential equations describing defect diffusion and reaction.

To supplement this work, we also perform quantum calculations (in collaboration with Prof. Elif Ertekin) to uncover the atomistic pathways and energies associated with defect injection and annihilation. Fig. 5 shows an example from a calculation for oxygen interstitial injection into TiO₂.

Fig.5