PNNL

Abstract

ZnO is a wide-gap oxide semiconductor of considerable current interest for electronics, optoelectronics, and, possibly, semiconductor spintronics. Through selective electronic doping, ZnO can be a transparent conducting oxide [1], a UV light emitter [2,3], and, when alloyed with a few to several atomic percent of Co, Mn or other transition metals with unpaired d electrons, is squarely at the center of controversy in the field of high-Tc ferromagnetic semiconductors [4]. Despite this range of interests, fundamental aspects of electronic doping in ZnO remain poorly understood. For instance, theoretical calculations suggest that H at either interstitial or substitutional sites is responsible for the persistent n-type conductivity that frustrates efforts to achieve p-type behavior, an essential requirement for the fabrication of pn light emitting diodes and hole-mediated ferromagnetic coupling of Mn dopants [5,6]. Yet, experimental studies do not in general align with this prediction. Another interesting aspect of n-type conduction in ZnO is the appearance of a near-surface conducting channel that appears to be present in bulk crystals [7,8]. Although it has been a few years since the original observation of this phenomenon was described, detailed understanding is lacking. In this paper, David Look presents a transport study aimed at gaining insight into the phenomenon of near-surface conductivity in bulk ZnO. All samples investigated were unintentionally n-type, as is typical. The measured Hall data were interpreted using a two-layer conduction model based on standard charge balance equations. Look fit the temperature dependence of the electron mobility and carrier concentration and extracted the thicknesses of the surface conducting layer, which dominates at lower temperatures, for bulk crystals synthesized by different methods and marketed by different companies. By making reasonable assumptions about the surface acceptor density, the lower limit of the surface conducting layer thickness was extracted from the transport data. From a surface science perspective, two aspects of the investigation are of particular interest: (i) the donor concentration was enhanced by annealing in forming gas, which is 5% H2, and, (ii) despite relatively weak variation in the sheet carrier concentration, the surface conducting layer thickness (dsurf) varied from as large as 28 nm to as small as 1.5 nm, depending on the method of preparation and supplier. The ongoing exploration of this phenomenon begs for the kind of experiments that practitioners of surface, interface and thin-film science can readily conduct. In bulk crystals, is H doping really the cause? Answering this question represents a major challenge because the donor concentration is in the 1017-1018 cm-3 range. Determining the donor identity in this concentration range is a major experimental hurdle. How does dsurf depend on crystallographic orientation? How does the phenomenon depend on the extent of band bending? Does persistent photoconductivity come into play? Can the effect be modified by band bending modification via surface photovoltage effects? All of the above questions pertain to epitaxial films as well, but other questions also arise. How do the film thickness and strain state affect dsurf? Can the surface conduction effect be enhanced by judicious design and growth of quantum well structures made from ZnO, MgxZn1-xO, and/or CdxZn1-xO? Answering these and other questions that arise will expand our understanding of this fascinating and potentially important material, as well as pave the way for device applications.