Session A25: Focus Session: Dopants and Defects in Semiconductors - ZnO

Magnetic Resonance Studies of Oxygen and Zinc-Vacancy Native Defects in Bulk ZnO Crystals

ZnO is currently attracting increasing attention as a key material for a wide variety of electronic and optoelectronic applications. Optical, electrical, and magnetic properties of ZnO are believed to be strongly influenced by native defects. However, unambiguous experimental evidence confirming the formation of these defects in as-grown ZnO as well as evaluations of defect densities is currently sparse. In this talk we shall review our recent results from comprehensive defect characterization of as-grown bulk ZnO. By using electron paramagnetic resonance (EPR) and optically detected magnetic resonance (ODMR) spectroscopies, we show that both oxygen and zinc vacancies are formed in ZnO grown from melt without subjecting to irradiation. Defect concentrations are also determined. Based on spectral dependences of its EPR and ODMR signals, the $V_{Zn}^- $ defect is concluded to act as a deep acceptor responsible for the red emission peaking at around 1.6 eV, but does not participate in the green emission as commonly believed. The energy level position of the $V_{Zn}$ corresponding to the (2-/-) transition is determined to be at $E_v$+1.0 eV. The center is also shown to exhibit a strong JT distortion with a JT energy of 0.8 eV. On the other hand, oxygen vacancies are probably less important in carrier recombination since they were only detected in EPR but not in ODMR. Annealing properties of both defects were also studied and higher thermal stability of the Zn vacancy was concluded. It was also suggested that annealing of the $V_{Zn}$ centers is facilitated by thermally-activated diffusion of impurity atoms to the $V_{Zn}$ sites. The obtained results are of importance for a better understanding of the defects in ZnO. They also provide useful information on control of electrical properties and defect-reaction induced degradation during device processing and operation, in the material that is commonly used as a substrate for epitaxial growth of layered device structures based on ZnO.   

A pathway to p-type wide-band-gap semiconductors

The development of solid-state ultraviolet-light sources in the form of light-emitting diodes and laser diodes to replace mercury lamps and bulky gas lasers is currently hindered by doping issues in wide-band-gap semiconductors such as AlN, ZnO, and ZnMgO alloys. While $n$-type AlN, ZnO, and ZnMgO can be achieved by using traditional doping concepts, i.e. swapping host atoms for impurities with an extra valence electron, $p$-type is still a major challenge. Based on first-principles calculations we devise an alternative approach to $p$-type doping in AlN, ZnO, and ZnMgO. Instead of searching for acceptors on the left of the host atoms in the periodic table, we propose to search on the far right. We find that F placed at interstitial sites in AlN, ZnO, and ZnMgO acts as a shallow acceptor, leaving a hole in an effective-mass state near the valence-band maximum. We discuss the stability of F impurities, and propose a procedure to selectively introduce F in the interstitial lattice sites of the above wide-band-gap semiconductors.   

Quantum Monte Carlo calculations of defects in ZnO

The semiconductor ZnO holds much promise for many applications due to its ability to display a wide variety of properties. However, an understanding of the structure and electronic structure of the defects which lead to the material's properties has proven difficult. Previous density functional theory (DFT) calculations by Van De Walle et. al. [1] using the local density approximation, suggest that hydrogen interstitials and oxygen vacancies may be the dominant defect present in the as-grown material leading to its n-type conductivity. However, the corrections accounting for DFT's band gap problem lead to a wide variance in defect formation energies reported and uncertainty in the qualitative results. More recent calculations by Oba et. al. [2] using the HSE hybrid functional confirm the presence of these defect structures and their electronic transition levels. We are performing Quantum Monte Carlo calculations for the defects in ZnO to determine the defect formation energies and transition levels and to verify the previous DFT calculations. [1] C. G. van de Walle, Phys. Rev. Lett. 85, 1012 (2000). [2] F. Oba, Phys. Rev. B 77, 245202 (2008).   

Why nitrogen cannot lead to $p$-type conductivity in ZnO

Nitrogen is often considered as the most suitable dopant for achieving $p$-type ZnO since it has a similar size to oxygen and has successfully been used as a shallow acceptor in ZnSe. In spite of many published reports on $p$-type ZnO achieved by nitrogen doping, reproducibility and stability still seem to be major issues, and devices based on \textit{pn} junctions have remained elusive. We have studied the electronic and structural properties of the nitrogen acceptor in ZnO using state-of-the-art first-principles calculations based on hybrid functionals (Heyd-Scuseria-Ernzernhof). We find the nitrogen acceptor level to be 1.3 eV above the valence-band maximum. Nitrogen therefore cannot lead to hole conductivity in ZnO. We have analyzed the optical properties (absorption and luminescence), which should offer characteristic experimental signatures of nitrogen in the lattice. Previous experimental reports are discussed in light of our new results.   

Nitrogen doping of single crystal ZnO

Zinc oxide (ZnO) is a wide band gap semiconductor with a range of potential optical, electronic, and mechanical applications. The lack of control over defects, in particular reliable $p$-type doping, is a problem that needs to be overcome. Nitrogen is a promising acceptor dopant. We incorporated nitrogen-hydrogen (N-H) complexes in ZnO during chemical vapor transport (CVT) growth, using ammonia as an ambient. Using commercial ZnO as a seed, we obtained bulk single-crystal growth. The N-H bond-stretching mode gives rise to an infrared (IR) absorption peak at 3150.6 cm$^{-1}$ at liquid helium temperatures. Isotopic substitutions result in the expected frequency shifts, providing an unambiguous identification of these complexes. The N-H complexes are stable up to $\sim $700\r{ }C. Annealing in oxygen appears to enhance the dissociation of the N-H bond, an observation that could prove useful in achieving reliable $p$-type conductivity.   

Theoretical simulation of doping mechanisms in p-type ZnO

The efficiency of doping in crystalline semiconductors is related to the high concentrations of a certain type of defects. We have developed a universal approach to relate the concentrations of desirable defects as a function of the doping process experimental parameters such as the growth temperature. As shown in our previous work (J. Appl. Phys. 105, 113711 (2009)) our calculations reproduce well the experimental results for the $p$-type doping of ZnO with Phosphorus prepared by RF Sputtering, PLD and MBE. For the $p$-type doping of ZnO, at least one supplementary complication arises as the oxygen vacancies created during the process greatly diminish the number of holes. For the case of $p$-type doping of ZnO by implantation with +1 charged P and As ions, we calculated the efficiency of doping, together with the extent of O vacancy contamination, for a range of substrate temperatures from 30 to 700 \r{ }C. The results show that As gives raise to hole densities three orders of magnitude higher than P (therefore recommending As a more efficient $p$-type dopant than P) and that many O vacancies are created during the ion implantation process. The high density of O vacancies resulting from our calculations may explain the poor stability and reproducibility of $p$-type ZnO.   

Robust Conductivity Changes in ZnO and MgZnO Nanoparticle Films from Annealing in Hydrogen

We report changes observed in the I-V characteristics of ZnO and MgZnO nanoparticle thin films after annealing in H$_{2}$ at sufficiently high temperatures. The nanoparticles were grown on insulating silicon substrates and had an average diameter of 30 nm. The devices were of a two terminal design, where the terminals consisted of two 25 $\mu $m diameter gold wires laid parallel to each other on the nanoparticle film to measure the current passing through the film. When exposed to H$_{2}$ gas at room temperature, no significant changes in the current-voltage behavior of the nanoparticles were observed relative to measurements done in vacuum. Annealing in H$_{2}$ below 140\r{ }C also resulted in no significant change in the current. When annealed above 140\r{ }C, we observed an increase of about a factor of ten that was semi-permanent. The origin of the change in I-V characteristics of ZnO and MgZnO nanoparticles when annealed in H$_{2}$ will be discussed.   

Zinc Vacancy Formation and its Effect on the Conductivity of ZnO

Exposing single crystal ZnO to 193-nm ArF excimer laser radiation can produce metallic zinc nanoparticles along the surface. The particle production mechanism appears to involve interstitial-vacancy pair formation in the near-surface bulk. Conductivity measurements made with one probe inside the laser spot and the other outside show evidence for rectifying behavior. Positron annihilation spectroscopy confirms the presence of Zn vacancies. We suggest that Zn vacancies are a possible source of p-type behavior in irradiated ZnO. Quadrupole mass spectroscopy shows that both oxygen and zinc are emitted during irradiation. Electron-hole pair production has previously been invoked to account for particle desorption from ZnO during UV illumination. Our results suggest that preexisting and laser-generated defects play a critical role in particle desorption and Zn vacancy formation.   

The delocalized nature of holes in (Ga, N) cluster doped ZnO

(Ga, N) cluster-doping in ZnO has been considered as a valid approach to enhance the p-type doping of ZnO. So far, the argument on the enhancement is based on the reduction of dopant formation energy and ionization energy. Here we present spin-polarized density functional theory calculation to reveal that in (Ga, N) cluster doping of ZnO, a hole state created by a (Ga, N) cluster contains the contribution from all N atoms in the cluster, meaning a more delocalized nature of the hole as compared to the case of mono N doping of ZnO. Our results explain from the electronic point of view why (Ga, N) cluster doping enhances the p-type doping of ZnO. We will compare both LDA+U and hybrid density functional calculations.   

Enhanced formation of Zn-interstitials by their attractive interactions with O-vacancies in ZnO

O-deficiency is known to give n-type doping in ZnO without intentional dopants. Even though the native defects, O-vacancies and Zn-interstitials, have been excluded theoretically as the main sources of the n-type doping, experiments have still shown the close relationship between the n-type doping and the O-deficiency. In this work, we investigated the interactions between the O-vacancies and Zn-interstitials in ZnO based on density-functional theory calculations, and propose that the formation of Zn-interstitials can be significantly enhanced by the attractive interactions with O-vacancies. The enhanced formation of Zn-interstitials can be an important source of the n-type doping in O-deficient ZnO.   

Size-dependent induced magnetism in carbon-doped ZnO nanostructures

The role of quantum confinement on the carbon doped ZnO nanocrystals will be examined using a real-space, first-principles method. We find that the spin polarization energy of carbon defects in nanocrystals is sensitive to size owing to the partial occupancy of a less-localized minority spin state. In addition, we find that the partial occupancy of a minority spin state results in a direct exchange mechanism between the carbon defects even under strong confinement. The analysis suggests that the semiconductor with non-magnetically induced magnetic moments should provide a significant advantage over the traditional dilute magnetic semiconductors when applied to nanometer-scaled applications.   

First-Principles Theoretical Analysis of Doping in II-VI Compound Semiconductor Nanocrystals

Doping of II-VI compound semiconductor nanocrystals allows for precise control of their optoelectronic and magneto-optical properties. Despite the significant progress in the synthesis of doped nanocrystals, the underlying doping mechanisms in colloidal nanocrystals still remain elusive. Using first-principles density functional theory calculations, we have carried out a fundamental quantitative study of semiconductor nanocrystal doping based on the analysis of adsorption and diffusion of dopants on nanocrystal surface facets. We focus on Mn doping of ZnSe and ZnO nanocrystals with characteristic sizes of $\sim $5 nm that have polyhedral shapes with well-defined facets. Our surface kinetic growth model takes into account the equilibrium nanocrystal shape, nanocrystal surface facet structure, nanocrystal composition, surfactants in the growth solution, and dopant surface coverage. Our theoretical results are consistent with recent experimental reports on doping efficiencies and provide an explanation for the doping difficulties during nanocrystal growth.   

A non-LSW Model for Semiconductor Nanocrystal Growth

While the efficient production of nanocrystals has been studied for decades, the mechanism by which the nanocrystals grow is still poorly understood.~ Current models do not extend far beyond the very general Lifshitz-Slyozov-Wagner theory of grain growth.~ Unfortunately, many nanocrystal materials do not rigorously obey Lifshitz-Slyozov-Wagner dynamics, specifically materials such as ZnO and CdSe in certain growth conditions. In an effort to understand these materials better, we approach the nanocrystal growth problem \textit{ab initio} by including the fact that the precursor molecule is being chemically produced as the nanocrystals are being formed.~ We apply this assumption to the kinetics of nanocrystal growth, and develop an analytical derivation of the time-dependent size distribution of nanocrystals in this scenario.~ With this mathematical model, we compare our results directly to nanocrystal experiments for ZnO and for CdSe, where both the average and FHWM of the size distribution evolve with time, to determine the chemical mechanism underlying the formation of these materials.