Research

This is a brief summary of some of the subjects that I have worked on in recent years as well as a list of topics that I would like to pursue in the future.

Doping and diffusion in nano-structures

There has been much effort recently to develop nano-scale technology because semiconductor devices are approaching limiting dimensions. Much progress has been made regarding the precise processing of nano-structures. Some progress has also been made in our ability to selectively dope these nano-structures, although there are still many material systems and structures that challenge our ability to introduce dopants.

In the study of bulk semiconductor technology, not only is the introduction of dopants important, but also the control of the placement. A literature search for "Si and diffusion" nets over 13,309 journal papers, ranging over 30 years, whereas a search for "Si and nanowire and diffusion" nets only 37 papers, most of which focus on surface diffusion. Clearly before nano-technology becomes viable, more research is necessary.

Because diffusion is mediated by intrinsic defects, I have begun my work investigating the properties of intrinsic impurities, interstitial atoms and vacancies. I have found that interstitial atoms are highly unstable, and are quickly drawn to the surface. Vacancies also prefer to be near the surfaces, although for special high-symmetry geometries it is possible to trap vacancies in the interior of nano-crystals. The figure shows the distribution of the occupied wave function associated with such high symmetry vacancy structures. I have found that the energy barrier to vacancy diffusion is slightly reduced from the value in bulk. From this evidence one might conclude that intrinsic defects will only be found on the surfaces, thereby depleting the interior of intrinsic defects and decreasing the diffusivity inside of nano-structures. However, I have also found that the energy of formation for vacancies and the energy to nucleate a Frenkel-pair (a vacancy and interstitial pair) are substantially lower than in bulk.

By coupling the enhanced creation and mobility of vacancies, one can hypothesize that there internal stirring. This internal mixing of the nano-structure would be detrimental to controlling the dopant placement. The next phase of this research involves creating a kinetic Monte Carlo model to gain an understanding of the self-diffusion process. The Monte Carlo model will be parameterized from the energies calculated by ab initio methods.

Defects in crystals

As a graduate student at the University of California-Berkeley, my focus was defects in crystals. Since that time I have had a continued enthusiasm for defects. This subject is central to my research interests because I see many material technology problems as being defect problems. For the development of a technological material we must first be able to remove the impurities that poison the properties we desire. Second we introduce our own defects that act to tailor the properties. Without the ability to control defects the material choices for engineering design would be severely limited. Semiconductor technology, high-tech steels and advanced ceramics are powerful examples of the importance of defects. Even at the most primitive level, our ability to work with basic metals and pottery depends on the control of atomic scale defects.

Much of my work involves defects in semiconductors because the rapid growth of the semiconductor industry affords theoretical researchers many opportunities to work directly with experimentalists on cutting-edge technologies. The techniques that I have developed and employed can be applied to other material systems. Not only have I studied the properties of dopants and impurity atoms in semiconductors, but I have also investigated multi-dimensional defects. I have examined the segregation of dopants to stacking faults and have studied the structure and properties of dislocation core. The figure shown here, is taken from PRL V95 P145501 (2005). The left image is an experimental HRTEM image of a partial dislocation in GaAs, and the right is an HRTEM image simulated from an atomic structure that I calculated. Upon detailed analysis of the two images it is found that the column positions match to better than 18 pm.

I am continuing my study of defects. I have begun studying the properties of dislocation in II-VI semiconductors and the preliminary results suggest that the reconstruction is very different from other semiconductor systems, due to the increased ionic nature of the compound. I am also presently investigating the effect of a dislocation's strain field on the local polarization of BTO ferroelectric crystals. For this study I am employ an effective Hamiltonian method.

Ferroelectric PZT thin films

Currently most computers use a form of DRAM for memory. DRAM is a volatile memory, meaning that it must be refreshed continually, approximately once every 60 ms. This requires an applied power source. There is a desire to move to non-volatile memory that would allow for the persistent storage of data even when the computer is turned off. This would not only allow for a substantial reduction of energy consumption during operation, but would also significantly improve the speed of boot-up. Unfortunately the existing non-volatile RAM does not perform as well as the DRAM. The non-volatile flash memory that is used for USB memory sticks lacks the speed of DRAM. Also the number of times that flash memory can be cycled is limited (around 100,000) because the rewrite process involves the conduction of charge carriers across an insulating region either by hot carrier injection or a quantum tunneling mechanism, which ultimately damages the insulator.

Ferroelectric RAM (FeRAM) is one potential non-volatile memory device that has the potential to out perform both DRAM and flash memory. The speed of FeRAM is, in principle, as fast as DRAM (around 1 ns access time), however in practice FeRAM has yet to achieve this speed. In addition to the speed, FeRAM has a power consumption on the order of flash memory, but has a better read/write cycling lifetime.

The ferroelectric alloy PZT has potential for application as FeRAM. It is desirable to be able to tailor the properties of the ferroelectric, particularly the energy and applied field necessary to reverse the polarization of the data bit. Without adding a fifth component to PZT, the two simplest parameters to vary are the relative ratio of Ti to Zr in the alloy and the applied biaxial strain due to the lattice mismatch between the ferroelectric thin film and the substrate.

I began working on evaluating PZT during the summer of 2007 and will update this description as results are obtained. I am employing the DFT/pseudpotential methods encoded in ABINIT. I am investigating the effect of alloying by creating correlated special quasi-random structures. The polarization is determined by calculating the Berry phase.