Session W32: Focus Session: Dielectric, Ferroelectric, and Piezoelectric Oxides - Bandgaps, Surfaces, Oxides on Semiconductors

Band Gap and Edge Engineering of SrTiO$_{3}$ and Related Compounds via Ferroic Distortion and Anisotropic Strain

Due to its electronic band edge energies and its stability in water, the perovskite strontium titanate (SrTiO$_{3}$) is a promising water splitting photocatalyst. However, its ability to use solar photons to split water into hydrogen and oxygen would be more efficient if its wide optical band gap (3.2 eV) better matched the solar spectrum. Therefore, there is interest in modifying the crystal structure of SrTiO$_{3}$ (e.g., via chemical doping or epitaxial strain) to tune its electronic and optical properties. We use density functional theory (DFT) and many-body perturbation theory within the GW approximation to calculate the effects of structural and chemical modifications of SrTiO$_{3}$ on its band gap and edges. Much of our work (Berger, Fennie, and Neaton, Phys. Rev. Lett. 107, 146804 [2011]) focuses on the effects of epitaxial strain and the associated ferroic distortions. Anisotropic strains are shown to reduce the SrTiO$_{3}$ gap by breaking degeneracies at the band edges. Ferroic distortions are shown to widen the gap by allowing new band edge orbital mixings. To reduce the SrTiO$_{3}$ gap, one must lower the symmetry from cubic while suppressing ferroic distortions. Our calculations indicate that for engineered orientation of the growth direction along [111], the SrTiO$_{3}$ gap can be controllably and considerably reduced at room temperature. Chemical doping and substitution can, in combination with strain and distortion, further tune the band gap and edges. Our results and their favorable qualitative agreement with experiment suggest achievable paths toward engineering efficient solar water splitting catalysts, and more generally provide fundamental insight into the relationships between crystal and electronic structure in the property-rich perovskite family.   

Epitaxial strain tuning of polarization and band gap in perovksite SnTiO$_3$

Lead toxicity has motivated theoretical studies of a tin-based perovskite ferroelectric material. Density-functional calculations predict a polar perovksite ground state for SnTiO$_3$. Simulated epitaxial strain up to $\pm$2\% tunes both the magnitude of the polar distortion, its direction, and the electronic band gap --- compressive bi-axial strain creates the largest polar distortions, which occur entirely in the growth direction, while tensile strain reorients the polar displacements, enlarging the band gap. Projected densities of states indicate that the broken four-fold symmetry of the non-growth-oriented distortion allows Ti $d_{xy}$ bands to mix with O $p_x$ bands, further separating the valence band maximum and conduction band minimum. Comparing Sn and Pb in the perovskite titanate phases shows similar trends and suggests that SnTiO$_3$ ferroelectrics may be viable thin-film alternatives to Pb-based oxides.   

Ferroelectric Properties of TiO$_2$ Rutile Bulk and Surfaces Investigated by Ab Initio Calculations

TiO$_2$ rutile is an incipient ferroelectric material [Lee et al., Phys. Rev. B \textbf{50}, 13379 (1994)] and theoretical studies have shown that a ferroelectric transition can be enforced [Montanari et al., J. Phys. Condens. Matter \textbf{16}, 273 (2004)]. The experimental realization of ferroelectric TiO$_2$ (110) surfaces would have great technical impact and already the tuning of the electric permittivity would be of interest, e.g., for optical coating. In order to get an insight into the ferroelectric trends, we have studied the atomic and electronic structure, as well as phonon spectra, dipolar interactions and the polarization, using density functional theory (VASP and PWscf). Accordingly, ferroelectric states polarized along (110) and (001) can be stabilized within bulk. We demonstrate the different strain dependencies of the corresponding polar modes, which opens up the possibility of strain engineering the polarization direction, and the resulting dielectric response [Gr\"unebohm et al., Phys. Rev. B \textbf{84}, 132105 (2011) and Gr\"unebohm et al. arxiv:1111.2575]. Although the dipolar interaction and the short range repulsion are both modified at the (110) surface, stable local dipoles are obtained within the surface planes, which increase with increasing strain.   

XPS measurements of interface dipole switching at the a-Al2O3/Si interface

In this work, we describe work on polarization switching at a high-k oxide-silicon interface. The procedure involves inserting a monolayer (ML) of ZrO2 in-between an amorphous-Al2O3 and Si. Theoretical calculations using density functional theory (DFT) predict that the ZrO2 should display two stable configurations of the polarization. Deposition of the ZrO2 in UHV is used to avoid SiO2 formation. The device is transferred in vacuum and the interface chemistry analyzed using x-ray photoelectron spectroscopy (XPS) to determine the oxidation state of the Si. When the ZrO2 is in direct contact with the Si, chemical shifts as large as 0.58 [eV] are observed, implying a polar interface. In addition, XPS measurements on devices under applied voltage, along with electron transport measurements, show a switching of the interface dipole of 0.25 [eV].~ These voltage dependent XPS results are consistent with the magnitude and direction of hysteresis loops observed in Capacitance-Voltage measurements. Finally, the microscopic structure has been investigated using extended x-ray absorption fine structure (EXAFS) at the Zr K-edge. The results are compared to DFT-calculated atomic positions.   

Macroscopic electrostatics at the nanoscale: From ferroelectric capacitors to confined electron gases

Complex oxides are characterized by a multitude of coupled electronic and lattice degrees of freedom, and therefore constitute an unusually rich playground for experimentalists and theoreticians alike. These microscopic variables manifest themselves in macroscopically measurable quantities such as polarization, magnetization and strain, whose mutual coupling is sought after for applications in multifunctional electronic devices. To understand the interplay of these many factors, density functional theory (DFT) has proven an invaluable tool. However, the treatment of the macroscopic electrical variables (electric fields and polarization), which are a crucial ingredient in describing the experimentally observed response properties, has traditionally been difficult within first-principles calculations. In this talk I will first review a number of recent methodological developments that removed this limitation, thus extending the scopes of first-principles theory to the simulation of realistic devices within arbitrary electrical boundary conditions. Next, I will discuss the evolution of the band offset at a metal/ferroelectric interface as a function of polarization, and its implications for the electrical properties of nanocapacitors. Finally, I will show that, depending on the polarization of the film, a problematic regime might occur where the metallic carriers populate the energy bands of the insulator, making it metallic. As the most common approximations of density functional theory are affected by a systematic underestimation of the fundamental band gap of insulators, this scenario is likely to be an artifact of the simulation. I will discuss a number of criteria to systematically identify this situation in simulations, and effective modeling strategies to describe this peculiar charge compensation mechanism.   

Band Alignment of Plasma-Enhanced ALD High-k Dielectrics on Gallium Nitride

GaN-based transistors have shown immense promise because of their high saturation velocity and breakdown field, but their performance is limited by the high gate leakage. This limitation is mitigated with the use of metal/high-$k$ oxide/III-N structures. This experiment investigates three promising high-$k$ dielectrics deposited by plasma enhanced ALD: Al$_{2}$O$_{3}$, HfO$_{2}$, and La$_{2}$O$_{3}$. The band gaps of these materials are 6.5eV, 5.8eV, and 4.3eV, while the dielectric constants are 9, 20, and 27, respectively. The large band gap associated with Al$_{2}$O$_{3}$ reduces the leakage current; however, the lower dielectric constant increases the equivalent oxide thickness. The band alignment of the high-$k$ oxide/GaN interface plays a critical role in determining the confinement properties of semiconductor carriers and ultimately device performance. In situ photoemission gave valence band offsets for Al$_{2}$O$_{3}$, HfO$_{2}$, and La$_{2}$O$_{3}$ with GaN as 1.8eV, 1.3eV, and 0.9eV. The results are described by the charge neutrality level and interface dipole models. We also investigated the use of Al$_{2}$O$_{3}$ as an interfacial passivation layer between HfO$_{2}$ and GaN. This research is supported by the Office of Naval Research.   

High-k dielectrics on n-Al$_{0.25}$Ga$_{0.75}$N via atomic layer deposition

AlGaN/GaN and AlInN/GaN high-electron-mobility transistors (HEMTs) are promising devices for high-temperature and high-power electronics applications. A key issue with these devices is the high gate leakage current, particularly for enhancement-mode HEMTs. There has been an increased interest in developing high quality gate insulators to reduce gate leakage current. Al$_{2}$O$_{3 }$and HfO$_{2}$ layers (21nm thick)$_{ }$were deposited via atomic layer deposition on n-Al$_{0.25}$Ga$_{0.75}$N pretreated with one of two different surface preparations, H$_{2}$O$_{2}$:H$_{2}$SO$_{4}$ (1:5) (piranha) or HF:H$_{2}$O (1:3). Dielectrics were characterized using spectroscopic ellipsometry, X-ray photoelectron spectroscopy, atomic force microscopy (AFM), and capacitance-voltage (C-V) measurements. AFM shows that Al$_{2}$O$_{3 }$and HfO$_{2}$ layers are continuous and uniform in thickness on both HF and piranha pretreated surfaces. However, C-V measurement shows smaller (15{\%}) hysteresis for HF pretreated samples. The estimated dielectric constants ($\varepsilon )$ are 9 and 18 for Al$_{2}$O$_{3}$ and HfO$_{2}$ on HF pretreated surfaces, respectively, in general agreement with theoretical values of 9 and 25. Al$_{2}$O$_{3}$ layers on Al$_{0.25}$Ga$_{0.75}$N exhibited a lower leakage (7x10$^{-8}$ A/cm$^{2}$ at 5 V) current and higher forward breakdown voltage of 7.5 MV/cm compared to that of HfO$_{2}$ layer. The higher breakdown voltage and lower leakage current for Al$_{2}$O$_{3}$ is due to larger conduction band offset with Al$_{0.25}$Ga$_{0.75}$N.   

The growth of c-axis oriented BaTiO$_{3}$ films on Si and Ge for a non-volatile field-effect transistor

The reorientable polarization of a ferroelectric material can be utilized in a variety of applications, including the development of novel memory devices. Of particular interest is the use of a ferroelectric's spontaneous polarization to maintain ``on'' and ``off'' conductivity states in a field effect transistor. BaTiO$_{3}$ has been proposed as a Pb-free ferroelectric for such an application. Direct coupling of the ferroelectric polarization with the channel of a field effect transistor requires c-axis oriented BaTiO$_{3}$ films to be grown on Si or Ge. However, due to the small thermal expansion coefficients of Si and Ge, BaTiO$_{3}$ films tend to be a-axis oriented, having the polarization lying in the plane of the film. In order to achieve c-axis oriented BaTiO$_{3}$ films, we have developed a graded buffer layer that imparts in-plane compressive strain to overcome the incompatibility in thermal expansion. Ferroelectric, c-axis oriented, BaTiO$_{3}$ films with thicknesses exceeding 120 nm have been achieved. We will discuss the growth and characterization of these films in the development of a non-volatile, ferroelectric transistor.   

First principles analysis of the capacitance density of high-k nanocapacitors utilizing the orbital separation approach

In order to realize further scaling of electronic devices, capacitors and transistors with higher capacitance density are necessary. To this end, higher-k films of nanometer thickness are being considered for next-generation devices. However, high-k/metal interfaces often suffer from degraded dielectric properties due to, e.g., contamination, defects, and the intrinsic dead layer effect [1]. In this work, we developed a new first principles approach based on the Kohn-Sham formalism that we call orbital separation approach [2] to calculate the capacitance of nano-sized capacitors and examine the limit in capacitance density. In this method, the Kohn-Sham orbitals around the Fermi level of a metal/insulator/metal slab are separated into left and right electrode. The separated orbitals are occupied according to different Fermi levels to simulate the effect of bias voltage. We applied this approach to Au/MgO/Au and SrRuO$_3$/SrTiO$_3$/SrRuO$_3$ systems to examine the effect of the interface on the dielectric response. We confirmed that this method gives reasonable results. The impact of the intrinsic dead layer and that of defects on the capacitance are also examined. \\[4pt] [1] M. Stengel and N. Spaldin, Nature 443, 679 (2006).\\[0pt] [2] S. Kasamatsu et al., Phys. Rev. B 84, 085120 (2011).   

Properties of Ultrathin Al$_{2}$O$_{3}$-TiO$_{2}$ Nanolaminate Films for Gate Dielectric Applications Deposited by Plasma-Assisted Atomic Layer Deposition

High permittivity dielectrics such as Al$_{2}$O$_{3}$, HfO$_{2}$, Ta$_{2}$O$_{5}$, TiO$_{2}$, etc., are an essential component of aggressively-scaled III-V and graphene field effect transistors (FETs) where insulators are necessary to reduce gate leakage current while maintaining high gate capacitance and charge control of the channel. Atomic layer deposition (ALD) has the capability to deposit hybrid films, or nanolaminates, of two or more dielectrics that have unique properties. Thin [Al$_{2}$O$_{3}$+TiO$_{2}$] nanolaminates with varying TiO$_{2}$ and Al$_{2}$O$_{3}$ content were deposited on $n$-Si substrates at $\sim $225-300\r{ }C using ALD. A nanolaminate is composed of bilayers, defined as the sum of (x)Al$_{2}$O$_{3}$ and (y)TiO$_{2}$, where x, and y indicate the number of times a component monolayer is repeated. While the overall thickness of the dielectric was held at $\sim $ 17-20 nm, the relative ratio of Al$_{2}$O$_{3}$ to TiO$_{2}$ in the bilayer stack was varied to evaluate changes in the material properties and electrical performance of the oxides. C-V and I-V measurements on various [(x)TiO$_{2}$+(y)Al$_{2}$O$_{3}$] MOS capacitors were taken. The high-TiO$_{2}$-content films show limited evidence of oxide charge trapping and relatively large dielectric constants ($\kappa \sim $15), whereas the high-Al$_{2}$O$_{3}$-content films offer a larger optical bandgap and improved suppression of leakage current. We will discuss the properties of very thin nanolaminates and their possible use as gate oxides. Morphological, electrical, and XPS composition assessments will be presented.   

Ferroelectric Surface Chemistry: FIrst-principle study of NOx Decomposition

NOx molecules are critical and regulated air pollutants produced during automotive combustion. As part of a long-term effort to design viable catalysts for NOx decomposition that operate at higher temperatures and thus would allow for greater fuel efficiency, we are studying NOx chemistry on ferroelectric perovskite surfaces. Changing the direction of the ferroelectric polarization can modify surface properties and thus can lead to switchable surface chemistry. We will discuss our results for NO and NO2 on the polar (001) surfaces of PbTiO3 as function of ferroelectric polarization, surface stoichiometry, and various molecular or dissociated binding modes.