arXiv subjects

Christopher Künneth

Publications and source records attributed to Christopher Künneth.

8 recordsLinked to original sources

arXiv2021

Copolymer Informatics with Multi-Task Deep Neural Networks

Polymer informatics tools have been recently gaining ground to efficiently and effectively develop, design, and discover new polymers that meet specific application needs. So far, however, these data-driven efforts have largely focused on homopolymers. Here, we address the property prediction challenge for copolymers, extending the polymer informatics framework beyond homopolymers. Advanced polymer fingerprinting and deep-learning schemes that incorporate multi-task learning and meta-learning are proposed. A large data set containing over 18,000 data points of glass transition, melting, and degradation temperature of homopolymers and copolymers of up to two monomers is used to demonstrate the copolymer prediction efficacy. The developed models are accurate, fast, flexible, and scalable to more copolymer properties when suitable data become available.

cond-mat.mtrl-sci↗

arXiv2020

Polymer Informatics with Multi-Task Learning

Modern data-driven tools are transforming application-specific polymer development cycles. Surrogate models that can be trained to predict the properties of new polymers are becoming commonplace. Nevertheless, these models do not utilize the full breadth of the knowledge available in datasets, which are oftentimes sparse; inherent correlations between different property datasets are disregarded. Here, we demonstrate the potency of multi-task learning approaches that exploit such inherent correlations effectively, particularly when some property dataset sizes are small. Data pertaining to 36 different properties of over $13, 000$ polymers (corresponding to over $23,000$ data points) are coalesced and supplied to deep-learning multi-task architectures. Compared to conventional single-task learning models (that are trained on individual property datasets independently), the multi-task approach is accurate, efficient, scalable, and amenable to transfer learning as more data on the same or different properties become available. Moreover, these models are interpretable. Chemical rules, that explain how certain features control trends in specific property values, emerge from the present work, paving the way for the rational design of application specific polymers meeting desired property or performance objectives.

cond-mat.mtrl-sci↗

arXiv2018

Piezospectroscopy and first-principles calculations of the nitrogen-vacancy center in gallium arsenide

The nitrogen-vacancy (NV) center occurs in GaAs bulk crystals doped or implanted with nitrogen. The local vibration of nitrogen gives rise to a sharp infrared absorption band at 638 cm$^{-1}$, exhibiting a fine structure due to the different masses of neighboring $^{69}$Ga and $^{71}$Ga host isotopes. Piezospectroscopic investigations in the crystallographic <100> direction prove that the center has C$_\text{3v}$ point symmetry, which is weakly perturbed by the isotope effect. The stress-induced shifts of some band components show an unusual non-linear behavior that can be explained by coupling between the isotope and the stress splitting. First-principles density-functional theory calculations are in full accordance with the experiments and confirm the C$_\text{3v}$ symmetry, caused by relaxation of the nitrogen atom from the anion lattice site towards the nearest-neighbor Ga plane. The NV center in GaAs is structurally analogous to the same center in diamond. The $-3$ charge state is most stable for nearly all Fermi-level positions.

cond-mat.mtrl-sci↗

arXiv2018

Symmetry and structure of carbon-nitrogen complexes in gallium arsenide from infrared spectroscopy and first-principles calculations

Molecular-like carbon-nitrogen complexes in GaAs are investigated both experimentally and theoretically. Two characteristic high-frequency stretching modes at \num{1973} and \SI{2060}{cm^{-1}}, detected by Fourier transform infrared absorption (FTIR) spectroscopy, appear in carbon- and nitrogen-implanted and annealed layers. From isotopic substitution it is deduced that the chemical composition of the underlying complexes is CN$_2$ and C$_2$N, respectively. Piezospectroscopic FTIR measurements reveal that both centers have tetragonal symmetry. For density functional theory (DFT) calculations linear entities are substituted for the As anion, with the axis oriented along the \hkl<100> direction, in accordance with the experimentally ascertained symmetry. The DFT calculations support the stability of linear N-C-N and C-C-N complexes in the GaAs host crystal in the charge states ranging from $+3$ to $-3$. The valence bonds of the complexes are analyzed using molecular-like orbitals from DFT. It turns out that internal bonds and bonds to the lattice are essentially independent of the charge state. The calculated vibrational mode frequencies are close to the experimental values and reproduce precisely the isotopic mass splitting from FTIR experiments. Finally, the formation energies show that under thermodynamic equilibrium CN$_2$ is more stable than C$_2$N.

cond-mat.mtrl-sci↗

arXiv2018

Impact of Four-Valent Doping on the Crystallographic Phase Formation for Ferroelectric HfO$_2$ from First-Principles: Implications for Ferroelectric Memory and Energy-Related Applications

The ferroelectric properties of nanoscale silicon doped HfO$_2$ promise a multitude of applications ranging from ferroelectric memory to energy-related applications. The reason for the unexpected behavior has not been clearly proven and presumably include contributions from size effects and doping effects. Silicon incorporation in HfO$_2$ is investigated computationally by first-principles using different density functional theory (DFT) methods. Formation energies of interstitial and substitutional silicon in HfO$_2$ paired with and without an oxygen vacancy prove the substitutional defect as the most likely. Within the investigated concentration window up to 12.5 formula unit %, silicon doping alone is not sufficient to stabilize the polar and orthorhombic crystal phase (p-o-phase), which has been identified as the source of the ferroelectricity in HfO$_2$. On the other hand, silicon incorporation is one of the strongest promoters of the p-o-phase and the tetragonal phase (t-phase) within the group of investigated dopants, confirming the experimental ferroelectric window. Besides silicon, the favoring effects on the energy of other four-valent dopants, C, Ge, Ti, Sn, Zr and Ce, are examined, revealing Ce as a very promising candidate. The evolution of the volume changes with increasing doping concentration of these four-valent dopants shows an inverse trend for Ce in comparison to silicon. To complement this study, the geometrical incorporation of the dopants in the host HfO$_2$ lattice was analyzed.

cond-mat.mtrl-sci↗

arXiv2017

A computational study of hafnia-based ferroelectric memories: from ab initio via physical modeling to circuit models of ferroelectric device

The discovery of ferroelectric properties of binary oxides revitalized the interest in ferroelectrics and bridged the scaling gap between the state-of-the-art semiconductor technology and ferroelectric memories. However, before hitting the markets, the origin of ferroelectricity and in-depth studies of device characteristics are needed. Establishing a correlation between the performance of the device and underlying physical mechanisms is the first step toward understanding the device and engineering guidelines for a novel, superior device. Therefore, in this paper a holistic modeling approaches which lead to a better understanding of ferroelectric memories based on hafnium and zirconium oxide is addressed. Starting from describing the stabilization of the ferroelectric phase within the binary oxides via physical modeling the physical mechanisms of the ferroelectric devices are reviewed. Besides, limitations and modeling of the multilevel operation and switching kinetics of ultimately scaled devices as well as the necessity for Landau-Khalatnikov approach are discussed. Furthermore, a device-level model of ferroelectric memory devices that can be used to study the array implementation and their operational schemes are addressed. Finally, a circuit model of the ferroelectric memory device is presented and potential further applications of ferroelectric devices are outlined.

cond-mat.mtrl-sci↗

arXiv2017

The impact of charge compensated and uncompensated strontium defects on the stabilization of the ferroelectric phase in HfO$_2$

Different dopants with their specific dopant concentration can be utilized to produce ferroelectric HfO$_2$ thin films. In this work it is explored for the example of Sr in a comprehensive first-principles study. Density functional calculations reveal structure, formation energy and total energy of the Sr related defects in HfO$_2$. We found the charge compensated defect including an associated oxygen vacancy Sr$_\text{Hf}$V$_\text{O}$ to strongly favour the non-ferroelectric, tetragonal P4$_\text{2}$/mnc phase energetically. In contrast, the uncompensated defect without oxygen vacancy Sr$_\text{Hf}$ favours the ferroelectric, orthorhombic Pca2$_\text{1}$ phase. According to the formation energy the uncompensated defect can form easily under oxygen rich conditions in the production process. Low oxygen partial pressure existing over the lifetime promotes the loss of oxygen leading to V$_\text{O}$ and, thus, the destabilization of the ferroelectric, orthorhombic Pca2$_\text{1}$ phase accompanied by an increase of the leakage current. This study attempts to fundamentally explain the stabilization of the ferroelectric, orthorhombic Pca2$_\text{1}$ phase by doping.

cond-mat.mtrl-sci↗

arXiv2015

The Origin of Ferroelectricity in Hf$_{x}$ Zr$_{1-x}$ O$_2$: A Computational Investigation and a Surface Energy Model

The structural, thermal, and dielectric properties of the ferroelectric phase of HfO$_2$, ZrO$_2$ and Hf$_{0.5}$ Zr$_{0.5}$ O$_2$ (HZO) are investigated with carefully validated density functional computations. We find, that the free bulk energy of the ferroelectric orthorhombic Pca2$_{1}$ phase is unfavorable compared to the monoclinic P2$_{1}$/c and the orthorhombic Pbca phase for all investigated stoichiometries in the Hf$_χ$Zr$_{1-χ}$O$_2$ system. To explain the existence of the ferroelectric phase in nanoscale thin films we explore the Gibbs / Helmholtz free energies as a function of stress and film strain and find them unlikely to become minimal in HZO films for technological relevant conditions. To assess the contribution of surface energy to the phase stability we parameterize a model, interpolating between existing data, and find the Helmholtz free energy of ferroelectric grains minimal for a range of size and stoichiometry. From the model we predict undoped HfO$_2$ to be ferroelectric for a grain size of about 4 nm and epitaxial HZO below 5 nm. Furthermore we calculate the strength of an applied electric field necessary to cause the antiferroelectric phase transformation in ZrO$_2$ from the P4$_2$/nmc phase as 1 MV/cm in agreement with experimental data, explaining the mechanism of field induced phase transformation.

cond-mat.mtrl-sci↗