Techn. Hochsch., Diss., 2002--Aachen.

Ferroelectrics are promising for nonvolatile memories. However, the difficulty of fabricating ferroelectric layers and integrating them into complementary metal oxide semiconductor (CMOS) devices has hindered rapid scaling. Hafnium oxide is a standard material available in CMOS processes. Ferroelectricity in Si-doped hafnia was first reported in 2011, and this has revived interest in using ferroelectric memories for various applications. Ferroelectric hafnia with matured atomic layer deposition techniques is compatible with three-dimensional capacitors and can solve the scaling limitations in 1-transistor-1-capacitor (1T-1C) ferroelectric random-access memories (FeRAMs). For ferroelectric field-effect-transistors (FeFETs), the low permittivity and high coercive field Ec of hafnia ferroelectrics are beneficial. The much higher Ec of ferroelectric hafnia, however, makes high endurance a challenge. This article summarizes the current status of ferroelectricity in hafnia and explains how major issues of 1T-1C FeRAMs and FeFETs can be solved using this material system.

Ferroelectric HfO₂ attracts a huge amount of attention not only for memory and negative capacitance, but also for programmable logic including memory-in-logic and neuromorphic applications. However, the understanding of material fundamentals still needs to be improved. This paper gives material fundamentals and new insights to this ferroelectric material for future device applications. In particular, the key role of dopants, effects of the interface on the ferroelectric phase, and a detailed discussion of the switching kinetics are of central focus. Based on material properties newly obtained, we discuss opportunities of ferroelectric HfO₂ for device applications.

Manifold research has been done to understand the detailed mechanisms behind the performance instabilities of ferroelectric capacitors based on hafnia. The wake-up together with the imprint might be the most controversially discussed phenomena so far. Among crystallographic phase change contributions and oxygen vacancy diffusion, electron trapping as the origin has been discussed recently. In this publication, we provide evidence that the imprint is indeed caused by electron trapping into deep states at oxygen vacancies. This impedes the ferroelectric switching and causes a shift of the hysteresis. Moreover, we show that the wake-up mechanism can be caused by a local imprint of the domains in the pristine state by the very same root cause. The various domain orientations together with an electron trapping can cause a constriction of the hysteresis and an internal bias field in the pristine state. Additionally, we show that this local imprint can even cause almost anti-ferroelectric like behavior in ferroelectric films.

An abstract is not available for this content.

Visualization of domain structure evolution under an electrical bias has been carried out in ferroelectric La:HfO2 capacitors by a combination of Piezoresponse Force Microscopy (PFM) and pulse switching techniques to study the nanoscopic mechanism of polarization reversal and the wake-up process. It has been directly shown that the main mechanism behind the transformation of the polarization hysteretic behavior and an increase in the remanent polarization value upon the alternating current cycling is electrically induced domain de-pinning. PFM imaging and local spectroscopy revealed asymmetric switching in the La:HfO2 capacitors due to a significant imprint likely caused by the different boundary conditions at the top and bottom interfaces. Domain switching kinetics can be well-described by the nucleation limited switching model characterized by a broad distribution of the local switching times. It has been found that the domain velocity varies significantly throughout the switching process indicating strong interaction with structural defects.

Ferroelectricity is very attractive for nonvolatile memories since it allows non-volatility paired with a field driven switching mechanism enabling a very low-power write operation. Non-volatile memories based on ferroelectric lead-zirconium-titanate (PZT) (see fig. la) are available on the market for more than a quarter of a century now [1]. Yet they are limited to niche applications due to the compatibility issues of the ferroelectric material with CMOS processes and the associated limited scalability [2]. The discovery of ferroelectricity in doped hafnium oxide has revived the activities towards a variety of scalable ferroelectric nonvolatile memory devices

Morphotrope PZT-Keramiken sind Festkörperlösungen aus Bleizirkonat und Bleititanat mit chemischen Zusammensetzungen um die 47% Ti-Anteil. Sie weisen im gepolten Zustand die größten piezoelektrischen Koppelkonstanten auf und sind daher von speziellem Interesse. Zur Vorhersage des Polungszustandes und der Bauteilfestigkeit in komplexen Bauteilen werden elektromechanisch gekoppelte Materialmodelle benötigt. In dieser Arbeit wird ein mikromechanischer Modellansatz aus der Literatur aufgegriffen. Ausgangspunkt ist ein dreidimensionales tetragonales Modell, welches ein repräsentatives Volumenelement des Kornverbundes und ein mikroskopisches Kornmodell vereint. Damit gelingt die Beschreibung der Korninteraktionen infolge unterschiedlicher Polungszustände der Körner. Die Domänenstruktur der Körner wird mittels der Volumenanteile der kristallographischen Varianten dargestellt. Ein vereinfachter Satz an mikroskopischen Materialkonstanten wird anhand experimenteller Daten und theoretischer Betrachtungen aus der Literatur abgeleitet. Die für zwei Lastfälle berechneten makroskopischen Materialantworten zeigen explizit, dass das tetragonale Modell nicht in der Lage ist, das Verhalten morphotroper PZT-Keramiken nachzubilden. Aus diesem Grund wird das Modell im Hinblick auf die besondere kristallographische Struktur morphotroper PZT-Keramiken um eine rhomboedrische Phase in veränderlichen Anteilen erweitert. Die somit berechneten makroskopischen Antworten stimmen sowohl quantitativ als auch qualitativ gut mit experimentellen Ergebnissen überein. Der Einfluss der im Modell berücksichtigten Kristallstruktur auf die makroskopische Materialantwort wird in der Arbeit ausführlich analysiert. / Morphotropic PZT ceramics are solid solutions made of lead zirconate and lead titanate with chemical composition around 47% Ti-content. When poled they possess the greatest piezoelectric coupling constants for which they are of special interest. Predicting the poling condition and the strength in complex devices requires electromechanically coupled material models. Within this work, a micromechanical modelling approach is utilised. Starting point is a three-dimensional tetragonal model, which combines a representative volume element of the grain compound together with a microscopic grain model. This allows the consideration of grain interaction due to different poling conditions of the grains. The domain structure of the grains is captured by volume fractions of the crystallographic variants. A simplified set of microscopic material constants is derived from experimental and theoretical data given in the literature. The macroscopic material response, which is computed for two load cases, shows explicitly that the tetragonal model is not capable of reproducing the behaviour of morphotropic PZT ceramics. Therefore, the model is extended by the rhombohedral phase in varying quantity with view of the specific crystallographic structure of morphotropic PZT ceramics. The so computed macroscopic response shows a quantitatively as well as qualitatively good agreement with experimental results. The effect of the crystallographic structure which is considered within the model on the macroscopic material response is extensively analysed.

Morphotrope PZT-Keramiken sind Festkörperlösungen aus Bleizirkonat und Bleititanat mit chemischen Zusammensetzungen um die 47% Ti-Anteil. Sie weisen im gepolten Zustand die größten piezoelektrischen Koppelkonstanten auf und sind daher von speziellem Interesse. Zur Vorhersage des Polungszustandes und der Bauteilfestigkeit in komplexen Bauteilen werden elektromechanisch gekoppelte Materialmodelle benötigt. In dieser Arbeit wird ein mikromechanischer Modellansatz aus der Literatur aufgegriffen. Ausgangspunkt ist ein dreidimensionales tetragonales Modell, welches ein repräsentatives Volumenelement des Kornverbundes und ein mikroskopisches Kornmodell vereint. Damit gelingt die Beschreibung der Korninteraktionen infolge unterschiedlicher Polungszustände der Körner. Die Domänenstruktur der Körner wird mittels der Volumenanteile der kristallographischen Varianten dargestellt. Ein vereinfachter Satz an mikroskopischen Materialkonstanten wird anhand experimenteller Daten und theoretischer Betrachtungen aus der Literatur abgeleitet. Die für zwei Lastfälle berechneten makroskopischen Materialantworten zeigen explizit, dass das tetragonale Modell nicht in der Lage ist, das Verhalten morphotroper PZT-Keramiken nachzubilden. Aus diesem Grund wird das Modell im Hinblick auf die besondere kristallographische Struktur morphotroper PZT-Keramiken um eine rhomboedrische Phase in veränderlichen Anteilen erweitert. Die somit berechneten makroskopischen Antworten stimmen sowohl quantitativ als auch qualitativ gut mit experimentellen Ergebnissen überein. Der Einfluss der im Modell berücksichtigten Kristallstruktur auf die makroskopische Materialantwort wird in der Arbeit ausführlich analysiert. / Morphotropic PZT ceramics are solid solutions made of lead zirconate and lead titanate with chemical composition around 47% Ti-content. When poled they possess the greatest piezoelectric coupling constants for which they are of special interest. Predicting the poling condition and the strength in complex devices requires electromechanically coupled material models. Within this work, a micromechanical modelling approach is utilised. Starting point is a three-dimensional tetragonal model, which combines a representative volume element of the grain compound together with a microscopic grain model. This allows the consideration of grain interaction due to different poling conditions of the grains. The domain structure of the grains is captured by volume fractions of the crystallographic variants. A simplified set of microscopic material constants is derived from experimental and theoretical data given in the literature. The macroscopic material response, which is computed for two load cases, shows explicitly that the tetragonal model is not capable of reproducing the behaviour of morphotropic PZT ceramics. Therefore, the model is extended by the rhombohedral phase in varying quantity with view of the specific crystallographic structure of morphotropic PZT ceramics. The so computed macroscopic response shows a quantitatively as well as qualitatively good agreement with experimental results. The effect of the crystallographic structure which is considered within the model on the macroscopic material response is extensively analysed.

Different causes for ferroelectric properties in hafnium oxide were discussed during the last decade including various dopants, stress, electrode materials, and surface energy from different grain sizes. Recently, the focus shifted to the impact of oxygen vacancies on the phase formation process. In this progress report, the recent understanding of the influence of oxygen supplied during deposition on the structural phase formation process is reviewed and supplemented with new data for mixed HfₓZr₁₋ₓOᵧ films. Even though polar and non-polar HfₓZr₁₋ₓOᵧ thin films are well characterized, little is known about the impact of oxygen exposure during the deposition process. Here, a combination of structural and electrical characterization is applied to investigate the influence of the oxygen and zirconium content on the crystallization process during ALD deposition in comparison to other deposition techniques. Different polarization properties are assessed which correlate to the determined phase of the film. Optimized oxygen pulse times can enable the crystallization of HfₓZr₁₋ₓOᵧ in a polar orthorhombic phase rather than a non-polar monoclinic and tetragonal phase.