Skip to main content

Role of oxygen vacancies in ferroelectric or resistive switching hafnium oxide

Abstract

HfO2 shows promise for emerging ferroelectric and resistive switching (RS) memory devices owing to its excellent electrical properties and compatibility with complementary metal oxide semiconductor technology based on mature fabrication processes such as atomic layer deposition. Oxygen vacancy (Vo), which is the most frequently observed intrinsic defect in HfO2-based films, determines the physical/electrical properties and device performance. Vo influences the polymorphism and the resulting ferroelectric properties of HfO2. Moreover, the switching speed and endurance of ferroelectric memories are strongly correlated to the Vo concentration and redistribution. They also strongly influence the device-to-device and cycle-to-cycle variability of integrated circuits based on ferroelectric memories. The concentration, migration, and agglomeration of Vo form the main mechanism behind the RS behavior observed in HfO2, suggesting that the device performance and reliability in terms of the operating voltage, switching speed, on/off ratio, analog conductance modulation, endurance, and retention are sensitive to Vo. Therefore, the mechanism of Vo formation and its effects on the chemical, physical, and electrical properties in ferroelectric and RS HfO2 should be understood. This study comprehensively reviews the literature on Vo in HfO2 from the formation and influencing mechanism to material properties and device performance. This review contributes to the synergetic advances of current knowledge and technology in emerging HfO2-based semiconductor devices.

Graphical Abstract

1 Introduction

Among various candidate metal oxides, HfO2 has shown the most promise for emerging semiconductor devices, including ferroelectric and RS memories. Table 1 compares the various properties of insulating metal oxides frequently used or potentially considered in the semiconductor industry. HfO2 has an electrical bandgap as high as 5.0–6.0 eV and a sufficiently high crystallographic-structure-dependent dielectric constant of 20–40. Although the bandgap of HfO2 is lower than those of SiO2 (approximately 9 eV) and Al2O3 (approximately 7 eV), it can form a significantly high Schottky barrier upon contact with high-work-function metals. Moreover, the dielectric constant of monoclinic or amorphous HfO2 is approximately 4–5 times higher than that of SiO2. This value can be further enhanced with the higher-k metastable phases such as tetragonal, cubic, and orthorhombic phases. Therefore, HfO2 is suitable for insulating high-k materials while preventing direct electrical conduction by enabling the penetration of a relatively high fraction of electric field to the neighboring materials (e.g., the semiconductor channel in metal oxide semiconductor field effect transistors, MOSFETs). Hence, HfO2 has emerged as the most frequently and early considered high-k material among numerous candidates.

Table 1 Comparison of properties of various metal oxides (potentially) utilized in semiconductor processes

HfO2 became the first material to replace SiO2 as the gate insulator in a commercial semiconductor chipset with its adoption in Intel’s Penryn processor in 2007 via atomic layer deposition (ALD). The ALD technique is suitable for homogeneous and uniform film deposition even on complex nanostructures. Moreover, the technique controls thickness with atomic-level accuracy, enabling ferroelectricity or RS in nanoscale semiconductor devices for ultra-large-scale integrated circuits. Various Hf precursors and reactants were adopted to deposit ferroelectric or RS HfO2 as well as conventional gate-insulating HfO2 in MOSFETs [1,2,3,4].

The characteristic electrical, physical, and chemical properties of HfO2 are based on their fluorite crystallographic structure. It is known that HfO2 exists in various polymorphs, including the monoclinic (P21/c), tetragonal (P42/nmc), orthorhombic (Pca21 or Pbca), and cubic (Fm-3m) phases. The eight oxygen ions are located at the tetrahedral sites and have three or four ionic bonds with the surrounding metal ions, and the hafnium ions have seven or eight coordination numbers. The differences in the chemical bonding and the ensuing crystallographic structures are particularly important for the ferroelectric properties because ferroelectricity originates from broken centrosymmetry, which results in reversible spontaneous polarization states. Figure 1 shows the crystallographic structure of various polymorphs of HfO2. The cubic phase is the fluorite-structure-based crystallographic phase with the highest multiplicity and the largest number of symmetry elements. This phase is known to be stable when the temperature is higher than 2,573 K owing to its high configurational entropy [5]. All the other polymorphs can be constructed from the cubic phase by partially removing the symmetry elements of the cubic phase. The scope of this study does not permit a discussion on all possible polymorphs of HfO2. Hence, we only expound on the frequently observed crystallographic phases such as monoclinic, tetragonal, cubic, and Pca21 and Pbca orthorhombic phases.

Fig. 1
figure 1

Frequently observed crystallographic phase of HfO2, presenting the lattice parameters, unit cell volume, and coordination number of Hf and oxygen ions in each phase

Given the coordination numbers of oxygen and hafnium ions, these polymorphs can be categorized into two groups, one comprising the monoclinic, and Pca21 and Pbca orthorhombic phases and the other comprising the tetragonal and cubic phases. In the former group, the coordination number of hafnium ion is seven, whereas that of oxygen ions is three or four. In the latter group, the coordination number of hafnium ion is eight, whereas that of oxygen ion is four. When the lattice parameters and unit cell volumes of the polymorphs are considered, the monoclinic phase is distinct from the other phases with larger lattice parameters and beta angle higher than 90°. Therefore, the resulting unit cell volume is also considerably different (~ 3.6 to 6.9% larger). To summarize, the ferroelectric Pca21 orthorhombic phase is similar to the monoclinic phase when considering the ionic coordination numbers, whereas it is similar to the tetragonal or cubic phase when considering the unit cell volumes and lattice angles.

Moreover, the Pbca orthorhombic phase is highly relevant to the Pca21 orthorhombic phase. Materlik et al. [6] reported that inducing two opposite polarization states in Pca21 orthorhombic phase unit cells generates the Pbca orthorhombic phase. Thus, the Pbca orthorhombic phase can be called antipolar phase, which is frequently considered for the origin of the antiferroelectric properties according to the classical Kittel’s model [7]. However, according to experimentally observed temperature-dependent changes in ferroelectric–paraelectric transitions, the antiferroelectricity in HfO2 is believed to originate from the reversible field-induced phase transition between the tetragonal and Pca21 orthorhombic phases rather than that between the Pbca and Pca21 orthorhombic phases [8, 9]. Interestingly, the antipolar Pbca orthorhombic phase has a lower free energy than the polar Pca21 orthorhombic phase under almost every temperature and pressure range examined in previous computational simulations. Therefore, the formation of the Pbca phase is highly probable. Recently, it was elucidated that the formation of the electrical cycling driven formation of the Pbca phase is the crystallographic origin of low field fatigue observed in ferroelectric Hf0.5Zr0.5O2 (HZO) thin films [9]. Although this phenomenon was observed at room temperature, a similar phenomenon is expected over wide ranges of temperature and pressure, when small free-energy and configurational-entropy differences between the two orthorhombic phases arise from their crystallographic similarity. Thus, it is believed that the antipolar Pbca orthorhombic phase should be further studied to understand the complex nanoscale polymorphism of HfO2-based thin films.

Notably, other interesting non-centrosymmetric crystallographic phases of HfO2 were reported in theoretical and experimental works. The Pmn21 orthorhombic phase is another interesting polar crystallographic phase with remarkably high spontaneous polarization and a small energy difference with the stable monoclinic or polar Pca21 orthorhombic phase. However, to the best of the authors’ knowledge, this phase has not been experimentally observed [10]. Rhombohedral R3 or R3m phases are also other interesting polar phases, which have been observed in epitaxial HfO2-based ferroelectric thin films fabricated using pulsed laser deposition (PLD) [11]. However, the Pca21 orthorhombic phase would be more focused on in this study.

The RS characteristics in HfO2 are less sensitive to polymorphism, unlike ferroelectricity that is deterministically affected by the crystallographic phase. Conversely, the degree of Vo formation in the different fluorite crystallographic structures of HfO2 strongly influences the RS characteristics. For example, there is a difference in the magnitude of initial electroforming voltage depending on the number of Hf–O bonds of a unit cell in the crystalline phase, as reported by S. U. Sharath et al. [12]. However, in the case of Vo-mediated RS characteristics, both the crystallographic phase and stoichiometry or oxygen deficiency determined during the deposition or fabrication process can be considered significant factors influencing the intrinsic Vo defect. The electromigration or accumulation of Vo under an applied voltage and the consequent formation of a specific conductive pathway called the conductive filament (CF) should also be essentially considered. HfO2 has been studied as a RS material since its early research stage because of its CMOS compatibility and the absence of an oxygen-deficient subphase, such as the Magnéli phase in TiO2. In RS memory devices, the HfO2 layer is used as a switching layer between an inert and a reactive metal electrode, which illustrates the RS operation attributed to Vo dynamics. From these studies, it can be concluded that the RS characteristics can be controlled by adjusting the formation, migration, and accumulation of Vo. In addition, it is easy to form the amorphous and crystalline phases, including the grain and grain boundary engineering, depending on the deposition and post-processing conditions, resulting in different properties according to each influence. The subsequent switching characteristics after electroformation are also strongly correlated with these above oxygen vacancy-related factors. Several outstanding properties of HfO2-based RS memory devices, such as large memory window, multi-level storage ability, low power consumption [13,14,15], and CMOS compatibility [16] with possible high-density or three-dimensional integration, have attracted a considerable research strong interest for their application in neuromorphic computing [17,18,19], in particular for the development of artificial synapses or neurons in neural networks. Most studies that reported Vo-mediated RS behavior in HfO2 focused on Vo-related factors and their relationships with electroformation and the subsequent switching performance to realize high-precision multilevel operation, and on the analog weight update in the synaptic application to exploit the resistance dynamics of memory devices. Similar to biological synapses, nonvolatile RS devices act as memory components, storing the strength of synaptic connections without power. However, for these RS devices to fully mimic synaptic behavior, they must meet specific requirements, including forming voltage, variation, switching behavior, switching speed, switching voltage, conductance modulation, endurance, and retention. As the Vo dynamics in an RS device significantly affect these multiple performance aspects, improving Vo controllability is crucial to accurately implement synaptic operations in artificial systems. In this review, we discuss several technical solutions to control the Vo dynamics for enhancing the performance of HfO2-based RS devices.

As mentioned earlier in this section, Vo might significantly influence the ferroelectricity and RS of HfO2, suggesting that a meticulous theory of the chemical, physical, and electrical properties of Vo and their behavior in HfO2 must be developed. Furthermore, from the similarity between the operating electric field and various pieces of experimental evidence, a comprehensive comparison and in the Vo effects on ferroelectricity and RS HfO2 would be helpful for synergistically advancing our knowledge on HfO2-based ferroelectric and RS memories. In this review, therefore, the physical/electrical effects, formation mechanism, and ferroelectricity and RS effects of Vo in HfO2-based thin films are discussed.

The structure of the review is as follows. In Chapter 2, the Vo formation, migration, and accumulation in HfO2 thin films are detailed as investigated in extant literature. In Chapters 3 and 4, the effects of Vo on ferroelectricity and RS in HfO2 thin films are reviewed, and the strategies to improve the ferroelectricity or RS are discussed based on the literature. In Chapter 5, Vo-mediated ferroelectricity and RS behavior are expounded. Based on this, we provide strategies to engineer devices for ferroelectric and RS memory. Finally, the conclusions and future outlook are presented in Sect. 5.

2 Oxygen vacancy formation, migration, and accumulation in HfO2

Vo is the most frequently observed intrinsic point defect in metal oxide thin films or bulk. The existence of Vo induces an imperfect bonding state in crystals with three-dimensional periodic arrays of atoms, ions, molecules, or their groups. Consequently, the enthalpy of materials increases with the increasing concentration of Vo. However, the configurational entropy would decrease for a specific range of Vo concentration. Thus, there is an equilibrium Vo concentration which is strongly affected by temperature. Generally, the equilibrium vacancy concentration increases with increasing temperature because the impact of the entropy term on the Gibbs free energy is enhanced with temperature. Furthermore, once formed vacancies tend to be remained without oxygen providing atmosphere, so the high temperature processes during the material fabrication is critical for the Vo concentration. To summarize, thermodynamically, finite Vo should exist in metal-oxide thin films, and it is same for the fluorite-structure oxides such as hafnia and zirconia, which are frequently utilized in classical (MOSFET or dynamic random-access memory, DRAM) or emerging semiconductor devices (ferroelectric memories and memristors).

The existence of Vo strongly influences the physical/chemical properties of metal oxides. Electrostatically, the Vo form electron trap levels in the forbidden bandgap and contribute to the local conduction, which is generally considered harmful for the insulating layer [20]. For memristor applications, however, the filament formation by accumulated Vo is one of the RS mechanisms [21]. For the case of ferroelectric (Hf,Zr)O2-based films, Vo have been reported as one of the many factors affecting the polymorphism that produces ferroelectricity [22, 23]. Charged Vo are mobile under a high electric field (in the order of MV/cm), so their drift is considered a major parameter impacting the dynamic changes in performances of electronic devices with metal oxides.

This discussion highlights the importance of understanding the physics and chemistry of Vo in (Hf,Zr)O2-based materials for improving the performance of semiconductor devices, such as ferroelectric memories and memristors. In this section, the factors influencing their formation, such as deposition conditions, dopants, and electrode materials, are discussed. Subsequently, the observation of Vo migration through a transmission electron microscope (TEM) and their roles are addressed. Lastly, the aggregation of Vo and the underlying conduction mechanism are discussed.

2.1 Factors affecting the formation of Vo

2.1.1 Deposition condition

Various deposition techniques such as chemical solution deposition (CSD) [24,25,26,27], sputtering [28,29,30,31], PLD [32], and ALD [1] are used to deposit the (Hf,Zr)O2 film. In the CSD method, the precursor chemicals are dissolved in a solvent, and the solution is deposited on the substrate through spin or spray coating. The subsequent drying removes the solvent, resulting in an amorphous film [25]. The CSD is economical, suitable for mass production, and does not require a vacuum environment; however, it is difficult to decrease the concentration of impurities to extremely low levels required for modern CMOS technology. Various research groups have investigated CSD-grown HfO2-based thin films [24, 26, 27, 33, 34]. However, the effect of oxygen vacancies, which is the main topic of this review, has not been studied frequently in CSD-grown HZO thin films compared to that grown using other deposition techniques. Starschich et al. demonstrated that the drift of oxygen vacancies is the origin of both the wake-up effect and RS in CSD grown ferroelectric HZO thin films [35], indicating that understanding the effects of oxygen vacancies is critical for both ferroelectric and resistive memories based on (Hf,Zr)O2 thin films.

Sputtering is a physical vapor deposition (PVD) technique in which ionized gas molecules in plasma are accelerated toward the target to induce ion bombardment. Subsequently, the atoms in the target eject and fly from the target to the substrate. Sputtering offers a low-pressure deposition process favorable for fabricating films with a low concentration of impurities such as C, H, and N [31]. Likewise, PLD is a PVD technique wherein a pulsed laser beam is focused on a target, resulting in the formation of a plasma plume and the growth of the film on the substrate [36]. In PVD techniques, the ratio of the O2/Ar gas flow and pressure can affect the concentration of Vo and the film properties [29, 31, 37]. Jaszewski et al. studied the effect of the O2/Ar gas flow ratio on the concentration of Vo in the HfOx film using reactive sputtering. The value of x in HfOx obtained from the low-loss electron energy loss spectroscopy (EELS) spectra increased from 1.51 to 1.68 as the proportion of O2 gas in the plasma increased from 7.4 to 8.0%. Thus, the amount of Vo in the HfO2 film decreases with an increase in the O2 gas proportion in the plasma during reactive sputtering. Song et al. suggested that, because of the high-energy PLD plasma, a higher number of Vo is expected in PLD when the total pressure of the gases is low [36].

Meanwhile, ALD is a modified CVD technique that has garnered interest from both industry and academia as a thin-film deposition technique for its great step coverage and atomic-level thickness control, which are essential for designing high-aspect-ratio and three-dimensional (3D) nanoscale structures [38]. ALD typically involves precursor injection, precursor purge, reactant injection, and reactant purge, which enables the growth of a monolayer (practically sub-monolayer) with its self-saturated growth mechanism characteristic based on surface chemistry. The Vo concentrations are strongly influenced by deposition conditions such as the O reactant and deposition temperature, tuning the appropriate parameters is important for improving the performance of ALD-grown HfO2-based semiconductor devices.

Hsain et al. studied the effect of the O reactant on the growth behavior and resulting physical and electrical properties of ferroelectric HZO film [39]. Figure 2a presents the results obtained from the time-of-flight secondary ion mass spectroscopy (ToF–SIMS) and TEM analyses by comparing the use of O2 plasma (O2*) and H2O as oxygen sources. A 10-nm HZO film was deposited on the TiN electrode and annealed at 800 °C for 30 s for film crystallization. The ToF–SIMS results revealed that films deposited by O2* exhibited a higher 50TiO intensity, indicating the presence of a thicker TiOx interfacial layer compared to that of the HZO film grown with H2O as the reactant. Typically, the interface layer is formed by scavenging O from the HZO film which increases the concentration of Vo within the HZO film. However, the TiN electrode is oxidized in the initial step of ALD with the highly reactive O reactant, thereby forming a TiOx layer. This additional TiOx layer acts as a physical barrier between the TiN electrode and the HZO film, suppressing the formation of Vo within the HZO film during the subsequent steps in the ALD process. Moreover, the radicals generated by O2* enhance the reactivity of the oxygen source, resulting in a denser film with a lower concentration of Vo.

Fig. 2
figure 2

a Time-of-flight secondary ion mass spectroscopy (ToF–SIMS) depth profile and high-angle annular bright-field (HAABF) TEM image of TiN/Hf0.5Zr0.5O2/TiN capacitor using H2O and O2 plasma (O2*) reactant. b Intensity fraction of Hf 4f oxide (blue), sub-oxide (red) peak depending on the deposition temperature. c Vo formation energy with various dopants. Dopants are classified by chemical group. d, e XPS spectra of 2-nm Hf0.5Zr0.5O2 on Mo and TiN electrodes, respectively. Stoichiometric (HfO2) and non-stoichiometric (HfO2–x) peaks are deconvoluted from the Hf 4f spectrum. a reproduced with permission from [39]. b data from [42]. c data from [57]. d, e data from [66]

However, plasma does not necessarily reduce the Vo concentration. Martínez-Puente et al. compared the HfO2 film properties using thermal ALD (TALD), remote plasma ALD (RPALD), and direct plasma ALD (DPALD) [40]. H2O and O2* were used as O reactants for TALD and plasma ALD (RPALD and DPALD), respectively. In RPALD, the plasma source was located away from the substrate, preventing the film from directly interacting with the plasma. Conversely, in DPALD, the substrate participated in plasma generation, and the plasma was generated near the film surface. They calculated the stoichiometry of HfOx films by deconvoluting the sub-oxide peak from X-ray photoelectron spectroscopy (XPS) Hf 4f and O 1s spectra, resulting in the O/Hf ratios of 1.84, 1.91, and 1.80 for TALD-, RPALD-, and DPALD-deposited films, respectively. The RPALD film exhibited a higher O/Hf ratio with an enhanced reactivity of O2* than that for the TALD film with the H2O reactant. Meanwhile, the DPALD film exhibited the smallest O/Hf ratio, which can be attributed to the broken Hf–O chemical bond within the film caused by the plasma-induced damage. These facts suggest that employing a stronger O reactant such as O2* promotes the reaction between the precursor and the reactant, resulting in a reduced Vo concentration in the HZO film. Nevertheless, it should be noted that the adoption of plasma does not always lead to a decreased Vo concentration and can damage the film.

The deposition temperature is another important factor that influences the properties of the HZO film, including the average grain size and impurity concentration, which subsequently affecting the distribution of Vo [41]. Choi et al. studied the effect of deposition temperature on the Vo concentration by deconvoluting the Hf 4f binding energy peak, which is the overlap of the HfO2-x sub-oxide and stoichiometric HfO2 peaks, analyzed using XPS results [42]. They deposited 10-nm-thick HZO films on TiN electrodes using tetrakis (ethylmethylamino) hafnium, tetrakis (ethylmethylamino) zirconium, and H2O reactant by varying the deposition temperature from 120 to 250 °C. The results demonstrated that, at a deposition temperature of 120 °C, the relative areal ratio of the sub-oxide peak was 16.13%. At 250 °C, it increases to 21.42% (Fig. 2b). Similar findings were observed when deconvoluting the sub-oxide peak from O 1 s spectra, where the ratio of sub-oxide peak in O 1s increased from 2.9 to 6.15% with the increasing deposition temperature. This can be attributed to the enhanced diffusion of O from the HZO film to the TiN electrodes with an increase in deposition temperature. This phenomenon was consistently observed with various Hf precursors and reactants [43]. Kukli et al. reported the same behavior of a HfO2 film directly grown on an Si substrate using a tetrakis (dimethylamide) precursor and H2O reactant [43]. The O/Hf ratio confirmed through the time-of-flight elastic recoil detection analysis revealed that it decreased from 1.96 to 1.83 with an increase in the deposition temperature from 350 to 400 °C, indicating an increased Vo concentration within the film.

These results lead to the conclusion that deposition conditions such as O reactant species and deposition temperature are critical for modulating the Vo concentration because the formation of Vo is strongly correlated to the interfacial redox chemistry at the oxide/metal interfaces. This implies that the concentration and distribution of the Vo can be modulated by optimizing ALD conditions, which improves the physical and electrical properties of the device. Such strategic approaches are discussed in Chapters 3 and 4 for the ferroelectric and RS memory devices based on existing literature.

2.1.2 Doping

Doping is a commonly employed strategy to improve the film properties. For example, doping enhances the thermal stability and reduces the leakage currents in metal oxides [44]. In ferroelectric HfO2 thin films, the dopant species and concentration are the key factors that deterministically affect the polymorphism and resulting physical and electrical properties. The introduction of dopants in HfO2, such as Si, Al, Zr, Sr, La, Gd, or Y, has been reported to induce the formation of a metastable polar orthorhombic phase (Pca21) at room temperature within an adequate range of dopant concentration, depending on the dopant species [45,46,47,48,49,50,51]. However, insufficient or excess doping can result in the formation of stable monoclinic or tetragonal/cubic phases [52]. Metal dopants can be introduced to manipulate the structural and electronic properties of HfO2. Several studies have revealed the close relationship between dopants and Vo. Based on their comprehensive study of the CSD-grown doped HfO2 thin films, Starschich et al. [53] suggested that the radius of dopant ions is a critical factor affecting the polymorphism and ferroelectric properties. They suggested that strong ferroelectricity is expected when the dopants of which radius is larger than that of Hf. Similarly, Batra et al. reported that a clear trend was observed where dopants with a large ionic radius and low electronegativity stabilize the ferroelectric orthorhombic phase in HfO2 [54]. They suggested that lanthanide series elements and the lower half of the alkaline earth metals (e.g., Ca, Sr, and Ba) are favorable dopants for inducing strong ferroelectricity in HfO2. However, it should be noted that the dopant is not the only factor that stabilizes the polar orthorhombic phase, and therefore, many other factors (e.g., Vo and surface energy) should be considered to make the polar orthorhombic phase the most stable phase. Schroeder et al. [55] reported that the doping concentration range required for inducing ferroelectricity in ALD-grown HfO2 thin films strongly depended on the valence number and diameter of the dopants. Especially, when the valence number of dopants is different from the host cation (Hf for the case of HfO2), to satisfy charge neutrality, a certain amount of Vo must be formed.

In addition, the effect of dopants on the migration barrier of Vo was investigated by Zhang et al. using the density functional theory (DFT) calculations based on the generalized gradient approximation for monoclinic HfO2 and ZrO2 containing 96 atoms [56]. The migration barrier for Vo toward the dopant increased as the dopant radius increased, as these vacancies are unfavorable to pass through the large cations. Conversely, the migration barrier for Vo moving away from the dopant decreased when trivalent dopants (e.g., Al and La) were introduced due to the larger lattice relaxation compared to tetra- or penta-valent dopants, which facilitates the movement of Vo into a specific direction. Thus, it is straightforward that the dopant species and concentration should be strongly correlated to the Vo concentrations as well as their movement within the film.

Zhao et al. studied the Vo formation energy in HfO2 doped with different dopants using the Vienna ab initio simulation package (VASP) [57]. The metal dopants are divided into two groups, interstitial and substitutional according to their relative position in parent HfO2, by comparing the energetically more stable site of dopants within the HfO2 film. Dopants such as Sc, La, Ti, Zr, Nb, and Ta exhibit lower formation energies at the substitutional site, where they chemically bond with O atoms. In contrast, dopants such as Mg, Al, Ni, Cu, and Ag have lower formation energies at the interstitial sites, where they merely occupy the space between atoms. Figure 2c illustrates the formation energies of Vo for substitutional dopants based on the number of valence electrons. Trivalent dopants exhibit the lowest Vo formation energy, attributed to the instability of metal–oxygen bonds caused by a deficit of valence electrons compared to Hf atoms. However, tetra- or penta-valent dopants exhibit Vo formation energies similar to that of undoped-HfO2 because their metal–oxygen bonds are already occupied by electrons, making it difficult for these dopants to form Vo compared to that with trivalent dopants.

To summarize, Vo concentrations are strongly correlated to the valence number of dopants, because Vo concentration is influenced by doping of bi- or tri-valent chemical species to meet the charge neutrality condition. Another important factor in this correlation is the mobility of Vo. Particularly, since the Vo-dopant complex is known to be stable at lower energies, the mobility of Vo near bi- or tri-valent dopants should be significantly slower than that of Vo spatially far from the dopants. Thus, the effect of dopants on Vo migration and agglomeration should be studied more carefully.

2.1.3 Electrode

With the state-of-the-art dimensional scaling of memory cells, electrodes with low resistivity and large work function are considered pivotal to reducing the leakage current. Moreover, the influence of electrodes on the formation of the ferroelectric orthorhombic phase was demonstrated [58, 59]. This was attributed to the application of the mechanical in-plane stress as well as the interfacial redox chemical reactions during the fabrication processes. Consequently, beyond their electrostatic role as electrical contacts to the film, electrodes affect the concentration of Vo by serving as a source of O to the neighboring HfO2 film. In addition, the ratio of N atoms in nitride electrodes such as TiN or TaN can affect the concentration of Vo. The N atoms can diffuse into the HfO2 thin film through the interface and form Hf-N bonds, causing the unwanted distribution of space charge. This can cause an asymmetric internal electric field and degrade ferroelectricity [60]. Meanwhile, more Vo are generated in the HfO2 thin film with an increase in the ratio of N atoms in TiN or TaN electrodes [61, 62]. Therefore, it is necessary to comprehend the stoichiometry of nitride electrodes and their impact on HfO2 thin films.

Electrodes frequently used for HfO2 thin films are categorized into three different groups: nitride, oxide, and metallic electrodes. TiN, a nitride electrode, has been already commercially adopted in conventional DRAM cell capacitors. However, a noteworthy concern arises in ultra-thin films, wherein an undesired TiOxNy layer can form at the interface between the TiN electrode and the HZO film [63]. This can be attributed to the oxygen-scavenging effect of the TiN electrode, leading to oxidization of the electrode and an increase in the Vo concentration within the dielectric oxide film [64].

To address the reliability issues associated with the interface layer formation, extensive research has been conducted on alternative electrodes. It has been observed that conducting oxide electrodes, such as IrO2 and RuO2, reduce to Ir and Ru, respectively, at the interface between the electrode and (Hf,Zr)O2 [65]. Notably, Goh et al. reported that, when a RuO2 electrode is employed on 10-nm HZO instead of a TiN electrode, the concentration of Vo in the HZO film decreases. They confirmed it through XPS analysis, comparing the sub-oxide HfO2-x peak intensity ratios of the two electrodes. This can be attributed to the oxide electrode supplying additional O atoms to the HZO film, resulting in a reduction of Vo concentration. However, Ir and Ru are less cost effective to be adopted in commercial products.

Additionally, the influence of metallic electrodes, specifically Mo and W electrodes, on the Vo concentration in HZO films was investigated. Figure 2d, e present the XPS analyses of 2-nm HZO films grown via ALD on Mo and TiN electrodes [66]. The non-stoichiometric HfO2–x (yellow line) XPS peak was deconvoluted from the XPS Hf 4f spectrum (red line in Mo electrode, blue line in TiN electrode) to calculate the ratio of the sub-oxide peaks. The analyses revealed that the HZO grown on the Mo electrode covered a smaller portion of the HfO2–x spectrum (12%) than the HZO film grown on the TiN electrode (44%). This trend persisted after rapid thermal annealing (RTA) with a decrease in the Vo concentration. This can be attributed to the oxidized MoO3 formed at the initial ALD cycles, undergoing reduction during annealing, which results in the formation of Mo and MoO2 species. This reduction process contributes to the decrease in Vo concentration within the HZO film. Consequently, the relative areal ratio corresponding to the sub-oxide peaks decreased from 12 to 3%, representing a significantly lower value than that under the pre-annealing condition. Furthermore, the impact of W electrode on the formation of Vo was also investigated. Yang et al. [67] employed the concept of standard formation enthalpy per oxygen of 1 mol (ΔHf per O) to elucidate the reduction behavior of metal oxides. They found that the ΔHf per O of WO3 was—280.97 kJ/mol, while HfO2 had a ΔHf per O value of—556.6 kJ/mol. This difference suggests that WO3, which is formed in the initial ALD cycles, is reduced to produce W owing to its lower ΔHf per O than HfO2.

2.2 Migration of Vo and their roles in HfO2

In conventional MOSFET, the drift of Vo in high-k HfO2 induces hysteresis in the drain current–gate voltage transfer curve [68]. Under repeated pulses, the reversible migration of the charged Vo can generate the hysteresis loop, which degrades the reliability of the device. Therefore, suppressing the formation of Vo is desirable to enhance the performance of the electronic device.

However, the soft and reversible dielectric breakdown in the electroforming process and subsequent RS operation is associated with Vo migration in RRAM, which is an RS device. The resistance of the device is determined between the high resistance state (HRS) and low resistance state (LRS) under opposite voltages based on the movement of Vo. Furthermore, the wake-up and fatigue behaviors observed in ferroelectric (Hf,Zr)O2 are known to originate from the Vo redistribution within the film (See subsection 3.2). Therefore, understanding the mechanisms and consequences of Vo migration is essential for comprehending the dynamics of HfO2 film and advancing the development of emerging devices.

In early studies, the migration of Vo could not be directly observed due to technological and hardware limitations. Consequently, the migration of Vo had to be indirectly confirmed through alternative ways. Nagata et al. utilized hard X-ray photoelectron spectroscopy (HX-PES) to observe the spectra at the interface between the Pt electrode and PLD-grown 30-nm HfO2 film under bias operation, providing evidence for Vo migration [69]. With the forward bias, the Pt–O bonding peak in O 1s spectra increased, whereas the Hf–O peak decreased, indicating the migration of Vo toward the electrode. In addition, Starschich et al. calculated the Vo migration distance during field cycling in CSD-prepared Y-doped HfO2 [35]. From the Mott–Gurney equation, charged Vo migrate approximately 6.5 nm at an electrical field of 3.25 MV/cm for 0.5 ms. While these findings provided insights into Vo migration, the migration process itself could not be directly captured.

However, advanced TEM techniques enable the direct imaging of light (O) and heavy (Hf) elements simultaneously. The popular high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) collects the electrons scattered at high angles using an annular detector. This technique allows the detection of materials composed of a single atomic species or elements with similar atomic numbers. However, in the case of metal oxides with elements of different atomic numbers, detection of the scattering strength of light elements adjacent to heavy atoms is challenging. To address this issue, integrated differential phase contrast scanning transmission electron microscopy (iDPC-STEM) was introduced [70]. By combining the scattered electrons that fall inside the bright-field disk and HAADF detector, they imaged both Hf and O atoms simultaneously [71].

Figure 3a illustrates the iDPC-STEM images of rhombohedral 6-nm-thick HZO epitaxially grown on an La0.67Sr0.33MnO3 (LSMO) electrode under 0- and 4-V bias. The two red arrows in Fig. 3a indicate two O ions chemically bonded to Hf and Zr cations. Vo migrates toward the bottom electrode with an increase in bias, exhibiting both vertical and in-plane migrations (Fig. 3b). Moreover, upon increasing the bias to 4 V, the epitaxial rhombohedral phase transforms into polycrystalline orthorhombic and monoclinic phases, as indicated in the inset of the fast Fourier transform (FFT) image in Fig. 3a. The stabilized rhombohedral phase under slightly O2-deficient condition transforms into more stoichiometric monoclinic and orthorhombic phases with oxygen supply from the bottom LSMO electrode. However, the rhombohedral phase is recovered under − 4 V through the reversible migration of O atoms (vacancies). In addition, by employing a noble electrode (Au), the study discovered that the HZO film itself acted as a Vo supplier, resulting in a phase transition between the rhombohedral and monoclinic phases. These observations suggest that the migration of Vo can induce phase transitions in epitaxial rhombohedral HZO and are dependent on the ferroelectricity in the HZO film.

Fig. 3
figure 3

Integrated differential phase contrast scanning transmission electron microscopy (iDPC-STEM) image of rhombohedral Hf0.5Zr0.5O2 film under a 0 V (left), 4 V (right). b Out-of-plane displacement of Vo under a positive voltage. Inset shows the in-plane and out-of-plane displacements. c High-angle annular dark-field scanning transmission electron microscopy image overlaid with O K line profile and EDS mapping of O K edges of oxygen-deficient channel on the HfO2 film. a, b reproduced with permission from [71]. c reproduced with permission from [75]

In RS devices, the electric field-dependent migration of Vo plays a crucial role in the entire switching process. In the Vo-mediated filamentary switching type, the switching mechanism involves the formation of a Vo CF, which results from the localized migration of Vo when voltage is applied. The forming process is required to initialize the formation of CFs, and this initialization is triggered by the electric field-driven dielectric breakdown process. During the forming process, oxygen ions dissociate from the HfO2 matrix and migrate from the cathode toward the anode under an external electric field, which initializes the nucleation and growth of Vo CFs in the HfO2 layer. Under subsequential operation (RESET and SET) biases, the Vo migrate, and the migration is driven by voltage. This phenomenon triggers repeated CF rupture, connection, and operating OFF/ON in the RS devices. The accurate observation and understanding of Vo dynamics are important for implementing controllable RS operations because these electroforming and subsequent RS operations are associated with Vo migration.

Unlike the highly mobile Ag or Cu in electrochemical metallization cell-type devices, the motion of oxygen in RS devices cannot be observed easily as it is a light element and suffers from ambient occlusion. Therefore, studies used TEM for observing the chemical composition and morphology of the device to directly identify Vo migration. In these studies, the change in the chemical composition in the Vo region was studied by EELS or high-resolution energy dispersive X-ray spectroscopy (EDS). The EELS is an appropriate method for light element detection because its contrast is not proportional to the atomic number but related to the inelastic interactions of the element with the materials. Therefore, EELS analysis can be used to observe changes in the oxygen concentration after RS operation with a low-energy-loss spectrum. Prior studies exploited EELS with STEM to observe electrical stress-mediated Vo migration in HfO2-based RS devices (Fig. 3c) [72, 73]. Based on the local chemical composition and phase, the STEM-EELS chemical maps of the Vo migration region can be recorded with high energy (1.6 eV) and spatial resolution (0.5 nm), which is evidence for the localization regions of the individual Vo in HfO2.

Notably, the detailed characteristics, which include atomic structure and phase transformation in and near the Vo migration region, were demonstrated through high-resolution transmission electron microscopy (HR-TEM) observations [74]. Yin et al. studied crystallite kinetics coupled with Vo migration using TEM in a W/HfOy/HfOx/Pt stacked structure. After the formation process, they discovered directionally aligned crystalline regions consisting of monoclinic and orthorhombic oxygen-deficiency phases in the initial amorphous HfO2. Their TEM observations demonstrated that the extrusion of Vo migration resulted in structural modifications involving crystallite separation, phase transformation, and misalignment. Further, Yang et al. attributed the difficulty in the direct visualization of oxygen ion motion to the commonly used electron microscope imaging focusing on the mass properties of ions [75]. Therefore, they combined TEM with in situ electrostatic force microscopy (EFM), which exploited both the mass and charge attribution of oxygen ions. Both accumulation and Vo migration can be detected by the electrostatic force between the probe and the sample because EFM is sensitive to charge accumulation. Moreover, the formation of conduction channels within the HfO2 layer was directly detected by high-resolution STEM and EDS analyses. Their results can be utilized for probing the ion-transport dynamics in solid electrolytes and for understanding the switching mechanism of RS devices.

Previous studies indirectly confirmed Vo migration through theoretical computations and alternative methods such as spectroscopy. However, recent advances in TEM techniques, specifically iDPC-STEM, enabled the direct imaging of Vo migration. In RS devices, accurately observing and understanding Vo dynamics is essential because electroforming and the subsequent RS operations rely on Vo migration. Observation techniques such as EELS, HRTEM, and in situ EFM were employed to study Vo migration and its effects on the behavior of the device, and the results provided valuable information on the chemical composition, phase transformation, and structural modifications caused by the migration.

2.3 Aggregation of Vo in HfO2

Understanding the aggregation behavior of Vo is crucial for comprehending the conduction mechanism because the formation of Vo chains can significantly impact the conductivity of the film. Figure 4a depicts the distribution of defect levels within the bandgap of a monoclinic-phase HfO2 film concerning the charge states of Vo at their most favorable sites: positively charged Vo+2, Vo+1 threefold coordinated sites, and neutral and negatively charged Vo0, Vo−1, and Vo−2 fourfold coordinated sites [76]. Gavartin et al. employed the B3LYP hybrid density functional electronic structure calculations on monoclinic-phase HfO2 with 96 atoms. The results indicate that Vo+2 creates one empty state 0.85 eV below the conduction band (CB), while Vo−2 generates two fully occupied states—one 1.46 eV below the CB and the other near the intrinsic Fermi level. For Vo+1 and Vo−1, they become magnetic defects, occupying half of the defect states. In case of a Vo0, it generates fully occupied state near intrinsic Fermi level. These findings suggest that Vo serves as an electron- and hole-localization center and a medium for trapping and transporting carriers. Consequently, the aggregation of Vo generates conductive paths or preferential sites for charge accumulation, causing film degradation or breakdown. The aggregation of Vo can occur through two distinct mechanisms: the clustering of pre-existing Vo or the creation of new Vo near the pre-existing ones under a strong electric field (1–2 MV/cm), weakening the Hf–O bond and promoting the generation of new Vo and O interstitials.

Fig. 4
figure 4

a Bandgap of HfO2 with defect states depending on the charge state of Vo. b Reconstructed 3D C-AFM images of Vo evolution and migration in HfOx-based RS device. c Atomic structure of monoclinic HfO2 with GB at the center (dashed line). The 4-Å-wide shadow region near the GB is a favorable region for Vo aggregation. d Current scan map using C-AFM of polycrystalline HfO2. e Topography (blue) and current (red) data along the green solid line in d. Schematic of conduction mechanism in HfO2 f before and g after breakdown. h Schematic and HRTEM of a complete CF resulting from Vo aggregation in an LRS device with a typical polymorphous HfOx region, namely h-Hf6O and m-HfO2 regions a reproduced with permission from [76]. b reproduced with permission from [80]. c reproduced with permission from [83]. d, e reproduced with permission from [184]. h reproduced with permission from [79]

Gao et al. conducted a DFT-based calculation to investigate Vo aggregation in amorphous HfO2 by employing 9 periodic models, containing 324 atoms each. They discovered that the Vo formation energy adjacent to the pre-existing Vo was lower than the formation energy of a single Vo [77]. This energy decreased with an increase in the number of adjacent Vo implying that the rate of vacancy formation increased significantly once Vo is generated. This can be attributed to distortion from the surrounding system, which facilitates the generation of new Vo. Furthermore, the aggregation of Vo amid electron injection was examined [78]. Injection of electrons from the electrode is inevitable under the high-electric-field condition which causes thermal fluctuations in O ions. It dislodges O atoms from their original sites and causes the formation of Vo–O interstitial pairs. The study also determined that the production of a Vo–O interstitial pair required energy of 1.19 eV from the perfect cell. This energy cost reduces to 0.96 eV near Vo. Consequently, Vo are more likely to form near the existing ones, favoring aggregation rather than distributing uniformly throughout the film.

Extensive nanoscale physicochemical analyses have been conducted to directly observe nanoscale Vo aggregation [79,80,81,82]. However, despite its importance, observation of Vo aggregation is challenging due to the small size of the filaments and minimal compositional difference from the surrounding matrix material. Among the various tools used for observing Vo aggregation, conductive atomic force microscopy (c-AFM) is widely employed because it can locally measure the electrical properties of sample surfaces with high spatial resolution. Celano et al. directly observed the morphology of a single CF during the forming process in a scaled Ru/Hf/HfO2/TiN device using a technique [81] that integrates AFM-based tomography (scalpel-SPM) with the high lateral resolution of c-AFM and sub-nanometer vertical resolution through a controlled removal procedure. This slice-and-view tomography technique provided a 3D characterization of a single CF, demonstrating its size (< 10 nm) and constriction regions determined by Vo aggregation under an electric field. According to their observations, the CF exhibited a conical shape located at the interface of the oxide-inert electrode, and this morphology correlated with the electrical conditions. Their results elucidated the morphological features of CF resulting from Vo aggregation under specific electric field conditions.

Although the study demonstrated the aggregation of Vo in a single CF during the forming process, it does not cover all the RS states resulting from various operations, such as forming, SET, and RESET. Wei et al. analyze CFs in the post-forming, post-SET, and post-RESET states in a Pt/HfOx/TaOy/TiN device using c-AFM techniques with 3D reconstruction (Fig. 4b) to comprehensively analyze CF changes in all three resistance states, including the quantity, morphology, and rupture location in the high-resistance state [80]. They acquired the reconstructed 3D images of CFs by combining the two-dimensional (2D) conductivity profiles collected from each exposed surface of the top electrode (TE) and switching layer, removed by a conductive diamond tip at different heights. Furthermore, multiple conductive channels in each resistance state could be observed because the 2D conductivity profiles cover the entire device. The CFs after operation under each state exhibited typical morphologies, including hourglass, inverted-cone, and short-cone, respectively. During the forming process, a positive voltage was applied to the bottom electrode (BE), causing oxygen ions to dissociate from the HfOx layer and migrate toward the BE, while leaving Vo in the HfOx layer. Positive feedback from the electric field accelerates oxygen ion migration toward the oxygen reservoir layer (oxygen-deficient TaOy layer in the bilayer structure HfOx/TaOy), which results in Vo aggregation, thereby forming tapered CFs. The mechanism of Vo migration during the SET operation is similar to the forming process; however, thicker CFs are observed after the operation. Subsequently, during the RESET process, an opposite external electric field is applied, causing oxygen ions to migrate out of the oxygen reservoir layer and upward to merge with Vo, which results in the rupturing of Vo CFs from the upper end near the TE and switching of the device from LRS to HRS. Subsequently, this leads to the formation of broken/newly generated cone-shaped filaments at the HfOx/oxygen reservoir layer interface.

Zhang et al. employed HRTEM to comprehensively investigate the crystal structure of a CF resulting from Vo aggregation, and they revealed that the Vo CF possesses the core–shell structure (oxygen-deficient CF and the corresponding oxygen-rich shell) consisting of metallic hexagonal-Hf6O (h-Hf6O) and its crystalline surroundings (monoclinic and tetragonal HfOx) [79]. Based on their theoretical and experimental investigation, they reported that the RS process of HfO2-based devices involve a phase transition of the CF shell from the monoclinic to the tetragonal phase. This transition occurs in the CFs surrounding both the complete and ruptured h-Hf6O. This core–shell structure was observed by another study while analyzing a HfOx-based RS device using synchrotron-based scanning transmission X-ray microscopy (STXM) [82]. The STXM system is employed for the nondestructive observation of the in situ switching of HfO2-based RS devices and the analysis of the variations in the chemical composition and positions associated with Vo aggregation. After completing the ON/OFF switching cycles, they examined a cell by capturing images at various X-ray energies specifically tailored to the O K-edge (570 eV). The images revealed a localized region consisting of a darker ring (low-conductivity, oxygen-rich region) surrounding a brighter center (high-conductivity, oxygen-deficient region).

Notably, Vo aggregation occurs in specific regions rather than arbitrary locations. Figure 4c shows the atomic structure of monoclinic HfO2 with the dashed line in the middle indicating the grain boundary (GB) [83]. Mckenna et al. employed nudged elastic band (NEB) calculations and the VSAP code to comprehensively characterize the behavior of Vo across the film. They observed a reduction of diffusion energy barrier approximately 0.7 eV near the GB compared to the bulk, indicating increased stability of this region with a lower formation energy compared to other regions [84, 85]. They attributed the cause of this phenomenon to the variations in bond length and electrostatic potential near the GB, which in turn enhanced the relaxation of ions.

Another study employed c-AFM to experimentally verify the aggregation of Vo. Figure 4d illustrates the CAFM current scan map of polycrystalline HfO2 using a conductive tip as the TE. Stronger currents were observed in specific regions, which corresponded to areas with lower topographical heights. These depressions were produced by thermal grooving, which is frequently observed at GBs [86, 87]. The difference in free energy between the GB and the surface of the film results in the formation of a topographical groove with a certain angle within the polycrystalline film. Therefore, the observed strong current near the GB can be attributed to the increased tunneling current and aggregation of Vo.

Finally, Fig. 4f illustrates the conduction mechanisms responsible for the leakage current produced in the HfO2 film. Extensive studies have worked on these mechanisms, including the Poole–Frenkel (PF) emission, Schottky emission, trap-assisted tunneling (TAT), Fowler–Nordheim (FN) tunneling, and space charge limited current [20, 88,89,90]. However, the precise mechanism could not be explained without considering the contribution of traps. The Schottky emission and FN tunneling model contain abnormal parameters and fail to generate an accurate interpretation of the experimental results, which demonstrate an increase in leakage current with higher Vo concentrations [20]. Trapped electrons can transition from their localized state using two mechanisms: thermal fluctuation and tunneling. In PF emission, thermal fluctuations provide sufficient energy for electrons to migrate to the CB and relax into a different localized state. Notably, under strong electric fields, electron hopping can be facilitated without significant thermal fluctuations [91]. Meanwhile, the TAT model describes the electron tunneling phenomenon between traps providing additional tunneling paths that increase the probability of electrons tunneling through the barriers [92]. These observations demonstrate the relationship between Vo concentration and leakage current, i.e., a high Vo concentration leads to a high leakage current. However, the repetitive generation of Vo under a high electric field and electron injection form conductive pathways across the film, ultimately resulting in a hard breakdown (Fig. 4g). In summary, aggregated Vo near the GB serve as a charge-transportation medium, while excessive Vo concentration can cause the hard breakdown of the film.

To summarize, the aggregation behavior of Vo in HfO2 films plays a crucial role in their conductivity and conduction mechanisms. Vo chains significantly influence the film’s conductivity by acting as carrier traps and transport centers. Several studies have utilized computational methods to investigate the behavior of Vo and revealed the charge states of Vo create different degrees of defect within a bandgap. Vo aggregates through two mechanisms: clustering of pre-existing Vo and generation of new Vo near existing ones under strong electric fields. Experimental techniques like c-AFM and HRTEM were employed to observe Vo aggregation at the nanoscale. The formation of Vo chains and their morphological features, including hourglass, inverted-cone, and short-cone, have been directly observed. The crystal structure of Vo chains is a core shell with h-Hf6O and its crystalline surroundings (Fig. 4h). Particularly, Vo aggregation preferentially occurs at GBs owing to reduced diffusion barriers and lower formation energies. Moreover, the aggregation of Vo near GBs produces higher tunneling currents and increased conductivity. Leakage currents in HfO2 films are influenced by the Vo concentration, i.e., high concentrations result in increased leakage. However, excessive Vo generation under high electric fields and electron injection has two outcomes: a soft breakdown, which induces the formation of CFs for RS behavior, or a hard dielectric breakdown of the film. Therefore, understanding Vo aggregation will help comprehend the conduction mechanisms, degradation processes, and RS behavior in HfO2 films, as it impacts their conductivity and leads to the formation of CF or film breakdown.

3 Effects of Vo on electrical properties reliability of ferroelectric HfO2

The spatial distribution and average concentration of Vo within HfO2 thin films significantly influence their polymorphism and the resulting ferroelectric properties, as well as reliability metrics such as switching endurance that are the major performance parameters of semiconductor devices based on HfO2-based ferroelectrics [29, 37, 93]. Studies reported that the concentration of Vo influences the relative free energy of the metastable crystallographic phase, including the ferroelectric orthorhombic phase [23, 94]. Interestingly, their relative location could also influence the polymorphism of HfO2-based thin films [23]. Moreover, charged Vo could be drifted by repeated electrical field cycling, causing wake-up, fatigue, and hard breakdown, which are typically observed during the endurance tests of ferroelectric HfO2-based capacitors [35, 95, 96]. Therefore, understanding the effect of Vo concentration and spatial distribution on the ferroelectric properties and reliability of HfO2 devices is important for advancing HfO2-based ferroelectric memory devices.

The effect of Vo in HfO2-based RS memories is even more comprehensible owing to the pivotal role played by filament formation and rupture resulting from Vo accumulation and electrically or thermally driven migration in driving the resistive change. Precise control over Vo is crucial for guarantee the reliable memory-switching behavior of HfO2-based RS devices. Therefore, the possible factors that can influence Vo-mediated RS behavior must be identified, and reliable methods to control Vo dynamics must be established.

Therefore, in this section, the effects of location, concentration, state of charge of Vo on the phase stabilization of HfO2, and migration of Vo induced by electrical field cycling are comprehensively reviewed and critically discussed based on extant literature. Furthermore, we also provide some perspectives on strategies to improve the performances and reliability of semiconductor devices based on ferroelectric HfO2 by controlling the Vo concentration based on the current understanding of them.

3.1 Effect of Vo on ferroelectric properties of HfO2

Zhou et al. investigated the effect of Vo on the crystal phase of HfO2 using the first-principles method based on DFT calculations. Figure 5a illustrates the total energy of the various crystal phases of HfO2 in the bulk regions of the HfO2 thin film as a function of Vo concentration [23]. Here, t, f, o, and m refer to the tetragonal P42/nmc, polar orthorhombic Pca21, nonpolar orthorhombic Pbca, and monoclinic P21/c phases, respectively. For the bulk HfO2 examined in Fig. 5a, the monoclinic phase is the most stable phase with the lowest total energy when the Vo concentration is zero. According to the phase diagrams of HfO2 and ZrO2, the relative free energies of the tetragonal and cubic phases decrease with increasing temperature. Consequently, the thermodynamically stable phase sequentially changes from the monoclinic phase to the tetragonal and cubic phases at 1,943 and 2,573 K in HfO2 and at 1,473 and 2,743 K in ZrO2, respectively [5, 97].

Fig. 5
figure 5

Total energies of the various crystal phases of HfO2 with different Vo concentrations of the a bulk and b interface region. c Energies of the orthorhombic Pca21 phase of HfO2 with different Vo states of charge and relaxation conditions relative to the monoclinic phase (ΔE) as a function of Vo concentration. Phase content of 10-nm HZO samples as a function of the d ozone dose time during the ALD process and e oxygen flow during the sputtering. The phase portions are extracted and calculated by the GIXRD pattern. f Polarization–electric field curves of the 10-nm samples vs. ozone dose time during ALD process. a, b reproduced with permission from [23]. c reproduced with permission from [94]. d, e reproduced with permission from [227]. f reproduced with permission from [37]

However, as the Vo concentration increases, the differences in total energy between the monoclinic and other metastable phases, including tetragonal and polar orthorhombic phases, decrease. As the crystallographic origin of ferroelectricity in polycrystalline HfO2 thin films is the formation of the polar orthorhombic phase [10, 52, 98], an adequate range of Vo concentration in the films would be beneficial for inducing ferroelectricity. According to previous studies, in epitaxial doped (Hf,Zr)O2-based films, the crystallographic origin of ferroelectricity is the formation of the rhombohedral phase (space group: R3 or R3m). However, in this study, polycrystalline films are the focus. Notably, several aspects related to the effect of Vo on polymorphism and the resulting ferroelectricity calculated from the computational simulation should be carefully considered. First, the maximum Vo concentration that does not crystallographically distort the original crystal structure should be considered, similar with the effect of chemical doping. Batra et al. [54] examined via high throughput calculations the effect of doping various metallic cations into HfO2 thin films. They showed that an excessive dopant concentration severely distorted the crystallographic structure by breaking the original symmetry in the parent phase. Similar effects are also expected with a high Vo concentration, which should be considered. Moreover, excessive Vo concentration could be the reason for the high leakage current produced by the formation of localized energy levels in HfO2 thin films, which would be another limiting factor of Vo for inducing ferroelectricity in HfO2. Second, an excessive Vo concentration could destabilize the polar orthorhombic phase because the total energy of the tetragonal phase tends to more rapidly decrease with increasing Vo concentration [50]. Third, most computational studies examined the total energy at 0 K, so the effect of entropy-related terms should also be considered. Particularly, high-temperature phases such as the tetragonal and cubic phases have entropy values significantly higher than those of the polar orthorhombic and monoclinic phases, which was proven by the energy crossover observed in previous studies when an entropy term was added to Gibbs free energy [6].

Despite these limitations, the effect of Vo on the relative free energy of various polymorphs is scientifically meaningful and consistent with experimental observations. The effect of Vo concentration would thermodynamically and/or kinetically alter the final polymorphs formed in HfO2-based thin films. The free energy difference between the polar orthorhombic and stable monoclinic phases is, in fact, excessively large to be overcome solely by the Vo effect. However, its synergy with various factors, such as surface energy, doping, and stress, could cause the formation of the polar orthorhombic phase. According to the kinetic model, the precursor tetragonal phase formed at elevated temperatures is a key core process, which might be facilitated by the effect of Vo. The subsequent competition between the tetragonal and polar orthorhombic phases should also be strongly correlated with this effect. Therefore, for both models, the Vo should be a critical factor influencing polymorphism and the resulting ferroelectricity [99,100,101].

In addition to the concentration of Vo, spatial distribution is another critical factor determining the ferroelectricity of HfO2. Zhou et al. examined the effect of Vo on the total energy of various polymorphs of HfO2 and revealed that the free energy values of the polymorphs depended on the location of Vo. Figure 5b illustrates the total energy of the tetragonal, polar orthorhombic, and monoclinic phases as a function of the number of Vo at the interface through DFT calculation was with a 1 × 1 × 4 supercell, where the total number of Hf and O atoms were 32 and 16, respectively. In the absence of Vo (perfect) in the interfacial region, the magnitude of reduction in free energy of the polar orthorhombic phase could further widen, resulting in a greater stabilization of the phase. However, with increasing number of Vo beyond 1, the monoclinic phase becomes the most stable phase. Here, 1, 2, and 3 Vo refer to the 1, 2, and 3 Vo relative to the total oxygen position in the supercell. This led us to interpret that the ferroelectricity within this phase can be stabilized by maintaining a low concentration of Vo at the interface of the thin film, unlike at the bulk region.

The Vo in oxide materials such as HfO2 exist in a neutral or positive-charge state [102, 103]. Therefore, the effect of the state of charge of Vo on the relative free energy of various polymorphs of HfO2 must be investigated. Figure 5c illustrates the difference in total energy between the polar orthorhombic and monoclinic phases of HfO2 due to the concentration of Vo with neutral or positive states of charge reported by He et al. [94]. They calculated the total energy using the DFT method. To examine the effect of the state of charge of Vo on phase stability, they defined ΔE as the difference in energy between the orthorhombic Pca21 phase and monoclinic P21/c phases and a function of the concentration of Vo. The orthorhombic Pca21 phase and monoclinic P21/c phases are modeled with 2 × 3 × 2 supercells with 96 O atoms. Here, removal of one oxygen atom is expressed at HfO2–x, where x = 2.08%. In addition, to understand the effect of atomic relaxation, the scenarios of before (dash) and after (solid) lattice relaxation are compared in Fig. 5c. The red line in Fig. 5c indicates that, even as the concentration of neutral Vo increases and the energy difference decreases, the polar orthorhombic phase continues to possess a higher total energy than the monoclinic phase. Conversely, the blue line, which corresponds to the change in energy of the polar orthorhombic phase induced by Vo2+ with a positive state of charge, indicates that, for the trend, the energy difference between the polar orthorhombic and monoclinic phases decreased significantly. In addition, an energy crossover was observed at approximately 2.2% and 3.1% for the relaxed and unrelaxed lattices, respectively. Hence, positively charged Vo2+ has a markedly stronger effect than neutral Vo. Additionally, a greater stabilization of the polar orthorhombic phase can be observed at lower Vo concentrations under the relaxed state of the lattice compared to that under the unrelaxed state.

Furthermore, He et al. [94] reported that a positively charged Vo2+ had a lower diffusion activation energy than a neutral Vo. The Vo diffusion activation energy of the positive and neutral Vo, calculated through DFT calculations, were 2.49–2.89 and 0.89–0.98 eV, respectively, which are consistent with the values reported in previous studies [104,105,106,107]. He et al. explained that in case of a neutral Vo, localized electrons within the HfO2 band gap interfere with the migration of the Vo. In contrast, for positive Vo, the bandgap does not contain effective charge, which results in its significantly lower Vo diffusion activation energy. This phenomenon was also observed in ZrO2 [108, 109] and other oxides [110, 111].

In summary, a high concentration of charged Vo could reduce the total energy difference between the polar orthorhombic phase and the monoclinic phase. Additionally, positively charged Vo2+ have a lower diffusion activation energy than neutral Vo, which could explain the Vo migration experimentally observed in different studies. The effect of oxygen vacancies on the ferroelectric properties of the HfO2 film requires should be investigated further. Jaszewski et al. conducted XPS and positron annihilation spectroscopy analyses for Vo characterization of 20 nm HfO2 deposited by magnetron sputtering [30]. They confirmed that the concentration of neutral Vo increased after the heat treatment of the thin film. However, ferroelectricity with a maximum of 2Pr of about 17.4 μC/cm2 appeared in the HfO2 thin film, suggesting that neutral Vo can also have an important influence on the ferroelectric properties of HfO2. Vo, which is commonly discussed in HfO2 ferroelectric applications, is a positively charged defect, and therefore, the positively charged Vo is discussed in the subsequent sections.

Ferroelectric HfO2 thin films reported in various previous studies were typically deposited via the ALD process [112,113,114,115,116]. As discussed in Sect. 2.1, the ALD process parameters, such as the O source or deposition temperature, distinctively alter the concentration of Vo in HfO2 thin films. Furthermore, as the concentration and migration of Vo can impact the crystal phase of the thin film, the oxygen source density or/and injection time during the ALD process can have a pronounced effect on the ferroelectric properties [117, 118].

Mitmann et al. investigated the ferroelectric properties of a HZO thin film by varying the ozone dose time during ALD process [37]. Figures 5d, e display the evolution of phase portion of TiN/HZO/TiN capacitors influenced by the oxygen flow or ozone dose time in the sputtering or ALD process, respectively. Figure 5f shows the polarization–voltage curves of the TiN/HZO/TiN capacitor deposited via the ALD process. Here, the relative phase fractions were extracted from grazing incidence X-ray diffraction (GIXRD) measurement results. Oxygen flow is the flow rate of injected oxygen gas during the sputtering process, and ozone dose time is the time taken to inject ozone flow during the ALD process. Although precise quantification may be difficult, it is reasonable that the concentration of Vo in the HZO thin film might decrease due to the increase in the oxygen flow and ozone dose time. For the HZO thin film deposited via the sputtering process, an increase in the monoclinic phase fraction and a decrease in the orthorhombic phase were observed with an increase in the oxygen flow. In case of the HZO thin film deposited by the ALD process, similarly, Similarly, as the ozone dose time increased, the monoclinic phase fraction increased, and the orthorhombic phase fraction decreased. The similar trend was observed by Pal et al. where the fraction of the monoclinic phase increased and leakage current decreased with an increase in the ozone dose time during ALD [119]. This behavior may be explained by the significant decrease in the concentration of Vo in the thin film due to excessive injection of oxygen sources. Additionally, this result alludes to the following tendencies: (1) An increase in the concentration of Vo in the HZO thin film may further stabilize the monoclinic phase; and (2) excessively high or low Vo concentrations may rather make the orthorhombic phase unstable. Therefore, an oxygen dosage optimized for the Vo concentration and enhancing the ferroelectric properties should be supplied to the HZO thin films.

In the polarization–voltage curve of Fig. 5e, the HZO thin film yields the maximum double remanent polarization (2Pr) value of approximately 42 μC/cm2 for an ozone dose time of 0.5 s. In addition, a distinctly low Pr value is observed in films with the shortest and longest ozone doses of 0.1 and 20 s, respectively. The notable point is the difference between before (black line, pristine) and after (blue line, wake up) the wake-up cycle of HZO in Fig. 5e. The ferroelectric HfO2 thin films show phenomenon in which Pr value increases by repeated electric-field cycling, so-called the wake-up effect [120,121,122,123,124]. The origin of this effect is still debatable, but the general consensus is that it is the redistribution of concentrated Vo in the interface region and transition from the tetragonal to orthorhombic phases [95, 100, 125]. Based on this, the wake-up effect according to the phase fraction can be identified by the polarization–voltage curve in Fig. 5e. The HZO thin film with an ozone dose time of 0.1 s exhibits a pinched hysteresis loop with a relatively low Pr value in the pristine state. After the wake-up cycle, however, distinct changes, such as an increase in Pr and an open hysteresis loop, are observed. On the other hand, in the HZO thin film with an ozone dose time of more than 5 s, where the tetragonal phase fraction is remarkably reduced, no evidence of the evolution of Pr value and/or hysteresis loop was found during the wake-up field cycling. This result could be understood through the decrease in the tetragonal phase fraction in the pristine state, one of the hypothetical origins of the wake-up effect [100, 123, 124]. The decreased tetragonal phase fraction can be explained from a previous report that the relative fraction of the tetragonal phase may increase as the Vo concentration in the ferroelectric HfO2 thin film is increased [50]. It could be the reason for the strongest wake-up effect in the HZO thin film, which suffered the shortest ozone dose time, shows the highest Vo concentration. Furthermore, if the ozone dose time is increased, the injection of more oxygen atoms can decrease the relative fraction of the t-phase. This can lead to a high Pr value and relaxation of the wake-up effect. The electrical properties, including the wake-up effect, of the HfO2 thin film under electric-field cycling will be expounded in Sect. 3.2. Notably, The effect of Vo can appear as the formation of a local electric field or domain-wall pinning, which is different from the effect on the relative free energy of the various polymorphs discussed earlier. However, with any intermediate mechanism, the high Vo concentration in the pristine state would cause a rather strongly pinched hysteresis owing to the wake-up effect.

A phase change from tetragonal to the orthorhombic phase is considered the origin of the wake-up effect of ferroelectric HfO2, as stated previously. However, a recent study reported that the wake-up effect and fatigue phenomenon could be attributed to the phase transition from the antiferroelectric orthorhombic Pbca phase to the ferroelectric orthorhombic Pca21 phase [9]. Cheng et al. reported the movement of the oxygen atom and the phase transition mechanism according to the electric field cycling of the HZO thin film. The STEM-annular bright-field (ABF) mode, which has a high detection sensitivity for oxygen ions, was used for the analysis. The coexistence of Pbca and Pca21 phases is confirmed in the pristine state of HZO film; however, the Pca21 phase becomes more dominant than the Pbca phase in the wake-up state. From the STEM-HAADF analysis, the thickness of tetragonal phase increased from 0.75–1.0 nm to 0.94–2.11 nm after electric field cycling. This result contrasts the common theory that the HZO thin film suffers phase transition from the tetragonal phase to the orthorhombic phase after electric field cycling. The increase in the tetragonal phase fraction after electric field cycling can be attributed to an increase in the concentration of Vo, which stabilizes the tetragonal phase at the interface between the HZO thin film and the TiN electrode. This result is a significantly different interpretation from the wake-up and fatigue mechanisms of previously reported HfO2 thin films. Therefore, further investigation is needed to understand the wake-up effect and the fatigue mechanism.

In summary, the spatial distribution, concentration, and charge state of Vo in a HfO2 thin film strongly influence the polymorphism of HfO2. In the bulk region of HfO2, Vo help stabilize the polar orthorhombic phase. At the interface region, however, the lower concentration of Vo could produce a more stable polar orthorhombic phase. Vo2+ with a positive charge state can contribute more to the stabilization of the polar orthorhombic phase than neutral Vo. The stabilization of orthorhombic phase by the manipulation of Vo concentration is also strongly associated with the enhancement of the ferroelectric properties. The ferroelectric properties, such as Pr value and wake-up effect, can also be controlled by this approach. These results suggest that the design of Vo is key to improving the ferroelectricity of HfO2.

3.2 Effect of Vo on reliability of HfO2

To apply HfO2 ferroelectrics to practical memory devices, the behavior of Vo under repetitive application of electric fields must be understood. Particularly, compared to perovskite ferroelectric materials (coercive field Ec = approximately 10–100 kV/cm), the high Ec of HfO2-based ferroelectrics (approximately 1,000–2,000 kV/cm) can sufficiently mobilize the Vo through the thin film [10, 63, 126,127,128]. Moreover, various reliability issues of ferroelectric HfO2 thin films, such as wake-up effect and fatigue, are closely related to the formation and migration of Vo. Therefore, the behavior of Vo under the external electric field of the ferroelectric HfO2 thin film should be meticulously considered for the application of the HfO2 to the ferroelectric memory application.

The wake-up effect is a well-known reliability challenge associated with HfO2-based ferroelectrics. Per research, the redistribution of Vo, which are initially concentrated at the interface of the HfO2 thin film, under electric field cycling and the subsequent phase transition from non-ferroelectric tetragonal phase to ferroelectric orthorhombic phase are plausible origins of the wake-up effect [63, 95, 100, 123,124,125]. Additional electric field cycling after the wake-up step causes a decrease in Pr, which is known as fatigue [95, 96, 129]. One of the origins causes of fatigue could be the charge trap and ferroelectric domain-wall pinning due to the accumulation of Vo [96]. The wake-up effect and fatigue evidently influence the ferroelectricity and reliability of HfO2. In this section, therefore, the behavior of Vo in HfO2 thin films by external electric fields cycle and the evolution of ferroelectricity will be discussed.

Figure 6a illustrates the typical polarization–electric field and current–electric field curves of a HfO2-based ferroelectric thin film after wake-up and fatigue steps. In the pristine state, two distinct current peaks are observed, which are the origins of the pinched hysteresis loop. After electric-field cycling, however, the peaks merge and the hysteresis loop gradually opens and evolves into a single hysteresis loop. In the fatigue step, a decrease in the current peak intensity and Pr value were observed.

Fig. 6
figure 6

a Polarization–electric field and current–electric field curves of a HfO2 ferroelectric thin film for pristine, wake-up, and fatigue states. Evolution of the Preisach/switching density (ρ) of the HfO2 thin film after b ~ 1, c ~ 103, and d ~ 10.6 cycles corresponding to the pristine, wake-up, and fatigue states, respectively. e Evolution of the 2Pr, leakage current, defect concentration, and Ebias through repeated electric-field cycling. b-e reproduced with permission from [100]

The wake-up effect, fatigue, and electrical breakdown are strongly correlated to the formation and migration of Vo, so it is necessary to understand the behavior of Vo under electrical cycling. Pešić et al. discussed the migration of Vo under electric-field cycling using the first-order reversal curve (FORC) measurement technique [100]. The FORC measurement is an effective technique that can electrically analyze the redistribution of Vo by measuring changes in the electric field inside a material [130]. Figures 6b–d show the FORC measurement results of a TiN/Gd:HfO2/TiN capacitor after several electrical cycling. The results after 0, ~ 104, and ~ 106 cycles correspond to pristine, wake-up, and fatigue, respectively. In the pristine state, two separate Preisach/switching current density (ρ) peaks appear. These peaks are the origins of the two current peaks observed in the HfO2 thin film in the pristine state, as depicted by the current–electric field curve in Fig. 6a. The mechanism of this peak separation is the internal bias from the Vo concentration on a specific region (generally the interface region) [63, 100, 123]. After ~ 103 electric-field cycles, which corresponds to the wake-up cycle, the two peaks merge into one. In addition, as shown in Fig. 6a, the pinched hysteresis loop opens and the two current–electric field curve peaks integrate. This means that Vo, which had spatially inhomogeneity, was redistributed by repetitive application of electric field to fix the asymmetry of the internal bias. Additional electric-field cycling causes fatigue and a decrease in the intensity of the current–density peak. Considering the low Pr value of the fatigued state and the decline of peak intensity of the current–electric field curve in Fig. 6a, fatigue evidently degrades the ferroelectric properties of the HfO2 thin film. These results suggest that the migration of Vo under repeated external electric field cycling has a dominant effect on the ferroelectricity and reliability, such as Pr value, wake-up effect and fatigue, of HfO2 ferroelectric thin film.

Figure 6e highlights the variation of the electrical properties of HfO2 with the number of electric field cycles. The wake-up and fatigue steps correspond to the stage before and after ~ 103 cycles, respectively. In the wake-up effect step, a gradual increase in Pr and a decrease in the internal bias (Ebias) were observed. Conversely, after the fatigue step, the degradation of device characteristics deteriorates, such as a decrease in Pr value and an increase in leakage current density. These trends can be explained through the change and redistribution of defect (Vo) concentration. In the wake-up step, the concentration of Vo is maintained at an approximately constant value. Meanwhile, the asymmetry of Ebias is gradually alleviated with the number of electric-field cycles. This is because no new Vo were produced during the wake-up step, and the Vo located at the interface region were redistributed into the bulk region of the HfO2 thin film. On the other hand, in the fatigue step, the concentration of Vo rapidly increases, and the internal bias did not change. This indicates the formation of new Vo in all regions of the thin film. An appropriate Vo concentration in the HfO2 thin film can stabilize the ferroelectric orthorhombic phases, whereas excessive Vo can rather degrade the stability of orthorhombic phases [99, 131]. Moreover, an increase in the leakage current density is related to the concentration of charge defects, such as Vo in ferroelectric thin films [131, 132]. In summary, the main mechanisms of the wake-up effect and fatigue in ferroelectric HfO2 films are the redistribution and additional formation of Vo under repetitive electric field, respectively. Electric-field cycling alleviates the inhomogeneous distribution of Vo and triggers the wake-up effect, which decreases the internal bias and increases the Pr value. Excessive electric-field cycles might cause fatigue with the additional generation of Vo, which causes an increase in the leakage current and a decrease in the Pr value. As Vo strongly influences the ferroelectricity and electrical properties of HfO2, which are associated with changes in the concentration and migration of Vo. Therefore, to stably apply the HfO2 thin film to a ferroelectric memory device, it is necessary to carefully control the concentration and migration of the Vo.

The concentration, location, and migration of Vo have a strong effect on various properties such as ferroelectricity, polymorphism, and reliability of the HfO2 thin film. We will review suitable approaches for improving improve ferroelectricity by understanding the influence of Vo.

3.3 Strategies to control Vo of HfO2 and enhance ferroelectricity

As discussed in the previous sections, Vo in the appropriate concentration can stabilize and enhance the ferroelectric properties of HfO2, whereas excessive concentrations or asymmetric distribution of Vo can deteriorate the ferroelectricity and/or reliability. Therefore, the optimized fabrication and/or engineering method to induce the appropriate Vo concentration in HfO2 should be prudently considered. To apply ferroelectric HfO2-based thin film to the ferroelectric memory devices, strategies to control the Vo concentration in HfO2-based thin films have been proposed by several studies [133,134,135,136,137,138]. In this section, we review several previous studies that have reported improvement of ferroelectricity and reliability by controlling the concentration of Vo via various engineering idea.

Chen et al. controlled the Vo concentration at the HZO thin film–TiN electrode interface by NH3 plasma treatment [137]. Figure 7a illustrates an EDS map of a TiN/HZO/TiN capacitor with and without NH3 plasma treatment. It is notable that NH3 plasma treatment increases the oxygen atom fraction on the interface region of HZO, while reducing the oxygen atom fraction of the interface region of TiN electrode side. This indicates that NH3 plasma treatment can decrease the Vo concentration on the HZO side of the interface. Figure 7b shows the evolution of various electrical properties with the number of electric-field cycles of ~ 106 cycles for the HZO thin film with and without NH3 plasma treatment. The HZO thin film without NH3 plasma treatment exhibits the typical wake-up effect and fatigue. Over approximately 103 cycles, the Pr value gradually increased, while the leakage current did not change significantly. This may be because of the redistribution of Vo without the formation of new Vo, indicating the wake-up effect. After 103 cycles, there are clear signs of fatigue, an increase in the leakage current and a decrease in the Pr value. On the other hand, the HZO thin film with NH3 plasma treatment exhibits low leakage current density as well as mitigated wake-up effect and fatigue. Consequently, treatment of HZO with NH3 plasma is a robust engineering approach for producing stable memory devices without drastic changes by electric-field cycles. This result is meaningful in that it demonstrates the improvement of reliability of HfO2 ferroelectric thin films via manipulation of Vo concentration.

Fig. 7
figure 7

a EDS depth profile, b Evolution of Pr and leakage current as functions of number of cycles for TiN/Hf0.5Zr0.5O2/TiN capacitors with and without NH3 plasma treatment. c Polarization–voltage curves. d Endurance test of Cu/VOx/Hf0.5Zr0.5O2/TiN capacitor with various sol–gel solutions concentration. e Transient currents during constant voltage stress. f Endurance test of TiN/Hf0.5Zr0.5O2/TiN capacitor annealed with 1- 2-step RTA processes. a, b reproduced with permission from [137]. c, d reproduced with permission from [138]. e, f reproduced with permission from [145]

Zhang et al. reported improvement in the ferroelectricity and endurance of the HZO thin film by inserting a vanadium oxide (VOx) capping between the TE and the HZO thin film [138]. Figure 7c illustrates the polarization–voltage curve of a Cu/VOx/HZO/TiN capacitor with a VOx capping layer fabricated with sol–gel solutions (concentrations of 1, 1.67 and 2 mg/mL), respectively. The inset depicts the stack of the capacitor. Here, the samples are named HZO/VOx-1, HZO/VOx-1.67, and HZO/VOx-2, depending on the concentration of the sol–gel solution (1, 1.67, and 2 mg/mL, respectively) for VOx. The sample without a VOx layer was named HZO/0. The highest 2Pr value of ~ 36.9 μC/cm2 was observed in the HZO/VOx-1.67. Figure 7d shows the endurance test results for HZO/0 and HZO/VOx-1.67. HZO/0 exhibits fatigue after ~ 106 cycles and a deteriorated Pr value of ~ 0 μC/cm2 after ~ 109 cycles. On the other hand, HZO/VOx-1.67 with the VOx capping layer exhibited enhanced fatigue resistance, with a decrease in Pr of only ~ 10.9% even after ~ 109 cycles. This result can be explained by the difference in the binding energy with various metal and oxygen. The V–O binding energy (625.0 ± 19.0 kJ/mol) energy is lower than those of Hf–O (810.0 ± 13.0 kJ/mol) and Zr–O (766.1 ± 10.6 kJ/mol). Hence, the VOx capping layer could supply the oxygen atom to the HZO thin film and control the oxygen concentration [139]. Other studies have reported a similar phenomenon in HZO thin films using oxide electrodes, such as oxides of Ru and Mo [133]. MoO3 and RuO2 have lower formation enthalpy than HfO2 and ZrO2 from room temperature to RTA temperature (~ 800 ℃) [140,141,142,143,144]. By supplying oxygen atoms to the HZO thin film during thermal processes such as ALD and RTA, these oxide electrodes serve as oxygen supplier and thus control the Vo concentration. Therefore, by reducing the Vo concentration in the interface area of HfO2, the engineering technique of supplying oxygen atoms to HfO2 thin film can contribute to the enhancement of endurance properties and orthogonal stabilization.

Su et al. proposed a strategy to control the Vo concentration of the TiN/HZO/TiN capacitor by splitting the crystallization process into two sequential steps [145]. The 1-step RTA sample, which refers to the general RTA process for a HfO2 thin film, was conducted in an oxygen-free atmosphere at 600 ℃. Meanwhile, the 2-step RTA sample for controlling the Vo concentration was first crystallized at 600 ℃ and subjected to additional low-temperature heat treatment at 400 ℃ for approximately 30 min in O2 atmosphere. Figures 7e, f show the electrical stress resistance and endurance measurement results for the 1-step and 2-step RTA samples. For a constant voltage stress of 3 V, the electrical breakdown of the 2-step RTA sample occurred after that of the 1-step RTA sample.

Moreover, the 2-step RTA sample shows an ~ 102 -times improved endurance cycle compared to the 1-step RTA sample. It was also attractive that the 2-step RTA sample exhibits a lower leakage current and a higher 2Pr value than the 1-step RTA sample for the same voltage stress time and number of electric-field cycles. An XPS analysis by Su et al. revealed that 1-step and 2-step RTA samples had Vo concentrations of 2.8% and 0.2%, respectively. This observation can be explained by the endurance of the relatively low Vo concentration by the 2-step RTA sample compared to that by the 1-step RTA under a longer electrical stress time. Furthermore, the increased Pr value and switching endurance indicate that the 2-step RTA process evidently contributed to the enhancement of ferroelectric and reliability properties of HfO2. This suggests that the 2-step RTA can improve the reliability of the HfO2 device without degrading its ferroelectricity and leakage current characteristics.

In summary, various engineering techniques can be implemented to regulate the Vo concentration in an HfO2 thin film for enhancing of Pr value, mitigating the wake-up effect and/or fatigue, and improving endurance. Strategies such as plasma treatment, artificial interface layer insertion, and RTA process improvement can effectively mitigate the nonideal effects of excess Vo. If the endurance of the HfO2 thin films can be further improved without critically deteriorating Pr values, the HfO2 thin films could be a promising candidate for future ferroelectric memory applications.

4 Effect of Vo in HfO2 base RS devices

The advantages of HfO2, which include its CMOS compatibility, low process temperature, high-k dielectric property, and process scalability, were extensively discussed and demonstrated in the literature [146]. These attributes establish HfO2 as a highly suitable material for a switching matrix in nonvolatile memory and neuromorphic computing applications, offering reliable and high-performance memory functionalities. The intrinsic factors of HfO2 can be seamlessly controlled, including the crystallinity, GB, stoichiometry, and oxide layer structure, through deposition or post-treatment processes because HfO2 is an established material that acts as a switching layer in Vo-mediated RS devices among CMOS-compatible substances. In addition, extrinsic alterations such as doping, insertion of an interfacial layer, and engineering of oxide–electrode interface can manipulate the Vo dynamics for improving device performance.

However, the HfO2-based RS devices still need to be improved in terms of their performances such as operation energy, uniformity, gradual conduction modulation, and switching linearity for large-scale integration; the deliberate modulation of Vo dynamics and CF formation are crucial for optimizing HfOx-based RS devices [147,148,149,150,151,152,153,154,155]. Therefore, numerous studies have been conducted for improving the uniformity and multilevel RS characteristics of HfOx-based RS devices via different engineering approaches. In Sect. 4.1, the valance change mechanism (VCM), which is a filament-evolution mechanism based on Vo dynamics, is described. In Sect. 4.2, the effect of Vo on VCM-based RS device performance is discussed. Based on the results, Sect. 4.3 introduces the various engineering strategies that have been utilized for controlling the Vo concentration in HfO2 and improving the RS characteristics.

4.1 Valance change mechanism (VCM)

The switching behavior of RS devices is related to the various phenomena of CF evolution. The CF evolution occurs via electrochemical metallization (ECM) or VCM, which depends on the materials that compose the CF [156,157,158,159,160]. In HfO2-based RS devices that lack an active electrode for delivering highly mobile cations, the Vo-mediated VCM is of vital importance for both device operation and CF formation. Vo are defects in the crystal lattice with missing oxygen atoms, which result in a change in the valence state of the nearby hafnium atoms. Thus, the presence, electromigration, and aggregation of Vo enable the formation of conductive paths within the insulating HfO2 material, which in turn produces the switching behavior observed in RS devices.

The switching mechanism of VCM-driven RS devices is associated with the formation, migration, aggregation, and rearrangement of Vo [21, 79]. The VCM is commonly observed in metal oxides that exhibit a remarkable mobility of ions within the crystal lattice, which includes the migration of oxygen ions. The migration of these ions alters the local stoichiometry of the switching layer, thereby inducing a valence change in the cation sublattice and a change in electronic conductivity. Vo can be deliberately introduced for operating RS devices by applying a specific set of electrical pulses. These pulses are referred to as “forming” pulses and used to initialize the device by generating a CF. The forming process involves the migration of Vo, which cluster to form a conductive pathway [161]. Once the CF has been formed, the movement of Vo becomes crucial for the subsequent switching operation. Under the application of an external electric field, Vo can migrate within the switching layer, which is facilitated by the presence of defects, GBs, or other structural irregularities in the material. Owing to the electrostatic forces, Vo tend to migrate toward the negatively biased electrode (cathode), which accumulates at the interface between the HfO2 layer and the electrode to form a CF. This establishes a low-resistance pathway by bridging the electrodes, i.e., switching ON the cell (LRS). The cell can be switched OFF (HRS) by repelling Vo from the electrode by applying a positive voltage. The evolution of CF with Vo migration dynamics at each operational stage of the switching mechanism is illustrated in Fig. 8a.

Fig. 8
figure 8

a Schematic of the working principle of a VCM device. b Various strategies to control the effect of Vo properties on the metrics of the RS device performance

4.2 Effect of Vo on RS device performance

The presence of Vo in an RS device has a significant effect on its multiple performance aspects. It plays a crucial role in forming and dissolving CFs, which directly affects the switching behavior between HRS and LRS. The migration and accumulation of Vo affect the stability and endurance of the device, as well as the switching speed and voltage. In this section, the device performance is explicitly divided into various metrics and the influence of Vo on them is analyzed [162, 163].

4.2.1 Forming voltage

The forming voltage is the initial voltage required to create a conductive pathway for operating an RS device. For VCM-driven RS devices, the resistance value in the HRS is significantly affected by the Vo generated during the forming process [164, 165]. A high forming voltage leads to the generation of excess Vo, thereby forming a strong CF and exhibiting higher current levels in the HRS. Furthermore, the forming process induces large device-to-device variation, making the switching behavior unreliable. As Vo plays an important role in facilitating the formation of CFs by providing mobile charge carriers, the forming voltage depends on the concentration of the intrinsic Vo of the switching layer and other extrinsic factors, which include electrode material, interfacial layer, and dopants. That is, CF-free RS devices can be designed with appropriate switching materials and device structures to improve Vo controllability.

4.2.2 Variation

The nonuniform distribution of Vo in the switching matrix, which results in different filament paths and resistive states, causes performance variation [166, 167]. Fluctuations in the Vo concentration affects RS parameters such as the resistance level, switching voltage, and endurance characteristics; this results in device-to-device and cell-to-cell variations. Studies have explored various techniques, which include optimizing the deposition and distribution of Vo within the resistive layer, controlling the fabrication process parameters, and developing materials with smoother Vo profiles for addressing these nonuniformities and improving the device performance. The device-to-device, cell-to-cell, and cycle-to-cycle variations in Vo-mediated RS devices can be reduced by achieving more precise control over the Vo distribution. Knowledge and regulation of the distribution of Vo can help mitigate these nonuniformities and improve the overall performance and reliability of RS devices.

4.2.3 Switching behavior

Switching behavior can be classified into abrupt and gradual behaviors based on its characteristics. While abrupt switching is characterized by a fast switching speed, it is limited in terms of endurance and device-to-device variation caused by the nonuniform formation and rupture of CFs [168, 169]. The transition from the abrupt to gradual RS behavior is influenced by the quantity of Vo in the switching layer [170, 171]. The transition between the two resistance states is sudden and discontinuous when the layer consists of fewer Vo, which results in abrupt switching behavior. The limited number of Vo restricts the movement of ions and the formation of CFs, leading to a distinct and rapid change in resistance. Conversely, when the layer contains sufficient Vo, the RS behavior becomes gradual because a higher concentration of Vo enables a more continuous transition between HRS and LRS. This gradual switching behavior is due to the enhanced ion migration and CF formation, allowing for an analog change in resistance. That is, the availability of Vo directly influences the switching behavior in RS devices.

4.2.4 Switching speed and switching voltage

The migration of Vo and formation/dissolution of CFs determine the switching speed between HRS and LRS. The presence of sufficient Vo facilitates the formation and rupture of CFs, enabling faster transitions between different resistive states and faster switching speeds [172]. Moreover, the concentration and distribution of Vo influence the switching voltage required to trigger the RS operation. The higher concentrations of Vo tend to lower the switching voltage of the device because of the increased availability of charged species, such as oxygen ions, which can migrate more easily in the presence of Vo. Thus, a lower voltage is required to initiate the movement of ions and the formation of CFs, enabling RS. Therefore, controlled concentrations of Vo can enhance the switching speed by promoting ion movement and filament formation. In addition, higher Vo concentrations can reduce the switching voltage required for RS.

4.2.5 Conductance modulation

The development of RS devices that exhibit linear conductance modulation is pivotal for their applications in multilevel memory or neuromorphic computing systems. The enlargement or disruption of Vo CFs in an RS device is governed by two mechanisms with different growth and dissolution rates: a fast redox process and a slow Vo diffusion process [173, 174]. RS devices exhibit nonlinear conduction modulation under the influence of these two processes with different reaction rates. The abrupt variation of conductance results from the inhomogeneous growth/dissolution of Vo filaments, which is in turn caused by the simultaneous involvement of the two mechanisms with different growth/dissolution rates. In this context, the growth/dissolution of CF should be controlled by either mechanism to improve the linearity of conductance modulation. The diffusion reaction becomes the predominant growth mechanism of Vo CFs when only a few Vo are present [175]. In contrast, the redox reaction takes precedence when numerous Vo exist [173]. Hence, the number of Vo in the RS device must be controlled so that only one mechanism is involved in the growth of CF to ensure linear conductance modulation.

4.2.6 Endurance and retention

Endurance is a crucial factor in ensuring reliable and long-lasting device performance. During the SET/RESET switching cycle in RS devices, the irreversible defect generated by excess Vo accumulation deteriorates the oxide layer, which leads to device failure. The generation of irreversible defects in the switching layer must be curtailed because the accumulation of additional Vo leads to failure with the generation of new CFs [169, 176]. Moreover, the retention properties of RS devices, which reflect their ability to retain stored information, are influenced by various factors, including device structure, materials, and current. Particularly, the activation energy, which is determined by the energy of the highest transition state during the Vo diffusion process, is a critical factor for the retention behavior [177].

In summary, exploiting the technical solution to control the Vo dynamics is vital for improving device performance. Therefore, the engineering approach for improving the performance of the HfO2-based RS device, which involves intrinsic factors (crystallinity, stoichiometry) and extrinsic factors (doping, interfacial layer, electrode, and environment) (Fig. 8b) is discussed in the following section.

4.3 Strategies to control the Vo of HfO2 and enhance RS characteristics

4.3.1 Crystallinity (GB)

The crystallinity of HfOx thin films, which includes the size and shape of the grains and the presence of GBs, is one of the most significant factors influencing the Vo motion dynamics for RS behavior. In the early stage of the Vo-mediated RS research on HfO2, several studies were conducted to identify the active switching region where the Vo is generated, migrated, and agglomerated, and thus formed CFs. Therefore, several theoretical and experimental studies investigated the relationship between the crystallinity of HfO2 and the Vo motion dynamics and its RS behavior. Most theoretical studies were based on DFT calculations [83, 85, 178, 179], suggesting that GBs in the crystalline HfO2 layer were the preferred locations for generating and migrating Vo caused by the lower diffusion barrier for oxygen ions (vacancies) diffusion [149]. Based on the Ab initio model, Bersuker et al. demonstrated the evolution of a current pathway at the GBs in monoclinic HfO2 during the forming process [84]. In their study, pre-existing Vo were energetically favored to segregate near the GBs, eventually forming the preferential leakage current pathway [180,181,182]. This conjecture from theoretical calculations was verified by c-AFM experiments. By comparing the electrical and topographical data of HfO2 stacks in amorphous and polycrystalline HfO2 phase thin films, Lanza et al. analyzed the effects of crystallinity on the RS behavior [178, 183, 184]. In the case of amorphous HfO2, leaky sites were arbitrarily distributed throughout the measured area. In contrast, the current increased at the area with the GBs of crystalline HfO2, indicating a correlation between topographic features and output current. When a voltage is applied to the c-AFM probe tip at a random spot, it induces two types of electroforming processes. In most regions, the process occurs with a high voltage (> 12 V), while in specific regions, such as the GBs, it occurs at significantly lower formation voltages (4 < V < 6.5). (Fig. 9(a)). After the high-voltage forming process, no RS behavior was detected in the subsequent I–V curves for those most regions. However, in areas where the forming voltage is low, a typical bipolar RS behavior is observed (Fig. 9b). When the probe tip was placed on the grain, it required a high voltage to continue the forming process, resulting in a hard dielectric breakdown of the device. Conversely, when the probe was placed on the GB, it required a low forming voltage to generate the CFs without causing a dielectric breakdown of the device. In addition, the study captured the topographic and output current profiles after iteratively performing the write–read–erase–read operations. The profiles confirm that the conductive pathways at the GBs can be erased through the RESET operation (Fig. 9c).

Fig. 9
figure 9

Low forming voltage exhibited on low-energy GBs in HfO2-based RS devices. a Forming process in annealed polycrystalline HfO2 at specific locations. b Typical bipolar RS behavior observed at the low voltage forming site (GB location). c Current map measurement in a write–read–erase–read cycle, which allows the locations to recover their insulating properties. a, b, c reproduced with permission from [184]. d, e reproduced with permission from [187]

The degree of crystallinity and the GB of the HfO2 thin film share strong correlations with the RS properties. Several studies have explored methods to enhance the RS functionality by controlling the degree of crystallinity and the GBs in HfO2 thin films [83, 185, 186]. Petzold et al. demonstrated the development of forming-free RS devices by engineering the GBs of HfO2 [187]. The study focused on the influence of Vo with respect to the GB and fabricated a device based on an epitaxial stack combination of TiN (111) and monoclinic HfO2 (11 \(\overline{1 }\)) in TiN/HfO2/Pt stacks. They grew TiN and HfO2 layers on a (0001)-oriented Al2O3 substrate using molecular beam epitaxy (MBE). This resulted in a defined subset of GBs with high symmetry (Fig. 9d). The electrical characterization of the devices reveal that they exhibit reliable and forming-free RS behavior (Fig. 9e). According to the study, the uniform distribution of forming and switching voltage might have emerged from the formation of CFs on a predefined path formed along the low-energy GBs, which would require a low forming voltage, thereby forming uniform filaments. According to these studies, GBs can be used to precisely control the Vo motion dynamics, resulting in improved RS performance. However, the randomized GBs in each cell can impact the variability of the forming voltage, which can hinder the device-to-device uniformity. Hence, the formation of a well-defined GB connecting the TE and BE is considered important for precisely controlling the Vo in RS devices [188, 189].

4.3.2 Stoichiometry (Oxygen deficiency)

The stoichiometry of the HfOx material layer can be adjusted by controlling the deposition condition during the fabrication process or during post-processing. This significantly influences the RS characteristics as it requires adjusting the concentration of Vo (oxygen deficiency) in the RS layer. Therefore, stoichiometry must be considered in addition to the Vo motion dynamics induced by the voltage application during RS operation. Several studies have indicated that a well-controlled oxygen stoichiometry determines the formation of Vo CFs [190, 191]. Employing oxygen engineering, Kaiser et al. identified the oxygen-dependent phase transitions from stoichiometric hafnia (m-HfO2) to hexagonal phase hcp-HfO0.7 [192]. Oxygen vacancies were introduced into the crystalline matrix with a decrease in oxidation conditions, thereby forming Vo defects. Therefore, through oxygen engineering, a Vo-driven significant decrease in resistivity from the insulating m-HfO2 (~ 1010 Ωm) over cubic c-HfO1.7 (~ 10–4 Ωm) to hexagonal hcp-HfO0.7 (\(\sim 7.3\times {10}^{-6}\) Ωm) is induced. McKenna et al. emphasized the importance of sub-stoichiometric HfOx in ensuring the uniform nucleation and growth of CFs during the forming process [193]. Through a first-principles study, they identified that an optimal HfOx stoichiometry with x in the range 1.50–1.75 was necessary for the efficient nucleation and growth of stable CFs during the forming process. The activation energy required for the nucleation of the CF is reduced within this optimal stoichiometric range, thereby facilitating the formation of clusters that serve as seeds for CF growth. Further, the DFT calculations demonstrated the stability of Hf-rich precipitates and their growth propensity by the outward diffusion of oxygen ions. This supplemented additional evidence that the range of 1.50–1.75 was conducive to the formation of CFs. In addition, the relationship of the forming voltage with the stoichiometry of HfO2 was investigated by comparing the cases of the stoichiometric (m-HfO2) and oxygen-deficient (t-HfO2–x) films [194]. Both films were epitaxially grown by reactive molecular beam epitaxy (RMBE) under different oxidation conditions. Figure 10a shows the forming voltage of oxygen-deficient devices, which are composed of sub-stoichiometric HfO2-x, is independent of film thicknesses up to 200 nm; however, the forming voltage of stoichiometric HfO2-based devices increases linearly with the thickness after overcoming an initial energy threshold to activate the ionic transport for CF formation. In the sub-stoichiometric HfO2–x, which is a layer with a thin oxidized top layer (Fig. 10b), the forming voltage was constant because only one filament needs to be formed in the oxidized top surface layer. The reason for the constant forming voltage is that the surface layer maintains a constant thickness because of using the same sample treatment method. These results show that a highly oxygen deficient layer has sufficient conductivity and does not contribute to the increase in forming voltage until the total oxide layer thickness is increased to 200 nm. Consequently, the oxygen-deficient layer acts as a Vo reservoir by employing well-defined oxides with controlled oxygen stoichiometry as the switching layers, which enables the implementation of forming-free switching behavior.

Fig. 10
figure 10

Reduced forming voltage in oxygen stoichiometry-engineered HfO2-based RS devices. a Forming process in annealed polycrystalline HfO2 according to the specific location. Low variation of forming voltage with thickness for oxygen deficient HfO2–x film. b Model for the thickness dependence of forming voltage in stoichiometric (left) and oxygen deficient HfO2–x (right). c Switching characteristics tuned in oxygen deficient t-HfO2–x. Schematic model of filament formation in oxygen engineered HfOx-based RS devices. a, b reproduced with permission from [228]. c reproduced with permission from [12]

Sharath et al. used MBE to precisely engineer different phases of HfO2, including hcp-Hf(O), t-HfOx, and m-HfO2, with varying oxygen stoichiometries [12]. The RS devices using the t-HfOx and m-HfO2 phases exhibited distinct forming and switching operations (Fig. 10c). In t-HfOx-based devices, intrinsic Vo caused a significant reduction in the forming voltage. Meanwhile, in m-HfO2, a higher number of Hf–O bonds were broken to generate Vo, which resulted in a higher forming voltage (VF ≈ − 7.0 V) and stronger filaments with more accumulation of oxygen ions near the TiN interface. Unformed and pristine t-HfO1.5 already contains a homogeneous distribution of numerous Vo, and therefore, the filament is formed under a considerably weaker electric field (VF ≈ − 2.2 V), where only a few Hf–O bonds needed to be broken because of the presence of pre-existing Vo.

The stoichiometric control of HfO2 in RS devices has significant implications for improving their RS properties. The literature demonstrated that the precise engineering of the oxygen stoichiometry in HfOx-based metal–insulator–metal structures can help significantly reduce the forming voltage. Consequently, comprehending the material conditions and RS operation aids in establishing the correlation between the structural and functional properties of RS device materials. These approaches enable the development of reliable RS devices by leveraging the ability to engineer the oxygen stoichiometry in HfOx.

4.3.3 Doping

In addition to the intrinsic properties of the HfO2-based RS layer, such as crystallinity and stoichiometry, extrinsic factors can be engineered to change the properties of the layer. These factors facilitate the highly controllable Vo dynamics to improve the RS performance. The Vo motion dynamics must be modulated to precisely control the rupture and reformation process of Vo-based CFs in the HfO2 RS device during operation. Several studies have proposed a few viable solutions, including element doping, metal/oxide interface modulation, and multilayer structures.

Zhao et al. utilized FT calculations to investigate the effect of metal dopants in these devices [57], and they analyzed the interaction between metal dopants and Vo based on formation energy. Consequently, they discovered that p-type metal dopants had a considerable effect on the formation of Vo CFs. Unstable metal–oxygen bonds were formed because of the lack of valence electrons in the p-type metal dopants, which in turn generated Vo around the dopants. In addition, metal dopants facilitate control over the mobility of Vo via the modulation of the activation barrier. Li et al. investigated the effect of Mg doping in HfOx by comparing TiN/HfOx/Pt and TiN/Mg:HfOx/Pt memristors [195]. The comparison showed that the Mg dopant tended to regulate the activation barrier energy of the adjacent Vo and alter the defect levels in monoclinic HfO2 [196]. This implies that a mutual-ion effect might occur between Vo2+ and Mg2+ during the RS operation when HfOx is doped with Mg. As shown in Fig. 11a, when Mg2+ migrates toward the Vo chain, it perturbs the consecutive channel of the original defect levels. The study found that regulation of the rate of Vo migration and the defect states of CFs enhance the controllability of the RS operation. Lee et al. investigated the doping effects of various aliovalent ions (Mg2+, Al3+, Nb+5) into HfO2 on RS characteristics [197]. Doping with aliovalent elements caused an increase in the nonlattice oxygen concentration and a reduction in the grain size in HfO2. The RS characteristics of the doped HfO2 are significantly influenced because the grain boundaries can act as favorable diffusion paths for atomic diffusion; therefore, the characteristics exhibit low forming voltage and improved uniformity in doped-HfO2 films. Moreover, Roy et al. demonstrated the beneficial effects of Al doping on the enhancement of the HfO2-based memory device performances [198]. They investigated the effect of Al doping on the formation of Vo, demonstrating that the 16.5% Al doping concentration enhances the RS properties of the device. The loss of oxygen in the HfO2 layer is induced with an increase in Al doping concentration, which increases the formation of Vo and decreases the forming voltage. The comprehensive experimental analysis involving TEM and operando HAXPS indicated that Al-doping enhances the formation of Vo in HfO2 and improves RS performances, thereby demonstrating the achievement of synaptic simulation. The effect of doping on RS performance was demonstrated with a Au-nanoparticle-doped HfO2-based device, in which the metal nanoparticles/crystals were embedded in a HfO2 layer [199]. Wu et al. explained that the particles induce defects (Vo) inside the oxide layer when metal particles were embedded in HfOx. These particles can enhance the local electric field, which is also a key factor in reducing the forming voltage [200, 201]. The improved overall device performance metrics are summarized in Fig. 11b.

Fig. 11
figure 11

Doping and interfacial engineering of HfO2-based RS devices. a Mg doping to modify Vo migration kinetics in HfO2. The multilevel I–V cycles with varying reset voltages of TiN/Mg:HfOx/Pt devices. b Improvement in performance of HfOx-based RS device using doping. c Enhanced switching characteristics for TiN/HfO2/Ti/HfO2/Pt/Ti stack RS devices, which use Ti as the interlayer. d Forming voltage distribution for PVD-TiN (left) devices and ALD-TiN (right) devices. e Structural modulation of HfO2 switching matrix as nanorod structures for exploiting the environment as the Vo reservoir. a reproduced with permission from [195]. b reproduced with permission from [199]. c reproduced with permission from [213]. d reproduced with permission from [215]. e reproduced with permission from [217]

Hence, studies demonstrated the effect of doping on Vo motion dynamics using various factors, including the energy required to generate or dissipate Vo in the HfO2 lattice, Vo mobility for migration, and intensity of the localized electrical field in the HfO2 layer. As the specific impact of doping on Vo in HfO2-based RS devices depends on different factors such as the dopant species, concentration, distribution, and the overall device design, experimental investigations need to be conducted to determine the optimal doping conditions that enhance the RS characteristics.

4.3.4 Interfacial layer

The incorporation of interfacial layers, such as 2D materials or other buffer layers, into the metal/oxide interface is a breakthrough approach for optimizing the interface properties in RS devices [202,203,204]. 2D materials were widely employed as interfacial layers in RS devices to reduce power consumption and mitigate detrimental surface effects [202, 205]. For example, graphene has been frequently employed as an interfacial layer to block atomic diffusion between the metal and the oxide. Zhu et al. used first-principles calculations to investigate the mechanism of the integration of graphene into the Au/HfO2 interface for enhanced interface properties [206]. They discovered that graphene influenced Vo diffusion by preventing strong interactions between the Au electrode and the HfO2 layer. This interfacial graphene effectively prevented the interface from becoming metallic and improved the dielectric reliability of HfO2. However, the study did not experimentally demonstrate the improvement of RS characteristics according to the graphene interfacial layer. Mannequin et al. experimentally investigated the improvement of RS properties in TiN/Pt/HfO2/graphene/Au stacks [207]. They discovered that the graphene interfacial layer acted as an oxygen reservoir, stabilizing oxygen ions released during the SET operation, which consequently improved the ON state retention.

The integration of a metal interlayer between the HfO2 layer and an electrode can impact the Vo behavior in different ways [208]. First, the metal interlayer can act as a diffusion barrier, limiting the migration of Vo from the HfO2 layer toward the electrode. This confinement of Vo within the HfO2 layer can stabilize the CF formation and improve the switching uniformity and endurance of the device. Moreover, the metal interlayer can facilitate oxygen exchange between the HfO2 layer and the electrode. Depending on the materials involved, it can capture oxygen from the electrode side or provide oxygen to the HfO2 layer. This exchange process influences the Vo concentration within the HfO2 layer, which will influence the RS characteristics. In addition, the metal interlayer can react with the HfO2 layer or the electrode material at the interface. These reactions can alter the local Vo concentration, modify the electrical properties at the interface, and subsequently impact the RS behavior. For example, utilizing Ti as the oxygen-gettering layer to produce TiOx/HfOx helped engineer highly reliable HfO2-based RS devices with exceptional performances [209,210,211,212]. Further, the required forming voltage and set voltage were significantly reduced. Lee et al. investigated the characteristics of RS in TiN/HfO2/Ti/HfO2/Pt/Ti stacks [213] and demonstrated that embedding reactive metallic layers, which act as an oxygen buffer layer, into the HfO2 films can effectively optimize the RS behavior. The employed TEM and EELS mapping and directly observed the accumulation of oxygen ion at the Ti/HfO2 interfaces and successfully improved the RS characteristics, such as a lower SET voltage and larger memory window (Fig. 11c). Furthermore, the significance of the position of the Ti layer was demonstrated [208]. By integrating the Ti adlayer on the top and bottom of the HfO2 layer in TiN/HfO2/TiN device stacks, Walczyk et al. investigated the impact of the Ti layer’s deposition position in RS operation. The XPS depth profile data in that study revealed that a Ti layer deposited on the top of HfO2 effectively attracts oxygen at the interface, and when deposited on the bottom of HfO2, the layer exhibits lower reactivity in attracting oxygen. This is because the Ti bottom adlayer was terminated by oxygen when the Hf(NMeEt)4 precursor was introduced and oxidized during atomic vapor deposition, thereby reducing the reactivity of the layer with the subsequently grown HfO2 film. However, as additional Ti layers are deposited on the top of the HfO2 layer, Ti atoms and clusters with unsaturated bonds are deposited, thereby increasing its affinity toward oxygen from HfO2. Consequently, the deposition of the Ti top adlayer led to the formation of a nonstoichiometric Ti/HfO2–x interfacial structure, which played an important role in achieving reliable RS.

The interfacial layer in HfO2-based RS devices can affect the Vo concentration and motion within the HfO2 layer through oxygen diffusion barrier effects, exchange processes, and interfacial reactions. These modifications can improve the RS characteristics, such as enhanced switching uniformity, stability, and endurance. These investigations highlight the criticality of interface management in influencing device performance. Establishing well-defined approaches for manipulating the physical and chemical characteristics of the interfacial layer/HfO2 interface is integral to attaining consistent and reliable RS performances.

4.3.5 Electrode

The choice of electrode material is vital for enhancing the controllability of Vo dynamics. Several studies have investigated the effect of the electrode material on RS operations. Padovani et al. employed Ti metal electrodes as a reactive buffer layer on the anodic side to attract oxygen atoms from the HfO2 layer, which promoted the formation of a sub-stoichiometric HfO2 region and improved the forming process in the RS operation [214]. This approach oxygen atoms to diffuse from the HfO2 layers to the Ti metal electrode, which in turn resulted in substantial oxygen deficiency in HfOx (x ≈ 1.4). The oxygen in the HfO2 layer shifted to the Ti layer and created an HfO2–x layer at the interface because Ti exhibits high oxygen-gettering activity. This Ti-induced sub-stoichiometric HfOx layer is important for developing low-voltage operating devices.

The accurately measured concentration of Vo is essential for improving its controllability. However, the overproduction of Vo poses a challenge to the reliability of the device; i.e., an optimal amount of Vo must be generated. Therefore, techniques that can prevent excess Vo must be explored. Yong et al. examined the relationship between RS and physicochemical properties of a TiN metal BE. They compared two types of HfOx-based RS devices fabricated with TiN BEs and deposited them using PVD and ALD [215]. The HfOx layer on the PVD-TiN electrode was more oxygen deficient than that on the ALD-TiN electrode. As the ALD-TiN electrode was fabricated with O2 plasma, the residual oxygen produced TiOyNz on its surface. Meanwhile, the PVD-TiN electrode was deposited in a stronger vacuum; the vacuum chamber was not connected to any oxygen source, which resulted in an oxygen-deficient HfOx layer on top. As illustrated in Fig. 11d, the PVD-TiN device exhibits smaller forming voltage values, which can be attributed to a higher concentration of Vo at the HfOx/PVD–TiN interface that effectively narrows the gap to be bridged for filament formation. Moreover, a higher initial defect density near the HfOx/PVD–TiN interface reduced filament base widening, which resulted in a thicker filament base than that of the HfOx/ALD–TiN interface. According to a simulation study by Carlo et al. [216], a thicker filament base at the bottom metal electrode interface can lead to progressive RESET behavior characterized by a gradual thinning of the CF and a moderate increase in the CF resistance. In addition, the larger area of the ruptured filament surface caused by a larger filament base resulted in a reduced memory window. Therefore, the formation of a thicker filament base at the HfOx/PVD–TiN interface led to gradual switching and a reduced memory window in PVD-TiN-based devices, as compared to the abrupt switching and large memory window caused by the sharp rupture in the thin CF formed within ALD-TiN devices. These studies indicated that the nature of constituent materials and deposition process influenced the Vo concentration at the interface of HfO2/electrode, and therefore, they played the definitive role of RS operation in HfOx-based RS devices.

4.3.6 Environment

During the switching operations, seamless gas exchange with the atmosphere improved the controllability of the redox reaction, which improved device performance. Kwon et al. demonstrated highly linear and symmetrical conductance modulation in HfO2-based RS devices by employing unique structures of HfO2 nanorods as a switching layer [217]. These nanorods enhanced the controllability of the redox reaction by permitting oxygen circulation between the oxide and ambient atmosphere (space between the nanorods) (Fig. 11e). The environment acted as a Vo reservoir and outlet for the emission of oxygen gas. Thus, this unique device structure precisely controlled the generation of Vo (oxygen ions) during RS operation, minimizing the randomness of the switching behavior.

5 Conclusion and outlook

Vo exerts both positive and negative effects on the ferroelectricity of HfO2-based thin films. The positive effect is the suppression of the formation of stable non-ferroelectric monoclinic phase. Although oxygen deficiency is helpful in reducing the monoclinic phase fraction of HfO2-based thin films deposited using ALD or sputtering, as a crystallographic defect, it forms localized trap sites that deteriorate the insulation properties of the device by increasing the local conduction and nonideal local field near the Vo. Moreover, the Vo retards the propagation of domain walls by decreasing the polarization switching speed, which is critical for maintaining the operation speed of ferroelectric memories. An increased Vo concentration enhances the relative stability and fraction of the non-ferroelectric tetragonal phase, which can degrade the ferroelectricity of HfO2-based thin films. Hence, an optimized Vo concentration range can help achieve both enhanced ferroelectricity and robust insulation free from charge trapping; moreover, the polymorphism of HfO2 can be controlled by the synergy of various factors including doping, thermal process, surface energy effect, and strain.

The three main types of ferroelectric memories categorized by their cell structure are as follows: (1) 1 transistor 1 capacitor ferroelectric random-access memories (FeRAM), (2) 1 transistor ferroelectric field-effect transistor (FeFET), and (3) 1 resistor ferroelectric tunnel junction (FTJ). Details of the ferroelectric memories with the different cell types are available in other reviews [218, 219]. Although the operation principles of the three devices are different, the properties they require to be considered as alternatives for state-of-the-art memory devices are similar: (1) high information density with ultra-large-scale integrated circuits with dimensional scalability, (2) high operation speed, (3) sufficient reliability including high switching endurance and low cycle-to-cycle variability, and (4) low device-to-device variability correlated to the spatial uniformity.

A high Vo concentration would be detrimental to achieving high information density with dimensional scaling. Such concentrations are frequently observed at interfacial regions near the electrodes or semiconductor channels. With decreasing film thickness for dimensional scaling, the increased fraction of the oxygen-deficient interfacial layers would result in an increased fraction of non-ferroelectric metastable phases, such as the tetragonal phase; the nonideal leakage current density through the ferroelectric HfO2 film would also increase. Moreover, a high concentration is not conducive to achieving high operation speeds, because it would decelerate the propagation of domain walls resulting from the attraction between defects. The frequently reported read-after-write latency [220, 221] is also strongly correlated to charge trapping/de-trapping, which is generally significantly slower than the polarization switching dominated by the domain wall propagation [222,223,224,225]. Furthermore, a high Vo concentration decreases the average grain size in oxide thin films. Therefore, the increased density of GBs is another cause of the decelerating switching speed of ferroelectric HfO2 thin films [31].

The main failure mechanism of the ferroelectric HfO2 is the hard breakdown that arises from the increased concentration and agglomeration of Vo with CFs forming in metal/ferroelectric/metal capacitors of FeRAM and FTJs or charge trapping and interfacial trap generation in the metal/ferroelectric/semiconductor gate stack of FeFETs. Therefore, neither mechanism should contain a high Vo concentration. The origin of the cycle-to-cycle variability with wake-up, fatigue, and soft or hard breakdown observed during repetitive polarization switching is also the change in Vo concentration and its drift, which results in its redistribution. Therefore, Vo concentration should be decreased to achieve reliable ferroelectric memories.

In addition to the Vo concentration, the spatial distribution of the concentration should also be controlled to decrease device-to-device variability. It should be also noted that numerous factors can affect the device-to-device variability of ferroelectric memory arrays, including polymorphism and film texture [226]. However, Vo distribution is a key factor that impacts this variability.

Therefore, the right strategy to achieve the optimized electrical properties of ferroelectric memories should be to decrease the Vo concentration by adequately controlling the other factors to suppress polymorphism, which is in turn achieved by suppressing the stable monoclinic phase formation. As described in the preceding paragraphs, the high Vo concentration has various detrimental effects on the material properties and the resulting device performance and reliability. However, the monoclinic phase fraction should be decreased by modulating factors other than the Vo concentration.

Meanwhile, it is necessary to consider the requirements of Vo concentration in RS phenomena from a different perspective. As previously reported, RS devices based on HfO2 employ a capacitor structure similar to that of FeRAM or FTJs, which possess a metal/RS layer/metal configuration. The equivalent of the failure mechanism in ferroelectric research is an electroforming process in RS research that triggers the subsequent RS behavior. Because the electroforming process does not contain an oxygen-deficient sub-phase, the RS behavior is determined solely by the generation, electromigration, and agglomeration of Vo. The Vo dynamics are influenced by intrinsic defects and those generated under voltage applications, which results in the formation of a conductive pathway known as the CF; this CF enables seamless RS. This switching mechanism is called a VCM because the presence of Vo in the crystal lattice of HfO2 leads to a change in the valence state. The migration of these oxygen ions or vacancies changes the local stoichiometry of the switching layer, inducing a valence change in the cation sublattice and modifying the electronic conductivity. In VCM devices, the switching behavior is associated with various phenomena related to a CF’s evolution, which occur through the formation, migration, agglomeration, and rearrangement of Vo. Once a CF has been formed, the movement of Vo becomes critical to the switching operation. Under an external electric field, Vo migrate within the switching layer and toward the positively biased electrode (anode) owing to electrostatic forces. This migration process leads to RESET (LRS to HRS) and SET (HRS to LRS) operations, which in turn results in the rupture and reformation of CFs.

Therefore, in addition to the concentration of Vo, factors such as the mobility of oxygen ions and structural irregularities within the switching material play a crucial role in facilitating the migration and agglomeration of Vo, affecting the RS performance. The concentration, migration, and agglomeration of intrinsic Vo defects are strongly influenced by crystallinity, stoichiometry, metal doping, interfacial layers, and electrode material. Particularly, in crystallized HfO2, GBs are considered active regions for RS or CF formation because they provide an environment where Vo can more easily form, migrate, and agglomerate than in bulk areas. The lower diffusion barrier for Vo at the GBs facilitates their movement and agglomeration. During the HfO2 deposition process, introducing certain metals as dopants can form oxygen-deficient regions owing to the availability of Vo and the ease of breaking metal-O bonds. The presence of Vo and the altered bonding contributes to the formation and stabilization of CFs. In addition, the selection of an appropriate interfacial layer or electrode can significantly influence the creation of an external reservoir for the Vo. This reservoir increases the reliability of the device during repeated switching events. By acting as a source or sink for the Vo, the interfacial layer or electrode helps maintain the stability and consistency of CF formation and rupture during the switching process. Therefore, understanding and controlling the GBs, stoichiometry, and appropriate interfaces/electrodes are important in the design and performance optimization of crystallized HfO2-based RS devices. Consequently, investigating those relationships has been a major focus of research. These factors are determined during the deposition and fabrication processes, as well as by post-processing and were detailed in this review. The critical parameters of RS devices—operating voltage, current, switching speed, gradual conductance modulation, endurance, device reliability, interdevice and intradevice variations, and long-term retention properties—can be optimized by precisely tuning these factors, which were also elaborately discussed.

Consequently, precise control over the Vo concentration in HfO2 can enhance both the ferroelectric properties and RS performances. As shown in Fig. 12, the concentration and migration of Vo may modify the properties of ferroelectric and RS devices, and it can be controlled by adjusting several key parameters. In the field of ferroelectric research, the main focus of studies thus far has been on the wake-up effect and fatigue behavior prior to soft or hard breakdown. Meanwhile, in RS research, the critical concern has been on comprehending the switching performance after a soft breakdown (CF formation). Due to these divergent objectives, controlling Vo by a unified method is not possible or necessary. Nevertheless, it is worthwhile to study the influence of factors on the generation, migration, and agglomeration of Vo during the deposition of HfO2 and fabrication of the device, and the effect on the structural, chemical, and, thus, electrical properties under voltage application. This insight can provide valuable guidance for future endeavors aimed at achieving desirable behaviors in both ferroelectric and RS properties.

Fig. 12
figure 12

Comprehensive schematic of the ferroelectric and RS HfO2 memory device and the effect of Vo for their performances. a Ferroelectric switching mechanism based on the movement of oxygen atoms (up) and Vo-mediated RS mechanism (down). b Effects of Vo and their roles in ferroelectric and resistive memory performance. c Key parameters affecting the concentration and motion of Vo in the film

Availability of data and materials

The review is based on the published data and sources of data upon which conclusions have been drawn can be found in the reference list.

Abbreviations

RS:

Resistive switching

Vo :

Oxygen vacancy

MOSFET:

Metal oxide semiconductor field effect transistor

ALD:

Atomic layer deposition

CSD:

Chemical solution deposition

CMOS:

Complementary metal-oxide semiconductor

HZO:

Hf0.5Zr0.5O2

PLD:

Pulsed laser deposition

CF:

Conductive filament

DRAM:

Dynamic random-access memory

TEM:

Transmission electron microscope

3D:

Three-dimensional

ToF–SIMS:

Time-of-flight secondary ion mass spectroscopy

O2 * :

O2 plasma

HAABF:

High-angle annular bright-field

TALD:

Thermal atomic layer deposition

RPALD:

Remote plasma atomic layer deposition

DPALD:

Direct plasma atomic layer deposition

XPS:

X-ray photoelectron spectroscopy

DFT:

Density functional theory

VASP:

Vienna ab initio simulation package

RF:

Radio frequency

RTA:

Rapid thermal annealing

ΔHf per O:

Enthalpy per oxygen of 1 mol

RRAM:

Resistive random-access memory

HRS:

High resistance state

LRS:

Low resistance state

HX-PES:

Hard X-ray photoelectron spectroscopy

HAADF-STEM:

High-angle annular dark-field scanning transmission electron microscopy

iDPC-STEM:

Integrated differential phase contrast scanning transmission electron microscopy

LSMO:

La0.67Sr0.33MnO3

FFT:

Fast Fourier transform

EELS:

Electron energy loss spectroscopy

EDS:

Energy dispersive X-ray spectroscopy

HR-TEM:

High-resolution transmission electron microscope

EFM:

Electrostatic force microscopy

CB:

Conduction band

c-AFM:

Conductive atomic force microscopy

2D:

Two-dimensional

TE:

Top electrode

BE:

Bottom electrode

STXM:

Scanning transmission x-ray microscopy

GB:

Grain boundary

NEB:

Nudged elastic band

PF:

Poole–Frenkel

TAT:

Trap-assisted tunneling

FN:

Fowler–Nordheim

GIXRD:

Grazing incidence X-ray diffraction

2Pr :

Double remanent polarization

Ec :

Coercive field

FORC:

First-order reversal curve

Ebias :

Internal bias

VOx :

Vanadium oxide

VCM:

Valence change mechanism

ECM:

Electrochemical metallization

MBE:

Molecular beam epitaxy

RMBE:

Reactive molecular beam epitaxy

PVD:

Physical vapor deposition

FeRAM:

Ferroelectric random-access memory

FTJ:

Ferroelectric tunnel junction

FeFET:

Ferroelectric field effect transistor

References

  1. X. Liu, S. Ramanathan, A. Longdergan, A. Srivastava, E. Lee, T.E. Seidel, J.T. Barton, D. Pang, R.G. Gordon, ALD of hafnium oxide thin films from tetrakis (ethylmethylamino) hafnium and ozone. J. Electrochem. Soc. 152(3), G213 (2005)

    Article  CAS  Google Scholar 

  2. K. Kukli, T. Pilvi, M. Ritala, T. Sajavaara, J. Lu, M. Leskelä, Atomic layer deposition of hafnium dioxide thin films from hafnium tetrakis (dimethylamide) and water. Thin Solid Films 491(1–2), 328–338 (2005)

    Article  CAS  Google Scholar 

  3. K. Kukli, M. Ritala, T. Sajavaara, J. Keinonen, M. Leskelä, Atomic layer deposition of hafnium dioxide films from hafnium tetrakis (ethylmethylamide) and water. Chem. Vap. Depos. 8(5), 199–204 (2002)

    Article  CAS  Google Scholar 

  4. H.-B. Kim, M. Jung, Y. Oh, S.W. Lee, D. Suh, J.-H. Ahn, Superior and stable ferroelectric properties of hafnium-zirconium-oxide thin films deposited via atomic layer deposition using cyclopentadienyl-based precursors without annealing. Nanoscale 13(18), 8524–8530 (2021)

    Article  CAS  Google Scholar 

  5. D. Shin, R. Arróyave, Z.-K. Liu, Thermodynamic modeling of the Hf–Si–O system. Calphad 30(4), 375–386 (2006)

    Article  CAS  Google Scholar 

  6. R. Materlik, C. Künneth, A. Kersch, The origin of ferroelectricity in Hf1− xZrxO2: a computational investigation and a surface energy model. J. Appl. Phys. 117(13), 134109 (2015)

    Article  Google Scholar 

  7. C. Kittel, Theory of antiferroelectric crystals. Phys. Rev. 82(5), 729 (1951)

    Article  CAS  Google Scholar 

  8. M. Lederer, R. Olivo, D. Lehninger, S. Abdulazhanov, T. Kämpfe, S. Kirbach, C. Mart, K. Seidel, L.M. Eng, On the origin of wake-up and antiferroelectric-like behavior in ferroelectric hafnium oxide. Phys. Status Solidi-Rapid Res. Lett. 15(5), 2100086 (2021)

    Article  CAS  Google Scholar 

  9. Y. Cheng, Z. Gao, K.H. Ye, H.W. Park, Y. Zheng, Y. Zheng, J. Gao, M.H. Park, J.-H. Choi, K.-H. Xue, C.S. Hwang, H. Lyu, Reversible transition between the polar and antipolar phases and its implications for wake-up and fatigue in HfO2-based ferroelectric thin film. Nat. Commun. 13(1), 645 (2022)

    Article  CAS  Google Scholar 

  10. T.D. Huan, V. Sharma, G.A. Rossetti Jr., R. Ramprasad, Pathways towards ferroelectricity in hafnia. Phys. Rev. B. 90(6), 064111 (2014)

    Article  CAS  Google Scholar 

  11. Y. Wei, P. Nukala, M. Salverda, S. Matzen, H.J. Zhao, J. Momand, A.S. Everhardt, G. Agnus, G.R. Blake, P. Lecoeur, B.J. Kooi, J. Íñiguez, B. Dkhill, B. Noheda, A rhombohedral ferroelectric phase in epitaxially strained Hf0.5Zr0.5O2 thin films. Nat. Mater. 17(12), 1095–1100 (2018)

    Article  CAS  Google Scholar 

  12. S.U. Sharath, S. Vogel, L. Molina-Luna, E. Hildebrandt, C. Wenger, J. Kurian, M. Duerrschnabel, T. Niermann, G. Niu, P. Calka, M. Lehmann, H.-J. Kleebe, T. Schroeder, L. Alff, Control of switching modes and conductance quantization in oxygen engineered HfOx based memristive devices. Adv. Funct. Mater. 27(32), 1700432 (2017)

    Article  Google Scholar 

  13. T.-Y. Wang, J.-L. Meng, M.-Y. Rao, Z.-Y. He, L. Chen, H. Zhu, Q.-Q. Sun, S.-J. Ding, W.-Z. Bao, P. Zhou, D.W. Zhang, Three-dimensional nanoscale flexible memristor networks with ultralow power for information transmission and processing application. Nano Lett. 20(6), 4111–4120 (2020)

    Article  CAS  Google Scholar 

  14. T. Wang, J. Meng, X. Zhou, Y. Liu, Z. He, Q. Han, Q. Li, J. Yu, Z. Li, Y. Liu, H. Zhu, Q. Sun, D.W. Zhang, P. Chen, H. Peng, L. Chen, Reconfigurable neuromorphic memristor network for ultralow-power smart textile electronics. Nat. Commun. 13(1), 7432 (2022)

    Article  CAS  Google Scholar 

  15. H. Abbas, Y. Abbas, G. Hassan, A.S. Sokolov, Y.-R. Jeon, B. Ku, C.J. Kang, C. Choi, The coexistence of threshold and memory switching characteristics of ALD HfO2 memristor synaptic arrays for energy-efficient neuromorphic computing. Nanoscale 12(26), 14120–14134 (2020)

    Article  CAS  Google Scholar 

  16. W. Banerjee, X. Zhang, Q. Luo, H. Lv, Q. Liu, S. Long, M. Liu, Design of CMOS compatible, high-speed, highly-stable complementary switching with multilevel operation in 3D vertically stacked novel HfO2/Al2O3/TiOx (HAT) RRAM. Adv. Electron. Mater. 4(2), 1700561 (2018)

    Article  Google Scholar 

  17. S. Brivio, S. Spiga, D. Ielmini, HfO2-based resistive switching memorydevices for neuromorphic computing. Neuromorphic Comput Eng (2022). https://doi.org/10.1088/2634-4386/ac9012

    Article  Google Scholar 

  18. Y. Park, M.-K. Kim, J.-S. Lee, Emerging memory devices for artificial synapses. J. Mater. Chem. C. 8(27), 9163–9183 (2020)

    Article  CAS  Google Scholar 

  19. Z. Peng, F. Wu, L. Jiang, G. Cao, B. Jiang, G. Cheng, S. Ke, K.C. Chang, L. Li, C. Ye, HfO2-based memristor as an artificial synapse for neuromorphic computing with Tri-Layer HfO2/BiFeO3/HfO2 design. Adv. Funct. Mater. 31(48), 2107131 (2021)

    Article  CAS  Google Scholar 

  20. V.A. Gritsenko, T.V. Perevalov, D.R. Islamov, Electronic properties of hafnium oxide: a contribution from defects and traps. Phys. Rep. 613, 1–20 (2016)

    Article  CAS  Google Scholar 

  21. S. Dirkmann, J. Kaiser, C. Wenger, T. Mussenbrock, Filament growth and resistive switching in hafnium oxide memristive devices. ACS Appl. Mater. Interfaces 10(17), 14857–14868 (2018)

    Article  CAS  Google Scholar 

  22. L.-Y. Ma, S. Liu, Structural polymorphism kinetics promoted by charged oxygen vacancies in HfO2. Phys. Rev. Lett. 130(9), 096801 (2023)

    Article  CAS  Google Scholar 

  23. Y. Zhou, Y. Zhang, Q. Yang, J. Jiang, P. Fan, M. Liao, Y. Zhou, The effects of oxygen vacancies on ferroelectric phase transition of HfO2-based thin film from first-principle. Comput. Mater. Sci. 167, 143–150 (2019)

    Article  CAS  Google Scholar 

  24. S. Starschich, D. Griesche, T. Schneller, U. Böttger, Chemical solution deposition of ferroelectric hafnium oxide for future lead free ferroelectric devices. ECS J. Soild State Sc. 4(12), P419 (2015)

    CAS  Google Scholar 

  25. Z. Wang, T. Kumagai, H. Kokawa, M. Ichiki, R. Maeda, Preparation of hafnium oxide thin films by sol–gel method. J. Electroceram. 21, 499–502 (2008)

    Article  CAS  Google Scholar 

  26. S. Starschich, D. Griesche, T. Schneller, R. Waser, U. Böttger, Chemical solution deposition of ferroelectric yttrium-doped hafnium oxide films on platinum electrodes. Appl. Phys. Lett. 104, 20 (2014)

    Article  Google Scholar 

  27. Z. Li, D. Zhou, J. Wang, N. Sun, W. Zhang, Ferroelectricity in hafnium oxide films doped with magnesium by chemical solution deposition. J. Appl. Phys. 131, 7 (2022)

    Article  Google Scholar 

  28. A. Callegari, E. Cartier, M. Gribelyuk, H. Okorn-Schmidt, T. Zabel, Physical and electrical characterization of Hafnium oxide and Hafnium silicate sputtered films. J. Appl. Phys. 90(12), 6466–6475 (2001)

    Article  CAS  Google Scholar 

  29. T. Mittmann, M. Materano, P.D. Lomenzo, M.H. Park, I. Stolichnov, M. Cavalieri, C. Zhou, C.C. Chung, J.L. Jones, T. Szyjka, M. Müller, A. Kersch, T. Mikolajick, U. Schroeder, Origin of ferroelectric phase in undoped HfO2 films deposited by sputtering. Adv. Mater. Interfaces. 6(11), 1900042 (2019)

    Article  Google Scholar 

  30. S.T. Jaszewski, E.R. Hoglund, A. Costine, M.H. Weber, S.S. Fields, M.G. Sales, J. Vaidya, L. Bellcase, K. Loughlin, A. Salanova, D.A. Dickie, S.L. Wolfley, M.D. Henty, J.-P. Maria, J.L. Jones, N. Shukla, S.J. McDonnell, P. Reinke, P.E. Hopkins, J.M. Howe, J.F. Ihlefeld, Impact of oxygen content on phase constitution and ferroelectric behavior of hafnium oxide thin films deposited by reactive high-power impulse magnetron sputtering. Acta Mater. 239, 118220 (2022)

    Article  CAS  Google Scholar 

  31. Y.H. Lee, H.J. Kim, T. Moon, K. Do Kim, S.D. Hyun, H.W. Park, Y.B. Lee, M.H. Park, C.S. Hwang, Preparation and characterization of ferroelectric Hf0.5Zr0.5O2 thin films grown by reactive sputtering. Nanotechnology 28(30), 305703 (2017)

    Article  Google Scholar 

  32. M. Filipescu, N. Scarisoreanu, V. Craciun, B. Mitu, A. Purice, A. Moldovan, V. Ion, O. Toma, M. Dinescu, High-k dielectric oxides obtained by PLD as solution for gates dielectric in MOS devices. Appl. Surf. Sci. 253(19), 8184–8191 (2007)

    Article  CAS  Google Scholar 

  33. S. Starschich, T. Schenk, U. Schroeder, U. Boettger, Ferroelectric and piezoelectric properties of Hf1-xZrxO2 and pure ZrO2 films. Appl. Phys. Lett. 110, 18 (2017)

    Article  Google Scholar 

  34. Y. Yao, D. Zhou, S. Li, J. Wang, N. Sun, F. Liu, X. Zhao, Experimental evidence of ferroelectricity in calcium doped hafnium oxide thin films. J. Appl. Phys. 126, 15 (2019)

    Article  Google Scholar 

  35. S. Starschich, S. Menzel, U. Böttger, Evidence for oxygen vacancies movement during wake-up in ferroelectric hafnium oxide. Appl. Phys. Lett. 108(3), 032903 (2016)

    Article  Google Scholar 

  36. T. Song, R. Solanas, M. Qian, I. Fina, F. Sánchez, Large enhancement of ferroelectric polarization in Hf0.5Zr0.5O2 films by low plasma energy pulsed laser deposition. J. Mater. Chem. C. 10(3), 1084–1089 (2022)

    Article  CAS  Google Scholar 

  37. T. Mittmann, M. Materano, S.-C. Chang, I. Karpov, T. Mikolajick and U. Schroeder, Impact of oxygen vacancy content in ferroelectric HZO films on the device performance. In proceeding of the 2020 IEEE International Electron Devices Meeting (IEDM), 2020.

  38. M. Sung, K. Rho, J. Kim, J. Cheon, K. Choi, D. Kim, H. Em, G. Park, J. Woo, Y. Lee, J. Ko, M. Kim, G. Lee, S. w. Ryu, D. S. Sheen, Y. Joo, S. Kim, C. H. Cho, M.-H. Na and J. Kim, "Low Voltage and High Speed 1Xnm 1T1C FE-RAM with Ultra-Thin 5nm HZO," in proceeding of the 2021 IEEE International Electron Devices Meeting (IEDM), 2021

  39. H.A. Hsain, Y. Lee, S. Lancaster, M. Materano, R. Alcala, B. Xu, T. Mikolajick, U. Schroeder, G.N. Parsons, J.L. Jones, Role of oxygen source on buried interfaces in atomic-layer-deposited ferroelectric hafnia-zirconia thin films. ACS Appl. Mater. Interfaces 14(37), 42232–42244 (2022)

    Article  CAS  Google Scholar 

  40. M. Martínez-Puente, P. Horley, F. Aguirre-Tostado, J. López-Medina, H. Borbón-Nuñez, H. Tiznado, A. Susarrey-Arce, E. Martínez-Guerra, ALD and PEALD deposition of HfO2 and its effects on the nature of oxygen vacancies. Mater. Sci. Eng. B 285, 115964 (2022)

    Article  Google Scholar 

  41. D.H. Lee, G.T. Yu, J.Y. Park, S.H. Kim, K. Yang, G.H. Park, J.J. Ryu, J.I. Lee, G.H. Kim, M.H. Park, Effect of residual impurities on polarization switching kinetics in atomic-layer-deposited ferroelectric Hf0.5Zr0.5O2 thin films. Acta Mater. 222, 117405 (2022)

    Article  CAS  Google Scholar 

  42. Y. Choi, C. Han, J. Shin, S. Moon, J. Min, H. Park, D. Eom, J. Lee, C. Shin, Impact of chamber/annealing temperature on the endurance characteristic of Zr: HfO2 ferroelectric capacitor. Sensors 22(11), 4087 (2022)

    Article  CAS  Google Scholar 

  43. Y. Senzaki, S. Park, H. Chatham, L. Bartholomew, W. Nieveen, Atomic layer deposition of hafnium oxide and hafnium silicate thin films using liquid precursors and ozone. J. Vac. Sci. Technol., A: Vac., Surf. Films 22(4), 1175–1181 (2004)

    Article  CAS  Google Scholar 

  44. S. Chen, Z. Liu, L. Feng, X. Che, Influence of cerium-doping on the structural and electrical properties of hafnium oxide gate dielectric. J. Mater. Sci.: Mater. Electron. 25, 749–753 (2014)

    Google Scholar 

  45. T. Böscke, J. Müller, D. Bräuhaus, U. Schröder, U. Böttger, Ferroelectricity in hafnium oxide thin films. Appl. Phys. Lett. 99(10), 102903 (2011)

    Article  Google Scholar 

  46. S. Mueller, J. Mueller, A. Singh, S. Riedel, J. Sundqvist, U. Schroeder, T. Mikolajick, Incipient ferroelectricity in Al-doped HfO2 thin films. Adv. Funct. Mater. 22(11), 2412–2417 (2012)

    Article  CAS  Google Scholar 

  47. M. Hyuk Park, H. Joon Kim, Y. Jin Kim, W. Lee, T. Moon, C.S. Hwang, Evolution of phases and ferroelectric properties of thin Hf0.5Zr0.5O2 films according to the thickness and annealing temperature. Appl. Phys. Lett. 102(24), 242905 (2013)

    Article  Google Scholar 

  48. T. Schenk, S. Mueller, U. Schroeder, R. Materlik, A. Kersch, M. Popovici, C. Adelmann, S. Van Elshocht and T. Mikolajick, Strontium doped hafnium oxide thin films: Wide process window for ferroelectric memories. In proceeding of the 2013 Proceedings of the European Solid-State Device Research Conference (ESSDERC), 2013

  49. U. Schroeder, C. Richter, M.H. Park, T. Schenk, M. Pesic, M. Hoffmann, F.P. Fengler, D. Pohl, B. Rellinghaus, C. Zhou, Lanthanum-doped hafnium oxide: a robust ferroelectric material, lnorg. Chem. 57(5), 2752–2765 (2018)

    CAS  Google Scholar 

  50. M. Hoffmann, U. Schroeder, T. Schenk, T. Shimizu, H. Funakubo, O. Sakata, D. Pohl, M. Drescher, C. Adelmann, R. Materlik, A. Kersch, T. Mikolajick, Stabilizing the ferroelectric phase in doped hafnium oxide. J. Appl. Phys. 118(7), 072006 (2015)

    Article  Google Scholar 

  51. J. Müller, U. Schröder, T. Böscke, I. Müller, U. Böttger, L. Wilde, J. Sundqvist, M. Lemberger, P. Kücher, T. Mikolajick, L. Frey, Ferroelectricity in yttrium-doped hafnium oxide. J. Appl. Phys. 110(11), 114113 (2011)

    Article  Google Scholar 

  52. M.H. Park, Y.H. Lee, H.J. Kim, Y.J. Kim, T. Moon, K.D. Kim, J. Mueller, A. Kersch, U. Schroeder, T. Mikolajick, C.S. Hwang, Ferroelectricity and antiferroelectricity of doped thin HfO2-based films. Adv. Mater. 27(11), 1811–1831 (2015)

    Article  CAS  Google Scholar 

  53. S. Starschich, U. Boettger, An extensive study of the influence of dopants on the ferroelectric properties of HfO2. J. Mater. Chem. C. 5(2), 333–338 (2017)

    Article  CAS  Google Scholar 

  54. R. Batra, T.D. Huan, G.A. Rossetti Jr., R. Ramprasad, Dopants promoting ferroelectricity in hafnia: insights from a comprehensive chemical space exploration. Chem. Mater. 29(21), 9102–9109 (2017)

    Article  CAS  Google Scholar 

  55. U. Schroeder, E. Yurchuk, J. Müller, D. Martin, T. Schenk, P. Polakowski, C. Adelmann, M.I. Popovici, S.V. Kalinin, T. Mikolajick, Impact of different dopants on the switching properties of ferroelectric hafnium oxide. Jpn. J. Appl. Phys. 53(8S1), 08LE02 (2014)

    Article  CAS  Google Scholar 

  56. H. Zhang, B. Gao, S. Yu, L. Lai, L. Zeng, B. Sun, L. Liu, X. Liu, J. Lu and R. Han, Effects of ionic doping on the behaviors of oxygen vacancies in HfO2 and ZrO2: a first principles study. In proceeding of the 2009 International Conference on Simulation of Semiconductor Processes and Devices, 2009.

  57. Y. Zhao, J. Wang, J. Xu, F. Yang, Q. Liu, Y. Dai, Metal dopants in HfO2-based RRAM: first principle study. J. Semicond. 35(4), 042002 (2014)

    Article  CAS  Google Scholar 

  58. Y. Lee, Y. Goh, J. Hwang, D. Das, S. Jeon, The influence of top and bottom metal electrodes on ferroelectricity of hafnia. IEEE Trans. Electron Devices 68(2), 523–528 (2021)

    Article  CAS  Google Scholar 

  59. Y. Lee, H.W. Jeong, S.H. Kim, K. Yang, M.H. Park, Effect of stress on fluorite-structured ferroelectric thin films for semiconductor devices. Mater. Sci. Semicond. Process. 160, 107411 (2023)

    Article  CAS  Google Scholar 

  60. P.D. Lomenzo, Q. Takmeel, C. Zhou, C.M. Fancher, E. Lambers, N.G. Rudawski, J.L. Jones, S. Moghaddam, T. Nishida, TaN interface properties and electric field cycling effects on ferroelectric Si-doped HfO2 thin films. J. Appl. Phys. 117, 13 (2015)

    Article  Google Scholar 

  61. Y. Li, R. Liang, B. Xiong, H. Liu, R. Zhao, J. Li, T. Liu, Y. Pang, H. Tian, Y. Yang, T.-L. Ren, TiNx/Hf0.5Zr0.5O2/TiNx ferroelectric memory with tunable transparency and suppressed wake-up effect. Appl. Phys. Lett. 114, 5 (2019)

    Google Scholar 

  62. S.S. Fields, S.W. Smith, C.M. Fancher, M.D. Henry, S.L. Wolfley, M.G. Sales, S.T. Jaszewski, M.A. Rodriguez, G. Esteves, P.S. Davids, S.J. McDonnell, J.F. Ihlefeld, Metal nitride electrode stress and chemistry effects on phase and polarization response in ferroelectric Hf0.5Zr0.5O2 Thin Films. Adv. Mater. Interfaces 8(10), 2100018 (2021)

    Article  CAS  Google Scholar 

  63. K. Yang, S.H. Kim, H.W. Jeong, D.H. Lee, G.H. Park, Y. Lee, M.H. Park, Perspective on ferroelectric devices: lessons from interfacial chemistry. Chem. Mater. 35(6), 2219–2237 (2023)

    Article  CAS  Google Scholar 

  64. W. Hamouda, A. Pancotti, C. Lubin, L. Tortech, C. Richter, T. Mikolajick, U. Schroeder, N. Barrett, Physical chemistry of the TiN/Hf0.5Zr0.5O2 interface. J. Appl. Phys. 127(6), 064105 (2020)

    Article  CAS  Google Scholar 

  65. T. Mittmann, T. Szyjka, H. Alex, M.C. Istrate, P.D. Lomenzo, L. Baumgarten, M. Müller, J.L. Jones, L. Pintilie, T. Mikolajick, U. Schroeder, Impact of iridium oxide electrodes on the ferroelectric phase of thin Hf0.5Zr0.5O2 films. Phys. Status Solidi-Rapid Res. Lett. 15(5), 2100012 (2021)

    Article  CAS  Google Scholar 

  66. Y. Lee, S.H. Kim, H.W. Jeong, G.H. Park, J. Lee, Y.Y. Kim, M.H. Park, Mitigation of field-driven dynamic phase evolution in ferroelectric Hf0.5Zr0.5O2 films by adopting oxygen-supplying electrode. Appl. Surf. Sci. 648, 158948 (2023)

  67. K. Yang, G.-Y. Kim, J.J. Ryu, D.H. Lee, J.Y. Park, S.H. Kim, G.H. Park, G.T. Yu, G.H. Kim, S.Y. Choi, M.H. Park, Wake-up-mitigated giant ferroelectricity in Hf0.5Zr0.5O2 thin films through oxygen-providing, surface-oxidized W electrode. Mater. Sci. Semicond. Process. 164, 107565 (2023)

    Article  CAS  Google Scholar 

  68. Y. Cheng, M. Zheng, X. Zhang, H. Dong, Y. Jiang, J. Wu, J. Qi, Z. Yin, Possible extrinsic ferroelectric-like signals originated from the oxygen vacancy drift in HfO2-based films. ACS Appl. Electron. Mater. (2023). https://doi.org/10.1021/acsaelm.3c00181

    Article  Google Scholar 

  69. T. Nagata, M. Haemori, Y. Yamashita, H. Yoshikawa, Y. Iwashita, K. Kobayashi, T. Chikyow, Oxygen migration at Pt/HfO2/Pt interface under bias operation. Appl. Phys. Lett. 97(8), 082902 (2010)

    Article  Google Scholar 

  70. S. de Graaf, J. Momand, C. Mitterbauer, S. Lazar, B.J. Kooi, Resolving hydrogen atoms at metal-metal hydride interfaces. Sci. Adv. 6(5), eaay4312 (2020)

    Article  Google Scholar 

  71. P. Nukala, M. Ahmadi, Y. Wei, S. De Graaf, E. Stylianidis, T. Chakrabortty, S. Matzen, H.W. Zandbergen, A. Björling, D. Mannix, D. Carbone, B. Kooi, B. Noheda, Reversible oxygen migration and phase transitions in hafnia-based ferroelectric devices. Science 372(6542), 630–635 (2021)

    Article  CAS  Google Scholar 

  72. S. Privitera, G. Bersuker, B. Butcher, A. Kalantarian, S. Lombardo, C. Bongiorno, R. Geer, D. Gilmer, P. Kirsch, Microscopy study of the conductive filament in HfO2 resistive switching memory devices. Microelectron. Eng. 109, 75–78 (2013)

    Article  CAS  Google Scholar 

  73. P. Calka, E. Martinez, V. Delaye, D. Lafond, G. Audoit, D. Mariolle, N. Chevalier, H. Grampeix, C. Cagli, V. Jousseaume, C. Guedj, Chemical and structural properties of conducting nanofilaments in TiN/HfO2-based resistive switching structures. Nanotechnology 24(8), 085706 (2013)

    Article  CAS  Google Scholar 

  74. J. Yin, F. Zeng, Q. Wan, F. Li, Y. Sun, Y. Hu, J. Liu, G. Li, F. Pan, Adaptive crystallite kinetics in homogenous bilayer oxide memristor for emulating diverse synaptic plasticity. Adv. Funct. Mater. 28(19), 1706927 (2018)

    Article  Google Scholar 

  75. Y. Yang, X. Zhang, L. Qin, Q. Zeng, X. Qiu, R. Huang, Probing nanoscale oxygen ion motion in memristive systems. Nat. Commun. 8(1), 15173 (2017)

    Article  Google Scholar 

  76. J. Gavartin, D. Muñoz Ramo, A. Shluger, G. Bersuker, B. Lee, Negative oxygen vacancies in HfO2 as charge traps in high-k stacks. Appl. Phys. Lett. 89, 8 (2006)

    Article  Google Scholar 

  77. D.Z. Gao, J. Strand, M.S. Munde, A.L. Shluger, Mechanisms of oxygen vacancy aggregation in SiO2 and HfO2. Front. Phys. 7, 43 (2019)

    Article  Google Scholar 

  78. S.R. Bradley, A.L. Shluger, G. Bersuker, Electron-injection-assisted generation of oxygen vacancies in monoclinic HfO2. Phys Rev Appl. 4(6), 064008 (2015)

    Article  Google Scholar 

  79. Y. Zhang, G.-Q. Mao, X. Zhao, Y. Li, M. Zhang, Z. Wu, W. Wu, H. Sun, Y. Guo, L. Wang, X. Zhang, Q. Liu, H. Lv, K.-H. Xue, G. Xu, X. Miao, S. Long, M. Liu, Evolution of the conductive filament system in HfO2-based memristors observed by direct atomic-scale imaging. Nat. Commun. 12(1), 7232 (2021)

    Article  CAS  Google Scholar 

  80. T. Wei, Y. Lu, F. Zhang, J. Tang, B. Gao, P. Yu, H. Qian, H. Wu, Three-dimensional reconstruction of conductive filaments in hfox-based memristor. Adv. Mater. 35(10), 2209925 (2023)

    Article  CAS  Google Scholar 

  81. U. Celano, L. Goux, R. Degraeve, A. Fantini, O. Richard, H. Bender, M. Jurczak, W. Vandervorst, Imaging the three-dimensional conductive channel in filamentary-based oxide resistive switching memory. Nano Lett. 15(12), 7970–7975 (2015)

    Article  CAS  Google Scholar 

  82. S. Kumar, Z. Wang, X. Huang, N. Kumari, N. Davila, J.P. Strachan, D. Vine, A.D. Kilcoyne, Y. Nishi, R.S. Williams, Conduction channel formation and dissolution due to oxygen thermophoresis/diffusion in hafnium oxide memristors. ACS Nano 10(12), 11205–11210 (2016)

    Article  CAS  Google Scholar 

  83. K. McKenna, A. Shluger, The interaction of oxygen vacancies with grain boundaries in monoclinic HfO2. Appl. Phys. Lett. 95(22), 222111 (2009)

    Article  Google Scholar 

  84. G. Bersuker, D. Gilmer, D. Veksler, P. Kirsch, L. Vandelli, A. Padovani, L. Larcher, K. McKenna, A. Shluger, V. Iglesias, M. Porti, M. Nafría, Metal oxide resistive memory switching mechanism based on conductive filament properties. J. Appl. Phys. 110(12), 124518 (2011)

    Article  Google Scholar 

  85. K.-H. Xue, P. Blaise, L. Fonseca, G. Molas, E. Vianello, B. Traoré, B. De Salvo, G. Ghibaudo, Y. Nishi, Grain boundary composition and conduction in HfO2: an ab initio study. Appl. Phys. Lett. 102(20), 201908 (2013)

    Article  Google Scholar 

  86. W. Mullins, The effect of thermal grooving on grain boundary motion. Acta Metall. 6(6), 414–427 (1958)

    Article  Google Scholar 

  87. W.W. Mullins, Theory of thermal grooving. J. Appl. Phys. 28(3), 333–339 (1957)

    Article  CAS  Google Scholar 

  88. X. Cheng, C. Zhou, B. Lin, Z. Yang, S. Chen, K.H. Zhang, Z. Chen, Leakage mechanism in ferroelectric Hf0.5Zr0.5O2 epitaxial thin films. Appl. Mater. Today 32, 101804 (2023)

    Article  Google Scholar 

  89. D.R. Islamov, A.G. Chernikova, M.G. Kozodaev, T.V. Perevalov, V.A. Gritsenko, O.M. Orlov, A.M. Markeev, Leakage currents mechanism in thin films of ferroelectric Hf0.5Zr0.5O2. ECS Trans. 75(32), 123 (2017)

    Article  CAS  Google Scholar 

  90. D.R. Islamov, V.A. Gritsenko, T.V. Perevalov, V.A. Pustovarov, O.M. Orlov, A.G. Chernikova, A.M. Markeev, S. Slesazeck, U. Schroeder, T. Mikolajick, G.Y. Krasnikov, Identification of the nature of traps involved in the field cycling of Hf0.5Zr0.5O2-based ferroelectric thin films. Acta Mater. 166, 47–55 (2019)

    Article  CAS  Google Scholar 

  91. R.M. Hill, Poole-Frenkel conduction in amorphous solids. Philos. Mag. 23(181), 59–86 (1971)

    Article  CAS  Google Scholar 

  92. I. Lundström, C. Svensson, Tunneling to traps in insulators. J. Appl. Phys. 43(12), 5045–5047 (1972)

    Article  Google Scholar 

  93. F. Gunkel, D.V. Christensen, Y. Chen, N. Pryds, Oxygen vacancies: the (in) visible friend of oxide electronics. Appl. Phys. Lett. 116(12), 120505 (2020)

    Article  CAS  Google Scholar 

  94. R. He, H. Wu, S. Liu, H. Liu, Z. Zhong, Ferroelectric structural transition in hafnium oxide induced by charged oxygen vacancies. Phys. Rev. B. 104(18), L180102 (2021)

    Article  CAS  Google Scholar 

  95. E.D. Grimley, T. Schenk, X. Sang, M. Pešić, U. Schroeder, T. Mikolajick, J.M. LeBeau, Structural changes underlying field-cycling phenomena in ferroelectric HfO2 thin films. Adv. Electron. Mater. 2(9), 1600173 (2016)

    Article  Google Scholar 

  96. A.K. Tagantsev, I. Stolichnov, E. Colla, N. Setter, Polarization fatigue in ferroelectric films: basic experimental findings, phenomenological scenarios, and microscopic features. J. Appl. Phys. 90(3), 1387–1402 (2001)

    Article  CAS  Google Scholar 

  97. J.C. Garcia, L.M.R. Scolfaro, A.T. Lino, V. Freire, G. Farias, C. Silva, H.L. Alves, S. Rodrigues, E. da Silva Jr, Structural, electronic, and optical properties of ZrO2 from ab initio calculations. J. Appl. Phys. 100(10), 104103 (2006)

    Article  Google Scholar 

  98. X. Sang, E.D. Grimley, T. Schenk, U. Schroeder, J.M. LeBeau, On the structural origins of ferroelectricity in HfO2 thin films. Appl. Phys. Lett. 106(16), 162905 (2015)

    Article  Google Scholar 

  99. D. Islamov, T. Perevalov, Effect of oxygen vacancies on the ferroelectric Hf0.5Zr0.5O2 stabilization: DFT simulation. Microelectron. Eng. 216, 111041 (2019)

    Article  CAS  Google Scholar 

  100. M. Pešić, F.P.G. Fengler, L. Larcher, A. Padovani, T. Schenk, E.D. Grimley, X. Sang, J.M. LeBeau, S. Slesazeck, U. Schroeder, T. Mikolajick, Physical mechanisms behind the field-cycling behavior of HfO2-based ferroelectric capacitors. Adv. Funct. Mater. 26(25), 4601–4612 (2016)

    Article  Google Scholar 

  101. A. Kashir, S. Oh, H. Hwang, Defect engineering to achieve wake-up free HfO2-based ferroelectrics. Adv. Eng. Mater. 23(1), 2000791 (2021)

    Article  CAS  Google Scholar 

  102. D.M. Ramo, A. Shluger, J. Gavartin, G. Bersuker, Theoretical prediction of intrinsic self-trapping of electrons and holes in monoclinic HfO2. Phys. Rev. Lett. 99(15), 155504 (2007)

    Article  Google Scholar 

  103. D.M. Ramo, J. Gavartin, A. Shluger, G. Bersuker, Spectroscopic properties of oxygen vacancies in monoclinic HfO2 calculated with periodic and embedded cluster density functional theory. Phys. Rev. B. 75(20), 205336 (2007)

    Article  Google Scholar 

  104. N. Capron, P. Broqvist, A. Pasquarello, Migration of oxygen vacancy in HfO2 and across the HfO2SiO2 interface: a first-principles investigation. Appl. Phys. Lett. 91(19), 192905 (2007)

    Article  Google Scholar 

  105. C. Tang, B. Tuttle, R. Ramprasad, Diffusion of O vacancies near Si:HfO2 interfaces: an ab initio investigation. Phys. Rev. B. 76(7), 073306 (2007)

    Article  Google Scholar 

  106. Y. Dai, Z. Pan, F. Wang, X. Li, Oxygen vacancy effects in HfO2-based resistive switching memory: first principle study. AIP Adv. 6(8), 085209 (2016)

    Article  Google Scholar 

  107. S. Zafar, H. Jagannathan, L.F. Edge, D. Gupta, Measurement of oxygen diffusion in nanometer scale HfO2 gate dielectric films. Appl. Phys. Lett. 98(15), 152903 (2011)

    Article  Google Scholar 

  108. A.S. Foster, V.B. Sulimov, F. Lopez Gejo, A.L. Shluger, R.M. Nieminen, Structure and electrical levels of point defects in monoclinic zirconia. Phys. Rev. B. 64, 22 (2001)

    Article  Google Scholar 

  109. J. Yang, M. Youssef, B. Yildiz, Oxygen self-diffusion mechanisms in monoclinic Zr O 2 revealed and quantified by density functional theory, random walk analysis, and kinetic Monte Carlo calculations. Phys. Rev. B. 97(2), 024114 (2018)

    Article  Google Scholar 

  110. A. Kyrtsos, M. Matsubara, E. Bellotti, Migration mechanisms and diffusion barriers of vacancies in Ga 2 O 3. Phys. Rev. B. 95(24), 245202 (2017)

    Article  Google Scholar 

  111. M.Y. Yang, K. Kamiya, B. Magyari-Köpe, M. Niwa, Y. Nishi, K. Shiraishi, Charge-dependent oxygen vacancy diffusion in Al2O3-based resistive-random-access-memories. Appl. Phys. Lett. 103(9), 093504 (2013)

    Article  Google Scholar 

  112. U. Schroeder, M.H. Park, T. Mikolajick, C.S. Hwang, The fundamentals and applications of ferroelectric HfO2. Nat. Rev. Mater. 7(8), 653–669 (2022)

    Article  Google Scholar 

  113. A. Chernikova, D. Kuzmichev, D. Negrov, M. Kozodaev, S. Polyakov, A. Markeev, Ferroelectric properties of full plasma-enhanced ALD TiN/La: HfO2/TiN stacks. Appl. Phys. Lett. 108(24), 242905 (2016)

    Article  Google Scholar 

  114. L. Yao, X. Liu, Y. Cheng, B. Xiao, A synergistic interplay between dopant ALD cycles and film thickness on the improvement of the ferroelectricity of uncapped Al: HfO2 nanofilms. Nanotechnology 32(21), 215708 (2021)

    Article  CAS  Google Scholar 

  115. M.H. Park, C.C. Chung, T. Schenk, C. Richter, M. Hoffmann, S. Wirth, J.L. Jones, T. Mikolajick, U. Schroeder, Origin of temperature-dependent ferroelectricity in Si-doped HfO2. Adv. Electron. Mater. 4(4), 1700489 (2018)

    Article  Google Scholar 

  116. M.H. Park, T. Schenk, C.M. Fancher, E.D. Grimley, C. Zhou, C. Richter, J.M. LeBeau, J.L. Jones, T. Mikolajick, U. Schroeder, A comprehensive study on the structural evolution of HfO2 thin films doped with various dopants. J. Mater. Chem. C. 5(19), 4677–4690 (2017)

    Article  CAS  Google Scholar 

  117. R. Alcala, C. Richter, M. Materano, P.D. Lomenzo, C. Zhou, J.L. Jones, T. Mikolajick, U. Schroeder, Influence of oxygen source on the ferroelectric properties of ALD grown Hf1-xZrxO2 films. J. Phys. D: Appl. Phys. 54(3), 035102 (2020)

    Article  Google Scholar 

  118. S.J. Kim, J. Mohan, H.S. Kim, J. Lee, S.M. Hwang, D. Narayan, J.-G. Lee, C.D. Young, L. Colombo, G. Goodman, A.S. Wan, P.-R. Cha, S.R. Summerfelt, T.S. San, J. Kim, Effect of hydrogen derived from oxygen source on low-temperature ferroelectric TiN/Hf0.5Zr0.5O2/TiN capacitors. Appl. Phys. Lett. 115(18), 182901 (2019)

    Article  Google Scholar 

  119. A. Pal, V.K. Narasimhan, S. Weeks, K. Littau, D. Pramanik, T. Chiang, Enhancing ferroelectricity in dopant-free hafnium oxide. Appl. Phys. Lett. 110, 2 (2017)

    Article  Google Scholar 

  120. P. Jiang, Q. Luo, X. Xu, T. Gong, P. Yuan, Y. Wang, Z. Gao, W. Wei, L. Tai, H. Lv, Wake-up effect in HfO2-based ferroelectric films. Adv. Electron. Mater. 7(1), 2000728 (2021)

    Article  CAS  Google Scholar 

  121. T.Y. Lee, K. Lee, H.H. Lim, M.S. Song, S.M. Yang, H.K. Yoo, D.I. Suh, Z. Zhu, A. Yoon, M.R. MacDonald, X. Lei, H.Y. Jeong, D. Lee, K. Park, J. Park, S.C. Chae, Ferroelectric polarization-switching dynamics and wake-up effect in Si-doped HfO2. ACS Appl. Mater. Interfaces 11(3), 3142–3149 (2018)

    Article  Google Scholar 

  122. M.H. Park, Y.H. Lee, T. Mikolajick, U. Schroeder, C.S. Hwang, Review and perspective on ferroelectric HfO2-based thin films for memory applications. MRS Commun. 8(3), 795–808 (2018)

    Article  CAS  Google Scholar 

  123. H.J. Kim, M.H. Park, Y.J. Kim, Y.H. Lee, T. Moon, K. Do Kim, S.D. Hyun, C.S. Hwang, A study on the wake-up effect of ferroelectric Hf0.5Zr0.5O2 films by pulse-switching measurement. Nanoscale 8(3), 1383–1389 (2016)

    Article  CAS  Google Scholar 

  124. M.H. Park, H.J. Kim, Y.J. Kim, Y.H. Lee, T. Moon, K.D. Kim, S.D. Hyun, F. Fengler, U. Schroeder, C.S. Hwang, Effect of Zr content on the wake-up effect in Hf1–xZrxO2 films. ACS Appl. Mater. Interfaces 8(24), 15466–15475 (2016)

    Article  CAS  Google Scholar 

  125. D. Zhou, J. Xu, Q. Li, Y. Guan, F. Cao, X. Dong, J. Müller, T. Schenk, U. Schröder, Wake-up effects in Si-doped hafnium oxide ferroelectric thin films. Appl. Phys. Lett. 103(19), 192904 (2013)

    Article  Google Scholar 

  126. H.-Y. Huang, M. Wu, L.-J. Qiao, The behaviour of 180° polarization switching in BaTiO3 from first principles calculations. Comput. Mater. Sci. 82, 1–4 (2014)

    Article  Google Scholar 

  127. X. Lu, H. Li, W. Cao, Landau expansion parameters for BaTiO3. J. Appl. Phys. 114(22), 224106 (2013)

    Article  Google Scholar 

  128. Z. Fan, J. Chen, J. Wang, Ferroelectric HfO2-based materials for next-generation ferroelectric memories. J. Adv. Dielectr. 6(02), 1630003 (2016)

    Article  CAS  Google Scholar 

  129. F.P. Fengler, R. Nigon, P. Muralt, E.D. Grimley, X. Sang, V. Sessi, R. Hentschel, J.M. LeBeau, T. Mikolajick, U. Schroeder, Analysis of performance instabilities of Hafnia-based ferroelectrics using modulus spectroscopy and thermally stimulated depolarization currents. Adv. Electron. Mater. 4(3), 1700547 (2018)

    Article  Google Scholar 

  130. T. Schenk, M. Hoffmann, J. Ocker, M. Pesic, T. Mikolajick, U. Schroeder, Complex internal bias fields in ferroelectric hafnium oxide. ACS Appl. Mater. Interfaces 7(36), 20224–20233 (2015)

    Article  CAS  Google Scholar 

  131. D.R. Islamov, T.M. Zalyalov, O.M. Orlov, V.A. Gritsenko, G.Y. Krasnikov, Impact of oxygen vacancy on the ferroelectric properties of lanthanum-doped hafnium oxide. Appl. Phys. Lett. 117(16), 162901 (2020)

    Article  CAS  Google Scholar 

  132. V. Gritsenko, A. Gismatulin, Charge transport mechanism in La: HfO2. Appl. Phys. Lett. 117(14), 142901 (2020)

    Article  CAS  Google Scholar 

  133. Y. Goh, S.H. Cho, S.-H.K. Park, S. Jeon, Oxygen vacancy control as a strategy to achieve highly reliable hafnia ferroelectrics using oxide electrode. Nanoscale 12(16), 9024–9031 (2020)

    Article  CAS  Google Scholar 

  134. R. Zhao, T. Liu, X. Zhao, H. Liu, M. Shao, Q. Feng, X. Sun, X. Wu, Y. Yang, T.-L. Ren, Impact of molybdenum oxide electrode on the ferroelectricity of doped-hafnia oxide capacitors. IEEE Trans. Electron Devices 69(3), 1492–1496 (2022)

    Article  CAS  Google Scholar 

  135. M. Yadav, A. Kashir, S. Oh, R.D. Nikam, H. Kim, H. Jang, H. Hwang, High polarization and wake-up free ferroelectric characteristics in ultrathin Hf0.5Zr0.5O2 devices by control of oxygen-deficient layer. Nanotechnology 33(8), 085206 (2021)

    Article  Google Scholar 

  136. S.H. Kim, G.T. Yu, G.H. Park, D.H. Lee, J.Y. Park, K. Yang, E.B. Lee, J.I. Lee, M.H. Park, Interfacial engineering of a Mo/Hf0.3Zr0.7O2/Si capacitor using the direct scavenging effect of a thin Ti layer. Chem. Commun. 57(93), 12452–12455 (2021)

    Article  CAS  Google Scholar 

  137. K.-Y. Chen, P.-H. Chen, R.-W. Kao, Y.-X. Lin, Y.-H. Wu, Impact of plasma treatment on reliability performance for HfZrOx-based metal-ferroelectric-metal capacitors. IEEE Electron Device Lett. 39(1), 87–90 (2017)

    Article  CAS  Google Scholar 

  138. Y. Zhang, Z. Fan, D. Wang, J. Wang, Z. Zou, Y. Li, Q. Li, R. Tao, D. Chen, M. Zeng, X. Gao, D. Jiyan, G. Zhou, X. Lu, J.-M. Liu, Enhanced ferroelectric properties and insulator-metal transition-induced shift of polarization-voltage hysteresis loop in VOx-Capped Hf0.5Zr0.5O2 thin films. ACS Appl. Mater. Interfaces. 12(36), 40510–40517 (2020)

    Article  CAS  Google Scholar 

  139. J.M. Simoes, J. Beauchamp, Transition metal-hydrogen and metal-carbon bond strengths: the keys to catalysis. Chem. Rev. 90(4), 629–688 (1990)

    Article  CAS  Google Scholar 

  140. A. Kornilov, I. Ushakova, E. Huber Jr., C. Holley Jr., The enthalpy of formation of hafnium dioxide. J. Chem. Thermodyn. 7(1), 21–26 (1975)

    Article  CAS  Google Scholar 

  141. I. Barin, G. Platzki, Thermochemical data of pure substances (no 334) (Wiley Online Library, Hoboken, 1989)

    Google Scholar 

  142. E. Cordfunke, R. Konings, The enthalpy of formation of RuO2. Thermochim. Acta 129(1), 63–69 (1988)

    Article  CAS  Google Scholar 

  143. E. Cordfunke, R. Konings, Thermochemical data for reactor materials and fission products: the ECN database. J. Ph. Equilibria 14, 457–464 (1993)

    Article  CAS  Google Scholar 

  144. M.W. Chase, N.I.S. Organization, (NIST-JANAF thermochemical tables (American Chemical Society, Washington, DC, 1998)

    Google Scholar 

  145. M. Su, S. Gao, Z. Weng, L. Zhao, C. Lee, Y. Zhao, Improvement of ferroelectricity and reliability in Hf0.5Zr0.5O2 thin films with two-step oxygen vacancy engineering. IEEE Electron Device Lett. 43(7), 1057–1060 (2022)

    Article  CAS  Google Scholar 

  146. Y. Chen, H. Lee, P. Chen, C. Tsai, P. Gu, T. Wu, K. Tsai, S. Sheu, W. Lin, C. Lin, P. Chiu, W. Chen, F. Chen, C. Lien and M. Tsai, Challenges and opportunities for HfOX based resistive random access memory. In proceeding of the 2011 International Electron Devices Meeting, 2011

  147. H. Wu, X.H. Wang, B. Gao, N. Deng, Z. Lu, B. Haukness, G. Bronner, H. Qian, Resistive random access memory for future information processing system. Proc. IEEE 105(9), 1770–1789 (2017)

    Article  CAS  Google Scholar 

  148. C. Li, B. Gao, Y. Yao, X. Guan, X. Shen, Y. Wang, P. Huang, L. Liu, X. Liu, J. Li, C. Gu, J.K. Kang, R. Yu, Direct observations of nanofilament evolution in switching processes in HfO2-based resistive random access memory by in situ TEM studies. Adv. Mater. 29(10), 1602976 (2017)

    Article  Google Scholar 

  149. A. Padovani, L. Larcher, O. Pirrotta, L. Vandelli, G. Bersuker, Microscopic modeling of HfOx RRAM operations: from forming to switching. IEEE Trans. Electron Devices 62(6), 1998–2006 (2015)

    Article  CAS  Google Scholar 

  150. C. Chen, A. Fantini, L. Goux, R. Degraeve, S. Clima, A. Redolfi, G. Groeseneken and M. Jurczak, Programming-conditions solutions towards suppression of retention tails of scaled oxide-based RRAM. In proceeding of the 2015 IEEE International Electron Devices Meeting (IEDM), 2015

  151. Q. Wang, G. Niu, S. Roy, Y. Wang, Y. Zhang, H. Wu, S. Zhai, W. Bai, P. Shi, S. Song, Z. Song, Y.-H. Xie, Z.-G. Ye, C. Wenger, X. Meng, W. Ren, Interface-engineered reliable HfO2-based RRAM for synaptic simulation. J. Mater. Chem. C. 7(40), 12682–12687 (2019)

    Article  CAS  Google Scholar 

  152. F. Xu, B. Gao, Y. Xi, J. Tang, H. Wu and H. Qian, Atomic-device hybrid modeling of relaxation effect in analog RRAM for neuromorphic computing. In proceeding of the 2020 IEEE International Electron Devices Meeting (IEDM), 2020

  153. X. Xu, L. Tai, T. Gong, J. Yin, P. Huang, J. Yu, D. N. Dong, Q. Luo, J. Liu, Z. Yu, X. Zhu, X. L. Wu, Q. Liu, H. Lv and M. Liu, 40× retention improvement by eliminating resistance relaxation with high temperature forming in 28 nm RRAM chip. In proceeding of the 2018 IEEE International Electron Devices Meeting (IEDM), 2018

  154. K.-H. Xue, Y. Li, H.-L. Su, J.-H. Yuan, Y. Li, Z.-R. Wang, B. Zhang, X.-S. Miao, Theoretical investigation of the Ag filament morphology in conductive bridge random access memories. J. Appl. Phys. 124(15), 152125 (2018)

    Article  Google Scholar 

  155. G.-Q. Mao, K.-H. Xue, Y.-Q. Song, W. Wu, J.-H. Yuan, L.-H. Li, H. Sun, S. Long, X.-S. Miao, Oxygen migration around the filament region in HfOx memristors. AIP Adv. 9(10), 105007 (2019)

    Article  Google Scholar 

  156. D.-H. Kwon, K.M. Kim, J.H. Jang, J.M. Jeon, M.H. Lee, G.H. Kim, X.-S. Li, G.-S. Park, B. Lee, S. Han, M. Kim, C.S. Hwang, Atomic structure of conducting nanofilaments in TiO2 resistive switching memory. Nat. Nanotechnol. 5(2), 148–153 (2010)

    Article  CAS  Google Scholar 

  157. J.-Y. Chen, C.-L. Hsin, C.-W. Huang, C.-H. Chiu, Y.-T. Huang, S.-J. Lin, W.-W. Wu, L.-J. Chen, Dynamic evolution of conducting nanofilament in resistive switching memories. Nano Lett. 13(8), 3671–3677 (2013)

    Article  CAS  Google Scholar 

  158. B. Choi, D.S. Jeong, S. Kim, C. Rohde, S. Choi, J. Oh, H. Kim, C. Hwang, K. Szot, R. Waser, B. Reichenberg, S. Tiedke, Resistive switching mechanism of TiO2 thin films grown by atomic-layer deposition. J. Appl. Phys. 98(3), 033715 (2005)

    Article  Google Scholar 

  159. C.-F. Kang, W.-C. Kuo, W. Bao, C.-H. Ho, C.-W. Huang, W.-W. Wu, Y.-H. Chu, J.-Y. Juang, S.H. Tseng, L. Hu, J.-H. He, Self-formed conductive nanofilaments in (Bi, Mn)Ox for ultralow-power memory devices. Nano Energy 13, 283–290 (2015)

    Article  CAS  Google Scholar 

  160. S. Long, L. Perniola, C. Cagli, J. Buckley, X. Lian, E. Miranda, F. Pan, M. Liu, J. Suñé, Voltage and power-controlled regimes in the progressive unipolar RESET transition of HfO2-based RRAM. Sci. Rep. 3(1), 2929 (2013)

    Article  Google Scholar 

  161. J.Y. Chen, C.W. Huang, C.H. Chiu, Y.T. Huang, W.W. Wu, Switching kinetic of VCM-based memristor: evolution and positioning of nanofilament. Adv. Mater. 27(34), 5028–5033 (2015)

    Article  CAS  Google Scholar 

  162. H. Meng, S. Huang, Y. Jiang, The role of oxygen vacancies on resistive switching properties of oxide materials. Inform. Technol. 1, 2 (2020)

    Google Scholar 

  163. K. Zhang, J. Wang, Y. Huang, L.-Q. Chen, P. Ganesh, Y. Cao, High-throughput phase-field simulations and machine learning of resistive switching in resistive random-access memory. npj Comput. Mater. 6(1), 198 (2020)

    Article  CAS  Google Scholar 

  164. B. Moirangthem, P.N. Meitei, A.K. Debnath, N.K. Singh, Forming-free RRAM device based on HfO2 thin film for non-volatile memory application using E-beam evaporation method. J. Mater. Sci. Mater. Electron. 34(4), 306 (2023)

    Article  CAS  Google Scholar 

  165. Y.-S. Chen, T.-Y. Wu, P.-J. Tzeng, P.-S. Chen, H.-Y. Lee, C.-H. Lin, F. Chen and M.-J. Tsai, Forming-free HfO2 bipolar RRAM device with improved endurance and high speed operation. In proceeding of the 2009 International Symposium on VLSI Technology, Systems, and Applications, 2009

  166. T. Wang, Y. Shi, F.M. Puglisi, S. Chen, K. Zhu, Y. Zuo, X. Li, X. Jing, T. Han, B. Guo, K. Bukvišová, L. Kachtík, M. Kolíbal, C. Wen, M. Lanza, Electroforming in metal-oxide memristive synapses. ACS Appl. Mater. Interfaces 12(10), 11806–11814 (2020)

    Article  CAS  Google Scholar 

  167. T. Li, H. Yu, S.H.Y. Chen, Y. Zhou, S.-T. Han, The strategies of filament control for improving the resistive switching performance. J. Mater. Chem. C. 8(46), 16295–16317 (2020)

    Article  CAS  Google Scholar 

  168. B. Chen, J.F. Kang, B. Gao, Y.X. Deng, L.F. Liu, X.Y. Liu, Z. Fang, H.Y. Yu, X.P. Wang, G.Q. Lo, D.L. Kwong, Endurance degradation in metal oxide-based resistive memory induced by oxygen ion loss effect. IEEE Electron Device Lett. 34(10), 1292–1294 (2013)

    Article  CAS  Google Scholar 

  169. B. Chen, Y. Lu, B. Gao, Y. Fu, F. Zhang, P. Huang, Y. Chen, L. Liu, X. Liu, J. Kang, Y. Wang, Z. Fang, H. Yu, X. Li, X. Wang, N. Singh, G. Lo and D. Kwong, Physical mechanisms of endurance degradation in TMO-RRAM. In proceeding of the 2011 International electron devices meeting, 2011

  170. J. Xu, H. Wang, Y. Zhu, Y. Liu, Z. Zou, G. Li, R. Xiong, Tunable digital-to-analog switching in Nb2O5-based resistance switching devices by oxygen vacancy engineering. Appl. Surf. Sci. 579, 152114 (2022)

    Article  CAS  Google Scholar 

  171. W.S. Choi, M.S. Song, H. Kim, D.H. Kim, Conduction mechanism analysis of abrupt-and gradual-switching InGaZnO memristors. Micromachines 13(11), 1870 (2022)

    Article  Google Scholar 

  172. M.-J. Lee, C.B. Lee, D. Lee, S.R. Lee, M. Chang, J.H. Hur, Y.-B. Kim, C.-J. Kim, D.H. Seo, S. Seo, U.-I. Chung, I.-K. Yoo, K. Kim, A fast, high-endurance and scalable non-volatile memory device made from asymmetric Ta2O5−x/TaO2−x bilayer structures. Nat. Mater. 10(8), 625–630 (2011)

    Article  CAS  Google Scholar 

  173. S.-M. Park, H.-G. Hwang, J.-U. Woo, W.-H. Lee, S.-J. Chae, S. Nahm, Improvement of conductance modulation linearity in a Cu2+-doped KNbO3 memristor through the increase of the number of oxygen vacancies. ACS Appl. Mater. Interfaces 12(1), 1069–1077 (2019)

    Article  Google Scholar 

  174. Y. Pyo, J.-U. Woo, H.-G. Hwang, S. Nahm, J. Jeong, Effect of oxygen vacancy on the conduction modulation linearity and classification accuracy of Pr0.7Ca0.3MnO3 memristor. Nanomaterials 11(10), 2684 (2021)

    Article  CAS  Google Scholar 

  175. Z. Wang, M. Yin, T. Zhang, Y. Cai, Y. Wang, Y. Yang, R. Huang, Engineering incremental resistive switching in TaOx based memristors for brain-inspired computing. Nanoscale 8(29), 14015–14022 (2016)

    Article  CAS  Google Scholar 

  176. Y. Lu, B. Chen, B. Gao, Z. Fang, Y. Fu, J. Yang, L. Liu, X. Liu, H. Yu and J. Kang, Improvement of endurance degradation for oxide based resistive switching memory devices correlated with oxygen vacancy accumulation effect. In proceeding of the 2012 IEEE International Reliability Physics Symposium (IRPS), 2012

  177. B. Traoré, P. Blaise, E. Vianello, H. Grampeix, A. Bonnevialle, E. Jalaguier, G. Molas, S. Jeannot, L. Perniola, B. DeSalvo and Y. Nishi, Microscopic understanding of the low resistance state retention in HfO2 and HfAlO based RRAM. In proceeding of the 2014 IEEE International Electron Devices Meeting, 2014.

  178. K. McKenna, A. Shluger, V. Iglesias, M. Porti, M. Nafría, M. Lanza, G. Bersuker, Grain boundary mediated leakage current in polycrystalline HfO2 films. Microelectron. Eng. 88(7), 1272–1275 (2011)

    Article  CAS  Google Scholar 

  179. K. Pey, R. Ranjan, C. Tung, L. Tang, V. Lo, K. Lim, D. Ang, Breakdowns in high-k gate stacks of nano-scale CMOS devices. Microelectron. Eng. 80, 353–361 (2005)

    Article  CAS  Google Scholar 

  180. V. Yanev, M. Rommel, M. Lemberger, S. Petersen, B. Amon, T. Erlbacher, A.J. Bauer, H. Ryssel, A. Paskaleva, W. Weinreich, C. Fachmann, J. Heitmann, U. Schroeder, Tunneling atomic-force microscopy as a highly sensitive mapping tool for the characterization of film morphology in thin high-k dielectrics. Appl. Phys. Lett. 92(25), 252910 (2008)

    Article  Google Scholar 

  181. V. Iglesias, M. Porti, M. Nafría, X. Aymerich, P. Dudek, G. Bersuker, Dielectric breakdown in polycrystalline hafnium oxide gate dielectrics investigated by conductive atomic force microscopy. J Vac. Sci. Technol. B 29(1), 01AB02 (2011)

    Article  Google Scholar 

  182. L. Vandelli, A. Padovani, L. Larcher, G. Bersuker, Microscopic modeling of electrical stress-induced breakdown in poly-crystalline hafnium oxide dielectrics. IEEE Trans. Electron Devices 60(5), 1754–1762 (2013)

    Article  CAS  Google Scholar 

  183. M. Lanza, A review on resistive switching in high-k dielectrics: a nanoscale point of view using conductive atomic force microscope. Materials 7(3), 2155–2182 (2014)

    Article  Google Scholar 

  184. M. Lanza, G. Bersuker, M. Porti, E. Miranda, M. Nafría, X. Aymerich, Resistive switching in hafnium dioxide layers: Local phenomenon at grain boundaries. Appl. Phys. Lett. 101(19), 193502 (2012)

    Article  Google Scholar 

  185. M. Lanza, K. Zhang, M. Porti, M. Nafría, Z. Shen, L. Liu, J. Kang, D. Gilmer, G. Bersuker, Grain boundaries as preferential sites for resistive switching in the HfO2 resistive random access memory structures. Appl. Phys. Lett. 100(12), 123508 (2012)

    Article  Google Scholar 

  186. G. Bersuker, J. Yum, L. Vandelli, A. Padovani, L. Larcher, V. Iglesias, M. Porti, M. Nafría, K. McKenna, A. Shluger, P. Kirsch, R. Jammy, Grain boundary-driven leakage path formation in HfO2 dielectrics. Solid State Electron 65, 146–150 (2011)

    Article  Google Scholar 

  187. S. Petzold, A. Zintler, R. Eilhardt, E. Piros, N. Kaiser, S.U. Sharath, T. Vogel, M. Major, K.P. McKenna, L. Molina-Luna, L. Alff, Forming-free grain boundary engineered hafnium oxide resistive random access memory devices. Adv. Electron. Mater. 5(10), 1900484 (2019)

    Article  CAS  Google Scholar 

  188. N. Raghavan, A. Fantini, R. Degraeve, P. Roussel, L. Goux, B. Govoreanu, D. Wouters, G. Groeseneken, M. Jurczak, Statistical insight into controlled forming and forming free stacks for HfOx RRAM. Microelectron. Eng. 109, 177–181 (2013)

    Article  CAS  Google Scholar 

  189. R. Degraeve, A. Fantini, N. Raghavan, L. Goux, S. Clima, B. Govoreanu, A. Belmonte, D. Linten, M. Jurczak, Causes and consequences of the stochastic aspect of filamentary RRAM. Microelectron. Eng. 147, 171–175 (2015)

    Article  CAS  Google Scholar 

  190. H.-Y. Lee, P.-S. Chen, C.-C. Wang, S. Maikap, P.-J. Tzeng, C.-H. Lin, L.-S. Lee, M.-J. Tsai, Low-power switching of nonvolatile resistive memory using hafnium oxide. Jpn. J. Appl. Phys. 46(4S), 2175 (2007)

    Article  CAS  Google Scholar 

  191. J.J. Yang, M.D. Pickett, X. Li, D.A. Ohlberg, D.R. Stewart, R.S. Williams, Memristive switching mechanism for metal/oxide/metal nanodevices. Nat. Nanotechnol. 3(7), 429–433 (2008)

    Article  CAS  Google Scholar 

  192. N. Kaiser, T. Vogel, A. Zintler, S. Petzold, A. Arzumanov, E. Piros, R. Eilhardt, L. Molina-Luna, L. Alff, Defect-stabilized substoichiometric polymorphs of hafnium oxide with semiconducting properties. ACS Appl. Mater. Interfaces 14(1), 1290–1303 (2021)

    Article  Google Scholar 

  193. K.P. McKenna, Optimal stoichiometry for nucleation and growth of conductive filaments in HfOx. Modell. Simul. Mater. Sci. Eng. 22(2), 025001 (2014)

    Article  CAS  Google Scholar 

  194. S. Sharath, T. Bertaud, J. Kurian, E. Hildebrandt, C. Walczyk, P. Calka, P. Zaumseil, M. Sowinska, D. Walczyk, A. Gloskovskii, T. Schroeder, L. Alff, Towards forming-free resistive switching in oxygen engineered HfO2−x. Appl. Phys. Lett. 104(6), 063502 (2014)

    Article  Google Scholar 

  195. L.-H. Li, K.-H. Xue, L.-Q. Zou, J.-H. Yuan, H. Sun, X. Miao, Multilevel switching in Mg-doped HfOx memristor through the mutual-ion effect. Appl. Phys. Lett. 119(15), 153505 (2021)

    Article  CAS  Google Scholar 

  196. D. Duncan, B. Magyari-Köpe, Y. Nishi, Properties of dopants in HfOx for improving the performance of nonvolatile memory. Phys. Rev. Appl. 7(3), 034020 (2017)

    Article  Google Scholar 

  197. K. Lee, Y. Kim, H. Na, H. Sohn, Effect of aliovalent impurities on the resistance switching characteristics of sputtered hafnium oxide films. J. Vac. Sci. Technol. B 33, 3 (2015)

    Article  Google Scholar 

  198. S. Roy, G. Niu, Q. Wang, Y. Wang, Y. Zhang, H. Wu, S. Zhai, P. Shi, S. Song, Z. Song, Z.-G. Ye, C. Wenger, T. Schroeder, Y.-H. Xie, X. Meng, W. Luo, W. Ren, Toward a reliable synaptic simulation using Al-doped HfO2 RRAM. ACS Appl. Mater. Interfaces 12(9), 10648–10656 (2020)

    Article  CAS  Google Scholar 

  199. Q. Wu, W. Banerjee, J. Cao, Z. Ji, L. Li, M. Liu, Improvement of durability and switching speed by incorporating nanocrystals in the HfOx based resistive random access memory devices. Appl. Phys. Lett. 113(2), 023105 (2018)

    Article  Google Scholar 

  200. W. Banerjee, S. Maikap, S. Rahaman, A. Prakash, T.-C. Tien, W.-C. Li, J.-R. Yang, Improved resistive switching memory characteristics using core-shell IrOx nano-dots in Al2O3/WOx bilayer structure. J. Electrochem. Soc. 159(2), H177 (2011)

    Article  Google Scholar 

  201. W. Banerjee, S. Maikap, C.-S. Lai, Y.-Y. Chen, T.-C. Tien, H.-Y. Lee, W.-S. Chen, F.T. Chen, M.-J. Kao, M.-J. Tsai, J.-R. Yang, Formation polarity dependent improved resistive switching memory characteristics using nanoscale (1.3 nm) core-shell IrOx nano-dots. Nanoscale Res. Lett. 7(1), 1–12 (2012)

    Article  Google Scholar 

  202. F. Hui, E. Grustan-Gutierrez, S. Long, Q. Liu, A.K. Ott, A.C. Ferrari, M. Lanza, Graphene and related materials for resistive random access memories. Adv. Electron. Mater. 3(8), 1600195 (2017)

    Article  Google Scholar 

  203. H. Tian, H.-Y. Chen, T.-L. Ren, C. Li, Q.-T. Xue, M.A. Mohammad, C. Wu, Y. Yang, H.-S.P. Wong, Cost-effective, transfer-free, flexible resistive random access memory using laser-scribed reduced graphene oxide patterning technology. Nano Lett. 14(6), 3214–3219 (2014)

    Article  CAS  Google Scholar 

  204. H. Tian, H.-Y. Chen, B. Gao, S. Yu, J. Liang, Y. Yang, D. Xie, J. Kang, T.-L. Ren, Y. Zhang, H.-S.P. Wong, Monitoring oxygen movement by Raman spectroscopy of resistive random access memory with a graphene-inserted electrode. Nano Lett. 13(2), 651–657 (2013)

    Article  CAS  Google Scholar 

  205. S. Bertolazzi, P. Bondavalli, S. Roche, T. San, S.Y. Choi, L. Colombo, F. Bonaccorso, P. Samorì, Nonvolatile memories based on graphene and related 2D materials. Adv. Mater. 31(10), 1806663 (2019)

    Article  Google Scholar 

  206. L. Zhu, X. Zhang, J. Zhou, Z. Sun, Interfacial graphene modulated energetic behavior of the point-defect at the Au/HfO2 interface. Appl. Surf. Sci. 489, 608–613 (2019)

    Article  CAS  Google Scholar 

  207. C. Mannequin, A. Delamoreanu, L. Latu-Romain, V. Jousseaume, H. Grampeix, S. David, C. Rabot, A. Zenasni, C. Vallee, P. Gonon, Graphene-HfO2-based resistive RAM memories. Microelectron. Eng. 161, 82–86 (2016)

    Article  CAS  Google Scholar 

  208. C. Walczyk, C. Wenger, D. Walczyk, M. Lukosius, I. Costina, M. Fraschke, J. Dabrowski, A. Fox, D. Wolansky, S. Thiess, E. Miranda, B. Tillack, T. Schroeder, On the role of Ti adlayers for resistive switching in HfO 2-based metal-insulator-metal structures: top versus bottom electrode integration. J. Vac. Sci. Technol. B 29(1), 01AD02 (2011)

    Article  Google Scholar 

  209. P. Calka, M. Sowinska, T. Bertaud, D. Walczyk, J. Dabrowski, P. Zaumseil, C. Walczyk, A. Gloskovskii, X. Cartoixà, J. Suñé, T. Schroeder, Engineering of the chemical reactivity of the Ti/HfO2 interface for RRAM: experiment and theory. ACS Appl. Mater. Interfaces 6(7), 5056–5060 (2014)

    Article  CAS  Google Scholar 

  210. H. Lee, P. Chen, T. Wu, Y. Chen, C. Wang, P. Tzeng, C. Lin, F. Chen, C. Lien and M.-J. Tsai, Low power and high speed bipolar switching with a thin reactive Ti buffer layer in robust HfO2 based RRAM. In proceeding of the 2008 IEEE International Electron Devices Meeting, 2008

  211. C. Walczyk, T. Schroeder, M. Lukosius, M. Fraschke, A. Fox, D. Wolansky, B. Tillack and C. Wenger, Resistive switching in TiN/HfO2/Ti/TiN MIM devices for future nonvolatile memory applications. In proceeding of the 2009 10th Annual Non-Volatile Memory Technology Symposium (NVMTS), 2009

  212. A.S. Sokolov, Y.-R. Jeon, S. Kim, B. Ku, D. Lim, H. Han, M.G. Chae, J. Lee, B.G. Ha, C. Choi, Influence of oxygen vacancies in ALD HfO2-x thin films on non-volatile resistive switching phenomena with a Ti/HfO2-x/Pt structure. Appl. Surf. Sci. 434, 822–830 (2018)

    Article  CAS  Google Scholar 

  213. D. Lee, Y. Sung, I. Lee, J. Kim, H. Sohn, D.-H. Ko, Enhanced bipolar resistive switching of HfO2 with a Ti interlayer. Appl. Phys. A 102, 997–1001 (2011)

    Article  CAS  Google Scholar 

  214. A. Padovani, L. Larcher, P. Padovani, C. Cagli and B. De Salvo, Understanding the role of the Ti metal electrode on the forming of HfO2-based RRAMs. In proceeding of the 2012 4th IEEE International Memory Workshop, 2012

  215. Z. Yong, K.-M. Persson, M.S. Ram, G. D’Acunto, Y. Liu, S. Benter, J. Pan, Z. Li, M. Borg, A. Mikkelsen, L.-E. Wernersson, R. Timm, Tuning oxygen vacancies and resistive switching properties in ultra-thin HfO2 RRAM via TiN bottom electrode and interface engineering. Appl. Surf. Sci. 551, 149386 (2021)

    Article  CAS  Google Scholar 

  216. D. Berco, T.-Y. Tseng, A numerical analysis of progressive and abrupt reset in conductive bridging RRAM. J. Comput. Electron. 15, 586–594 (2016)

    Article  Google Scholar 

  217. J.U. Kwon, Y.G. Song, J.E. Kim, S.Y. Chun, G.H. Kim, G. Noh, J.Y. Kwak, S. Hur, C.-Y. Kang, D.S. Jeong, S.J. Oh, J.H. Yoon, Surface-dominated HfO2 nanorod-based memristor exhibiting highly linear and symmetrical conductance modulation for high-precision neuromorphic computing. ACS Appl. Mater. Interfaces 14(39), 44550–44560 (2022)

    Article  CAS  Google Scholar 

  218. G.H. Park, D.H. Lee, H. Choi, T. Kwon, Y.H. Cho, S.H. Kim, M.H. Park, Emerging fluorite-structured antiferroelectrics and their semiconductor applications. ACS Appl. Electron. Mater. 5(2), 642–663 (2023)

    Article  CAS  Google Scholar 

  219. D.H. Lee, G.H. Park, S.H. Kim, J.Y. Park, K. Yang, S. Slesazeck, T. Mikolajick, M.H. Park, Neuromorphic devices based on fluorite-structured ferroelectrics. InfoMat. 4(12), e12380 (2022)

    Article  CAS  Google Scholar 

  220. M. Hoffmann, A.J. Tan, N. Shanker, Y.-H. Liao, L.-C. Wang, J.-H. Bae, C. Hu, S. Salahuddin, Write disturb-free ferroelectric FETs with non-accumulative switching dynamics. IEEE Electron Device Lett. 43(12), 2097–2100 (2022)

    Article  CAS  Google Scholar 

  221. K. Ni, P. Sharma, J. Zhang, M. Jerry, J.A. Smith, K. Tapily, R. Clark, S. Mahapatra, S. Datta, Critical role of interlayer in Hf0.5Zr0.5O2 ferroelectric FET nonvolatile memory performance. IEEE Trans. Electron Devices 65(6), 2461–2469 (2018)

    Article  CAS  Google Scholar 

  222. D. Kleimaier, H. Mulaosmanovic, S. Dünkel, S. Beyer, S. Soss, S. Slesazeck, T. Mikolajick, Demonstration of a p-type ferroelectric FET with immediate read-after-write capability. IEEE Electron Device Lett. 42(12), 1774–1777 (2021)

    Article  CAS  Google Scholar 

  223. H. Mulaosmanovic, E. T. Breyer, T. Mikolajick and S. Slesazeck, Switching and charge trapping in HfO2-based ferroelectric FETs: An overview and potential applications. In proceeding of the 2020 4th IEEE Electron Devices Technology & Manufacturing Conference (EDTM), 2020

  224. S. Deng, Z. Zhao, Y.S. Kim, S. Duenkel, D. MacMahon, R. Tiwari, N. Choudhury, S. Beyer, X. Gong, S. Kurinec, K. Ni, Unraveling the dynamics of charge trapping and de-trapping in ferroelectric FETs. IEEE Trans. Electron Devices 69(3), 1503–1511 (2022)

    Article  CAS  Google Scholar 

  225. B.H. Kim, S.H. Kuk, S.K. Kim, J.P. Kim, Y.J. Suh, J. Jeong, D.M. Geum, S.H. Baek, S.H. Kim, Oxygen scavenging in HfZrOx-Based n/p-FeFETs for switching voltage scaling and endurance/retention improvement. Adv. Electron. Mater. 9(5), 2201257 (2023)

    Article  CAS  Google Scholar 

  226. S. De, B.-H. Qiu, W.-X. Bu, M.A. Baig, P.-J. Sung, C.-J. Su, Y.-J. Lee, D.D. Lu, Uniform crystal formation and electrical variability reduction in hafnium-oxide-based ferroelectric memory by thermal engineering. ACS Appl. Electron. Mater. 3(2), 619–628 (2021)

    Article  CAS  Google Scholar 

  227. M. Materano, T. Mittmann, P.D. Lomenzo, C. Zhou, J.L. Jones, M. Falkowski, A. Kersch, T. Mikolajick, U. Schroeder, Influence of oxygen content on the structure and reliability of ferroelectric HfxZr1–xO2 layers. ACS Appl. Electron. Mater. 2(11), 3618–3626 (2020)

    Article  CAS  Google Scholar 

  228. S. Sharath, J. Kurian, P. Komissinskiy, E. Hildebrandt, T. Bertaud, C. Walczyk, P. Calka, T. Schroeder, L. Alff, Thickness independent reduced forming voltage in oxygen engineered HfO2 based resistive switching memories. Appl. Phys. Lett. 105(7), 073505 (2014)

    Article  Google Scholar 

  229. M. Balog, M. Schieber, M. Michman, S. Patai, Chemical vapor deposition and characterization of HfO2 films from organo-hafnium compounds. Thin Solid Films 41(3), 247–259 (1977)

    Article  CAS  Google Scholar 

  230. J. Robertson, High dielectric constant gate oxides for metal oxide Si transistors. Rep. Prog. Phys. 69(2), 327 (2005)

    Article  Google Scholar 

  231. E.O. Filatova, A.S. Konashuk, Interpretation of the changing the band gap of Al2O3 depending on its crystalline form: connection with different local symmetries. J. Phys. Chem. C 119(35), 20755–20761 (2015)

    Article  CAS  Google Scholar 

  232. J.P. Chang, Y.-S. Lin, K. Chu, Rapid thermal chemical vapor deposition of zirconium oxide for metal-oxide-semiconductor field effect transistor application. J. Vac. Sci. Technol. B 19(5), 1782–1787 (2001)

    Article  CAS  Google Scholar 

  233. Z. Wang, U. Helmersson, P.-O. Käll, Optical properties of anatase TiO2 thin films prepared by aqueous sol–gel process at low temperature. Thin Solid Films 405(1–2), 50–54 (2002)

    Article  CAS  Google Scholar 

  234. D. Vanderbilt, X. Zhao, D. Ceresoli, Structural and dielectric properties of crystalline and amorphous ZrO2. Thin Solid Films 486(1–2), 125–128 (2005)

    Article  CAS  Google Scholar 

  235. J.W. McPherson, J. Kim, A. Shanware, H. Mogul, J. Rodriguez, Trends in the ultimate breakdown strength of high dielectric-constant materials. IEEE Trans. Electron Devices 50(8), 1771–1778 (2003)

    Article  CAS  Google Scholar 

  236. L. Berberich, M. Bell, The dielectric properties of the rutile form of TiO2. J. Appl. Phys. 11(10), 681–692 (1940)

    Article  CAS  Google Scholar 

  237. L.V. Gurvich, I. Veyts, Thermodynamic properties of individual substances: elements and compounds (CRC Press, Boca Raton, 1990)

    Google Scholar 

  238. S. Elder, F. DiSalvo, L. Topor, A. Navrotsky, Thermodynamics of ternary nitride formation by ammonolysis: application to lithium molybdenum nitride (LiMoN2), sodium tungsten nitride (Na3WN3), and sodium tungsten oxide nitride (Na3WO3N). Chem. Mater. 5(10), 1545–1553 (1993)

    Article  CAS  Google Scholar 

  239. P.J. Linstrom, W.G. Mallard, The NIST Chemistry WebBook: a chemical data resource on the internet. J. Chem. Eng. Data 46(5), 1059–1063 (2001)

    Article  CAS  Google Scholar 

Download references

Acknowledgements

Not applicable

Funding

This work was supported by the National Research Foundation (NRF), funded by the Korean Ministry of Science and ICT (MSICT) (Grant Nos. 2020R1C1C1008193, 2022M3F3A2A01073562, 2021M3F3A2A02037889, 2021M3F3A2A01037738 and 2022R1C1C1004176), and the Korea Institute of Science and Technology (2E32491 and 2V09802).

Author information

Authors and Affiliations

Authors

Contributions

JL, KY and JYK contributed equally to this work. JL, KY, and JYK wrote the original draft manuscript. JEK, DIH and DHL reviewed the manuscript. JHY and MHP conceptualized the article and edited the manuscript. All authors read and approved the final manuscript.

Corresponding authors

Correspondence to Jung Ho Yoon or Min Hyuk Park.

Ethics declarations

Competing interests

The authors declare that they have no competing interests.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Lee, J., Yang, K., Kwon, J.Y. et al. Role of oxygen vacancies in ferroelectric or resistive switching hafnium oxide. Nano Convergence 10, 55 (2023). https://doi.org/10.1186/s40580-023-00403-4

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s40580-023-00403-4

Keywords