The mechanisms leading to wake-up and fatigue in ferroelectric hafnium zirconium oxide thin film devices with symmetric RuO2 electrodes are investigated via polarization, relative permittivity, dielectric nonlinearity, pyroelectric coefficient, and microfocus x-ray diffraction (XRD) measurements. The devices are observed to wake-up for up to 103 bipolar pulsed field cycles, after which fatigue occurs with polarization approaching zero following 108 cycles. Wake-up is accompanied by a decrease in both high-field permittivity and hysteresis loop pinching and an increase in the pyroelectric coefficient, indicating that the wake-up process involves a combination of transformations from the tetragonal to the orthorhombic phase and domain depinning from defect redistribution. Fatigue is observed to coincide with an increase in irreversible domain wall motion and a decrease in pyroelectric coefficient. Finite pyroelectric coefficients are measured on fully fatigued devices, indicating that domain pinning is a strong contributor to fatigue and that fatigued devices contain domain structures that are unable to switch under the fields applied for measurement. Microfocus XRD patterns measured on each device reveal that the phase constitution is qualitatively unaffected by field cycling and resultant polarization fatigue. These data indicate that the wake-up process has contributions from both phase transformations and domain depinning, whereas the fatigue process is driven primarily by domain pinning, and the near-zero measured switchable polarization is actually a poled device with immobile domains. These observations provide insight into the physical changes occurring during field cycling of HfO2-based ferroelectrics while examining a possible oxide electrode material for silicon CMOS device implementation.

The first report of ferroelectricity in HfO2-based thin films in 20111 generated interest for the implementation of this scalable2,3 and silicon compatible4,5 material for device applications, including ferroelectric random access memory,6,7 energy harvesting,8 non-linear optics,9,10 infrared sensing,11–13 and negative differential capacitance field effect transistors.14,15 The ferroelectric properties of this system have been attributed to a metastable, space group Pca21 orthorhombic phase,16,17 which is stabilized with respect to the room-temperature equilibrium, linear dielectric P21/c monoclinic and high-temperature, field-induced ferroelectric P42/nmc tetragonal phases with doping,18–20 reduced oxygen content,21,22 the presence of biaxial stress,23,24 and utilization of sub-30 nm film thicknesses.25,26 Of the many HfO2-based ferroelectric dopants, zirconium oxide has been identified as an alloying material (HfxZr1−xO2, HZO) that stabilizes the orthorhombic phase through nearly the entire composition window20,27 and reduces the thermal budget required for its formation compared to many other doped HfO2 compositions28 and so has received significant attention.2,20,27,29

Investigations of the ferroelectric and field cycling properties of HfO2-based thin films, including HZO, most commonly incorporate binary nitride electrode materials, including TiN1,30,31 and TaN,32 into metal-insulator-metal (MIM) device structures. These electrodes have been reported to induce tensile biaxial stress18,33,34 within and scavenge oxygen from the adjacent HfO2-based ferroelectric layer,35,36 resulting in the stabilization of the orthorhombic phase. With initial electric field cycling of these MIM devices above their coercive field, redistribution of oxygen vacancies within the ferroelectric layer has been reported to unpin ferroelectric domains37–40 and drive a transformation from the tetragonal phase to the orthorhombic phase,41,42 both of which result in an increased remanent polarization (Pr) and wake-up. With extended field cycling, domain pinning due to defect generation within the ferroelectric layer39,43 and transformations from the tetragonal and orthorhombic phases to the monoclinic phase42 cause a reduction in Pr and fatigue. Further field cycling often results in increasing defect concentration and the formation of conductive paths between the electrodes, causing dielectric breakdown.44 

The field cycling behavior of ferroelectric hafnia has been observed to vary depending upon the utilized electrode material. A direct comparison between TiN and TaN electrodes, and various combinations thereof, observed that the utilization of TaN electrodes yielded films that only woke up with field cycling and did not fatigue.32 Devices with TaN electrodes also maintained higher Pr; however, they experienced a reduction in cycles-to-breakdown. The same study found that TiN electrodes, alternatively, increased the number of cycles-to-breakdown and yielded devices that both woke up and fatigued. These differences in field cycling behavior have been attributed to differences in oxygen vacancy concentration within the ferroelectric layers due to oxygen scavenging by the neighboring electrodes,35,45 providing the potential for control of the wake-up, fatigue, and breakdown behavior via electrode engineering.

Beyond binary nitrides, conductive oxide electrodes, such as RuO246–48  and IrO2,49,50 have been explored for ferroelectric HfO2-based thin film applications. These and other conductive oxide electrode materials were experimentally shown to yield fatigue performance superior to noble metal electrodes in perovskite ferroelectrics, such as Pb(Zr,Ti)O3, with the improvements attributed to a reduction in defect-driven domain pinning.51–55 It was concluded that a large contributor to fatigue was an interfacial buildup of oxygen vacancies, which migrate to the ferroelectric/electrode interfaces during field cycling. Utilization of conductive oxide electrodes aided in mitigating this vacancy buildup by allowing oxygen vacancies to cross the ferroelectric/electrode interface, thus improving the fatigue performance.56 

To date, investigations evaluating hafnia-based ferroelectrics with IrO2 and RuO2 electrodes have reported that the fatigue performance is not improved compared to more widely utilized TiN and TaN electrodes.46,47,49,50 Reports of fatigue performance of HZO with symmetric IrO2 electrodes observed 107 cycles to breakdown and fatigue onset after only 105 cycles, with low measured remanent polarizations (∼5 μC cm−2).49,50 An investigation of asymmetric TiN/HZO/RuO2 devices observed polarization fatigue following 105 cycles, low endurance (∼107 cycles to breakdown), and low remanent polarizations (∼5 μC cm−2), all of which were attributed to a reduction in the RuO2 top electrode to metallic ruthenium during processing due to the TiN bottom electrode.46 A more recent work examining HZO with symmetric RuO2 electrodes reported remanent polarizations of 15 μC cm−2 and with consistent wake-up during field cycling.48 This work indicated that RuO2 may be an advantageous oxide electrode for devices compared to IrO2 owing to the larger reported Pr values and apparent lack of fatigue. This study did not report the utilized field cycling waveforms or procedure but showed 108 cycles-to-breakdown. In contrast to these reports of devices with symmetric and asymmetric oxide electrodes, symmetric TiN electrode devices have been shown to yield HZO devices that demonstrate a higher cycles-to-breakdown of ∼1010 and maintain modest remanent polarizations of ∼8-10 μC cm−2.57 

In this study, the mechanisms leading to fatigue in symmetric RuO2/HZO/RuO2 devices cycled with 50% duty cycle, 10 kHz, and 2.0 MV cm−1 bipolar pulses are investigated. Polarization, switching current, relative permittivity, dielectric nonlinearity, and leakage current measurements are used in conjunction with pyroelectric measurements and area detector x-ray diffraction (XRD) analyses to elucidate how the phase constitution and domain dynamics evolve within the ferroelectric layer with field cycling and wake-up/fatigue. Understanding the mechanisms leading to the low remanent polarizations and cycles-to-breakdown reported for HZO between conductive oxide electrodes may facilitate the development of electrode materials that balance the requirements for ferroelectric layer oxygen content and oxygen vacancy-driven degradation mitigation.

RuO2/HZO/RuO2 devices were prepared with 20 nm-thick HZO films between symmetric 120 nm-thick RuO2 electrodes on silicon substrates. The planar bottom RuO2 electrode was deposited via pulsed DC reactive sputtering (30 kHz/4 μs reverse time) with a power density of 4.6 W cm−2 in a Denton Discovery 550 sputtering system. Following bottom electrode deposition, HZO with a targeted composition of Hf0.5Zr0.5O2 was deposited via thermal atomic layer deposition at a temperature of 150 °C within an Ultratech Savannah instrument utilizing tetrakis(dimethylamino)hafnium (TDMA HF, at 75 °C) and tetrakis(dimethylamino)zirconium (TDMA Zr, at 75 °C) as hafnium and zirconium precursors, respectively, and H2O as an oxidant. Due to per-cycle growth rate differences between HfO2 and ZrO2, the 5:5 super-cycle ratio resulted in a final film composition of Hf0.58Zr0.42O2.11 After HZO growth, planar RuO2 electrodes were deposited at a thickness of 120 nm utilizing the same conditions as used for the bottom electrode. After top electrode deposition, samples were rapid thermal annealed in a Surface Science Integration Solaris 150 instrument at 600 °C for 30 s in a N2 atmosphere. Following annealing, samples meant for pyroelectric current and XRD measurements received 1000 μm-diameter platinum contacts through a shadow mask, whereas samples meant for electrical characterization received 100 μm-diameter photolithographically defined platinum contacts. After contact deposition, the RuO2 top electrodes were reactive ion etched to isolate discreet devices for characterization, utilizing the platinum contacts as a hard mask.

Electrical characterization, including polarization-electric field hysteresis loop [P(E)], switching current, and positive-up negative-down (PUND) pulsed polarization measurements, was carried out using a Radiant Technologies Precision LC II instrument. Nested P(E) (1.0–2.5 MV cm−1 maximum fields) and switching current (2.5 MV cm−1 maximum field) measurements were performed with a period of 1 ms, whereas nested PUND (1.0–2.5 MV cm−1 maximum fields) measurements were performed with a pulse width of 1 ms and pulse delay of 100 ms. Dielectric nonlinearity (Rayleigh) and capacitance–voltage (CV) measurements were completed with a Keysight E4980A Precision LCR Meter. A 0.05 V, 10 kHz oscillator was utilized for CV measurements, while Rayleigh measurements were performed at 10 kHz with an oscillator amplitude ranging from 0.001 to 0.09 MV cm−1. CV measurements were conducted between ±2.5 MV cm−1 with a step size of 0.05 MV cm−1. Leakage currents were measured from −2.5 to 2.5 MV cm−1 with a step size of 0.05 MV cm−1 using a Keysight B2901A Precision Source Measure Unit. Pyroelectric coefficients were measured using the technique described by Sharp and Garn58 by measuring pyroelectric current with a Keithley 6514 Electrometer while oscillating the device temperature using a Peltier cooler (TE Technology model VT-127-1.0-1.3-71) connected to a Keysight 33500B Waveform Generator, similar to measurements described elsewhere.11,58,59 For temperature oscillation, a 10 V (peak-to-peak) sinusoidal 1 mHz waveform was applied to the Peltier cooler, resulting in temperature amplitudes of approximately 9 °C. Fatigue cycling was completed with an electric field of 2.0 MV cm−1 at a frequency of 10 kHz utilizing a 50% duty cycle waveform with a Keysight 33500B Waveform Generator. The fatigue pulses were meant to simulate application-relevant waveform shapes and had a rise and fall time of 2 μs and hold time of 300 ms.60 A modified Sawyer–Tower circuit with the load capacitor replaced with a 50 Ω resistor was used to analyze the switching behavior of the samples with both the small and large contact diameters to verify that full switching was occurring at the 10 kHz pulsing frequency (data and description available in Fig. S1 in the supplementary material).

Grazing incidence XRD (GIXRD) measurements were performed in a parallel beam geometry using a Rigaku Smartlab diffractometer utilizing Cu Kα radiation with ω incident angle fixed at 0.7°. Microfocused area detector XRD measurements were completed on individually field-cycled capacitors using a Bruker D8 Venture diffractometer equipped with an Incoatec IμS 3.0 Cu Kα radiation source and a Photon III detector with MgO powder adhered to device surfaces as a stress-free height alignment standard.33 For these measurements, the ω angle was fixed at 20° and a 0.2 mm collimator was utilized to produce a beam footprint diameter of 600 μm on device surfaces.

Based upon the GIXRD pattern measured on the RuO2/HZO/RuO2 sample, shown in Fig. 1, the processed devices contain predominantly monoclinic phase. The intensity of the superimposed (111)/(101) orthorhombic/tetragonal peak, present at 30.3° in 2θ, had a low intensity compared to the (111) monoclinic peak at 31.5°. The (1¯11) monoclinic peak, typically indexed at 28.5°, was not observable due to the presence of an intense (110) RuO2 reflection at 28.1° from the 120 nm-thick bottom electrode. Based upon this diffraction pattern, it is apparent that the samples contain comparatively more monoclinic phase than films with similar composition and processing but with TaN electrodes.11 The higher amount of monoclinic phase for the samples in this study suggests that the HZO has a lower oxygen vacancy concentration,21,22 which is likely related to processing with oxide electrodes.

FIG. 1.

GIXRD pattern measured on the RuO2/HZO/RuO2 sample with indexing for each peak provided above the panel. A logarithmic intensity scale was used owing to the high intensity of the RuO2 110 reflection.

FIG. 1.

GIXRD pattern measured on the RuO2/HZO/RuO2 sample with indexing for each peak provided above the panel. A logarithmic intensity scale was used owing to the high intensity of the RuO2 110 reflection.

Close modal

Nested P(E) measurements performed on the pristine device and after 5 × 104 field cycles, shown in Figs. 2(a) and 2(b), respectively, revealed a decrease in loop pinching and an increase in polarization response consistent with wake-up. All subsequent P(E) measurements revealed decreasing polarizations and fatigue. As shown in Fig. 2(c), the device response appeared as a linear dielectric following 107 bipolar pulses. The same trends are evident in the switching current. The switching current measured on the pristine device is plotted in Fig. 2(d) and shows two distinct switching peaks that condense to a single peak in the measurement made following 5 × 104 cycles [Fig. 2(e)]. This merging of switching current peaks is consistent with wake-up.32,49 After this initial wake-up, the switching current decreases with each additional field cycling interval until the response approaches that of a linear dielectric. The measurement made following 107 cycles, which lacks significant switching current peaks, is shown in Fig. 2(f). The coercive fields (Ec) from hysteresis and switching current measurements are shown in Fig. S2 in the supplementary material. The values were observed to increase from 0.65/−0.40 to 1.12/−0.87 MV cm−1 following initial hysteresis de-pinching. The values then remained relatively constant until 105 cycles, at which point they decreased throughout the fatigue process to 0 MV cm−1. The low polarizations and switching currents are consistent with the GIXRD phase analysis and indicate that the films contain substantial fractions of the non-ferroelectric monoclinic phase. Regardless, the electrical responses indicate that these devices wake-up with 104 bipolar pulses, fatigue with subsequent field cycling, and yield well-saturating polarization and switching current responses irrespective of field cycling progress.

FIG. 2.

Nested polarization hysteresis measurements made following (a) 0, (b) 50 000, and (c) 10 000 000 cycles. Current loop measurements made following (d) 0, (e) 50 000, and (f) 10 000 000 cycles.

FIG. 2.

Nested polarization hysteresis measurements made following (a) 0, (b) 50 000, and (c) 10 000 000 cycles. Current loop measurements made following (d) 0, (e) 50 000, and (f) 10 000 000 cycles.

Close modal

Wake-up and fatigue are also evident in the change in the switchable polarization (Psw) with cycling, as assessed from PUND measurements and shown in Fig. 3. The initial Psw of 4.26 μC cm−2 increased to 6.38 μC cm−2 with wake-up through the first 103 cycles and then decreased to 0.14 μC cm−2 following 108 cycles during fatigue. The switchable polarization quantity characterizes the difference in polarization magnitude between the up-poled and down-poled states. Thus, while the switchable polarization decreases with field cycling, the PUND measurement does not necessarily indicate a lack of polar material, domain structure, or the orientation of any domain structure that is unable to switch under the fields applied for measurement. The magnitudes of these Psw values are consistent with previous reports that examined HZO with symmetric IrO2 electrodes49,50 and lower than investigations of HZO MIM devices with TiN and TaN electrodes. These values are also lower than values reported in the other recent study examining symmetric RuO2 electrodes,48 which may be related to the 20 nm HZO thickness utilized for this study, as thinner films typically contain more of the ferroelectric orthorhombic phase.26 In contrast to the devices with IrO2 electrodes that began to fatigue following 105 cycles, the fatigue in these RuO2 devices began after 103 cycles, and they did not suffer dielectric breakdown after the switchable polarization became zero following 108 cycles.

FIG. 3.

Switchable polarization quantified from PUND measurements as a function of the number of 50% duty cycle, 2.0 MV cm−1, and 10 kHz bipolar pulses.

FIG. 3.

Switchable polarization quantified from PUND measurements as a function of the number of 50% duty cycle, 2.0 MV cm−1, and 10 kHz bipolar pulses.

Close modal

While the decrease in Psw reveals that the concentration of switchable domains decreases during fatigue, the mechanism by which this decrease is occurring is not evident based on field cycling-resolved P(E), switching current, and PUND measurements. Analyses of CV and Rayleigh measurements provide insight into the mechanism. CV data are shown in Fig. 4(a). Similar to switchable polarizations, the relative permittivity maxima at the coercive fields also reveal a decrease in magnitude, and the behavior appears more consistent with a linear dielectric response following 108 cycles. The high-field permittivities extracted from the CV profiles are plotted in Fig. 4(b) and reveal a decrease from 21.1 to 21.0 through 104 cycles during wake-up. The magnitude of these permittivities is lower than those reported in investigations utilizing TiN,57 TaN,33 and IrO249  electrodes and is likely related to the high content of the monoclinic phase in these samples. The decrease in high-field permittivity with wake-up is suggestive of a phase transformation from the tetragonal to the orthorhombic phase, as has been observed using synchrotron XRD measurements of W/HZO/W and TaN/HZO/TaN devices42 and TEM analyses of field-cycled Gd-doped HfO2 with TiN electrodes,41 respectively. However, this decrease is likely too small to completely account for the entire increase in Psw of 153% with field cycling, indicating that domain depinning processes, widely observed during initial cycling of HfO2-based ferroelectrics,40,43 are also contributing to the observed wake-up. After decreasing during wake-up, the high-field relative permittivity is observed to increase during fatigue from 21.0 to 21.1. By comparison, the same investigations, which examined field cycling of HZO (with Pt electrodes)42 and Gd-doped HfO2 (with TiN electrodes),41 reported decreases in permittivity magnitude of approximately 3–5 throughout the fatigue process. In the case of HZO with Pt electrodes, it was determined from combined microdiffraction and electrical analysis that a phase transformation from the orthorhombic and tetragonal phases to the monoclinic phase accompanied the polarization fatigue,42 whereas in the Gd-doped HfO2 investigation with TiN electrodes, it was concluded that an increase in oxygen vacancy concentration resulted in domain pinning.41 While the polarization decreases similarly in the Pt/HZO/Pt, TiN/Gd-doped HfO2/TiN, and RuO2/HZO/RuO2 devices during fatigue, the permittivity behavior is different. Polarization fatigue in devices with noble metal and binary nitride electrodes coincides with a decrease in permittivity, while the fatigue coincides with a permittivity increase in these devices with RuO2 electrodes. This difference in permittivity behavior indicates that the fatigue mechanisms in devices with oxide electrodes may be different from that in devices with binary nitride or noble metal electrodes. The minimal permittivity change with fatigue, particularly in comparison to the Pt/HZO/Pt investigation, is also evidence that the phase constitution is not substantially changing with field cycling in these devices with RuO2 electrodes.

FIG. 4.

(a) Relative permittivity (lines, left axis) and loss (open circles, right axis) measured vs the applied DC field at intervals between 0 and 108 field cycles, with the number of cycles indicated by the color bar above the panel. (b) Relative permittivity measured at +2.5 MV cm−1 (closed circles) and −2.5 MV cm−1 (open circles) at intervals between 0 and 108 bipolar waves.

FIG. 4.

(a) Relative permittivity (lines, left axis) and loss (open circles, right axis) measured vs the applied DC field at intervals between 0 and 108 field cycles, with the number of cycles indicated by the color bar above the panel. (b) Relative permittivity measured at +2.5 MV cm−1 (closed circles) and −2.5 MV cm−1 (open circles) at intervals between 0 and 108 bipolar waves.

Close modal

Figure 5(a) shows dielectric nonlinearity measurements that also reflect this decrease in permittivity magnitude and provide further insight into the mechanisms causing the polarization fatigue. The reversible Rayleigh coefficient values (ɛinit), extracted at the threshold AC field (∼0.02 MV cm−1), are plotted in Fig. 5(b). ɛinit was observed to remain effectively constant at 20.95 through the first 104 bipolar pulses and then decrease, coinciding with the start of the polarization fatigue, to 20.75. Given that phase transformations were not obviously observed in the high DC field relative permittivities, the decrease in ɛinit at the onset of fatigue is evidence that the domain boundaries move less freely as fatigue progresses. Figure 5(b) shows the irreversible Rayleigh coefficient (α′) with field cycling. The irreversible Rayleigh coefficient can characterize the energetic barrier landscape encountered by domain boundaries as they traverse the ferroelectric film under an AC field below the coercive field.61 larger α′ values are indicative of higher irreversible contributions to domain wall movement, which is consistent with increased domain wall pinning.62 In these devices, α′ was observed to increase from 0.40 to 0.70 cm MV−1 through field cycling; once again, a notable increase is observed at 104 cycles. Thus, the increase in α′ coincides with the onset of fatigue observed in Psw, P(E), switching current, and high field permittivity measurements and is evidence that the energy landscape for domain wall motion is being altered at the same time that the switchable polarization decreases. Between 107 and 108 cycles, α′ then decreased from 0.70 to 0.65 cm MV−1, which may be related to increased energy barriers for domain movement as the final steps of fatigue are approached.

FIG. 5.

(a) Relative permittivity (left axis) and loss tangent (right axis) as a function of the AC oscillator level measured at various intervals of bipolar waves, as indicated by the color bar above the plot. (b) Reversible Rayleigh coefficient, ɛinit (left axis, closed points), and irreversible Rayleigh coefficient, α′ (right axis, open points), measured at various cycling intervals.

FIG. 5.

(a) Relative permittivity (left axis) and loss tangent (right axis) as a function of the AC oscillator level measured at various intervals of bipolar waves, as indicated by the color bar above the plot. (b) Reversible Rayleigh coefficient, ɛinit (left axis, closed points), and irreversible Rayleigh coefficient, α′ (right axis, open points), measured at various cycling intervals.

Close modal

Despite the diminished switchable polarizations in the highly fatigued devices, the presence of a finite α′ value is indicative that a domain structure remains within the ferroelectric layer. Given that PUND measurements can only evaluate the amount of switchable polarization and do not indicate the degree to which or direction in which poled, but unswitchable domains exists in the film, pyroelectric coefficient measurements were made on devices cycled at each interval to further examine the evolution of polarization magnitude with field cycling. Pyroelectric measurements were made after initial poling/cycling and after 500 h at rest to minimize possible contributions from electret effects.59 The phase differences between the sinusoidal current and temperature signals were fit to be −90° ± 3°, indicating that the current contributions were solely from crystallographic dipoles. The pyroelectric current generated by the initially poled device (i.e., after only a single PUND measurement and no fatigue cycling) is shown in Fig. 6(a). A current density amplitude of 3.7 × 10−10 A cm−2 was measured while the temperature oscillated with an amplitude of 9 °C. With 104 cycles and wake-up, the current density amplitude remained approximately the same, as shown in Fig. 6(b). Following 107 cycles and fatigue, the current density amplitude decreased to 2.2 × 10−10 A cm−2, as shown in Fig. 6(c). Based on an analysis of the phase difference between the pyroelectric current and temperature, it was determined that the devices are poled in an up orientation during each measurement (i.e., the positive end of the dipole is oriented toward the top electrode), which is consistent with sequence and polarity of PUND pulses utilized to measure the devices prior to pyroelectric current measurement.

FIG. 6.

Pyroelectric current (light red, left axis) and temperature oscillation (light gray, right axis) and associated sinusoidal fits (dark red and black, respectively) measured on devices cycled with (a) 1, (b) 50 000, and (c) 10 000 000 bipolar pulses.

FIG. 6.

Pyroelectric current (light red, left axis) and temperature oscillation (light gray, right axis) and associated sinusoidal fits (dark red and black, respectively) measured on devices cycled with (a) 1, (b) 50 000, and (c) 10 000 000 bipolar pulses.

Close modal

Given that the device that had been cycled with 107 waves was measured to have a significantly reduced Psw of 0.62 μC cm−2, the presence of a measurable pyroelectric current is evidence that a poled domain structure remains in the HZO layer in spite of the severe fatigue. It is hypothesized that the decrease in pyroelectric current with field cycling is due to the presence of domains that are pinned in the poled-down direction and thus provide a current oscillation that partially cancels the response of the majority poled-up domains. Some decrease in the pyroelectric current could also be attributed to field-cycling-related phase transformations from the orthorhombic to monoclinic phases, although this is not supported by the increasing high-field relative permittivity coinciding with fatigue. Thus, given that the pyroelectric current decreases with field cycling, but does not disappear completely, it is evident that domain pinning is a major contributor to the polarization decrease, and a domain structure persists in the device following complete fatigue. The derived pyroelectric coefficients are shown in Fig. 7 along with Psw measured on each individual device. It is apparent that the 1000 μm-diameter devices on which pyroelectric measurements were performed have similar fatigue behavior to the 100 μm-diameter devices used for the previous electrical measurements. The pyroelectric coefficient was initially measured to be −11.5 μC m−2 K−1. This value is consistent with previous measurements of the HZO system with sizable fractions of the monoclinic phase.11 Upon wake-up, the pyroelectric coefficient increased to −13.2 μC m−2 K−1 and then decreased to −7.0 μC m−2 K−1 through 108 cycles during fatigue. The fatigued pyroelectric coefficient represents a decrease of 39% from the initial value, while Psw decreased by 95% for these devices.

FIG. 7.

Pyroelectric coefficients (closed circles, left axis) and switchable polarizations (open circles, right axis) measured on devices cycled with various intervals of bipolar pulses.

FIG. 7.

Pyroelectric coefficients (closed circles, left axis) and switchable polarizations (open circles, right axis) measured on devices cycled with various intervals of bipolar pulses.

Close modal

To corroborate that the phases and phase fractions present within the RuO2/HZO/RuO2 structures remain essentially unchanged by fatigue, x-ray microdiffraction data were collected with an area detector on devices cycled with each interval of bipolar pulses. An example detector frame and an unwarped 2D pattern are shown in Fig. S3 in the supplementary material for the singly poled device. Debye rings, which correspond to the RuO2 electrodes, the HZO layer, the platinum contact, and the MgO powder height alignment standard, were present in each pattern. Each area detector pattern was integrated through all captured γ angles to produce 1D diffraction patterns36 for devices pulsed at each cycling interval, as shown in Fig. 8. Owing to the constrained beam footprint necessary to characterize the individual devices, the integrated counts for each pattern were too low to effectively fit and compare peak positions and intensities. Regardless, peaks corresponding to the superimposed (111) orthorhombic/(101) tetragonal and (111) monoclinic HZO reflections were indexed at 30.3° and 31.4°, respectively, while (110), (011), and (200) RuO2 reflections were indexed at 28.1°, 35.2°, and 39.7°, respectively. A (111) MgO reflection, which was used for sample alignment, was indexed at 36.9°. All angles refer to the 2θ axis. Given the low integrated intensities and the presence of the intense (110) RuO2 reflection overlapping the (1¯11) monoclinic HZO peak, quantitative phase fraction evaluations36,42 were not possible. However, peaks consistent with all three HZO phases were observed regardless of the number of field cycles, and the relative intensity ratios of the (111) orthorhombic/(101) tetragonal to (111) monoclinic peaks appear to be qualitatively similar for each device. Thus, diffraction patterns indicate that the orthorhombic phase is likely present in the highly fatigued samples and its volume fraction is qualitatively consistent regardless of state of wake-up or fatigue. Furthermore, the strong pyroelectric response in the fatigued devices is evidence that the orthorhombic phase remained present.

FIG. 8.

Diffraction patterns integrated from area detector patterns collected via microfocus XRD on individual devices following various intervals of bipolar waves, as indicated by the color bar on the right. Peak indices are provided above the plot and identified above the panel. A logarithmic intensity scale was used to increase the visibility of HZO peaks.

FIG. 8.

Diffraction patterns integrated from area detector patterns collected via microfocus XRD on individual devices following various intervals of bipolar waves, as indicated by the color bar on the right. Peak indices are provided above the plot and identified above the panel. A logarithmic intensity scale was used to increase the visibility of HZO peaks.

Close modal

While domain pinning was found to be responsible for the observed polarization fatigue in these devices, the cause of the pinning was not revealed by polarization, relative permittivity, pyroelectric, or diffraction measurements. An increase in the number of point defects, which are known to be pinning centers for ferroelectric domains,63 could be responsible for the observed increase in irreversible domain wall motion and polarization fatigue. Leakage current measurements have previously been utilized to estimate defect concentrations by other investigations22,64 where the measured current is associated with trap-assisted tunneling occurring due to the presence of point defects. Larger leakage currents have been attributed to high defect concentrations.65 Leakage current profiles measured following each field cycling interval on the HZO devices with RuO2 electrodes are shown in Fig. 9(a). Throughout the first 105 cycles, the leakage currents remain effectively at the noise level of the instrumentation, 10−7 A cm−2. Following 105 cycles, however, the measured leakage currents increase, with a maximum measured at ∼10−3 A cm−2 following 108 cycles and complete polarization fatigue. This trend is shown in Fig. 9(b), in which the leakage current magnitudes at ±2.5 MV cm−1 are plotted. The beginning of fatigue and the increase in irreversible domain wall motion coincide with an increase in leakage current, indicating that the increased energy barriers for domain motion may be caused by electrically active point defects. The specific chemistry of these point defects is not directly evident based upon the leakage current measurements. While the generation of oxygen vacancies is known to occur due to field cycling in HfO2-based ferroelectrics,39 the likelihood of these point defects remaining within the 20 nm-thick, fast-oxygen-ion conducting HZO layer66 during field cycling is low given the conductive RuO2 electrodes, which have been observed to mitigate interfacial oxygen vacancy buildup in other systems.51 In tandem, the investigation of the fatigue behavior of HZO with an oxygen-blocking electrode, Pt, revealed that a phase transformation due to field cycling was causing the polarization decrease in the Pt/HZO/Pt devices.42 Such a phase transformation was not observed in high-field relative permittivity or diffraction measurements in these devices. Furthermore, these devices were not observed to break down even after extensive cycling. These observations support the notion that oxygen vacancies and their accumulation may not be the point defects predominantly causing the polarization fatigue. Alternatively, electron accumulation at domain walls, known to both occur in and cause conduction in conventional perovskite oxide ferroelectrics,67,68 may be responsible for the fatigue. Electron trapping at domain walls has been reported as a contributor to domain pinning and fatigue in HfO2-based ferroelectrics69 via an examination of activation energies for polarization regeneration and is more likely to be causing the fatigue in these RuO2/HZO/RuO2 devices. While the exact nature of the accumulating defects can only be inferred based upon the electrode materials, a clear pyroelectric response was measured in devices that have near-zero switchable polarization. These data suggest that the devices maintain a spontaneous polarization, but that polarization cannot be switched due to domain pinning or domain seed nucleation inhibition.

FIG. 9.

(a) Leakage current profiles measured on devices cycled at various intervals, as indicated by the color bar at the top of the panel. (b) Leakage current densities at ±2.5 MV cm−1 extracted from leakage current profiles as a function of the number of field cycles.

FIG. 9.

(a) Leakage current profiles measured on devices cycled at various intervals, as indicated by the color bar at the top of the panel. (b) Leakage current densities at ±2.5 MV cm−1 extracted from leakage current profiles as a function of the number of field cycles.

Close modal

In conclusion, the field cycling behavior and mechanisms leading to observed polarization wake-up and fatigue in RuO2/HZO/RuO2 ferroelectric devices have been investigated. Utilizing 50% duty cycle, 10 kHz, 2.0 MV cm−1 bipolar pulses, polarization wake-up, from Psw = 4.26 to 6.38 μC cm−1, was observed through 103 cycles, followed by fatigue to ∼0 μC cm−1 through 108 cycles. Pyroelectric coefficients increased from −11.5 m−2 K−1 to −13.2 μC m−2 K−1 during wake-up and then decreased to −7.0 μC m−2 K−1 during fatigue through 108 cycles. The polarization wake-up, which coincided with a decrease in high field permittivity and increased pyroelectric coefficient, was determined to be related to a combination of domain depinning and transformations from the tetragonal to the orthorhombic phase, likely due to redistribution of point defects or transport of oxygen vacancies into the oxide electrodes. The polarization fatigue was shown to coincide with an increase in α′, related to more irreversible domain wall movement. The pyroelectric coefficients were observed to decrease from −13.24 to −7.0 μC m−2 K−1, whereas the switchable polarization decreased to nearly zero. The finite pyroelectric coefficient indicates that fully fatigued devices contain an unswitchable domain structure and remain polar. The existence of the orthorhombic phase was qualitatively confirmed in devices cycled at each interval via microfocused XRD measurements, corroborating that the phase constitution necessary for pyroelectric response existed in the highly fatigued devices. Thus, domain pinning appears to be a significant factor in the fatigue of these RuO2/HZO/RuO2 devices. These observations reveal the origin of switchable polarization changes in HZO with RuO2 electrodes due to field cycling while evaluating this electrode material for next-generation ferroelectric device implementation into silicon CMOS technology.

See the supplementary material for description and data from modified Sawyer–Tower measurement of 100 μm-diameter and 1000 μm-diameter contacts, change in coercive field with field cycling, and an area detector frame and an unwarped pattern corresponding to the singly poled device.

The authors wish to acknowledge Dr. Glen Fox for his guidance on application-relevant fatigue waveforms. This research is supported by the Laboratory Directed Research and Development program at Sandia National Laboratories (film growth and device preparation), the Semiconductor Research Corporation's (SRC) Nanomanufacturing Materials and Processes Program (material characterization, including x-ray diffraction and electrical measurements), the U.S. National Science Foundation's Major Research Instrumentation program under Grant No. CHE-2018870 (area detector x-ray diffractometer), and as part of the Center for 3D Ferroelectric Microelectronics (3DFeM), an Energy Frontier Research Center funded by the U.S. Department of Energy, Office of Science, Basic Energy Sciences under Award No. DE-SC0021118 (fatigue protocols and instrumentation). S.T.J. acknowledges support from the U.S. National Science Foundation's Graduate Research Fellowship Program under Grant No. DGE-1842490. Sandia National Laboratories is a multimission laboratory managed and operated by National Technology and Engineering Solutions of Sandia, LLC, a wholly owned subsidiary of Honeywell International, Inc., for the U.S. Department of Energy's National Nuclear Security Administration under Contract No. DE-NA0003525. This paper describes objective technical results and analysis. Any subjective views or opinions that might be expressed in the paper do not necessarily represent the views of the U.S. Department of Energy or the United States Government.

The data that support the findings of this study are available within the article and its supplementary material.

1.
T. S.
Böscke
,
J.
Müller
,
D.
Bräuhaus
,
U.
Schröder
, and
U.
Böttger
,
Appl. Phys. Lett.
99
,
102903
(
2011
).
2.
S. S.
Cheema
,
D.
Kwon
,
N.
Shanker
,
R.
dos Reis
,
S.-L.
Hsu
,
J.
Xiao
,
H.
Zhang
,
R.
Wagner
,
A.
Datar
,
M. R.
McCarter
,
C. R.
Serrao
,
A. K.
Yadav
,
G.
Karbasian
,
C.-H.
Hsu
,
A. J.
Tan
,
L.-C.
Wang
,
V.
Thakare
,
X.
Zhang
,
A.
Mehta
,
E.
Karapetrova
,
R. V.
Chopdekar
,
P.
Shafer
,
E.
Arenholz
,
C.
Hu
,
R.
Proksch
,
R.
Ramesh
,
J.
Ciston
, and
S.
Salahuddin
,
Nature
580
,
478
482
(
2020
).
3.
S. J.
Kim
,
J.
Mohan
,
H. S.
Kim
,
J.
Lee
,
C. D.
Young
,
L.
Colombo
,
S. R.
Summerfelt
,
T.
San
, and
J.
Kim
,
Appl. Phys. Lett.
113
,
182903
(
2018
).
4.
K. J.
Hubbard
and
D. G.
Schlom
,
J. Mater. Res.
11
,
2757
2776
(
1996
).
5.
D. G.
Schlom
,
S.
Guha
, and
S.
Datta
,
MRS Bull.
33
,
1017
1025
(
2008
).
6.
J.
Müller
,
T. S.
Böscke
,
D.
Bräuhaus
,
U.
Schröder
,
U.
Böttger
,
J.
Sundqvist
,
P.
Kücher
,
T.
Mikolajick
, and
L.
Frey
,
Appl. Phys. Lett.
99
,
112901
(
2011
).
7.
S.
Mueller
,
S. R.
Summerfelt
,
J.
Muller
,
U.
Schroeder
, and
T.
Mikolajick
,
IEEE Electron Device Lett.
33
,
1300
1302
(
2012
).
8.
M. H.
Park
,
H. J.
Kim
,
Y. J.
Kim
,
T.
Moon
,
K. D.
Kim
, and
C. S.
Hwang
,
Nano Energy
12
,
131
140
(
2015
).
9.
J. F.
Ihlefeld
,
T. S.
Luk
,
S. W.
Smith
,
S. S.
Fields
,
S. T.
Jaszewski
,
W. T.
Riffe
,
S.
Bender
,
C.
Constantin
,
M. V.
Ayyasamy
,
P. V.
Balachandran
,
P.
Lu
,
M. D.
Henry
, and
P. S.
Davids
,
J. Appl. Phys.
128
,
034101
(
2020
).
10.
J.
Qin
,
F.
Huang
,
X.
Li
,
L.
Deng
,
T.
Kang
,
A.
Markov
,
F.
Yue
,
Y.
Chen
,
X.
Wen
,
S.
Liu
,
Q.
Xiong
,
S.
Semin
,
T.
Rasing
,
D.
Modotto
,
R.
Morandotti
,
J.
Xu
,
H.
Duan
, and
L.
Bi
,
ACS Nano
13
,
1213
(
2019
).
11.
S. W.
Smith
,
A. R.
Kitahara
,
M. A.
Rodriguez
,
M. D.
Henry
,
M. T.
Brumbach
, and
J. F.
Ihlefeld
,
Appl. Phys. Lett.
110
,
072901
(
2017
).
12.
S.
Pandya
,
G.
Velarde
,
L.
Zhang
, and
L. W.
Martin
,
Phys. Rev. Mater.
2
(
12
),
124405
(
2018
).
13.
C.
Mart
,
M.
Czernohorsky
,
S.
Zybell
,
T.
Kämpfe
, and
W.
Weinreich
,
Appl. Phys. Lett.
113
,
122901
(
2018
).
14.
H.
Mulaosmanovic
,
J.
Ocker
,
S.
Müller
,
U.
Schroeder
,
J.
Müller
,
P.
Polakowski
,
S.
Flachowsky
,
R.
van Bentum
,
T.
Mikolajick
, and
S.
Slesazeck
,
ACS Appl. Mater. Interfaces
9
,
3792
3798
(
2017
).
15.
T. S.
Boscke
,
J.
Muller
,
D.
Brauhaus
,
U.
Schroder
, and
U.
Bottger
, in
2011 International Electron Devices Meeting
(
IEEE
,
Washington, DC
,
2011
), pp.
24.5.1
24.5.4
.
16.
R.
Batra
,
T. D.
Huan
,
J. L.
Jones
,
G.
Rossetti
, and
R.
Ramprasad
,
J. Phys. Chem. C
121
,
4139
(
2017
).
17.
X.
Sang
,
E. D.
Grimley
,
T.
Schenk
,
U.
Schroeder
, and
J. M.
LeBeau
,
Appl. Phys. Lett.
106
,
162905
(
2015
).
18.
U.
Schroeder
,
C.
Richter
,
M. H.
Park
,
T.
Schenk
,
M.
Pešić
,
M.
Hoffmann
,
F. P. G.
Fengler
,
D.
Pohl
,
B.
Rellinghaus
,
C.
Zhou
,
C.-C.
Chung
,
J. L.
Jones
, and
T.
Mikolajick
,
Inorg. Chem.
57
,
2752
2765
(
2018
).
19.
J.
Müller
,
U.
Schröder
,
T. S.
Böscke
,
I.
Müller
,
U.
Böttger
,
L.
Wilde
,
J.
Sundqvist
,
M.
Lemberger
,
P.
Kücher
,
T.
Mikolajick
, and
L.
Frey
,
J. Appl. Phys.
110
,
114113
(
2011
).
20.
J.
Müller
,
T. S.
Böscke
,
U.
Schröder
,
S.
Mueller
,
D.
Bräuhaus
,
U.
Böttger
,
L.
Frey
, and
T.
Mikolajick
,
Nano Lett.
12
,
4318
4323
(
2012
).
21.
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
, and
U.
Schroeder
,
Adv. Mater. Interfaces
6
,
1900042
(
2019
).
22.
A.
Pal
,
V. K.
Narasimhan
,
S.
Weeks
,
K.
Littau
,
D.
Pramanik
, and
T.
Chiang
,
Appl. Phys. Lett.
110
,
022903
(
2017
).
23.
T.
Shiraishi
,
K.
Katayama
,
T.
Yokouchi
,
T.
Shimizu
,
T.
Oikawa
,
O.
Sakata
,
H.
Uchida
,
Y.
Imai
,
T.
Kiguchi
,
T. J.
Konno
, and
H.
Funakubo
,
Appl. Phys. Lett.
108
,
262904
(
2016
).
24.
T.
Shiraishi
,
K.
Katayama
,
T.
Yokouchi
,
T.
Shimizu
,
T.
Oikawa
,
O.
Sakata
,
H.
Uchida
,
Y.
Imai
,
T.
Kiguchi
,
T. J.
Konno
, and
H.
Funakubo
,
Mater. Sci. Semicond. Process.
70
,
239
245
(
2017
).
25.
T.
Mittmann
,
F. P. G.
Fengler
,
C.
Richter
,
M. H.
Park
,
T.
Mikolajick
, and
U.
Schroeder
,
Microelectron. Eng.
178
,
48
51
(
2017
).
26.
M.
Hyuk Park
,
H.
Joon Kim
,
Y.
Jin Kim
,
W.
Lee
,
T.
Moon
, and
C.
Seong Hwang
,
Appl. Phys. Lett.
102
,
242905
(
2013
).
27.
M. H.
Park
,
H. J.
Kim
,
Y. J.
Kim
,
Y. H.
Lee
,
T.
Moon
,
K. D.
Kim
,
S. D.
Hyun
,
F.
Fengler
,
U.
Schroeder
, and
C. S.
Hwang
,
ACS Appl. Mater. Interfaces
8
,
15466
15475
(
2016
).
28.
H. A.
Hsain
,
Y.
Lee
,
G.
Parsons
, and
J. L.
Jones
,
Appl. Phys. Lett.
116
,
192901
(
2020
).
29.
H.
Yu
,
C.-C.
Chung
,
N.
Shewmon
,
S.
Ho
,
J. H.
Carpenter
,
R.
Larrabee
,
T.
Sun
,
J. L.
Jones
,
H.
Ade
,
B. T.
O’Connor
, and
F.
So
,
Adv. Funct. Mater.
27
,
1700461
(
2017
).
30.
S. J.
Kim
,
D.
Narayan
,
J.-G.
Lee
,
J.
Mohan
,
J. S.
Lee
,
J.
Lee
,
H. S.
Kim
,
Y.-C.
Byun
,
A. T.
Lucero
,
C. D.
Young
,
S. R.
Summerfelt
,
T.
San
,
L.
Colombo
, and
J.
Kim
,
Appl. Phys. Lett.
111
,
242901
(
2017
).
31.
Y.
Lee
,
H.
Alex Hsain
,
S. S.
Fields
,
S. T.
Jaszewski
,
M. D.
Horgan
,
P. G.
Edgington
,
J. F.
Ihlefeld
,
G. N.
Parsons
, and
J. L.
Jones
,
Appl. Phys. Lett.
118
,
012903
(
2021
).
32.
M.
Hoffmann
,
U.
Schroeder
,
T.
Schenk
,
T.
Shimizu
,
H.
Funakubo
,
O.
Sakata
,
D.
Pohl
,
M.
Drescher
,
C.
Adelmann
,
R.
Materlik
,
A.
Kersch
, and
T.
Mikolajick
,
J. Appl. Phys.
118
,
072006
(
2015
).
33.
S. S.
Fields
,
D. H.
Olson
,
S. T.
Jaszewski
,
C. M.
Fancher
,
S. W.
Smith
,
D. A.
Dickie
,
G.
Esteves
,
M. D.
Henry
,
P. S.
Davids
,
P. E.
Hopkins
, and
J. F.
Ihlefeld
,
Appl. Phys. Lett.
118
,
102901
(
2021
).
34.
T.
Schenk
,
C. M.
Fancher
,
M. H.
Park
,
C.
Richter
,
C.
Künneth
,
A.
Kersch
,
J. L.
Jones
,
T.
Mikolajick
, and
U.
Schroeder
,
Adv. Electron. Mater.
5
,
1900303
(
2019
).
35.
W.
Hamouda
,
A.
Pancotti
,
C.
Lubin
,
L.
Tortech
,
C.
Richter
,
T.
Mikolajick
,
U.
Schroeder
, and
N.
Barrett
,
J. Appl. Phys.
127
,
064105
(
2020
).
36.
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
, and
J. F.
Ihlefeld
,
Adv. Mater. Interfaces
8
,
2100018
(
2021
).
37.
D.
Zhou
,
J.
Xu
,
Q.
Li
,
Y.
Guan
,
F.
Cao
,
X.
Dong
,
J.
Müller
,
T.
Schenk
, and
U.
Schröder
,
Appl. Phys. Lett.
103
,
192904
(
2013
).
38.
S.
Starschich
,
S.
Menzel
, and
U.
Böttger
,
Appl. Phys. Lett.
108
,
032903
(
2016
).
39.
M.
Pešić
,
F. P. G.
Fengler
,
L.
Larcher
,
A.
Padovani
,
T.
Schenk
,
E. D.
Grimley
,
X.
Sang
,
J. M.
LeBeau
,
S.
Slesazeck
,
U.
Schroeder
, and
T.
Mikolajick
,
Adv. Funct. Mater.
26
,
4601
4612
(
2016
).
40.
A.
Chouprik
,
M.
Spiridonov
,
S.
Zarubin
,
R.
Kirtaev
,
V.
Mikheev
,
Y.
Lebedinskii
,
S.
Zakharchenko
, and
D.
Negrov
,
ACS Appl. Electron. Mater.
1
,
275
287
(
2019
).
41.
E. D.
Grimley
,
T.
Schenk
,
X.
Sang
,
M.
Pešić
,
U.
Schroeder
,
T.
Mikolajick
, and
J. M.
LeBeau
,
Adv. Electron. Mater.
2
,
1600173
(
2016
).
42.
S. S.
Fields
,
S. W.
Smith
,
P. J.
Ryan
,
S. T.
Jaszewski
,
I. A.
Brummel
,
A.
Salanova
,
G.
Esteves
,
S. L.
Wolfley
,
M. D.
Henry
,
P. S.
Davids
, and
J. F.
Ihlefeld
,
ACS Appl. Mater. Interfaces
12
,
26577
26585
(
2020
).
43.
F. P. G.
Fengler
,
M.
Pešić
,
S.
Starschich
,
T.
Schneller
,
C.
Künneth
,
U.
Böttger
,
H.
Mulaosmanovic
,
T.
Schenk
,
M. H.
Park
,
R.
Nigon
,
P.
Muralt
,
T.
Mikolajick
, and
U.
Schroeder
,
Adv. Electron. Mater.
3
,
1600505
(
2017
).
44.
S.
Starschich
,
S.
Menzel
, and
U.
Böttger
,
J. Appl. Phys.
121
,
154102
(
2017
).
45.
W.
Hamouda
,
C.
Lubin
,
S.
Ueda
,
Y.
Yamashita
,
O.
Renault
,
F.
Mehmood
,
T.
Mikolajick
,
U.
Schroeder
,
R.
Negrea
, and
N.
Barrett
,
Appl. Phys. Lett.
116
,
252903
(
2020
).
46.
M. H.
Park
,
H. J.
Kim
,
Y. J.
Kim
,
W.
Jeon
,
T.
Moon
, and
C. S.
Hwang
,
Phys. Status Solidi Rapid Res. Lett.
8
,
532
535
(
2014
).
47.
Y.
Goh
,
S. H.
Cho
,
S.-H. K.
Park
, and
S.
Jeon
,
IEEE Trans. Electron Devices
67
,
3431
3434
(
2020
).
48.
Y.
Goh
,
S. H.
Cho
,
S.-H. K.
Park
, and
S.
Jeon
,
Nanoscale
12
,
9024
9031
(
2020
).
49.
T.
Mittmann
,
T.
Szyjka
,
H.
Alex
,
M. C.
Istrate
,
P. D.
Lomenzo
,
L.
Baumgarten
,
M.
Müller
,
J. L.
Jones
,
L.
Pintilie
,
T.
Mikolajick
, and
U.
Schroeder
,
Phys. Status Solidi Rapid Res. Lett.
15
,
2100012
(
2021
).
50.
T.
Szyjka
,
L.
Baumgarten
,
T.
Mittmann
,
Y.
Matveyev
,
C.
Schlueter
,
T.
Mikolajick
,
U.
Schroeder
, and
M.
Müller
,
Phys. Status Solidi Rapid Res. Lett.
15
,
2100027
(
2021
).
51.
D. P.
Vijay
and
S. B.
Desu
,
J. Electrochem. Soc.
140
,
L7
(
1993
).
52.
E.
Bouyssou
,
P.
Leduc
,
G.
Guégan
, and
R.
Jérisian
,
J. Phys.: Conf. Ser.
10
,
317
320
(
2005
).
53.
I. K.
Yoo
and
S. B.
Desu
,
Phys. Status Solidi A
133
,
565
573
(
1992
).
54.
R.
Ramesh
,
H.
Gilchrist
,
T.
Sands
,
V. G.
Keramidas
,
R.
Haakenaasen
, and
D. K.
Fork
,
Appl. Phys. Lett.
63
,
3592
3594
(
1993
).
55.
R.
Dat
,
D. J.
Lichtenwalner
,
O.
Auciello
, and
A. I.
Kingon
,
Appl. Phys. Lett.
64
,
2673
2675
(
1994
).
56.
A. K.
Tagantsev
,
I.
Stolichnov
,
E. L.
Colla
, and
N.
Setter
,
J. Appl. Phys.
90
,
1387
1402
(
2001
).
57.
M. G.
Kozodaev
,
A. G.
Chernikova
,
E. V.
Korostylev
,
M. H.
Park
,
R. R.
Khakimov
,
C. S.
Hwang
, and
A. M.
Markeev
,
J. Appl. Phys.
125
,
034101
(
2019
).
58.
E. J.
Sharp
and
L. E.
Garn
,
J. Appl. Phys.
53
,
8980
8987
(
1982
).
59.
S. W.
Smith
,
M. D.
Henry
,
M. T.
Brumbach
,
M. A.
Rodriguez
, and
J. F.
Ihlefeld
,
Appl. Phys. Lett.
113
,
182904
(
2018
).
60.
J. A.
Rodriguez
,
K.
Remack
,
K.
Boku
,
K. R.
Udayakumar
,
S.
Aggarwal
,
S. R.
Summerfelt
,
F. G.
Celii
,
S.
Martin
,
L.
Hall
,
K.
Taylor
,
T.
Moise
,
H.
McAdams
,
J.
McPherson
,
R.
Bailey
,
G.
Fox
, and
M.
Depner
,
IEEE Trans. Device Mater. Reliab.
4
,
436
449
(
2004
).
61.
D. V.
Taylor
and
D.
Damjanovic
,
J. Appl. Phys.
82
,
1973
1975
(
1997
).
62.
D.
Damjanovic
,
Phys. Rev. B
55
,
R649
R652
(
1997
).
63.
T.
Rojac
,
M.
Kosec
,
B.
Budic
,
N.
Setter
, and
D.
Damjanovic
,
J. Appl. Phys.
108
,
074107
(
2010
).
64.
D. R.
Islamov
,
T. M.
Zalyalov
,
O. M.
Orlov
,
V. A.
Gritsenko
, and
G. Y.
Krasnikov
,
Appl. Phys. Lett.
117
,
162901
(
2020
).
65.
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
, and
G. Y.
Krasnikov
,
Acta Mater.
166
,
47
55
(
2019
).
66.
S.
Zafar
,
H.
Jagannathan
,
L. F.
Edge
, and
D.
Gupta
,
Appl. Phys. Lett.
98
,
152903
(
2011
).
67.
J.
Seidel
,
L. W.
Martin
,
Q.
He
,
Q.
Zhan
,
Y.-H.
Chu
,
A.
Rother
,
M. E.
Hawkridge
,
P.
Maksymovych
,
P.
Yu
,
M.
Gajek
,
N.
Balke
,
S. V.
Kalinin
,
S.
Gemming
,
F.
Wang
,
G.
Catalan
,
J. F.
Scott
,
N. A.
Spaldin
,
J.
Orenstein
, and
R.
Ramesh
,
Nat. Mater.
8
,
229
234
(
2009
).
68.
T.
Rojac
,
A.
Bencan
,
G.
Drazic
,
N.
Sakamoto
,
H.
Ursic
,
B.
Jancar
,
G.
Tavcar
,
M.
Makarovic
,
J.
Walker
,
B.
Malic
, and
D.
Damjanovic
,
Nat. Mater.
16
,
322
327
(
2017
).
69.
F.
Huang
,
X.
Chen
,
X.
Liang
,
J.
Qin
,
Y.
Zhang
,
T.
Huang
,
Z.
Wang
,
B.
Peng
,
P.
Zhou
,
H.
Lu
,
L.
Zhang
,
L.
Deng
,
M.
Liu
,
Q.
Liu
,
H.
Tian
, and
L.
Bi
,
Phys. Chem. Chem. Phys.
19
,
3486
3497
(
2017
).

Supplementary Material