The emerging field of ferroelectric hafnium zirconium oxide has garnered increased attention recently for its wide array of applications from nonvolatile memory and transistor devices to nanoelectromechanical transducers. Atomic layer deposition is one of the preferred techniques for the fabrication of hafnium zirconium oxide thin films, with a standard choice of oxidizer being either O3 or H2O. In this study, we explore various oxidizing conditions and report on the in situ treatment of hydrogen plasma after every atomic layer during the deposition of hafnium zirconium oxide to increase the virgin state polarization. Three different oxidization methods were utilized during the fabrication of the Hf0.5Zr0.5O2 films: H2O, O2 plasma, and O2 plasma followed by H2 plasma. The 10 and 8 nm thick films oxidized with only O2 plasma result in initially anti-ferroelectric films. Comparatively, the addition of H2 plasma after every O2 plasma step results in films with strong ferroelectric behavior. Peak shifting of the GIXRD pattern suggests that the sequential O2-H2 plasma films tend more to the orthorhombic phase as compared to the O2 plasma and H2O oxidized films.

Ferroelectric hafnium zirconium oxide is quickly establishing itself as a candidate for emerging thin film devices such as ferroelectric tunnel junctions1–3 and ferroelectric random access memory3,4 for non-volatile memory applications, ferroelectric field effect transistors for semiconductor devices,3,5,6 and even wireless communication devices through recently demonstrated nanoelectromechanical resonators.7,8 The field of hafnium oxide based ferroelectrics has been explored extensively over the past decade, but began with the discovery of a hysteresis and remanent polarization in silicon doped hafnium oxide.9 Varying the silicon doping has been shown to result in an array of films from ferroelectric to anti-ferroelectric.9–11 Similarly, in the hafnia-zirconia solid solution, varying the ratio of HfO2 to ZrO2 produces a wide range of ferroelectric and anti-ferroelectric films.12,13 Aside from dopants alone, ferroelectricity was shown to be affected by a number of factors including deposition temperature,14 post-metallization annealing,15–17 oxidation pulse time during atomic layer depositions (ALD),18–20 and the oxygen flow rate during sputter depositions.21,22 In this work, the application of O2 plasma, sequential O2-H2 plasma, and H2O oxidation methods and the effects that these methods can have on the ferroelectricity of 1:1 hafnium zirconium oxide (HZO) films are presented.

Atomic layer deposition of the HZO films was carried out at 200 °C using tetrakis(dimethylamido)hafnium(IV) (TDMAH) and tetrakis(dimethylamido)zirconium(IV) (TDMAZ). Three different oxidizing treatments were utilized: H2O-oxidized (labeled H2O films throughout), O2 plasma-oxidized (labeled O2 plasma films), and O2 plasma-oxidized followed by H2 plasma-treated (labeled sequential O2-H2 plasma films). For the sequential O2-H2 plasma films, O2 plasma was first applied after the precursor pulse and then followed by H2 plasma. Using each of the three oxidizing conditions, 4.5, 6, 8, and 10 nm thick films were grown by employing the tiered deposition method described previously.23 From ellipsometry measurements of the ∼10 nm thick HZO films grown directly on silicon, the growth rate for H2O oxidized films was determined to be 1 Å per cycle, whereas the growth rate for both O2 and sequential O2-H2 plasma films was slightly lower at 0.8 Å per cycle.

Bottom electrodes of 10 nm thick TiN were fabricated on p+ silicon using tetrakis(dimethylamido)titanium(IV) (TDMAT) and nitrogen plasma. Top electrodes of TiN were similarly deposited, sandwiching the above HZO films to form a metal-ferroelectric-metal (MFM) film stack. MFM capacitors were fabricated by sputtering 50 nm thick platinum which, following lift-off, was used as a hard mask to etch the exposed TiN in H2O2 heated to 65 °C, creating device areas ranging from 1600 to 14 400 μm2. A rapid thermal anneal (RTA) was carried out at 500 °C for 20 s in N2. Hysteresis and positive-up-negative-down (PUND) tests24 were carried out with an Agilent 33500B waveform generator and a Tektronix TDS5104B oscilloscope. Hysteresis measurements were taken at 1 and 2 kHz, while PUND tests were executed with 1–10–1 μs pulses. Wake-up cycling was performed at 100 kHz with a bipolar square wave. GIXRD patterns were collected with a PANalytical XPert MRD at a grazing incidence angle of 0.5°. Scans had a step size of 0.03° and a count time of 2 s at each step. Narrow scans from 28.5° to 32.5° were collected three times and averaged, with a phi rotation performed between scans to obtain a more representative average of the polycrystalline structure.

Hysteresis results shown in Fig. 1 demonstrate the effect of the sequential O2-H2 plasma on the remanent polarization (Pr). For film thicknesses of 10, 8, and 6 nm, the sequential O2-H2 plasma films have the highest virgin state Pr of 18, 16, and 12 μC/cm2 compared to 12, 9, and 2 μC/cm2 for H2O films and <1 μC/cm2 for O2 plasma films at 10 and 8 nm (where the films are anti-ferroelectric) and 0 μC/cm2 at 6 and 4.5 nm (where the films are paraelectric). Further still, in the virgin state, the sequential O2-H2 plasma film at 4.5 nm displays anti-ferroelectricity whereas the H2O and O2 plasma films at this thickness are paraelectric.

FIG. 1.

Hysteresis curves of the HZO films prepared using H2O, O2 plasma, and sequential O2-H2 plasma oxidizing conditions during ALD. The hysteresis curves were collected in the virgin state before applying any electric field cycling.

FIG. 1.

Hysteresis curves of the HZO films prepared using H2O, O2 plasma, and sequential O2-H2 plasma oxidizing conditions during ALD. The hysteresis curves were collected in the virgin state before applying any electric field cycling.

Close modal

To study the evolution of the polarization with repetitive switching cycles, endurance tests of the films were carried out via PUND pulses. Figure 2 shows the net switched polarization obtained from the integral of the positive-minus-up pulses for the films across the four thicknesses deposited. As with the hysteresis results, the large virgin state polarization is readily apparent for the sequential O2-H2 plasma films, whereas the H2O films require wake-up cycling of 106 switching cycles to match the switched polarization of 34 μC/cm2 attained by the sequential O2-H2 plasma film at 10 and 8 nm. At 4.5 nm, only the sequential O2-H2 plasma film showed a switched polarization, appearing initially anti-ferroelectric but waking up to a switched polarization of 7.6 μC/cm2 after 107 switching cycles. The O2 plasma films are initially anti-ferroelectric at 10 and 8 nm but eventually develop non-zero switched polarizations of 12 and 7.4 μC/cm2 after 108 switching cycles. The onset of fatigue is seen in the sequential O2-H2 plasma films after 106 cycles; however, the rate of fatigue is reduced, allowing for a switched polarization to be maintained above 30 μC/cm2 for the 10, 8, and 6 nm films after 108 switching cycles. Interestingly, at 10 and 8 nm, the O2 plasma films did not show any fatigue and continued to wake-up, up to 108 switching cycles.

FIG. 2.

Endurance testing of the H2O (orange triangles), O2 (green squares), and sequential O2-H2 (blue circles) plasma films. The integrated positive minus up values are reported.

FIG. 2.

Endurance testing of the H2O (orange triangles), O2 (green squares), and sequential O2-H2 (blue circles) plasma films. The integrated positive minus up values are reported.

Close modal

The differences in switched polarization and wake-up between the films can be illustrated by examining the raw voltage responses from the PUND pulses. Figure 3(a) shows the applied and measured voltages for the positive and up pulses of the 10 nm films in the virgin state. Both the H2O and sequential O2-H2 plasma films have similar strong voltage responses for the first positive pulse (P), indicating a high polarization saturation. However, the voltage response for the up pulse (U) of the H2O film nearly matches that of the previous positive pulse (P), whereas for the sequential O2-H2 plasma film, the voltage response for the up pulse (U) is greatly reduced. Hence, the net switched polarization [(P-Pa)–(U-Ua)],24 which is the applicable polarization required for memory applications, is low for the H2O film but high for the sequential O2-H2 plasma film. This implies that the H2O film has significant reverse switching in the virgin state. That is, the maximum switching polarization remains high but once the external voltage is removed a large portion of this polarization reverses, reducing the overall remanence. This reverse switching is also illustrated by the disparity seen between the hysteresis and switched polarization measurements. The 10 nm H2O film, for example, has a Pr from Fig. 1 of 12 μC/cm2, which equates to a 2Pr of 24 μC/cm2. However, from the PUND tests in Fig. 2, the virgin state switched polarization of the 10 nm H2O film is only 10 μC/cm2, less than half of the expected value. This is due in part to the large reverse switching which occurs in H2O films during the PUND tests, since the delay time used between PUND pulses is 1 s. Hysteresis tests operate continuously and are often not able to capture the full extent of reverse switching, leading to differences in 2Pr values and switched polarizations. For the sequential O2-H2 plasma film, from which the PUND pulses show significantly less reverse switching, this difference is minimal with the virgin state 2Pr for the 10 nm equal to 36 μC/cm2, only 9 μC/cm2 higher than the virgin state switched polarization of 27 μC/cm2.

FIG. 3.

(a) Raw voltage response of the 10 nm films for the positive and up pulses in the virgin state with insets after wake-up following 106 switching cycles. (b) Dynamic switching current of the 10 nm films before and after 106 wake-upswitching cycles.

FIG. 3.

(a) Raw voltage response of the 10 nm films for the positive and up pulses in the virgin state with insets after wake-up following 106 switching cycles. (b) Dynamic switching current of the 10 nm films before and after 106 wake-upswitching cycles.

Close modal

The insets in Fig. 3(a) show the voltage response after 106 wake-up cycles. After cycling, the up pulse (U) response for the H2O film is reduced, resulting in the higher polarization seen for the 10 nm film in Fig. 2 after 106 switching cycles. For the virgin O2 plasma film, both the positive (P) and up (U) pulses overlap, giving a net positive-minus-up switched polarization of 0 μC/cm2, whereas after wake-up, a slight difference is seen between the positive (P) and up (U) pulses, correlating to the small 5 μC/cm2 switched polarization observed in Fig. 2 at 106 switching cycles. The dynamic switching currents in Fig. 3(b) further support the PUND results as the 10 nm H2O film initially has two switching peaks per polarity which merge after cycling. This indicates the presence of an internal bias in the virgin state which reduces with cycling, leading a reduction of reverse switching and an increase in the switched polarization.25 The sequential O2-H2 plasma film has one merged peak per polarity which improves after wake-up, correlating to its high virgin state and woken up switched polarizations. The O2 plasma film has two small peaks per polarity which move toward each other after cycling but never merge into a single strong peak per polarity.

To help interpret the electrical results with crystal phase analysis, the GIXRD patterns collected for the 10 nm thick HZO films are shown in Fig. 4(a). The peak locations and intensities across the 10 nm H2O, O2 plasma, and sequential O2-H2 plasma films are strikingly similar. The highest intensity peak observed at a 2θ between 29° and 32° is the typical peak examined to determine the presence of the orthorhombic and tetragonal phases; however, distinguishing between the two phases is challenging due to their similar structures and broadening induced from finite size effects.15 Indeed, a Rietveld refinement was not performed due to the unavoidable challenges incurred when measuring thin films (<11 nm) under grazing incidence, such as anisotropic and asymmetric peak shifting and peak broadening, for which no Rietveld model can adequately describe. Additionally, modest to strong crystallographic textures cannot be appropriately modeled using such data because the measured patterns do not contain enough Bragg peaks to refine all the coefficients in a texture model.

FIG. 4.

(a) GIXRD patterns of the H2O, O2 plasma, and O2-H2 plasma films. (b)–(d) shows the average of three scans from 28.5° to 32.5° for the three films.

FIG. 4.

(a) GIXRD patterns of the H2O, O2 plasma, and O2-H2 plasma films. (b)–(d) shows the average of three scans from 28.5° to 32.5° for the three films.

Close modal

One definitive phase result which can be made for all films is the absence of any peaks associated with the monoclinic phase near 28.5° and 31.6°. Previous studies, which explored the effect of oxygen concentration on the ferroelectric response and GIXRD pattern of hafnium based films, have shown a suppression of the monoclinic phase by lowering the oxygen content. This was achieved by either decreasing the oxygen flow rate in PVD depositions21,22 or in ALD by under pulsing the oxidizers such as H2O or O3.19,20

With no monoclinic phase observed in the O2 plasma film, its suppression in the H2O and sequential O2-H2 plasma films as the cause for the improved polarization are ruled out. This notion is supported by the hysteresis results in Fig. 1, as the presence of the monoclinic phase in the O2 plasma films would lead to a reduced polarization, but not a pinching of the hysteresis loop. On the contrary, the O2 plasma films in the virgin state exhibit strong anti-ferroelectricity associated with the tetragonal phase, which transforms to the orthorhombic phase with a high enough applied field, and then back to tetragonal once the field is removed.26 The virgin state sequential O2-H2 plasma film, however, exhibits strong ferroelectricity attributed to the orthorhombic phase in the as fabricated state. The H2O films fall in-between with a ferroelectric but pinched hysteresis, indicative of a mixture of anti-ferroelectric and ferroelectric properties, which could be realized through a mixture of tetragonal and orthorhombic crystal phases.

To examine the 2θ peak between 29° and 32°, the GIXRD patterns were re-measured for the 10 nm thick films from 28.5° to 32.5° three times and averaged [Figs. 4(b)–4(d)]. Both the O2 plasma and H2O films have similar peak locations with 2θ values of 30.621° ± 0.007° and 30.634° ± 0.007°, respectively. (The error presented is the worst case error encountered while fitting. More information on the scans and fits can be found in the supplementary material.) The sequential O2-H2 plasma film, however, has a slightly lower peak position, with a 2θ of 30.587° ± 0.007°. The lower 2θ position for the sequential O2-H2 plasma film could be indicative of a shift toward the orthorhombic phase, which in thin film HZO has a 2θ location of 30.4°.15 The O2 plasma and H2O films, with the higher 2θ position over the sequential O2-H2 plasma film, could tend more to the tetragonal phase, which is located at a 2θ of 30.8°.15 Another factor to consider is the FWHM of peaks, which is 0.877° ± 0.007° for the O2 plasma film, 0.919° ± 0.007° for the sequential O2-H2 plasma film, and 0.961° ± 0.007° for the H2O film. The O2 plasma film has the narrowest peak centered around 30.621° ± 0.007° and the lowest virgin state polarization, both the features are consistent with the tetragonal phase. The sequential O2-H2 plasma film has the highest virgin state polarization and a peak centered around 30.587° ± 0.007°. These features align with a shift to the orthorhombic phase. Finally, the H2O film has the broadest peak centered around 30.634° ± 0.007° and a virgin state polarization higher than the O2 plasma films but lower than the sequential O2-H2 plasma film. The peak location itself is more consistent with the tetragonal phase, but the peak broadening seen in the H2O film could be caused by an overlap of multiple underlying peaks, or a mix of orthorhombic and tetragonal crystal phases, which is consistent with the electrical results of a mix of anti-ferroelectric and ferroelectric behavior.

The major electrical difference between the H2O and the sequential O2-H2 plasma films lies in the often studied wake-up effect. Both films achieve comparable polarizations after cycling, but the sequential O2-H2 plasma films exhibit less of a wake-up effect, with the 8 and 10 nm films having switched polarizations over 20 μC/cm2 before any electric field cycling. The wake-up effect in ferroelectric hafnium oxide is still an area of great academic interest, but a variety of experiments and papers have been published elucidating some of the fundamental mechanisms. Pešić et al. examined Sr:HfO2 and attributed wake-up to a redistribution of defects which reduced the built-in field and led to phase transformations at the interfaces and bulk.27 Lomenzo et al.28 and Kim et al.29 also suggested a phase transformation from tetragonal to orthorhombic with electric field cycling. Recently, Chouprik et al. utilized piezoresponse force microscopy (PFM) to examine domain structure changes per electric field cycle. They concluded wake-up was due to the redistribution of oxygen vacancies which reduced the internal bias, leading to a reduction of “anomalous” and static domains, and an increase in normal domains which comprise the remanent polarization.30 The application of H2 plasma has the possibility of inducing point defects through oxygen vacancies generation. Indeed, H2 plasma is used in ALD processes for the deposition of single-element materials31 and has been experimentally shown for the case of Al2O3 and IrO2 to reduce the deposited oxide.32,33 Additionally, in an experiment on resistive switching in hafnium oxide, it was shown that hydrogen plasma treatment during the top TiN deposition partially reduced the underlying hafnium oxide layer and generated oxygen vacancies.34 These potential defects in the sequential O2-H2 plasma films, already distributed throughout the film due to the layer by layer deposition, could increase the virgin state switched polarization either through a tetragonal to orthorhombic phase transition or by a reduction of “anomalous” and static domains and an increase in normal domains. This is in contrast to the H2O films, which require upwards of 106 electric field cycles of wake-up to match the switched polarizations of the sequential O2-H2 plasma films. Thus, the application of the H2 plasma seems to generate an as-fabricated wake-up effect over other traditional oxidation methods.

It should not be ignored that the sequential O2-H2 plasma films, even with large virgin state switched polarizations, still exhibit further wake-up with electric field cycling. This effect is an unavoidable consequence from the fabrication of the films, as oxygen scavenging by the electrodes during ALD28,35 or the post-deposition anneal36 would create defect clusters at the interfaces.37 Reduction of the defect clusters with electric field cycling would lead to an increase in the remanent polarization by converting the dead layers at the interface to ferroelectric switching layers.38 As the films are scaled down to 4 and 6 nm, the interfaces become the majority of the film; thus, the wake-up effect would become more pronounced as observed. For the O2 plasma films, which exhibit anti-ferroelectricity and a GIXRD pattern consistent with the tetragonal phase, the wake-up effect would then be mostly due to changes at the interface, which explains the slow growth and low magnitude of the switched polarization as a function of switching cycles.

In situ H2 plasma during the ALD of ferroelectric HZO films provides a technique to minimize the wake-up effect and increase the virgin state polarization and may present a path forward for defect engineering. The sequential O2-H2 plasma films had large virgin state switched polarizations of 27 and 21 μC/cm2 for the 10 and 8 nm thick films and a strong ferroelectric thickness scalability with the 10, 8, and 6 nm films exhibiting switched polarizations of ∼34 μC/cm2 after 106 switching cycles. Among the other oxidation methods explored, the O2 plasma films were initially anti-ferroelectric and required 108 switching cycles to open the hysteresis loop. The H2O films in the virgin state showed a pinched hysteresis but could be cycled to match the switched polarization of the sequential O2-H2 plasma films at 10 and 8 nm after 106 switching cycles. The peak shifting observed in the GIXRD pattern from 30.621° ± 0.007° for the O2 plasma film to 30.587° ± 0.007° for the sequential O2-H2 plasma films suggests that the addition of the in situ H2 plasma step during ALD helps to stabilize the orthorhombic phase. The application of the H2 plasma during the deposition may partially reduce the previously deposited oxide, generating oxygen vacancy point defects and increasing the virgin state switched polarization through a tetragonal to orthorhombic phase transition, or by altering the internal bias and contributing to an increase in normal domains.

See the supplementary material for full plots and fits of the GIXRD scans.

This work was supported in part by NSF Grant No. ECCS 1610387. The authors acknowledge the use of the Research Service Centers at the Herbert Wertheim College of Engineering at the University of Florida and would like to thank Dr. Brent Gila, Andres Trucco, and Kristy Schepker for their invaluable insights and support.

1.
Y.
Goh
and
S.
Jeon
,
Appl. Phys. Lett.
113
,
052905
(
2018
).
2.
B.
Max
,
M.
Hoffmann
,
S.
Slesazeck
, and
T.
Mikolajick
, in
48th European Solid-State Device Research Conference
(
2018
), p.
142
.
3.
M. H.
Park
,
Y. H.
Lee
,
T.
Mikolajick
,
U.
Schroeder
, and
C. S.
Hwang
,
MRS Commun.
8
,
795
(
2018
).
4.
S.
Mueller
,
S. R.
Summerfelt
,
J.
Müller
,
U.
Schroeder
, and
T.
Mikolajick
,
IEEE Electron Device Lett.
33
,
1300
(
2012
).
5.
T.
Ali
,
P.
Polakowski
,
S.
Riedel
,
T.
Buttner
,
T.
Kampfe
,
M.
Rudolph
,
B.
Patzold
,
K.
Seidel
,
D.
Lohr
,
R.
Hoffmann
,
M.
Czernohorsky
,
K.
Kuhnel
,
P.
Steinke
,
J.
Calvo
,
K.
Zimmermann
, and
J.
Muller
,
IEEE Trans. Electron Devices
65
,
3769
(
2018
).
6.
B.
Zeng
,
W.
Xiao
,
J.
Liao
,
H.
Liu
,
M.
Liao
,
Q.
Peng
,
S.
Zheng
, and
Y.
Zhou
,
IEEE Electron Device Lett.
39
,
1508
(
2018
).
7.
M.
Ghatge
,
G.
Walters
,
T.
Nishida
, and
R.
Tabrizian
, in
IEEE Internationl Electron Devices Meeting
(
IEEE
,
2018
), pp.
4.6.1
4.6.4
.
8.
M.
Ghatge
,
G.
Walters
,
T.
Nishida
, and
R.
Tabrizian
,
Nat. Electron.
2
,
506
(
2019
).
9.
T. S.
Böscke
,
J.
Müller
,
D.
Bräuhaus
,
U.
Schröder
, and
U.
Böttger
,
Appl. Phys. Lett.
99
,
102903
(
2011
).
10.
P. D.
Lomenzo
,
Q.
Takmeel
,
C.
Zhou
,
Y.
Liu
,
C. M.
Fancher
,
J. L.
Jones
,
S.
Moghaddam
, and
T.
Nishida
,
Appl. Phys. Lett.
105
,
072906
(
2014
).
11.
P. D.
Lomenzo
,
Q.
Takmeel
,
S.
Moghaddam
, and
T.
Nishida
,
Thin Solid Films
615
,
139
(
2016
).
12.
J.
Muller
,
T. S.
Boscke
,
U.
Schroder
,
S.
Mueller
,
D.
Brauhaus
,
U.
Bottger
,
L.
Frey
, and
T.
Mikolajick
,
Nano Lett
12
,
4318
(
2012
).
13.
M. H.
Park
,
H. J.
Kim
,
Y. J.
Kim
,
Y. H.
Lee
,
T.
Moon
,
K.
Do Kim
,
S. D.
Hyun
,
F.
Fengler
,
U.
Schroeder
, and
C. S.
Hwang
,
ACS Appl. Mater. Interfaces
8
,
15466
(
2016
).
14.
K. D.
Kim
,
M. H.
Park
,
H. J.
Kim
,
Y. J.
Kim
,
T.
Moon
,
Y. H.
Lee
,
S. D.
Hyun
,
T.
Gwon
, and
C. S.
Hwang
,
J. Mater. Chem. C
4
,
6864
(
2016
).
15.
M.
Hyuk Park
,
H.
Joon Kim
,
Y.
Jin Kim
,
W.
Lee
,
T.
Moon
, and
C.
Seong Hwang
,
Appl. Phys. Lett.
102
,
242905
(
2013
).
16.
M. H.
Park
,
C.-C.
Chung
,
T.
Schenk
,
C.
Richter
,
K.
Opsomer
,
C.
Detavernier
,
C.
Adelmann
,
J. L.
Jones
,
T.
Mikolajick
, and
U.
Schroeder
,
Adv. Electron. Mater.
4
,
1800091
(
2018
).
17.
A.
Shekhawat
,
G.
Walters
,
C.-C.
Chung
,
R.
Garcia
,
Y.
Liu
,
J.
Jones
,
T.
Nishida
, and
S.
Moghaddam
,
Thin Solid Films
677
,
142
(
2019
).
18.
A.
Pal
,
V. K.
Narasimhan
,
S.
Weeks
,
K.
Littau
,
D.
Pramanik
, and
T.
Chiang
,
Appl. Phys. Lett.
110
,
022903
(
2017
).
19.
J.-D.
Luo
,
H.-X.
Zhang
,
Z.-Y.
Wang
,
S.-S.
Gu
,
Y.-T.
Yeh
,
H.-T.
Chung
,
K.-C.
Chuang
,
C.-Y.
Liao
,
W.-S.
Li
,
Y.-S.
Li
,
K.-S.
Li
,
M.-H.
Lee
, and
H.-C.
Cheng
,
Jpn. J. Appl. Phys., Part 1
58
,
SDDE07
(
2019
).
20.
U.
Schroeder
,
M.
Materano
,
T.
Mittmann
,
P. D.
Lomenzo
,
T.
Mikolajick
, and
A.
Toriumi
,
Jpn. J. Appl. Phys., Part 1
58
,
SL0801
(
2019
).
21.
Y. H.
Lee
,
H. J.
Kim
,
T.
Moon
,
K.
Do Kim
,
S. D.
Hyun
,
H. W.
Park
,
Y.
Bin Lee
,
M. H.
Park
, and
C. S.
Hwang
,
Nanotechnology
28
,
305703
(
2017
).
22.
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
).
23.
G.
Walters
,
A.
Shekhawat
,
N. G.
Rudawski
,
S.
Moghaddam
, and
T.
Nishida
,
Appl. Phys. Lett.
112
,
192901
(
2018
).
24.
A. G.
Acosta
,
J.
Rodriguez
,
B.
Obradovic
,
S.
Summerfelt
,
T.
San
,
K.
Green
,
T.
Moise
, and
S.
Krishnan
,
in IEEE International Reliability Physics Symposium
(
2010
), p.
689
.
25.
T.
Schenk
,
M.
Hoffmann
,
J.
Ocker
,
M.
Pesic
,
T.
Mikolajick
, and
U.
Schroeder
,
ACS Appl. Mater. Interfaces
7
,
20224
(
2015
).
26.
S. E.
Reyes-Lillo
,
K. F.
Garrity
, and
K. M.
Rabe
,
Phys. Rev. B
90
,
140103
(
2014
).
27.
M.
Pesic
,
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
(
2016
).
28.
P. D.
Lomenzo
,
Q.
Takmeel
,
C.
Zhou
,
C. M.
Fancher
,
E.
Lambers
,
N. G.
Rudawski
,
J. L.
Jones
,
S.
Moghaddam
, and
T.
Nishida
,
J. Appl. Phys.
117
,
134105
(
2015
).
29.
H. J.
Kim
,
M. H.
Park
,
Y. J.
Kim
,
Y. H.
Lee
,
T.
Moon
,
K.
Do Kim
,
S. D.
Hyun
, and
C. S.
Hwang
,
Nanoscale
8
,
1383
(
2016
).
30.
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
).
31.
S. M.
George
,
Chem. Rev.
110
,
111
(
2010
).
32.
M.
Morales-Masis
,
L.
Ding
,
F.
Dauzou
,
Q.
Jeangros
,
A.
Hessler-Wyser
,
S.
Nicolay
, and
C.
Ballif
,
APL Mater.
2
,
096113
(
2014
).
33.
M.
Mattinen
,
J.
Hämäläinen
,
M.
Vehkamäki
,
M. J.
Heikkilä
,
K.
Mizohata
,
P.
Jalkanen
,
J.
Räisänen
,
M.
Ritala
, and
M.
Leskelä
,
J. Phys. Chem. C
120
,
15235
(
2016
).
34.
Y. Y.
Chen
,
L.
Goux
,
J.
Swerts
,
M.
Toeller
,
C.
Adelmann
,
J.
Kittl
,
M.
Jurczak
,
G.
Groeseneken
, and
D. J.
Wouters
,
IEEE Electron Device Lett.
33
,
483
(
2012
).
35.
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
).
36.
A.
Kerber
and
E. A.
Cartier
,
IEEE Trans. Device Mater. Reliab.
9
,
147
(
2009
).
37.
S.
Starschich
,
S.
Menzel
, and
U.
Böttger
,
Appl. Phys. Lett.
108
,
032903
(
2016
).
38.
E. D.
Grimley
,
T.
Schenk
,
X.
Sang
,
M.
Pešić
,
U.
Schroeder
,
T.
Mikolajick
, and
J. M.
LeBeau
,
Adv. Electron. Mater.
2
,
1600173
(
2016
).

Supplementary Material