We report on the soft x-ray absorption spectroscopy investigation of thin film capacitors using a modified total electron yield detection mode. This mode utilizes two ammeters instead of one as commonly employed in the classical total electron yield scheme to measure photocurrents of devices under soft x-ray irradiation. The advantage of this configuration over the surface sensitive classical total electron yield mode is that it can provide information from buried layers and interfaces up to a thickness equal to the penetration depth of soft x-rays. The method can be easily adapted to existing synchrotron end stations. We investigate dielectric capacitors with dissimilar electrodes to assess the feasibility of the modified total electron yield method. Furthermore, in operando soft x-ray absorption spectroscopy measurements are performed on ferroelectric capacitors under bias and using two ammeters. The experimental results are discussed in terms of the external and internal photoemission processes and their distribution in thin film capacitors under an external bias condition. The proposed detection method opens the way to perform electronic and chemical state analyses of the buried interfaces and layers in various devices like multiferroic tunnel junctions, memristive devices, etc., during operation under an applied bias.

With the advent of brilliant sources of synchrotron radiation, x-ray absorption spectroscopy (XAS) has become a widely used tool to probe electronic and structural properties of materials.1–4 Specifically, soft x-ray absorption spectroscopy (SXAS) with photon energies below 2 keV is a unique and powerful technique to study the electronic structure of conventional metals, semiconductors, and insulators5,6 and more advanced quantum materials such as topological insulators or highly correlated electron materials7 as well as interfaces and devices.8,9 By applying high-intensity soft x-ray radiation of tunable energy, SXAS measures the electron transition probability between an atomic core level and one of the unoccupied electronic states, while complying with dipole selection rules. The appropriate choice of photon energies for excitation of selected core levels is one of the key features of SXAS and enables probing of element specific density of states.10–13 

Three methods are available to perform SXAS: transmission mode, fluorescence yield detection, and electron yield detection.14,15 In the transmission mode, the x-ray intensity is measured before and after transmission through the sample. The other two modes can be performed as total and partial yield detection depending on whether all photons or electrons are measured or only those in a specific energy window. The total electron yield (TEY) mode is characterized by a simple experimental setup using an ammeter [Fig. 1(a)]. The total fluorescence yield method16 (TFY) measures the photon flux from the sample [Fig. 1(a)] and, therefore, requires a more sophisticated photodetector setup in comparison to the TEY detection. Due to a large photon attenuation length of typically over 50 nm, TFY enables the monitoring of bulk properties. In contrast, the nondispersive TEY method measures the neutralization current that flows to the sample due to electron emission at the irradiated sample surface. The short mean probing depth of photoelectrons of typically less than 5 nm coming from the sample surface makes the TEY mode a more surface sensitive technique.17 Often both TEY and TFY modes are applied simultaneously.18 

FIG. 1.

(a) Standard SXAS setup for TFY and TEY methods. (b) Schematic of the TC-TEY method using one ammeter in the top (ITop) and another in the bottom (IBot) wiring line and showing the electron flow for photon energies matching those of core level resonances in the tracer layers of the top and bottom electrodes. The gray arrows indicate the compensating electron flow from the top and bottom wires. The optional voltage source UBias is used for in operando measurements with bias potentials. Contributions to photocurrent: “1” external and internal and “2” internal photoelectric effect. Figure adapted from Ref. 29.

FIG. 1.

(a) Standard SXAS setup for TFY and TEY methods. (b) Schematic of the TC-TEY method using one ammeter in the top (ITop) and another in the bottom (IBot) wiring line and showing the electron flow for photon energies matching those of core level resonances in the tracer layers of the top and bottom electrodes. The gray arrows indicate the compensating electron flow from the top and bottom wires. The optional voltage source UBias is used for in operando measurements with bias potentials. Contributions to photocurrent: “1” external and internal and “2” internal photoelectric effect. Figure adapted from Ref. 29.

Close modal

Here, we report a modified TEY method to experimentally probe buried layers and interfaces in planar capacitors. In the two-channel TEY (TC-TEY) mode, a second ammeter is affixed to the bottom electrode of the device [Fig. 1(b)] and enables the detection of photoexcited electrons in deeply buried layers and interfaces. In principle, the TC-TEY extends the effective mean probing depth of synchrotron radiation, penetrating deeper into the sample. In this work, layers and interfaces as deep as 185 nm from the sample surface are explored. TC-TEY also includes an additional voltage source that enables in operando studies [Fig. 1(b)]. By applying a bias voltage, the sign and intensity of internal photocurrents can be tuned, offering the opportunity to enhance element specific information across the sample. In general terms, the method belongs to the class of in operando spectroscopy, which has been successfully applied to investigate electronic devices such as memristors or ferroelectric devices.19–21 We demonstrate the functionality of the TC-TEY mode on three different types of samples: one dielectric and two ferroelectric capacitors. In the first two samples, buried interfaces are probed, while in the last sample, ferroelectric properties are investigated. The method can readily be applied to investigate other devices such as solar cells, batteries, or memristive devices.9,19,21–23

In a conventional TEY setup, only the external photoelectric effect contributes to the measured signal. To overcome this limitation, the contributions of the internal photoelectric effects need to be measured. This is accomplished in the TC-TEY configuration. The basic experimental setup is sketched in Fig. 1(b). The device consists of a metal-insulator–metal capacitor-like structure, which optionally includes thin metallic tracer layers at the interfaces. The top and bottom electrodes are grounded via two ammeters. For in operando investigations, we use an additional voltage source connected between the top electrode and the top ammeter [Fig. 1(b)]. Soft x-rays are predominantly absorbed through excitations of core electrons to empty states above the Fermi level. Since the core electron excitation is element specific and strongly dependent on photon energy,2 electrons can be photoinjected into selected layers. Depending on the electron energy, interfacial barrier heights, and bias voltage, drift and diffusion currents can then flow through the insulating layer and be measured by the two ammeters [Fig. 1(b) and supplementary material Sec. 1].

All results presented here were obtained under ultrahigh vacuum conditions at the synchrotron radiation beamline BW3 [Deutsches Elektronen-Synchrotron (DESY), Hamburg], while preliminary experiments were conducted at the Advanced Light Source's beamline 8.0.1 (ALS, Lawrence Berkeley National Laboratory). The DESY undulator beamline was equipped with a high-flux SX-700 plane grating monochromator that allowed for a reasonably high energy resolution of ΔE =104·hν at the high photon flux required for the measurements (of more than 1012 photons/s).24 The absolute photon energy scale of the beamline was calibrated prior to the SXAS measurements by comparing core-level photoemission lines in a photoelectron spectroscopy setup. The x-rays absorption signals were acquired by tuning the energy of the incoming soft x-rays across an absorption edge while simultaneously measuring the currents in the top (ITop) and bottom (IBot) leads. Keithleys 6517 and 6485 were used as top and bottom ammeters, respectively. Here, we use the conventional current definition (opposite to the actual electron flow). The acquired signals were normalized to the incident photon flux as measured from a gold mesh located at the end of the beamline. The dielectric sample was deposited by sputtering while the ferroelectric samples were synthesized by pulsed laser deposition (supplementary material Sec. 2).

To demonstrate the concept of the TC-TEY method, a dielectric capacitor consisting of a 150 nm SiO2 film and referred to as sample SiO2-150 (supplementary material Sec. 2) is investigated, as sketched in Fig. 2(a). SXAS measurements of the SiO2-150 sample include photon energy scans from 451 to 474 eV and from 846 to 881 eV to be sensitive to the Ti and Ni tracer elements at the bottom and top electrodes, respectively. For these experiments, the voltage source is set to zero. The measurements acquired from the two ammeters are each divided into two main components. The first component is a constant background across the absorption edge and is determined by the current in the pre-edge region at 453 and 848 eV for the Ti and Ni spectral regions, respectively. The background currents retained from the Ti spectral region are IBot0.082 nA and ITop5.130 nA. The corresponding values in the Ni spectral region are IBot0.403 nA and ITop7.625 nA. By applying a linear background subtraction, we obtain the spectra shown in Fig. 2(c) exhibiting maxima of the Ti L3,2 absorption at 458.2 and 463.5 eV, respectively (the second component of the measured currents). The spectral shape and width indicate predominantly metallic Ti.25,26 The Ti L3,2 signal from the ammeter connected to the bottom electrode IBot is positive, whereas the spectrum from the ammeter connected to the top electrode ITop is negative. A negative current corresponds to electrons flowing from the bottom electrode to the top ammeter. The sum ITop+IBot [Fig. 2(c), gray crosses] is approximately zero.

FIG. 2.

(a) Electrical wiring and layer structure of sample SiO2-150. Bold font indicates the tracer elements from which the studied signals originate. (b) Schematic depiction of the currents resulting from core electron excitation at photon energies corresponding to the maxima of the Ti (left) and Ni (right) L3 absorption edges. Arrows indicate the direction of electron flow. (c) Ti and (d) Ni L3,2 TC-TEY x-ray absorption spectra, as measured with the two ammeters recording ITop and IBot. Baseline currents (BLC) and linear backgrounds were subtracted. The insets show the position of the chemical tracer layers. Figure adapted from Ref. 29.

FIG. 2.

(a) Electrical wiring and layer structure of sample SiO2-150. Bold font indicates the tracer elements from which the studied signals originate. (b) Schematic depiction of the currents resulting from core electron excitation at photon energies corresponding to the maxima of the Ti (left) and Ni (right) L3 absorption edges. Arrows indicate the direction of electron flow. (c) Ti and (d) Ni L3,2 TC-TEY x-ray absorption spectra, as measured with the two ammeters recording ITop and IBot. Baseline currents (BLC) and linear backgrounds were subtracted. The insets show the position of the chemical tracer layers. Figure adapted from Ref. 29.

Close modal

Applying the same procedure, the Ni L3,2 signal is obtained [Fig. 2(d)]. The top-electrode spectrum again presents two peaks: the first at 853.2 eV and the second at 870.7 eV. The signal from the bottom electrode is negative and 34% lower in amplitude than the top-electrode signal. Thus, the sum ITop+IBot [Fig. 2(d), gray crosses] shows a non-zero difference of the amplitudes of two signals.

The observed spectra can be explained by considering external and internal photoemission. In addition, drift and diffusion currents inside the capacitor structure under short circuit condition need to be taken into account.

The Ti-edge spectra from the Ti tracer layer located at the bottom electrode exhibit an increased absorption at photon energies corresponding to the Ti L3,2 peaks [Fig. 2(c)]. At these energies, Ti 2p core electrons are excited in the Ti layer [supplementary material Sec. 1]. Some of these electrons, having sufficient energy to be excited into the SiO2 conduction band, diffuse across the SiO2 dielectric barrier. In short circuit conditions, these charges are compensated externally via the two ammeters and result in ITop and IBot currents having the same amplitude and opposite signs.

In the case of the Ni signal [Fig. 2(d)], the excitation of Ni 2p core electrons at photon energies corresponding to the Ni L3,2 peaks takes place at the top electrode, and consequently, the direction of the external compensation current is reversed [Fig. 1(b) and supplementary material Sec. 1]. This explains the observed sign reversal in the ITop and IBot currents for the two cases. Moreover, the Ni tracer layer is part of the Au top electrode, and the inelastic mean free path for electrons with energies slightly above the threshold is in the range of 5–10 nm in Au.27,28 A finite number of these low-energy electrons can, therefore, reach the surface and contribute to external photoemission. This additional external photoemission contribution can, thus, lead to the observed imbalance in the amplitudes of ITop and IBot. These results demonstrate the potential of the TC-TEY method to probe core-level excitation in the selected buried layers.

In the following, two PbZr0.52Ti0.48O3 (PZT)-based ferroelectric capacitor structures, referred to as PZT-80 and PZT-150 (supplementary material Sec. 2), are investigated to demonstrate the in operando capability of SXAS in the TC-TEY mode. In these samples, SrRuO3 (SRO) is used as a bottom electrode, CoFe2O4 (CFO) as a tracer, and Au as a top electrode. Ni48–Cu52 is used as a tracer in PZT-80 only.

As PZT-80 contains chemical tracers in the top and bottom metallic layers, similar to Si-150, it is used first to specifically verify the enhanced depth sensitivity of TC-TEY for ferroelectric systems. The PZT-150 sample is then employed to study the influence of an electric field of up to ±20 kV/mm on the PZT layer and the TC-TEY spectra.

In PZT-80, schematized in Fig. 3(a), Co and Ni are used as chemical markers for the bottom and top electrodes, respectively. The corresponding normalized (background corrected) absorption signals reflecting the Co and Ni L3,2 edges are shown in Figs. 3(b) and 3(c). Currents measured by the top and bottom ammeters are again represented by blue and red curves, and the sum of both currents is plotted in gray.

FIG. 3.

(a) Electrical wiring and layer structure of sample PZT-80. Bold font indicates the tracer materials from which the studied signals originate. (b) Co and (c) Ni L3,2 TC-TEY x-ray absorption spectra, as measured by the two ammeters recording ITop and IBot. BLC and linear background were subtracted. The insets show the positions of the chemical tracer layers. Figure adapted from Ref. 29.

FIG. 3.

(a) Electrical wiring and layer structure of sample PZT-80. Bold font indicates the tracer materials from which the studied signals originate. (b) Co and (c) Ni L3,2 TC-TEY x-ray absorption spectra, as measured by the two ammeters recording ITop and IBot. BLC and linear background were subtracted. The insets show the positions of the chemical tracer layers. Figure adapted from Ref. 29.

Close modal

These results are similar to those obtained on SiO2-150, where the Co L3,2 and Ni L3,2 peaks can be observed in the spectra acquired with both ammeters and present equal intensities and opposite signs due to internal photocurrents. The additional external photocurrent observed in SiO2-150, expected to be of similar amplitude here, is within the noise level in PZT-80. The higher noise observed in PZT-80 is attributed to a higher background signal, caused by higher leakage. This demonstrates the capability of TC-TEY to study buried thin films in a capacitor-like structure with a ferroelectric insulating material and suitable chemical tracers.

The effect of an applied bias is investigated next on sample PZT-150, illustrated in Fig. 4(a). The ferroelectric nature of these capacitors as well as the integrity of the measured devices (for applying bias voltages) are demonstrated by performing polarization vs voltage (PV) hysteresis loop measurements at room temperature. The hysteresis loops are recorded at a frequency of 1 kHz under an excitation signal of triangular shape using a Radiant ferroelectric analyzer. Figure 4(b) presents a typical PV loop for the PZT capacitor of sample PZT-150. The impact of the bias potential applied between the two electrodes of sample PZT-150 on the spectroscopic measurements is depicted in Figs. 4(c) and 4(d). The Ti L3 and Ti L2 absorption edge regions are recorded to extract the response of the ferroelectric PZT layer. Since the bias potential influences both the specific SXAS signals and the baseline currents (BLCs), raw data are shown for different bias potentials in Figs. 4(c) and 4(d). Here, four prominent Ti peaks can be seen, which are related to the crystal field splitting of Ti4+ in an octahedral symmetry. The first two peaks belong to the L3 edge (t2g and eg), while the third and fourth peaks belong to the L2 edge (t2g and eg).

FIG. 4.

(a) Electrical wiring and layer structure of sample PZT-150 with a floating voltage source UBias. (b) Polarization vs electric field hysteresis loop. (c) and (d) Ti L3,2 x-ray absorption spectra (raw data) for different bias potentials, as measured by the bottom (IBot) and top (ITop) ammeter, respectively. BLC are indicated as dotted lines for UBias=±3 V. Figure adapted from Ref. 29.

FIG. 4.

(a) Electrical wiring and layer structure of sample PZT-150 with a floating voltage source UBias. (b) Polarization vs electric field hysteresis loop. (c) and (d) Ti L3,2 x-ray absorption spectra (raw data) for different bias potentials, as measured by the bottom (IBot) and top (ITop) ammeter, respectively. BLC are indicated as dotted lines for UBias=±3 V. Figure adapted from Ref. 29.

Close modal

The sample is poled prior to the measurements with a bias potential of −3.5 V. SXAS measurements are performed starting with a 0 V bias potential, which is then decreased down to −3 V. The subsequent measurement series contains the bias voltage region from +0.25 to +3 V. We can see that the magnitude of the absorption signals and the BLCs increases with increasing bias potential, and the signs of the currents measured by the top and bottom ammeters are opposite. The BLCs extracted at a photon energy of 453 eV and plotted against the bias potential are illustrated in Fig. 5(a).

FIG. 5.

Bias potential-dependent currents as extracted from Figs. 4(c) and 4(d). (a) BLC at a photon energy of 453 eV. (b) Mean current as averaged over the Ti L3,2 absorption spectra [Figs. 4(c) and 4(d), filled gray areas for UBias=±3 V]. Arrows at the top indicate the measurement sequence. The vertical arrow in (a) indicates the bias voltage at which the current measured at the bottom ammeter equals zero. Figure adapted from Ref. 29.

FIG. 5.

Bias potential-dependent currents as extracted from Figs. 4(c) and 4(d). (a) BLC at a photon energy of 453 eV. (b) Mean current as averaged over the Ti L3,2 absorption spectra [Figs. 4(c) and 4(d), filled gray areas for UBias=±3 V]. Arrows at the top indicate the measurement sequence. The vertical arrow in (a) indicates the bias voltage at which the current measured at the bottom ammeter equals zero. Figure adapted from Ref. 29.

Close modal

At a negative bias potential of −3 V, the BLC measured at the top ammeter (blue line) reaches a maximum before decreasing almost linearly with changing toward a positive bias. The BLC measured at the bottom ammeter shows the opposite trend [Fig. 5(a), red line]. The sum of both BLCs exhibits an almost constant value of about 6 nA [Fig. 5(a), gray line]. This constant current sum comes from the contribution of the external photoelectric effect of the Au top layer and is nearly equal to the external photocurrent measured in sample SiO2-150 at 455 eV. The top and bottom BLCs exhibit a mostly linear current–voltage characteristic that is shifted by an inner electric field to a negative bias potential of about −660 mV, as depicted in Fig. 5(a). Therefore, at a bias potential of −660 mV, the BLC from the bottom ammeter is zero, and the current flowing through the top ammeter neutralizes the external photocurrent. This inner electric field in the PZT films results from the ferroelectric polarization, different electrode materials, and the different photon absorption densities of the electrodes. At this point, the external bias cancels out the inner electric potential, and no internal current is flowing between the electrodes [Fig. 5(a)]. The inner electric potential, thus, has a value of 660 mV.

The average of the background corrected SXAS signals [see gray filled spectra in Figs. 4(c) and 4(d) as an example] is then plotted against the applied bias potential as shown in Fig. 5(b) for the top and bottom ammeters. In this way, the contribution of Ti 2p core electron photoemission only is extracted, and all other current contributions are mathematically subtracted. The averaged signal measured by the top ammeter at a −3 V bias potential is −2 nA and rises proportionally with increasing applied bias until it reaches 4 nA for a 3 V bias voltage. The averaged signal from the bottom ammeter behaves in the opposite way, decreasing from 2 to −4 nA in the same bias voltage range. This is confirmed by the sum of the two measured currents ITop+IBot, which depicts a nearly constant value of 0 nA, as presented in Fig. 5(b) (gray line). This result is expected, since the buried PZT film is not involved in external photoemission.

In conclusion, we demonstrated how a modified TEY method, the TC-TEY method, can be exploited to measure SXAS signals of films buried as deep as 185 nm using an additional ammeter and chemical tracers in capacitor-like patterned devices measured under an applied bias voltage. The TC-TEY method is explained in terms of drift and diffusion currents inside an insulator layer, which are generated by element-specific and layer-selective internal photoemission, and enables the determination of the potential barriers. The in operando capability of TC-TEY is exploited to probe a ferroelectric capacitor structure with inner electric fields under applied external bias potentials. An enhancement of the TC-TEY signal is obtained by applying an external bias voltage on capacitor-like structures during SXAS measurements in this TC-TEY configuration. This method is suitable for capacitor-like devices based on insulators and wide bandgap semiconductors as long as dielectric leakage remains lower than the photocurrent.

See the supplementary material for a schematic of the band diagram and more details regarding the devices' structure and fabrication.

We thank R. Ramesh for carefully reading the manuscript. We gratefully acknowledge the skillful assistance of the staff at the synchrotron radiation facility DESY (Hamburg). G.K. acknowledges the postdoctoral fellowship from the Alexander von Humboldt foundation. This research used resources of the Advanced Light Source, a U.S. DOE Office of Science User Facility under Contract No. DEAC02-05CH11231. This work was funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation)—Project ID: 434434223-SFB 1461.

The authors have no conflicts to disclose.

E.K. and A.P. contributed equally to this work.

The data that support the findings of this study are available from the corresponding author upon reasonable request.

1.
J.
Stöhr
,
J. Vac. Sci. Technol.
16
,
37
(
1979
).
2.
J.
Stöhr
, in
NEXAFS Spectroscopy
, Springer Series in Surface Sciences Vol.
25
, edited by
G.
Ertl
,
R.
Gomer
,
D. L.
Mills
, and
H. K. V.
Lotsch
(
Springer
,
Berlin/Heidelberg
,
1992
).
3.
A.
Erbil
,
G.
Cargill Iii
,
R.
Frahm
, and
R.
Boehme
,
Phys. Rev. B
37
,
2450
(
1988
).
4.
F. M.
de Groot
,
H.
Elnaggar
,
F.
Frati
,
R.-p.
Wang
,
M. U.
Delgado-Jaime
,
M.
van Veenendaal
,
J.
Fernandez-Rodriguez
,
M. W.
Haverkort
,
R. J.
Green
,
G.
van der Laan
 et al,
J. Electron Spectrosc. Relat. Phenom.
249
,
147061
(
2021
).
5.
J.
Sherman
,
Spectrochim. Acta
7
,
283
(
1955
).
6.
J.
Guo
,
Int. J. Nanotechnol.
1
,
193
(
2004
).
7.
B.
Keimer
and
J.
Moore
,
Nat. Phys.
13
,
1045
(
2017
).
8.
P.
Vishwakarma
,
M.
Gupta
,
V. R.
Reddy
,
D. M.
Phase
, and
A.
Gupta
,
Phys. Status Solidi RRL
14
,
2000177
(
2020
).
9.
X.
Liu
,
W.
Yang
, and
Z.
Liu
,
Adv. Mater.
26
,
7710
(
2014
).
10.
G.
Hao
,
T. K.
Ekanayaka
,
A. S.
Dale
,
X.
Jiang
,
E.
Mishra
,
C.
Mellinger
,
S.
Yazdani
,
J. W.
Freeland
,
J.
Zhang
,
R.
Cheng
 et al,
Magnetochemistry
7
,
135
(
2021
).
11.
P.
Machado
,
I.
Cano
,
C.
Menéndez
,
C.
Cazorla
,
H.
Tan
,
I.
Fina
,
M.
Campoy-Quiles
,
C.
Escudero
,
M.
Tallarida
, and
M.
Coll
,
J. Mater. Chem. C
9
,
330
(
2021
).
12.
C.-C.
Chiu
,
Y.-W.
Chang
,
Y.-C.
Shao
,
Y.-C.
Liu
,
J.-M.
Lee
,
S.-W.
Huang
,
W.
Yang
,
J.
Guo
,
F. M.
de Groot
,
J.-C.
Yang
, and
Y.-D.
Chuang
,
Sci. Rep.
11
,
5250
(
2021
).
13.
N.
Isomura
,
K.
Oh-ishi
,
N.
Takahashi
, and
S.
Kosaka
,
J. Synchrotron Radiat.
28
,
1820
(
2021
).
14.
M.
Abbate
,
J.
Goedkoop
,
F.
De Groot
,
M.
Grioni
,
J.
Fuggle
,
S.
Hofmann
,
H.
Petersen
, and
M.
Sacchi
,
Surf. Interface Anal.
18
,
65
(
1992
).
15.
A. K.
Poswal
,
C.
Basak
,
D.
Udupa
, and
M.
Deo
,
AIP Conf. Proc.
2265
,
030203
(
2020
).
16.
J.
Jaklevic
,
J.
Kirby
,
M.
Klein
,
A.
Robertson
,
G.
Brown
, and
P.
Eisenberger
,
Solid State Commun.
23
,
679
(
1977
).
17.
K.
Wojtaszek
,
W.
Blachucki
,
K.
Tyrala
,
M.
Nowakowski
,
M.
Zajac
,
J.
Stepien
,
P.
Jagodzinski
,
D.
Banas
,
W.
Stanczyk
, and
J.
Czapla-Masztafiak
,
J. Phys. Chem. A
125
,
50
(
2021
).
18.
T.
Schneller
,
H.
Kohlstedt
,
A.
Petraru
,
R.
Waser
,
J.
Guo
,
J.
Denlinger
,
T.
Learmonth
,
P.-A.
Glans
, and
K.
Smith
,
J. Sol-Gel Sci. Technol.
48
,
239
(
2008
).
19.
C.
Baeumer
,
C.
Schmitz
,
A. H.
Ramadan
,
H.
Du
,
K.
Skaja
,
V.
Feyer
,
P.
Müller
,
B.
Arndt
,
C.-L.
Jia
,
J.
Mayer
 et al,
Nat. Commun.
6
,
8610
(
2015
).
20.
E.
Kröger
,
A.
Petraru
,
A.
Quer
,
R.
Soni
,
M.
Kalläne
,
N. A.
Pertsev
,
H.
Kohlstedt
, and
K.
Rossnagel
,
Phys. Rev. B
93
,
235415
(
2016
).
21.
W.
Sun
,
B.
Gao
,
M.
Chi
,
Q.
Xia
,
J. J.
Yang
,
H.
Qian
, and
H.
Wu
,
Nat. Commun.
10
,
3453
(
2019
).
22.
S.
Kumar
,
C. E.
Graves
,
J. P.
Strachan
,
A. D.
Kilcoyne
,
T.
Tyliszczak
,
Y.
Nishi
, and
R. S.
Williams
,
J. Appl. Phys.
118
,
034502
(
2015
).
23.
D.
Ielmini
and
R.
Waser
,
Resistive Switching: From Fundamentals of Nanoionic Redox Processes to Memristive Device Applications
(
John Wiley & Sons
,
2015
).
24.
C.
Larsson
,
A.
Beutler
,
O.
Björneholm
,
F.
Federmann
,
U.
Hahn
,
A.
Rieck
,
S.
Verbin
, and
T.
Möller
,
Nucl. Instrum. Methods Phys. Res., Sect. A
337
,
603
(
1994
).
25.
G.
Akgül
,
Bull. Mater. Sci.
37
,
41
(
2014
).
26.
R.
Singh
,
M.
Gupta
,
D.
Phase
, and
S.
Mukherjee
,
Mater. Res. Express
6
,
116449
(
2019
).
27.
S.
Sze
,
J.
Moll
, and
T.
Sugano
,
Solid-State Electron.
7
,
509
(
1964
).
28.
O. Y.
Ridzel
,
V.
Astašauskas
, and
W. S.
Werner
,
J. Electron Spectrosc. Relat. Phenom.
241
,
146824
(
2020
).
29.
E.
Kröger
, “
In-operando-Röntgenspektroskopie von oxidischen Materialien und Grenzflächen
,” Ph.D. thesis (Christian-Albrechts-Universität zu Kiel,
2015
).

Supplementary Material