We report the electrical resistance switching of YbFe2O4, which is one of the electronic ferroelectrics and shows multi-step polarization reversal. The electrical resistance of the single crystal bulk of YbFe2O4 was measured along the c-axis. Three kinds of resistance states were observed: high resistance state (HRS), low resistance state (LRS), and extra low resistance state (ELRS). The as-grown sample was in HRS. The resistance state switched from HRS to LRS under an electric field of ∼180 V/cm. HRS was reproduced under the same magnitude of the electric field in the opposite direction. The switching from LRS to ELRS was assisted by a current flow in the ab-plane, which is orthogonal to the measurement axis (c-axis). The switching from LRS to ELRS was observed under an electric field of ∼140 V/cm during the resistance measurement along the c-axis after a current flow in the ab-plane for a few seconds. The resistance ratio in HRS, LRS, and ELRS was ∼2:1.5:1, which is regarded as a large value considering that the sample size was sub-millimeter order. This multi-step resistive switching is likely due to the multi-step reversal of electric polarization, accompanied by a multi-step change in Schottky barrier height between the sample and electrodes. The currents in the ab-plane are considered to release some pinnings and assist in the polarization reversal.

Memory devices based on resistive switching have attracted much attention. Various mechanisms have been reported. A current path “filament” is often formed between electrodes by oxygen vacancies in oxides.1–3 When a large voltage is applied to an oxide thin film, oxygen ions within the oxide can migrate, leading to a current path of oxygen vacancies between the two electrodes. This current path is called a filament. The interface effect is also reported.4–6 The oxygen deficiency or oxygen ions are often drifted by the electric field and collected at the interface between the oxide and the electrode, changing the electrical resistance at the interface. Ferroelectricity also affects resistance.7–9 The ferroelectric film is sandwiched between different kinds of electrodes to form a Schottky junction at one interface and an ohmic junction at the other interface. Since the energy levels of polarized ferroelectrics are tilted inside the ferroelectric, the height of the Schottky barrier between the ferroelectric and the electrode changes according to the direction of polarization, enabling resistance switching by polarization reversal.

In the case of resistive switching devices using ferroelectrics, their performance depends on the polarization switching properties of ferroelectrics. Polarization switching in ordinary ferroelectrics is caused by ionic displacement. In contrast, the polarization switching of “electronic ferroelectrics”10 can be achieved only by electron transfer without ionic displacement. Therefore, the polarization switching of electronic ferroelectrics is expected to be faster, less energy consuming, and more iteratively tolerant.

Rare earth ion ferrite RFe2O4 (R = Y, Dy, Ho, Er, Tm, Yb, Lu, Sc, or In) is an electronic ferroelectric whose ferroelectricity arises from polar charge order.11–13  RFe2O4 consists of double stacked iron triangular layers (W-layers) separated by a single rare earth triangular layer. Equal amounts of Fe2+ and Fe3+ exist on the triangular lattice and form charge ordering at room temperature. Each W-layer has electric polarization since the Fe2+ and Fe3+ are arranged without inversion symmetry. In-phase stacking of polarized W-layers results in ferroelectricity.

We reported a single hysteresis P–E loop in the YbFe2O4 single crystal at 10 K.12 A multi-step hysteresis P–E loop was also observed, indicating that the polarization switches in multiple steps. It was observed that the electric polarization was reversed by the reversal of the applied electric field. However, the reversal process did not show a simple one-step change but consists of some medium stages. We call this phenomenon the multi-step reversal of polarization, which indicates a multi-step change in the net polarization of the entire sample. When the net polarization changes in multiple steps, the height of the Schottky barrier at the interface with the electrode is also expected to change in multiple steps, resulting in the multi-step switching of the resistance. The multi-step resistance switching could be applied to multi-level ReRAM.

Combining the advantages of multi-step polarization inversion and electronic ferroelectricity, YbFe2O4 could be applied to multi-level ReRAM with high speed, low energy consumption, and a long lifetime. In this study, we investigated resistive switching in YbFe2O4 bulk single crystal. In general, bulk samples are less sensitive to interface effects than film samples. However, for RFe2O4, it is known that the Schottky junction with the electrode significantly affects the electrical properties, even in bulk.14 

High quality single crystal bulk of YbFe2O4 was grown using the floating zone melting method, in the same way as in Ref. 12. The sample showed spotty superlattice diffraction signals, which promise high stoichiometry and the formation of the charge order.15 The samples were cut into 1.3 × 1.0 × 0.55 mm3 sizes (sample A) and 2.672 × 2.062 × 0.725 mm3 sizes (sample B) for the electric measurements. The largest surface of the sample corresponds to the ab-plane. Silver paste was used for the electrodes. The electric measurements were performed along the c-axis at 300 K, as shown in Fig. 1(a). The voltage was measured under the current flow in order to reduce self-heating by the Joule effect. Keithley 224 and Keithley 2000 were used as the current source and voltmeter, respectively.

FIG. 1.

(a) Experimental setup for sample A. (b) E–J curves in HRS and LRS in sample A. The circle symbols and blue line indicate the E–J curve in HRS (as-grown). The square symbols and green line indicate the E–J curve in LRS. The triangle symbols and red line indicate the E–J curve in the reproduced HRS. The dotted arrows indicate the direction of the hysteresis curve. (c) E–J curve during the switching process from HRS to LRS. (d) E–J curve during the process from LRS to HRS. (e) Time dependences of the applied current density (gray line without symbols) and measured electric field (blue and green lines with circle and square symbols) during the process from HRS to LRS. The dotted gray lines indicate the electric field in HRS for comparison. (f) Time dependences of the applied current density and measured electric field (green and red lines with square and triangle symbols) during the process from LRS to HRS.

FIG. 1.

(a) Experimental setup for sample A. (b) E–J curves in HRS and LRS in sample A. The circle symbols and blue line indicate the E–J curve in HRS (as-grown). The square symbols and green line indicate the E–J curve in LRS. The triangle symbols and red line indicate the E–J curve in the reproduced HRS. The dotted arrows indicate the direction of the hysteresis curve. (c) E–J curve during the switching process from HRS to LRS. (d) E–J curve during the process from LRS to HRS. (e) Time dependences of the applied current density (gray line without symbols) and measured electric field (blue and green lines with circle and square symbols) during the process from HRS to LRS. The dotted gray lines indicate the electric field in HRS for comparison. (f) Time dependences of the applied current density and measured electric field (green and red lines with square and triangle symbols) during the process from LRS to HRS.

Close modal

Figures 1(b)1(d) show the electric field as a function of the current density (E–J curve) of sample A. Before discussing resistive switching, we note the characteristics of the E–J curve common to all resistance states. All E–J curves exhibited nonlinearity and hysteresis characteristics. First, we describe nonlinearity. With increasing current, the slope of the E–J curve decreases, becomes negative, and finally asymptotically approaches zero, as shown in Figs. 1(c) and 1(d). This nonlinearity is mainly due to the Joule effect.16 The current flow and the voltage during the electrical measurements generate Joule heat, which increases the sample temperature and decreases the sample resistance. The resistance decreases rapidly with increasing current since the Joule effect is larger for larger current and voltage. Therefore, the slope of the E–J curve decreases with increasing current and eventually becomes negative. However, the decrease in resistance gradually slows down for much larger currents since Joule heat decreases with decreasing voltage. As a result, the slope of the E–J curve approaches zero asymptotically.

Next, we describe the hysteresis characteristic of the E–J curves. The electric field is larger with increasing current density than with decreasing it, as shown by the dotted arrows in Fig. 1(b). This hysteresis character is also derived from the Joule effect. In the process of increasing the current, the sample temperature increases and the resistance decreases. If the current is increased faster than the temperature increases, the resistance and the voltage show higher values. In the process of decreasing the current, the resistance and the voltage show a lower value if the current is decreased faster. These nonlinearity and hysteresis characteristics are maintained before and after the switching of the resistance.

Figure 1(b) shows E–J curves in the high resistance state (HRS) and low resistance state (LRS) of sample A. The as-sintered sample was in HRS, as shown by the circle symbol and blue line in Fig. 1(b). Figures 1(c) and 1(e) show the process from HRS to LRS. In the negative current flow, the electric field dropped down to around −5.5 A/cm2 (−185 V/cm), indicating the switching from HRS to LRS. HRS was reproduced by a positive current flow. Figures 1(d) and 1(f) show the switching process from LRS to HRS. The electric field jumped up to around +5.5 A/cm2 (185 V/cm). In other words, a current flow of 5.5 A/cm2 (a voltage of 185 V/cm) in the negative direction caused a switching from HRS to LRS, and the same magnitude of current and voltage in the opposite direction caused a switching from LRS to HRS. The E–J curve in the reproduced HRS is in good agreement with that of the as-sintered sample, as shown in Fig. 1(b). The ratio of resistivity between HRS and LRS is about 2:1 at 0.75 A/cm2. This value is regarded as large, considering that the sample size was sub-millimeter order.

To investigate whether multi-step resistance switching occurs, resistance was measured with larger currents. However, no further resistive switching was observed. Further resistive switching was likely to be inhibited by the voltage limitation caused by the Joule effect. A larger electric field is likely to be required for multi-step resistive switching, considering that larger voltages were required for the multi-step P–E loop observations. However, the max electric field is about 400 V/cm under the current density of 0.5 A/cm2, as shown in Figs. 1(c) and 1(d). The voltage decreases due to the Joule effect, even under a larger current flow.

Therefore, we investigated ways to switch polarization other than applying a large voltage. We assume that the application of a large electric field reverses the polarization by detaching some pinnings, such as the domain wall. The current in the ab-plane may be able to remove the pins if the domain wall is running along the measurement direction (c-axis), which is orthogonal to the ab-plane. Therefore, sample B was prepared with electrodes for current flow in the ab-plane, in addition to the electrodes for resistance measurement along the c-axis. Sample B was cut from the same rod as sample A. Figure 2(a) shows a schematic diagram of the measurement set up for sample B.

FIG. 2.

(a) Experimental setup for sample B. (b) E–J curves in HRS, LRS, and ELRS in sample B. The circle symbols and blue line indicate the E–J curve in HRS (as-grown). The square symbols and green line indicate the E–J curve in LRS. The diamond-shaped symbols and orange line indicate the E–J curve of ELRS. (c) E–J curves during the switching process from HRS to LRS.

FIG. 2.

(a) Experimental setup for sample B. (b) E–J curves in HRS, LRS, and ELRS in sample B. The circle symbols and blue line indicate the E–J curve in HRS (as-grown). The square symbols and green line indicate the E–J curve in LRS. The diamond-shaped symbols and orange line indicate the E–J curve of ELRS. (c) E–J curves during the switching process from HRS to LRS.

Close modal

Figures 2(b) and 2(c) show the E–J curves of sample B. First, we describe the overall trend of the E–J curves of sample B while comparing it to that of sample A. Non-linearity and hysteresis characteristics were observed in sample B as well as sample A. The resistivity of sample B was smaller than that of sample A. The difference is noticeable when a current exceeding 0.1 A/cm2 is applied, as shown in Figs. 1(b) and 2(c). This difference is likely due to the larger size of sample B compared to sample A. The current flowing through the entire sample is larger in sample B, even at the same current density as sample A, since sample B has a larger electrode area. This difference in the sample size results in a larger Joule heat and lower resistance in sample B than in sample A. Therefore, the difference in resistance between samples A and B is not considered to be an essential difference.

We describe the HRS and LRS of Sample B. The E–J curve of as-grown sample B is indicated by the circle symbol and blue line in Fig. 2(b). As-grown sample B was in HRS as well as sample A. Figure 2(c) shows the switching process from HRS to LRS. The resistance state switched from HRS to LRS with a current density of 0.43 A/cm2 and an electric field of 183 V/cm along the c-axis. However, no further resistive switching was observed even with a larger current along the c-axis as well as in sample A.

Then, a current of 0.1 A/cm2 was applied in the ab-plane for a few seconds. The resistance was measured in the range of ±0.1 A/cm2 along the c-axis after stopping the current flow in the ab-plane. The resistance state was still LRS. However, the resistance dropped down during the measurement with a large current (up to 1 A/cm2) along the c-axis. The current density and the electric field were +0.11 A/cm2 and +136 V/cm, respectively, when the resistance switched. In other words, a state with even lower resistance than the LRS was realized by a large current flow along the measurement direction (c-axis), assisted by current flow in the ab-plane. We name this resistance state “extra low resistance state” (ELRS), which has lower resistance than that in LRS. The E–J curve in ELRS is shown by diamond-shape plots and an orange line in Fig. 2(b). The resistance ratio between HRS, LRS, and ELRS was 2:1.5:1 at 0.1 A/cm2.

We discuss the mechanism of this multi-step resistive switching in YbFe2O4. Three mechanisms have been proposed for resistance switching in oxides: first, the formation of current path “filaments” by oxygen vacancies;1–3 second, the trapping of oxygen ions or oxygen vacancies at the interface;4–6 and third, the switching of resistance due to polarization reversal.7–9 First, we discuss the possibility of filament formation. The current path “filaments” are often formed between electrodes by oxygen vacancies in the oxide film.1–3 When a large voltage is applied to an oxide thin film, oxygen ions within the oxide film can migrate, leading to a current path of oxygen vacancies between the two electrodes. This current path is called a filament. However, this scenario is unlikely for our results. Our sample is a 0.55 or 0.725 mm thick bulk; it is unlikely that atoms or ions would move enough to form such a long filament. Furthermore, the switching ratio would be much larger if such a long filament could be formed in the bulk.

Next, we discuss the possibility of resistive switching due to the trapping of oxygen ions or oxygen vacancies at the interface. The oxygen deficiency or oxygen ions are often drifted by the electric field and collected at the interface with the electrode, changing the electrical resistance at the interface.4–6 However, this model also does not suit YbFe2O4 because it cannot explain the three-step resistance switching. Furthermore, it is unlikely that atoms or ions would move such a long distance (0.55 or 0.725 mm), as discussed for the filaments.

Finally, we discuss the possibility of resistive switching due to the reversal of polarization. The resistive switching is reported in ferroelectric films with a Schottky junction between the films and the electrodes.7–9 Since the energy levels of polarized ferroelectrics are tilted inside the ferroelectric, the height of the Schottky barrier changes according to the direction and magnitude of polarization, enabling resistance switching by polarization reversal. A Schottky barrier is formed between the ferroelectric YbFe2O4 and the silver electrodes.14 The Schottky barrier height is expected to change depending on the magnitude of the net polarization of YbFe2O4. Therefore, the stepwise changes in net polarization in YbFe2O4 are expected to cause stepwise changes in Schottky barrier height and resistance. The currents in the ab-plane are considered to release the pinning and assist in the polarization reversal. This hypothesis is supported by the magnitude of the electric field required for resistive switching. The electric field for switching should have common values for samples A and B because the electric field corresponds to the coercive electric field in the P–E hysteresis if switching is due to polarization reversal. The electric field required for switching is ∼180 V/cm, which is common to samples A and B. This hypothesis is also supported by the small resistive switching ratio. The polarization reversed at one time was as small as 1 nC/cm2, which is the step height observed in the multi-step P–E loop. This small polarization is consistent with the small resistance change. Therefore, the multi-step resistive switching in YbFe2O4 is highly likely to originate from multi-step polarization switching.

We report multi-step resistive switching of YbFe2O4, although further investigations are required to confirm the mechanism. YbFe2O4 has the potential for application as a new multi-level ReRAM with high speed, low energy consumption, and a long lifetime if the resistive switching arises from polarization switching.

YbFe2O4 is one of the electronic ferroelectrics in which the electric polarization switches only by electron transfer without ionic displacement. Multi-step polarization switching has also been observed. These characteristics have the potential to be applied to multi-level ReRAM with high speed, low energy consumption, and a long lifetime. In this study, the electric resistance of the single crystal bulk of YbFe2O4 was measured along the c-axis. Three kinds of resistance states were observed: HRS, LRS, and ELRS. The as-grown sample was in HRS. The resistance state switched from HRS to LRS under an electric field of ∼180 V/cm. HRS was reproduced under the same magnitude of the electric field in the opposite direction. The switching from LRS to ELRS was assisted by a current flow in the ab-plane. The switching from LRS to ELRS was observed at an electric field of ∼140 V/cm during the resistance measurement along the c-axis after a current flow in the ab-plane. The resistance ratio in HRS, LRS, and ELRS was ∼2:1.5:1, which is regarded as a large value considering that the sample size was submillimeter order. This multi-step resistive switching is likely due to the multi-step reversal of electric polarization, accompanied by a multi-step change in Schottky barrier height between the sample and electrodes. The currents in the ab-plane are considered to release the pinning and assist in the polarization reversal. A more detailed investigation will open up the possibility of innovative ReRAM applications for YbFe2O4.

This work was partially supported by JSPS KAKENHI Grant Nos. 22K20361 and 22H01942.

The authors have no conflicts to disclose.

Tomoko Nagata: Conceptualization (lead); Investigation (lead); Visualization (lead); Writing – original draft (lead). Naoshi Ikeda: Resources (equal).

The data that support the findings of this study are available within the article.

1
X.
Gao
,
Y.
Xia
,
J.
Ji
,
H.
Xu
,
Y.
Su
,
H.
Li
,
C.
Yang
,
H.
Guo
,
J.
Yin
, and
Z.
Liu
, “
Effect of top electrode materials on bipolar resistive switching behavior of gallium oxide films
,”
Appl. Phys. Lett.
97
,
193501
(
2010
).
2
I. G.
Baek
,
M. S.
Lee
,
S.
Seo
,
M. J.
Lee
,
D. H.
Seo
,
D.-S.
Suh
,
J. C.
Park
,
S. O.
Park
,
H. S.
Kim
,
I. K.
Yoo
,
U.-I.
Chung
, and
J. T.
Moon
, “
Highly scalable nonvolatile resistive memory using simple binary oxide driven by asymmetric unipolar voltage pulses
,” in
IEDM Technical Digest. IEEE International Electron Devices Meeting, San Francisco
(
IEEE
,
2004
), p.
587
, https://ieeexplore.ieee.org/abstract/document/1419228.
3
D. C.
Kim
,
S.
Seo
,
S. E.
Ahn
,
D.-S.
Suh
,
M. J.
Lee
,
B.-H.
Park
,
I. K.
Yoo
,
I. G.
Baek
,
H.-J.
Kim
,
E. K.
Yim
,
J. E.
Lee
,
S. O.
Park
,
H. S.
Kim
,
U.-I.
Chung
,
J. T.
Moon
, and
B. I.
Ryu
, “
Electrical observations of filamentary conductions for the resistive memory switching in NiO films
,”
Appl. Phys. Lett.
88
,
202102
(
2006
).
4
A.
Baikalov
,
Y. Q.
Wang
,
B.
Shen
,
B.
Lorenz
,
S.
Tsui
,
Y. Y.
Sun
,
Y. Y.
Xue
, and
C. W.
Chu
, “
Field-driven hysteretic and reversible resistive switch at the Ag–Pr0.7Ca0.3MnO3 interface
,”
Appl. Phys. Lett.
83
,
957
(
2003
).
5
S.
Tsui
,
A.
Baikalov
,
J.
Cmaidalka
,
Y. Y.
Sun
,
Y. Q.
Wang
,
Y. Y.
Xue
,
C. W.
Chu
,
L.
Chen
, and
J.
Jacobson
, “
Field-induced resistive switching in metal-oxide interfaces
,”
Appl. Phys. Lett.
85
,
317
319
(
2004
).
6
X.
Chen
,
N. J.
Wu
,
J.
Strozier
, and
A.
Ignatiev
, “
Direct resistance profile for an electrical pulse induced resistance change device
,”
Appl. Phys. Lett.
87
,
233506
(
2005
).
7
N.
Setter
,
D.
Damjanovic
,
L.
Eng
,
G.
Fox
,
S.
Gevorgian
,
S.
Hong
,
A.
Kingon
,
H.
Kohlstedt
,
N. Y.
Park
,
G. B.
Stephenson
,
I.
Stolitchnov
,
A. K.
Taganstev
,
D. V.
Taylor
,
T.
Yamada
, and
S.
Streiffer
, “
Ferroelectric thin films: Review of materials, properties, and applications
,”
J. Appl. Phys.
100
,
051606
(
2006
).
8
L.
Jiang
,
W. S.
Choi
,
H.
Jeen
,
S.
Dong
,
Y.
Kim
,
M.-G.
Han
,
Y.
Zhu
,
S. V.
Kalinin
,
E.
Dagotto
,
T.
Egami
, and
H. N.
Lee
, “
Tunneling electroresistance induced by interfacial phase transitions in ultrathin oxide heterostructures
,”
Nano Lett.
13
,
5837
5843
(
2013
).
9
H. J.
Mao
,
P. X.
Miao
,
J. Z.
Cong
,
C.
Song
,
B.
Cui
,
J. J.
Peng
,
F.
Li
,
G. Y.
Wang
,
Y. G.
Zhao
,
Y.
Sun
,
L. R.
Xiao
, and
F.
Pan
, “
Interface-modification-enhanced tunnel electroresistance in multiferroic tunnel junctions
,”
J. Appl. Phys.
116
,
053703
(
2014
).
10
T.
Portengen
,
T.
Ostreich
, and
L. J.
Sham
, “
Theory of electronic ferroelectricity
,”
Phys. Rev. B
54
,
17452
(
1996
).
11
N.
Ikeda
,
H.
Ohsumi
,
K.
Ohwada
,
K.
Ishii
,
T.
Inami
,
K.
Kakurai
,
Y.
Murakami
,
K.
Yoshii
,
S.
Mori
,
Y.
Horibe
, and
H.
Kito
, “
Ferroelectricity from iron valence ordering in the charge-frustrated system LuFe2O4
,”
Nature
436
,
1136
1138
(
2005
).
12
T.
Nagata
,
P.-E.
Janolin
,
M.
Fukunaga
,
B.
Roman
,
K.
Fujiwara
,
H.
Kimura
,
J.-M.
Kiat
, and
N.
Ikeda
, “
Electric spontaneous polarization in YbFe2O4
,”
Appl. Phys. Lett.
110
,
052901
(
2017
).
13
K.
Fujiwara
,
Y.
Fukada
,
Y.
Okuda
,
R.
Seimiya
,
N.
Ikeda
,
K.
Yokoyama
,
H.
Yu
,
S.
Koshihara
, and
Y.
Okimoto
, “
Direct evidence of electronic ferroelectricity in YbFe2O4 using neutron diffraction and nonlinear spectroscopy
,”
Sci. Rep.
11
,
4277
(
2021
).
14
T.
Kambe
,
Y.
Fukada
,
J.
Kano
,
T.
Nagata
,
H.
Okazaki
,
T.
Yokoya
,
S.
Wakimoto
,
K.
Kakurai
, and
N.
Ikeda
, “
Magnetoelectric effect driven by magnetic domain modification in LuFe2O4
,”
Phys. Rev. Lett.
110
,
117602
(
2013
).
15
K.
Fujiwara
,
T.
Karasudani
,
M.
Fukunaga
,
H.
Kobayashi
,
J.
Kano
,
P.-E.
Janolin
,
J.-M.
Kiat
,
Y.
Nogami
,
R.
Kondo
, and
N.
Ikeda
, “
Possible charge order structure of stoichiometric YbFe2O4
,”
Ferroelectrics
512
,
85
91
(
2017
).
16
T.
Nagata
,
Y.
Fukada
,
M.
Kawai
,
J.
Kano
,
T.
Kambe
,
E.
Dudzik
,
R.
Feyerherm
,
P.-E.
Janolin
,
J.-M.
Kiat
, and
N.
Ikeda
, “
Nonlinear electric conductivity of charge ordered system RFe2O4 (R = Lu, Yb)
,”
Ferroelectrics
442
,
45
49
(
2013
).