CuI - Adv Funct Materials - 2022 - Mishra - Light Mediated Multi Level Flexible Copper Iodide Resistive Random Access Memory For-1
CuI - Adv Funct Materials - 2022 - Mishra - Light Mediated Multi Level Flexible Copper Iodide Resistive Random Access Memory For-1
www.afm-journal.de
Adv. Funct. Mater. 2023, 33, 2211022 2211022 (1 of 12) © 2022 Wiley-VCH GmbH
16163028, 2023, 8, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adfm.202211022 by Chungnam National University, Wiley Online Library on [20/02/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
www.advancedsciencenews.com www.afm-journal.de
understanding of CuI film properties for memory applications CuI film, particularly the grain size and degree of roughness.[31]
has been absent up to now. A variety of CuI synthesis methods Accordingly, in the present work, the iodine sublimation
have been employed for electronic applications, including the scheme was conducted under vacuum (≈1 mtorr) at room tem-
use of reactive sputtering,[24] spin coating of CuI solution,[25] dip perature to precisely control the rate of iodination of Cu film
coating,[26] thermal evaporation of CuI,[27] and others.[28–30] according to the vacuum condition.[32,33] Comparatively, vapor
Surprisingly, a CuI NVM has rarely been reported,[22,23] iodination of Cu at room temperature results in a rough sur-
and further in-depth studies, including long-term endurance, face and narrower grain-sized CuI films.[34] On the other hand,
charge transport mechanisms, and others, are limited, pos- vacuum-driven solid iodination results in highly transparent,
sibly due to the poor stability of the film. This low stability is fine grains, broader grain size distribution, and smoother CuI
caused by mediocre CuI film surface and morphology, which film due to the room temperature conditions (see Figure S1a,
depends on the synthesis route. Recently, a solid iodination Supporting Information).[32]
method yielded high-quality CuI film with a columnar grain- Figure 1b displays the X-ray diffraction (XRD) patterns,
like morphology and nanoscopic holes-free surface, due to revealing the diffraction peaks at 25.5°, 42.21°, and 49.9°, cor-
vacuum-driven processing conditions, which effectively aided responding to the assigned (111), (220), and (311) planes, respec-
in high RS performance.[31] Interestingly, operationally stable tively. These match γ-phase CuI, as reported in JCPDS No.
CuI RRAMs with a multi-level scheme based on the solid iodi- 06–0246.[35] The zinc blende, Marshite crystal structure of CuI
nation technique have not been reported, even though the pre- is depicted in the inset of Figure 1b.
sented CuI synthesis process and its decent compatibility with The scanning survey XPS spectrum confirmed the Cu and I
optimum NVM configuration should open up tremendous as well as the absence of any other extra peaks in the CuI film
research opportunities. (Figure S1b, Supporting Information ).[36] The core level Cu 2p
The present work first reports reliable CuI RRAM devices and I 3d orbitals, ascribed to spin-orbit interactions, were split
based on the solid iodination technique. Exhaustive studies up into two peaks, at 932.1 eV (Cu 2p5/2) and 951.9 eV (Cu 3d3/2)
were performed on their promising properties which include (Figure 1c), while the I 3d core-level spectra split up into two
i) low-voltage resistive switching at less than ± 0.5 V with out- peaks at 619.3 and 631.4 eV, corresponding to I 3d5/2 and I 3d3/2
standing endurance (103 cycles), ii) decent retention (5 × 104 s) (Figure 1d), respectively. Furthermore, the XPS peak spectra
with a consistent On/Off ratio (≈104), iii) reproducible photonic were free from shakeups or satellite peaks. The data collectively
response in HRS, which may make them suitable for light- indicate that the oxidation state of Cu is + 1, reaffirming that
assisted multi-level memory storage, iv) flexible CuI RRAMs the Cu2+ oxidation state was not formed.[37,38]
on PET substrates, exhibiting similar performance to rigid Figure 1e–g present the planar scanning electron microscopy
substrates, and v) securing reliable operation of CuI RRAMs (SEM) images, EDS elemental mapping, and cross-sectional
via optimized PMMA encapsulation, subsisting the RS per- SEM images of the CuI film on ITO/glass. The planar FE-SEM
formance over 15 days in air. In addition, the CuI RRAMs image (Figure 1e; Figure S2a, Supporting Information ) and the
presented in this study are so reliable that comprehensive EDS elemental mapping (Figure 1f) of the CuI film confirm
studies on temperature-dependent I–V characteristics were the homogeneous, pinhole-free, transparent crystalline CuI,
executed in the range from 200 to 343 K. This unravels acti- and the uniform distribution of both Cu and I throughout the
vation energy, Schottky barrier height (ϕb), and other electrical film. The EDS mapping affirms the uniform distribution of Cu
properties which help to understand conduction and resistive and I throughout the thickness of the CuI film. The average
switching properties in the CuI RRAM device. Herein, the pro- grain size of the CuI film, calculated from the FE-SEM image,
posed binary halide CuI shows outstanding characteristics for is 120 nm, which is in line with the reported literature.[32] The
nonvolatile RRAMs and photomemristive characteristics with cross-sectional SEM images (Figure 1g) depict the formation
decent reliability, offering a prominent contribution towards of a uniform, columnar, broader grain-like CuI film. This con-
next-generation invisible, flexible, multivalued NVM storage firms that the deposited 60 nm thick Cu film was converted
applications. into a 270 nm thick CuI film, which matches the theoretical
volume ratio of Cu to CuI (≈4.5–5) within the range reported in
the literature (see Figure S2, Supporting Information).[39]
2. Results and Discussion
2.1. CuI Film Synthesis and Characterizations 2.2. Resistive Switching Characteristics
2.1.1. CuI synthesis, Crystal Structure, Phase, and Morphology 2.2.1. RRAM Characteristics
Figure 1a illustrates the process flows for synthesis of CuI Figure 2a shows typical I–V characteristics in a semi-loga-
film. First, the Cu (60 nm) was deposited on a cleaned ITO/ rithmic scale under the applied bias range of ± 0.5 V for the
glass substrate via thermal evaporation at room temperature ITO/CuI/Ag-based RRAM device. (Inset: Figure 2a illustrates
(Figure 1a(i)). Thereafter, the Cu film was exposed to solid the vertically stacked architecture of the ITO/CuI/Ag-based
iodine in an iodine-loaded vacuum desiccator for 30 min under RRAM devices). The ITO/CuI/Ag device displays bipolar
vacuum, followed by annealing at 120 °C, yielding transparent switching behaviors with a positive bias at the top electrode
CuI film, as schematically illustrated in Figure 1a(ii–iv). The (Ag) with a current compliance (Icc) of 100 µA, while the
iodination route hugely affects the surface morphology of the bottom electrode (ITO) was grounded. Initially, the device is in
Adv. Funct. Mater. 2023, 33, 2211022 2211022 (2 of 12) © 2022 Wiley-VCH GmbH
16163028, 2023, 8, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adfm.202211022 by Chungnam National University, Wiley Online Library on [20/02/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
www.advancedsciencenews.com www.afm-journal.de
Figure 1. a) Schematic of CuI film synthesis on ITO-glass by a room temperature, vacuum-driven iodination method. b) XRD pattern and JCPDS file
of as-prepared CuI films; Inset is γ-phase (Marshite) zinc blende crystal structure of CuI. c,d) XPS spectra of Cu 2p and I 3d core levels of CuI film.
e) Surface FE-SEM images of CuI thin films. f) EDS elemental mapping of Cu and I element in the CuI film. g) Cross-sectional SEM images of fabricated
ITO/CuI vertical structure on ITO-glass (Inset: low magnification of cross-section of CuI/ITO-glass.
the high resistance state (HRS), and showed abrupt switching cycles. The remarkable RS cycling by the ITO/CuI/Ag devices
to a low resistance state (LRS) at a voltage of around + 0.3 V demonstrated a consistent On/Off current ratio (≈104) without
(Vset) during a positive voltage sweep from 0 to 0.5 V, coming any significant degradation for a consecutive 1000 cycles (see
back to 0 V. On the other hand, a negative voltage sweep (0 to Figure S3, Supporting Information ), exhibiting excellent
−0.5) reverts the device to the HRS at an ultra-low reset voltage endurance performance.
(Vreset) (≈−0.15 ± 0.02 V). The forming free RS prevents chem- Moreover, Figure 2c shows the statistically analyzed box-
ical and thermal disintegration of the active layer, which is prev- whisker plots for the set or reset voltage (Vset/reset), realizing
alent during the electroforming process, leading to damage-free excellent operational uniformity. The Vset/reset distribution is
operation and negligible variation in Vset or Vreset. Possibly, the ≈0.29 ± 0.03 V (or −0.12 ± 0.07 V), respectively, as clearly visual-
forming-free switching behavior might be accredited to the ized in the cumulative probability plot in the inset of Figure 2c.
easy migration of Ag+ cations. As a result, Ag+ cations could The CuI RRAMs are expected to be appealing for memory appli-
hop over the barrier height through the CuI film and set up a cations due to their ultra-low power consumption. For instance,
conducting path through the active layer, leading to uniform RS the average power consumption for the setting (or resetting)
behavior.[40] was estimated to be 2.7 pW (or 16 pW), according to the relation
Figure 2b depicts the HRS and LRS, deduced at of Pset = Vset × Icc (or Vreset × Icc), respectively. Data retention can
Vread ≈± 0.05 V from the I–V characteristics of 1000 consecutive be a very crucial parameter when evaluating RRAMs from the
Adv. Funct. Mater. 2023, 33, 2211022 2211022 (3 of 12) © 2022 Wiley-VCH GmbH
16163028, 2023, 8, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adfm.202211022 by Chungnam National University, Wiley Online Library on [20/02/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
www.advancedsciencenews.com www.afm-journal.de
Figure 2. Resistive switching characteristics of ITO/CuI/Ag-based memory devices. a) Typical I–V characteristics of ITO/CuI/Ag device; Inset shows
schematic of vertically stacked configuration of ITO/CuI/Ag device. b) HRS and LRS for 1000 RS cycles of ITO/CuI/Ag device. c) Statistics of the set
and reset voltage distributions for 1000 RS cycles, depicted as a box-whisker plot; inset shows the cumulative set and reset voltage for ITO/CuI/Ag
device. d) Data retention characteristics measured for duration 5 × 104 s at Vread ± 0.05 V for the On and Off states. e) Multi-level I–V characteristics
under different compliance currents (Icc at 1, 10, 100 µA, and 1 mA) for ITO/CuI/Ag device. f). HRS and LRS at different current compliances (Icc), 1,
10, 100 µA, and 1 mA.
perspective of practical applications. In addition, the data reten- ITO bottom electrode (ITO) in the CuI active layer.[42] Further-
tion characteristics, measured at Vread ≈ ± 0.05 V for the ITO/ more, the reset for the condition of ICC (≈100 µA) occurred at
CuI/Ag device, are shown in Figure 2d. For a sufficiently long an ultra-low voltage (−0.02 to −0.1 V), which is highly appealing
period of 5 × 105 s, the ITO/CuI/Ag devices retained the HRS for the low-power memristor application.
and LRS without discernable degradation. Herein, the CuI RRAMs demonstrate tangible four-level
The multi-level I-V characteristics and their corresponding storage capability, enhancing data storage density, and favorable
resistance states are explored in Figures 2e,f. Multi-level data leverage in analog memristors.[43] To assess the reproduc-
storage in the CuI-based RRAM devices can be achieved due to ibility of the RS, the cell-to-cell variation for the ITO/CuI/
their abrupt switching and high On/Off ratios, by controlling Ag-based memory device was evaluated using 50 assembles
the Icc. Figure 2e shows the I-V characteristics of the ITO/CuI/ with the same physical condition (See Figure S4a–c, Supporting
Ag device under different current compliance, ICC (1 µA, 10 µA, Informaiton). The HRS/LRS value and Vset/reset distributions
100 µA, and 1 mA), illustrating four different LRS states with exhibited high uniformity for 50 cell-to-cell measurements,
similar HRS (See Figure 2f). Furthermore, even with different substantiating operational consistency and overall device
Icc conditions, the CuI RRAM device shows the ability to switch yield.
from HRS to LRS at a similar voltage Vset. These multi-level
I-V characteristics with the similar Vset signify that the RS is
independent of the current compliance. This observation shows 2.3. Current Conduction and Resistive Switching Mechanism
that the applied electric field predominantly dictates the RS in
memory devices.[41] The current conduction mechanism and switching mechanism
The Vset for Icc from 1 to 100 µA shows no significant varia- are presented in Figure 3. The abrupt switching from HRS to
tion, except for a marginal change for Icc = 1 mA. However, the LRS (set) and vice versa (reset) is attributed to the filamentary
Vreset during reversible RS exhibited slight fluctuations at the type switching instigated by the electrochemical reaction and its
negative bias sweep. At lower compliance (≈1 and 10 µA), LRS rupture due to thermal effects.[44,45] For a better understanding
to HRS switching is noticed at ≈0 V, which is attributed to the of the current conduction mechanism for HRS and LRS, the
formation of a weak conducting filament. On the other hand, I–V characteristics were plotted in a double logarithmic scale,
the higher current compliance (1 mA) requires a higher reset as shown in Figure 3a. The governing current transport mech-
voltage because of the stronger filament formation. This is due anisms in HRS and LRS have distinct natures. The disparity
to the required high Icc and more cations accumulated near the concerning the electrical conductions in the same voltage range
Adv. Funct. Mater. 2023, 33, 2211022 2211022 (4 of 12) © 2022 Wiley-VCH GmbH
16163028, 2023, 8, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adfm.202211022 by Chungnam National University, Wiley Online Library on [20/02/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
www.advancedsciencenews.com www.afm-journal.de
Figure 3. Current conduction mechanism and resistive switching mechanism. a) Double-logarithm plot of the I-V characteristics of ITO/CuI/Ag device.
b) ln I–V plot for LRS and ln I–V1/2 plot for HRS (straight lines are linear fittings). Schematics of energy level diagram for Ohmic contact between Ag or
(ITO) with CuI c) before contact, and d) after contact illustrating the current conduction via thermally activated ion hopping in the HRS. e) Schematic
of the proposed switching model demonstrating filamentary-based electrochemical metallization type in CuI-based RRAM.
(0 to 0.5 V) suggests different mechanisms are involved for cur- where I is current, e for the dielectric constant, q for the elec-
rent conduction in HRS and LRS. trical charge, d for the thickness of CuI film, and V is the
In the On state, i.e., the LRS region, the LRS fits linearly for applied voltage. During current conduction in HRS, the applied
ln I versus ln V with a slope of 1.04 during the positive sweep electric field assists easy transport of Ag+ ions over the Schottky
(0 to + 0.5 to 0 V), which implies that the LRS state follows barrier at the interface of the top electrode, Ag and CuI film.[50]
Ohmic conduction behavior.[46] On the other hand, the Off state, The governing Schottky conduction in HRS is based on
i.e., the HRS region, follows a non-linear path with a slope of Equation (2):
0.51, closely consistent with the Schottky conduction mecha-
nism.[47] To elucidate the current conduction mechanisms the qE
I–V curves were replotted in Figure 3b, illustrating the Ohmic −q ϕ b −
( 4 πεr ε o )
conduction in LRS and Schottky emission in HRS. Figure 3b(i) J = A * T 2 exp (2)
shows the linearly fitted plot for the I versus V plot in the LRS kT
region, confirming the Ohmic conduction mechanism. This is
attributed to the formation of a conducting filament between
the top and bottom electrodes through the CuI resistive layer, where A* is for effective Richardson constant, q for electrical
resulting in abrupt switching. However, for the assessment of charge, ϕb for the Schottky barrier height, εo for vacuum per-
the non-linear HRS region, I–V characteristics were replotted mittivity, and εr for relative permittivity.[48,49]
as logarithmic current versus the square root of voltage The energy band diagram before physical contact formation
(ln I vs V1/2) relation in Figure 3b(ii). This region (HRS) lin- between Ag and ITO with CuI is shown in Figure 3c, revealing
early fitted in the ln I versus V1/2 plot, substantiating that the information on the band diagram necessary for the Ag/CuI and
Schottky emission is predominant in the Off state. CuI/ITO junction. Overall, the switching behaviors decently
According to Equation (1), the migration of diffused cations agree with electrochemical metallization (ECM) type switching,
(Ag+) from the top electrode is governed by the conduction and thus, the proposed switching model is assessed with the
mechanism of thermionic Schottky emission.[48,49] energy band diagram after the physical contact formation, for
CuI sandwiched between Ag and ITO (Figures 3d,e). Also, the
speculated electro-physical kinetics are schematically illustrated
q3 in Figure 3e, and explained in relation to the above-mentioned
lnI ∝ kT × V (1)
4 πε d energy band diagram in Figure 3d.
Adv. Funct. Mater. 2023, 33, 2211022 2211022 (5 of 12) © 2022 Wiley-VCH GmbH
16163028, 2023, 8, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adfm.202211022 by Chungnam National University, Wiley Online Library on [20/02/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
www.advancedsciencenews.com www.afm-journal.de
The switching process initiates with the oxidation of Ag when of Au, top electrodes. This was attributed to the high stability
a positive bias from 0 to 0.5 V is applied to the top electrode. of the Au top electrodes, which suppresses conducting fila-
This initiates the formation of Ag+ ions followed by the gradual ment formation. This suggests that Au in the top electrode in
diffusion of generated Ag+ cations into the CuI. Thereafter, the the CuI film did not participate in the RS behaviors. Very low
ions move toward the ITO electrode through the CuI resis- on/off switching behavior, less than one order of magnitude,
tive layers. The migration of cations (Ag+) in the CuI layer is was observed in CuI with the Au-top electrode. This might be
speculated to be associated with thermally assisted ion hopping due to point defects associated with possible candidates, such
as one of the main transport mechanisms for the HRS state, as vacancies (Vcu and VI), or interstitial defects (Cui and Ii)
as depicted in the energy band diagram (See Figure 3d,e(ii)). present in the CuI film, reaffirming that ECM-type filamen-
Thus, the main HRS conduction is highly relevant to trap tary switching mainly governs the kinetics in the ITO/CuI/
states-mediated conduction in the switching media of CuI, Ag device. The detailed top and bottom electrode study is pre-
which is consistent with previously reported results based on sented in Figure S6c,d (Supporting Information).
thermally excited electrons.[51] In addition, when Ag+ ions reach The thermal stability of the ITO/CuI/Ag device was exam-
the bottom electrodes (Figure 3e(iii)), the Ag+ ions undergo ined in the temperature range from 203 to 343 K to investigate
reduction to Ag, which tends to form a conducting filament. the correlation between the charge transport of carriers, RS
This results in abrupt switching from HRS to LRS. For the mechanism, and operational stability in the stipulated tem-
reset process, i.e., switching from LRS to HRS, the conducting perature range. The temperature-dependent I–V characteristics
path is supposed to be ruptured due to the Joule-heating dis- during RS confirmed the operational stability of the ITO/CuI/
solution of the Ag filament electrodes (Figure 3e(iv)).[46,52] Thus, Ag device from 203 to 343 K, with a step size of 20 K (Figure 4a).
the switching behavior and capability of CuI can be achieved at For analytical understanding, the HRS current was replotted
ultra-low bias voltage, strongly hinting that the conducting fila- according to temperature in Figure 4b in the voltage range 0.02
ments are very narrow and thermally unstable. to 0.1 V, demonstrating there was a consistent increase in the
Selection of the electrode pair plays a critical role in HRS current value with temperature. This is possibly due to the
achieving these remarkable and reproducible switching prop- synergistic consequence of metal cation (Ag+) migration from
erties. Several metals with different work functions were used the top electrode and the point defects present in the switching
for the top and bottom electrodes to understand the switching media (CuI).[53] The migration speed of the Ag+ cations prob-
mechanism during RS. The ITO/CuI/Au devices exhibited ably governs the current level in the high-temperature regions,
different I–V characteristics than the ITO/CuI/Ag devices (as while the trap states in the CuI film predominantly contribute
shown in Figure S6a,b, Supporting Information), which is pos- to the HRS current variation in the low-temperature region.[46]
sibly due to the absence of the electrochemically inert nature The dominance of point defects in the low-temperature range
Figure 4. Temperature Study of the ITO/CuI/Ag device. a) Temperature-dependent I-V characteristics in a temperature range, of 203–343 K. b) Tem-
perature-dependent HRS state in the voltage range 0.02–0.1 V of the ITO/CuI/Ag device. c) Set and reset voltage distributions with temperature d)
Arrhenius (ln I vs 1/T) plot. e,f) Energy band diagram for the low-temperature and high-temperature region for the ITO/CuI/Ag device.
Adv. Funct. Mater. 2023, 33, 2211022 2211022 (6 of 12) © 2022 Wiley-VCH GmbH
16163028, 2023, 8, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adfm.202211022 by Chungnam National University, Wiley Online Library on [20/02/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
www.advancedsciencenews.com www.afm-journal.de
is experimentally validated by the temperature-dependent elec- slope of the Richardson plots (Figure S8b, Supporting Informa-
trical conductivity, indicating thermally assisted transport of tion). For the ITO/CuI/Ag devices, the obtained barrier heights,
charge carriers is prevalent in the CuI film (See Figure S7, Sup- ϕb at the junction of CuI/Ag were calculated to be 210 meV (or
porting Information).[46,53] 80 meV) for low (and high) temperatures, respectively. This
Figure 4c shows the gradual decreasing tendency of Vset suggests that the HRS current level arises because there are dif-
or (Vreset) with the rise in temperature, which is relevant to ferent Schottky barrier heights for the low and high-tempera-
the indirect temperature effects, coming from various factors ture regions. (See Figure S8c, Supporting Information).[50]
such as i) oxidation and rate of nucleation of the top electrode Here are possible scenarios: the Schottky contact formed at
(Ag), ii) the migration speed of cations (Ag+) in the switching the interface of the Ag and CuI layer, and then the Ag metal
layer, and iii) reduction at the bottom electrode (ITO). The tem- cation, dissolved in the CuI layer, is supposed to function as
perature increase promotes the easy migration of Ag+ cations charge carriers from the top electrode. In addition, Ag metal
through the CuI resistive layer and aids in reducing Ag+ to Ag, cations reduce the Schottky barrier height between the Ag and
resulting in lower Vset.[52,53] However, the Vreset is associated CuI. The experimentally obtained Ea values suggest that the
with the rupture of filaments via Joule heating-assisted disso- HRS current in the low-temperature region solely contributes
lution. Subsequently, the conducting filament is narrower at to the trap and defect states (copper and iodine vacancies) pre-
high temperature, exhibiting lower thermal stability leading to sent in the CuI active layer. Accordingly, the VI and Vcu defects
the reduced Vreset. The temperature-dependent Vset or (Vreset) happen to be the main contributing factors to the charge trans-
values, observed during the switching, substantiate that the port in the low-temperature region (203 to 273 K).
ECM mechanism in the ITO/CuI/Ag devices is predominant, The energy band diagram for low temperature (see
which is consistent with the previously reported literature.[54] Figure 4e) illustrates that VI and Vcu defects majorly contribute
The Arrhenius plot for the HRS state reveals two distinct to the charge transport for the HRS state. Furthermore, the
regions, implying different charge transport mechanisms might charge transport in the high-temperature region (273 to 343 K)
be involved in the measured temperature range (Figure 4d). is governed by the migration of Ag+ cations and point defects
Kaushik et al. reported that thermally evaporated CuI exhibited via hopping conduction, as shown in the energy band diagram
distinct charge transport behaviors in the temperature range (Figure 4f). In other words, the HRS current in the high-tem-
from 150 to 300 K.[27,55] The estimated activation energies (Ea) perature region shows an increasing tendency due to the ele-
at a different voltage from 0.02 to 0.1 V for the high-temper- vated migration speed of Ag+ cations and increased hopping,
ature and low-temperature regions in the HRS are shown in along with distributed trap sites in the CuI film. These results
Figure 4d. The extracted activation energies for high-tempera- are consistent with the Schottky emission-based mechanism
ture (or low-temperature) regions were found to be 0.38 eV (or reported in the literature for the HRS region.[62]
0.18 eV), respectively (see Figure S8a, Supporting Information).
Notably, the trap states are located at a specific energy state
within the optical bandgap of the CuI film, relevant to point 2.4. Evaluation of Light-Mediated CuI Memristor and its
defects such as vacancies (Vcu and VI), interstitial defects (Cui Demonstration as a Flexible Device
and Ii), and substitutional defects, as previously reported in the
literature.[56–58] According to density functional theory calcula- Notably, the competence of CuI as a photo-sensing material
tions, the formation enthalpy (ΔH) for VI (or Vcu) defects in CuI has been actively reported in the literature.[21,31,57,63] For the
films is in the range of ΔH ≤ 0.5–0.1 eV for VI (or 0.005–0.03 eV affirmation of light-assisted multi-level states in the CuI-based
for Vcu), respectively.[59] devices, their I-V characteristics were evaluated under blue
The extracted Ea for charge transport implies that the Ag light (λ = 455 nm) illumination (Figure 5). Figure 5a shows
cation migration and Cu vacancies act as trap sites for ion hop- a schematic cartoon of the electrical and optical analyzer, uti-
ping conduction associated with the localized states and its lized to evaluate the optoelectronic properties of the ITO/CuI/
transition through them in the low-temperature region.[60,61] Ag RRAM devices. The I–V characteristics of the ITO/CuI/
However, the difference in activation energy observed at dif- Ag-based device were evaluated for different power intensities
ferent temperature ranges (i.e., regions I and II) might be under blue light irradiation (Figure 5b), demonstrating a dis-
ascribed to the combined effects on transport behaviors rel- crete HRS current, corresponding to the dark state, and dis-
evant to the Ag+ cations and trap states in CuI films. There- tinct illumination power of blue light from 10% to 100%. This
fore, we speculate that one of the possible causes of the vari- authenticates the practical sensitivity of CuI-based memory
ation in conductivity with temperature is related to the point devices, toward the wavelength of 455 nm, with respect to light-
defects and non-stoichiometry present in the CuI film, which assisted multi-level data storage. These results suggest repro-
needs further study. Accordingly, there is a high chance that VI ducible, discrete current levels and multi-level HRS can be
and Vcu defects significantly contribute to the charge transport achieved by controlling both the illumination power intensity
in the low-temperature region (203 to 273 K).[24,27] and wavelength for the optoelectronic device in an engineering
The Richardson plot of ln (I/T2) versus 1000/T at applied mode.
voltages of 0.02, 0.04, 0.06, and 0.1 V exhibit two distinct linear Figure 5c shows the corresponding LRS and HRS in the dark
regimes for high and low temperatures, respectively, and state, at 50% and 100% illumination power intensity. The LRS
accordingly, the calculated barrier height, ϕb, can be extracted shows consistent behavior for dark and light conditions with
for each bias from the slope of these linear plots. In this way, no variation. However, the HRS current level exhibits three
the trend of Schottky emission distance was assessed from the different stages under light exposure from 20 to 56 nA at a
Adv. Funct. Mater. 2023, 33, 2211022 2211022 (7 of 12) © 2022 Wiley-VCH GmbH
16163028, 2023, 8, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adfm.202211022 by Chungnam National University, Wiley Online Library on [20/02/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
www.advancedsciencenews.com www.afm-journal.de
Figure 5. a) Schematic illustration of the experimental setup for photo-memristive characterization of ITO/CuI/Ag devices. b) Typical I-V character-
istics of the ITO/CuI/Ag device measured in the dark condition and under a blue light (light wavelength: 455 nm) with irradiance varying from 3.3 to
6.6 mW cm−2. c) HRS and LRS value in dark conditions, 50% and 100% power illumination with 455 nm (blue light). d) Light-dependent I-V charac-
teristics of the ITO/CuI/Ag devices under consecutive dark-blue-dark conditions illustrating complete recovery of the HRS. e,f) Set and reset voltage
distribution of the ITO/CuI/Ag devices in dark and illumination conditions. g) Photoresponse of CuI RRAM under different illumination power (455 nm)
at 10%, 50%, and 100%. h,i) Schematics energy band diagram for the dark and light conditions for the ITO/CuI/Ag devices.
read voltage of 0.05 V, enabling light-assisted multi-level data Figure 5e,f. The Vset in both dark and light conditions exhib-
storage. The wavelength dependency test reemphasizes the ited normal distribution, with an average value of 0.313 V for
response of CuI toward the blue light, due to direct band gap dark and 0.29 V for illumination conditions, respectively. How-
properties and various trap states (as discussed in Figure S9, ever, the average Vreset had a decreasing tendency for the blue
Supporting Information). In particular, the illumination at dif- light condition from −0.20 to −0.13 V, which demonstrates an
ferent excitation wavelengths (λ = 355, 540, 656, 740, 850, and increase in uniformity in reset behavior in the light illumina-
940 nm) either failed to induce the repetitive RS (λ = 355 nm) tion condition. Conclusively, the variation of distribution for
or did not show any distinguishable HRS sta in the ITO/CuI/ Vset/reset was noticeably reduced under blue light illumination,
Ag devices. leading to better device operationality.
The I-V characteristics of the ITO/CuI/Ag devices dis- Figure 5g shows the time-dependent photoresponse of CuI
played near-perfect recovery after consecutive cycling in dark for different illumination powers of 10%, 50%, and 100%.
conditions, blue light illumination, and again dark condi- The Off current was 8 nA, which increased to 13.9, 22.04, and
tion, as shown in Figure 5d. Notably, the average Vset/reset for 28 nA corresponding to the respective illumination powers
the dark and light conditions was evaluated by measuring of 10%, 50%, and 100%, within the 0.2 s duration. The decay
over 50 assembles (i.e., ITO/CuI/Ag device cells), as shown in time to reach the Off state current level (8 nA), defined as 10%
Adv. Funct. Mater. 2023, 33, 2211022 2211022 (8 of 12) © 2022 Wiley-VCH GmbH
16163028, 2023, 8, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adfm.202211022 by Chungnam National University, Wiley Online Library on [20/02/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
www.advancedsciencenews.com www.afm-journal.de
of the On current level under illuminations, was consistently substrate at various bending radii, 10.8, 6.6, and 4.5 mm.
observed within 0.02 s, which corroborates the outstanding Effective RS performance was observed with no noticeable deg-
photonic property of CuI. However, a 10–15% decay in the radation even after bending the substrate at a high bending
photocurrent during the light turn-on period was noticed, and radius and tensile strain, ε = 0.83%, corresponding to 4.55 (the
the time constants for the same were estimated to be 12.8, detailed bending stress is elaborated in Figure S11a–d, Sup-
13.4, and 13.9 s, corresponding to the 10%, 50%, and 100% porting Information).
power illumination levels (see Figure S10a, Supporting Infor- As shown in Figure 6d, consecutive RS cycling was evalu-
mation). The photocurrent decay and change in Vset or (Vreset) ated for 30 cycles at different bending radii of 10.8 and 6.6 mm,
(reported in Figure 5g) reaffirm the light-assisted migration of and 40 cycles of extreme bending at a radius of 4.5 mm. The
point defects, implying the movement of charge carriers. How- ITO/CuI/Ag devices displayed no sign of noticeable degrada-
ever, negligible persistent photocurrent effects were observed, tion in terms of RS performance such as HRS, LRS, On/Off
meaning the RRAM under illumination can be appealing for ratio, or set or reset voltage, demonstrating outstanding flexible
practical application. performance. Despite the extreme bending in the last 40 cycles
The variation in HRS current levels under blue light at a 4.5 mm radius, only a marginal change in the LRS was
(455 nm) illumination was studied in detail with the help of observed even though other RS properties were successfully
band diagrams (as shown in Figure 5h,i), respectively, for retained.
the dark and light conditions. During the dark condition, the Long-term stability is crucial for the successful operation
HRS current levels were ascribed to the migration of Ag+ of nonvolatile memory in practical applications. Ag is a well-
ions and point defects. In contrast, under illumination, the known electrochemically active metal, which can readily react
photo-generated charge carriers, defects, and trap states (i.e., with CuI and form AgIx, degrading device performance over
VCu, VI, CuI, and ICu) present in CuI contribute to the cur- time.[65–67] Thus, the I–V characteristics of the ITO/CuI/Ag
rent level at HRS.[30] The possible mechanism involved in the device were examined for up to 15 days to evaluate (Figure S12,
blue light illumination, as indicated in Figure 5i, well matches Supporting Information) the short-term lifespan associated
the inter-band transition and transition-related trap level near with the device longevity, without passivation, and with PMMA
the conduction band edge of VCu located near the valence encapsulation. For the non-passivated configuration, the ITO/
band.[30,64] CuI/Ag devices completely lost their RS performance within
To address practical aspects of the CuI RRAMs, we suc- 5 days (Figure S12a–c, Supporting Information). The I–V char-
cessfully demonstrated flexible ITO/CuI/Ag RRAM devices acteristics of the ITO/CuI/Ag devices started to show noticeable
fabricated on a PET/ITO substrate. The schematic and picto- degradation on the second day after device fabrication. How-
rial image of the flexible configuration of a CuI-based RRAM ever, the ITO/CuI/Ag/PMMA devices with encapsulation suc-
is illustrated in Figure 6a,b. Figure 6c shows the I–V charac- cessfully retained the RS performance for over 15 days without
teristics of ITO/CuI/Ag devices during 100 cycles on a flexible significant changes in the HRS and LRS value even in ambient.
Figure 6. Flexible Device Demonstration. a,b) Schematic and optical image of the flexible ITO/CuI/Ag device fabricated on PET substrate. c) Typical I-V
characteristics of 100 cycles of the PET-/ITO/CuI/Ag device. d) Reversible HRS and LRS of ITO/CuI/Ag device on a flexible PET substrate for 100 cycles
at a different bending radius of 10.8, 6.6, and 4.5 mm.
Adv. Funct. Mater. 2023, 33, 2211022 2211022 (9 of 12) © 2022 Wiley-VCH GmbH
16163028, 2023, 8, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adfm.202211022 by Chungnam National University, Wiley Online Library on [20/02/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
www.advancedsciencenews.com www.afm-journal.de
In addition, CuI is highly susceptible to humidity, which rap- use in-memory technologies which require ultra-low power
idly degrades CuI film when exposed to air ambient. The mois- consumption for next-generation computing systems.
ture present in air can trigger a reaction between the CuI and
Ag, resulting in accelerated degradation.[68] Obviously, the rapid
degradation of the memory device is a result of the reaction of 4. Experimental Section
the top Ag metal electrode with the CuI layer, which eventually
becomes hyperactive in the presence of moisture in ambient Film Formation and Device Fabrication: ITO-coated glass substrates
(sheet resistance of 10 ohm sq−1, thickness 0.7 mm, purchased from
conditions. This could hinder the reliable performance of CuI
Omniscience Korea) were utilized as the bottom electrodes. The
in an ambient atmosphere. ITO-coated substrates were rinsed via ultrasonication by acetone,
Among a variety of encapsulation methods, PMMA over- isopropanol, and deionized water and then dried up with N2 gas before
coating, followed by additional annealing, has been popularly synthesizing CuI films. CuI film was synthesized on the cleaned ITO-
employed as a convenient way of improving short-term device glass substrates by transforming copper (Cu) into copper iodide at
longevity in various applications by blockading moisture dif- room temperature. First, the Cu film was deposited on the cleaned
fusion.[69,70] Thus, in the present work, a similar scheme of ITO-glass substrate by thermal evaporation at a rate of 0.15 Å s−1 under
the vacuum (1 × 10−6 Torr). The evaporated copper film was set in an
PMMA encapsulation was adopted for CuI RRAMs, and the iodine atmosphere under a vacuum-driven iodine evaporation technique
encapsulated devices with the ITO/CuI/Ag/PMMA configura- commonly known as the Iodination process.[74] The reaction between Cu
tion exhibited better operating durability. Figure S12d–f (Sup- and solid iodine occurs in the vacuum desiccator under a medium-level
porting Information) shows the I–V characteristics of a PMMA vacuum. The vacuum level not only prevents the oxidation of copper
passivated ITO/CuI/Ag device in air, confirming the prolonged and iodine but is also responsible for driving the iodine evaporation
operating stability of the RRAM device even after 15 days. The rate. At high temperature growth conditions, the aggregation of CuI film
leads to uneven growth of CuI. Thus, the room temperature vacuum-
PMMA encapsulation, which is frequently employed for low driven iodination process out matches film quality, obtained through
dimensional carbon nanotubes (or air-sensitive thin film) tran- high-temperature process.[31] The iodination time directly impacts the
sistors, effectively prevents moisture penetration to the surface thickness and roughness of CuI film.[39] However, as the thickness
of the CuI/Ag film.[71–73] The PMMA encapsulated device con- increases, the CuI film epitaxial growth might be curtailed due to the
sequentially showed stable RS, which ensures the reduction strain effect, even though the RRAM device demonstrates a wider
in the reaction rate of the CuI layer with the top Ag electrode. memory window. Later, the shadow mask is utilized to deposit the top
electrode (90 nm thick Ag) by thermal evaporation.
Conclusively, PMMA encapsulation can effectively minimize
Device Stability: The device stability of CuI-based RRAM was enhanced
the degradation of device performance, leading to preserving by coating with a PMMA (Poly (methyl methacrylate)) encapsulation
the RS performance CuI-based memory device for longevity, layer. The PMMA solution 2 wt.% (weight/volume ratio, d 1.17 g ml−1,
besides allowing device characterization at low temperatures. purchased from Sigma Aldrich) was spin-coated over ITO/CuI/Ag device
configuration, followed by baking at 110 °C for 1 h in air. The passivated
devices were stored under ambient atmospheric conditions to evaluate
the effect of humid air and moisture. In the case of non-passivated
3. Conclusions ITO/CuI/Ag-based devices, the degradation is accelerated owing to the
catalytic effect of moisture owing to the reaction between CuI film and
Forming-free, light-mediated binary halide CuI RRAMs with Ag-top electrodes.
an operating capability within 0.5 V were implemented using Film Characterization: The CuI film was grown on ITO-glass substrate
CuI film which developed column-shaped crystal features via a followed by vacuum annealing at 120 °C. X-ray diffraction (XRD)
solid iodination process. This device scheme for CuI RRAMs is measurements on a RIgaku Smart Lab with Cu Kα radiation (λ = 1.54 Å),
anticipated to open light-mediated multifunctional analog (or/ characterized the phase and crystalline structure. XRD peak data were
recorded in a 2θ range of 20°–80° at room temperature with a step size
and digital) circuit systems, and furthermore, support a desir- of 0.03° and a scan speed of 3 min−1. Field emission scanning electron
able increase in NVM memory density. This study contributes microscopy (FE-SEM) (JEOL/JSM-7800F) investigated the surface
to the development of the envisioned application, required and cross-sectional images. The elemental mapping and images were
for ultra-low power and extreme memory density, such as obtained at an accelerating voltage of 10.0 kV. XPS elemental study was
autonomous driving, and neuromorphic (or security)-relevant performed on ULVAC-PHI 5000 Versa Probe PHI at 10–8 Pa with the
systems embedded with PIM. A conventional PMMA encap- incident energy of 600 eV.
Device Characterization: The electrical properties were characterized
sulation scheme was employed for the CuI RRAMs to allow
by a semiconductor parameter analyzer (Agilent 4155B) in ambient air.
low-temperature characterization. The results revealed vital For the low-temperature measurement, the I–V characteristics were
parameters (e.g., activation energy, Schottky barrier height, etc) measured by a semiconductor analyzer (4200-SCS, Keithley) using a
and elucidated the underlying mechanisms of current conduc- low-temperature cryogenic vacuum chamber (CRX-6.5K, Lake Shore) at
tion, which were associated with point defects and RS mecha- a vacuum level of 6 × 10−2 Torr. The photoconductive effect on the RS
nism. These findings were consistent with the migration and of CuI RRAM devices was studied under laser light source by the multi-
mobility of silver ions, device longevity, and other features. In wavelength fiber-coupled LED source (Mightex) with various wavelengths
such as 455 (blue), 540 (green), 656 (red), 740, 850, and 940 nm. The
parallel, we successfully demonstrated a flexible RRAM and the calibrated power intensity for 455 nm laser was 6.60 mW cm−2.
photo memory effect on RRAM, exploiting the optoelectronic
properties of the CuI. The results pave the way for application
in various internet-of-things (IoT), including flexible, wear-
able, and smart electronics. This research work is expected Supporting Information
to subdue the understanding of CuI and other halide-based Supporting Information is available from the Wiley Online Library or
RRAM for low-voltage RS and illustrate their potentially wide from the author.
Adv. Funct. Mater. 2023, 33, 2211022 2211022 (10 of 12) © 2022 Wiley-VCH GmbH
16163028, 2023, 8, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adfm.202211022 by Chungnam National University, Wiley Online Library on [20/02/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
www.advancedsciencenews.com www.afm-journal.de
Acknowledgements [18] N. Yamada, Y. Kondo, R. Ino, Phys. Status Solidi A 2019, 216,
1700782.
This research was supported by the Incheon National University [19] J. A. Christians, R. C. M. Fung, P. V. Kamat, J. Am. Chem. Soc. 2014,
Research grant in 2020, Incheon, Republic of Korea. 136, 758.
[20] W. Y. Chen, L. L. Deng, S. M. Dai, X. Wang, C. B. Tian, X. X. Zhan,
S. Y. Xie, R. Bin Huang, L. S. Zheng, J. Mater. Chem. A 2015, 3,
19353.
Conflict of Interest [21] Y. Zhang, S. Li, W. Yang, M. K. Joshi, X. Fang, J. Phys. Chem. Lett.
The authors declare no conflict of interest. 2019, 10, 2400.
[22] A. Bala, P. Pujar, D. Daw, Y. Cho, M. Naqi, H. Cho, S. Gandla,
S. Kim, ACS Appl. Electron. Mater. 2022, 4, 3973.
[23] Y. Zhang, L. Liu, Z. Wang, Y. Yang, F. Xing, J. Alloys Compd. 2022,
Data Availability Statement 900, 163456.
[24] C. Yang, M. Kneib, M. Lorenz, M. Grundmann, Proc. Natl. Acad. Sci.
The data that support the findings of this study are available from the
USA 2016, 113, 12929.
corresponding author upon reasonable request.
[25] M. Rahman, M. A. Newaz, B. K. Mondal, A. Kuddus, M. A. Karim,
M. M. Rashid, M. H. K. Rubel, J. Hossain, Mater. Res. Bull. 2019,
118, 110518.
Keywords [26] M. Grundmann, F. L. Schein, M. Lorenz, T. Böntgen, J. Lenzner,
H. Von Wenckstern, Phys. Status Solidi A 2013, 210, 1671.
CuI-based nonvolatile memory, forming-free, photo responsive RRAM [27] D. K. Kaushik, M. Selvaraj, S. Ramu, A. Subrahmanyam, Sol. Energy
with multi-level data storages, ultra-low power RRAM devices Mater. Sol. Cells 2017, 165, 52.
[28] V. Gottschalch, S. Blaurock, G. Benndorf, J. Lenzner,
Received: September 22, 2022 M. Grundmann, H. Krautscheid, J. Cryst. Growth 2017, 471, 21.
Revised: October 29, 2022 [29] P. M. Sirimanne, M. Rusop, T. Shirata, T. Soga, T. Jimbo, Chem.
Published online: November 28, 2022 Phys. Lett. 2002, 366, 485.
[30] K. X. Zhang, S. W. Wang, L. Y. Bai, Y. Wang, K. Ou, Y. W. Zhang,
L. X. Yi, J. Lumin. 2019, 214, 116522.
[31] S. D. Baek, D. K. Kwon, Y. C. Kim, J. M. Myoung, ACS Appl. Mater.
[1] Y. Li, Q. Qian, X. Zhu, Y. Li, M. Zhang, J. Li, C. Ma, H. Li, J. Lu, Interfaces 2020, 12, 6037.
Q. Zhang, InfoMat 2020, 2, 995. [32] B. M. M. Faustino, D. Gomes, J. Faria, T. Juntunen, G. Gaspar,
[2] Y. Sun, J. A. Rogers, Adv. Mater. 2007, 19, 1897. C. Bianchi, A. Almeida, A. Marques, I. Tittonen, I. Ferreira, Sci. Rep.
[3] K. Kobashi, R. Hayakawa, T. Chikyow, Y. Wakayama, Nano Lett. 2018, 8, 6867.
2018, 18, 4355. [33] Y. An, H. Fei, G. Zeng, X. Xu, L. Ci, B. Xi, S. Xiong, J. Feng, Y. Qian,
[4] A. Kumar, M. Krishnaiah, D. Mishra, H. Jo, S. H. Jin, J. Alloys Nano Energy 2018, 47, 503.
Compd. 2022, 922, 166199. [34] K. kumar Chinnakutti, V. Panneerselvam, D. Govindarajan,
[5] A. Q. Jiang, C. Wang, K. J. Jin, X. B. Liu, J. F. Scott, C. S. Hwang, S. T. Salammal, Mater Today Proc 2019, 23, 34.
T. A. Tang, H. Bin Lu, G. Z. Yang, Adv. Mater. 2011, 23, 1277. [35] M. Kneiß, C. Yang, J. Barzola-Quiquia, G. Benndorf, H. von Wencks
[6] J. J. Yang, M. D. Pickett, X. Li, D. A. A. Ohlberg, D. R. Stewart, tern, P. Esquinazi, M. Lorenz, M. Grundmann, Adv. Mater. Interfaces
R. S. Williams, Nat. Nanotechnol. 2008, 3, 429. 2018, 5, 1701411.
[7] A. Wedig, M. Luebben, D.-Y. Cho, M. Moors, K. Skaja, V. Rana, [36] M. Sun, J. Hu, C. Zhai, M. Zhu, J. Pan, ACS Appl. Mater. Interfaces
T. Hasegawa, K. K. Adepalli, B. Yildiz, R. Waser, I. Valov, Nat. Nano- 2017, 9, 13223.
technol. 2016, 11, 67. [37] C. Roy, J. Galipaud, L. Fréchette-Viens, S. Garbarino, J. Qiao,
[8] M.-J. Lee, C. B. Lee, D. Lee, S. R. Lee, M. Chang, J. H. Hur, Y.-B. Kim, D. Guay, Electrochim. Acta 2017, 246, 115.
C.-J. Kim, D. H. Seo, S. Seo, U.-I. Chung, I.-K. Yoo, K. Kim, Nat. [38] M. Zi, J. Li, Z. Zhang, X. Wang, J. Han, X. Yang, Z. Qiu, H. Gong,
Mater. 2011, 10, 625. Z. Ji, B. Cao, Phys. Status Solidi A 2015, 212, 1466.
[9] F. Zeng, Y. Guo, W. Hu, Y. Tan, X. Zhang, J. Feng, X. Tang, ACS Appl. [39] F. L. Schein, H. Von Wenckstern, M. Grundmann, Appl. Phys. Lett.
Mater. Interfaces 2020, 12, 23094. 2013, 102, 092109
[10] J. Sheng, T. Hong, H.-M. Lee, K. Kim, M. Sasase, J. Kim, H. Hosono, [40] X. N. Zhao, Z. Q. Wang, W. T. Li, S. W. Sun, H. Y. Xu,
J.-S. Park, ACS Appl. Mater. Interfaces 2019, 11, 40300. P. Zhou, J. Q. Xu, Y. Lin, Y. C. Liu, Adv. Funct. Mater. 2020, 30,
[11] H.-Y. Chen, S. Yu, B. Gao, P. Huang, J. Kang, H.-S. P. Wong, in 1910151.
2012 International Electron Devices Meeting, IEEE, New York 2012, [41] J. Choi, S. Park, J. Lee, K. Hong, D. H. Kim, C. W. Moon, G. Do
pp. 20.7.1–20.7.4. Park, J. Suh, J. Hwang, S. Y. Kim, H. S. Jung, N. G. Park, S. Han,
[12] S. Wrede, H. Tian, Phys. Chem. Chem. Phys. 2020, 22, 13850. K. T. Nam, H. W. Jang, Adv. Mater. 2016, 28, 6562.
[13] B. A. D. Williamson, J. Buckeridge, J. Brown, S. Ansbro, [42] H. Y. Li, X. Di Huang, J. H. Yuan, Y. F. Lu, T. Q. Wan, Y. Li, K. H. Xue,
R. G. Palgrave, D. O. Scanlon, Chem. Mater. 2017, 29, 2402. Y. H. He, M. Xu, H. Tong, X. S. Miao, Adv. Electron. Mater. 2020, 6,
[14] S. Ves, D. Glötzel, M. Cardona, H. Overhof, Phys. Rev. B 1981, 24, 2000309.
3073. [43] K. M. Kim, S. R. Lee, S. Kim, M. Chang, C. S. Hwang, Adv. Funct.
[15] H. Wang, L. Zhang, X. Tan, C. M. B. Holt, B. Zahiri, B. C. Olsen, Mater. 2015, 25, 1527.
D. Mitlin, J. Phys. Chem. C 2011, 115, 17599. [44] L. Pauling, J. Phys. Chem. 1941, 45, 1142.
[16] S. Lee, H. J. Lee, Y. Ji, S. M. Choi, K. H. Lee, K. Hong, J. Mater. [45] J. J. O’Dwyer, The Theory of Electrical Conduction and Breakdown in
Chem. C 2020, 8, 9608. Solid Dielectrics, Clarendon Press, Oxford 1973.
[17] C. Yang, D. Souchay, M. Kneiß, M. Bogner, H. M. Wei, M. Lorenz, [46] J. S. Han, Q. Van Le, J. Choi, K. Hong, C. W. Moon,
O. Oeckler, G. Benstetter, Y. Q. Fu, M. Grundmann, Nat. Commun. T. L. Kim, H. Kim, S. Y. Kim, H. W. Jang, Adv. Funct. Mater. 2018, 28,
2017, 8, 4. 1705783.
Adv. Funct. Mater. 2023, 33, 2211022 2211022 (11 of 12) © 2022 Wiley-VCH GmbH
16163028, 2023, 8, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adfm.202211022 by Chungnam National University, Wiley Online Library on [20/02/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
www.advancedsciencenews.com www.afm-journal.de
[47] D. J. J. Loy, P. A. Dananjaya, X. L. Hong, D. P. Shum, W. S. Lew, Sci. [61] A. N. Ionov, I. S. Shlimak, in Hopping Transport in Solids
Rep. 2018, 8, 14774. (Eds.: M. Pollak, B. Shklovskii), 28th eds, Elsevier, Amsterdam 1991,
[48] R. Waser, R. Dittmann, G. Staikov, K. Szot, Adv. Mater. 2009, 21, pp. 397–421.
2632. [62] K. H. Chen, T. M. Tsai, C. M. Cheng, S. J. Huang, K. C. Chang,
[49] F. C. Chiu, Adv. Mater. Sci. Eng. 2014, 2014, 578168. S. P. Liang, T. F. Young, Materials 2017, 11, 43.
[50] X. Guan, W. Hu, M. A. Haque, N. Wei, Z. Liu, A. Chen, T. Wu, Adv. [63] M. Krishnaiah, A. Kumar, D. Mishra, A. K. Kushwaha, S. H. Jin,
Funct. Mater. 2018, 28, 1704665. J. T. Park, J. Alloys Compd. 2021, 887, 161326.
[51] H. Kim, J. S. Han, S. G. Kim, S. Y. Kim, H. W. Jang, J. Mater. Chem. [64] S. Koyasu, M. Miyauchi, J. Electron. Mater. 2020, 49, 907.
C 2019, 7, 5226. [65] Y. Kato, L. K. Ono, M. V. Lee, S. Wang, S. R. Raga, Y. Qi, Adv. Mater.
[52] T. Tsuruoka, K. Terabe, T. Hasegawa, M. Aono, Nanotechnology Interfaces 2015, 2, 1500195.
2010, 21, 425205. [66] J. Sun, Q. Liu, H. Xie, X. Wu, F. Xu, T. Xu, S. Long, H. Lv, Y. Li,
[53] J. S. Han, Q. Van Le, J. Choi, H. Kim, S. G. Kim, K. Hong, L. Sun, M. Liu, Appl. Phys. Lett. 2013, 102, 053502
C. W. Moon, T. L. Kim, S. Y. Kim, H. W. Jang, ACS Appl. Mater. Inter- [67] S. Gao, C. Song, C. Chen, F. Zeng, F. Pan, Appl. Phys. Lett. 2013,
faces 2019, 11, 8155. 102, 141606
[54] T. Tsuruoka, K. Terabe, T. Hasegawa, M. Aono, Nanotechnology [68] A. Crovetto, H. Hempel, M. Rusu, L. Choubrac, D. Kojda,
2011, 22, 254013. K. Habicht, T. Unold, ACS Appl. Mater. Interfaces 2020, 12, 48741.
[55] R. M. Hill, Phys. Status Solidi A 1976, 34, 601. [69] S. D. Stranks, G. E. Eperon, G. Giulia, M. Christopher,
[56] S. Ge, Y. Huang, X. Chen, X. Zhang, Z. Xiang, R. Zhang, W. Li, M. J. P. Alcocer, L. Tomas, L. M. Herz, P. Annamaria, H. J. Snaith,
Y. Cui, Adv. Mater. Interfaces 2019, 6, 1802071. Science 2013, 342, 341.
[57] S. Koyasu, N. Umezawa, A. Yamaguchi, M. Miyauchi, J. Appl. Phys. [70] F. Li, C. Ma, H. Wang, W. Hu, W. Yu, A. D. Sheikh, T. Wu, Nat.
2019, 125, 115101 Commun. 2015, 6, 8238.
[58] J. Wang, J. Li, S. S. Li, J. Appl. Phys. 2011, 110, 054907 [71] S. Tappertzhofen, I. Valov, T. Tsuruoka, T. Hasegawa, R. Waser,
[59] H. Chen, C. Y. Wang, J. T. Wang, Y. Wu, S. X. Zhou, Phys. B 2013, M. Aono, ACS Nano 2013, 7, 6396.
413, 116. [72] T. Tsuruoka, K. Terabe, T. Hasegawa, I. Valov, R. Waser, M. Aono,
[60] B. Roling, in Charge Transport in Disordered Solids with Applications in Adv. Funct. Mater. 2012, 22, 70.
Electronics (Ed: S. Baranovski), John Wiley & Sons, Hoboken, New [73] N. Onofrio, D. Guzman, A. Strachan, Nat. Mater. 2015, 14, 440.
Jersey 2006, pp. 379–401. [74] N. Shimidzu, H. Masuda, S. Ohno, Chem. Lett. 1984, 13, 961.
Adv. Funct. Mater. 2023, 33, 2211022 2211022 (12 of 12) © 2022 Wiley-VCH GmbH