Journal of

Materials Chemistry C
View Article Online
PAPER View Journal | View Issue
Published on 28 August 2024. Downloaded by CHUNGBUK NATIONAL UNIVERSITY on 7/31/2025 9:18:29 AM.

Fabrication of a high performance memristor

device by metallization of Ag+ inside a solution
Cite this: J. Mater. Chem. C,
2024, 12, 16145 processed Li5AlO4 thin film†
Subarna Pramanik, Rajarshi Chakraborty, Sobhan Hazra, Utkarsh Pandey
and Bhola Nath Pal *

A solution processed Ag-ion-exchanged Li5AlO4 thin film has been used to fabricate a high performance
oxide memristor device with enhanced bistable switching and memory retention. For this device
fabrication, an Li5AlO4 thin film has been deposited on a highly p-doped silicon (p+-Si) substrate
followed by an ion exchange and reduction process that allows the conversion of the Li5AlO4 thin film
to an Ag(Ag+)–Al2O3 film. The Ag(Ag+)–Al2O3 film is a mixed phase consisting of an Ag nanoparticle
embedded Al2O3 film and silver ions and is responsible for resistive switching whereas the p+-Si
substrate works as the bottom electrode. A silver electrode has been used on top of Ag(Ag+)–Al2O3/p+-
Si that works as the top electrode of this memristor device. The I–V characteristics of the device
Received 17th June 2024, demonstrate a reversible switching behaviour that remains stable even after 200 consecutive operation
Accepted 27th August 2024 cycles. Furthermore, the ratio between the forward and reverse sweeps of the device exceeds 103 orders
DOI: 10.1039/d4tc02527j of magnitude and the device has a stable data retention capability of B4  104 seconds (B12 hours)
maintaining its on/off ratio of B104. Also, this device effectively emulates the properties of ReRAM up to
rsc.li/materials-c 200 iterations.

1. Introduction optoelectronic logic operations,17,18 optogenetic-inspired synaptic

functions,19,20 etc. Besides, memristor devices have various
The exponential growth of information volume has led to an advantages such as non-volatility, rapid switching speeds, and
escalating demand for information storage devices,1,2 necessi- high endurance, offering an exciting avenue for the develop-
tating extensive research and development in the field of ment of next-generation digital memory and neuromorphic
memory storage devices.3,4 Among the various categories of computing systems.21,22 In addition, a pivotal characteristic of
memory devices such as flash memory,5,6 FeFETs,7,8 FeTFTs,9,10 memristive devices is their ability to retain information without
and many more, considerable attention has been directed requiring a continuous power supply. These devices are com-
toward resistive switching memory11–13 devices as potential monly characterized by two distinct features: the zero-crossing
contenders for the next-generation data storage solutions. current–voltage (I–V) property and I–V hysteresis, situated
Moreover, resistive memory technology holds immense promise within the first and third quadrants of the I–V curve. Apart
for advanced applications, including memristor devices,14–16 from this, by leveraging the analog property of memristive
devices, electronic synapse components can be designed for
School of Materials Science and Technology, Indian Institute of Technology application in neuromorphic computing.23
(Banaras Hindu University), Varanasi 221005, India. However, in conventional metal–insulator–metal memristors,24,25
E-mail: [email protected] which operate on a filamentary switching mechanism, issues
† Electronic supplementary information (ESI) available: The SEM image of the
arise with respect to fluctuating operating voltages and resis-
cross-sectional view of the device is provided in Fig. S1, the TEM image of the
Ag(Ag+)–Al2O3 thin film sample is provided in Fig. S2, the I–V curve in linear scale
tance states due to the unpredictable formation and disintegration
with different concentrations is provided in Fig. S3, the I–V curve in semi- of conductive filaments. In recent times, a variety of materials
logarithmic scale with different top electrode areas is provided in Fig. S4, the exhibiting resistive switching behavior have been discovered, with
I–V curve and endurance property with a different electrode is provided in Fig. S5, a particular focus on materials like aluminum,26,27 tantalum,28,29
device to device variation data are provided in Fig. S6, the complete dataset for
and hafnium-based oxides.30,31 Despite their numerous advan-
pulse switching endurance cycles is available in Fig. S7, the statistical histogram
for the device p+-Si/(Ag/Ag+)–Al2O3/Ag is represented in Fig. S8, and the I–V curve
tages, oxide-based memory materials have encountered several
for the device p+-Si/Ag(Ag+)–Al2O3/Ag under hot air treatment is provided in drawbacks, including poor repeatability, non-uniformity, and lim-
Fig. S9. See DOI: https://doi.org/10.1039/d4tc02527j itations in data retention capabilities.32,33

This journal is © The Royal Society of Chemistry 2024 J. Mater. Chem. C, 2024, 12, 16145–16155 | 16145
 View Article Online

Paper Journal of Materials Chemistry C

In this work, a solution processed Li5AlO4 thin film has been 10 minutes) to remove organic substances and make the
deposited on a highly p-doped Si (p+-Si) substrate to fabricate a surfaces hydrophilic. The hydrophilic surfaces of p+-Si allow
high performance memristor device. Li5AlO4 is a well-known deposition of a pin-hole-free thin film of Li5AlO4 which is an
ion-conducting oxide electrolyte material that contains highly important step in the memristor device fabrication. After the
mobile Li+ through its crystal channel.34 The mobile Li+ of the cleaning process, the precursor sol of Li5AlO4 is spin coated
Published on 28 August 2024. Downloaded by CHUNGBUK NATIONAL UNIVERSITY on 7/31/2025 9:18:29 AM.

Li5AlO4 thin film is replaced by Ag+ through an ion-exchange over these substrates at a spinning speed of 2500 rpm for one
process that has been metallized after the reduction process minute under ambient atmospheric conditions and dried at
resulting in an Al2O3 embedded Ag nanoparticle thin film 90 1C for two minutes to remove the solvent. After spin coating,
Ag(Ag+)–Al2O3.35,36 A silver electrode has been deposited on the precursor film is annealed at 350 1C for 30 minutes. This
top of this Ag(Ag+)–Al2O3/p+-Si film to complete the device process is repeated three times to achieve the desired thickness
fabrication. The ambient condition I–V characteristics of this of the Li5AlO4 thin film. Finally, thin films are annealed
device indicate its bistable resistive memory switching with a at 500 1C for one hour to get a polycrystalline film of Li5AlO4.
high on/off ratio and repeatability. Besides, devices show high After that, the thin films are dipped inside a 100 mM silver
memory retention with high endurance capability. nitrate [AgNO3] solution for 45 minutes to initiate the ion
exchange process. Through the ion-exchange process, silver
ions (Ag+) from the solution interchange with mobile lithium
2. Experimental section ions (Li+) from the Li5AlO4 thin film to form an Ag ion-
2.1. Material synthesis exchanged film (Ag5AlO4). Ag5AlO4 is thermodynamically
unstable and partly converts to the Ag(Ag+)–Al2O3 phase
The Li5AlO4 dielectric thin films are deposited by a sol–gel (Ag/Ag+–Al2O3) which is confirmed through XRD (discussed in
technique followed by an annealing process.34 The thin film the Results and discussion section). After this ion exchange
synthesis method allows the precursors to react uniformly procedure, the samples are rinsed with DI water to eliminate
at the molecular level and form a polycrystalline single any excess AgNO3 from the film’s surface. After washing, the
phase compound of Li5AlO4. Aluminum nitrate nonahydrate substrates are immersed in a 300 mM NaBH4 solution at
[Al(NO3)39H2O] (acquired from S D Fine Chem Limited, ambient temperature for one hour. During this process, the
498.0% pure) and lithium(II) acetate [CH3COOLi] (acquired remaining Ag+ has been transformed to metallic silver (Ag0) to
from TCI, 498.0% pure) are used as precursor materials for the form Ag-NPs, and the initial Ag5AlO4 film is transformed to
synthesis process. In the beginning, the Yoldas method is form the final Al2O3 embedded Ag nanoparticle thin film
employed to prepare a 300 mM alumina sol. Besides, a 300 mM Ag(Ag+)–Al2O3. After reduction, the films are cleaned by dipping
lithium acetate solution is prepared by using 2-methoxy etha- inside DI water to eliminate any remaining NaBH4 from the
nol as a solvent. These two solvents are then mixed in a 1 : 5 surface of the film. Finally, Ag electrodes of thickness 100 nm
ratio, so that the final product (Li5AlO4) maintains its stoichio- are deposited through the thermal evaporation method on top
metric ratio. To create a transparent, homogeneous precursor of Ag(Ag+)–Al2O3 with an electrode area of B0.71 mm2
solution of Li5AlO4, the resulting mixture is stirred at room (0.84 mm  0.84 mm) (Fig. 1).
temperature for 30 minutes. This mixture is then fabricated Ion exchange and reduction reaction:
on a heavily boron-doped p-type silicon (p+-Si) substrate after
being filtered through a PVDF syringe filter (0.45 mm).
To perform the ion exchange and reduction process of the
Li5AlO4 thin films, a silver nitrate [AgNO3] (acquired from
(1)
Sigma-Aldrich, 98% purity) solution (100 mM) is prepared
using deionized water whereas an aqueous solution of
300 mM sodium borohydride (crystals) [NaBH4] (acquired from
Sisco Research Laboratories Pvt. Ltd 498.0% pure) has been (2)
prepared for the reduction process.

2.2. Device fabrication 2.3. Characterization

+
For the memristor device fabrication, heavily boron-doped p -Si A bench-top X-ray diffraction (XRD) (Rigaku, Smart Lab) setup
wafers (h100i) (15 mm  15 mm, resistivity B0.01–0.02 O cm, has been used to investigate the phase and crystallinity of
and thickness 705–745 mm) have been used as substrates that different materials. The UV-visible analysis is performed using
also work as the bottom electrode. Before spin coating, the Jasco V-650 spectrophotometers in the spectral range of 200–
substrates are cleaned with soap solution followed by rinsing in 800 nm. An NTMDT-NTEGRA-prima atomic force microscope
distilled (DI) water. Afterwards, substrates are cleaned using an (AFM) has been employed to analyze the surface topographies
ultrasonic bath with the help of DI water, acetone, and iso- of thin films. High-resolution scanning electron microscopy
propyl alcohol (IPA) respectively in different individual steps.37 (HR-SEM; NOVA NANOSEM 40, FEITM) is conducted for SEM
After wet cleaning, the substrates are placed inside an oxygen imaging. High-resolution transmission electron microscopy
plasma chamber for oxygen plasma treatment (20 W for (HR-TEM; Tecnai G2 20 TWIN, 200 kV) is utilized for TEM

16146 | J. Mater. Chem. C, 2024, 12, 16145–16155 This journal is © The Royal Society of Chemistry 2024
 View Article Online

Journal of Materials Chemistry C Paper

Published on 28 August 2024. Downloaded by CHUNGBUK NATIONAL UNIVERSITY on 7/31/2025 9:18:29 AM.

Fig. 1 Schematic configuration of the (a) reference device based on Li5AlO4 dielectric (p+-Si/Li5AlO4/Ag), (b) device after ion exchange based on
Ag5AlO4 dielectric (p+-Si/(Ag/Ag+)–Al2O3/Ag) and (c) proposed device based on Ag(Ag+)–Al2O3 dielectric (p+-Si/Ag(Ag+)–Al2O3/Ag) respectively.

imaging. A Thermo Scientific K-a apparatus equipped out of those the peaks at 2y E 31.921 and 57.671 correspond to
with a monochromatic Al X-ray source was used to conduct the Al2O3(220) and Al2O3(116) planes, matched from the JCPDS
X-ray photoelectron spectroscopy (XPS) studies. A semiconduc- data (file no. 79-1558). The other two peaks are located at 2y E
tor parametric analyzer (Keysight B1500A) and a dual 37.761 and 43.71 which correspond to Ag(111) and Ag(200)
source meter (Keysight B2912A) are used to perform all elec- respectively (JCPDS file no. 89-7322). These data indicate
trical characterization studies and retention time measure- metallic silver and Al2O3 formation in the Ag-ion exchange
ments of the memristor under ambient conditions. The set film. However, after reduction, only two highly intense XRD
and reset probability curves have been fitted by a Gaussian peaks located at 2y E 37.761 and 43.71 are observed which
curve using Python. The ReRAM study of the device has been correspond to the Ag(111) and Ag(200) planes (JCPDS 89-7322).
performed using an AFG 31000 SERIES Arbitrary Function These peak intensities are B3 times higher than those of ion-
Generator. exchange samples, implying metallization of most of the Ag+ in
the film. This comparative XRD study clearly indicates Ag+ are
partly formed Ag metal during ion-exchange process and mostly
3. Results and discussion remains in the ionic form ((Ag/Ag+)–Al2O3). However, after
reduction surface Ag+ are mostly form metallic silver (Ag) and
3.1. XRD and UV-Vis absorption
deeper Ag+ of this film remain in the ionic form which is
Crystalline phase identification of all different thin films has further investigated through the XPS study, is presenting as
been conducted through an XRD study which is shown in Ag(Ag+)–Al2O3.
Fig. 2(a). Initially, a Li5AlO4 thin film is prepared on a p+-Si Normalized UV-Vis absorption spectra of Li5AlO4 and
substrate exactly in a similar way to that of device fabrication. Ag(Ag+)–Al2O3 thin films that are deposited on a quartz sub-
The XRD data of the sample indicate polycrystalline thin film strate are depicted in Fig. 2(b). The data indicate strong UV-
formation with intense peaks at 2y E 21.041, 22.941, 31.31, absorption of the Li5AlO4 thin film in the 200–350 nm region,
39.461 and 44.761 corresponding to the (102), (112), (113), (132) whereas the absorption band of the Ag(Ag+)–Al2O3 thin film
and (142) planes of the crystalline phase of Li5AlO4, identified exhibits a strong visible range absorption. The absorption band
from the JCPDS data file 70-0432.34 However, after Ag-ion of the Ag(Ag+)–Al2O3 thin film is extended up to B600 nm with
exchange of the Li5AlO4 thin film, it shows a completely a peak position at 375 nm which originates due to the plas-
different XRD pattern, indicating a new phase formation from monic absorption of Ag NPs. Although the plasmonic peak for
the earlier. As it shows, four intense peaks are generated, and silver NPs is commonly observed in the range of 400–500 nm,

Fig. 2 (a) Thin film XRD patterns of Li5AlO4, (Ag/Ag+)–Al2O3, and Ag(Ag+)–Al2O3 thin films, coated on the silicon (p+-Si) substrate. (b) UV-Vis analysis of
Li5AlO4 and Ag(Ag+)–Al2O3 thin films, coated on a quartz substrate.

This journal is © The Royal Society of Chemistry 2024 J. Mater. Chem. C, 2024, 12, 16145–16155 | 16147
 View Article Online

Paper Journal of Materials Chemistry C

the blue shift in the plasmonic peak is caused due to the nanoparticles is around 10 nm as obtained from the TEM
change in the electronic environment around silver atoms image of the Ag(Ag+)–Al2O3 film sample (Fig. S2, ESI†).
caused due to the surrounding Al2O3 layer.38
3.3. X-ray photoemission spectroscopy (XPS)
To verify the chemical state of the containing element in the
3.2. Surface morphology study through AFM, particle size
Published on 28 August 2024. Downloaded by CHUNGBUK NATIONAL UNIVERSITY on 7/31/2025 9:18:29 AM.

Ag(Ag+)–Al2O3 thin film, X-ray photoelectron spectroscopy (XPS)

analysis and height profile is utilized. In this study, the Ag(Ag+)–Al2O3 thin film sample is
Atomic force microscopy (AFM) is employed to examine the prepared by spin-coating the Li5AlO4 precursor on a silicon
surface roughness of an ion-conducting dielectric thin film substrate and subsequent ion-exchange and reduction process,
under three distinct conditions. The initial thin film comprised exactly in a similar way as this memristor device is fabricated
solely of Li5AlO4 (p+-Si/Li5AlO4), the second device (p+-Si/(Ag/ excluding top electrode deposition. The XPS spectra of Ag 3d,
Ag+)–Al2O3) underwent an ion exchange process to produce the Al 2p, and O 1s are shown in Fig. 4 respectively. The two strong
(Ag/Ag+)–Al2O3 thin film, and the final device configuration (p+- peaks in Ag 3d shown in Fig. 4(a) are located at B374.1 and
Si/Ag(Ag+)–Al2O3) incorporated reduced Ag, resulting in the B368.1 eV, respectively, and represent the binding energies of
formation of Ag(Ag+)–Al2O3, denoted as S1, S2, and S3, respec- Ag 3d3/2 and Ag 3d5/2. Additionally, there are four distinct peaks
tively. Fig. 3 illustrates representative images of all samples. that may be separated from the 3d5/2 and 3d3/2 peaks. The
Sample S2 (1.07 nm) and S3 (1.03 nm) exhibited similar root- metallic Ag0 is responsible for the peaks at around B368.2 and
mean-square (RRMS) roughness values which are much higher B374.3 eV. The peaks at approximately 367.3 and 373.9 eV are
compared to that of sample S1 (0.76 nm). The increased RRMS related to Ag+.39,40 Since Ag+ has a higher peak intensity than
values in samples S2 and S3 can be attributed to the formation Ag0, it can be speculated that most of the silver atoms in the
of voids and defects in the film during the ion exchange and deeper part of the film are in ionic form. An intense peak at
reduction process. Cross-sectional SEM analysis, shown in around B74.25 eV may be seen in Al 2p, as shown in
Fig. S1 (ESI†), reveals an average thickness of 70 nm for the Fig. 4(b).39,41 Deconvolution of the O 1s spectrum yields two
Ag(Ag+)–Al2O3 layer, whereas the average particle size of silver peaks with binding energies of approximately 530.1 eV and

Fig. 3 (a) and (b) 2-D and 3-D AFM images of p+-Si/Li5AlO4, (c) and (d) 2-D and 3-D AFM images of p+-Si/(Ag/Ag+)–Al2O3, and (e) and (f) 2-D and 3-D
AFM images of p+-Si/Ag(Ag+)–Al2O3.

16148 | J. Mater. Chem. C, 2024, 12, 16145–16155 This journal is © The Royal Society of Chemistry 2024
 View Article Online

Journal of Materials Chemistry C Paper

Published on 28 August 2024. Downloaded by CHUNGBUK NATIONAL UNIVERSITY on 7/31/2025 9:18:29 AM.

Fig. 4 High-resolution XPS spectra of the Ag(Ag+)–Al2O3 sample: (a) Ag 3d, (b) Al 2p, and (c) O 1s.

531.3 eV, respectively, corresponding to lattice oxygen and of Device-2) and Ag NPs (in the case of Device-3) catalyses the
oxygen vacancy (Fig. 4c).39,42,43 The peak positions of all these formation of a conductive filament. This filamentary structure,
systems match those of the previously published Ag–Al2O3 under the influence of bias voltage, undergoes a cyclic for-
systems quite closely.39 Thus, the XPS analysis of the sample mation and rupture process. The formation of the filament
provides additional evidence supporting the thin-film Ag(Ag+)– represents the SET process of the device and is named the high
Al2O3 synthesis. It also clearly shows in this thin film the ratio resistance state (HRS). In contrast to this process, the rupture
of Ag+ is higher than Ag0. represents the RESET process and is termed the low resistance
state (LRS). Consequently, the inception of Ag ions and Ag NPs
empowers the device to prominently showcase memristive
3.4. Electrical characteristics properties. The device performance has also been studied by
Ambient atmosphere electrical characterization has been per- varying the concentration of the switching layer (100 mM,
formed to realize the resistive switching properties of these 200 mM, and 500 mM), as pictured in Fig. S3 (ESI†). The
devices. For this study, two reference devices have been fabri- data reveal no significant switching behaviour in any of the
cated: one of them by using solely the Li5AlO4 thin film and the devices, indicating that resistive switching (RS) occurs only at a
other one with the (Ag/Ag+)–Al2O3 film. Devices with active particular concentration of Ag(Ag+)–Al2O3. The performance
materials of Li5AlO4, (Ag/Ag+)–Al2O3 and Ag(Ag+)–Al2O3 are analysis has also been studied by altering the top electrode
referred to as Device-1, Device-2 and Device-3 respectively. area (Fig. S4, ESI†).
Fig. 5 shows the current–voltage (I–V) characteristics of these The endurance of a memristor that determines the stability
devices on a linear scale for a single current–voltage iteration of the device under consecutive resistive switching cycles is one
cycle within the voltage range of 4 V. The red-coloured arrows of the most important studies to realize the performance and
show the direction of the applied voltage sequences and the reliability of the device. Therefore, the resistive switching
corresponding current. In this investigation, Device-1 did not behaviour of these devices has been investigated through
exhibit any discernible switching behaviour (Fig. 5a), but repeated cyclic I–V characterization. As described in the earlier
Device-2 and 3 demonstrated distinctly robust switching beha- section, the resistive switching behaviour of Device-1 is not
viour (Fig. 5b and c) respectively). The switching characteristics observed; therefore, these endurance studies have been inves-
are found at B4 V of external bias, both for Device-2 and 3. tigated for Device-2 and 3. The semi-log plot of such I–V
This behavior implies that the introduction of Ag+ (in the case characteristics of Device-2 is plotted in Fig. 6(a). This study

Fig. 5 Linear plot of current vs. voltage curve for (a) p+-Si/Li5AlO4/Ag, (b) p+-Si/(Ag/Ag+)–Al2O3/Ag, and (c) p+-Si/Ag(Ag+)–Al2O3/Ag, respectively.

This journal is © The Royal Society of Chemistry 2024 J. Mater. Chem. C, 2024, 12, 16145–16155 | 16149
 View Article Online

Paper Journal of Materials Chemistry C

Published on 28 August 2024. Downloaded by CHUNGBUK NATIONAL UNIVERSITY on 7/31/2025 9:18:29 AM.

Fig. 6 (a) Current vs. voltage curve of p+-Si/(Ag/Ag+)–Al2O3/Ag on a semi-logarithmic scale and (b) endurance property of p+-Si/(Ag/Ag+)–Al2O3/Ag for
100 cycles of continuous measurement.

reveals that Device-2 is capable of maintaining its resistive An endurance test has also been performed by subjecting
switching up to 100 continuous sweeps with an HRS-to-LRS the devices to a set and reset voltage pulse of a pulse width of
ratio of B102. However, above this limit the device is totally 10 ms with a pulse height of 5 V. Device-2 exhibits the ability to
degraded. It can be noted that the RESET-step has a wider maintain two stable resistance states throughout cycle stress-
variation in different cycles with respect to the SET-step. ing, lasting up to 3  104 cycles, while Device-3 achieves
Current vs. number of iterations for both the LRS and HRS stability for an impressive 9  104 cycles. Beyond this point,
are shown in Fig. 6(b). there is no sudden switch from the low resistance state (LRS) to
A similar study has been performed with Device-3 which is the high resistance state (HRS). To provide a comprehensive
shown in Fig. 7. The data indicate that Device-3 can maintain view of the endurance curve, data have been chosen selectively
an HRS-to-LRS ratio of B103 with minimal cycle-to-cycle varia- for the LRS and HRS at regular intervals, as illustrated in
tion for continuous sweep up to the 100 cycles (Fig. 7a). Fig. 8(a and b). The complete dataset for endurance cycles is
Above this range, Device-3 can still show its resistive switching available in Fig. S7 of the ESI.†
behaviour but the HRS-to-LRS ratio reduces by one order The statistical distribution of set and reset voltages in
of magnitude (102). However, this ratio still maintains up to consecutive cyclic I–V characterization of a memristor device
200 consecutive cycles (Fig. 7b). This reduction may be due to can provide information about its switching voltage. Besides, it
the degradation of the device, caused by the continuous bias also provides insight into the switching mechanism of the
stress that crosses the tolerance limit of the device. Current vs. device. A narrow distribution of the plot is expected for a high
number of iterations for both the LRS and HRS are shown in performance memristor device. As depicted in Fig. 8(c and d),
Fig. 7(c). The cyclic stability and endurance of the device have the set voltage of Device-3 exhibits variability spanning from
also been documented by replacing the top electrode from Ag to 2.5 V to 4 V. The accompanying Gaussian curve fit highlights
Au, and the data are shown in Fig. S5 (ESI†). Besides this, a peak centred at 3.3 V, indicating that the mean set voltage for
device-to-device variation for switching voltages and HRS to the p+-Si/Ag(Ag+)–Al2O3/Ag configuration is B3.3 V. On the
LRS ratios is illustrated in Fig. S6 (ESI†). other hand, reset voltages exhibit a broad dispersion ranging

Fig. 7 (a) and (b) Current vs. voltage curve of p+-Si/Ag(Ag+)–Al2O3/Ag on a semi-logarithmic scale and (c) endurance property of p+-Si/Ag(Ag+)–Al2O3/
Ag for 200 cycles of continuous measurement.

16150 | J. Mater. Chem. C, 2024, 12, 16145–16155 This journal is © The Royal Society of Chemistry 2024
 View Article Online

Journal of Materials Chemistry C Paper

Published on 28 August 2024. Downloaded by CHUNGBUK NATIONAL UNIVERSITY on 7/31/2025 9:18:29 AM.

Fig. 8 Pulse switching endurance for two resistance states of (a) p+-Si/(Ag/Ag+)–Al2O3/Ag and (b) p+-Si/Ag(Ag+)–Al2O3/Ag respectively and (c) and (d)
the statistical histograms of set and reset voltages over 200 consecutive cycles of p+-Si/Ag(Ag+)–Al2O3/Ag. Lines are obtained by fitting to the Gaussian
distribution.

from 2.9 V to 4.2 V. The distribution curve demonstrates a the ON and OFF state currents. In contrast, Device-2 and
predominant peak at 3.55 V, implying an average reset voltage Device-3 both displayed notably favourable behavior, with the
of approximately 3.55 V. A similar study has been carried out ON state to OFF state current ratio of 104 which has remained
for the device p+-Si/(Ag/Ag+)–Al2O3/Ag represented in Fig. S8 stable over the span of 3 hours (103 s) and 12 hours (more than
(ESI†). 104 s) for these devices respectively. Such a long retention time
by maintaining an on/off ratio of 104 is rarely reported in the
3.5. Retention time literature. However, it is observed that the performance of
The data retention capability of the devices, namely Device-1, Device-2 highly diminishes after 103 s resulting in a negligible
Device-2 and Device-3, has been investigated as shown in Fig. 9. ON/OFF ratio, thus displaying its relatively poor performance as
The initial step involved applying a voltage of 3 V to attain the a data retention device. On the other hand, Device-3 displays no
high-resistance state (HRS) and a bias of 5 V to achieve the low such degradation in device performance and hence shows its
resistance state (LRS). Subsequently, a current vs. time (I–t) capability to be used as a memory storage device. These data
measurement is carried out. In the case of Device-1, it is have been projected over a period of at least 30 days, revealing
observed that no significant distinction is evident between the sustained preservation of the ON/OFF ratio even after this

Fig. 9 Retention time of (a) p+-Si/(Ag/Ag+)–Al2O3/Ag and (b) p+-Si/Ag(Ag+)–Al2O3/Ag, respectively.

This journal is © The Royal Society of Chemistry 2024 J. Mater. Chem. C, 2024, 12, 16145–16155 | 16151
 View Article Online

Paper Journal of Materials Chemistry C

Table 1 Figures of merits for memristors

Device ION/IOFF Endurance Retention time (s)

+
p -Si/Li5AlO4/Ag — — 250
p+-Si/Ag/Ag+–Al2O3/Ag B102 100 1000
p+-Si/Ag(Ag+)–Al2O3/Ag B103 200 4104
Published on 28 August 2024. Downloaded by CHUNGBUK NATIONAL UNIVERSITY on 7/31/2025 9:18:29 AM.

extended timeframe. Table 1 describes an analogy of the

different figures of merits for Device-1, Device-2 and Device-3,
respectively. A comparative summary of device parameters of
some recent publications on oxide memristors along with the
current work (Device-3) is given in Table 2.

3.6. ReRAM property

The present work has confirmed that the device satisfies the
basic requirement of resistive random-access memory (ReRAM)
properties. In order to conduct this investigation, pulse vol-
tages of 2.5 V, 0.5 V, 2.5 V, and 0.5 V are applied
repeatedly50,51 and the device current is recorded simulta-
neously (Fig. 10). After the application of 2.5 V, the device Fig. 10 ReRAM property of p+-Si/Ag(Ag+)–Al2O3/Ag where (a) current of
the device is measured simultaneously when (b) voltage pulses are applied
changes its conducting state to a low conducting state (LCS)
at the top electrode.
and then reads the current at 0.5 V, called read after erase.
Similarly, after the application of 2.5 V, it changes its state to
a high conducting state (HCS) and reads the current at the filament formation starts, the remaining conical shape guides
same read voltage of 0.5 V. It is evident that for the same read the filament formation to achieve the SET state. Therefore, the
voltage, two distinct conducting states are present. More than SET voltage remains almost the same as that of the earlier cycle.
200 cycles of these iterations have been performed to confirm Moreover, such guided filament formation by the remains of
this property of the device. the conical shape Ag enables us to reach high endurance by
maintaining almost the same LRS and HRS and long memory
3.7. Resistive switching mechanism retention time by maintaining a high on/off ratio.
The mechanism behind the creation and breakdown process of This resistive switching behaviour can also be explained
the complementary filament in the case of p+-Si/Ag(Ag+)–Al2O3/ with the help of the energy band model. In the initial state
Ag can be elucidated using a simplified device model. The isolated Ag nanoparticles (Ag NP) exist in the Al2O3 matrix of
device model representing the SET and RESET processes of the the top surface of the Ag(Ag+)–Al2O3 thin film which is pre-
memristor is depicted in Fig. 11. Initially, the Ag nanoparticles sented in Fig. 12(a). Upon application of a negative bias to the
remain in a disordered arrangement within the dielectric layer. bottom electrode, Ag NP faces a strong field and positive ions of
Upon applying a positive bias to the top electrode, the Ag+ ions Ag (Ag+) from the deeper part of the film begin to accumulate at
of the Ag nanoparticles begin to aggregate, forming a conical the interface, resulting in an extension of Ag filament as shown
structure. This conical structure evolves into a filament, initiat- in Fig. 12(b). This aggregation culminates in the establishment
ing conduction. This state is referred to as ‘SET’. Conversely, of a conductive filament, ultimately aligning the Fermi levels of
upon reversing the polarity, the connected Ag filament (Ag0) both silver (Ag) of the electrode and Ag(Ag+)–Al2O3, the bulk
becomes Ag+ and the filament undergoes rupture, reverting to part of the thin film as shown in Fig. 12(c). When a positive bias
individual Ag nanoparticles once again. However, the entire is applied to the bottom electrode, the electrode acquires
conical structure is not destroyed, rather the tip filament part is a positive charge, prompting the reformation of Ag+ from the
ruptured. This state is termed ‘RESET’. In the next cycle, when Ag filament (Ag0), thereby facilitating the rupture process, as

Table 2 Comparison of the different oxide-based memristors with important parameters

Device structure Fabrication process ION/IOFF Retention time (s) Endurance Ref.
Pt/HfO2/Pt Focused ion beam technique B10 4
103
100 30
Ag/NiOx/TiO2/FTO Spin-coated and sputtering-deposited B103 104 100 44
GaIn/NiO/CeO2/ITO Sol–gel spin coating B103 — 100 45
Ag/ZrO2/Pt Magnetic sputtering and spin coating — 3000 100 46
p+-Si/b-Ga2O3/Au VLS 10 — — 47
Ti/Pt/MoO3/Ti/Au e-Beam evaporation 102 B105 100 48
Pt/a-GMO/Pt PAD B103 B103 150 49
p+-Si/Ag(Ag+)–Al2O3/Ag Solution-processed B103 4104 200 This work

16152 | J. Mater. Chem. C, 2024, 12, 16145–16155 This journal is © The Royal Society of Chemistry 2024
 View Article Online

Journal of Materials Chemistry C Paper

Published on 28 August 2024. Downloaded by CHUNGBUK NATIONAL UNIVERSITY on 7/31/2025 9:18:29 AM.

Fig. 11 Schematic model showing the mechanism of filament formation and rupture at the (a) initial state; (b) SET process and (c) RESET process,
respectively.

Fig. 12 Band diagram depicting (a) the initial state, (b) Ag accumulation, (c) metallic contact, and (d) Ag depletion of p+-Si/Ag(Ag+)–Al2O3/Ag,
respectively.

demonstrated in Fig. 12(d). This orchestrated sequence deline-

ates the bidirectional phenomenon of filament formation and
rupture that the memristor undergoes. To realize the role of OH
(moisture) in this mechanism, we blow hot air on top of the
memristor device with a drier for different time periods ranging
from 0–10 minutes which can remove moisture mostly from the
sample. Immediately after the hot air blow, we evaluated the
I–V characteristics of the device and didn’t find any significant
change in the characteristics. The I–V characteristics are shown
in the supporting document (Fig. S9, ESI†).
The I–V characteristics of Device-3 are graphed on a double
logarithmic scale to clarify the conduction mechanism in both
Fig. 13 I–V characteristics of the p+-Si/Ag(Ag+)–Al2O3/Ag device in
the LRS and HRS (Fig. 13). In the LRS or ON state, an ohmic double logarithmic scale illuminating the conduction mechanism in both
behavior (Region I) is evident, displaying a linear relationship the HRS and LRS.
with the applied bias and a slope of B1.10. This behavior is
attributed to the formation of a conductive pathway between
the top and bottom electrodes through the Fowler–Nordheim In Region I, ohmic conduction occurs at a low applied voltage,
(FN) tunnelling process, resembling an ohmic conductor. where the low bias is insufficient to release trapped electrons from
Conversely, the HRS or OFF state exhibits three distinct regions. charge trapping centres, resulting in a linear current–voltage

This journal is © The Royal Society of Chemistry 2024 J. Mater. Chem. C, 2024, 12, 16145–16155 | 16153
 View Article Online

Paper Journal of Materials Chemistry C

relationship with a slope of B1.18. Region II, a smaller area, Acknowledgements

adheres accurately to the Mott Gurney law for semiconductors,
indicating that free charged particles in a semiconducting This work was supported by the ‘‘Science and Engineering
medium drift with a specific velocity (V = me) and the current Research Board,’’ India (CRG/2019/001826) and DST, India
density is proportional to the square of the applied bias (DST/INT/SWD/VR/P-12/2019). The authors thank IIT (BHU)
and its Central Instrument Facility. Subarna Pramanik is
Published on 28 August 2024. Downloaded by CHUNGBUK NATIONAL UNIVERSITY on 7/31/2025 9:18:29 AM.

(I B V2). The calculated slope in Region II is in excellent

agreement with a value of B2.70. Finally, Region III demon- thankful to DST-India for providing the PhD INSPIRE fellow-
strates a steep increase in current compared to a slight change ship. Rajarshi Chakraborty thanks IIT (BHU) for providing an
in applied bias. The current in this region follows the relation SRF fellowship.
I = Va, where a is approximately 33.12. Such notably high values
of a indicate a robust trap-controlled space-charge-limited
conduction (SCLC) mechanism. References
1 L. Zhou, J. Mao, Y. Ren, S. T. Han, V. A. Roy and Y. Zhou,
Small, 2018, 14, 1703126.
4. Conclusion 2 Z. Pan, P. Wang, X. Hou, L. Yao, G. Zhang, J. Wang, J. Liu,
In summary, an innovative methodology has been developed to M. Shen, Y. Zhang and S. Jiang, Adv. Energy Mater., 2020,
fabricate a solution-processed Ag(Ag+)–Al2O3 thin film that 10, 2001536.
contains Al2O3 embedded Ag nanoparticles. This Ag(Ag+)– 3 J. Q. Yang, Y. Zhou and S. T. Han, Adv. Electron. Mater., 2021,
Al2O3 thin film has been utilized to fabricate a memristor 7, 2001181.
device by utilising its pre-existing Ag NPs that reversibly form 4 Z. Zhang, Z. Wang, T. Shi, C. Bi, F. Rao, Y. Cai, Q. Liu, H. Wu
an Ag filament between two electrodes. This device has been and P. Zhou, InfoMat, 2020, 2, 261–290.
fabricated on a highly p-doped Si (p+-Si) substrate with a device 5 S. S. Kim, S. K. Yong, W. Kim, S. Kang, H. W. Park,
configuration of p+-Si/Ag(Ag+)–Al2O3/Ag that exhibits favourable K. J. Yoon, D. S. Sheen, S. Lee and C. S. Hwang, Adv. Mater.,
switching behavior, along with impressive stability that persists 2023, 35, 2200659.
for at least 200 cycles. In addition, the device demonstrates a 6 L. Liu, C. Liu, L. Jiang, J. Li, Y. Ding, S. Wang, Y.-G. Jiang,
long retention time, lasting up to 12 hours, while maintaining Y.-B. Sun, J. Wang and S. Chen, Nat. Nanotechnol., 2021, 16,
an ON/OFF ratio of up to 104. Moreover, the device also shows 874–881.
the ReRAM property up to 200 iterations, which is an additional 7 T. Ali, P. Polakowski, S. Riedel, T. Büttner, T. Kämpfe,
feature of the memristor. Overall, these findings suggest M. Rudolph, B. Pätzold, K. Seidel, D. Löhr and R.
that this technique has the potential to not only advance Hoffmann, Appl. Phys. Lett., 2018, 112, 222903.
our understanding of resistive switching mechanisms but also 8 A. I. Khan, A. Keshavarzi and S. Datta, Nat. Electron., 2020, 3,
pave the way for the development of next-generation, high- 588–597.
performance, multistate memory devices. 9 D. Kwon, E. C. Park, W. Shin, R.-H. Koo, J. Hwang, J.-H. Bae,
D. Kwon and J.-H. Lee, Adv. Intell. Syst., 2023, 2300125.
10 R. Chakraborty, N. Pal, U. Pandey, S. Pramanik, S. Paliwal,
S. Suman, A. Gupta, A. K. Singh, P. Swaminathan and
Author contributions P. K. Roy, Appl. Mater. Today, 2023, 33, 101862.
11 R. Waser and M. Aono, Nat. Mater., 2007, 6, 833–840.
Subarna Pramanik: most of the experimental work, draft
12 D.-H. Kwon, K. M. Kim, J. H. Jang, J. M. Jeon, M. H. Lee,
writing, and conceptualization; Rajarshi Chakraborty: partial
G. H. Kim, X.-S. Li, G.-S. Park, B. Lee and S. Han, Nat.
experimental work; Sobhan Hazra: partial experimental work;
Nanotechnol., 2010, 5, 148–153.
Utkarsh Pandey: partial experimental work; Bhola Nath Pal:
13 Y. Shan, Z. Lyu, X. Guan, A. Younis, G. Yuan, J. Wang, S. Li
conceptualization, draft writing/editing, supervising, and gene-
and T. Wu, Phys. Chem. Chem. Phys., 2018, 20, 23837–23846.
rating fund for the project.
14 S. H. Jo, T. Chang, I. Ebong, B. B. Bhadviya, P. Mazumder
and W. Lu, Nano Lett., 2010, 10, 1297–1301.
15 P. Yao, H. Wu, B. Gao, J. Tang, Q. Zhang, W. Zhang,
Data availability J. J. Yang and H. Qian, Nature, 2020, 577, 641–646.
16 S. Awasthi, S. Pramanik, K. K. Singh, A. Mohan and B. N.
https://drive.google.com/drive/folders/1q1n_p3pUFN2oI2VQGt Pal, ACS Appl. Nano Mater., 2024, 7, 6392–6400.
ZTG8bRPj5JBzDT?usp=drive_link. 17 Y. Li, N. An, Z. Lu, Y. Wang, B. Chang, T. Tan, X. Guo, X. Xu,
J. He and H. Xia, Nat. Commun., 2022, 13, 3138.
18 F. Zhou, Y. Liu, X. Shen, M. Wang, F. Yuan and Y. Chai, Adv.
Conflicts of interest Funct. Mater., 2018, 28, 1800080.
19 X. Zhao, Z. Wang, W. Li, S. Sun, H. Xu, P. Zhou, J. Xu, Y. Lin
The authors assert that they have no conflicting interests. and Y. Liu, Adv. Funct. Mater., 2020, 30, 1910151.

16154 | J. Mater. Chem. C, 2024, 12, 16145–16155 This journal is © The Royal Society of Chemistry 2024
 View Article Online

Journal of Materials Chemistry C Paper

20 Y. Sun, Y. Ding, D. Xie, J. Xu, M. Sun, P. Yang and Y. Zhang, 36 U. Qamar, S. Hazra, C. Kant, U. U. Ghosh, B. N. Pal and
Adv. Opt. Mater., 2021, 9, 2002232. S. Das, Microchem. J., 2024, 196, 109686.
21 S. Kumar, X. Wang, J. P. Strachan, Y. Yang and W. D. Lu, 37 N. K. Chourasia, A. Sharma, V. Acharya, N. Pal, S. Biring and
Nat. Rev. Mater., 2022, 7, 575–591. B. N. Pal, J. Alloys Compd., 2019, 777, 1124–1132.
22 W. Huh, D. Lee and C. H. Lee, Adv. Mater., 2020, 32, 2002092. 38 T. Chaieb, L. Delannoy, C. Louis and C. Thomas, Catal. Lett.,
Published on 28 August 2024. Downloaded by CHUNGBUK NATIONAL UNIVERSITY on 7/31/2025 9:18:29 AM.

23 D. Marković, A. Mizrahi, D. Querlioz and J. Grollier, Nat. 2016, 146, 2622–2629.

Rev. Phys., 2020, 2, 499–510. 39 W. Ji, H. Zhou, Y. Ye, J. Zhao, Y. Xiao, S. Lei and B. Cheng,
24 L. Qingjiang, A. Khiat, I. Salaoru, C. Papavassiliou, X. Hui Adv. Opt. Mater., 2019, 7, 1901154.
and T. Prodromakis, Sci. Rep., 2014, 4, 4522. 40 H. Karimi-Maleh, B. G. Kumar, S. Rajendran, J. Qin,
25 Y. J. Lee, Y. Kim, H. Gim, K. Hong and H. W. Jang, Adv. S. Vadivel, D. Durgalakshmi, F. Gracia, M. Soto-
Mater., 2023, 2305353. Moscoso, Y. Orooji and F. Karimi, J. Mol. Liq., 2020, 314,
26 S. Samanta, K. Han, S. Das and X. Gong, IEEE Electron Device 113588.
Lett., 2020, 41, 924–927. 41 J. Raja, C. P. T. Nguyen, C. Lee, N. Balaji, S. Chatterjee,
27 E. Carlos, J. Deuermeier, R. Branquinho, C. Gaspar, K. Jang, H. Kim and J. Yi, IEEE Electron Device Lett., 2016, 37,
R. Martins, A. Kiazadeh and E. Fortunato, J. Mater. Chem. 1272–1275.
C, 2021, 9, 3911–3918. 42 S. Jain, J. Shah, N. S. Negi, C. Sharma and R. K. Kotnala, Int.
28 W. Wang, F. Yin, H. Niu, Y. Li, E. S. Kim and N. Y. Kim, Nano J. Energy Res., 2019, 43, 4743–4755.
Energy, 2023, 106, 108072. 43 C.-H. Huang, T.-S. Chou, J.-S. Huang, S.-M. Lin and
29 M. Kim, M. A. Rehman, D. Lee, Y. Wang, D.-H. Lim, Y.-L. Chueh, Sci. Rep., 2017, 7, 2066.
M. F. Khan, H. Choi, Q. Y. Shao, J. Suh and H.-S. Lee, ACS 44 G. Zhou, L. Xiao, S. Zhang, B. Wu, X. Liu and A. Zhou,
Appl. Mater. Interfaces, 2022, 14, 44561–44571. J. Alloys Compd., 2017, 722, 753–759.
30 Y. Zhang, G.-Q. Mao, X. Zhao, Y. Li, M. Zhang, Z. Wu, 45 J.-C. Li, H.-P. Cui and X.-Y. Hou, J. Alloys Compd., 2018, 752,
W. Wu, H. Sun, Y. Guo and L. Wang, Nat. Commun., 2021, 247–252.
12, 7232. 46 X. Yan, C. Qin, C. Lu, J. Zhao, R. Zhao, D. Ren, Z. Zhou,
31 S. Li, M. E. Pam, Y. Li, L. Chen, Y. C. Chien, X. Fong, D. Chi H. Wang, J. Wang and L. Zhang, ACS Appl. Mater. Interfaces,
and K. W. Ang, Adv. Mater., 2022, 34, 2103376. 2019, 11, 48029–48038.
32 Y. Liu, C. Ye, K. C. Chang, L. Li, B. Jiang, C. Xia, L. Liu, 47 T.-F. Weng, M.-S. Ho, C. Sivakumar, B. Balraj and P.-F.
X. Zhang, X. Liu and T. Xia, Small, 2020, 16, 2004619. Chung, Appl. Surf. Sci., 2020, 533, 147476.
33 H. X. Li, Q. X. Li, F. Z. Li, J. P. Liu, G. D. Gong, Y. Q. Zhang, 48 R. A. Martins, E. Carlos, A. Kiazadeh, R. Martins and
Y. B. Leng, T. Sun, Y. Zhou and S. T. Han, Adv. Mater., 2023, J. Deuermeier, ACS Appl. Eng. Mater., 2024, 2, 298–304.
2308153. 49 X. Dai, X. Zhang, D. Gong and G. Xiang, Adv. Funct. Mater.,
34 A. Sharma, N. K. Chourasia, A. Sugathan, Y. Kumar, S. Jit, 2023, 33, 2304749.
S.-W. Liu, A. Pandey, S. Biring and B. N. Pal, J. Mater. Chem. C, 50 Z. Xu, Y. Li, Y. Xia, C. Shi, S. Chen, C. Ma, C. Zhang and
2018, 6, 790–798. Y. Li, Adv. Funct. Mater., 2023, 2312658.
35 S. Hazra, S. V. Singh, S. Dahiya, P. K. Aich and B. N. Pal, ACS 51 A. Betal, J. Bera and S. Sahu, J. Mater. Chem. C, 2023, 11,
Appl. Nano Mater., 2023, 6, 15119–15127. 4674–4682.

This journal is © The Royal Society of Chemistry 2024 J. Mater. Chem. C, 2024, 12, 16145–16155 | 16155