Ultralow Voltage Resistive Switching in Hafnium–Zirconium Oxide

Ultralow SET and RESET voltage are essential for high‐density, low‐power, and small heat dissipation nonvolatile random‐access memory (NVRAM) elements. A nanoscale polycrystalline Hf0.75Zr0.25O2 (HZO) thin films on Pt/Si substrate are fabricated and investigated for suitability for bipolar resistive switching. The device illustrates monoclinic and tetragonal/orthorhombic phases with weak ferroelectricity and robust resistive switching. Small remanent polarization (≈0.1 μC cm−2) may assist in the height reduction of barrier height and ease the electron for transport. Remarkably, the Al/HZO/Pt/Si device, consisting of thin films with 10 and 5 nm thicknesses, exhibits a switching voltage below −30 mV from a low‐resistance state (LRS) to a high‐resistance state (HRS). It shows a significant ROFF/RON ratio of 106, making it suitable for low power consumption and minimal heat dissipation devices. Moreover, the utilization of an ultrathin film (5 nm) results in an improved reduction (< 0.7 V) of the operating window at the positive voltage. Direct tunneling and the Fowler–Nordheim tunneling model are performed in current–voltage (I–V) data to study the charge transportation behavior over a trapezoidal and triangular potential barrier. These results of the HZO candidate may stimulate the futuristic nonvolatile resistive random‐access memory (ReRAM) in the optoelectronic industry.


Introduction
Information storage is a big challenge today, a more trending area in integrated circuits. [1]For several decades, researchers have been trying to get advancement in silicon-based transistor scaling. [2]However, these nanoscale devices are unreliable when a small applied voltage creates a large electric field on the device that may cause current leakage and data loss.Among various DOI: 10.1002/apxr.202300123commercially available emerging storage technologies, resistive random-access memory (ReRAM) has received significant attention due to its low power consumption, fast switching speed, large endurance, and high retention time. [3,4]][13][14][15][16][17][18] Several parameters are creating issues regarding the application of information, such as low storage, low retention, high power consumption, and entire transmission mechanism, that need to be addressed. [1]oreover, the resistive switching phenomenon was exhibited in various smart materials by introducing different mechanisms, such as the migration of oxygen vacancies and oxygen ions with the help of the electrochemical reaction, [8,[19][20][21] electrochemical metallization of the electrode, [22][23][24] and ferroelectric resistive switching by switching of polarization.These atomic phenomena assisted in the formation and annihilation of nano-conducting filament and barrier height/width modulation with the support of the charge screening effect. [18,25]However, these mechanisms play an essential role in the present study.Corresponding to these mechanisms' severe disadvantages like less reliability and highpower consumption are present, [20] which may concern robust devices.To do that, researchers tried to achieve the resistive switching phenomenon in different transition metal oxides such as in NiO x , [26] Al 2 O 3 , [27] ZrO 2 , [28] etc., that involved with high operating SET, RESET, and electroforming voltage.Another trending material, HfO 2 , was introduced and made a promising contribution to the futuristic nonvolatile memory application for computer data storage and neuromorphic computing.It is compatible with semiconductor technology along with physical properties and scaling capability.Several industries are devoting their efforts to making it commercial. [29][32] Therefore, doping zirconium into the HfO 2 may be more suitable for achieving the low power consumption ReRAM device.Zirconium and hafnium have almost analogues in chemistry regarding the same crystal phase structure, crystal chemistry, and electron affinity. [33]Especially Hf x Zr 1-x O 2 based structure was investigated in past studies due to its robust ferroelectricity nature [34] with different stoichiometric ratios.As per the previous study, lots of work has been done on Hf 0.5 Zr 0.5 O 2 .[37] The issues elucidated above emphasize the considerable attention directed toward ferroelectric HZO materials.Particularly, the focus lies on leveraging the potential of these materials to achieve reduced power consumption.This is accomplished by modulating the barrier height at the interface by applying an external electric field.Such modulation induces nonvolatile changes in resistance.However, researchers are trying different strategies to reduce the operating voltage and enhance the resistance ON/OFF ratio between HRS and LRS, which may assist in lowering barrier height and thickness with the help of direct and Fowler-Nordheim tunneling. [38]n the proposed study, PLD-grown HZO thin films of different thicknesses were deposited on a platinum (Pt) substrate to obtain polarization-induced bipolar resistive switching.To enhance the ferroelectric effect in HZO, the growth of monoclinic and tetragonal/orthorhombic phase formation was obtained.We performed polarization measurements to see the effect of ferroelectric polarization on the barrier height at the metal/insulator interface.The Al/HZO/Pt/Si device revealed interesting bipolar resistive switching with ultralow RESET voltage.Direct and Fowler-Nordheim tunneling mechanisms dominate the linear fitting of measured I-V data.The charge transfer mechanism through barrier has been studied, which convinces the barrier modulation.The polarization attracts the charge carrier at the interface.Its effect can be seen in the modulation of barrier height by which injected charge carriers move through oxygen vacancies by the hopping process.This hopping process will start once the charge carrier reaches the conduction band of HZO in F-N tunneling.

Results and Discussion
The PLD-deposited ultrathin film HZO was characterized using the grazing incident X-ray diffraction (GIXRD) to determine the phase structure.These 5 and 10 nm thin films showed similar XRD patterns; only 10 nm thin film XRD is given for clarity and better presentation.[41][42][43] The monoclinic phase dominates HZO compared to the tetragonal and orthorhombic phases [44] .
In addition, the surface morphology of HZO thin film was studied by atomic force microscopy (AFM) over the scan area of 1 × 1 m 2 , as shown in Figure 1b,c.Thin films of 10 and 5 nm have heights of 36 and 11 nm, respectively.The calculated rootmean-square (r.m.s.) roughness for 10 and 5 nm thin film is 3 and 1.47 nm, respectively.Figure 1b,c shows the grain boundaries, which may assist in making an electroforming-free device and suggest that these grain boundaries provide a predefined dielectric breakdown for electroforming-free ReRAM devices. [45]At the atomic resolution, HfO 2 provided the defects within grain boundaries.The electrical measurements of Al/HZO/Pt/Si were performed at room temperature to obtain the switching phenomenon by sweeping the voltage sweep cycles.Figure 2a is a schematic representation of the Al/HZO/Pt/Si thin film device, which represents how we made the electrical connection with microprobes.In a pristine device, when positive voltage was applied to Al and negative to the Pt electrode and run the sweep from 0 to 6 V, the device initially starts from the LRS state, as illustrated in Figure 2b.The inset of Figure 2b is the actual microscopic image of a fabricated Al/HZO/Pt/Si device.
Subsequently, the voltage cycle was swept from 0 to −30 mV.The device remained in the previous LRS state up to a certain millivolt range, as shown in Region 1 of Figure 2c. Figure 2d is the magnified area of Figure 2c, representing the ultralow RESET voltage (V RESET ), which is not clearly visible in Figure 2c due to the very small negative voltage scale on the x-axis in respect of the large positive voltage scale.The arrows in Figure 2c,d show the sweeping voltage direction.In the particular case of V RESET, the device switches its resistance from LRS to HRS, as shown in Region 2, and later on, it remains in the HRS state.When the voltage was swept from −30 mV to 0 V, the device did not switch back to the LRS state, as shown in Region 3.This V RESET window is a bit wide in the range of −4 mVto −30 mV.It may be affected when multiple polarization domains do not completely return to their original orientation.
Furthermore, we started the voltage sweep cycle from 0 to 6 V.The device stayed in the HRS state at low voltage regimes, as shown in Region 4. At high voltage, the device was again switched to the LRS state at a certain voltage called SET voltage (V SET ), and again, the current was increased abruptly (region 6), hitting the compliance current 100 A where the resistance of the device completely switched from HRS to LRS.The LRS state has remained in the same state as shown in region 6 over the voltage sweep cycle from 6 to 0 V, as illustrated in Figure 2c.The gap between the two resistance states represents the nonvolatile memory window in terms of resistance.The process was repeated 20 times and always showed a similar bipolar resistive switching behavior.Sometimes, the current begins to increase slowly when the sweep cycle starts in the positive voltage regime, which may be due to the contribution of charge carriers from the thermionic injection.Therefore, a thermionic and non-thermionic injection procedure may be one of the reasons behind the repeatability issues.
Additionally, we also investigate thickness-dependent resistive switching.At the lower thickness (5 nm) of the HZO device, V SET reduced to the window of 0.4 to 0.7 V, and V RESET is almost the same as for 10 nm, as shown in Figure 2e,f.We also observed that sometimes V RESET goes up to a certain volt, which is not mentioned in the graph.The short conduction channel or weak polarization may cause this higher V RESET .If the polarization is a bit high, it can assist more in reducing barrier height and will ease the carrier tunneling through the insulating (HZO) layer.Although to improve the wider voltage and higher power consumption in the SET process of 10 nm thin film, we thoroughly investigate the resistive switching in 5 nm thin film.As shown in Figure 2e, the window of V SET was reduced by reducing film thickness, and now it is 0.4 to 0.7 V for switching devices.But V RESET is not much affected in this case.Another reason for the reduction in the V SET window may be grain boundaries.The grain boundaries, which can be connected with the electrodes, are the novel way to bring it a low operating voltage window in 5 nm because the density of grain boundaries in 5 nm thin film is higher as compared to 10 nm, as illustrated in Figure 1c.
Additionally, in bipolar resistive switching of HZO, the weak ferroelectricity and oxygen vacancies migration are the joint participation.For understanding, a ferroelectric polarization effect was investigated in HZO thin film, which may assist in barrier modulation for the charge transfer process between electrodes through an insulating layer.Figure 3a is the measured symmetry hysteresis loop of the polarization-electric field (P-E) at 1 kHz.The maximum polarization (P * ) and remanent polarization (P r ) is 0.33 and 0.1 C cm -2 , respectively.Furthermore, the coercive field (E c ) was realized at 80.8 kV cm −1 during the positive bias and 126 kV cm −1 during the negative bias.The difference between the positive and negative E c was obtained due to the two asymmetric interfaces, Al/HZO and HZO/Pt, widely observed in other metal-ferroelectric-metal structures. [1]o obtain a more accurate polarization charge, positive-up and negative-down (PUND) measurement was performed, as shown in Figure 3b.Five square pulse waveforms were applied for PUND measurement with pulse width and pulse delay of 1 ms, respectively.The first pre-set pulse is used to switch the polarization pointing toward the top Al electrode, which will drive the device into the HRS state.A second positive pulse is applied for negative initialization, which integrates the switchable and nonswitchable charges.A following up-pulse was applied to measure the nonferroelectric polarization, which integrates only nonswitchable charge (including contribution from relaxed polarization and leakage current).Similarly, the same principle was applied for negative and down-pulse. [46,47]The maximum polarization was slightly higher than that obtained from the P-E hysteresis loop.The difference between P * and P ˆis not so much that it may be affected by the metal-HZO interface and creates a large switching window.The endurance of the ferroelectric ReRAM device was measured up to 10 5 pulse cycles, as illustrated in Figure 3c.In the inset of Figure 3c, the magnified scale of remanent polarization (P r and −P r ) shows a few fluctuations that may be caused by a large number of defects (oxygen vacancies) migrating to the ferroelectric/electrode interface or domain wall and thus pinning the switching. [46]With this, the retention time of HRS and LRS was measured up to 10 4 s at a read voltage of 10 mV, showing two well-stable states with resistance ratio (R OFF /R ON ) ≈ 10 6 , as illustrated in Figure 3d.

Interface Limited Conduction Mechanism
In order to explore the charge carrier mechanism involved in ferroelectric resistive switching, diverse conduction mechanisms were examined by fitting I-V data.This was done to elucidate the carrier transport mechanism underlying resistive switching induced by polarization in HZO devices.Space charge limited conduction (SCLC), Pool Frenkel conduction mechanism (P-F), Schottky emission (S-E) conduction mechanism, direct tunneling (D-tunneling), and Fowler-Nordheim (F-N) tunneling mechanism were studied.D-tunneling and F-N tunneling are interface-limited conduction mechanisms when the inserting layer thickness goes beyond 10 nm.It depends on the electrodedielectric interface's electrical properties. [48]In D-tunneling, electrons will directly tunnel through the Pt/HZO interface at low electric fields, Figure 4a.High electric fields may aid in modifying the potential barrier structure from a rectangular to a triangular one, which is the significance of F-N tunneling, as shown in Figure 4b, and electrons can tunnel from the electrode to the conduction band of HZO. [49,50]The I-V data was analyzed using SCLC, S-E, and P-F conduction models for 10 and 5 nm thin films.Figures S1 and S2 (Supporting Information) display the fitting results.However, it was observed that the SCLC, S-E, and P-F conduction mechanisms did not align well with the fitting criteria.This discrepancy was attributed to the low slope values, indicating an unrealistic dielectric constant.
Furthermore, another conduction mechanism, such as Dtunneling and F-N tunneling was investigated using I-V data.We obtained there is the transition of D-tunneling to F-N tunneling, called point of inflection, which is the sign of a reduction in barrier height (the energy offset between the fermi level of the electrode and minimum conduction band of an insulating layer) at the interface of HZO/Pt layer.3][54][55][56] Where J is measured current density, q is the electronic charge, ∅ B (∅ B = q∅) is the barrier height, E is the applied electric field, h is Planck's constant (ℏ = h/2), and m* is the effective tunneling mass of HZO.J F − N is inversely proportional to the film thickness in the tunneling process, which can define tunneling. [57]he ln(J E −2 ) versus 1/E graph was generated using the I-V data from Figure 2c,e.In Figure 5a,b, linear curve fitting was applied, with the data fitted using Equation 1 during positive and negative voltage sweeps, respectively.At low electric field, the conduction mechanism follows D-tunneling, where electrons drift across the trapezoidal barrier.As the electric field increases, F-N tunneling comes into play, characterized by a point of inflection where a stream of electrons drifts over the triangular barrier from the HZO/Pt interface.The inflection point [50,55,58] divides the ln(J E −2 ) versus 1/E plot into two regions (F-N and D-tunnelling), as illustrated in Figure 5a.The fitting plot of Figure 5a agrees with I-V data for the electron tunneling process.Moreover, a linear relation exists between ln(J E −2 ) versus 1/E with a negative slope, with an R-square value of> 0.9.In the inset of Figure 5a is the magnified area of a black circle to get clear visibility for the point of inflection.Figure 5c,d are the linear curve fitting for a 5 nm HZO thin film device.The concentration of injected free electrons from the Pt electrode through HZO is higher, abruptly leading to switching from the HRS to the LRS.Similarly, when voltage was swept from 6 to 0 V, the device sustained the LRS state where F-N tunneling and D-Tunneling followed at high and low electric fields, respectively.For switching the device from LRS to HRS, the voltage was swept from 0 to −30 mV; again, the same tunneling behavior happened.Electrons will tunnel from the opposite interface (Al/HZO), as shown in Figure 5b, and lead the resistance switch from LRS to HRS.Additionally, during the switching of the device from HRS to LRS and LRS to HRS in F-N tunneling region, ∅ B was calculated from Figure 5a,b for 10 nm.Further ∅ B for 5 nm was calculated from Figure 5c,d by the slope fitting equation of F-N tunneling is expressed as. [54] Slope = −6.83× 10 7 √ ( where m o is the mass of the electron.When positive voltage was applied to the top Al electrode and negative voltage to the bottom of the electrode, polarization direction pointed toward the Pt electrode, which can assist the band bending at the HZO/Pt interface.The value of ∅ B is 1.73 and 0.31 eV for 10 and 5 nm, respectively, at the HZO/Pt interface.Similarly, the polarization pointed toward the Al electrode when negative to the Pt electrode and positive to the Al electrode.The calculated value of ∅ B is 5.8 and 5 meV for 10 and 5 nm thin film, respectively, at Al/HZO interface.Consequently, the device was switched from LRS to HRS at ultralow voltage due to low potential barrier height.During the calculation, the value of effective mass (m * ) taken for HZO is 0.4 m o [59] .

Energy Band Diagram and Barrier Modulation
We conducted a comprehensive analysis of the energy band diagram, depicted in Figure 6, and performed the calculations to determine the values of ∅ B .At the Al/HZO interface, we found a value of 1.48 eV, while at the HZO/Pt interface, the value was determined to be 2.85 eV for the tunneling process.However, it is worth noting that these calculated values for tunneling are higher than the experimental values indicated in the mechanism model of Figure 7.One possible explanation for the observed deviation is the impact of polarization, which likely contributes to the practical reduction in ∅ B during the tunneling process.Figure 7a-c) shows the mechanism model of the Al/HZO/Pt/Si structure at a different region of I-V curves under the applied positive and negative voltage.When we applied the positive terminal to the top Al electrode (TE) and negative to the bottom Pt electrode (BE) and started the voltage sweep from 0 to 6 V. Initially, the device remained in an LRS state, as shown in (a) where we can say that the conducting channel between two electrodes was formed prior to applying electric field.We anticipate the homogeneous polarization in the ferroelectric tunnel barrier with the initial polarization remaining in the downward state, which can assist in reducing barrier height at the HZO/Pt interface. [60]After that, we connected the negative terminal to the TE and the positive to the BE.The polarization direction switched to the upward direction, which has been assisting in the reduction of barrier height at the Al/HZO interface, and again, the oxygen atom will receive two electrons at the Al/HZO interface, leading to the recombination of O 2 − and V o .. as shown (b) with the help of reaction equation 3.This recombination process will eradicate the conducting filament with the help of reaction equation 4. (3) After that, the device goes into the HRS state completely, as shown in Figure 7b.Furthermore, we applied the positive terminal to TE and the negative to the BE.The polarization direction was pointed downward, which may assist in reducing barrier height at the HZO/Pt interface. [50]The oxygen atom receives the coming electrons from the interface and will reduce into the oxygen vacancies and oxygen ions in the oxide layer [61] ; the electrochemical reaction occurs (5) where O is the oxygen on the regular site, V o .. denotes the oxygen vacancies and O 2 − represents oxygen ion.This electrochemical reaction separates the oxygen ions from the regular oxygen site and drifts toward TE. [1]At low biased voltage, thermionic injected electrons or non-thermionic injected electrons will tunnel directly through the insulating layer.At high voltage, electrons do not tunnel directly, but they tunnel through the triangular barrier into the conduction band of the HZO layer.These electrons will move toward the Al/HZO interface through the hopping process, as illustrated in Figure 7c.Again, when we sweep the voltage from 0 to 6 V, the same reaction 5 will start, leading the device switch to the complete LRS state, as shown in Figure 7a.
][64][65][66][67][68] Notably, our device demonstrated switching within an exceptionally low V RESET window, ranging from −4 to −30 mV.This level of performance at such a low voltage scale has not been reported before.V SET window (1-6 V) is high in 10 nm thin film, but further, it brings down to 0.4 to 0.7 V with the help of thickness reduction (5 nm).The V SET may be further reduced by changing the bottom electrode because Pt has a high work function.Additionally, the R OFF /R ON ratio was obtained to be very large (10 6 ) with nonvolatile properties at room temperature, including retention time in the proposed novel structure, as shown in Figure 4d.Other hafniumbased reported structures do not show a high R OFF /R ON ratio, as illustrated in Figure 8 with red circles.
We have compared three important properties in Figure 8 with different combinations of hafnium oxide or hafnium zirconium oxide materials that have been reported for resistive switching.The ferroelectric resistive switching properties of Hf 0.75 Zr 0.25 O 2 with other Hf 0.5 Zr 0.5 O 2 and HfZrO 2 materials have been investigated for ferroelectric switching.The P * , P r , coercive voltage (V c ), memory window (in ferroelectric loop and I-V loop), V SET , V RESET , and retention time are all compared.High-work function electrodes were used for high ferroelectric polarization to sustain high electric field strength and high temperature, as shown in (Table S1, Supporting Information).However, they were not able to exhibit good resistive switching.Good ferroelectric resistive switching does not require high polarization.However, in the proposed HZO device, the polarization is incorporated only to assist the barrier modulation at the interface and cause the reduction in operating voltage for resistive switching.Therefore, an unsymmetric structure intentionally reduces the operating voltage with the help of polarization.The polarization in the proposed device is very weak due to the low work function of the top aluminum electrode and bulk amount of nonpolar monoclinic phase, but it can be improved by improving the tetragonal phase formation by either reducing the thickness or altering the stochiometric ratio.Table S1 (Supporting Information) indicates that different HZO-based devices corresponding to different types of electrodes have high polarization.However, the key is that the proposed HZO device can switch at extremely low V RESET with good retention time and R OFF /R on ratio.Table S1 (Supporting Information) demonstrates that the devices with high polarization and very low leakage current have high operating voltage, which may not be suitable for RRAM devices.Additionally, the V RESET window in HZO has not changed for the 5 nm thin film device, which is in the voltage window of −4-−30 mV.The V SET has reduced from the voltage window amid 1.5V and 6 V to 0.4V and 0.7 V, which is quite helpful for the futuristic low power consumption resistive switching device.It may be possible to further reduce the operating voltage in ultrathin HZO films with thicknesses lower than 5 nm.

Conclusion
We report resistive switching in thin film devices with ultrathin HZO (10 and 5 nm) as dielectric layer.The polycrystalline HZO thin film device was deposited on Pt/Si substrate using PLD, and it mainly constitutes monoclinic and tetragonal/orthorhombic phases and has a weak ferroelectric effect.During the I-V sweep, the device switched at ultralow V RESET, which may be suitable for futuristic nonvolatile memory components at low power.The 5 nm ultralow thin film reduced the SET operating voltage window by a factor of 10 (upper voltage limit 6 to 0.7 V).The P-E and PUND measurement was performed to check the ferroelectric effect inside the HZO thin film to assist in barrier modulation.D-T and F-N tunneling curve fitting was performed in I-V data to study the charge transport behavior during the switching phenomenon.An overall observation of Al/HZO/Pt/Si thin film devices at ultralow scale may contribute to their exploitation for futuristic ultralow power consumption resistive switching devices.

Experimental Section
Device Fabrication: The Pt/Si substrate was cleaned with acetone for 20 min and then cleaned with Isopropyl Alcohol for 10 min to remove the impurity from the surface of the substrate.Initially, the one-inch target of HZO (Ultrananotech #99.99%) was mounted on the target holder of the PLD system.The vacuum chamber was evacuated until the base pressure reached 10 −6 mbar.The oxygen (99.999%) gas pressure was inserted inside the chamber to preserve the pressure of 0.18 mbar.After that, a ferroelectric HZO thin film was grown on the Pt/Si substrate using the krypton fluoride (KrF) excimer laser, which was incident on the rotating target at 45 ○ .The laser energy density was fixed at 2 J cm −2 to ablate the material from the mounted target.The laser shots were implied on the target surface to achieve a 10 and 5 nm thin layer with a repetition rate 5 Hz.The distance between the target and substrate was kept 4.5 cm.The ablated material was contentiously grown on the surface of the substrate at 650 °C.After the growth process, the film was immediately annealed at 650 °C for 30 min to form the crystalline and phase structures under the ambient 400 mbar oxygen partial pressure.Later the HZO sample was cooled at the rate of 10 °C min −1 .
Electrical Measurements: From electrical measurements point of view, the top Aluminum (Al) electrode was deposited by using the thermal evaporation technique using a shadow mask of 100 × 100 m 2 .The electrical characterizations, such as I-V and retention time, were carried out by using the Keithley 2400 and Keithley 2614B source meters.P-E, PUND, and stress cycles were carried out using a ferroelectric tester (Radiant Technologies).
For Crystal Structure and Surface Topography: The crystal structure and phase formation were confirmed using GIXRD.For 10 and 5 nm surface topography, deposited thin film on Pt/Si substrate was studied using AFM.The film thickness was measured using the profilometer (NanoMap 500ES)

Figure 1 .
Figure 1.a) XRD graph of ultrathin HZO film.b,c) AFM images of 10 and 5 nm films.The scan area is 1 × 1 m 2 , and the calculated surface roughness for 10 and 5 nm is ≈ 3 and 1.47 nm, respectively.

Figure 2 .
Figure 2. a) Schematic Al/HZO/Pt/Si device representation.b) First, I-V sweep presents the device in an LRS state.Inset, Schematic of proposed novel structure and its actual microscopic image.c) I-V characteristics of bipolar resistive switching device (10 nm).d) Magnified image of area under blue ellipse of (c) during negative voltage sweep cycle.e) I-V characteristics of bipolar resistive switching device (5 nm).d) Magnified image of e) during negative voltage sweep cycle.

Figure 3 .
Figure 3. a) P-E hysteresis loop, b) PUND measurement by applying five switchable and nonswitchable pulses.c) Stress cycle measure of Al/HZO/Pt/Si device up to 10 5 pulse cycle.Inset, magnified area of remanent polarization (P r and −P r ), which is covered under the area of a rectangle.d) Retention time measurement to check the retentivity of HRS and LRS state with 10 6 R OFF /R ON ratio.

Figure 4 .
Figure 4. a) Schematic of step potential barrier at the low electric field.b) The effect of a strong electric field on the potential barrier might occur in electron tunneling through the barrier.

Figure 5 .
Figure 5. Illustration of D-tunneling and F-N tunneling at low and high electric fields, respectively, for determining potential barriers.a) linear curve fitting of F-N tunneling when voltage swept from 0 to 6 V, and the device switched from HRS to LRS. b) linear curve fitting for the LRS state, which switches from LRS to HRS (0-−30 mV).c) linear curve fitting for 5 nm thin film when voltage was sweep from 0 to 1V which follows the D-tunneling and F-tunneling at low and high applied electric field, respectively.The device was switched from HRS to LRS. d) Linear curve fitting when voltage sweeps from 0--20 mV.

Figure 6 .
Figure 6.A schematic representation of the potential energy band diagram of Al/HZO/Pt/Si thin-film device.

Figure 7 .
Figure 7. Schematic illustration of energy diagram.a) When the device is initially in LRS state with filament formation.For HRS to LRS. b) Illustration of filament annihilation when the device switches from LRS to HRS. c) Filament formation when the device switches from HRS to LRS state.

Figure 8 .
Figure 8. Schematic illustration of comparison of previously reported work with our proposed work.