www.acsami.org Research Article Experimental and Modeling Study of Metal−Insulator Interfaces to Control the Electronic Transport in Single Nanowire Memristive Devices Gianluca Milano,* Enrique Miranda, Matteo Fretto, Ilia Valov, and Carlo Ricciardi* Cite This: ACS Appl. Mater. Interfaces 2022, 14, 53027−53037 Read Online ACCESS Metrics & More Article Recommendations * sı Supporting Information ABSTRACT: Memristive devices relying on redox-based resistive switching mechanisms represent promising candidates for the development of novel computing paradigms beyond von Neumann architecture. Recent advancements in understanding physicochemical phenomena underlying resistive switching have shed new light on the importance of an appropriate selection of material properties required to optimize the performance of devices. However, despite great attention has been devoted to unveiling the role of doping concentration, impurity type, adsorbed moisture, and catalytic activity at the interfaces, specific studies concerning the effect of the counter electrode in regulating the electronic flow in memristive cells are scarce. In this work, the influence of the metal−insulator Schottky interfaces in electrochemical metallization memory (ECM) memristive cell model systems based on single-crystalline ZnO nanowires (NWs) is investigated following a combined experimental and modeling approach. By comparing and simulating the electrical characteristics of single NW devices with different contact configurations and by considering Ag and Pt electrodes as representative of electrochemically active and inert electrodes, respectively, we highlight the importance of an appropriate choice of electrode materials by taking into account the Schottky barrier height and interface chemistry at the metal−insulator interfaces. In particular, we show that a clever choice of metal−insulator interfaces allows to reshape the hysteretic conduction characteristics of the device and to increase the device performance by tuning its resistance window. These results obtained from single NW-based devices provide new insights into the selection criteria for materials and interfaces in connection with the design of advanced ECM cells. KEYWORDS: metal−insulator interfaces, memristive devices, resistive switching, nanowires, schottky barriers ■ INTRODUCTION Memristive technologies are considered promising candidates in ECM cells can be either volatile or nonvolatile.16−18 While nonvolatility represents a prerequisite for the development of for the development of alternative memory devices for data memristive memories, great attention have been recently storage and for the development of computing architectures devoted to volatile devices that endow the capability of beyond von Neumann architecture, paving the way for in- temporal processing of the input signal thus allowing the memory computing and neuromorphic systems.1−11 Redox- implementation of synaptic functionalities and unconventional based memristive devices are essentially two-terminal metal− computing paradigms.19−23 insulator−metal cells in which the internal resistance state can In the last years, important advancements in the under- be modulated by ionic effects under the action of an external standing of the physicochemical processes regulating mem- electrical stimulation.12−14 In particular, resistive switching ristive dynamics and functionalities have been achieved. In this mechanism underlaying memristive functionalities in electro- context, memristive devices based on nanowires (NWs) have chemical metallization cells (ECMs) relies on the formation/ demonstrated their suitability as model systems to investigate dissolution of a metallic filament that bridges the two the intrinsic mechanism of switching in memristive cells by electrodes.15 In this case, a positive voltage applied to the taking advantage of their planar structure and high localization electrochemically active electrode can cause the dissolution of metal atoms to form metal ions that start to migrate toward the counter electrode under the action of the applied electric field. Received: June 21, 2022 This leads to the formation of a metallic filament after Accepted: August 25, 2022 recrystallization, causing the switching of the memristive cell Published: November 17, 2022 from a high-resistance state (HRS) to a low-resistance state (LRS). Depending on the operational conditions that control the lifetime of the metallic filament, the switching mechanism © 2022 The Authors. Published by American Chemical Society https://doi.org/10.1021/acsami.2c11022 53027 ACS Appl. Mater. Interfaces 2022, 14, 53027−53037 ACS Applied Materials & Interfaces www.acsami.org Research Article Figure 1. Single ZnO NW devices. Single NWs contacted by (a) symmetric Pt−Pt electrodes, (b) asymmetric Pt−Ag electrodes, and (c) symmetric Ag−Ag electrodes (scale bars, 250 nm) and corresponding I−V characteristics in the low-voltage range (panels d−f). For each electrode configuration, the electrical response of at least 10 different NW-based devices is reported. of the switching events.17,24−30 For redox-based switching conduction mechanism in the pristine state, we show that devices, notable accomplishments have been reported in interfaces can be appropriately tuned to modify the hysteretic connection with half-cell redox reactions and nanoscale behavior and the resistance window of the device. Within this electrochemistry31,32 together with new insights into the framework, the crucial role of regulating the electronic formation/dissolution of the filament by in situ and ex situ conduction mechanism of the metal−insulator at the counter characterization techniques.33−36 In addition, the impact of electrode is highlighted. In particular, it is shown that a proper introducing dopants and impurities in the oxide matrix,37 the choice of the counter electrode with a blocking character influence of moisture,38−41 and the combination of electrodes allows to reshape the hysteretic conduction characteristic and and electrolyte materials have been extensively investigated.42 to enlarge the resistance window. Importantly, significant interest has been devoted to the understanding of the role played by interfaces that, due to the nanoscale nature of the involved materials, can be more relevant in regulating the switching characteristic of the device ■ RESULTS AND DISCUSSION Single NW memristive cells were realized by contacting single- than the active matrix itself.43,44 Though not always explicitly crystalline and hexagonal-shaped ZnO NWs (diameter of discussed, the use of different electrodes and doping levels ∼100 nm) grown by low-pressure chemical vapor deposition reflects in a change of the properties of interfaces that regulate (LP-CVD) with metal electrodes through a combined optical both the ionic and the electronic flux. In valence change and electron beam lithography (EBL), as detailed in our memories (VCMs), modifying the Schottky electrode−oxide previous works46−48 (details of the ZnO NW synthesis and interface using a variety of metal electrodes/oxides has been device fabrication can be found in the Experimental Section). proposed as a way of altering the resistance of the cell.43 Concerning device scalability, it is worth noticing that the ZnO Instead, while substantial efforts have been devoted to NW diameter can be further reduced by exploiting peculiar understand the influence of the catalytic activity of the counter growth strategies that allows the growth of ZnO NWs with electrode on the nanoionic processes taking place at the ECM diameters below 10 nm, as for example reported by Yin et al.49 cells,45 studies on the influence of the counter electrode in the that exploited a catalyst-assisted growth technique. The role of regulation of the electronic flux are scarce. the metal−insulator interfaces on the electronic transport Here, we investigate through a combined experimental and properties and switching behavior of single NW memristive modeling approach the role of metal−insulator interfaces in cells was evaluated by exploiting Ag and Pt as representative single-crystalline ZnO NW-based devices, used as memristive electrochemically active and inert electrodes, respectively, both model systems. The influence of the metal−NW interface was in symmetric and asymmetric configurations as discussed in analyzed by comparing devices with different metal electrodes the next section. both in symmetric and asymmetric configurations and by Metal−Insulator Interfaces and Electronic Transport exploiting Ag and Pt as representative electrochemically active Mechanisms. Representative scanning electron microscopy and inert metal electrodes, respectively. Besides comparing the (SEM) images of single NW memristive cells with symmetric influence of metal−NW interfaces in regulating the electronic Pt electrodes, asymmetric Pt and Ag electrodes, and symmetric 53028 https://doi.org/10.1021/acsami.2c11022 ACS Appl. Mater. Interfaces 2022, 14, 53027−53037 ACS Applied Materials & Interfaces www.acsami.org Research Article Figure 2. Experimental and modeling electronic transport properties of single ZnO NW devices in the pristine state. I−V characteristics of ZnO NWs contacted with (a) symmetric Pt−Pt electrodes, (b) asymmetric Pt−Ag electrodes (positive voltage is applied to the Ag electrode), and (c) symmetric Ag−Ag electrodes. The equivalent circuital representations of devices represented by back-to-back connected Schottky junctions and the NW series resistance (RNW) are reported as insets. Ag electrodes are illustrated in Figure 1a−c, respectively. In the of the single ZnO NW memristive cell comprising two metal− low-voltage range, symmetric Pt/ZnO NW/Pt devices yield an ZnO interfaces can be electrically represented as two back-to- almost linear I−V characteristic (Figure 1d), asymmetric Ag/ back connected Schottky barriers in series with the NW ZnO NW/Pt devices resulted in a rectifying I−V characteristic resistance, where each Schottky barrier is characterized by its (Figure 1e), while symmetric Ag/ZnO NW/ZnO device particular value of I0 and η. This electrical configuration does resulted in a blocking characteristics (Figure 1f). Notably, not have analytic solution for dissimilar devices, but it can be the electrical characteristics of at least 10 devices for each approximately solved using a piecewise model with nonlinear configuration are reported in panels d−f of Figure 1 revealing series resistance correction, as detailed in a previous work.55 that, despite device-to-device variability in current levels that According to this simplified model and considering the can be ascribed to variations in contact geometries and impedance of the electrochemical reaction smaller than the junction areas,50 comparable behavior is observed in different Schottky barrier before electroforming, the voltage across the devices with the same electrode configuration. In this context, memristive cell can be simply expressed as note also that small variations in nanosized contact geometries and junction areas at the two ends of the NW are responsible ij I yz for nonperfectly symmetric I−V characteristics of symmetric V=± L,R VT lnjjjj± + 1zzzz + IR± devices. The comparison of the pristine state electrical k I0L,R { (2) characteristics of NW-based devices contacted with different electrode configurations clearly demonstrates the influence of where ηL and ηR are the ideality factors and I0L and I0R are the the metal−insulator interface in regulating the electronic saturation currents of the two Schottky barriers, respectively. conduction. R± is an effective series resistance that endows the NW A combined experimental and modeling approach was resistance (RNW) and contact effects. + and − represent the exploited for a more detailed investigation of the role of sign of the applied voltage. From (2), the current flowing metal−insulator interfaces on the electronic transport proper- through a single NW device can be described by the more ties. In purely electronic transfer (i.e., without electrochemical complex expression55 reactions and ionic transport), the metal-insulating interface at ÄÅ É the nanocontact can be modeled as a Schottky barrier, as a V ÅÅ I R± Å ij |V | + I R± yzÑÑÑÑ I ± = ± ± W ÅÅÅÅ expjjjj zzzÑÑÑ I0L,R L,R T 0L,R 0L,R consequence of the semiconducting nature of the intrinsically ÅÅ L,R VT j zzÑÑ {ÑÑÑÖ R ÅÅÇ V n-type doped ZnO NWs.51−54 In this case, electronic transport k L,R T at the metal−ZnO interface of the nanocontact occurs through (3) thermionic emission, where current can be phenomenologi- cally expressed according to the relationship where W is the Lambert function, i.e., the solution of the ÄÅ ÉÑ transcendental equation W(x) exp (W(x)) = x. As discussed in ÅÅ i V y ÑÑ a previous work,55 a first-order correction term R± = R±0 + R±1 | Å Å jj zz Ñ I = I0ÅÅexpjj zz 1ÑÑÑ ÅÅ j VT z ÑÑ V| resulting in a voltage-dependent resistance needs to be ÅÇ k { ÑÖ (1) introduced in the model to account for the interface states at where V is the applied voltage, I0 is the inverse saturation the contacts. It is important to remark that expression (3) current, η is the ideality factor, VT = kT/e is the thermal considers different parameters for the two metal−insulator voltage at temperature T, k is the Boltzmann constant, and e is interfaces, thus allowing the model to represent both the electron charge. Note that the saturation current I0 symmetric and asymmetric contact device configurations. implicitly depends on the contact area, as well as on the Though this heuristic approximation provides a simple analytic Schottky barrier height, thus relying on the specific choice of treatment of the conduction problem, a more comprehensive the metal contact. Similarly, the ideality factor η representing and accurate solution can be found by directly solving the the deviation of the electrical characteristics from the ideal conduction characteristics of the combined devices using a thermionic emission theory relies on the specific metal circuit simulator, as presented later in Section 2.2 and considered for the contact. In this context, the pristine state described in detail in the Experimental Section. 53029 https://doi.org/10.1021/acsami.2c11022 ACS Appl. Mater. Interfaces 2022, 14, 53027−53037 ACS Applied Materials & Interfaces www.acsami.org Research Article Experimental data and modeling of a representative I−V Metal−Insulator Interfaces and Memristive Behavior. characteristic using (3) in an extended voltage range for a In the NW-based electrochemical metallization cell previously single NW device contacted by symmetric Pt electrodes is discussed, ionic movement can be coupled to electronic reported in Figure 2a. As can be seen, the I−V characteristics transport depending on the operating conditions. Indeed, a exhibit a sigmoidal-like shape, in which the linear behavior can positive voltage applied to an electrochemically active only be detected in a small voltage span around 0 V, while electrode causes the dissolution of atoms from the contact to current saturation is reached for larger applied voltages. This form ions that start to migrate along the NW surface toward behavior is a consequence of the current limitation imposed by the counter electrode under the action of the applied electric the inverse saturation currents of the diode. Notably, it can be field. As a consequence, a metallic filament bridging the two observed that the proposed model closely agrees with the electrodes can be formed.17 In single ZnO NW memristive experimental curve in the whole voltage range of analysis. cells, dissolution and migration of metal ions can be observed Instead, Figure 2b reports experimental data and modeling by positively biasing the Ag electrode due to the electro- results for a typical I−V characteristic in an extended voltage chemical activity of this metal, while the higher extraction range of an asymmetrically contacted ZnO NW with Pt and Ag barrier of Pt atoms from the contacts makes it an electro- electrodes. In this case, it can be observed that the I−V chemically inert electrode.58 Therefore, ionic migration of Ag+ characteristic shows, besides a rectifying behavior with ions with the consequent formation of an Ag conductive negligible current flowing through the device when the Ag/ filament along the ZnO NW can be observed by biasing the Ag ZnO interface is inversely biased, a tendency to reach the electrode in both asymmetric Pt/ZnO NW/Ag and symmetric current saturation regime when the Ag/ZnO interface is Ag/ZnO NW/Ag devices. The investigation of the NW-based positively biased. Also in this case, experimental data are well device through morphological and elemental analyses after interpolated by modeling. Finally, as reported in Figure 2c, resistive switching confirmed the ECM nature of the switching devices symmetrically contacted with Ag show a negligible mechanism related to the formation of an Ag conductive electronic conductance (not modeled) even when stimulated filament along the ZnO NW surface (experimental evidence in an extended voltage range. These results show that are reported in previous works17,59−61). The presence of Ag electronic transport in NW-based memristive cells in the nanoclusters after resistive switching mainly accumulated near pristine state can be well represented by the physics-based the Ag-biased electrode, as experimentally shown in a previous model described by eq 3. In particular, the close agreement work,25 suggests that the filament growth proceeds from the between the thermionic emission model and the experimental biased electrode and grows toward the counter electrode as discrete nanoclusters, as expected in resistive switching devices data is testified by the low values assumed for the ideality characterized by low cation mobility. Indeed, in low mobility, factors (approximately from 1 to 2). With these values, NW ions can pile and reach the critical nucleation conditions before resistances in the range of tens of MΩ to a few MΩ were reaching the counter electrode and filament growth can found, but this strongly depends also on the considered proceed by cluster displacement via repeated splitting− saturation currents. merging processes as discussed in ref 62. In this context, it is So far, the reported results demonstrate that the metal− worth noticing also that previous works reported resistive insulator interface in electrochemical metallization memories switching in single NWs symmetrically contacted by electro- can be used to regulate the electronic conductivity of the chemical interelectrodes such as Ti or Ti/Au electrodes.63−66 memristive cell in the pristine state. In particular, the electronic However, in these cases, the switching mechanism can be conduction in NW-based cells is observed to be mainly dominated by the TiO2 interfacial layer formed at the Ti/ZnO controlled by the inversely biased junction. Indeed, the interface, related to the higher affinity of Ti compared to Zn inversely biased junction acts as a blocking barrier that limits with O (the presence of this layer was experimentally revealed the overall electronic current flowing through the device. In by transmission electron microscopy (TEM) analyses by single ZnO NW devices, results show a more pronounced Chiang et al.66). The presence of this layer can strongly impact blocking action of the Ag/ZnO junction with respect to the the switching mechanism which in this case can be related to Pt/ZnO junction. This can be clearly seen from the data the migration of oxygen-related species at the ZnO/metal reported in Figure 2, where the current is negligible for all of interface and not along the NW. In our case, no resistive the cases where the Ag/ZnO junction is inversely biased (refer switching was observed in Pt/ZnO NW/Pt devices.17 The to Figure 2b,c). On the other hand, an inversely biased Pt/ absence of resistive switching phenomena in Pt/ZnO NW/Pt ZnO junction results in a nonnegligible saturation current that symmetric devices testify that the switching mechanism in Ag- was observed to be in the range of hundreds of nA (refer to contacted devices is purely ECM and is not related to the Figure 2a,b). It is worth noticing that a higher Schottky barrier migration of O species at the interface and/or along the NW, is theoretically expected from the Schottky−Mott rule at the as expected in the valence change memory (VCM) Pt/ZnO junction as a consequence of the higher Pt work mechanism. function (5.12 eV) compared to the Ag work function (4.26 After an initial assessment of the conductive filament eV). However, the ideal Schottky−Mott rule does not consider generation by an electroforming step (details given in Figure possible chemical interactions occurring at the interfaces S1), these devices exhibit resistive switching behavior because during the deposition of the metallic contacts.56 For example, of the formation/rupture of the Ag conductive pathway. It is the formation of interfacial chemical bonds at the Pt/ZnO worth noticing that the electroforming voltage depends on the junction can be related to the formation of eutectics in electrode spacing, where a substantial reduction of the voltage between Pt and Zn,57 causing a deviation from the ideal required for electroforming can be obtained by reducing the Schottky−Mott rule. It follows that the electronic conduction electrode spacing that, for the same applied voltage, results in a mechanism can be regulated by choosing contact materials higher electric field.17,25 In this context, the established metallic with appropriate work functions and interface chemistry. filamentary path along the NW surface provides an additional 53030 https://doi.org/10.1021/acsami.2c11022 ACS Appl. Mater. Interfaces 2022, 14, 53027−53037 ACS Applied Materials & Interfaces www.acsami.org Research Article Figure 3. Experimental and modeling ionic transport properties and memristive behavior of single ZnO NW devices. (a) Schematization of the ZnO NW memristive cells with asymmetric Pt and Ag electrodes and its resistive switching mechanism based on the migration of Ag+ ions from the Ag electrode along the NW toward the Pt counter electrode under the action of the applied electric field. The equivalent circuital representation of the cell is represented by the back-to-back connected Schottky diodes in an asymmetric configuration with the NW series resistance RNW (channel b), all in parallel with the variable resistance of the conductive filament (channel a). Experimental and modeled (b) memristive I−V characteristics of a Pt/ZnO NW/Ag device, (c) corresponding R−V characteristics showing the resistance window and (d) the ILRS/IHRS ratio as a function of the applied voltage. (e) Schematization of the ZnO NW memristive cells with symmetric Ag electrodes and its resistive switching mechanism based on the migration of Ag+ ions along the NWs under the action of the applied electric field. The equivalent circuital representation of the cell is represented by the back-to-back connected Schottky diodes in a symmetric configuration with the NW series resistance RNW (channel b), all in parallel with the variable resistance of the conductive filament (channel a). Experimental and modeled (f) memristive I−V characteristics of a Ag/ ZnO NW/Ag device, (g) corresponding R−V characteristics showing the resistance window and (h) the ILRS/IHRS ratio as a function of the applied voltage. transport channel for electronic conduction. Indeed, when the progressive arrival and departure of conducting species conductive filament is established, electrons can flow parallel to following a self-saturation process. Interestingly, the memory the pathway consisting in the NW resistance and the Schottky state eq 4 can be recursively solved by assuming a simulation interfaces previously discussed. The conductance properties of timestep Δt > 0 as the filament-related electronic conduction channel are P ( P+ D) t ( P + D) t regulated by ionic dynamics underlying its formation/rupture gt = [1 e ] + gt e 1 under the action of the applied electric field. The hysteretic P + D (6) conduction characteristic and memory state dynamics resulting from the formation/rupture of the conductive filament can be Note that eq 6, which is not expressed as a continuous function modeled by a voltage-dependent rate balance equation that of t, allows modeling the dynamics of the Ag conductive takes into account also short-term dynamics and volatility of filament that regulates the evolution of the internal memory the internal memory state related to the spontaneous relaxation state of the memristive cell for an arbitrary stimulation voltage, of the Ag conductive filament.67 provided Δt is small enough. Moreover, expression (6) In this framework, the memory state resulting from filament captures the essence of hysteretic conduction since the current dynamics can be described through the equation transport state (gt) depends on the previous condition (gt−1). For a constant applied voltage, eq 5 has analytic solution. Once dg the normalized conductance is found from eq 6, it can be = k p(1 g) kdg dt (4) straightforwardly used for computing the I−V characteristics of the device. In this connection, different models can be where g represents the memory state of the device in terms of considered for the electron transport, including, for example, its normalized conductance, while kp and kd are the a linear relationship I ∼ gV like in ref 67 or a more elaborated potentiation and depression coefficients, respectively, which expression such as I ∼ g sinh[α(V − IR)] where α is a fitting depend exponentially on the applied voltage as expected for constant and R is a series resistance, like the one used for the diffusive ionic processes model curves in Figure 3. Notice that in this latter case, both k p,d = k p0,d0 exp( p,d V) (5) for low voltages (HRS) and high currents (LRS), the I−V curve becomes linear as well because of the role played by R. It where ηp and ηd are the transition rates and kp0 and kd0 are the is worth stressing that the model equations describe the fitting constants. Equation 4 phenomenologically describes the formation/rupture of the filament without distinguishing the 53031 https://doi.org/10.1021/acsami.2c11022 ACS Appl. Mater. Interfaces 2022, 14, 53027−53037 ACS Applied Materials & Interfaces www.acsami.org Research Article Figure 4. Effects of the metal−insulator interface at the counter electrode on the memristive characteristics. (a) Schematization of the ECM cell. Until the filament bridges the two electrodes (i.e., d > 0), the electronic conduction mechanism is dominated by the inversely polarized metal− semiconductor interface (M−S) and the current flowing into the device is determined by the saturation current at the Schottky interface. (b) Evolution of the I−V memristive hysteretic loop, (c) corresponding R−V characteristics and (d) the ILRS/IHRS ratio for ECM cells by progressively increasing (arrow direction) the saturation current I0 at the metal−semiconductor interface of the counter electrode by modeling. contributions of each individual physicochemical processes adjusting the operation conditions such as the compliance involved, such as redox processes, nucleation, diffusion, current level (details in Supporting Information, Figure S3). Butler−Volmer transport, field-accelerated transport, chemical Note also that NW-based devices can be programmed by dissolution, Gibbs−Thomson effect etc. voltage pulses17 (Supporting Information, Figure S4). A schematization of the working principle of symmetric and Importantly, our modeling framework includes the effects of asymmetric memristive cells with their corresponding electrical adding a current compliance to the experimental setup. As circuital representations and resistive switching characteristics detailed in the Experimental Section, this limiting mechanism are reported in Figure 3a−h, respectively. As a consequence of is modeled through the use of a variable resistor in series with the coupling between ionics and electronics, asymmetric and the memristive structure. This allows to limit the current symmetric single NW-based cells can be electrically modeled in beyond the required maximum by adjusting the input voltage a first approximation as a parallel combination of two to the device, similarly to what happens in the experimental conductive channels for electrons (Figure 3a,e), i.e., the setup. During the time the current compliance is activated, the metallic filament channel along the NW (channel a) and the memory state of the device is allowed to evolve so that the ZnO semiconductor channel (channel b). In this framework, emerging point of the I−V characteristics when reducing the the internal cell voltage that can arise in asymmetrically applied voltage (RESET) is no longer the same as the entrance contacted electrodes related to the nanobattery effect68 is point corresponding to the increasing voltage (SET). This is considered negligible. While the electron flow through channel clearly seen, for example, in Figure 3f. a depends on the dynamics of the filament internal resistance Despite similar switching characteristics, symmetric and state regulated by the applied voltage (in turn described by the asymmetric devices exhibit a different behavior in the HRS. memory state equation and its solution, eqs 4 and 6, the flow Indeed, before SET events, the asymmetric device exhibits a of electrons through channel b is regulated by the NW sigmoidal-like behavior, while the symmetric device exhibits a resistance itself and the Schottky barriers at the interfaces as negligible current flow. This means that in the HRS the device described by the piecewise model with nonlinear series conductance is regulated by metal−ZnO interfaces, similarly to resistance correction, eq 3). Alternatively, a more compre- the pristine state case, where the inversely polarized junction at hensive outcome can be obtained by representing and solving the counter electrode regulates the electron flow. Thus, in the the whole system in a circuit simulator. The details of this HRS, where the filament is not formed, electrons flow mainly numerical approach are provided in the Experimental Section through the conduction channel b. Instead, similar behavior of this work. was observed in the LRS where the Ag conductive filament Experimental data and modeling results for representative I− shorts the two electrodes in both symmetric and asymmetric V resistive switching characteristics of asymmetrically con- configurations, and electrons flow mainly through the tacted Pt/ZnO NW/Ag and symmetrically contacted Ag/ZnO conduction channel a. NW/Ag devices are reported in Figure 3b,f, respectively The different electronic conduction mechanisms in the HRS (resistive switching characteristics of different devices from the of asymmetric and symmetric devices lead to distinctive same batch and devices from different batches are reported in resistance window characteristics, as reported in Figure 3c,g, Supporting Information, Figure S2). As can be observed, both respectively. Due to the nonlinearity of the electronic devices exhibit hysteretic behavior related to the device conduction mechanism, here the HRS exhibits a maximum switching from a high-resistance state (HRS) to a low- followed by a decrease as a function of the applied voltage. resistance state (LRS) in correspondence of the SET voltage. Instead, a monotonic increase of the HRS can be observed in The switching characteristics are in both cases volatile as a symmetric devices, as a consequence of the blocking character consequence of the spontaneous dissolution of the Ag of the reversely polarized Ag/ZnO contact. Thus, a wider conductive filament, as previously discussed. While in this resistance window can be observed in the R−V loop of work we focus on volatile switching characteristics usually symmetrically contacted NWs. This in turn reflects in a higher required for temporal processing of the input signal, a ON/OFF ratio of the symmetrically contacted device, as it can nonvolatile switching characteristic can be induced in both be observed by comparing the LRS/HRS current ratio as a symmetric and asymmetric device configurations by properly function of the voltage in the hysteresis loop for asymmetric 53032 https://doi.org/10.1021/acsami.2c11022 ACS Appl. Mater. Interfaces 2022, 14, 53027−53037 ACS Applied Materials & Interfaces www.acsami.org Research Article Figure 5. Memristive behavior of single ZnO NWs with symmetric Ag electrodes. (a) I−V characteristics showing the resistive switching behavior over cycling. All of the 250 cycles are reported in light gray, while the median curve is reported in bold blue. (b) Endurance characteristics (read voltage of 0.4 V), (c) SET voltage distribution, and (d) distribution of conductance values in the LRS normalized by the fundamental quantum of conductance G0. Data of panels (b−d) were extracted from 250 full-sweep cycles. The distributions in panels (c) and (d) are interpolated by normal distributions. (Figure 3d) and symmetric (Figure 3h) devices. It is worth corresponding resistance window as well as the LRS and HRS noticing also that, since the electronic transport properties in current ratio, as revealed by the model results reported in the HRS are mainly regulated by the reversely polarized Figure 4b−d, respectively. In this context, it is important to Schottky junction at the metal/ZnO interface and not by the remark that, as experimentally shown in the single NW device, ZnO NW itself, no significant differences are expected at the the choice of the counter electrode material has to take into Schottky barrier in the pristine state and after resistive account both the theoretical Schottky barrier height expected switching loops. This hypothesis is supported by experimental by considering the counter electrode metal work function and data that shows a qualitative good overlapping of I−V traces the interface chemistry. relative to the electroforming step before forming and the HRS In single NWs, results show that the choice of Ag as a of the device before SET (Supporting Information, Figure S5). counter electrode allows to reduce the saturation current when In addition, these data also show no significant changes over reversely polarized, enhancing the ON/OFF switching ratio, as cycling of the reversely polarized Schottky barrier that well as the separation of resistance states. In this framework, it regulates the HRS, as testified by the good overlapping of I− is worth stressing that the switching characteristics of these V traces in the HRS region over cycling. symmetrically contacted devices with Ag electrodes are More in general, these results indicate that the hysteretic repeatable over cycling, as reported in Figure 5a. High stability memristive behavior can be tuned by a proper choice of the and low dispersion of both HRS and LRS were observed in the metal−insulator Schottky barrier at the counter electrode. endurance characteristics acquired during 250 full-sweep Indeed, before the conductive filament bridges the two metallic cycles, as reported in Figure 5b. In addition, a narrow electrodes, the reversely polarized Schottky interface at the dispersion of the SET voltage over cycling was observed. counter electrode controls the flow of electrons in the device, According to experiments, the SET voltage distribution where the electron flow corresponds to the saturation current reported in Figure 5c has a mean value of 0.69 V with a of the reversely biased junction (Figure 4a). It turns out that, standard deviation of 0.13 V (details of the SET voltage as a by progressively decreasing the saturation current of the function of cycling are reported in Supporting Information, junction through a clever choice of the electrode material, it is Figure S6). Interestingly, the distribution of the device possible to enlarge not only the hysteretic loop but also the conductance over cycling in the LRS normalized with respect 53033 https://doi.org/10.1021/acsami.2c11022 ACS Appl. Mater. Interfaces 2022, 14, 53027−53037 ACS Applied Materials & Interfaces www.acsami.org Research Article to the fundamental quantum of conductance G0 reported in with a submillimetric prepatterned probe circuit. A two-step electron Figure 5d shows that the device internal state of conductance is beam lithography and metal deposition were performed to obtain lower than G0 (G/G0 mean value of 0.22, standard deviation of asymmetrically contacted NW-based memrsitive cells. A detailed 0.03). While values close to G0 are expected when the filament description of the synthesis of ZnO NWs and fabrication steps of single NW devices can be found in our previous work.46 size reaches the atomic scale favoring the formation of a Electrical Characterization. Electrical characterization was quantum point contact,69 values of G < G0 mean that the performed in two-terminal configuration by a Keithley 6430 formed filament does not allow the complete transmission of subfemtometer sourcemeter with a remote preamplifier, a Keithley the electron wave packet. This corresponds to the tunneling 2636A, and a probe station. Electrical measurements of memristive regime through a subband. The effect can be explained by the devices in the pristine state in an extended voltage range (Figure 2) formation of a noncontinuous metallic filament with the were performed on devices with a high electrode spacing to prevent presence of an extremely narrow gap in the form of a broken- ionic migration, thus avoiding resistive switching phenomena. All down filament and/or in the form of a sparse concentration of electrical measurements were performed in air at room temperature. scatterers. Modeling. A memristive cell model consisting of two back-to-back Schottky barriers in series with a nonlinear resistance (channel a) in ■ CONCLUSIONS In conclusion, we have experimentally investigated and combination with a parallel conducting channel exhibiting memristive properties (channel b) was implemented in LTspice (circuit simulator from Analog Devices). The script used for modeling is reported in modeled the role played by metal−insulator interfaces in Supporting Information, Figure S7 and contains five parts: parameter ECM cells, by considering single-crystalline ZnO NW-based definition, memory state equation, current−voltage characteristic for devices as model systems. To this end, a number of symmetric the memristive device, current−voltage characteristic for the nano- and asymmetric device configurations have been considered, wire, and auxiliary functions. The memristive behavior is based on an adaptation of the memdiode model for resistive switching devices where Ag and Pt are exploited as representative electrochemi- reported in ref 70. The model includes the snapback effect and an cally active and inert electrodes, respectively. A comparison of internal series resistance. A schematic representation of the circuit the electronic conduction properties of different devices used for modeling is reported in Supporting Information, Figure S8. highlights that the electron flow is mainly regulated by the metal−insulator interfaces. In this framework, experiments are in close agreement with the modeling of the memristive cells considered as two parallel conducting channels for electrons. ■ * ASSOCIATED CONTENT sı Supporting Information The first channel, which dominates in the pristine state and in The Supporting Information is available free of charge at the HRS, is represented by back-to-back connected Schottky https://pubs.acs.org/doi/10.1021/acsami.2c11022. diodes in series with the NW resistance, while the conductance of the metallic filament channel, which dominates in the LRS, Typical electroforming characteristics of a single NW relies on the filament dynamics. Results show that the memristive cell by applying a positive bias to an hysteretic characteristics can be modified by appropriate electrochemically active Ag electrode (Figure S1); choice of the counter electrode metal that controls the flow resistive switching behavior of different Ag/ZnO NW/ of electrons in the HRS, as confirmed by experiments and Pt devices from the same batch and from a different modeling. In particular, a blocking counter electrode was batch (Figure S2); nonvolatile resistive switching in observed to increase the resistance window, as reported in symmetrically contacted ZnO NWs with Ag electrodes symmetrically contacted devices with Ag electrodes that are (Figure S3); resistive switching behavior of an Ag/ZnO shown to exhibit also low SET voltages, together with NW/Pt device observed under voltage pulse stimulation reproducible switching characteristics. Besides highlighting (Figure S4); comparison of I−V traces of electroforming the importance of the counter electrode in memristive devices, and resistive switching cycles of an Ag/ZnO NW/PT our results show that it is possible to cleverly design the device (Figure S5); SET voltage as a function of the hysteretic response of the memristive cell by controlling not switching cycle (Figure S6); script for modeling the only ionic transport properties but also the electronic aspect of memristive cell with two back-to-back Schottky barrier the problem. To this aim, an appropriate choice of the contact systems (Figure S7); and LTspice circuit representation materials should be done by taking into account not only the (Figure S8) (PDF) Schottky barrier height at the interface but also the interface chemistry. More in general, these results clearly point out that a rational design of memristive cells should rely on the selection of materials that properly regulate the electronic ■ AUTHOR INFORMATION Corresponding Authors transport properties at the metal−insulator interfaces. Gianluca Milano − Advanced Materials Metrology and Life ■ EXPERIMENTAL SECTION Single NW Memristive Cell Fabrication. ZnO NWs were Sciences Division, INRiM (Istituto Nazionale di Ricerca Metrologica), 10135 Torino, Italy; orcid.org/0000-0002- 1983-6516; Email: g.milano@inrim.it synthesized through low-pressure chemical vapor deposition by a self- Carlo Ricciardi − Department of Applied Science and seeding vapor−solid (VS) mechanism that involves the nucleation of Technology, Politecnico di Torino, 10129 Torino, Italy; ZnO and subsequent NW growth on a Pt substrate from the orcid.org/0000-0002-4703-7949; evaporation of a Zn source in an O2 environment. The median NW length and diameter were assessed to be 1.6 μm and 100 nm, Email: carlo.ricciardi@polito.it respectively. Each nanowire is a single crystal with high chemical Authors purity and is characterized by a clean surface without any amorphous surface layers. Fabrication of single NW memristive cells was Enrique Miranda − Departament d’Enginyeria Electrònica, performed by combined optical and electron beam lithography Universitat Autònoma de Barcelona (UAB), 08193 (EBL), by connecting dispersed NWs on an insulating SiO2 substrate Cerdanyola del Vallès, Spain 53034 https://doi.org/10.1021/acsami.2c11022 ACS Appl. Mater. Interfaces 2022, 14, 53027−53037 ACS Applied Materials & Interfaces www.acsami.org Research Article Matteo Fretto − Advanced Materials Metrology and Life Emerging Neuromorphic Devices: Fundamentals, Progress, and Sciences Division, INRiM (Istituto Nazionale di Ricerca Challenges. Adv. Mater. 2019, 31, No. 1902761. Metrologica), 10135 Torino, Italy (7) Yang, J. Q.; Wang, R.; Ren, Y.; Mao, J. Y.; Wang, Z. P.; Zhou, Y.; Ilia Valov − JARA�Fundamentals for Future Information Han, S. T. Neuromorphic Engineering: From Biological to Spike- Technology, 52425 Jülich, Germany; Peter-Grünberg-Institut Based Hardware Nervous Systems. Adv. Mater. 2020, 32, No. 2003610. (PGI 7), Forschungszentrum Jülich, 52425 Jülich, Germany; (8) Wang, Z.; Wu, H.; Burr, G. W.; Hwang, C. S.; Wang, K. L.; Xia, orcid.org/0000-0002-0728-7214 Q.; Yang, J. J. Resistive Switching Materials for Information Complete contact information is available at: Processing. Nat. Rev. Mater. 2020, 5, 173−195. https://pubs.acs.org/10.1021/acsami.2c11022 (9) Lanza, M.; Sebastian, A.; Lu, W. D.; le Gallo, M.; Chang, M.-F.; Akinwande, D.; Puglisi, F. M.; Alshareef, H. N.; Liu, M.; Roldan, J. B. Notes Memristive Technologies for Data Storage, Computation, Encryption, The authors declare no competing financial interest. and Radio-Frequency Communication. Science 2022, 376, No. eabj9979. The data that support the findings of this study are available on (10) Xiong, X.; Xiong, F.; Tian, H.; Wang, Z.; Wang, Y.; Tao, R.; Zenodo (https://doi.org/10.5281/zenodo.6976318). All other Klausen, L. H.; Dong, M. Ultrathin Anion Conductors Based data are available from the authors. Memristor. Adv. Electron. Mater. 2022, 8, No. 2100845. ■ ACKNOWLEDGMENTS Part of this work was supported by the European project (11) Khot, A. C.; Dongale, T. D.; Park, J. H.; Kesavan, A. V.; Kim, T. G. Ti 3 C 2 -Based MXene Oxide Nanosheets for Resistive Memory and Synaptic Learning Applications. ACS Appl. Mater. Interfaces 2021, 13, 5216−5227. MEMQuD, code 20FUN06. This project (EMPIR 20FUN06 (12) Dittmann, R.; Strachan, J. P. Redox-Based Memristive Devices MEMQuD) has received funding from the EMPIR program for New Computing Paradigm. APL Mater. 2019, 7, No. 110903. cofinanced by the Participating States and from the European (13) Waser, R.; Dittmann, R.; Staikov, G.; Szot, K. Redox-Based Union’s Horizon 2020 research and innovation program. Part Resistive Switching Memories-Nanoionic Mechanisms, Prospects, and of this work has been carried out at NanoFacility Piemonte Challenges. Adv. Mater. 2009, 21, 2632−2663. INRiM, a laboratory supported by the “Compagnia di San (14) Waser, R.; Aono, M. Nanoionics-Based Resistive Switching Paolo” Foundation, and at the QR Laboratories, INRiM. The Memories. Nat. Mater. 2007, 6, 833−840. support of Thomas Poessinger with graphics is gratefully (15) Valov, I.; Waser, R.; Jameson, J. R.; Kozicki, M. N. acknowledged. Electrochemical Metallization Memories�Fundamentals, Applica- tions, Prospects. Nanotechnology 2011, 22, No. 289502. ■ REFERENCES (1) Christensen, D. V.; Dittmann, R.; Linares-Barranco, B.; (16) Wang, W.; Wang, M.; Ambrosi, E.; Bricalli, A.; Laudato, M.; Sun, Z.; Chen, X.; Ielmini, D. Surface Diffusion-Limited Lifetime of Silver and Copper Nanofilaments in Resistive Switching Devices. Nat. Sebastian, A.; Le Gallo, M.; Redaelli, A.; Slesazeck, S.; Mikolajick, Commun. 2019, 10, No. 81. T.; Spiga, S.; Menzel, S.; Valov, I.; Milano, G.; Ricciardi, C.; Liang, S.- (17) Milano, G.; Luebben, M.; Ma, Z.; Dunin-Borkowski, R.; J.; Miao, F.; Lanza, M.; Quill, T. J.; Keene, S. T.; Salleo, A.; Grollier, Boarino, L.; Pirri, C. F.; Waser, R.; Ricciardi, C.; Valov, I. Self-Limited J.; Markovic, D.; Mizrahi, A.; Yao, P.; Yang, J. J.; Indiveri, G.; Single Nanowire Systems Combining All-in-One Memristive and Strachan, J. P.; Datta, S.; Vianello, E.; Valentian, A.; Feldmann, J.; Li, Neuromorphic Functionalities. Nat. Commun. 2018, 9, No. 5151. X.; Pernice, W. H.; Bhaskaran, H.; Furber, S.; Neftci, E.; Scherr, F.; (18) Ohno, T.; Hasegawa, T.; Tsuruoka, T.; Terabe, K.; Gimzewski, Maass, W.; Ramaswamy, S.; Tapson, J.; Panda, P.; Kim, Y.; Tanaka, J. K.; Aono, M. Short-Term Plasticity and Long-Term Potentiation G.; Thorpe, S.; Bartolozzi, C.; Cleland, T. A.; Posch, C.; Liu, S.-C.; Mimicked in Single Inorganic Synapses. Nat. Mater. 2011, 10, 591− Panuccio, G.; Mahmud, M.; Mazumder, A. N.; Hosseini, M.; 595. Mohsenin, T.; Donati, E.; Tolu, S.; Galeazzi, R.; Christensen, M. (19) Milano, G.; Pedretti, G.; Montano, K.; Ricci, S.; Hashemkhani, E.; Holm, S.; Ielmini, D.; Pryds, N. 2022 Roadmap on Neuromorphic S.; Boarino, L.; Ielmini, D.; Ricciardi, C. In Materia Reservoir Computing and Engineering. Neuromorphic Comput. Eng. 2022, 2, Computing with a Fully Memristive Architecture Based on Self- No. 022501. Organizing Nanowire Networks. Nat. Mater. 2022, 21, 195−202. (2) Berggren, K.; Xia, Q.; Likharev, K. K.; Strukov, D. B.; Jiang, H.; (20) Zhong, Y.; Tang, J.; Li, X.; Gao, B.; Qian, H.; Wu, H. Dynamic Mikolajick, T.; Querlioz, D.; Salinga, M.; Erickson, J. R.; Pi, S.; Xiong, Memristor-Based Reservoir Computing for High-Efficiency Temporal F.; Lin, P.; Li, C.; Chen, Y.; Xiong, S.; Hoskins, B. D.; Daniels, M. W.; Madhavan, A.; Liddle, J. A.; McClelland, J. J.; Yang, Y.; Rupp, J.; Signal Processing. Nat. Commun. 2021, 12, No. 408. Nonnenmann, S. S.; Cheng, K.-T.; Gong, N.; Lastras-Montaño, M. A.; (21) Midya, R.; Wang, Z.; Asapu, S.; Zhang, X.; Rao, M.; Song, W.; Talin, A. A.; Salleo, A.; Shastri, B. J.; de Lima, T. F.; Prucnal, P.; Tait, Zhuo, Y.; Upadhyay, N.; Xia, Q.; Yang, J. J. Reservoir Computing A. N.; Shen, Y.; Meng, H.; Roques-Carmes, C.; Cheng, Z.; Bhaskaran, Using Diffusive Memristors. Adv. Intell. Syst. 2019, 1, No. 1900084. H.; Jariwala, D.; Wang, H.; Shainline, J. M.; Segall, K.; Yang, J. J.; Roy, (22) Moon, J.; Ma, W.; Shin, J. H.; Cai, F.; Du, C.; Lee, S. H.; Lu, W. K.; Datta, S.; Raychowdhury, A. Roadmap on Emerging Hardware D. Temporal Data Classification and Forecasting Using a Memristor- and Technology for Machine Learning. Nanotechnology 2021, 32, Based Reservoir Computing System. Nat. Electron. 2019, 2, 480−487. No. 012002. (23) Milano, G.; Pedretti, G.; Fretto, M.; Boarino, L.; Benfenati, F.; (3) Li, H.; Wang, S.; Zhang, X.; Wang, W.; Yang, R.; Sun, Z.; Feng, Ielmini, D.; Valov, I.; Ricciardi, C. Brain-Inspired Structural Plasticity W.; Lin, P.; Wang, Z.; Sun, L.; Yao, Y. Memristive Crossbar Arrays for through Reweighting and Rewiring in Multi-Terminal Self-Organizing Storage and Computing Applications. Adv. Intell. Syst. 2021, 3, Memristive Nanowire Networks. Adv. Intell. Syst. 2020, 2, No. 2100017. No. 2000096. (4) Ielmini, D.; Pedretti, G. Device and Circuit Architectures for In- (24) Milano, G.; Porro, S.; Valov, I.; Ricciardi, C. Recent Memory Computing. Adv. Intell. Syst. 2020, 2, No. 2000040. Developments and Perspectives for Memristive Devices Based on (5) Xia, Q.; Yang, J. J. Memristive Crossbar Arrays for Brain-Inspired Metal Oxide Nanowires. Adv. Electron. Mater. 2019, 5, No. 1800909. Computing. Nat. Mater. 2019, 18, 309−323. (25) Milano, G.; Raffone, F.; Luebben, M.; Boarino, L.; Cicero, G.; (6) Tang, J.; Yuan, F.; Shen, X.; Wang, Z.; Rao, M.; He, Y.; Sun, Y.; Valov, I.; Ricciardi, C. Water-Mediated Ionic Migration in Memristive Li, X.; Zhang, W.; Li, Y.; Gao, B.; Qian, H.; Bi, G.; Song, S.; Yang, J. Nanowires with a Tunable Resistive Switching Mechanism. ACS Appl. J.; Wu, H. Bridging Biological and Artificial Neural Networks with Mater. Interfaces 2020, 12, 48773−48780. 53035 https://doi.org/10.1021/acsami.2c11022 ACS Appl. Mater. Interfaces 2022, 14, 53027−53037 ACS Applied Materials & Interfaces www.acsami.org Research Article (26) Nagashima, K.; Yanagida, T.; Oka, K.; Kanai, M.; Klamchuen, Cation-Based Nanoscale Resistive Switching Memories. ACS Nano A.; Kim, J.-S.; Park, B. H.; Kawai, T. Intrinsic Mechanism of 2013, 7, 6396−6402. Memristive Switching. Nano Lett. 2011, 11, 2114−2118. (46) Milano, G.; Boarino, L.; Valov, I.; Ricciardi, C. Memristive (27) Nagashima, K.; Yanagida, T.; Oka, K.; Taniguchi, M.; Kawai, Devices Based on Single ZnO Nanowires�from Material Synthesis to T.; Kim, J. S.; Park, B. H. Resistive Switching Multistate Nonvolatile Neuromorphic Functionalities. Semicond. Sci. Technol. 2022, 37, Memory Effects in a Single Cobalt Oxide Nanowire. Nano Lett. 2010, No. 034002. 10, 1359−1363. (47) Milano, G.; Porro, S.; Ali, M. Y.; Bejtka, K.; Bianco, S.; (28) Oka, K.; Yanagida, T.; Nagashima, K.; Kanai, M.; Kawai, T.; Beccaria, F.; Chiolerio, A.; Pirri, C. F.; Ricciardi, C. Unravelling Kim, J. S.; Park, B. H. Spatial Nonuniformity in Resistive-Switching Resistive Switching Mechanism in ZnO NW Arrays: The Role of the Memory Effects of Nio. J. Am. Chem. Soc. 2011, 133, 12482−12485. Polycrystalline Base Layer. J. Phys. Chem. C 2018, 122, 866−874. (29) Oka, K.; Yanagida, T.; Nagashima, K.; Tanaka, H.; Kawai, T. (48) Porro, S.; Risplendi, F.; Cicero, G.; Bejtka, K.; Milano, G.; Nonvolatile Bipolar Resistive Memory Switching in Single Crystalline Rivolo, P.; Jasmin, A.; Chiolerio, A.; Pirri, C. F.; Ricciardi, C. Multiple Nio Heterostructured Nanowires. J. Am. Chem. Soc. 2009, 131, 3434− Resistive Switching in Core−Shell ZnO Nanowires Exhibiting 3435. Tunable Surface States. J. Mater. Chem. C 2017, 5, 10517−10523. (30) Bandopadhyay, K.; Prajapati, K. N.; Mitra, J. Resistive (49) Yin, H.; Wang, Q.; Geburt, S.; Milz, S.; Ruttens, B.; Degutis, G.; Switching in Individual ZnO Nanorods: Delineating the Ionic D’Haen, J.; Shan, L.; Punniyakoti, S.; D’Olieslaeger, M.; Wagner, P.; Current by Photo-Stimulation. Nanotechnology 2018, 29, No. 105701. Ronning, C.; Boyen, H.-G. Controlled Synthesis of Ultrathin ZnO (31) Valov, I.; Lu, W. D. Nanoscale Electrochemistry Using Nanowires Using Micellar Gold Nanoparticles as Catalyst Templates. Dielectric Thin Films as Solid Electrolytes. Nanoscale 2016, 8, Nanoscale 2013, 5, 7046. 13828−13837. (50) Zhang, Z.; Yao, K.; Liu, Y.; Jin, C.; Liang, X.; Chen, Q.; Peng, (32) Lübben, M.; Valov, I. Active Electrode Redox Reactions and L.-M. Quantitative Analysis of Current-Voltage Characteristics of Device Behavior in ECM Type Resistive Switching Memories. Adv. Semiconducting Nanowires: Decoupling of Contact Effects. Adv. Electron. Mater. 2019, 5, No. 1800933. Funct. Mater. 2007, 17, 2478−2489. (33) Liu, Q.; Sun, J.; Lv, H.; Long, S.; Yin, K.; Wan, N.; Li, Y.; Sun, (51) Milano, G.; D’Ortenzi, L.; Bejtka, K.; Ciubini, B.; Porro, S.; L.; Liu, M. Real-Time Observation on Dynamic Growth/Dissolution Boarino, L.; Ricciardi, C. Metal−Insulator Transition in Single of Conductive Filaments in Oxide-Electrolyte-Based ReRAM. Adv. Crystalline ZnO Nanowires. Nanotechnology 2021, 32, No. 185202. Mater. 2012, 24, 1844−1849. (52) Janotti, A.; Van de Walle, C. G. Fundamentals of Zinc Oxide as (34) Celano, U.; Goux, L.; Belmonte, A.; Opsomer, K.; Franquet, A.; a Semiconductor. Rep. Prog. Phys. 2009, 72, No. 126501. Schulze, A.; Detavernier, C.; Richard, O.; Bender, H.; Jurczak, M.; (53) Chiu, S.-P.; Lin, Y.-H.; Lin, J.-J. Electrical Conduction Vandervorst, W. Three-Dimensional Observation of the Conductive Mechanisms in Natively Doped ZnO Nanowires. Nanotechnology Filament in Nanoscaled Resistive Memory Devices. Nano Lett. 2014, 2009, 20, No. 015203. 14, 2401−2406. (54) Tsai, L.-T.; Chiu, S.-P.; Lu, J. G.; Lin, J.-J. Electrical (35) Yang, Y.; Gao, P.; Gaba, S.; Chang, T.; Pan, X.; Lu, W. Conduction Mechanisms in Natively Doped ZnO Nanowires (II). Observation of Conducting Filament Growth in Nanoscale Resistive Nanotechnology 2010, 21, No. 145202. Memories. Nat. Commun. 2012, 3, No. 732. (55) Miranda, E.; Milano, G.; Ricciardi, C. Compact Modeling of (36) Sun, W.; Gao, B.; Chi, M.; Xia, Q.; Yang, J. J.; Qian, H.; Wu, H. the I-V Characteristics of ZnO Nanowires Including Nonlinear Series Understanding Memristive Switching via in Situ Characterization and Resistance Effects. IEEE Trans. Nanotechnol. 2020, 19, 297−300. Device Modeling. Nat. Commun. 2019, 10, No. 3453. (56) Liu, Y.; Guo, J.; Zhu, E.; Liao, L.; Lee, S.; Ding, M.; Shakir, I.; (37) Lübben, M.; Cüppers, F.; Mohr, J.; von Witzleben, M.; Breuer, U.; Waser, R.; Neumann, C.; Valov, I. Design of Defect-Chemical Gambin, V.; Huang, Y.; Duan, X. Approaching the Schottky−Mott Properties and Device Performance in Memristive Systems. Sci. Adv. Limit in van Der Waals Metal−Semiconductor Junctions. Nature 2020, 6, No. eaaz9079. 2018, 557, 696−700. (38) Valov, I.; Tsuruoka, T. Effects of Moisture and Redox (57) Brillson, L. J.; Lu, Y. ZnO Schottky Barriers and Ohmic Reactions in VCM and ECM Resistive Switching Memories. J. Phys. Contacts. J. Appl. Phys. 2011, 109, No. 121301. D: Appl. Phys. 2018, 51, No. 413001. (58) Raffone, F.; Cicero, G. Does Platinum Play a Role in the (39) Messerschmitt, F.; Kubicek, M.; Rupp, J. L. M. How Does Resistance Switching of ZnO Nanowire-Based Devices? Solid State Moisture Affect the Physical Property of Memristance for Anionic- Ionics 2017, 299, 93−95. Electronic Resistive Switching Memories? Adv. Funct. Mater. 2015, 25, (59) Yang, Y.; Zhang, X.; Gao, M.; Zeng, F.; Zhou, W.; Xie, S.; Pan, 5117−5125. F. Nonvolatile Resistive Switching in Single Crystalline ZnO (40) Milano, G.; Luebben, M.; Laurenti, M.; Boarino, L.; Ricciardi, Nanowires. Nanoscale 2011, 3, 1917. C.; Valov, I. Structure-Dependent Influence of Moisture on Resistive (60) Qi, J.; Huang, J.; Paul, D.; Ren, J.; Chu, S.; Liu, J. Current Self- Switching Behavior of ZnO Thin Films. Adv. Mater. Interfaces 2021, 8, Complianced and Self-Rectifying Resistive Switching in Ag- No. 2100915. Electroded Single Na-Doped ZnO Nanowires. Nanoscale 2013, 5, (41) Milano, G.; Luebben, M.; Laurenti, M.; Porro, S.; Bejtka, K.; 2651. Bianco, S.; Breuer, U.; Boarino, L.; Valov, I.; Ricciardi, C. Ionic (61) Milano, G.; Boarino, L.; Ricciardi, C. Junction Properties of Modulation of Electrical Conductivity of ZnO Due to Ambient Single ZnO Nanowires with Asymmetrical Pt and Cu Contacts. Moisture. Adv. Mater. Interfaces 2019, 6, No. 1900803. Nanotechnology 2019, 30, No. 244001. (42) Chen, S.; Valov, I. Design of Materials Configuration for (62) Yang, Y.; Gao, P.; Li, L.; Pan, X.; Tappertzhofen, S.; Choi, S.; Optimizing Redox-Based Resistive Switching Memories. Adv. Mater. Waser, R.; Valov, I.; Lu, W. D. Electrochemical Dynamics of 2022, 34, No. 2105022. Nanoscale Metallic Inclusions in Dielectrics. Nat. Commun. 2014, 5, (43) Valov, I. Interfacial Interactions and Their Impact on Redox- No. 4232. Based Resistive Switching Memories (ReRAMs). Semicond. Sci. (63) Zhang, R.; Pang, W.; Zhang, Q.; Chen, Y.; Chen, X.; Feng, Z.; Technol. 2017, 32, No. 093006. Yang, J.; Zhang, D. Enhanced Non-Volatile Resistive Switching in (44) Cho, D.-Y.; Luebben, M.; Wiefels, S.; Lee, K.-S.; Valov, I. Suspended Single-Crystalline ZnO Nanowire with Controllable Interfacial Metal−Oxide Interactions in Resistive Switching Memo- Multiple States. Nanotechnology 2016, 27, No. 315203. ries. ACS Appl. Mater. Interfaces 2017, 9, 19287−19295. (64) Lai, Y.; Xin, P.; Cheng, S.; Yu, J.; Zheng, Q. Plasma Enhanced (45) Tappertzhofen, S.; Valov, I.; Tsuruoka, T.; Hasegawa, T.; Multistate Storage Capability of Single ZnO Nanowire Based Waser, R.; Aono, M. Generic Relevance of Counter Charges for Memory. Appl. Phys. Lett. 2015, 106, No. 031603. 53036 https://doi.org/10.1021/acsami.2c11022 ACS Appl. Mater. Interfaces 2022, 14, 53027−53037 ACS Applied Materials & Interfaces www.acsami.org Research Article (65) Lai, Y.; Qiu, W.; Zeng, Z.; Cheng, S.; Yu, J.; Zheng, Q. Resistive Switching of Plasma−Treated Zinc Oxide Nanowires for Resistive Random Access Memory. Nanomaterials 2016, 6, No. 16. (66) Chiang, Y. D.; Chang, W. Y.; Ho, C. Y.; Chen, C. Y.; Ho, C. H.; Lin, S. J.; Wu, T. B.; He, J. H. Single-ZnO-Nanowire Memory. IEEE Trans. Electron Devices 2011, 58, 1735−1740. (67) Miranda, E.; Milano, G.; Ricciardi, C. Modeling of Short-Term Synaptic Plasticity Effects in ZnO Nanowire-Based Memristors Using a Potentiation-Depression Rate Balance Equation. IEEE Trans. Nanotechnol. 2020, 19, 609−612. (68) Valov, I.; Linn, E.; Tappertzhofen, S.; Schmelzer, S.; van den Hurk, J.; Lentz, F.; Waser, R. Nanobatteries in Redox-Based Resistive Switches Require Extension of Memristor Theory. Nat. Commun. 2013, 4, No. 1771. (69) Milano, G.; Aono, M.; Boarino, L.; Celano, U.; Hasegawa, T.; Kozicki, M.; Majumdar, S.; Menghini, M.; Miranda, E.; Ricciardi, C.; Tappertzhofen, S.; Terabe, K.; Valov, I. Quantum Conductance in Memristive Devices: Fundamentals, Developments, and Applications. Adv. Mater. 2022, 34, No. 2201248. (70) Aguirre, F. L.; Suñé, J.; Miranda, E. SPICE Implementation of the Dynamic Memdiode Model for Bipolar Resistive Switching Devices. Micromachines 2022, 13, 330. 53037 https://doi.org/10.1021/acsami.2c11022 ACS Appl. Mater. Interfaces 2022, 14, 53027−53037