Addressable Direct-Write Nanoscale Filament ... - ACS Publications

May 1, 2017 - Addressable Direct-Write Nanoscale Filament. Formation and Dissolution by Nanoparticle-. Mediated Bipolar Electrochemistry. Garrison M...
1 downloads 0 Views 4MB Size
Download PDF
Addressable Direct-Write Nanoscale Filament Formation and Dissolution by NanoparticleMediated Bipolar Electrochemistry Garrison M. Crouch,†,# Donghoon Han,†,# Susan K. Fullerton-Shirey,‡ David B. Go,*,†,§ and Paul W. Bohn*,†,∥ †

Department Department § Department ∥ Department ‡

of of of of

Chemical and Biomolecular Engineering, University of Notre Dame, Notre Dame, Indiana 46556, United States Chemical and Petroleum Engineering, University of Pittsburgh, Pittsburgh, Pennsylvania 15213, United States Aerospace and Mechanical Engineering, University of Notre Dame, Notre Dame, Indiana 46556, United States Chemistry and Biochemistry, University of Notre Dame, Notre Dame, Indiana 46556, United States

S Supporting Information *

ABSTRACT: Nanoscale conductive filaments, usually associated with resistive memory or memristor technology, may also be used for chemical sensing and nanophotonic applications; however, realistic implementation of the technology requires precise knowledge of the conditions that control the formation and dissolution of filaments. Here we describe and characterize an addressable directwrite nanoelectrochemical approach to achieve repeatable formation/dissolution of Ag filaments across a ∼100 nm poly(ethylene oxide) (PEO) film containing either Ag+ alone or Ag+ together with 50 nm Ag-nanoparticles acting as bipolar electrodes. Using a conductive AFM tip, formation occurs when the PEO film is subjected to a forward bias, and dissolution occurs under reverse bias. Formation−dissolution kinetics were studied for three film compositions: Ag|PEO-Ag+, Ag|poly(ethylene glycol) monolayer-PEO-Ag+, and Ag|poly(ethylene glycol) monolayer-PEOAg+/Ag-nanoparticle. Statistical analysis shows that the distribution of formation times exhibits Gaussian behavior, and the fastest average initial formation time occurs for the Ag|PEO-Ag+ system. In contrast, formation in the presence of Ag nanoparticles likely proceeds by a noncontact bipolar electrochemical mechanism, exhibiting the slowest initial filament formation. Dissolution times are log-normal for all three systems, and repeated reformation of filaments from previously formed structures is characterized by rapid regrowth. The direct-write bipolar electrochemical deposition/dissolution strategy developed here presents an approach to reconfigurable, noncontact in situ wiring of nanoparticle arraysthereby enabling applications where actively controlled connectivity of nanoparticle arrays is used to manipulate nanoelectronic and nanophotonic behavior. The system further allows for facile manipulation of experimental conditions while simultaneously characterizing surface conditions and filament formation/dissolution kinetics. KEYWORDS: solid polymer electrolyte, conductive filament, resistive switching, conductive AFM, formation−dissolution kinetics onductive filament growth by electrodeposition through polymer electrolytes is of particular interest as a potential approach to devices such as resistive random-access memory (ReRAM),1 electrochemical metallization memory (ECM),2 and chemical sensors,3,4 or for reconfigurable materials, such as nanoparticle arrays for nanoelectronic and nanophotonic applications.5 These devices and materials function by exhibiting a binary change in electrical resistance, with distinct “on” and “off” states typically separated by several orders of magnitude; this behavior is based on the redox formation, and subsequent dissolution, of a conductive filament between two fixed electrodes through an

insulating electrolyte.6 The scale of the filament plays an important role in determining device behavior; as the thickness of the filament approaches the electron wavelength in the filament material, the conductance can exhibit quantized behavior characteristic of ballistic charge transport due to the creation of a quantum-point contact.4 In devices with larger filaments outside of this quantum regime, evidence suggests

C

© 2017 American Chemical Society

Received: March 8, 2017 Accepted: May 1, 2017 Published: May 1, 2017 4976

DOI: 10.1021/acsnano.7b01657 ACS Nano 2017, 11, 4976−4984

Article

www.acsnano.org

Article

ACS Nano

Figure 1. Single filament formation. (A) Schematic representation of the four phases of filament behavior; initial nucleation at C-AFM tip with a positive substrate bias (i), leading to filament growth (ii), which contacts the surface at time Δτf (iii). Under reverse, i.e., negative, substrate bias, dissolution occurs (iv). (B) An I−V characteristic of a single filament, starting at a substrate bias of −1 V ((1) OFF state) and sweeping to +1 V ((2) Formation), showing no measurable current until the filament is formed ((3) ON state). The filament remains stable under a small negative bias before resetting to a nonconductive state ((4) Dissolution). Curve acquired at 0.6 V s−1, with a maximum readable current of ±600 nA. (C) Current vs time trace showing the time difference (Δτf) between application of a formation voltage and the resulting increase in current corresponding to filament formation. The junction is stable under an applied positive bias to the substrate, which is stepped down in −200 mV increments to −1 V, at which the filament remains conductive until dissolution occurs.

involves the rearrangement of relatively large number of atoms even for short filaments, dissolution only necessitates the removal of a few atoms from an existing filament in order to return to an insulating state. Thus, multiple possible mechanisms may contribute to dissolution, and separating the effect of each mechanism, either through simulation or direct observation, is challenging. Conductive atomic-force microscopy (C-AFM) is a powerful technique in nanoscale surface characterization and manipulation,23 and it is well-suited for controlling nanoscale electrodeposition and electrodissolution.24 C-AFM probes with tip radii approaching 10 nm are commercially available, raising the possibility of replacing the fixed electrodes in conventional filament-based applications with a movable nmscale voltage source. This could be used, for example, in the arbitrary patterning of filaments in 2D arrays as well as reconfiguring of active filaments through addressable dissolution, similar to scanning-tunneling microscopy (STM) demonstrations of controlled electrodeposition in polymer films by Bard et al.25 Recently, controlled electrodeposition using a scanning probe microscope (SPM) has been used to effect the controlled electrodeposition of Ag nanoclusters on superionic conductors.16,17,26 Similarly, C-AFM-mediated nanoparticle growth by reduction of both Au-24 and Ag-containing27 electrolyte solutions has been demonstrated with fine control over surface feature sizes and spacing. In these experiments, both electrical conductivity and surface morphology were altered, and the formation of clusters on the surface was evident in both the conductive and topographical images.

that submicroscopic structures of the electrolyte and electrode materials drastically impact device speed, switching current, and repeatability.7 However, the size of the fixed electrodes typically used for device-level experiments are >1 μm,1 much larger than the resolution achieved by advanced semiconductor manufacturing techniques. The ensuing relatively broad electric field distributions result in decreased control over filament positioning and increased variability in filament formation kinetics.8 Nanoscale electrodeposition and electrodissolution are dependent on the microscale to nanoscale features of the materials involved, and characterizing the atomic-scale effects that influence filament kinetics in situ is challenging. For a single filament, the process of formation has been studied experimentally through imaging with transmission-electron microscopy (TEM)8−13 and scanning-electron microscopy (SEM),14 by contact methods such as atomic-force microscopy,1,15−17 and through atomistic simulations.9,18,19 Prior work has usually examined filament formation as a function of the switching voltage, based on a voltage ramp7,15,20 or voltage pulses,21,22 with the general result that filaments formed at different voltages/currents also form at different times. However, these studies have not addressed either the complete initial formation of filaments, which is typically substantially different than subsequent formations, due to an initialization process,6 or the kinetic behavior of many filaments in parallel. In contrast to filament formation, filament dissolution has been less explored.7,14 To comprehensively characterize conductive filament behavior, both formation and dissolution processes must be studied. Whereas initial formation of a filament 4977

DOI: 10.1021/acsnano.7b01657 ACS Nano 2017, 11, 4976−4984

Article

ACS Nano

Figure 2. Formation/dissolution time distributions for simple filaments (system I). (A) Formation (τf) and (B) dissolution time (τd) distributions for simple filaments in 0.5 wt % (40 nm thickness, green), 1 wt % (80 nm thickness, blue) and 2 wt % (130 nm thickness, red) films of PEO with formation bias Vsubstrate − Vtip = +0.6 V, and dissolution bias of −1.0 V. The solid lines are skewnormal on standard time (formation) and log time (dissolution) curve fits. (C) Formation (τf) and (D) dissolution (τd) time distributions for simple filaments in a 1 wt % (80 nm) PEO film with a formation bias of +0.6 V and a dissolution bias of −0.6 V. Insets to (B) and (D): The dissolution time distributions are clearly log-normal when plotted directly vs time.

A common theme in exploring conductive filament formation in ReRAM devices is the addition of nanostructures or nanoparticles within the electrolyte as a means of modifying the electric field distribution within the electrolyte/insulating material.28 Prior experiments have utilized nanostructures either attached to, or formed from, an electrode8,22,26 or made from either insulators20 or conductors.22 In a suitable geometry, nanostructures could conceivably increase junction density by forming multiple junctions with one set of electrodes, more effectively using the available volume without additional electronics. Metal nanoparticles added to the interior of the insulating film are particularly interesting, because, in addition to their inherent photonic and plasmonic properties, they can function as nanoscale bipolar electrodes,29 substantially altering the mechanism of filament formation or dissolution, its position in space, and the conditions required for filament control. Certainly, the presence of a nanoparticle bipolar electrode would produce a more complex voltage profile in the presence of the C-AFM than a particle-free insulating film. Here we are motivated by the possibility of using nanoscale filaments for the addressable, reconfigurable, direct-write in situ wiring of nanoparticle arrays as a route to the creation of nanoelectronic and nanophotonic materials and devices requiring nanometer scale control of the dielectric response function. To support this, we present detailed experimental and computational studies of three silver (Ag) filament formation/

dissolution systems involving poly(ethylene oxide) (PEO) thin films: (system I) Ag|PEO-Ag+, (system II) Ag|poly(ethylene glycol) monolayer-PEO-Ag+, and (system III) Ag|poly(ethylene glycol) monolayer-PEO-Ag + /Ag-nanoparticle (AgNP). The automated experimental approach presented here makes it possible to measure hundreds of formation/ dissolution events in a single experiment, thereby enabling quantitative statistical analysis of formation/dissolution events. The kinetic characterization method is applied as a function of experimental conditions, to establish the relative ordering of formation and dissolution times as a function of film architecture, and give insight into the underlying mechanisms. Furthermore, the C-AFM approach presented here also lays the foundation for engineering designed reconfigurable nanofilament arrays and patterns in polyelectrolyte films.

RESULTS Fabrication and Structural Characterization. The basic device structure consists of a ∼ 100 nm layer of PEO containing AgNO3 supported on a Ag thin film substrate, which is used as a sacrificial electrode and is electrically connected to the AFM stage. A C-AFM tip is used as the addressable, directwrite working electrode. Starting in the OFF state, when a positive bias is applied to the substrate, Ag+ ions are reduced at the negative AFM tip via Ag+ + e− → Ag(s), while oxidation occurs at the sacrificial Ag electrode via Ag(s) → Ag+ + e−, Figure 1A(i). Thus, the global concentration of Ag+ in the PEO 4978

DOI: 10.1021/acsnano.7b01657 ACS Nano 2017, 11, 4976−4984

Article

ACS Nano

Figure 3. PDT-functionalized Ag-coated substrate after spin-coating a 1 wt % PEO layer (system II). (A) Plan-view SEM image. (B) FIB crosssectional image of sample, showing glass substrate, Au, Ag, and PEO layers. A protective Pt layer was deposited on the PEO layer to preserve film integrity during the cross-sectioning process. (C) Formation (τf) and (D) dissolution (τd) time distributions.

film remains constant as the conductive filament is formed, Figure 1A(ii), although localized concentration gradients will appear due to the diffusion and electromigration of Ag+ ions through the electrolyte, according to the Nernst−Planck equation (see Supporting Information, SI). After a sufficient amount of material has been deposited, the filament contacts the Ag-coated substrate, completing a conductive path from the C-AFM tip to the substrate (ON state), Figure 1A(iii), causing an abrupt increase in the current measured through the tip. Subsequently, the filament can be dissolved by reversing the applied potential, which, after a delay, resets the filament to a nonconducting state, Figure 1A(iv). Simple Filament Formation and Resistance Measurement. Filaments remain stable for substantial time periods under both steady and increasing bias and even under small reverse bias. This can be seen in both a current−voltage sweep (Figure 1B) as well as a current vs time measurement (Figure 1C). Behavior resembling the “pinched” I−V curve characteristic of memristive devices confirms prior experiments with similar systems.2,12,13,20,21,26,30,31 If the formation bias remains applied substantially longer than the time needed for filament formation, τf, nanoclusters can form on the surface, (Figure S1, SI) which are evidence of filament overgrowth, as observed previously.16,24 In contrast, topographical AFM images of scan areas after formation with the bias removed at τf (or immediately following) show no evidence of overgrowth, even after the formation of multiple filaments (Figure S2, SI). The resistance characteristics of typical filaments were obtained by averaging the I−V characteristics of ca. 30 filaments in the linear region before the compliance current of the C-AFM was

reached. These measurements indicate an average resistance of 200−400 kΩ (Figure S3, SI), consistent with published results obtained from devices with similar compliance currents.10 Formation and Dissolution Kinetics of Simple Filaments. To examine the impact of film thickness and dissolution voltage on formation and dissolution, the C-AFM was used to form and dissolve 870 individual filaments on each sample, in a rectangular array with a pitch of 200 nm. PEO and salt concentration, ionic strength, formation and dissolution voltages, filament spacing, presence of a supporting electrolyte, etc. all might be expected to affect formation and dissolution kinetics. For this work, thickness, ion concentration, and dissolution voltage were selected for detailed study. Three samples were prepared with 0.5, 1, and 2 wt % PEO in acetonitrile/AgNO3 solutions, corresponding to ether oxygen:Ag+ ratios of 180, 90 and 45:1. Spin-coating the electrolyte layer produced films of ca. 40, 80, and 130 nm, respectively, (Figure S4, SI). These films were tested with a + 0.6 V substrate voltage for formation and −1.0 V for dissolution (all voltages specified as Vsubstrate − Vtip). Formation, τf, and dissolution, τd, time histograms are shown in Figures 2A and 2B, respectively. It is evident that while τf follows a normal distribution for all three samples, the τd distributions of dissolution times are log-normal (see inset in Figure 2B). This is clear evidence of inherent mechanistic differences between the two processes. Figure 2A shows a marked increase in τf with increasing film thickness, and Figure 2B shows a similar shift for τd distributions. While the ion concentration will also affect the formation/dissolution times, the ionic conductivity of PEO is relatively insensitive to AgNO3 4979

DOI: 10.1021/acsnano.7b01657 ACS Nano 2017, 11, 4976−4984

Article

ACS Nano

Figure 4. AgNP-functionalized sample after spin-coating 1 wt % PEO film (system III). (A) Plan-view SEM image showing in-plane distribution of nanoparticles. (B) Cross-sectional SEM image showing vertical placement of nanoparticles (white arrows overlaid) in the PEO film. (C) Formation (τf) and (D) dissolution (τd) time distributions together with multipeak fits (black line, overlay; individual peaks shown below). Peaks are numbered corresponding to formation condition: (1) shorted nanoparticle against tip or substrate, (2) formation between tip and substrate in area away from AgNP, and (3) formation between tip and substrate through AgNP acting as a bipolar electrode.

can act as a sacrificial, “wireless” bipolar electrode. During filament formation, equal masses of Ag must deposit on the lower, and dissolve from the upper side of the nanoparticle to maintain net electrical neutrality. However, compared to filament formation in the undecorated film (system II), the presence of the AgNPs can affect the filament formation kinetics by altering the local spatial distribution of the electric field, as well as the local Ag+ spatial distributions in the PEO film. An α,ω-dithiolpoly(ethylene glycol) (PDT) linker was used to secure the AgNPs above the Ag-coated substrate. After selfassembling PDT on the surface, AgNPs were attached to the PDT layer and 1 wt % PEO/Ag+ was applied by spin-coating to complete fabrication of the PEO thin film. A control sample (system II) containing the PDT-functionalized Ag-coated substrate but without AgNPs was fabricated similarly. Layer morphology (Figure 3A) and thickness (Figure 3B) were confirmed by SEM. The presence of the PDT layer was found to produce shorter formation and longer dissolution times (Figures 3C and 3D, respectively) compared to the 80 nm system I sample, suggesting that the PDT layer stabilizes the filament such that formation is favored, and dissolution hindered, relative to the sample without PDT (system I). Placement of the nanoparticles in system III was confirmed by plan view SEM and FIB cross-section images, Figures 4A

concentration in the concentration range studied here.32 Therefore, the shift to longer formation/dissolution times in Figure 2A and B can be explained by increasing the film thickness from 40 to 80 to 130 nm. To test the effect of dissolution voltage, the 80 nm sample was tested with +0.6 V formation bias followed by a−0.6 V dissolution bias (compared to −1.0 V dissolution bias in Figure 2B). The resulting histograms, Figures 2C and 2D, show a small, but statistically significant, decrease in τf but a markedly longer average dissolution time, approaching that of the 130 nm sample. The displacement to longer τd values at smaller dissolution voltage is consistent with a kinetically limited process in which the effective barrier height (for Ag atom displacement near the quantum point contact) is surmounted more easily at larger dissolution biases. The combination of different distributions for τf and τd and the thickness dependence of both processes demonstrates that formation and dissolution kinetics can be controlled independently. Formation and Dissolution Kinetics through Ag Nanoparticle Bipolar Electrodes. The presence of nanoparticles in the polymer electrolyte may enable different control mechanisms for formation and dissolution. By positioning AgNPs in the body of the electrolyte film, with polymer between the tip and the nanoparticle and between the nanoparticle and the Ag substrate (system III), the nanoparticle 4980

DOI: 10.1021/acsnano.7b01657 ACS Nano 2017, 11, 4976−4984

Article

ACS Nano

Figure 5. Schematic representations of the filament formation conditions based on the vertical and horizontal position of the nanoparticle in the PEO film for system III. Arrow color gradients represent qualitative changes in electric potential within the film (quantitative simulation results given in Supporting Information). (A) AgNP shorted against the AFM tip or substrate. (B) Filament formation in the absence of a AgNP. The filament grows directly from the tip to the substrate as in system II. (C) The presence of a proximal nanoparticle deflects filament formation away from the nanoparticle, increasing τf. (D) With the tip positioned over a AgNP, the formation of two conductive filaments is required to produce a conductive pathway.

Figure 6. Repeated formation and dissolution of single filaments (cyclic formation/dissolution). Plot of dissolution times, (τd in ln[ms]), of single filaments vs repetition number for (A) system I and (C) systems II and III. (B) and (D) show distributions of the data from panels (A) and (C), respectively, with Gaussian fit lines overlaid.

and 4B, respectively. The τf distribution (Figure 4C) differs markedly from those obtained from systems I (Figure 2A) and II (Figure 3C), with a pronounced increase in the number of filaments at sub-100 ms formation times and a broader distribution at longer formation times. Similarly, the τd distribution for system III (Figure 4D) also differs from the corresponding distributions for systems I (Figure 2B) and II (Figure 3D). Both τf and τd distributions for system III can be resolved into three components. The τf histogram in Figure 4C can be fit with an unresolvable short component and two normal distributions in the 0.1−2.5 s range. The shorter of these latter two matches the single distribution observed in system II. We hypothesize that the multicomponent τf and τd distributions observed for system III arise from nanoparticles in three

generic locations, as shown in Figure 5. In Figure 5A the nanoparticle is shorted to an electrode, either through direct contact with the metal substrate or incomplete coverage of the PEO layer above the nanoparticle providing direct contact with the AFM tip. These particles are likely responsible for the sub100 ms τf component, because only one short filament needs to be formed. In the second location, Figure 5B, the closest NP is sufficiently far from the AFM tip that filaments grow directly from the tip to the substrate, thus behaving like system II (Figure 4C component 2). The third location places the nanoparticle in the interior of the polymer electrolyte, close to (Figure 5C) or directly in line with (Figure 5D) the tip, such that filament formation requires the formation of two separate filaments (below and above the AgNP) to achieve a conductive 4981

DOI: 10.1021/acsnano.7b01657 ACS Nano 2017, 11, 4976−4984

Article

ACS Nano

concentration of free ions in the electrolyte, or individually by varying the bias. Ag nanoparticles embedded in the PEO electrolyte film act as bipolar electrodes and increase the formation time. In fact, the fastest initial formation occurs for Ag|PEO-Ag+ (system I), while the Ag nanoparticle-containing system III exhibits the slowest initial filament formation. Furthermore, the nanoparticles, which cover 14% of the surface as measured by SEM, affect more than 70% of the filament formation, indicating that nanoparticle effects extend beyond the immediate volume proximal to the nanoparticle. Finally, the C-AFM measurement system developed here can be easily reconfigured to test the effect of experimental conditions on formation and dissolution kinetics without requiring additional lithography or fabrication. Thus, the direct-write bipolar electrochemical deposition/dissolution strategy promises to enable a host of applications where the need to connect/ disconnect physically inaccessible components is crucial.

pathway, resulting in the slowest formation distribution reflected in component 3 in Figure 4C. At first, it might be expected that the fraction of the system III τf distribution attributed to nanoparticles would be proportional to the total cross-sectional area presented by the nanoparticles. However, threshold-value analysis of the planview SEM image of system III (Figure S5, SI) shows that ca. 13% of the surface is covered by nanoparticles, while the fraction of Figure 4C attributed to nanoparticles constitutes 74% of the total area. To explore this discrepancy, finite element modeling (described in SI) of the potential distribution in the PEO-electrolyte nanoparticle composite was performed, indicating that when the C-AFM tip is positioned directly over a nanoparticle, both the potential distribution and the concentration gradient extend beyond the edge of the nanoparticle. In simulations with the C-AFM tip horizontally offset from the nanoparticle, the potential distribution through the electrolyte is altered such that the highest field gradient, and therefore the direction of filament growth, is not between the tip and the proximal point on the substrate, (Figure S6, SI) but is deflected away from the nanoparticle, as illustrated in Figure 5C. The longer filament length increases the formation time. Thus, both simulation and experiment support a picture in which AgNPs, acting as bipolar electrodes, affect filament growth over an area ca. 6× greater area than that occluded geometrically by the nanoparticle. Repeatability of Filament Formation and Dissolution. For each film condition, experiments were conducted to assess the repeatability of formation and dissolution over 30 repetitive cycles at the same position, Figure 6. In these experiments, the tip was held stationary, and the voltage was switched from formation to dissolution conditions after the current reached compliance and vice versa when the current fell below the “off” threshold. In nearly all cases, after initial filament formation, subsequent τf values (not shown) were below the 1 ms measurement time resolution. The rapid filament reformation demonstrates that conversion to a nonconductive state does not require a significant reset of the filament conditions, only a small rearrangement, perhaps involving only a few atoms.21 The dissolution kinetics obtained upon repeated regrowth and dissolution of the same filament are similar to the dissolution of the initial filament, clearly following a log-normal distribution, Figures 6B and 6D. However, dissolution of repetitively regrown filaments produces shorter (by >1 log(ms)) τd values than the initially grown filaments, especially in the presence of nanoparticles (system III). This suggests that the structure of the filament likely evolves with repeated growth/dissolution cycles, annealing to produce structures that can be dissolved more easily.

METHODS Chemicals and Materials. Anhydrous acetonitrile (ACN), silver nitrate (AgNO3), poly(ethylene glycol) dithiol, sulfuric acid (95%) and hydrogen peroxide (30%) were obtained from Sigma-Aldrich. Silver nanoparticles (AgNPs, 50 nm size) were purchased from Ted Pella, Inc. Poly(ethylene oxide) (PEO) with a molecular weight of 94 600 Da was purchased from PSS Polymer Standards Service, Inc. All reagents were used as received without further purification. Fabrication. Glass slides were cleaned in piranha solution (3:1 sulfuric acid (95%):hydrogen peroxide (30%)Caution! Strong oxidizer, use with extreme care), rinsed with deionized (DI) water, and dried at 110 °C. A 100 nm thick Au layer was deposited by electron-beam evaporation (UNIVEX 450B, Oerlikon) after deposition of a 10 nm Ti adhesion layer. Then, a 100 nm Ag layer was deposited on the same glass slide. In an argon-filled glovebox with oxygen and water concentrations controlled to less than 0.1 ppm, 10 mM and 20 mM solutions of AgNO3 in anhydrous ACN were prepared; similarly, PEO was dissolved in ACN to make 0.5, 1, and 2 wt % PEO solutions. The AgNO3 solution were added to the PEO solutions in a 1:9 volume ratio, yielding final solutions of AgNO3 in ACN with concentrations of 1 mM, 2 mM, and 2 mM for PEO wt % of 0.5, 1, and 2 wt %, respectively (ether oxygen:Ag+ ratios of 90:1, 90:1, and 45:1). Each solution was then spin-coated at 4000 rpm for 30 s onto the Ag-coated glass slides inside the glovebox. Film thickness was confirmed by cross-sectional SEM imaging (see SI in Figure S4); due to exposure to atmosphere, these samples were not returned to the glovebox for electrical analysis. In addition to sample preparation, electrical characterization via conductive AFM was carried out inside the glovebox. C-AFM Measurements. A Bruker Dimension Icon AFM was used in contact mode. Before each run, a force vs distance calibration curve was taken to ensure that the engagement set point was not distorted by the reflectivity of the surface. A custom script was written to (1) move the tip from point-to-point in a raster scan pattern at a preset point spacing; (2) apply the desired voltage; and (3) measure the current between the conductive AFM tip and the AFM chuck. The AFM tip is grounded and the voltage is applied to the chuck. PFTUNA tips consisting of a Pt−Ir coating on a Si cantilever with spring constant of 0.4 N/m, were used for all measurements. At each point, the chuck voltage was set to a formation voltage until the current increased above a set threshold value, after which the voltage was switched to the dissolution voltage until the current (now of the opposite sign) decreased in magnitude below a set threshold. Data were recorded both as a current vs time trace and as τf, τd pairs for each point. These experiments were implemented with a 200 nm pitch, a formation voltage of +0.6 V, dissolution voltage of −1 V (limited by the compliance current of the instrument) and a rest voltage of 0 V relative to ground. Formation and dissolution thresholds were chosen as +450 nA and −25 nA respectively. The instrument exhibits a compliance current of ca. 600 nA. The accuracy of the script was

CONCLUSION In conclusion, repeatable formation and dissolution of conductive filaments can be achieved through an addressable, direct-write, bipolar electrochemical deposition/dissolution strategy, presenting a path to reconfigurable noncontact in situ wiring of nanoparticle arrays. Such a capability would enable active control of the connectivity of nanoparticle arrays, which in turn could be used to regulate the behavior of nanoelectronic and nanophotonic materials. The kinetics of the C-AFM direct write process give insight into the underlying mechanisms of filament formation and dissolution that suggest straightforward control strategies. For example, formation and dissolution times can be altered simultaneously by changing the 4982

DOI: 10.1021/acsnano.7b01657 ACS Nano 2017, 11, 4976−4984

Article

ACS Nano

via Tip-Enhanced Electric Fields. Adv. Electron. Mater. 2016, 2, 1600233. (9) Qin, S.; Liu, Z.; Zhang, G.; Zhang, J.; Sun, Y.; Wu, H.; Qian, H.; Yu, Z. Atomistic Study of Dynamics for Metallic Filament Growth in Conductive-Bridge Random Access Memory. Phys. Chem. Chem. Phys. 2015, 17, 8627−8632. (10) Celano, U.; Goux, L.; Belmonte, A.; Opsomer, K.; Franquet, A.; Schulze, A.; Detavernier, C.; Richard, O.; Bender, H.; Jurczak, M. Three-Dimensional Observation of the Conductive Filament in Nanoscaled Resistive Memory Devices. Nano Lett. 2014, 14, 2401− 2406. (11) Krishnan, K.; Aono, M.; Tsuruoka, T. Kinetic Factors Determining Conducting Filament Formation in Solid Polymer Electrolyte Based Planar Devices. Nanoscale 2016, 8, 13976−13984. (12) Kwon, D.-H.; Kim, K. M.; Jang, J. H.; Jeon, J. M.; Lee, M. H.; Kim, G. H.; Li, X.-S.; Park, G.-S.; Lee, B.; Han, S. Atomic Structure of Conducting Nanofilaments in TiO2 Resistive Switching Memory. Nat. Nanotechnol. 2010, 5, 148−153. (13) Arita, M.; Takahashi, A.; Ohno, Y.; Nakane, A.; TsurumakiFukuchi, A.; Takahashi, Y. Switching Operation and Degradation of Resistive Random Access Memory Composed of Tungsten Oxide and Copper Investigated Using in-situ TEM. Sci. Rep. 2015, 5, 17103. (14) Guo, X.; Schindler, C.; Menzel, S.; Waser, R. Understanding the Switching-Off Mechanism in Ag+ Migration Based Resistively Switching Model Systems. Appl. Phys. Lett. 2007, 91, 133513. (15) Nardi, F.; Deleruyelle, D.; Spiga, S.; Muller, C.; Bouteille, B.; Ielmini, D. Switching of Nanosized Filaments in NiO by Conductive Atomic Force Microscopy. J. Appl. Phys. 2012, 112, 064310. (16) Lee, M.; O’Hayre, R.; Prinz, F. B.; Gür, T. M. Electrochemical Nanopatterning of Ag on Solid-State Ionic Conductor RbAg4I5 Using Atomic Force Microscopy. Appl. Phys. Lett. 2004, 85, 3552−3554. (17) Valov, I.; Sapezanskaia, I.; Nayak, A.; Tsuruoka, T.; Bredow, T.; Hasegawa, T.; Staikov, G.; Aono, M.; Waser, R. Atomically Controlled Electrochemical Nucleation at Superionic Solid Electrolyte Surfaces. Nat. Mater. 2012, 11, 530−535. (18) Bersuker, G.; Gilmer, D.; Veksler, D.; Kirsch, P.; Vandelli, L.; Padovani, A.; Larcher, L.; McKenna, K.; Shluger, A.; Iglesias, V. Metal Oxide Resistive Memory Switching Mechanism Based on Conductive Filament Properties. J. Appl. Phys. 2011, 110, 124518. (19) Menzel, S.; Kaupmann, P.; Waser, R. Understanding Filamentary Growth in Electrochemical Metallization Memory Cells Using Kinetic Monte Carlo Simulations. Nanoscale 2015, 7, 12673− 12681. (20) You, B. K.; Park, W. I.; Kim, J. M.; Park, K.-I.; Seo, H. K.; Lee, J. Y.; Jung, Y. S.; Lee, K. J. Reliable Control of Filament Formation in Resistive Memories by Self-Assembled Nanoinsulators Derived from a Block Copolymer. ACS Nano 2014, 8, 9492−9502. (21) Choi, B. J.; Torrezan, A. C.; Norris, K. J.; Miao, F.; Strachan, J. P.; Zhang, M.-X.; Ohlberg, D. A.; Kobayashi, N. P.; Yang, J. J.; Williams, R. S. Electrical Performance and Scalability of Pt Dispersed SiO2 Nanometallic Resistance Switch. Nano Lett. 2013, 13, 3213− 3217. (22) Hu, Q.; Park, M.; Abbas, Y.; Kim, J. S.; Yoon, T.-S.; Choi, Y. J.; Kang, C. J. Resistive Switching Properties of Manganese Oxide Nanoparticles with Hexagonal Shape. Semicond. Sci. Technol. 2015, 30, 015017. (23) Xie, X. N.; Chung, H.; Sow, C.; Wee, A. Nanoscale Materials Patterning and Engineering by Atomic Force Microscopy Nanolithography. Mater. Sci. Eng., R 2006, 54, 1−48. (24) Yanagi, H.; Ohno, T. Nanofabrication of Gold Particles in Glass Films by AFM-Assisted Local Reduction. Langmuir 1999, 15, 4773− 4776. (25) Craston, D. H.; Lin, C. W.; Bard, A. J. High Resolution Deposition of Silver in Nafion Films with the Scanning Tunneling Microscope. J. Electrochem. Soc. 1988, 135, 785−786. (26) Liu, Q.; Long, S.; Lv, H.; Wang, W.; Niu, J.; Huo, Z.; Chen, J.; Liu, M. Controllable Growth of Nanoscale Conductive Filaments in Solid-Electrolyte-Based ReRAM by Using a Metal Nanocrystal Covered Bottom Electrode. ACS Nano 2010, 4, 6162−6168.

verified with an external oscilloscope, triggered by the voltage applied to the tip in single-run edge mode, and shown to exhibit a time resolution of ∼1.5 ms.

ASSOCIATED CONTENT S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acsnano.7b01657. AFM images before/after deposition for nanoclusters and filaments, filament resistance data, AgNP image analysis, thickness characterization of PEO/AgNO3 thin films, and detailed description of finite-element simulations of electrochemical behavior (PDF)

AUTHOR INFORMATION Corresponding Authors

*E-mail: [email protected]. Tel.: +1 574 631 8394. Fax: +1 574 631 8341. *E-mail: [email protected]. Tel.: +1 574 631 1849. Fax: +1 574 631 8366. ORCID

Garrison M. Crouch: 0000-0002-0056-5269 Donghoon Han: 0000-0003-1870-3006 Susan K. Fullerton-Shirey: 0000-0003-2720-0400 Paul W. Bohn: 0000-0001-9052-0349 Author Contributions #

G.M.C. and D.H. contributed equally to the paper.

Notes

The authors declare no competing financial interest.

ACKNOWLEDGMENTS The authors wish to thank Dr. Paul Rumbach, Erich Kinder, and Prof. Alan Seabaugh for assistance with the gloveboxconfined AFM. The authors also wish to thank Zhongmou Chao and Kaiyu Fu for helpful discussions. This work was supported by a NASA Space Technology Research Fellowship NNX16AM45H (GMC) and the Defense Advanced Research Projects Agency FA8650-15-C-7546 (DH). REFERENCES (1) Yoon, J.; Lee, J.; Choi, H.; Park, J.-B.; Seong, D.-j.; Lee, W.; Cho, C.; Kim, S.; Hwang, H. Analysis of Copper Ion Filaments and Retention of Dual-Layered Devices for Resistance Random Access Memory Applications. Microelectron. Eng. 2009, 86, 1929−1932. (2) Valov, I.; Waser, R.; Jameson, J. R.; Kozicki, M. N. Electrochemical Metallization Memories-Fundamentals, Applications, Prospects. Nanotechnology 2011, 22, 254003. (3) Patolsky, F.; Zheng, G.; Lieber, C. M. Nanowire-Based Biosensors. Anal. Chem. 2006, 78, 4260−4269. (4) Hwang, T.-W.; Branagan, S. P.; Bohn, P. W. Chemical Noise Produced by Equilibrium Adsorption/Desorption of Surface Pyridine at Au−Ag−Au Bimetallic Atom-Scale Junctions Studied by Fluctuation Spectroscopy. J. Am. Chem. Soc. 2013, 135, 4522−4528. (5) Dong, S.; Zhang, K.; Yu, Z.; Fan, J. A. Electrochemically Programmable Plasmonic Antennas. ACS Nano 2016, 10, 6716−6724. (6) Hasegawa, T.; Terabe, K.; Tsuruoka, T.; Aono, M. Atomic Switch: Atom/Ion Movement Controlled Devices for Beyond VonNeumann Computers. Adv. Mater. 2012, 24, 252−267. (7) Midya, R.; Wang, Z.; Zhang, J.; Savel’ev, S. E.; Li, C.; Rao, M.; Jang, M. H.; Joshi, S.; Jiang, H.; Lin, P. Anatomy of Ag/Hafnia-Based Selectors with 1010 Nonlinearity. Adv. Mater. 2017, 29, 1604457. (8) Shin, K. Y.; Kim, Y.; Antolinez, F. V.; Ha, J. S.; Lee, S. S.; Park, J. H. Controllable Formation of Nanofilaments in Resistive Memories 4983

DOI: 10.1021/acsnano.7b01657 ACS Nano 2017, 11, 4976−4984

Article

ACS Nano (27) Hubert, C.; Amrani, H.; Ali Khan, M.; Vocanson, F.; Destouches, N. Electrical Growth of Metallic Nanoparticles in Mesoporous Silica Films Using Atomic Force Microscopy. Appl. Phys. Lett. 2012, 100, 241605. (28) Yoon, J. H.; Han, J. H.; Jung, J. S.; Jeon, W.; Kim, G. H.; Song, S. J.; Seok, J. Y.; Yoon, K. J.; Lee, M. H.; Hwang, C. S. Highly Improved Uniformity in the Resistive Switching Parameters of TiO2 Thin Films by Inserting Ru Nanodots. Adv. Mater. 2013, 25, 1987− 1992. (29) Fosdick, S. E.; Crooks, R. M. Bipolar Electrodes for Rapid Screening of Electrocatalysts. J. Am. Chem. Soc. 2011, 134, 863−866. (30) You, B. K.; Kim, J. M.; Joe, D. J.; Yang, K.; Shin, Y.; Jung, Y. S.; Lee, K. J. Reliable Memristive Switching Memory Devices Enabled by Densely Packed Silver Nanocone Arrays as Electric-Field Concentrators. ACS Nano 2016, 10, 9478−9488. (31) Gao, S.; Zeng, F.; Chen, C.; Tang, G.; Lin, Y.; Zheng, Z.; Song, C.; Pan, F. Conductance Quantization in a Ag Filament-Based Polymer Resistive Memory. Nanotechnology 2013, 24, 335201. (32) Verma, M. L.; Sahu, H. D. Ionic Conductivity and Dielectric Behavior of Peo-Based Silver Ion Conducting Nanocomposite Polymer Electrolytes. Ionics 2015, 21, 3223−3231.

4984

DOI: 10.1021/acsnano.7b01657 ACS Nano 2017, 11, 4976−4984