Lithography-Free Miniaturization of Resistive Nonvolatile Memory

Jan 18, 2017 - The scaling of nonvolatile memory (NVM) devices based on resistive filament switching to below a 100 nm2 footprint area without employi...
0 downloads 0 Views 683KB Size
Subscriber access provided by HACETTEPE UNIVERSITESI KUTUPHANESI

Communication

Lithography-free miniaturization of resistive non-volatile memory devices to the 100 nm scale by glancing angle deposition Giovanni Ligorio, Marco Vittorio Nardi, and Norbert Koch Nano Lett., Just Accepted Manuscript • DOI: 10.1021/acs.nanolett.6b04794 • Publication Date (Web): 18 Jan 2017 Downloaded from http://pubs.acs.org on January 22, 2017

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Nano Letters is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 12

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Nano Letters

Lithography-free miniaturization of resistive non-volatile memory devices to the 100 nm scale by glancing angle deposition Giovanni Ligorio1, Marco Vittorio Nardi1, and Norbert Koch1, 2† 1) Humboldt-Universität zu Berlin, Institut für Physik & IRIS Adlershof, Brook-Taylor Str. 6, 12489 Berlin (Germany) 2) Helmholtz-Zentrum Berlin für Materialien und Energie GmbH, Albert-Einstein Str. 15, 12489 Berlin, (Germany)

The scaling of non-volatile memory (NVM) devices based on resistive filament switching to below 100 nm2 footprint area without employing cumbersome lithography is demonstrated. Nanocolumns of the organic semiconductor 4,4bis[N-(1-naphthyl)-N-phenyl-amino]diphenyl (α-NPD) were grown by glancing angle deposition on a silver electrode. Individual NVM devices were electrically characterized by conductive atomic force microscopy with the tip of a conductive cantilever serving as second electrode. The resistive switching mechanism is unambiguously attributed to Ag filament formation between the electrodes. This sets the upper limit for filament diameter to well below 100 nm. Full functionality of these NVM nano-devices is evidenced, revealing a potential memory density of > 1 GB/cm2 in appropriate architectures. Keywords: non-volatile memory, nanostructure growth, organic semiconductor

The increasing demand for logic memory technology with improved high-density, high-speed and low-power consumption is pushing research toward new materials and architectures. In particular, novel non-volatile memory (NVM) devices are currently investigated to overcome the limitations of traditional memory technology.1,2 For instance, the dynamic random access memory (DRAM), commonly used in central processing units (CPUs), is very fast but has the drawback of high-energy consumption for refreshing the logic elements, as well as high production costs.1 On the other hand, the flash memories, extensively employed as mobile data storage, suffer from low operation speed, and low endurance.3 Among the candidates competing for new-generation memories, resistive switching NVM (R-NVM) devices are emerging and are at the edge of commercialization. An R-NVM is a simple two-terminal device formed by two electrodes separated by an insulating or semiconducting material [see Fig. 1(a)]. We focus here on bistable devices (not discussing “write once read many” or WORM devices)4 in which the resistance of the device can be changed reversibly between a high resistance state (HRS) and a low resistance state (LRS). The particular resistance state can be read non-destructively and no power consumption is needed to maintain the resistive state (non-volatility). For the realization of R-NVMs, organic materials as well as metal oxides have †

corresponding author email: [email protected]

ACS Paragon Plus Environment

1

Nano Letters

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 12

been employed as layer between electrodes.5–8 Different models have been suggested to explain the switching mechanism in R-NVMs. While some of these are specific to the nature of the materials used for device fabrication (e.g. oxygen vacancy in metal oxides based devices)9,10 the formation/rupture of conductive localized filaments is reported as playing a role in both inorganic- and organic-based R-NVM devices.11–14 When a metal M with appropriate redox potentials is used as electrode (e.g. Al, Cu, Ag),15 oxidation at the electrode interface occurs upon biasing (M → Mz+ + ze-) and metal ions (Mz+) migrate through the interlayer material [see Fig. 1(b)] towards the opposite electrode. There, ion reduction takes place with formation of conductive metallic structures, which can eventually bridge the electrodes and in doing so abruptly decrease the device resistance [see Fig. 1(c)]. Dissolution of the metallic filament via opposite bias or Joule dissipation leads to filament rupture and sets the device back to the former HRS. As layer separating the electrodes, organic materials, such as small molecules or polymers, provide many advantages over inorganic ones, particularly because of easy implementation on flexible substrates, printability, (close to) room temperature processing, and generally low fabrication costs. However, the need of high-density memory modules requires the miniaturization of organic-based R-NVM devices and verification that filament-switching is still possible in two-terminal devices with diameter on the order of 100 nm. A method to achieve laterally separated nanoscale organic molecular islands (columns) without cumbersome lithography techniques is glancing angle deposition (GLAD).16,17 Directing the molecular beam under an oblique incident-angle onto a substrate, which is in continuous rotation, results in arrays of separated columnar nanostructures due to shadowing effects. Without substrate rotation, the columns point towards the vapor source. Substrate rotation allows sculpturing the nanostructure morphology, and in particular enables orienting the columns perpendicularly to the substrate surface.18,19 Here, we used GLAD to fabricate arrays of nanocolumns with ca. 100 nm diameter of the organic

hole-transport

material

N,N′-Di(1-naphthyl)-N,N′-diphenyl-(1,1′-biphenyl)-4,4′-

diamine (α-NPD) on a silver electrode. With the conductive tip of a scanning force microscope (SFM) we formed the top contact for electrical characterization. We demonstrate the full functionality of these nanostructures as R-NVM elements and evidence that switching is indeed due to filament formation/rupture, as in similar macroscale devices20. Indium tin oxide (ITO) coated glass (1 cm2, Thin Film Device Inc) was used as substrate. Such coupons were cleaned by sequential ultra-sonication in acetone and isopropyl alcohol and by further exposure to ultraviolet-ozone treatment for 30 minutes. The ITO substrates were ACS Paragon Plus Environment

2

Page 3 of 12

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Nano Letters

mounted in a custom-made vacuum chamber (base pressure 1 × 10-8 mbar) on a sample holder attached to a computer-controlled stepper motor. The latter allowed for sample rotation during deposition. One corner of the ITO substrate was masked to allow for subsequent electrical contacting to perform conductive tip SFM. For Ag-coated ITO (necessary to achieve resistive switching), 20 nm of silver (MaTeck) was evaporated onto the ITO substrates. α-NPD (Aldrich) nanocolumns were fabricated by GLAD on both pristine ITO (serving as control without switching behavior) and Ag-coated ITO. α-NPD was sublimed from an effusion cell and the deposited mass thickness monitored with a quartz crystal microbalance. The nominal and surface-normal deposition rate was 1.0 Å/s. The substrate orientation was adjusted with respect to the incoming molecular flux by a precision rotary motion feedthrough (± 1.5°) and set to an incident angle θ of 80°. The sample was continuously rotated during molecular deposition by the computer-controlled stepper motor with a rotation speed  = 0.3 rpm (revolutions per minute). For the sake of comparison, an α-NPD film was grown at normal incidence angle (θ = 0°) onto an ITO substrate; this evaporation geometry corresponds to conventional physical vapor deposition. The schematic illustration of both evaporation geometries is shown in Fig. 2. The morphology of the samples was investigated with an SFM (JPK Nanowizard II) and scanning electron microscopy (SEM). Conductive-tip SFM (C-SFM; JPK conductive AFM module) was employed for the electrical characterization of both the reference (growth on Ag-free ITO) and resistivity switching nano-devices (growth on Ag-coated ITO). During the CSFM experiments, a DC bias (maximum range ± 11 V, resolution 0.1 V) was applied to the bottom electrode while the conductive tip was held at ground. The electrical connection with the bottom contact was made on the corner of the ITO/glass substrate that was masked during the molecular deposition (see above). To address one specific nanocolumn (corresponding to one nano-device), the lateral (or xy-) position of the cantilever was adjusted according to the topography image provided by the SFM scan in tapping mode. The Au-coated conductive tip (Bruker) was brought into contact with the nano-device to be characterized; i.e., the tip acted as top electrode. The vertical (or z-) position of the tip during the electrical measurement was maintained in position via force scan mode, in which a feedback system controls the force applied on the tip against the sample surface, according to a chosen set point. The current between the conductive tip and the sample was measured. Because of the reduced dimensions of the tip, the current was limited (compliance) to avoid damage to the conductive coating. The maximum current range of the instrument was ± 100 ACS Paragon Plus Environment

3

Nano Letters

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 12

nA (resolution 3 pA). SFM and C-SFM measurements were performed in air and at room temperature. In Figure 2 [(a) and (e), respectively] we compare the morphology of α-NPD deposited on ITO in normal and oblique angle deposition. Normal angle deposition (θ = 0°) shows a continuous and featureless film on the ITO substrate [see Fig. 1(b)–(d)], typical for amorphous films formed by α-NPD. In contrast, cylindrically shaped columns are randomly distributed on the ITO substrate when setting θ to 80° (GLAD). The nanocolumnar growth is a consequence of the evaporation under the condition of a glancing angle and is amplified by Volmer-Weber or Stranski–Krastanov growth mode.17 The arrival of the α-NPD vapour almost parallel to the substrate is a random ballistic process. Surface outdents represent nucleation points in a random ballistic process. As result of the shadowing effects, some nucleation point will shadow (and therefore suppress) the neighbours. The nuclei, as result of the further evaporation of molecular vapour, will rise into columns normally oriented to the surface. Analysis of the SFM data [with JPK data processing software see, Fig. 1(f)] yields an average diameter of the α-NPD nanocolumns of (98 ± 12) nm. The area-density of columns is (95 ± 15) columns/ µm2.

Similar

topography

values

have

been

reported

for

tris(8-

hydroxyquinolinato)aluminium (Alq3), which analogously forms amorphous nanocolumns on ITO when deposited via GLAD.18 The height scale in the SFM image [1(f)] suggests an average height for the α-NPD nanocolumns (~ 90 nm), substantially smaller than the height measured by the cross-section SEM image [~ 200 nm, see (h)]. This due to the conical shape of the SFM tip, making it unable to access the bottom between neighboring nanocolumns. The crosssection SEM image provides evidence that growth of α-NPD via GLAD produces laterally isolated nanocolumns. In the following, every nanocolumn will be considered as a single nanoscale device for electrical characterization. See Supporting Information for more details on the nanocolumnar growth of films via GLAD.

The schematic in Fig. 3(a) illustrates the C-SFM experimental set-up during the electrical characterization of the reference nano-device (α-NPD nanocolumns grown on pristine ITO). The material choice for the electrodes (Au and ITO with work function values of approximately 4.7 eV and 4.8 eV, respectively)21,22 enables hole injection from both electrodes at moderate voltages (c.f. the schematic energy level diagram in Fig. 3(b), but electron injection is not possible due to the high barrier.

ACS Paragon Plus Environment

4

Page 5 of 12

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Nano Letters

Fig. 3(c) shows representative electrical characteristics for a pristine reference nano-device. The applied bias (U), the measured current (I) are plotted as a function of time (t). For t < 150 s, while the bottom ITO electrode is increasingly stepwise negatively biased (reverse bias), no current above noise-level is measured (see I-t plot). For t > 150 s, the ITO electrode is positively biased (forward bias) with U increasing stepwise (steps of 1 V). Now the current (a few 10 pA) increases above noise for U > 5 V. The I-t plot shows that the current is not constant over time, i.e., even if the same bias is applied for tens of seconds, the current is subject to variation, most likely due to fluctuations of the contact geometry at room temperature in air. This is even more evident in Fig. 3(d), where a scan performed entirely in forward bias is reported. The asymmetry between reverse and forward bias is a consequence of the higher hole-injection barrier of the α-NPD/Au contact compared to that of α-NPD/ITO. It is worth notice that, the reference nano-device, even when exposed for extended time (up to 600 s, not shown) to the highest possible bias (i.e., 11 V), did not undergo any noticeable change in resistance, i.e. the nano-device remained in its HRS. To observe a resistance state change, and in particular the switch to a LRS, it is necessary to use an electrode material that allows filament formation, i.e., to employ an electrochemically active metal for one of the electrodes3,13,14, Ag in our case. Fig. 4(a) displays the structure of the nano-devices based on α-NPD nanocolumns grown on Ag-coated ITO during C-SFM measurements. Statistical analysis performed with the SFM images confirms the same size and area-density of the nanocolumns grown on Ag and those on ITO [see Fig. 4(b)]. Fig. 4(c) displays the applied bias and the measured current as a function of time (bottommost and topmost curve, respectively). In reverse bias conditions no current above noise level is measured. At forward bias (t > 120 s), the current increases with sharp peaks during the stepwise increase of the bias (1 V step). In particular, a remarkable change of the current magnitude is observed at 6 V (t ~ 270 s), with a plateau at 100 nA, corresponding to the instrumental compliance. This point at t ~ 270 s corresponds to the device switching from HRS to LRS. Subsequently lowering the bias to 0.5 V (representing the read-out bias, see t ~ 300 s), the device still displays a current limited by the compliance, indicating that it is still in the LRS state. Coherently, the device set into the LRS shows at small negative bias (- 0.5 V) a negative current (again limited by the compliance at -100 nA), see Fig. 3(d) at t ~ 310 s. In all cases studied, the switch of a pristine nano-device from HRS to LRS could only be triggered when Ag was positively biased, i.e., indicating the migration Ag cations.

ACS Paragon Plus Environment

5

Nano Letters

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 12

To better understand the process that supposedly leads to filament formation, a constant bias (5 V) was employed to switch a pristine device (never electrically addressed before) from HRS to LRS. The current was measured as a function of time, and the curve is plotted for different time scales (60 s, and 180 s) in Fig. 5. Note that in Fig. 5(a) the current scale is pA, while for Fig. 5(b) the scale is in nA. The conductivity changes over time, as the current varies over a few orders of magnitude. Noteworthy, a sudden jump of current occurs at t = 172 s. This abrupt change in conductivity indicates that a conductive path bridging the electrodes was formed, setting the device from HRS to LRS. Silver has often been reported as an active material able to form filaments in both organicand inorganic-based non-volatile memories.12,23–26 Filament formation arises from oxidation of the Ag atoms that leave the positively biased electrode, Ag → Ag+ + e–. The Ag cations drift within the dielectric (here α-NPD) upon the electric field applied between the electrodes. The neutralization of the cations occurs due to electron transfer at the opposite electrode (Ag+ + e– → Ag). The accumulation of metallic clusters forms a conductive path, effectively bridging the electrodes. To study the dynamics of filament formation related to transport of cations in the nanodevices, I-t curve measurements under different constant bias were carried out. Constant bias (4 V, 5 V, 6 V, 7 V, 8 V and 9 V) was applied, in order to switch pristine nano-devices from HRS to LRS. The current between the tip and the ground was measured and is displayed as a function of time in Fig. 5(c). The wait time for the switching event (τS) is strongly dependent on the applied bias, since it decreased almost one order of magnitude when the bias increased by 1 V. Fig. 5(d) shows the plot (in semi logarithmic scale) of the switching time τS vs. the constant applied bias. For each applied bias the switching time of 5 pristine devices were measured. The experimental data are in excellent agreement with an exponential decay27,28  =  exp(− /  ), where τ0 and V0 are the fit parameters. Such a dependence of the wait time on applied bias is expected due to the stochastic nature of switching following a Poisson distribution.27 This suggests, therefore, that the transport of the Ag cations is the limiting factor in the dynamics of filament formation. Deviations of some of the data from the exponential decay might be due to the slight differences in the height of the nanocolumns. In R-NVMs, once the filament is formed with consequent bridging of the electrodes, the device is set from the HRS to the LRS. An inverse filament-rupture process is needed, in order to reset the device in the HRS. Usually, filament rupture is achieved by applying a voltage higher than the threshold needed for the formation of the filament.3,29 Joule dissipation, due to the ACS Paragon Plus Environment

6

Page 7 of 12

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Nano Letters

increased current, leads to thinning and degradation of the filaments. Because of the intrinsic compliance of our setup (i.e., 100 nA), it was not possible to apply higher current through the Ag filaments. Nonetheless, spontaneous switching from HRS to LRS was observed when bias (> 8 V) was applied between the electrodes of the nano-devices for extended time (> 1000 s). The switching mechanisms of the here reported nano-devices, interpreted as current-driven dissolution of filaments, is supported by comparison with the electrical behavior of macroscopic NVM devices based on α-NPD reported in literature.13,20,30 As a matter of fact, the α-NPD based devices where proven to be able to be reversible switched between HRS and LRS upon current-bias programing cycles. In particular, the switching mechanism has been unequivocally identified as the formation/rupture of conductive localized filaments.

In conclusion, the potential ultimate miniaturization of resistive non-volatile memory devices based on filamentary switching was explored. Devices with ca. 100 nm2 footprint area were fabricated by employing GLAD. Morphological investigations were provided by AFM and SEM; electrical characterizations were performed with (C-AFM). In full analogy to macro- and micro-scopic devices it is found that resistive switching is possible only if a material allowing for electrochemical ion mass transport is present in the device. Organic semiconductor nanocolumns fabricated on a silver electrode allowed electrical bistability to be observed, in contrast to the behavior of nanocolumns fabricated on ITO. In line with the switch mechanism based on conductive filaments, it is found that the filament formation process is dynamically limited by Ag ion transport upon the application of an external electric field. Our nanoscale NVM devices exhibited full functionality as compared to large scale devices reported earlier, enabling potentially a memory density of 1.25 GB/cm2 with rather low-cost and large-area fabrication techniques. Certainly, the memory density will be lower in actual applications, as wiring and the use of rectifying diodes in cross-bar array geometry will consume further space.

Acknowledgments: The authors thank C. Klimm (Helmholtz-Zentrum Berlin, Institut für Silizium Photovoltaik) and Dr. P. Schaefer (Humboldt-Universität zu Berlin) for help with SEM measurements. Prof. Dr. J. P. Rabe (Humboldt-Universität zu Berlin) is acknowledged for granting access to the AFM. This work was supported by the European Commission FP7 Project HYMEC (Grant No. 263073).

ACS Paragon Plus Environment

7

Nano Letters

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 12

Supporting Information Available: Additional information on the fabrication of nanocolumns via GLAD, and on the time dependent switching of pristine nano-devices as a function of the applied bias. This material is available free of charge via the Internet at http://pubs.acs.org

References: (1)

Scott, J. C. Science 2004, 304 (5667), 62–63.

(2)

Meijer, G. I. Science 2008, 319 (5870), 1625–1626.

(3)

Waser, R.; Dittmann, R.; Staikov, G.; Szot, K. Adv. Mater. 2009, 21 (25–26), 2632–2663.

(4)

Möller, S.; Perlov, C.; Jackson, W.; Taussig, C.; Forrest, S. R. Nature 2003, 426 (6963), 166–169.

(5)

Yang, Y.; Ouyang, J.; Ma, L. P.; Tseng, R. J.-H. J.-H.; Chu, C.-W. Adv. Funct. Mater. 2006, 16 (8), 1001–1014.

(6)

Scott, J. C. C.; Bozano, L. D. Adv. Mater. 2007, 19 (11), 1452–1463.

(7)

Ling, Q.-D.; Liaw, D.-J.; Zhu, C.; Chan, D. S.-H.; Kang, E.-T.; Neoh, K.-G. Prog. Polym. Sci. 2008, 33 (10), 917–978.

(8)

Cho, B. B.; Kim, T.; Song, S.; Ji, Y.; Jo, M.; Hwang, H.; Jung, G.-Y.; Lee, T. Adv. Mater. 2010, 22 (11), 1228–1232.

(9)

Yang, J. J.; Pickett, M. D.; Li, X.; Ohlberg, D. a a; Stewart, D. R.; Williams, R. S. Nat. Nanotechnol. 2008, 3 (7), 429–433.

(10) Mundle, R.; Carvajal, C.; Pradhan, A. K. Langmuir 2016, 32 (19), 4983–4995. (11) Linn, E.; Rosezin, R.; Kügeler, C.; Waser, R. Nat. Mater. 2010, 9 (5), 403–406. (12) Yang, Y.; Gao, P.; Gaba, S.; Chang, T.; Pan, X.; Lu, W. Nat. Commun. 2012, 3, 732. (13) Nau, S.; Sax, S.; List-Kratochvil, E. J. W. Adv. Mater. 2014, 26 (16), 2508–2513. (14) Cho, B.; Yun, J.-M.; Song, S.; Ji, Y.; Kim, D.-Y.; Lee, T. Adv. Funct. Mater. 2011, 21 (20), 3976–3981. (15) Kügeler, C.; Rosezin, R.; Linn, E.; Bruchhaus, R.; Waser, R. Appl. Phys. A 2011, 102 (4), 791–809. (16) Brett, M. J.; Hawkeye, M. M. Science 2008, 319 (5867), 1192–1193. (17) Hawkeye, M. M.; Brett, M. J. J. Vac. Sci. Technol. A Vacuum, Surfaces, Film. 2007, 25 (5), 1317. (18) Zhang, J.; Salzmann, I.; Schäfer, P.; Oehzelt, M.; Duhm, S.; Rabe, J. P.; Koch, N. J. Mater. Res. 2011, 24 (4), 1492–1497. (19) Backholm, M.; Foss, M.; Nordlund, K. Nanotechnology 2012, 23 (38), 385708.

ACS Paragon Plus Environment

8

Page 9 of 12

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Nano Letters

(20) Sebastian, P.; Lindner, F.; Walzer, K.; Lü ssem, B.; Leo, K. J. Appl. Phys. 2011, 110 (8), 84508. (21) Sugiyama, K.; Ishii, H.; Ouchi, Y.; Seki, K. J. Appl. Phys. 2000, 87 (2000), 295. (22) Wan, A.; Hwang, J.; Amy, F.; Kahn, A. Org. Electron. 2005, 6 (1), 47–54. (23) Liao, Z. M.; Hou, C.; Zhang, H. Z.; Wang, D. S.; Yu, D. P. Appl. Phys. Lett. 2010, 96 (20), 94– 97. (24) Tian, X.; Wang, L.; Wei, J.; Yang, S.; Wang, W.; Xu, Z.; Bai, X. Nano Res. 2014, 7 (7), 1065– 1072. (25) Kwan, W. L.; Lei, B.; Shao, Y.; Prikhodko, S. V.; Bodzin, N.; Yang, Y. J. Appl. Phys. 2009, 105 (12), 124516. (26) Chen, J.; Ma, D. Appl. Phys. Lett. 2005, 87 (2), 23505. (27) Jo, S. H.; Kim, K. H.; Lu, W. Nano Lett. 2009, 9 (1), 496–500. (28) Gao, S.; Song, C.; Chen, C.; Zeng, F.; Pan, F. Appl. Phys. Lett. 2013, 102 (14), 141606. (29) Chang, Y.-C.; Wang, Y.-H. ACS Appl. Mater. Interfaces 2014, 6 (8), 5413–5421. (30) Ligorio, G.; Nardi, M. V.; Steyrleuthner, R.; Ihiawakrim, D.; Crespo-Monteiro, N.; Brinkmann, M.; Neher, D.; Koch, N. Appl. Phys. Lett. 2016, 108 (15), 153302.

ACS Paragon Plus Environment

9

Nano Letters

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 12

Figure 1. A schematic diagram of a typical R-NVM is depicted in (a). Upon bias, active electrode metal atoms oxidize and ions migrate toward the opposite inert electrode where deoxidize therein, (b). The formation of a conductive metallic filament bridges the electrodes, (c).

Figure 2. Schematics of the deposition geometry are shown in (a) and (e). The topmost pictures refer to the sample evaporated at normal incidence (θ = 0°), the bottommost pictures to the sample evaporated via GLAD (θ = 80°). SFM images of the samples are reported in (b) and (f), respectively. Top-view SEM images and crosssection SEM images are reported in (c), (g) and in (d), (h) for the two samples respectively.

Figure 3. The schematic in (a) depicts the C-SFM set-up during the measurement of a reference nano-device. The schematic in (b) represents the energy alignment. The inserts (c) and (d) reports the measured current vs. time (I-t) as a function of the applied bias vs. time (U-t). The dashed lines are guides for the eyes.

ACS Paragon Plus Environment

10

Page 11 of 12

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Nano Letters

Figure 4. The schematic diagram in (a) depicts the experimental setup during the electric characterization via CSFM of a nanocolumn grown on Ag. The morphology of α-NPD film grown via GLAD on Ag is obtained via SFM and reported in (b); each nanocolumns is a single nano-device. The typical electrical characterization of a nanocolumn is shown in (c). I-t curve and U-t curve are plotted in the topmost and bottommost insets, respectively. The device switches from the HRS to the LRS once 6 V are applied between the Ag electrode and the C-SFM tip. If the bias is decreased to smaller value the current is still limited by the compliance, as depicted in (d).

Figure 5. I-t curves of a pristine device upon constant bias of 5 V are reported in (a) and (b); the curves refer to the same measurement, i.e. the current is displayed by increasing the time scale. The pristine device lies in HRS; at time 170 s the current increases several orders of magnitude [(b) is displayed in nA], and the device is set in LRS. I-t curves of pristine devices upon constant bias in (c) show the dependency of the switching time (τ ) with the applied bias. Switching time as a function of the applied constant bias in (d); the dotted line is an exponential fit of the experimental data.

ACS Paragon Plus Environment

11

Nano Letters

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

112x48mm (72 x 72 DPI)

ACS Paragon Plus Environment

Page 12 of 12