Evidence for Ionic Liquid Gate-Induced Metallization of Vanadium

Apr 3, 2017 - It has recently been shown that the metal–insulator transition in vanadium dioxide epitaxial films can be suppressed and the material ...
0 downloads 11 Views 3MB Size
Letter pubs.acs.org/NanoLett

Evidence for Ionic Liquid Gate-Induced Metallization of Vanadium Dioxide Bars over Micron Length Scales Donata Passarello,†,‡ Simone G. Altendorf,‡ Jaewoo Jeong,† Charles Rettner,† Noel Arellano,† Teya Topuria,† Mahesh G. Samant,† and Stuart S. P. Parkin*,†,‡ †

IBM Research - Almaden, San Jose, California 95120, United States Max Plank Institute for Microstructure Physics, Weinberg 2, 06120 Halle (Saale), Germany



ABSTRACT: It has recently been shown that the metal−insulator transition in vanadium dioxide epitaxial films can be suppressed and the material made metallic to low temperatures by ionic liquid gating due to migration of oxygen. The gating is only possible on certain crystal facets where volume channels along the VO2’s rutile c-axis intersect the surface. Here, we fabricate bars with the c-axis in plane and oriented parallel to or perpendicular to the length of the bars. We show that only bars with the c-axis perpendicular to the bars, for which the volume channels are accessible from the sides of the bar, can be metallized by ionic liquid gating. Moreover, we find that bars up to at least 0.5 μm wide can be fully gated, demonstrating the possibility of the electric field induced migration of oxygen over very long distances, ∼5 times longer than previously observed. KEYWORDS: Metal insulator transition, ionic liquid gating, metallization, epitaxial oxide films, vanadium dioxide lectric field induced tuning of the electronic properties of many materials is an important tool both for exploring the fundamental origins of these properties as well as for their potential technological use.1−7 One such means is via the use of ionic liquid gating that can provide very large electric fields at an interface between the liquid and the material of interest via the formation of an electric double layer (EDL).8−13 This tool can be used for varying electrostatically the electron density in the material close to or within the EDL. But recently it has been shown that the electric fields within the EDL are so large that they can lead to electrochemical changes at interfaces with oxide thin films.14−18 In these cases the changes that take place occur over distances that are much longer than the thickness of the EDL.15,19 Perhaps one of the most interesting examples is VO2 that has been extensively studied since the discovery of a metal insulator transition (MIT) in this material near room temperature in the 1950s.20 The electric field control of the MIT in VO2 and the related V2O3 has been of significant interest for some time, but a purely electronic control of the MIT has been elusive, confined to extremely short time scales21 due to thermal effects which otherwise dominate.22 The recent observation of the ionic liquid gate induced metallization of thin VO2 films has been shown to be due to an electrochemical mechanism related to the removal of small quantities of oxygen from the VO2 films,14−17 rather than a purely electronic mechanism. In our earlier work, the metallization of VO2 by ionic gating was found to be strongly crystal facet-dependent, related to the presence of volume channels in its rutile structure along the rutile c-axis along which it is supposed that oxygen ions

E

© 2017 American Chemical Society

migrate.16 We refer to the rutile structure even though below the MIT the structure is distorted and becomes monoclinic. It is interesting to note that the diffusion of H is also reported to be highly anisotropic in rutile VO2 and TiO223−26 (as well as other species, for example Li in rutile TiO227), with orders of magnitude higher diffusivity along the rutile c-axis than perpendicular to it. The VO2 films are only gatable when the volume channels are connected to the EDL.16 In other oxide systems such as WO3, which has a perovskite structure with volume channels in all directions, the films can be ionic liquid gated to a metallic state independent of the crystal orientation.18 An important question is the length scale over which these oxides can be gated from the EDL. Previously we have shown that VO2 (001) films as thick as 90 nm can be volume metallized.19 Here we use VO2 bars of varying widths in which the volume channels are oriented in-plane to show that bars as wide as 0.5 μm can be fully liquid gated to the metallic state from the exposed edges (side walls) of the bars. Moreover, we show that bars are only gatable to the metallic state when the volume channels are oriented perpendicular to the length of the bars. Bars fabricated from the same film structure with the volume channels oriented along the bars are little affected by ionic liquid gating. The strong dependence of gating on the orientation of the volume channels within the bars and the large distances over which the gating can take place not only confirms the critical role of the volume channels to the Received: December 2, 2016 Revised: March 30, 2017 Published: April 3, 2017 2796

DOI: 10.1021/acs.nanolett.6b05029 Nano Lett. 2017, 17, 2796−2801

Letter

Nano Letters

Figure 1. VO2 bar devices. (a) Schematic diagram of part of one device that is composed of a series of VO2 bars oriented along the (010) rutile axis. The (100) VO2 layers are shown in blue, and the corresponding (100) TiO2 substrate is shown in pink. Schematic layouts are on a 10 × 10 mm2 area substrate of (b) T devices and (c) X devices. The different arrangements of the atoms within the rutile structure on the exposed (001) and (010) edges are schematically illustrated at the top and bottom of the layout in b. Also illustrated is the oxygen that exits the (001) but not the (010) edges during the ionic liquid gating. The inset in c illustrates the array of bars within a single device. (d) Typical atomic force microscopy image and (e) line scans showing cross sections of two sets of bars; top: 1 μm wide bars spaced 1 μm apart and bottom: 0.5 μm wide bars spaced 0.25 μm apart.

used to ionic liquid gate the bars. Since there are no volume channels that are connected to the top surface of the VO2, only the edges can gate them, based on our previous studies.16 Furthermore, the VO2 film deposited within the trench has no exposed edges and is therefore not gateable. In any case, by design, only the bars are electrically connected to the source and drain contacts, as discussed further below, and not the VO2 layers at the bottom of the trenches. Each bar forms an electric double layer transistor (EDLT). Many bars were connected to common source and drain contacts so that the electrical properties of an array of EDLTs was measured. This allowed for any variations in properties of the bars to be averaged out in a single measurement. Each of these arrays comprises a single device. Two distinct device types were fabricated for transportdevice Tand synchrotron X-ray diffractiondevice

metallization of the oxide material but could lead to important technological applications, such as neuromorphic devices and membranes for oxygen separation. To avoid patterning of the deposited VO2 films which can lead to damage affecting its electrical properties, we patterned the underlying TiO2 substrate before depositing the VO2 film to create the bars. The TiO2 (100) single crystal substrates were patterned by e-beam lithography to create arrays of mesas separated by 40 nm deep trenches. The trenches were 2 mm long. The widths of the mesas and the spacings between them were each designed to be 0.25, 0.5, or 1 μm. Subsequently a VO2 layer is deposited on the patterned TiO2 substrate to form arrays of VO2 bars on top of the mesas. The trench is much deeper than the thickness of the deposited VO2 layer. Thus, these VO2 bars have exposed edges (i.e., side walls) which are 2797

DOI: 10.1021/acs.nanolett.6b05029 Nano Lett. 2017, 17, 2796−2801

Letter

Nano Letters Xmeasurements. The arrays were 100 μm wide for device type T and 300 μm wide for device type X. Each of the devices was surrounded by a “moat”, namely, a trench 100 μm wide to ensure electrical isolation of each device. The moats were spaced 25 μm apart. Figure 1a shows a schematic diagram of one device that is covered with ionic liquid. For device type T, four sets of nine devices, each with trenches of different widths and spacings, and patterned in close proximity to each other, were formed, as shown in Figure 1b, two sets with the bars oriented along, and two sets with the bars oriented perpendicular to the in-plane rutile c-axis. For device type T each device was gated independently by placing a droplet of ionic liquid selectively on a single device with a micropipette. Another droplet of ionic liquid was placed on the gate electrode, and additional ionic liquid was dispensed to connect the device and gate droplets while avoiding contact to the other devices present on the substrate. For device X all the devices were gated in parallel. The layout of the devices and their connections are shown in Figure 1c. For the T devices two gate electrodes ∼1 × 10 mm2 were fabricated on opposite sides of the substrate. The trenches were patterned using e-beam lithography with poly(methyl methacrylate) resist (950k molecular weight, 5% solution in anisole). This resist was spun onto the 10 × 10 mm2 samples at 2000 rpm spin speed followed by baking at 120 °C for 120 s. In order to prevent charging during e-beam exposure, a charge dissipation top coat was added consisting of the aquaSAVE conducting polymer solution (Mitsubishi Rayon Co., Ltd.), spun at 1500 rpm followed by baking at 90 °C for 60 s. The patterns were exposed using a Vistec/Leica VB6 e-beam machine with a 5 nA beam, taking 150 min per sample. After the water rinse of the aquaSAVE and 120 seconds resist development in a 1:3 solution of methyl-isobutyl ketone− isopropyl alcohol solution, the samples were etched in a PlasmaTherm Versaline ICP etch tool under the following conditions: pressure of 5 mT, BCl3 flow rate of 45 sccm (standard cubic centimeter per minute), SF6 flow rate of 5 sccm, a coil RF power of 300 W, and a platen RF power of 50 W. The total etch time was 75 s. Following this etch, the resist was stripped in hot N-methyl-2-pyrrolidone (NMP) solvent, followed by cleaning in EKC 265 postetch residue remover (DuPont) for 30 min at 60 °C. After this etching procedure, the initially insulating TiO2 substrates were electrically conducting, presumably due to the formation of oxygen vacancies. Thus, the substrates were annealed in a tube furnace in an oxygen flow of ∼10 sccm at the furnace temperature of 200 °C for 3 h. After this treatment they were no longer conducting. The substrates were then sonicated in deionized (DI) water for 10 min and then etched in 7:1 HF− NH4F buffered oxide etch (BOE) solution for 50 s. Next, they were rinsed in a stream of running DI water to remove residual acid from the surface, and finally they were dried by blowing nitrogen gas over the surface.28 The substrates were then loaded into a pulsed laser deposition (PLD) chamber and annealed again for ∼90 min at 500 °C, under the same oxygen pressure as used during the subsequent growth of the VO2 films. VO2 (100) thin films were deposited at a deposition temperature of 500 °C which was chosen to achieve the growth of the highest quality films. The oxygen pressure used during growth was optimized to yield VO2 films with the maximum change in resistance at the metal to insulator transition temperature (TMIT). VO2 films, ∼14 nm thick, were

grown on the patterned TiO2 substrates at oxygen pressures between 13 and 16 mTorr. A Coherent LPX 305 excimer laser operating at a wavelength of 248 nm and a pulse rate of 2 Hz was used for PLD. The laser beam was focused to a size of ∼1 × 5 mm2 with an energy density at the VO2 target that was attenuated by ∼70% to ∼1 J/cm2 per pulse. During growth both the target and substrates were rotated continuously, and the laser spot was rastered across the target surface. The substrate to target distance in the PLD chamber was ∼7 cm. After deposition, the samples were cooled down in the same oxygen pressure as used during growth. After the samples were removed from the PLD chamber, electrical contacts were added by using the same lithographic approach used to fabricate the trenches with 5% PMMA and aquaSAVE to create aligned openings over the ends of the bars. After rinse and develop, the samples were ashed in a plasma etcher to remove resist scum and then ion milled for 90 s in an Oxford IonFab before ion beam deposition of 5 nm of ruthenium followed by 65 nm of gold. Finally, the samples underwent liftoff in hot NMP for 30 min. The gate electrodes, ∼1 × 10 mm2 in area, for the T devices were fabricated later by a shadow masking technique using the same metal layers as the contacts. Atomic force microscopy (AFM) images of two type T devices with bar widths of 1 and 0.5 μm and spacings of 1 and 0.25 μm, respectively, are shown in Figure 1d. The surface of the VO2 films on top of the bars is very smooth with a rootmean-square (rms) roughness of ∼0.5 nm, whereas it is slightly rougher, with an rms roughness of ∼0.7 nm within the trenches, where the TiO2 substrate had been ion-milled prior to the VO2 deposition. AFM line profiles shown in Figure 1e confirm the depth of the trench to be ∼40 nm. Transmission electron microscopy samples were prepared by cross-sectional focused ion beam milling using a Ga ion beam on selected bars. Typical results are shown in Figure 2 for a bar with channels perpendicular to the length of the bars. The images were taken using a JEOL ARM200 microscope. A low magnification image is shown in Figure 2a and a highresolution, high-magnification image is shown in Figure 2b and c. The high-resolution image along the ⟨010⟩ zone axis clearly shows that the rutile c-axis is oriented perpendicular to the length of the bar and in the plane of the bar (see schematic in Figure 2c). The high-resolution image shows that the VO2 film is epitaxial with the rutile TiO2(100) single crystal substrate so that the VO2 unit cell is expanded along the c-axis by ∼3%.29,30 The clamping of the VO2 lattice to the substrate likely accounts for the small structural changes found from in situ synchrotron X-ray diffraction studies. Ionic liquid (IL) gating transport experiments on type T devices were performed in a Quantum Design DynaCool Physical Property Measurements System. The IL used for these experiments was 1-hexyl-3-methylimidazolium bis(trifluormethylsulfonyl)imide (HMIM-TFSI). The devices and the IL were separately baked at 120 °C in a high vacuum chamber at 10−6 Torr prior to the gating experiments, to eliminate any contaminants, such as water. A droplet of IL was placed on the substrate such that it covered an entire set of nine devices and the adjacent gate electrode. A single device from the set was selected for IL gating by ultrasonically wire bonding 25 μm diameter gold wires to its source and drain contacts. A gate voltage (VG) of +2.6 V and a source drain voltage (VSD) of +0.1 V were applied at 300 K, where the VO2 is in its insulating state. After 8 h of gating VG was reduced to 0 V, and the device 2798

DOI: 10.1021/acs.nanolett.6b05029 Nano Lett. 2017, 17, 2796−2801

Letter

Nano Letters

Figure 2. Transmission electron microscopy images of bars. (a) Z contrast, low-magnification cross-sectional image and (b) highresolution image of a VO2 bar that was 0.25 μm in width and whose length is oriented along (010). The specimen is about 50 nm thick in the beam direction: the vertical lines are an artifact of the FIB preparation due to specimen thickness variations. The sample was coated with an amorphous thermally evaporated carbon layer followed by a FIB platinum layer to protect the bar. (c) Magnified image showing the rutile unit cell with channels along the horizontal direction but not along the vertical direction. A schematic diagram showing the arrangement of the atom columns that accounts for the volume channels. The unit cell contains two rows of the columns of V atoms that can be clearly seen in the image. The oxygen atoms cannot be seen.

Figure 3. Anisotropy and time dependence of gating of VO2 (100) bars. Transport measurements on sets of 140 bars, 0.5 μm wide spaced by 0.25 μm. Measurements for a set of bars parallel to the c-axis (a) and perpendicular to c-axis (b) after gating for ∼8 h. (c) Timedependent transport measurements on one set of 50 bars, 1 μm wide and 1 μm apart. The measurements were performed after gating for 3, 6, and 12 h. (a−c) In each case the devices were gated at a fixed VG = 2.6 V, but the resistance measurements were carried out after VG was set to zero. A temperature sweep of 3 K/min was used.

resistance was measured as a function of temperature at a scan rate of 3 K/min. Figure 3a and b shows typical resistance versus temperature measurements, before and after gating, for two devices, one with bars parallel and one with bars perpendicular to the VO2 rutile c-axis. These devices each have 140 bars, 0.5 μm wide, spaced by 0.75 μm. Figure 3a shows that gating of the device for which the rutile c-axis is parallel to the length of the bars results in small changes in the device’s electrical properties. The MIT of the device is slightly reduced in temperature. In contrast, Figure 3b shows that gating of the bars in which the rutile c-axis is perpendicular to their length results in a complete suppression of the device’s MIT, and the device becomes metallic to the lowest temperature measured. Moreover, upon gating, the resistance of the device at temperatures above its MIT increases from 3.6 to 4.5 kΩ. The fact that no remnant MIT is observed for the bars with the rutile c-axis perpendicular to their length means that the entire volume of all the bars in this device were metallized. Furthermore, this shows that the ionic liquid gate induced metallization occurs up to a depth of 0.5 μm, the width of the bars. The observation that the bars have a very different response to ionic liquid gating depending on the orientation of the rutile c-axis has several implications. First, these results are consistent with our previous studies that showed that VO2 is metallized, not by electrostatic effects, but rather by electric field induced migration of oxygen along the volume channels within the rutile structure along the c-axis.15,16,18 Second, it is possible to achieve a volume metallization of VO2 (100) by gating from sidewalls as long as the sidewalls are perpendicular to the rutile

c-axis. In this case the volume channels are oriented perpendicular to the sidewalls with access to the EDL and the ionic liquid. The sidewalls have VO2(001) facets. On the other hand, when the sidewalls have (010) facets, as is the case for the bars with the rutile c-axis oriented along their length, we find almost no gating effect. In both cases the top surfaces of the bars have (100) facets, and therefore, if these surfaces played a role in the gating, we would have seen metallization, independent of the orientation of the bar edges. Moreover, our findings on bars with rutile c-axis oriented along their length are consistent with a previous study where no IL gating effect was observed on nanobeams whose c-axis was oriented along the nanobeam axis.31 Third, the gating effect occurs over a very long length scale of at least 0.5 μm. This also indicates the high quality of the VO2 films used in this study as the open volume channels must extend over this same length scale. Figure 3c shows a series of IL gating experiments that were performed on a type T device that had wider bars. This device had 50 bars, each ∼1 μm wide, spaced ∼1 μm apart, with the rutile c-axis perpendicular to the length of the bars. Temperature dependent resistance measurements were performed on this device in its insulating state and after gating in vacuum for 2799

DOI: 10.1021/acs.nanolett.6b05029 Nano Lett. 2017, 17, 2796−2801

Letter

Nano Letters

= 2.6 V, no shift in the film’s diffraction peak is observed for the set of bars parallel to the c-axis, whereas a shift of ∼0.4° is observed for the bars perpendicular to the c-axis. A shift of the peak position to a lower 2θ angle represents an expansion of the out-of-plane lattice parameter. While this peak shift is very significant, it is much smaller than we have previously observed when gating a (001) oriented VO2 film from its surface in which the film is found to expand by 3% or more out-of-plane. Here the bars cannot expand in plane along the (001) axis since their lattice is clamped to the TiO2(100) substrate and is therefore already expanded by ∼3%. We have clearly demonstrated that bars of crystalline VO2 can be converted from an insulating to a metallic state by ionic liquid gating from the side walls of the bars over substantial distances of more than 0.5 μm. We note that the area of the surface of the bars exposed to the ionic liquid is about 20 times larger than the area of the side walls, yet our results show that negligible gating takes place through these surfaces. Gating occurs only when the side walls are properly crystallographically oriented, when the volume channels in the VO2’s rutile structure are oriented perpendicular to the side walls. These results are a clear manifestation of the oxygen vacancy formation mechanism that gives rise to the ionic liquid gating metallization. These results are important because not only do they demonstrate that oxygen vacancies can be induced even at large distances from the EDL but that the process takes place over considerable time. This could be useful for several technological applications, perhaps most interestingly for neuromorphic computing devices where analogue devices are needed. Another potential application is the use of VO2 membranes for oxygen separation driven by ionic liquid gating on one or both sides of the membrane. Once an oxygen vacancy gradient is established across the membrane the process could be much faster than the initial process of setting up this gradient.

3, 6, and 12 h. A significant reduction in the MIT was observed, but the metallization was not complete even after gating for 12 h. Note that between each gating experiment the sample was removed from vacuum into air and the IL was removed by rinsing in iso-propyl alcohol (IPA) for 5 min and blow drying in nitrogen. We found that during this process the device resets itself to its initial insulating state as confirmed by transport measurements. The sample was baked in vacuum at 100 °C for several hours to remove any residual water before performing the subsequent gating experiment. We note that TiO2 has the same rutile structure and correspondingly volume channels as VO2; earlier it has been found that ionic liquid gating has a similar crystal facet dependence, but the metallization is confined to a thin surface layer (∼2 nm thick), and the conductivity of this metallized layer is 100 times smaller than the conductivity of the metallized VO2.14 Moreover, the effect is volatile so that when VG = 0 the TiO2 has reverted to its insulating state. Thus, the TiO2 substrate does not play a role in our VO2 gating experiments. Type X devices were fabricated to investigate structural changes of the VO2 bars upon IL gating. X-ray diffraction (XRD) measurements were carried out, in situ, at the Stanford Synchrotron Radiation Laboratory (SSRL) at room temperature in a purpose-built electrochemical cell.16,32 Type X sample had a total of 18 devices with 9 devices in which the rutile c-axis of VO2 was parallel and 9 devices in which the rutile c-axis of VO2 was perpendicular to the length of the bars. The devices were 300 μm wide in order to accommodate the width of the focused X-ray beam. Source and drain contacts were formed from ∼1 mm × 10 mm wide stripes of 5 nm Ru/65 nm Au that were deposited along opposite sides of the 10 mm × 10 mm substrate. These contacts were connected to the source and drain of each of the devices, in parallel (see Figure 1c). Thus, all the devices on one Type X sample were gated simultaneously. The same ionic liquid, HMIM-TFSI, was used as for the transport measurements. The XRD measurements were carried out using a monochromatic X-ray beam with an energy of 12 keV. Figure 4a and b shows XRD θ−2θ patterns for two devices each with 300 bars, ∼2 mm long, 0.5 μm wide at a spacing of 0.5 μm. For one device the rutile c-axis is parallel, and for the other device, perpendicular to the length of the bars. For both devices, the XRD data in the out-of-plane direction show sharp substrate peaks at 53.4° (TiO2(400)) and a broader diffraction peak from the VO2 film at ∼54.7° (VO2(400)). After gating for 12 h at VG



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. ORCID

Stuart S. P. Parkin: 0000-0003-4702-6139 Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS Part of this research was carried out at the Stanford Synchrotron Radiation Lightsource, a Directorate of SLAC National Accelerator Laboratory and an Office of Science User Facility operated for the U.S. Department of Energy Office of Science by Stanford University. We thank Eugene Delinia for TEM sample preparation.



REFERENCES

(1) Hwang, H. Y.; Iwasa, Y.; Kawasaki, M.; Keimer, B.; Nagaosa, N.; Tokura, Y. Nat. Mater. 2012, 11, 103−113. (2) Waser, R.; Aono, M. Nat. Mater. 2007, 6, 833−840. (3) Zhang, J.; Averitt, R. D. Annu. Rev. Mater. Res. 2014, 44, 19−43. (4) Ueno, K.; Shimotani, H.; Yuan, H.; Ye, J.; Kawasaki, M.; Iwasa, Y. J. Phys. Soc. Jpn. 2014, 83, 032001. (5) Stemmer, S.; Allen, S. J. Annu. Rev. Mater. Res. 2014, 44, 151− 171. (6) Chakhalian, J.; Freeland, J. W.; Millis, A. J.; Panagopoulos, C.; Rondinelli, J. M. Rev. Mod. Phys. 2014, 86, 1189−1202.

Figure 4. Structural evolution of gated VO2 (100) bars. In situ XRD diffraction patterns in pristine state (black) and after gating for 12 h with a gate voltage VG = 2.6 V and a source drain voltage VSD = 0.3 V (red). The measurements are performed on two sets of bars 0.5 μm wide and 0.5 μm apart, one set parallel to the c-axis (a) and one set perpendicular to the c-axis (b). 2800

DOI: 10.1021/acs.nanolett.6b05029 Nano Lett. 2017, 17, 2796−2801

Letter

Nano Letters (7) Bjaalie, L.; Himmetoglu, B.; Weston, L.; Janotti, A.; Van de Walle, C. G. New J. Phys. 2014, 16, 025005. (8) MacFarlane, D. R.; Tachikawa, N.; Forsyth, M.; Pringle, J. M.; Howlett, P. C.; Elliott, G. D.; Davis, J. H.; Watanabe, M.; Simon, P.; Angell, C. A. Energy Environ. Sci. 2014, 7, 232−250. (9) Fujimoto, T.; Awaga, K. Phys. Chem. Chem. Phys. 2013, 15, 8983−9006. (10) Simon, P.; Gogotsi, Y. Nat. Mater. 2008, 7, 845−854. (11) Tsuda, T.; Hussey, C. L. Interface 2007, 42−49. (12) Galiński, M.; Lewandowski, A.; Stępniak, I. Electrochim. Acta 2006, 51, 5567−5580. (13) Goldflam, M. D.; Liu, M. K.; Chapler, B. C.; Stinson, H. T.; Sternbach, A. J.; McLeod, A. S.; Zhang, J. D.; Geng, K.; Royal, M.; Kim, B.-J.; Averitt, R. D.; Jokerst, N. M.; Smith, D. R.; Kim, H.-T.; Basov, D. N. Appl. Phys. Lett. 2014, 105, 041117. (14) Schladt, T. D.; Graf, T.; Jeong, J.; Aetukuri, N.; Li, M.; Fantini, A.; Jiang, X.; Samant, M.; Parkin, S. S. P. ACS Nano 2013, 7, 8074− 8081. (15) Jeong, J.; Aetukuri, N.; Graf, T.; Schladt, T. D.; Samant, M. G.; Parkin, S. S. P. Science 2013, 339, 1402−1405. (16) Jeong, J.; Aetukuri, N. B.; Passarello, D.; Conradson, S. D.; Samant, M. G.; Parkin, S. S. P. Proc. Natl. Acad. Sci. U. S. A. 2015, 112, 1013−1018. (17) Li, M.; Han, W.; Jiang, X.; Jeong, J.; Samant, M. G.; Parkin, S. S. P. Nano Lett. 2013, 13, 4675−4678. (18) Altendorf, S. G.; Jeong, J.; Passarello, D.; Aetukuri, N. B.; Samant, M. G.; Parkin, S. S. P. Adv. Mater. 2016, 28, 5284−5292. (19) Passarello, D.; Jeong, J.; Samant, M. G.; Parkin, S. S. P. Appl. Phys. Lett. 2015, 107, 201906. (20) Morin, F. J. Phys. Rev. Lett. 1959, 3, 34. (21) Gray, A. X.; Jeong, J.; Aetukuri, N. P.; Granitzka, P.; Chen, Z.; Kukreja, R.; Higley, D.; Chase, T.; Reid, A. H.; Ohldag, H.; Marcus, M. A.; Scholl, A.; Young, A. T.; Doran, A.; Jenkins, C. A.; Shafer, P.; Arenholz, E.; Samant, M. G.; Parkin, S. S. P.; Dürr, H. A. Phys. Rev. Lett. 2016, 116, 116403. (22) Brockman, J. S.; Gao, L.; Hughes, B.; Rettner, C. T.; Samant, M. G.; Roche, K. P.; Parkin, S. S. P. Nat. Nanotechnol. 2014, 9, 453−458. (23) Johnson, O. W.; Paek, S. H.; DeFord, J. W. J. Appl. Phys. 1975, 46, 1026−1033. (24) Bates, J. B.; Wang, J. C.; Perkins, R. A. Phys. Rev. B: Condens. Matter Mater. Phys. 1979, 19, 4130−4139. (25) Lin, J.; Ji, H.; Swift, M. W.; Hardy, W. J.; Peng, Z.; Fan, X.; Nevidomskyy, A. H.; Tour, J. M.; Natelson, D. Nano Lett. 2014, 14, 5445−5451. (26) Kasırga, T. S.; Coy, J. M.; Park, J. H.; Cobden, D. H. Nanotechnology 2016, 27, 345708. (27) Johnson, O. W. Phys. Rev. 1964, 136, A284−A290. (28) Martens, K.; Aetukuri, N.; Jeong, J.; Samant, M. G.; Parkin, S. S. P. Appl. Phys. Lett. 2014, 104, 081918. (29) Burdett, J. K.; Hughbanks, T.; Miller, G. J.; Richardson, J. W.; Smith, J. V. J. Am. Chem. Soc. 1987, 109, 3639−3646. (30) Longo, J. M.; Kierkegaard, P. Acta Chem. Scand. 1970, 24, 420− 426. (31) Ji, H.; Wei, J.; Natelson, D. Nano Lett. 2012, 12, 2988−2992. (32) Samant, M. G.; Toney, M. F.; Borges, G. L.; Blum, L.; Melroy, O. R. J. Phys. Chem. 1988, 92, 220−225.

2801

DOI: 10.1021/acs.nanolett.6b05029 Nano Lett. 2017, 17, 2796−2801