Sub-5-Nm Gaps Prepared by On-Wire Lithography: Correlating Gap Size with Electrical Transport
Lidong Qin,Jae-Won Jang,Ling Huang,Chad A. Mirkin
DOI: https://doi.org/10.1002/smll.200600386
IF: 13.3
2006-01-01
Small
Abstract:OWL – a wise choice: On-wire lithography (OWL) has been used to fabricate 2.5-nm gaps in high yield in Au nanowires. Current–voltage measurements of the gaps show tunneling transport behavior and calculated gap sizes that match those determined by electron microscopy (see figure). Such gaps prepared by OWL can be utilized to make measurements on nanomaterials, and even on single molecules. The development of the field of nano- and molecular electronics depends greatly on the ability to fabricate nanometer-sized gaps between electrodes.1, 2 For the field to advance both from a fundamental and technological standpoint, reliable, high-throughput, low-cost fabrication methods are needed.1, 3–8 With current lithography methods, such as photolithography, electron-beam lithography, and focused ion-beam lithography,9 it is difficult to fabricate structures smaller than 10 nm and even more difficult to build devices with dimensions that can address single molecules (less than 5 nm). To access this extraordinarily small molecular length scale, researchers have developed a series of techniques to prepare sub-5-nm gaps, including the break-junction method,3, 10 gap-narrowing procedures,11, 12 and a variety of unconventional nanopatterning techniques.13–17 Among these methods, the formation of junctions using either mechanical force3 or electron migration10 can yield gaps on the molecular length scale. However, these methods are hindered by the fact that the fabrication approaches employ high-cost electron-beam lithography to generate nanowire patterns on substrates, and the yields of working devices are extremely low (typically less than 5 %). While it is possible to obtain sub-5-nm gaps by narrowing procedures with micrometer-sized electrodes,11, 12 these processes are also complicated, difficult to control, and very low yielding. Recently, a variety of new approaches for synthesizing gap structures have been developed. All of them rely on the use of template synthetic methods to prepare one-dimensional wires with sacrificial materials strategically located along the long axis of the wire. One method developed by Bao and co-workers utilizes both high temperature and chemical-etching approaches to remove a sacrificial spacer layer between two Au ends.18 This results in two separate rods made of Au, separated by a gap as small as 20 nm. Keating and co-workers used an approach where they initially created a core/shell structure by encapsulating a wire with sacrificial layers with silica.19 Subsequent dissolution of the sacrificial layer with an etchant created rods with gaps defined by the size of the sacrificial layer. Although these structures are anticipated to have all sorts of interesting optical properties, it is difficult to envision using such structures for electrical applications since they are encapsulated with an insulating silica shell. We developed a procedure called on-wire lithography (OWL),20 which utilizes nanowires with sacrificial layers incorporated along the long axis of the wire. The wires are then coated on one face with silica. This is accomplished by dispersing the as-prepared wires on a glass substrate and then depositing silica on them by plasma-enhanced chemical vapor deposition (PECVD). Sonication releases the wires from the glass substrate and exposes the unprotected side of the wire. Dissolution of the sacrificial spacer layer with an etchant yields wires with gaps that can be precisely controlled and where the electrode ends are fixed to an underlying silica coating. In the first report of the OWL procedure,20 we demonstrated gap-fabrication capabilities down to 5 nm with electrical measurements on 13-nm gaps. The principle advantages of OWL for nanoelectrodes include the ability to fabricate sub-20-nm gaps in high yield, precise dimensional control (the diameter of the wire is controlled by template-pore diameter, and the length of the wire and size of the gap are controlled electrochemically), high batch-to-batch reproducibility, and the requirement of only simple and highly accessible fabrication instrumentation.20 In principle, one can prepare gaps as small as an atomic layer of the sacrificial material used in the OWL process, but in practice, thus far only 5-nm gaps have been fabricated. Moreover, the gap characterization has been primarily by field-emission scanning electron microscopy (FE-SEM) analysis, which only provides information about the gap at the exposed surface. From an electrical-measurement standpoint, the tunneling properties of the gaps are critical and are likely a better indicator of the actual gap size. Herein, we show that OWL can be used to fabricate gaps as small as 2.5 nm, and we use electrical measurements in addition to FE-SEM data to characterize and assign the actual gap size. To evaluate the range of gaps that can be fabricated by OWL, we synthesized Au wires with Ni stripes through the template synthesis method using an anodic aluminum oxide (AAO) template and literature procedures.20, 21 The Ni segment length, which was varied over the 40 to 2.5-nm range (Figure 1), 1can be controlled by controlling the applied charge during deposition. For example, 20-nm Ni segments are obtained when 200 mC of charge is applied during electrochemical deposition, while 2.5-nm-thick segments result from 25 mC of passed charge. A) Schematic diagram of nanowires with Ni segments of 40, 30, 20, 10, 7.5, 5, and 2.5 nm. B) Gapped nanowire structures after on-wire lithography. C) Plot of gap size versus charge passed during electrochemical deposition. Whatman Anodisc 25 alumina templates (0.02 μm, 25 mm) were used. Following the OWL process, we fabricated striped wires consisting of Au and Ni segments via the template synthesis method. After dissolution of the template, the resulting multisegmented nanowires were imaged by FE-SEM. Billions of nanowires can be obtained in one batch (based upon the number of pores in the AAO template), and the dimensions of the Ni segments appear to be quite uniform (Figure 2). 2A high-resolution image of a single nanowire demonstrates the high level of segment-length control down to 2.5 nm (Figure 2 B). 2In the next step of the OWL process, the Au/Ni multisegmented nanowires were coated on one side with a 50-nm layer of silica by PECVD, and the Ni segments were subsequently removed by wet-chemical etching (Figure 1 B 1and Figure 2 C,2 D). Due to the insulating silica layer, the resulting gaps often appear quite bright in the SEM (Figure 2 C). 2When a 10-nm layer of carbon was coated on wires in an attempt to avoid this charging effect, the gaps could be more clearly observed and appear as dark contrast areas (Figure 2 D).2 A) Large-area SEM image of Au/Ni nanowires with different Ni segments. B) SEM image of a single nanowire with Ni segment sizes of 40, 30, 20, 10, 7.5, 5, and 2.5 nm. Each Ni segment is separated by a 100-nm Au segment. A 40-nm Ni segment is used as a marker at one end near the 2.5-nm Ni segment. The inset is a high-magnification image of the smallest Ni segment. C) The same nanowires shown in (A) after dissolution of the Ni. D) A high-magnification image of a gapped nanowire. Gap sizes are 40, 30, 20, 10, 7.5, 5, and 2.5 nm (from top to bottom). The ability to prepare sub-3-nm gaps is significant, especially for those interested in studying transport in molecule-based systems. Therefore, we decided to more carefully evaluate the types of gap that can be prepared in this size regime. OWL was used to synthesize Au wires, each with a single 2.5-nm gap (Figure 3). 3High-resolution FE-SEM and transmission electron microscopy (TEM) images, before and after removal of the Ni that defines the resulting gap, suggest the formation of uniform structures that approach 2.5 nm (the variation is approximately 10 % based upon an analysis of 100 wires). TEM allows the gap to be clearly imaged, appearing as a bright line separating the two Au wire ends. Note the internal Au surfaces at both ends of the gaps are not atomically flat. The roughness is in part a result of the rough Ag used to electrochemically deposit the Au and Ni features. A) SEM image of Au–Ni–Au nanowires with 2.5-nm Ni segments. B) A high-magnification image showing gaps in the center of the nanowires. C) A single Ni segment of 2.5 nm. D) TEM image of a 2.5-nm gap, which is a result of the OWL process on nanowires shown in (A–C). To assign actual gap sizes, I–V measurements were carried out on gapped nanowires using a microelectrode configuration.22 In a typical experiment, the nanowires, suspended in water, were deposited onto 2-μm wide, 12-μm long Au microelectrodes separated by 2 μm. Upon drying, some of the nanowires form a bridge across the microelectrodes (Figure 4 A, 4inset), allowing for easy I–V measurements on them. The I–V curve of a typical wire with a 2.5-nm gap (as assigned by TEM and SEM) exhibits linear behavior over the −0.8 to +1 V region and agrees with the direct tunneling equation in the low-bias region (Figure 4 D, 4inset and Equation (1)). The measurements are quite reproducible from wire to wire and give similar levels of resistance (Figure 4 B). 4The data for seven different wires show a resistance variation of less than an order of magnitude and over ten orders of magnitude greater resistance than structures without a gap (Figure 4 B). 4The low measured resistance for the nanowires without gaps is consistent with the conclusion that the contact resistance between the Au ends of the nanowires and the microelectrode is negligible. As a control experiment, the resistance between two microelectrodes was measured (5×1013 Ω) and was found to be an order of magnitude greater than the average resistance value for the nanowires with gaps (5×1012 Ω). Collectively, these data show that the measured resistance for the nanogap structures are due to the gaps rather than contact resistance or resistance inherent in the microelectrode array configuration. A) I–V curve of a nanowire with a 2.5-nm gap. The solid line is a simulated line based upon the direct tunneling equation in the low-bias region. Inset: a gapped nanowire spanning two microelectrodes. B) Resistance measurements: across the microelectrodes, for nanowires with gaps (#1–#7) and without a gap (no gap). Resistances of nanowires and microelectrodes were measured several times. These box charts represent the distribution of the collected resistance data. The resistance of the microelectrode gap was measured five times and the resistances for the nanowires (#1–#7 and the no-gap nanowire) were measured 2, 1, 2, 2, 3, 1, 1, and 1 times, respectively. C) Arrhenius plots of current measured from nanogap nanowires as a function of temperature. D) Nanogap size calculated by fitting to a tunneling equation. To determine whether the transport is due to electron hopping or electron tunneling across the gaps within the nanowire system, measurements of current as a function temperature were carried out. Arrhenius plots showed no significant temperature dependence from 300 to 90 K (Figure 4 C). 4Therefore, it was concluded that a tunneling process, rather than capacitive hopping, is the dominant mode of electrical transport in these gapped nanostructures.23–26 In summary, we have developed an OWL-based method to produce gap structures on the sub-5-nm length scale. Electrical transport measurements and electron microscopy measurements suggest that we can make structures at least as small as 2.5 nm in a very reproducible manner. This simple and easily adoptable method is ideally suited to fabricate structures that can be used to make measurements on nanomaterials and perhaps even individual molecules. When one considers that the process can be used with many electrode materials, it should be particularly well-suited for matching the proper surface chemistry and electrode work functions with the molecules and materials to be studied. Electrochemical deposition: In all experiments, commercially available Ni plating solution (Ni sulfamate from Technic Inc.) was used for electrochemical deposition. For the growth of Ni segments thicker than 10 nm, a −900 mV deposition potential (Ag/AgCl reference, Pt counter electrode) was used. In order to precisely deposit segment lengths of 7.5, 5, and 2.5 nm, the Ni plating solution was diluted by a factor of 100 and a potential of −780 mV was applied. Note that careful rinsing of the electrochemical deposition cell was required when changing Au and Ni plating solutions. During the OWL process, segment lengths were controlled by monitoring the charge passed through the working electrode. Au segments between Ni were fixed at 100 nm. Ni segments, 40, 30, 20, 10, 7.5, 5, and 2.5 -nm long were made by passing 403, 302, 201, 102, 74, 51, and 25 mC of charge, respectively. A final 40-nm Ni segment was deposited as a marker at the end of the gap array. This was used to quickly identify the orientation of the wires and location of the gaps by SEM. Tunneling current measurement setup: I–V measurements of the nanowires were carried out with a Keithley 6517 A electrometer/high resistance meter and a two-probe configuration. Except where noted, all of the measurements were performed under vacuum (10−2 mbar) and at room temperature (≈293 K). The contacts between the nanowires and microelectrodes were inspected with an optical microscope prior to each measurement. The configurations where the wire rested on the microelectrodes with the insulating silica coating on the bottom were discarded.