DNA-Templated Nanowires in the 10 nm Regime

Nanoscale wires on the order of 10 nm thick, formed from or patterned on DNA are presented here. Uncoated DNA wires from 4 nm thick and continuously metal-clad DNA wires from 10 nm thick are observed. These wires are produced by deposition of either single or double stranded DNA across a trench in a SiN substrate produced through chemical etching or focused ion beam techniques. Some of these wires are then coated with AuPd or other metals by sputtering, producing visually continuous metal wires. All wires are observed by scanning electron microscopy. Multi-strand DNA ’ropes’ of a greater thickness (30-40 nm) are also observed. These wires present possibility of conducting or superconducting wires and devices on the 10 nm scale. There has been much interest recently in nanoscale wires, and one current focus of this research is nanostructures built from DNA or using DNA as a template for construction with other materials. The DNA’s thin crosssection and capability for self-assembled structures make it an ideal building material on this scale. Although DNA has semiconducting properties on a 10 nm length scale, on the order of 100 nm or more it is thoroughly insulating. Therefore, interest is high in creating the thinnest possible conductive coating on a DNA substrate. Efforts by Braun et al. produced conducting Ag wires grown on a DNA substrate on the order of 100 nm in width and 12 μm long by chemical means. Subsequent work by Richter et al. gave rise to 50 nm wide Pd-on-DNA conducting nanowires of a similar length. Here is presented the *electronic mail: [email protected] production of numerous DNA nanowires, both bare DNA and metallized by coatings of AuPd, Os, or MoGe. A variety of wires of width from 4 nm to 40 nm and length from 100 nm to 960 nm were created. Though no conductivity measurements have yet been made, visually continuous coated wires ~10 nm wide have been observed. In this experiment, 48,502 base pair (16μm) ?-DNA was used to make double stranded wires, and variable length poly-C DNA formed single stranded elements. Methods Multilayered substrates, consisting of a thick layer of Si, topped by 216 nm SiO2 and 56 nm SiN are patterned by Ebeam lithography to carve a 100 nm wide trench and periodic distance markers through the SiN. After cleaning with HNO3 in an ultrasonic bath for 10 min, rinsed with DI water and sonicated again in isopropyl alcohol for 10 min, HF is used to etch the SiO2 for 9 s until an undercut is formed. The sample is then rinsed immediately with DI water to remove HF, and subsequently cleaned by manual agitation in HNO3. This must be done carefully because of the fragile nature of the overhanging SiN, the severe health hazards of using HF and the potential to ruin a substrate by excessive etching. Note that further sonication will destroy the fragile undercut regions. This undercut allows DNA to suspend in free space without adhering to the sides and bottom of the trench (Fig. 1). After a final rinse in isopropyl alcohol and forced N2 drying, DNA can be deposited. Trenches in thin membranes of SiN carved by focused ion beam (FIB) techniques also provide a suitable substrate for DNA wires. A substrate of Si with several 100 μm by 100 μm holes is sputter-coated with 60 nm of SiN, providing large, thin membranes. These are then patterned with cuts of desired width, length, and shape (generally long straight lines of a variety of widths), and carved with FIB. This technique allows trenches ranging from ~20 nm in width up to many microns. Due to availability of many 100 nm width trenches from the chemical etching process above, only lines of 200 nm or greater were produced by FIB for this effort (Fig. 2.). Membrane samples can also benefit from chemical cleaning, but a reliably non-damaging cleaning method has not yet been found. Both ?-DNA (1.6 nM, Promega, Madison, Wisconsin), poly-C (2.8μM, Sigma, St. Louis, Missouri) were deposited on samples. For either 2 μL droplet of DNA is deposited directly over the trench in multilayer samples or over the holes carved in a membrane sample. The droplet is allowed to settle for 2 min, so that DNA will attach to the substrate. The droplet is then blown away with forced dry N2 to avoid depositing salts from the buffer solution. Metallized samples were prepared by sputtering of AuPd or MoGe, or osmium plasma coated (courtesy of Structure Probe, Inc.). Both metallized and uncoated DNA samples were then imaged with scanning electron mircroscopy (SEM), preferentially at 15kV and under high magnification. For uncoated or lightly coated single stranded samples (poly-C), high magnification and voltage destroy the thin wires, so lower settings of 5kV and <70,000x magnification were used. Fig. 1. Schematic of sample with undercut trench. Trench width is 100 nm. Si layer thicker than shown Fig. 2. Various width cuts made in SiN membrane by FIB. Note the wires crossing the 5 right (narrower) cuts. Results These techniques were successful in producing hundreds of bare or metallized DNA wires with widths ranging from 4-5 nm uncoated strands up to 30-40 nm well coated, multistranded ‘ropes’. As most of the samples were prepared by the undercut trench method, these wires necessarily had a length of 100 nm. However, instances of wires as long as 960 nm were seen on membrane samples, and there is hope that still longer wires will be produced in subsequent efforts. Wires made without a coating were found to have widths from about 4 nm to 10 nm (Fig. 3.). Even the thinnest of these is still substantially thicker than the known width of double stranded DNA (2 nm) and of single stranded DNA (1 nm). This can be attributed to a number of phenomena. Depositing of buffer salts onto DNA would thicken wires, but the presence of large quantities of such salt should be visible then on the surface of the SiN substrate, which it often was not. The imaging process itself thickens the wires as the SEM deposits amorphous carbon as it scans the surface. This surely plays a role in the increased width of wires, particularly those imaged several times high quality (slow speed) scans, but does not explain all seen cases. Further, uncoated samples are poorly conducting, leading to fainter SEM imaging as compared to well conducting metallized samples and thus wider, blurrier appearing wires. This does not adequately explain all overthick wires either, and it is suspected, then, that many wires that are thicker than expected contain more than one DNA strand. This was seen in some cases in a split end on one or both ends of a wire. The clearest example of this multi-stranded ‘rope’ of DNA is 40 nm wide, with only a 3 nm sputtered coating of AuPd (Fig. 4.). The thin membrane of this particular sample also allowed faint imaging of these many strands as they disperse away from the wire on either end of the cut (Fig. 5.). Fig. 3. Pair of uncoated ?-DNA wires on an undercut trench substrate, width of 7-8 nm each. Note the surface’s freedom from unwanted buffer salt deposits. Thickening of wires was primarily caused by carbon depositing by repeated SEM imaging. Fig. 4. Mutli-stranded ‘rope’ of ?-DNA, 3 nm AuPd coating, width ~40 nm, length 960 nm. This is the leftmost (longest) wire visible in (2). The majority of samples produced were metallized, creating thicker, sturdier, and more easily imaged wires. Most of these coatings were produced by AuPd sputtering, and several thicknesses of films were created. . Thicker coatings (~7 nm) (Fig. 6a.) appeared more even Fig. 5. Detail of same wire as (4), note strands diverging on membrane surface from end of wire.