4.2. Nanowire Definition

In this section, we present the experimental details of the three methods we have used to fabricate nanowires. These are shown schematically in Figure 2, and Figure 3 shows images of one sample fabricated by each of the methods, collected using a helium focussed ion-beam or scanning electron microscope.

Our first technique for defining nanowires employs a negative resist to define a mask. We use hydrogen silsesquioxane (HSQ) resist, diluted to 1% HSQ in methyl isobutyl ketone (MIBK) and spun at 2000 rpm to give a 35-nm HSQ layer. To achieve nanowires with widths as low as 15–20 nm, we used a 10-kV electron beam to write single-pixel lines in the resist using a line dose set at some multiple of a default value of 1280 pC/cm, followed by development for one minute in MF-26A developer. The nanowire is then formed using reactive ion etching (RIE) at 100 W and 100 mTorr using flows of 35 sccm of CHF₃ and 14 sccm of SF₆ for approximately 120 s in 20-s bursts to mitigate potential overheating during etching. Since removal of exposed HSQ requires the use of HF, we do not remove any HSQ remaining after RIE.

For the second technique we have used for defining nanowires via EBL, we utilise polymethyl methacrylate (PMMA), a positive-tone resist, and follow a “cut-out” strategy [30] to expose the edges of the nanowires to be removed, followed by RIE, using the same recipe as above. This method means that HSQ resist is not required, and the PMMA resist may be removed in acetone when etching is complete. Figure 3b shows an image of a nanowire prepared using this method. The ‘cut-out’ lines are typically ∼20 nm wide. The data shown in Figure 4 were obtained on nanowires prepared in this way. An advantage of this technique is that the ultimate width obtainable by the technique may be made smaller than the narrowest line obtainable by patterning the PMMA resist, since the linewidth is determined by the difference between the separation of the centres of the ‘cut-out’ lines and the width of the cut-out line, which may be chosen when the pattern is exposed by EBL. Note also that over-exposure in this geometry leads to a reduction in the width of the nanowire obtained, rather than an increase in the width as would be obtained when patterning using a negative resist. As for any EBL-based process in which ultimate resolution is sought, the ultimate linewidth is obtained following suitable dose tests and relies on the stability of results obtained by the EBL and subsequent development and etching.

The third fabrication technique we have used is based on neon focussed ion-beam milling. This does not rely on either an e-beam resist mask or RIE when defining the nanowire and allows milling with 5-nm resolution. Although some damage is inevitable in focussed ion-beam milling, since neon ions are inert (unlike the more commonly-used gallium), poisoning from implanted ions is expected to be avoided when milling using neon. We typically use a Ne ion beam accelerated to 15 kV, providing a beam current of ∼2 pA, to supply a dose of 0.5–1 nC/μm² to remove material from the NbN film to define the nanowire. Because of the relatively slow speed of milling, it is not practicable to fabricate the whole structure by FIB, so the deposited NbN film is first coated with PMMA and patterned by EBL, then etched by RIE to define the coarse features of the structure, in a similar way to the “cut-out” strategy. A wider nanowire (with a width of 300 nm) is patterned in this step, and this nanowire is then milled using the Ne-FIB to remove material in order to define a nanowire. In previously published work, we have also used this technique successfully to fabricate NbN nanowires as elements within superconducting co-planar waveguide resonators [40,41].