In Situ Monitoring of Etching Steps

Typical current traces in some Au/Ti double-layered samples observed during FIB etching are shown in Fig. 4. In this experiment, the thickness of the Au film was ~30 nm. Figure 4a shows a current trace measured during the nanowire formation step. The small steps observed in the current trace were caused by interruptions in the etching, during which a prefabricated marker pattern was detected and beam drift was minimized. After the FIB irradiation began, the current initially remained unchanged and then began to decrease. At the initial stage of etching, the sample resistance was dominated by the resistance of the lead regions, which resulted in a constant current. When the remaining film thickness in the irradiated regions became sufficiently thin, the sample current was mainly determined by the resistance of the nanowire region. It can be seen from Fig. 4a that the current exhibited a rapid decrease after initially remaining constant and then decreased more gradually. The rapid decrease in the current trace corresponded to the final stage of the nanowire formation etching. The gradual decrease in the current after the rapid decrease reflected the narrowing and thinning of the nanowire caused by etching at wire edges due to FIB tail components. The nanowire formation etching was terminated manually after the gradual decrease was observed in the current trace.

A current trace measured during the nanogap formation step is shown in Fig. 4b. Again, the two large steps caused by the interruptions in the etching for the drift correction are observed in the current trace. In the present step, the FIB etching was terminated by blanking the FIB automatically at a current level of ~120 nA, which was preset by Vref. The removal rate of the Au film for this example corresponded to ~3.8 nm/s or ~15 ML (monolayers)/s. The current decreasing rate (~1.2 μA/s) at the final etching stage suggests that the average remaining thickness should be ~1 to 2 ML. The inset in Fig. 4b shows the normalized current trace observed without the interruptions in the etching, along with the simulated current as a function of the normalized depth removed (which is the same with the normalized etching time). The current trace during the nanogap formation step could be reproduced reasonably by the numerical simulation. A gradual decrease in current with a smaller slope by the Ti to Au etching rate ratio should be observed at the end of the sharp current decrease if the thin Ti adhesion layer had the same magnitude of conductivity as the bulk value. The absence of such a slower decrease in the present experiments may suggest that the conductance of the thin Ti films was very low and was negligible compared to that of the Au film.

Monitoring Current Level

It is important to adjust the preset current level that controls the blanking of the FIB for the reproducible fabrication of nanogap electrodes using the present method. It was observed that narrower nanogaps were formed when a higher blanking current was used, as could be expected. As an example, Fig. 5a shows an SEM image of a nanogap electrode fabricated using a relatively higher blanking current of ~230 nA in a Au/Ti sample. Using the present blanking, a ~3-nm wide gap was formed, and the average remaining thickness at the time of the blanking was estimated to be 2 to 3 ML. However, we found that nanogaps fabricated with a blanking current higher than 100 nA were often broken after the current–voltage (I-V) characteristic measurements of the gap. The current density of the blanking current was estimated to be ~1011 A/m2, which is more than an order of magnitude lower than the current level needed to induce electromigration in Au films (Durkan et al. 1999). The break might have been caused by a discharge of static electricity during sample handling before the I-V characteristic measurement, electromigration during the I-V characteristic measurement, or by a combination of Joule heating and residual stress in the films. Figure 5b shows an SEM image of a nanogap electrode fabricated with a relatively low blanking current of ~40 nA in the Au/Ti sample and observed after the I-V characteristic measurements. The figure indicates that a ~5-nm wide gap was formed without a break. Using the present method, nanogaps with a width narrower than half of the FIB spot size (~12 nm) can be reproducibly obtained.

Electrode Thickness

Nanogaps with thinner electrodes are advantageous because the number of molecules that bridge electrodes is expected to decrease, which makes it possible to measure the conductivity of this smaller number of molecules connected in parallel. We examined the feasibility of fabricating thin electrode films by using Au/Ti double layer films with a thin Au electrode film with a thickness of ~10–15 nm. Figure 6a shows an SEM image of a fabricated nanogap electrode. A nanogap with a width of ~7 nm was successfully formed. However, electrodes were frequently broken during nanowire formation etching, as shown in Fig. 6b. In this experiment, the gap was formed accidentally before the nanogap formation step. This was because the effect of etching at the wires due the FIB tail and the thickness fluctuation caused by the grain became significant compared with thick Au films, and the thinnest region was first broken during the nanowire etching process. It should be noticed that this break was not caused by a complete removal from the etching but rather by a coupled effect that included the monitoring current and residual film stress. To reproducibly fabricate nanogap electrodes in thin electrode films, precise adjustments of the etching time and FIB focusing are needed.

Improvement of Fabrication Reliability

An unintentional defocusing of the FIB enhances the narrowing and thinning of nanowires and degrades the reliability in the nanogap fabrication using Au/Ti doublelayered samples. The reliability of nanogap fabrications can be improved using Ti/Au/Ti triple-layered samples. The representative current traces observed during nanowire formation steps in Au/Ti and Ti/Au/Ti samples are shown in Fig. 7a. The thickness of the Au films was ~30 nm. In these fabrications, rectangle-shaped patterns were used to form nanowires instead of the U-shaped patterns in order to obtain the side view image of nanogap electrodes. The increase of the etching time observed for the nanowire formation in the Ti/Au/Ti sample was due to the etching of the top Ti film that has much less etching rate than Au. It can be seen from Fig. 7a that the decrease rate of the currents after the nanowire formation (rapid decrease of current) in the Au/Ti samples differed among two traces. The higher decrease rate indicates that the thickness and width of the nanowire are decreased by the etching. This should occur by FIB tail component which may vary depending on operation conditions such as focusing and vacuum in the optics column. We also see that that the current decrease caused by the narrowing and thinning of the nanowire was considerably reduced in the Ti/Au/Ti sample. The current traces during the nanogap formation steps are shown in Fig. 7b. The current in the Au/Ti sample decreased immediately after FIB irradiation, while the current in the Ti/Au/Ti sample showed gradual decrease and then rapid decrease. The gradual decrease reflects the etching of the top Ti film, indicating that the Ti film remained after the nanowire formation etching.

Figure 8 shows SEM images of a nanogap electrode fabricated in the Au/Ti double-layered sample. The side view image was obtained by rotating the sample at a 70 angle with respect to the substrate normal. The current traces are shown in Fig. 7 by solid lines. As can be seen in Fig. 8a, a nanogap with a width of ~5 nm was formed. Using the Au/Ti double-layered samples with ~30-nm thick Au, we succeeded in the fabrication of nanogaps with widths of ~5–8 nm. However, gaps wider than 10 nm were sometimes observed. Hence, the fabrication yield for gaps of <10 nm was reduced to ~50 %. Another example of a nanogap electrode fabricated in the Au/Ti double-layered sample is shown in Fig. 9. The current trace observed for this fabrication is shown in Fig. 7 by the bold line. From the comparison of Figs. 8 and 9, it can also be seen that the effect of the FIB tail was clearly extended and the width and thickness of the nanowire were considerably reduced in this sample. The enhancement of the narrowing and thinning of a nanowire caused by the FIB blur presumably prevents the reproducible fabrication of narrow nanogaps.

Figure 10 shows SEM images of a nanogap electrode fabricated in the Ti/Au/Ti triple-layered sample. It is seen in Fig. 10a that an ~50-nm wide nanowire with an ~3-nm wide nanogap was formed. Although both edges of the nanowire were etched because of the FIB tail, no apparent narrowing and thinning of the nanowire were observed in most electrodes fabricated in Ti/Au/Ti samples. These results indicate that the excess etching due to the FIB blur at the edges of Au/Ti nanowires is suppressed by the top Ti film. In the Ti/Au/Ti samples, gaps of ~3–6 nm were reproducibly fabricated. The fabrication yield increased to ~90 %. By using the Ti protective layer, the controllability of the nanogap fabrication steps was considerably improved.

Gap Resistance

To measure the electrical properties of single molecules, an electrical insulating resistance of a nanogap electrode higher than 1 GΩ is desirable. We measured the electrical resistances of the fabricated nanogap electrodes. Figure 11a shows a histogram of the resistances of 14 nanogap electrodes fabricated in the Au/Ti samples. It was found that the resistances were scattered. The highest resistance observed in the electrodes was ~80 GΩ, while most of the electrodes showed resistances higher than a few GΩ. Figure 11b shows a histogram of the resistances of 19 nanogap electrodes fabricated in the Ti/Au/Ti samples. Most of the electrodes exhibited resistances of 10 G to 1 TΩ, which were more than one order of magnitude higher than those fabricated in the Au/Ti samples. The higher resistances of the Ti/Au/Ti nanogap electrodes were most likely caused by the decrease in the influence of the redeposition of the sputtered Au ions on the SiO2 surface around the tips of the electrodes during the nanogap formation step. This is because a higher dose was required to form nanogaps in the triple layers. At the higher dose, a deeper groove was formed around the tips of the electrodes, and the redeposition around this groove region was reduced as a result of the shadowing effect.