The scaling down of semiconductor devices and an increase in their integration have been emphasized to improve their performance in various applications. Accordingly, the characterization of nanoscale electric structures is crucial for nanoscale fabrication processes. Although conventional optical microscopes have a spatial resolution limitation ascribed to the diffraction limit, scanning electron microscopes (SEM) have been extensively employed for measuring nanoscale surface topography [1–4]. However, because conductive surfaces are typically examined using SEM, the surfaces of nonconductive specimens require metal coatings before analysis [5–7]. Various nanoscale surfaces on specimens require nondestructive characterization, and the atomic force microscope (AFM) is a remarkable characterization tool for measuring the topography of these surfaces without damaging them [8–10]. Furthermore, this scanning probe microscope can be extended to analyze nanoscale topography exhibiting various electrical, thermal, and magnetic properties, which are generally measured using an electric force microscope (EFM), scanning thermal microscope, and magnetic force microscope, respectively [11–15]. In particular, EFM can be used to measure the work function of a specimen surface with a scanning Kelvin probe microscope and the conductance variation on the specimen surface under tip gating with a scanning gate microscope (SGM). Recently, because the conductance variation with potential on the nanoscale area of graphene has been measured by tuning the Fermi level [16,17], the SGM characterization of nanoscale graphene nanoribbons (GNR), which are fabricated from two-dimensional graphene, is important for understanding the electrical properties of onedimensional carbon networks [18,19].

In this study, we introduce SGM measurement for GNR. The SGM is employed to measure the conductance of nanoscale metal and semiconductor channels by applying a drain-source bias voltage (V_DS). Because the SGM locally applies an electric gate field from the conductive tip to the nanoscale surface region, the nanoscale conductance properties can be determined by applying V_DS. To characterize the local conductance properties of the GNR under V_DS by the SGM, the GNR is prepared via plasma etching, using hydrogen silsesquioxane (HSQ), which is an insulator, as a mask on single-layer graphene (SLG). The electrical properties of the GNR under the HSQ dielectric layer, determined by SGM measurements, are used to explain the variation in carrier density depending on the width difference of the GNR. These results clarify the width-dependent doping differences in nanoscale GNR structures and suggest that GNR structures can be predicted from electrical property measurements.

The conductance of semiconductors or metal channels encapsulated by a dielectric layer can be measured through SGM characterization. As illustrated in Fig. 1, when the semiconductor channel is measured using two-terminal characterization, the conductance is tuned by electric gating via the dielectric layer. For the SGM, partial gating can be applied using a conductive AFM tip. The following steps were used to obtain the SGM images: Initially, the topography was measured in non-contact mode with a cantilever dithering frequency of ~100 kHz. Subsequently, by employing a constant-height mode with a distance of ~30 nm between the tip and the sample surface, a partial gate voltage (V_tip) was applied via a conductive tip (Au-coated SGM tip) in addition to measuring the drain-source current (I_DS) in the GNR channel. An SGM image was generated by plotting the I_DS values of the GNR channel for each scanning area of the SGM tip. The charge distribution in the channel was characterized based on the conductance image data. In particular, nanoscale channels have been reported to exhibit SGM signal variation owing to charge disorders or dimensional fluctuations of the channel, which change the electric field density with the SGM tip, resulting in current flow variation in the channel [20].

Figure 1. Schematic of SGM characterization leading to partial tip gating (V_tip) on conducting (semiconductor or metal) channels under a dielectric layer, with the evaluation of conductance by two-terminal (V_DS−I_DS) measurement.

Here, we introduced a partial conductance variation in the GNR via SGM characterization. The GNR was fabricated from SLG on a SiO₂ substrate (thickness = ~280 nm) [Fig. 2(a)]. This SLG on SiO₂ was prepared via mechanical exfoliation and then brought into contact with Cr/Au (0.5/40 nm thickness) electrodes via normal e-beam lithography process. The nanoscale etch mask on the SLG was fabricated with a negative-tone e-beam resist (i.e., HSQ), and the unprotected SLG area was etched by O₂ plasma treatment. The HSQ mask on GNR was prepared with a width of ~100 nm and length of 1 μm [Figs. 2 and 3(a)] [21–24]. Figure 2(b) illustrates an AFM image of the GNR. Consequently, we observed SGM images that exhibited three different current variation areas in the GNR, marked as A, B, and C in the SGM images presented in Fig. 3(b). The topography and SGM images were measured using a commercial AFM (XE-100, Park Systems Corp.) in air at room temperature. Upon applying V_tip via the SGM tip, I_DS of GNR was measured at V_DS = 0.5 V, leading to an initial current level of 8.7 μA. Notably, this initial I_DS is influenced by dangling bonds on the SiO₂ substrate, and the current level can be improved using the ultraclean surface of a hexagonal boron nitride substrate [25,26].

Figure 2. (a) Schematic of GNR fabrication with HSQ mask by O₂ plasma treatment. (b) Topography image of GNR with Au electrodes (The width and length of HSQ on GNR are measured to be approximately 100 nm and 1 μm, respectively).

Figure 3. (a) Schematic of SGM process for measuring GNR under HSQ mask on SiO₂. (b) SGM images of GNR with different V_tip (−3, −1, 0, 1, and 3 V) exhibiting substantial variations in current, with positions marked by A, B, and C. The dashed lines correspond to the GNR boundary, as confirmed from the topography [Fig. 2(b)]. (c) Current profiles extracted along the GNR length direction of SGM images with different V_tip (−3, −1, 0, 1, and 3 V) in (b).

To study the conductance variation of the GNR with V_tip, we investigated SGM images of GNR for different values of V_tip (−3, −1, 0, 1, and 3 V) [Fig. 3(b)]. The decrease and increase in the current levels for positive (3 and 1 V) and negative (−3 and −1 V) V_tip were observed as compared with the current level for V_tip = 0 V, respectively. The origin of these areas with considerable variations in the current (i.e., areas A and C) could be attributed to a partially undesired width fluctuation in the GNR, leading to a partial conductance variation on the GNR, as reported previously [16–18]. Herein, we focus on the width fluctuation in the GNR; additional experimental measurements are necessary to confirm the origin of the areas with substantial current variations, which will be investigated more comprehensively in the future. Owing to the indistinguishable topography of the HSQ, we can consider that the permeation of O₂ plasma under HSQ randomly etches the carbon bonds at the edge of graphene, leading to a width fluctuation in the GNR under the HSQ mask [19,20]. This partially induced current difference in the GNR can be manifested by employing the electric field density difference for different widths at positions A, B, and C of the GNR [Fig. 3(c)]. Since the electric field density between the SGM tip and area A (i.e., narrowest width region) is higher than that for areas B and C, the Fermi level tuning range in area A by V_tip is larger than that in areas B and C. Furthermore, the variation in I_DS induced by V_tip values of −3 to 3 V for position B is smaller than that for positions A and C, indicating that the width of B is the widest in the GNR. The larger Fermi level tuning under the same V_tip in positions A and C than in position B of the GNR underneath the HSQ structure can be attributed to a narrower width region on the GNR, leading to a high electric field density between the tip and the sample, as reported previously [20].

Consequently, we infer that the width fluctuations in the GNR with changes in the I_DS are induced through partial gating in the GNR. This is useful for predicting the structure through nondestructive inspection in an environment where the structure of the sample cannot be identified by topographic measurements using AFM.

Herein, we introduced the SGM and the corresponding measurement process. The conductance on the GNR underneath the HSQ was investigated. In particular, the current level of the GNR varied with the application of a partial gate via the SGM tip. Based on previous results [20–22], we propose that partial width fluctuations in the GNR vary the current in the GNR under the HSQ due to V_tip gating. This finding can be applied to analyze the electrical properties of nanoelectronic structures. The proposed method of evaluating conductance using the SGM is expected to contribute markedly to the development and performance optimization of nanostructure processes.

This work was supported by the Research Fund of the Chungnam National University.

The authors declare no conflicts of interest