We experimentally show the detailed electrical transport characteristics of an SnO₂ nanomaterial with three-branch junction morphology, termed the nanowire T-junction. By contacting the three branches of a T-junction as the gate, source and drain, some control and tunability has been achieved over the output current, without the need for an external gate, analogous to the operation of a field effect transistor. Our results suggest that SnO₂ T-junction nanostructures represent novel current splitters with potential electronic applications. The different resistivities for each branch across the junction can be mainly attributed to differences of the defects and dislocations at the junction. The gating effects are directly related to the resistances of each branch and are electric field dependent. These features mean that SnO₂ T-junctions may be useful in nanoelectronic devices.

Being an important n-type semiconductor with a wide band gap, SnO₂ possesses many unique electrical properties which are useful for sensing applications such as, a remarkable receptivity variation in gaseous environments, low resistivity, and excellent chemical stability¹. There has been an explosion of interest in electronic nanodevices based on SnO₂ nanomaterials, including field effect transistors (FET)², field emissions³, and gas sensors^{[3], [4], [5] and [6]}. However, most previous reports on electronically addressed nanodevices to date have focused on using traditional materials, such as SnO₂ particle films⁷, nanowire frameworks^{[6] and [8]}, individual SnO₂ nanowires^{[1], [9], [10], [11] and [12]}, etc. Instead of a single-wire SnO₂ nanodevice, we experimentally demonstrate the detailed electrical transport characteristics of a three-branch junction SnO₂ nanomaterial, termed the nanowire T-junction, with tungsten (W) tips directly in contact with each probing pad. By contacting the three terminals of a T-junction as the gate, source and drain, some control and tunability has been achieved over the output current, without the need for an external gate, analogous to the operation of a field effect transistor. We believe that such an SnO₂ T-junction with three-terminals has great potential to be used in constructing novel nanoelectronic sensing devices.

In the current study the SnO₂ T-junctions were synthesized by a thermal evaporation method (see Supplementary information for details). Collected SnO₂ nanostructures were ultrasonicated in ethanol solvent, and then dispersed on a silicon wafer for characterization. Fig. 1a provides a closer look at a typical sample using scanning electron microscopy (SEM). It can easily be seen that the branch is perpendicular to the stem, forming a T-junction with three terminals. Using the geometrical shape offered by the T-junction, where two terminals can be regarded as source and drain and the other as gate, the electrical transport characteristics can be tuned, as will be considered in more detail in the following section. Patterns recorded using x-ray powder diffraction (Fig. 1b) suggest that all diffraction peaks of the as-prepared product could be indexed using the tetragonal rutile structure of SnO₂ (a = b = 0.4738 nm and c = 0.3187 nm, JCPDS card. 41-1445). No other crystalline forms, such as Sn or other tin oxides, were detected. Fig. 1c is the corresponding transmission electron microscopy (TEM) image. The “T” shaped morphology is further highlighted. Three selective area electron diffraction (SAED) patterns corresponding to the stem, junction, and branch are shown in Figs. 1d, 1e and 1f, respectively. The results indicate that the SnO₂ T-junction has the same rutile crystalline lattice and a rectangular cross-section enclosed by ±(010) and ±(101¯) facet planes. These crystallographic configurations of synthesized nanowires are similar to those previously observed^{[13] and [14]}. The growth directions of the stem and branch are parallel to the [101] and [010] crystal direction, respectively. High resolution TEM (HRTEM) images provide consistent results with the SAED patterns and confirm the tetragonal rutile structure for T-shaped SnO₂ nanowires, as shown in Figs. 1g, 1h and 1i, respectively. The interplanar d-spacing marked in the HRTEM images is about 0.477 nm, which can be indexed to the (010) lattice plane.

The SnO₂ T-junction nanodevice was fabricated using the photolithographic technique, four-probe nanomanipulation technique and a focused ion beam system. The T-junction nanodevice was formed on a p-doped silicon-on-insulator (SOI) wafer. A 15 nm layer of titanium was grown on the active silicon area to form the adhesion layer, followed by 100 nm of gold to form the metallic electrodes. An individual SnO₂ T-junction was selected and transferred to the central of Au/Ti electrodes (Figs. 2a and 2b). Platinum microleads were deposited using a focused ion beam (FIB) system onto the metallic electrodes and T-junction (Fig. 2c). The nanodevice was electrically tested using the metal pads as contacts (Fig. 2d). (For further details about the fabrication process, see Supplementary information)

Prior to measurements of the gating effect of the SnO₂ T-junction, I−V characteristics of this T-junction nanodevice across different terminals were first investigated (Fig. 3). It is clear that the device exhibits good linear behavior at the bias voltage ranging from −1 to 1 V, which proves that there is a good ohmic contact between the SnO₂ nanowire and platinum microleads. Based on the I−V characteristics, the resistances of each branch of the T-junction have been calculated. For the oa, ob and oc branches, it is about 2.875×10⁷, 17.445×10⁷ and 6.715×10⁷ Ω, respectively. Combined with geometric parameters (length and diameter) of each branch, their resistivity (ρ_oa, ρ_oc and ρ_ob) has also been obtained. It is about 3.85 Ω·cm for the oa branch, 3.58 Ω·cm for the oc branch and 32.52 Ω·cm for the ob branch, which is larger than 0.941 Ω·cm for individual SnO₂ nanowires¹⁵. In addition, it can be found that ρ_oa of the oa branch is approximately equal to ρ_oc of the oc branch. However, ρ_ob of the ob branch is about 89 times of ρ_oa or ρ_oc. The different resistivities for each branch across the junction could be mainly attributed to the difference of the defects and dislocations at the junction. This contribution was further evidenced by the resistivity of the linear part of the oc branch, without the junction. The result was found to be smaller than that for the oa branch and the oc branch, which is similar to the value previously reported. However, it is also important to consider the effect of the anisotropy of crystalline materials on their electrical behavior. The anisotropic resistivity of various single crystal materials has been reported previously^{[16] and [17]}. Again, for the directions perpendicular and parallel to the tetragonal c-axis of single crystal SnO₂, the dielectric functions are also different and anisotropic in theoretical calculations¹⁸. Therefore, anisotropic behavior might also be a factor contributing to the different resistivities of the stem of oa and oc along the [101] direction and the branch of ob along the [010] direction for a T-shaped SnO₂ nanowire.

Analysis of how the source-drain (I_d) and gate current (I_g) varies with the gate (V_g) and source-drain voltage (V_d) in the T-junction is shown in Fig. 4. Analogous to the operation of the FET, three terminals of T-junction SnO₂ can be used as the gate, source and drain. In this experiment, b is the gate (G) while a and c are the drain (D) and source (S), respectively (Fig. 4a). Fig. 4b presents the corresponding equivalent circuit. In terms of the Kirchhoff's Voltage/Current Law, the current I_g and I_d for the equivalent circuit can be expressed by the following equations:

(1)

(2) where R_s, R_g and R_d are the resistances of each branch. According to the expression, the source-drain current I_d increases linearly while the gate current I_g decreases with increasing source-drain voltage, respectively, at a constant gate voltage V_g. Figs. 4c and d experimentally show how the source-drain (I_d) and gate current depends on the source-drain voltage in the same nanodevice as in Fig. 2d. I_d increases linearly when the source-drain voltage increases (Fig. 4c), whereas I_g decreases linearly (Fig. 4d). The gate voltage dependence on the source-drain and gate current is presented in Figs. 4e and f. In contrast to the behavior seen in Figs. 4c and d, as the gate voltage is increased, the source-drain current decreases (Fig. 4e); and gate current increases (Fig. 4f). This is consistent with that estimated by the above equations.

It is clear from these electrical characteristics that I_d and I_g can be modulated by the source-drain or gate voltage, despite this the operation mode of the device is quite different from that of a conventional FET that is controlled by the carrier density and the thickness of the channel¹⁹. The nature of the T-junction means that it possible for the junction to act as a current splitter that can be controlled by the gate or source-drain voltage. Again, the gate is not a perfect insulating gate. A leakage current would be diverted from one path to another when a gate voltage is applied, resulting in a shift in current. This is similar to that seen in the three-way electrical gating effect of Y-shaped carbon nanotubes^{[20], [21] and [22]}, Co-doped Y-shaped ZnO²³ and the ZnO tetrapod^{[19] and [24]}.

Fig. 5 shows the I−V characteristics of the SnO₂ T-junction using different terminals as the gate, source and drain, under different source-drain voltages. The schematic measurement approaches are presented in the inset of Figs. 5a, c and e. It is noted that the source-drain current is also observed to linearly increase (Figs. 5a, c and e) and the gate current is also observed to decrease (Figs. 5b, d, and f) when the source-drain voltage is increased. The main difference is the slopes of the I_d−V_d and I_g−V_d curves. Based on the current proposed equations ((1) and (2)), it is believed that the resulting slope differences are due to the resistances of each branch, which could be either “(R_s + R_g)/R” (i.e., I_d−V_d curve) or “−R_s/R” (i.e., I_g−V_d curve). This suggests that the gating effect behavior of T-junction SnO₂ nanostructures is directly related to the resistance of each branch, which is determined by the resistivity and geometry of each branch. Further, the differences of the proportional displacement of the I_d−V_d and I_g−V_d curves are also seen with increasing V_g. This is because the proportional displacement of the I_d−V_d and I_g−V_d curves depends on −R_s/R and (R_s + R_d)/R, at a given V_g, respectively.

To determine whether the I_d−V_g and I_g−V_g curves behavior is similar to that observed in the above experiments, I−V characteristics were measured under different source-drain voltages and the results are presented in Fig. 6. As expected, I_d decreases and I_g increases linearly with the increase of V_g, when the drain voltage V_d is fixed, respectively. The similar behavior was observed using the a terminal as the gate, and the I−V curves are shown in [Fig. S1] and [Fig. S2] (see Supplementary information). So far, it is necessary to point out that such a T-junction nanodevice might be more advanced than those double-terminal sensors if employed as a sensor. By making use of the current splitter feature, it could be designed as a multiterminal sensor, i.e., a sensor which can give multiple responses to a single signal at the same time. Further research on the sensing applications of SnO₂ T-junction nanodevices would be of value and is in hand.