Figure 1

Schematic of the SBMOSFET (not drawn to scale). The source and drain tunneling contacts consist of the metal silicide PtSi. The gate is formed from a conventional metal-oxide-semiconductor (MOS) capacitor. The substrate is lightly doped ($5\times {10}^{15}$ ${\mathrm{cm}}^{-3}$) with boron. The device operates in accumulation and transport is dominated by holes. The device dimensions used in this paper are width/length $=$ 3 μm/0.5 μm.Reuse & Permissions

Figure 2

(a) Differential conductance $\frac{\partial I_{\mathit{ds}}}{\partial V_{\mathit{ds}}}$ vs gate voltage |$V_g$| at $V_{\mathit{sd}}=0$ V and magnetic fields $B$ $=$ 2.7 and 4.7 T at 50 mK. There are eight antiresonances, labeled a–h, and two resonant peaks. These lineshapes are shifted in energy (either to higher or lower |$V_g$|) with $B$, attributed mostly to the Zeeman effect. Although the magnitude of the differential conductance decreases slightly with field, there is not a substantial change in the lineshape. The inset shows a more detailed plot of dip e. (b) Three-dimensional plot of the $\frac{\partial I_{\mathit{ds}}}{\partial V_{\mathit{ds}}}$ showing the displacement of the lineshapes in (a) as a function of magnetic field. Here we use a small range of |$V_g$| for clarity. The dotted lines are the linear fits obtained by extracting the local minimum (maximum) near the lineshape and the resultant slopes are reported in Table .Reuse & Permissions

Figure 3

Changes in the lineshape at different $V_{\mathit{sd}}$ of a resonant peak (a) and (c) and dip c (b) and (d). In both (a) and (c), the $\frac{\partial I_{\mathit{ds}}}{\partial V_{\mathit{ds}}}$ of the $V_{\mathit{ds}}=0$ V curve is significantly suppressed compared to higher bias. This suppression of the background current is due to electron-electron interactions. In (b) and (d), the lineshapes are scaled, respectively, by the $\frac{\partial I_{\mathit{ds}}}{\partial V_{\mathit{ds}}}$ at $V_{\mathit{ds}}=0$ V, $V_g=-1.951$ V, and $V_{\mathit{ds}}=0\hspace{0.5em}V$, $V_g=-1.913$ V. However, as shown in (b) and (d), the lineshapes do not scale in the same way as the background. All curves are shown at 50 mK and at 4.7 T.Reuse & Permissions

Figure 4

For each of the antiresonances in Fig. 2(a) at 4.7 T, we calculate the percentage the differential conductance dips below the accompanying shoulder. The frequency and percent change of the differential conductance decreases as $\left|V_g\right|$ increases. This provides evidence that the background current in the channel is increasingly dominated by incoherent direct tunneling through the barrier rather than ‘‘hot spots’’ resulting from resonant levels.Reuse & Permissions

Figure 5

(a) We fit the Fano lineshape [Eq. (2)] to dip c in Fig. 3 at 4.7 T and 50 mK. The value of the $V_{\mathit{ds}}$ from bottom to top is 0, ±0.1, ±0.2, ±0.3, and ±0.4 mV. Note that the curves are NOT offset.The $V_{\mathit{ds}}=0$ V curve is plotted in black with the fit indicated in gray, the positive $V_{\mathit{ds}}$ curves are plotted dark gray with fits in light gray and the negative $V_{\mathit{ds}}$ curves are plotted in light gray with fits in darkgray. (b) Plot of the asymmetry factor $q$ as a function of bias voltage $V_{\mathit{ds}}$. At $V_{\mathit{ds}}=0$ V, we observe an antiresonance, corresponding to an asymmetry parameter $q$ close to 0, and indicative of destructive interference. For $V_{\mathit{ds}}>0$ V, we find positive $q$ values and, for $V_{\mathit{ds}}<0$ V, negative $q$ values, as might be expected if the interference is changed by a 180° phase shift. As $\left|V_{\mathit{ds}}\right|$ is increased, the resonant tunneling component becomes more important and the interference becomes increasingly constructive, indicated by $q$ $=$ 1. Note that the standard deviation of the $q$-parameter fit is indicated by error bars, which demonstrate the robustness of the changes with |$V_{\mathit{ds}}$|.Reuse & Permissions