Experimental investigation of the surface acoustic wave electron capture mechanism

We have investigated surface acoustic wave charge transport through a one-dimensional channel, where the form of the potential is defined by two split gates placed in series. It is found that, when the second gate is set to shut off the acoustoelectric current, sweeping the first gate to negative voltages recovers the current and a set of current-quantization plateaux are observed. At a certain negative gate voltage, a sudden transition occurs and the current starts to decrease to zero, again showing plateaux. Calculation of the channel potential shows that this counterintuitive behavior is caused by the switching of maximum potential slope between two different locations that give different current dependencies. This result indicates that in the high transducer-power regime, the charge transport by surface acoustic wave is determined by the potential gradients.

Figure 1

A schematic diagram of the sample. The interdigitated transducer on the left produces a SAW running towards the mesa on the right (shown in gray) containing a 2DEG. The lithographic length of the channel defined by gates G1, G2, and G3 is approximately $1.7\mu \mathrm{m}$. The current through the channel is measured from the ohmic contacts shown as crossed boxes.

Figure 2

Acoustoelectric current, $I$, as a function of the voltage on G1, $V_{\mathrm{G}1}$. The voltage on G2, $V_{\mathrm{G}2}$, is decremented from $0\mathrm{V}$ (on the leftmost curve) to $-1.4\mathrm{V}$ by $0.05\mathrm{V}$ step. The inset shows the $I\text{−}{V}_{\mathrm{G}1}$ curve when $V_{\mathrm{G}2}=0\mathrm{V}$. The power on the transducer is $P=13.5\mathrm{dBm}$.

Figure 3

(a) Transducer power dependence of $I\text{−}{V}_{\mathrm{G}1}$ curve. The output power of the signal generator is incremented from $9\mathrm{dBm}$ (bottom curve) to $13.5\mathrm{dBm}$ (top curve) by $0.5\mathrm{dBm}$, with the voltages on G2 and G3 fixed at $-2.1\mathrm{V}$ and $-1.35\mathrm{V}$, respectively. (b) $V_{\mathrm{G}2}$ dependence of $I\text{−}{V}_{\mathrm{G}1}$ curve. $V_{\mathrm{G}2}$ is decremented from $-2\mathrm{V}$ (top curve) to $-3\mathrm{V}$ (bottom curve) by $0.1\mathrm{V}$ steps. The transducer power is set at $13\mathrm{dBm}$, and the voltage on G3 is $-1.35\mathrm{V}$.

Figure 4

(a)–(c) Three cases of the potential energy at the center of the channel ($450\mathrm{nm}$ away from G3) in the SAW traveling direction $x$: (a) $V_{\mathrm{G}1}=0\mathrm{V}$, (b) $V_{\mathrm{G}1}=-0.4\mathrm{V}$, and (c) $V_{\mathrm{G}1}=-0.65\mathrm{V}$. $V_{\mathrm{G}2}=-1.2\mathrm{V}$ for all the cases. Solid lines are the potential from the three-dimensional solution of Laplace’s equation, and their derivatives (i.e., proportional to the electric field) are plotted with the dotted line. The steepest part of the potential is marked by “×.” The SAW travels from left to right. (d) The experimental curve in Fig. 2 with $V_{\mathrm{G}2}=-1.2\mathrm{V}$. Points “$A$,” “$B$,” and “$C$” mark the voltages corresponding to (a), (b), and (c), respectively.

Figure 5

(a) The positions in $V_{\mathrm{G}1}$ of the current peak (crosses), the pinch off on the left of the peak (circles), and that on the right of the peak (triangles) of the data shown in Fig. 3a. The solid line is the theoretical position of the current peak. (b) SAW amplitude (peak to peak) derived from the calculated potential at the pinch off (see text for detail). (c) and (d) Similar plots for the data in Fig. 3b.

Figure 6

Grayscale plot of the power dependence of the acoustoelectric current. Here, the transducer power $P$ is swept, instead of gate voltage, to average out sample heating. $V_{\mathrm{G}1}$ is incremented by $5\mathrm{mV}$. The derivative of the curve is plotted in grayscale with white representing plateaux. Up to five current-quantization plateaux can be seen. For $P\lesssim 7\mathrm{dBm}$, random-telegraph-signal noise becomes dominant. The data were not taken in the blank region on the top-right and bottom-left corners.