Giant Room-Temperature Piezoresistance in a Metal-Silicon Hybrid Structure

Metal-semiconductor hybrids are artificially created structures presenting novel properties not exhibited by either of the component materials alone. Here we present a giant piezoresistance effect in a hybrid formed from silicon and aluminum. The maximum piezoresistive gage factor of 843, measured at room temperature, compares with a gage factor of $-93$ measured in the bulk homogeneous silicon. This piezoresistance boost is not due to the silicon-aluminum interface, but results from a stress induced anisotropy in the silicon conductivity that acts to switch current away from the highly conductive aluminum for uniaxial tensile strains. Its magnitude is shown, via the calculation of hybrid resistivity weighting functions, to depend only on the geometrical arrangement of the component parts of the hybrid.

Figure 1

(a) Schematic diagram of the aluminum-silicon hybrid. Stress is applied along the [110] crystal direction, perpendicular to the aluminum-silicon interface for a number of devices corresponding to different values of $b$. Photographs of three different geometries with (b) $b=2\mu \mathrm{m}$, (c) $b=5\mu \mathrm{m}$, and (d) $b=20\mu \mathrm{m}$. In each case the white scale bar is $20\mu \mathrm{m}$ long.

Figure 2

(a) Dependence of the zero strain resistance on $b$. As $b$ is reduced, more current passes through the metal shunt thereby reducing the resistance. (b) Change in resistance, $\Delta R$, under strain for a number of different device geometries. Depending on the device geometry $\Delta R$ can be large and positive (e.g., $b=8\mu \mathrm{m}$), close to zero ($b=2\mu \mathrm{m}$), or large and negative (shuntless control device).

Figure 3

Measured GF versus $b$, filled circles. A peak of 843 is observed at $b=5\mu \mathrm{m}$. This compares with a measured GF in the homogeneous silicon of $-93$ ($b=20\mu \mathrm{m}$ shuntless control device), open circle. The GF calculated for the homogenous silicon agrees with the measured value, and the dotted curve is the GF calculated for the hybrids using only the nonadjustable parameters ${\sigma }_0$, ${\sigma }_{0,Al}$, ${\pi }_{11}$, ${\pi }_{12}$, and ${\pi }_{44}$.

Figure 4

Finite element calculation of the HRWF for three geometries, (a) $b=2\mu \mathrm{m}$, (b) $5\mu \mathrm{m}$, and (c) $20\mu \mathrm{m}$. To better view the interface, only the first $5\mu \mathrm{m}$ of the metal above the interface (indicated by the black arrows) are shown. In each case, white (gray/black) areas contribute strongly (weakly) to the resistance. For $b=2\mu \mathrm{m}$, the metal strongly contributes to the resistance since it is close to the external contacts. For $b=5\mu \mathrm{m}$ and $b=20\mu \mathrm{m}$ the metal no longer contributes, and in the latter case, the semiconductor dominates the resistance. The interfacial layer contributes the most to the resistance at $b=5\mu \mathrm{m}$ (white shading) as confirmed by the integral $I$ of the HRWF line scan at the interface (d). At this value of $b$, a stress induced redirection of the current in the silicon will most effectively switch current from the metal to the semiconductor resulting in the largest possible piezoresistance.