High-Mobility, Low-Temperature, CMOS Back-End Compatible III-V Crystalline Material Growth

Abstract

In this work, we present a low temperature templated liquid phase (LT-TLP) growth process, that enables us to directly grow high optoelectronic quality single crystalline compound semiconductors on amorphous dielectrics at temperatures below 400 0C. Importantly, we show that when compared to other non-epitaxial growth techniques, this approach results in materials which exhibit 1-2 orders of magnitude improved mobility. This is attributed to the ability to grow single crystalline materials in desired geometries, removing the polycrystalline interfaces which dramatically degrade mobility of most crystalline materials. We demonstrate both InP and InAs, and interestingly observe a peak in quality when InP is grown at 300 oC, a temperature which is low enough to enable back-end-of-line growth on fully fabricated Si CMOS circuits. Using this low-temperature grown InP, we then develop a transistor fabrication process entirely carried out at 300 oC or below and demonstrate an indium phosphide nanoribbon field effect transistor with excellent on/off ratios, indicating low defect density in the material. Additionally, we demonstrate high electron mobility single crystal InAs mesas monolithically integrated on amorphous dielectric substrates at a growth temperature of 300°C. Critically, a room temperature mobility of ~5800 cm2/V-s was measured, the highest mobility reported for any thin-film semiconductor material system directly grown on a non-epitaxial substrate. Detailed modeling of the scattering mechanisms in the grown material indicates that the mobility is limited by surface roughness scattering, not the intrinsic material quality. We project that reducing the RMS surface roughness of the InAs from 1.8 nm to 1 nm would produce materials with room temperature mobilities of >10,000 cm2/V-s, and RMS roughness of 0.5 nm would result in mobility ~20,000 cm2/V-s, essentially identical to epitaxially grown materials. These results pave the way for growth of high-mobility materials directly onto the back-end of silicon CMOS wafers as well as other non-epitaxial substrates such as glass, and polymers for flexible electronics. Figure 1