Obtaining high-quality materials, based on nanocrystals, at low temperatures is one of the current challenges for opening new paths in improving and developing functional devices in nanoscale electronics and optoelectronics. Here we report a detailed investigation of the optimization of parameters for the in situ synthesis of thin films with high Ge content (50 %) into SiO2. Crystalline Ge nanoparticles were directly formed during co-deposition of SiO2 and Ge on substrates at 300, 400 and 500 °C. Using this approach, effects related to Ge–Ge spacing are emphasized through a significant improvement of the spatial distribution of the Ge nanoparticles and by avoiding multi-step fabrication processes or Ge loss. The influence of the preparation conditions on structural, electrical and optical properties of the fabricated nanostructures was studied by X-ray diffraction, transmission electron microscopy, electrical measurements in dark or under illumination and response time investigations. Finally, we demonstrate the feasibility of the procedure by the means of an Al/n-Si/Ge:SiO2/ITO photodetector test structure. The structures, investigated at room temperature, show superior performance, high photoresponse gain, high responsivity (about 7 AW−1), fast response time (0.5 µs at 4 kHz) and great optoelectronic conversion efficiency of 900% in a wide operation bandwidth, from 450 to 1300 nm. The obtained photoresponse gain and the spectral width are attributed mainly to the high Ge content packed into a SiO2 matrix showing the direct connection between synthesis and optical properties of the tested nanostructures. Our deposition approach put in evidence the great potential of Ge nanoparticles embedded in a SiO2 matrix for hybrid integration, as they may be employed in structures and devices individually or with other materials, hence the possibility of fabricating various heterojunctions on Si, glass or flexible substrates for future development of Si-based integrated optoelectronics.
Freehand to Digital Circuit Reconstruction Using HOG and SVM
