材料科学
光子学
氮化物
光电子学
绝缘体(电)
铝
集成光学
光学
纳米技术
复合材料
物理
图层(电子)
作者
Redouane Amrar,A. Soltani,Guillaume Beaudin,Gabriel Droulers,Simon Loquai,Paul G. Charette
摘要
Aluminum nitride (AlN) holds significant potential for near- and mid-infrared integrated photonics, particularly in high-speed telecommunications and sensors. However, relatively high material absorption losses in sputtered AlN films limit performance. This study investigates continuous and cyclic annealing protocols designed to reduce material losses in the near-infrared regime for an AlN-on-insulator (AlNOI) integrated photonics platform fabricated by pulsed-DC magnetron sputtering on 8-inch Si wafers. The effects of annealing on AlN microstructure and residual stress were characterized using X-ray diffraction, micro-Raman and infrared spectroscopies. Best results were obtained with a 5-cycle 900 °C - 1350 °C thermal treatment, resulting in a rocking curve of 0.85°, a c-axis tilt angle of 0.7° relative to the surface normal, and an 85% relaxation of residual tensile stress compared to unannealed samples, demonstrating a highly textured structure. Material absorption losses were characterized using strip waveguides fabricated by e-beam lithography. By conditioning the AlNOI wafers with the 5-cycle profile prior to waveguide fabrication, material absorption losses were reduced by 63%, from 1.44 dB/cm to 0.54 dB/cm (scattering from vertical sidewall roughness in waveguides fabricated by e-beam lithography contributed additional losses of 0.13 dB/cm). Finally, with regard to electro-optic applications, the effect of the annealing protocols on the electrical properties of the films was characterized using vertical metal-insulator-semiconductor (MIS) and planar metal-semiconductor-metal (MSM) structures. By conditioning the AlNOI wafers with cyclic annealing prior to device fabrication, film resistivity and breakdown field increased exponentially with the number of cycles to 2 × 10 14 Ω·cm and 2.95 MV/cm after five cycles, respectively, while the leakage current decreased by 2 to 3 orders of magnitude depending on the applied voltage.
科研通智能强力驱动
Strongly Powered by AbleSci AI