钝化
材料科学
硅
原子层沉积
场效应
图层(电子)
电场
电压
电荷密度
分析化学(期刊)
光电子学
纳米技术
电气工程
化学
物理
量子力学
工程类
色谱法
作者
Hemangi Patel,Christian Reichel,Armin Richter,Paul Masuch,Jan Benick,Stefan W. Glunz
标识
DOI:10.1016/j.apsusc.2021.152175
摘要
Multilayer stacks in the sequence c-Si/(Al2O3/SiO2)n with n = 1 (single multilayer stacks) or n = 3 (triple multilayer stacks) were deposited at different temperatures by plasma-enhanced atomic layer deposition (PEALD) with varying Al2O3 thickness to investigate the formation of dipoles or trapped charges at the interfaces and their effect on the surface passivation quality. It was shown that the field-effect passivation deteriorated in multilayer stacks compared to an Al2O3 single layer as flat-band voltage (Vfb) and total amount of effective charges (Qtot) were lower than Al2O3 reference layer (Al2O3 single layer) (Vfb = 2.36 V and Qtot = −1.13 × 1013 cm−3). However, choosing the right process conditions, the chemical passivation is comparable to the Al2O3 reference layer. Multilayer stacks deposited at 250 °C with 3 nm of Al2O3 interlayers exhibited a superior chemical passivation with effective lifetimes (τeff) of 11 ms, characterized by a very low interface trap density (Dit), which was comparable to the Al2O3 reference layer. Furthermore, it was observed that field-effect passivation could be tuned by manipulating Qtot in the multilayer stacks by applying a voltage stress (Vstress) of varying magnitude, duration and polarity (voltage of particular magnitude and sign applied for a specific time) before the capacitance–voltage measurement. The charge dynamics demonstrated that Qtot shifted from their original values under the effect of Vstress which showed a high Vfb shift (∼6.5 V), providing a gateway in tailoring the field-effect passivation quality of multilayer stacks on c-Si. Nevertheless, bias-photoconductance measurements reveal a degradation of τeff after the application of high bias voltages which correlated with a strong increase of Dit after the Vstress application.
科研通智能强力驱动
Strongly Powered by AbleSci AI