兴奋剂
材料科学
理论(学习稳定性)
X射线光电子能谱
无定形固体
光电子学
电子迁移率
机器学习
多孔性
人工智能
人工神经网络
薄膜晶体管
电子工程
探测器
光谱学
氧气
分析化学(期刊)
反向传播
半导体器件
作者
Weixin Cheng,Yuexin Yang,Han Li,Xiao‐Qin Wei,Honglong Ning,Guoping Su,Shaojie Jin,Chenbo Min,Rihui Yao,Junbiao Peng
摘要
ABSTRACT In 2 O 3 ‐based TFTs have garnered widespread attention due to their higher mobilities than amorphous silicon. Previous studies have indicated that rare earth doping can enhance the NBIS stability of TFTs, but this often results in a decrease in mobility. To improve the mobility of TFTs while maintaining stability, we incorporated Mo and Pr into In 2 O 3 , fabricating InPrMoO TFTs. Mo doping is believed to positively affect In 2 O 3 through reducing porosity and defects. Pr doping has been proposed as a potential strategy to enhance the NBIS stability of In 2 O 3 . A nondestructive μPCD detector was employed to characterize the local defect states of the film. X‐ray photoelectron spectroscopy data demonstrate that the InPrMoO film with 0.8 mol% Mo doping has the lowest concentration of oxygen vacancies (Vo). TFTs fabricated using the InPrMoO film doped with an optimized concentration of 0.8 mol% Mo exhibit superior electrical properties ( μ sat = 12.2 cm 2 /V·s, V th = 1.6 V, I on / I off = 2.17 × 10 6 , and SS = 0.47 V/dec) and the minimal Δ V th under NBS/PBS/NBIS = −0.65 V/0.79 V/−0.70 V. The synergistic effect of Mo and Pr doping has led to enhanced film uniformity and density, consequently improving the mobility and stability of the TFTs. To tackle the challenge of predicting optimal process parameters, a multiobjective prediction model integrating physical models and machine learning was developed. The predicted optimal parameters (0.78 mol% Mo doping, 381°C annealing) were experimentally verified, yielding < 5% relative error in most film properties. The prepared TFT exhibits a mobility of 13.5 cm 2 /V·s (10.6% improvement), an on/off current ratio of 3.82 × 10 6 , and an SS of 0.40 V/dec, demonstrating superior efficiency over conventional trial‐and‐error methods.
科研通智能强力驱动
Strongly Powered by AbleSci AI