Enhanced Mobility and Stability in Solution-Processed Mo-Pr Co-Doped In2O3 TFTs Guided by Machine Learning Optimization
Weixin Cheng , Yuexin Yang , Han Li , Xiaoqin Wei , Honglong Ning , Guoping Su , Shaojie Jin , Chenbo Min , Rihui Yao , Junbiao Peng
Electron ›› 2025, Vol. 3 ›› Issue (4) : e70020
In2O3-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 In2O3, fabricating InPrMoO TFTs. Mo doping is believed to positively affect In2O3 through reducing porosity and defects. Pr doping has been proposed as a potential strategy to enhance the NBIS stability of In2O3. 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 (μsat = 12.2 cm2/V·s, Vth = 1.6 V, Ion/Ioff = 2.17 × 106, and SS = 0.47 V/dec) and the minimal ΔVth 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 cm2/V·s (10.6% improvement), an on/off current ratio of 3.82 × 106, and an SS of 0.40 V/dec, demonstrating superior efficiency over conventional trial-and-error methods.
high NBIS stability / machine learning / metal oxide / molybdenum doping / oxygen vacancy suppressing
| [1] |
|
| [2] |
|
| [3] |
|
| [4] |
|
| [5] |
|
| [6] |
|
| [7] |
|
| [8] |
|
| [9] |
|
| [10] |
|
| [11] |
|
| [12] |
|
| [13] |
|
| [14] |
|
| [15] |
|
| [16] |
|
| [17] |
|
| [18] |
|
| [19] |
|
| [20] |
|
| [21] |
|
| [22] |
|
| [23] |
|
| [24] |
|
| [25] |
|
| [26] |
|
| [27] |
|
| [28] |
|
| [29] |
|
| [30] |
|
| [31] |
|
| [32] |
|
| [33] |
|
| [34] |
|
| [35] |
|
| [36] |
|
| [37] |
|
| [38] |
|
| [39] |
|
| [40] |
|
| [41] |
|
| [42] |
|
| [43] |
|
| [44] |
|
| [45] |
|
| [46] |
|
| [47] |
|
| [48] |
|
| [49] |
|
| [50] |
|
| [51] |
|
| [52] |
|
| [53] |
|
| [54] |
|
2025 The Author(s). Electron published by Harbin Institute of Technology and John Wiley & Sons Australia, Ltd.
/
| 〈 |
|
〉 |