Author Affiliations
1School of Microelectronics, Shanghai University, Shanghai, China2Laboratory of Thin Film Optics, Key Laboratory of Materials for High Power Laser, Shanghai Institute of Optics and Fine Mechanics, Chinese Academy of Sciences, Shanghai, China3Center of Materials Science and Optoelectronics Engineering, University of Chinese Academy of Sciences, Beijing, China4Hangzhou Institute for Advanced Study, University of Chinese Academy of Sciences, Hangzhou, Chinashow less
Fig. 1. THL-BPNN model with all neurons in adjacent layers connected, where x = [x1; x2], y1 and hij represent the input, output and intermediate processing signals, respectively.
Fig. 2. Accuracy of BPNNs with one to four hidden layers based on (a) the refractive index (at 355 nm), (b) layer thickness and (c) O/Hf ratio of PEALD-HfO2. The four columns in each subgraph represent the R2 values of the model in the training and validation sets and the RMSE values in the training and validation sets, respectively. The table indicates the number of neurons in each hidden layer of each model.
Fig. 3. Measured and predicted (a)–(c) refractive index, (d)–(f) layer thickness and (g)–(i) O/Hf ratio of HfO2 thin films. The data in the left-hand, middle and right-hand columns are predicted by the LR model, SVR model and THL-BPNN model, respectively. The blue line (with a slope of 1) serves as a guideline for perfect prediction.
Fig. 4. Correlations between properties of HfO2 thin films used in this section. Blue indicates a negative correlation, whereas red indicates a positive correlation. Darker colors and larger circles indicate higher correlations. The numbers inside the circles indicate the corresponding correlation coefficients of the two features.
Fig. 5. Comparison of measured and predicted LIDT values on the (a) training set and (b) validation set.
Fig. 6. Comparison of measured and predicted values of (a) the refractive index (at 355 nm), (b) the layer thickness and (c) the O/Si ratio for SiO2 thin films in the validation set.
|
---|
| HfO2 thin films | SiO2 thin films |
---|
|
---|
| Variables | Range | Variables | Range |
---|
Input | Annealing atmosphere* | 0–3 | Deposition temperature (°C) | 50–200 | | Annealing temperature (°C) | 0–800 | Precursor exposure time (s) | 0.1–0.7 | | Output | Refractive index(at 355 nm) | 1.83–2.24 | Refractive index(at 355 nm) | 1.48–1.49 | | Thickness (nm) | 34.7–50.3 | Thickness (nm) | 69.0–88.1 | | O/Hf ratio | 1.80–2.04 | O/Si ratio | 1.94–2.01 |
|
Table 1. Datasets for property prediction of HfO2 and SiO2 thin films.
|
---|
| | Range |
---|
|
---|
| Variables | HfO2 | SiO2 |
---|
| | thin films | thin films |
---|
Input | Type* | 1 | 2 | | Total impurity content | 5.4–13.5 | 0.6–1.1 | | (%, atomic fraction) | | | | Absorption (ppm) | 211–10892 | 3.8–5.8 | | Stoichiometric ratio | 1.81–2.06 | 1.94–2.01 | Output | LIDT (J/cm2) | 1.2–6.3 | 22.0–39.4 |
|
Table 2. Datasets for LIDT prediction of HfO2 and SiO2 thin films.
|
---|
| Refractive index | Layer thickness | O/Hf ratio |
---|
|
---|
| Training data | Validation data | Training data | Validation data | Training data | Validation data |
---|
|
---|
| R2 | RMSE | R2 | RMSE | R2 | RMSE | R2 | RMSE | R2 | RMSE | R2 | RMSE |
---|
LR | 0.72 | 0.06 | 0.66 | 0.08 | 0.74 | 2.24 | 0.48 | 2.88 | 0.43 | 0.08 | 0.48 | 0.08 | SVR | 0.71 | 0.06 | 0.52 | 0.10 | 0.75 | 2.22 | 0.56 | 2.64 | 0.84 | 0.04 | 0.74 | 0.05 | THL-BPNN | 0.99 | 0.01 | 0.99 | 0.01 | 0.94 | 1.08 | 0.91 | 1.18 | 0.94 | 0.03 | 0.90 | 0.03 |
|
Table 3. Evaluation of the LR, SVR and THL-BPNN models.
|
---|
| Training data | Validation data |
---|
|
---|
| R2 | RMSE | AA | RMSE |
---|
Refractive index | 0.99 | 0.00 | 1.00 | 0.00 | Layer thickness | 0.99 | 0.43 | 0.98 | 1.72 | O/Si ratio | 0.81 | 0.01 | 0.99 | 0.03 |
|
Table 4. Evaluation of the THL-BPNN model for SiO2 thin film properties.