Reactions of methyl radicals with hydrogen bromide CH3 + HBr → CH4 + Br (1) and bromine atoms CH3 + Br → CH3Br (2) were studied using excimer laser photolysis−transient UV spectroscopy at 297 ± 3 K over the 1−100 bar buffer gas (He) pressure range. Methyl radicals were produced by 193 nm (ArF) laser photolysis of acetone, (CH3)2CO, and methyl bromide, CH3Br. Temporal profiles of methyl radicals were monitored by UV absorption at 216.51 nm (copper hollow cathode lamp with current boosting). The yield of acetyl radicals in photolysis of acetone at 193 nm was found to be less than 5% at 100 bar He based on the transient absorptions at 222.57 and 224.42 nm. The measured rate constants for reaction 1 are k 1 = (2.9 ± 0.7) × 10-12, (3.8 ± 1.5) × 10-12, and (3.4 ± 1.3) × 10-12 cm3 molecule-1 s-1 at the buffer gas (He) pressures of 1.05, 11.2, and 101 bar, respectively. The rate data obtained in this study confirmed high values of the previous (low pressure) measurements and ruled out the possibility of interference of excited species. The measured rate constant is independent of pressure within the experimental error. The rate constant of reaction of methyl radicals with bromine atoms (2) was determined relative to the rate constant of methyl radical self-reaction, CH3 + CH3 → C2H6 (3) in experiments with photolysis of CH3Br:  k 2/k 3 = 0.92 ± 0.32, 1.15 ± 0.30, and 1.65 ± 0.26 at 1.05, 11.2, and 101 bar He, respectively. On the basis of the literature data for reaction 3, this yields k 2 = (5.8 ± 2.2) × 10-11, (7.4 ± 2.2) × 10-11, (10.7 ± 2.3) × 10-11, and (11.9 ± 2.5) × 10-11 cm3 molecule-1 s-1 at 1.05, 11.2, 101 bar (He), and in the high-pressure limit, respectively.

No abstract

A machine learning (ML) model by combing two autoencoders and one linear regression model is proposed to avoid overfitting and to improve the accuracy of Technology Computer-Aided Design (TCAD)-augmented ML for semiconductor structural variation identification and inverse design, without using domain expertise. TCAD-augmented ML utilizes TCAD simulations to generate sufficient data for ML model development when experimental data are inadequate. The ML model can then be used to identify semiconductor structural variation for given experimental electrical measurements. In this study, the variation of layer thicknesses in the p-i-n diode is used as a demonstration. An ML model is developed to predict the diode layer thicknesses based on a given Current-Voltage (IV) curve. Although the variations of interest can be incorporated easily in TCAD simulations to generate ML training data, the TCAD-augmented ML model generally is overfitted and cannot predict the variations in experiment well due to hidden variables which also alters the IV curves. We show that by using an autoencoder, this problem can be solved. To verify the effectiveness, another set of TCAD simulation data is generated with hidden variables (dopant concentration variation) to emulate experimental data. Testing on the second set of data shows that the proposed model can avoid overfitting and has up to 15 times improvement in accuracy in thickness prediction. Moreover, this model is used successfully to perform inverse design and can capture an underlying physics that cannot be described by a simple physical parameter.

No abstract

No abstract