Physics-Informed Neural Network for High Frequency Noise Performance in Quasi-Ballistic MOSFETs
Round 1
Reviewer 1 Report
Must verify the references, probable was omitted some references and level of non-referenced plagiarism high.
The idea of this paper excellent, modern and actual.
Good agreement of the prediction and the actual measurements.
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
- Literature review - is developed at a good level . The oldest entry is reference number 9 - Naveh, Y.; Likharev, K. K. Modeling of 10nm-scale ballistic MOSFET’s. IEEE Electron Device Lett. 2000, 21, 242-244 - however, this is an important item for the topic of the article. A lot of the latest literature on the topic in the article from the years 2000-2010.
- A very well-developed point number 2 of the article - 2. Theoretical Models. The author divided this issue into two sections and developed a good theoretical basis for these issues. These problems are shown by the formulas 1-18b and the drawing - Figure 1. Representation of energy barrier profile (a) along the y-axis in the source-drain
direction and (b) along the x-axis in the channel-gate direction. - In chapter 3, - 3. Artificial Neural Network Model the transistor model is shown in Fig. 3 - (Figure 2. High frequency equivalent circuit of a MOSFET including the neural-based drain, gate,
its correlation noise, and drain, gate shot noise model) - please use the appropriate resistor symbol - Rg. Please complete the descriptions of the individual elements of the equivalent transistor diagram below - Vgs, Cgd, ...., e.t.c. - Model verification, chapter no. worked out well. All the basic parameters of the transistor are shown in the graphs - fig. 4-8.
- The conclusions are only 5 sentences. Please refer in the conclusions to the difference between the measurements in the model in the graphs, for example 8 a - d. What can these slight discrepancies be caused by. (4 expanded sentences) What author of the paper plans further simulation or research.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
The author presents a physics-informed neural network (PINN) model to study nonlinear characteristics of high-frequency noise performance in quasi-ballistic MOSFETs by constructing a noise-equivalent circuit model which includes all the noise sources. The model exhibits a good capability for predicting the noise performance at high frequencies.
The manuscript is well organized and presents the results in a clear way. The graphs comparing neural networks’, experimental and numerical data support the effectiveness of the model.
As minor recommendations to the author, I would suggest being more accurate with definitions - $V_{gs}$ in Eq. (1b) is not defined, a reader has to assume that $I_d$ is a drain current. What is the reference for the first equations of the manuscript – is it [5]? Section 2 contains a lot of equations – the author should underline the most important ones and discuss them.
To summarize, this manuscript can be accepted for publication after minor revision.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 4 Report
Please see the attached file.
Comments for author File: Comments.pdf
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 4 Report
Good revision!
