Effect of Positive Bias and Pressure on Plasma Flow Characteristics in a Chemical Vapor Deposition Chamber
Round 1
Reviewer 1 Report
This is a simulation work on a specific setup of PECVD with H2. The authors focus on the effect of bias voltage and investigate the 2D spatially resolved plasma parameters such as the electron temperature. However, there is no informative and valuable analysis in this work and therefore, I don’t see any substantial contribution to the CVD community. Also, the manuscript contains a result report only, without a discussion section. The authors entitle their conclusion to the discussion section. Detailed comments are the following.
First, it is very common for PECVD to use microwave plasma, while a bias voltage is applied to make the deposition. Playing with the bias voltage and pressure is not new to the CVD community. We expect to see the novelty in this work but everything we find in this work had been discussed for many years.
Second, the simulation is too simple compared with other works. The authors have H2 in the chamber, but even with such a single species, the plasma chemistry can still be very complex. So far, we only see three chemical reactions. It is fine to assume no photon reactions, but the authors need at least contain e + H => H+ + e; 2H => H2; and e + H => H* + e. Also, the authors need to contain the difference in the excited states of H* and H2*.
Third, there are many unclear descriptions in the manuscript. In Line 123 and Figure 3, the authors mention “plasma flow characteristics”. There is no such terminology in the plasma world. It is ok to introduce a new term, but the authors show no definition of it. What plasma species we are looking at in Figure 3? By increasing the bias voltage, we expect to see the velocity difference between electrons and ions. Where is the stagnation of the velocity field on the target surface? Where is the sputtering velocity (the reflection of particles) on the target surface? In Figure 5(b) the maximum electron temperature is unusually and extremely high. Moreover, the authors even do have the definition of the “maximum electron temperature”. Such a variable is also not shown in any of the equations in this manuscript. As we all know that the electron has its energy distribution function (EEDF) where the high-energy side is an open end. What is the “maximum electron temperature” exactly mean? In Figure 8, the authors show the variation of pressure up to 90 Torr, but in the previous figures, the authors have only up to 70 Torr. The research plan is unclear, how many conditions are tested?
Overall, this work is too superficial with no novelty and no contribution to the PECVD Community. Also, the model is too simple due to the missing of many important collisions. There are also many unclear descriptions. Therefore, I cannot give any more positive comments but have to suggest a rejection.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Reviewer 2 Report
Dear Authors,
The submitted manuscript, provides insights on the 'Effect of Positive Bias and Pressure on Plasma Flow Characteristics in a Chemical Vapor Deposition Chamber'. The authors used COMSOL software in order to establish a two-dimensional axisymmetric plasma model of pure H2 discharge. They found, based on their model, that there is a influence of the intensity change of positive bias voltage and deposition pressure on plasma flow characteristics. The manuscript text is mainly well written, with good quality images and well established equations.
However, all figures must be made larger, with legends and larger font sizes (closer to the size of the text), for a good and clear visualization by the reader.
I also expect that the 'Discussion' paragraph be extended and that the conclusions clearly be drawn out and highlighted.
A quick check of the entire manuscript, after the above addressed issues, should be made prior to submission.
So, in present form, the manuscript could not be consider for publication in Processes Journal: MINOR REVISION
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Reviewer 3 Report
Accept in present form
Author Response
Thanks for your comments.
Round 2
Reviewer 1 Report
The authors tried to improve the manuscript but some of the responses and improvements are unclear and with mistakes. The manuscript is still at a very low quality.
First, as the authors mentioned in their revision, the flow fields of electrons and ions are not specified. This makes the plasma like a neutral gas, and thus significantly degrades this work. There are tons of PECVD research works had been published which contain detailed plasma behaviors. A study of PECVD is meaningless without considering the electrons and ions flow and collisions in an external electromagnetic field.
This raises the second issue. In Table I, the authors list the inelastic collisions, but there are no rate coefficients for these collisions. There is a “particle number variable” shown in Eq. (4). We would expect the variable equal to the summation of rate equations with rate coefficients contained. The authors need to expand and show the details of it. Meanwhile, even the collisions in Table I are also containing mistakes. The newly added collisions are not those I recommended in the last comments. The authors added two very strange collisions. Collision #4 is an electron quenching, while the authors marked it mistakenly as “excitation”. Electron-impact quenching reactions are very rare, I have never seen any of them in the plasma simulation papers. Collision #5 is a three-body collision of H2 association, while the authors also mistakenly marked it as “excitation”. Also, two H atoms can associate without electrons. Why do we need an electron here? This is also very rare. Even though the authors can add the reference of where they found these two strange collisions, the authors still need to add those more commonly occur as I suggested in the previous comment.
A more severe problem is Fig. 5. First, an electron temperature over 2500 eV is extremely high. I am not sure if the authors even understand what this value means. A 2500 eV temperature equals 2.89 x 10^7 K. This is a temperature 5000 times higher than the sun. As I mentioned, the EEDF has an open end, therefore, it is true that there is a non-zero probability of an extremely small number of electrons reaching such a high temperature, but the number is too small to be meaningful. Theoretically, according to the open end of EEDF, there is no way to have a maximum temperature for an electron. The definition of the “max value of electron temperature” is questionable. Second, Fig. 5 looks exactly the same from the first manuscript to the second version, while the authors added two collisions in the model. Why the electron temperature is still the same, or the difference is too small to be observed? Without the rate coefficients, we cannot easily estimate the contribution of these newly added collisions to the electron temperature. I request the editor to check the raw data of these figures. If the electron temperatures of the two versions of the manuscript are exactly the same, then the authors didn’t run their simulation again with the newly added collisions, but simply responded to the reviewers and modified the manuscript, pretending that the simulation is improved, which breaks the academic integrity.
Other minor issues:
1) Figure text is too small and unclear.
2) Fig. 8 contains non-English characters.
3) Dot products are not showing correctly in Eq. (4)-(6). They are all shown in squares.
Overall, considering these fatal problems, I have to suggest a rejection of this work.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 3
Reviewer 1 Report
The authors revised the manuscript and responded to my previous comments. However, there are still major issues.
The electron temperature is still incorrect. The authors explained the unusually high values as the result of a strong electric field near the chamber wall, with an undefined term "edge effect". Since electron temperature is playing a major role in this work, the authors must show the electric field to support their statements (Lines 179-188). Also, it would be helpful to show the electron density distribution in the chamber. One can thus estimate the power applied to these electrons. It would not be a difficult task to plot the electric field and the electron density in the COMSOL post-processing.
Also, the gas bulk temperatures shown in Fig. 7 are over 2000 K near the substrate, which is too high. The melting point of silicon is 1687 K at 1 atm. The pressure here is much lower than atmospheric, leading to an even lower melting point. Usually, the temperature near the substrate should at around several hundred kelvins. The temperature shown in Fig. 7 will melt Si and SiO2 for sure.
Considering the issues of electron temperature and gas bulk temperature, there might be something wrong with the energy equation setups in the simulation. The temperature results currently shown in the manuscript are very weird and different from other PECVD works. Also, as a routine of a simulation paper, the authors should add a figure of meshing and several figures to show the initial conditions of species distributions, boundary conditions of electric potential, and pressure (or velocity). The initial species distribution, along with the electric field and electron density figures, might be helpful to find out what was wrong in the simulation. The reviewers and journal editors are not responsible to help the authors debug and troubleshoot. Please seriously double-check the simulation setups.
Other minor issues:
· The authors mentioned the collision cross section in Eq. (7). However, each collision event has a unique cross section. Please specify the values or the math expressions. Usually, a cross section is also a function of collision energy. The authors can add these values in Table 1.
· Line 169: not sure what “cloud diagram” means. Fig. 4 shows the contour plots.
· Typo in the x-axis label in Fig. 5.
· The collision type of Reaction 5 in Table 1 should be “association”.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Round 4
Reviewer 1 Report
The authors responded to my previous comments, but the main issue is still unsolved. The authors claimed that the low potential of the side wall results in the high electron temperature up to 1700 eV which is an extremely and unusually high value. The authors also claimed that an ion sheath is formed near the wall. According to the newly added electron density figure and the electron temperature values, we can estimate the Debye length near the wall at around millimeter scale. This means that the high electron temperature region is within and near the sheath region. However, according to the Bohm criterion, such a high electron temperature means that the ion drift velocity must go over 100 km/s to form such a sheath, which is impossible. Therefore, the simulation work is still incorrect.
I am trying to help the authors to find the problem. There must be some incorrect setups in the simulation. How are the authors compute the heat flux q in the Eq. (6)? Also, the pressure gradient in the Eq. (5) should not be used for charged particles. As a momentum equation, Eq. (5) is a variation of the Navier-Stokes equation. For neutral gases, the pressure gradient moves the molecules from the high-pressure region to the low-pressure region. In the microscopic view, the pressure gradient makes such a movement by means of the momentum transfers during collisions. However, for the charged particles, the collisions are Coulomb collisions. Therefore, it is questionable to simply use the pressure gradient on plasmas in Eq. (5). The motion of electrons and ions might be computed and separated from the neutral gas molecules. If Eq. (5) is a build-in model of COMSOL, then using it in this work is improper. The authors need to use a more professional plasma model. Also, it will be helpful for the authors to cite references of experimental works which can have such a high electron temperature in the CVD chamber to show that these results are possible in reality.
Overall, the simulation results look incorrect obviously and not convincing. I suggest the authors seriously consider their model used in COMSOL, and the physical mechanism in the computation.