Simulation Data for Ion Interactions with Target Materials from Ion Beam and Plasma Systems

This area is for ADVANCED theory and discussions only. If you just joined fusor.net, chances are this is NOT the area for you to be posting.
Post Reply
Michael Bretti
Posts: 175
Joined: Tue Aug 01, 2017 12:58 pm
Real name: Michael Bretti

Simulation Data for Ion Interactions with Target Materials from Ion Beam and Plasma Systems

Post by Michael Bretti »

Over the past three weeks or so I have been working on a variety of simulations to better understand and provide data on a variety of interactions and mechanisms between ions and surfaces interacting with the ions. These simulations range from simple stopping ranges to much more sophisticated modeling of sputtering, implantation, and target damage due to ions from a variety of sources, primarily ion beams and plasmas. The software used is SRIM 2013, along with the various tools and functionalities associated with the program. A wide variety of simulations will be explored and elaborated on, including ion loading from beam on target systems, surface interactions with DC glow discharge plasmas, sputtering yields from magnetron sputtering sources, and a variety of target materials and combinations with different ions.

To start, below are some very simple tables of ion ranges for several different ions in stainless steel. Note that these represent only mono-energetic energy distributions, with a single impinging angle upon the target surface. Therefore this only looks at the estimated maximum penetration range at each energy level for an angle perpendicular to the surface. For each case, data from 100eV to 1MeV are presented. This data was generated using the SR (Stopping Range) functionality of SRIM. Stainless steel was chosen as the starting material since this is the most commonly employed and largest bulk metal seen in high vacuum systems dealing with ion and plasmas, especially for fusors. I will upload more data specifically for different target materials and layering combinations, such as the commonly employed titanium self-loading target for beam-on-target systems as I progress in this area, along with full TRIM simulation results.

Argon in Stainless Steel.pdf
(57.73 KiB) Downloaded 448 times
Helium in Stainless Steel.pdf
(58.2 KiB) Downloaded 541 times
Hydrogen in Stainless Steel.pdf
(58.29 KiB) Downloaded 516 times
Nitrogen in Stainless Steel.pdf
(57.85 KiB) Downloaded 499 times

The next set of data that will be posted shortly includes much more advanced simulations utilizing the TRIM functionality of SRIM, exploiting the use of custom .DAT input files to generate ion parameters for estimating the effects of diffuse glow discharge plasmas across a wider area of stainless steel surface. This is used to explore the effects of sputtering and surface loading of stainless steel chambers when utilizing glow discharge cleaning to prepare the vacuum surfaces. The first set of simulation data looks at three different ion energy distribution inputs for nitrogen on stainless steel. These range from lower bound unrealistic estimates, to upper bound estimates, to an intermediate energy distribution that should account for a wide variation of system fluctuations and uncertainty during operation. This is used to determine and select an energy distribution for the next set of simulations exploring the differences in ion implantation concentrations and sputtering yields between hydrogen (as an analog to deuterium), nitrogen (as an analog to air), and argon (as a general comparison for a common gas used in sputtering and surface cleaning). All of the data along with a detailed write-up is already posted on my website, but I will post and elaborate on the important parameters here relevant to sputtering and surface loading effects for DC glow discharge cleaning.
Michael Bretti
Posts: 175
Joined: Tue Aug 01, 2017 12:58 pm
Real name: Michael Bretti

Re: Simulation Data for Ion Interactions with Target Materials from Ion Beam and Plasma Systems

Post by Michael Bretti »

UPDATE 04/19/2018

Previously in this post I had written a very long and detailed description of simulations that were run using SRIM 2013/TRIM for looking at the effects of ion energy distribution for implantation depth and sputtering in a 500V DC glow discharge plasma, including various graphs and charts for these simulation results and other output data. Upon further reviewing the literature, the ion energy distributions presented here were not correct. Since this was a preliminary study looking at identifying the differences in simulation output, rather than the actual experimental simulation results, the data was not directly beneficial for work related to fusors and other plasma systems here. As a result, I have decided to edit this original post and remove the preliminary data to avoid confusion or misinformation on this subject. However, proper energy distributions have been identified and will be run for the proposed gases of hydrogen, nitrogen, and argon, for both low voltage and high voltage glow discharge cleaning that will be submitted here in subsequent posts.
Last edited by Michael Bretti on Thu Apr 19, 2018 5:54 pm, edited 2 times in total.
John Futter
Posts: 1848
Joined: Wed Apr 21, 2004 10:29 pm
Real name: John Futter
Contact:

Re: Simulation Data for Ion Interactions with Target Materials from Ion Beam and Plasma Systems

Post by John Futter »

Michael
I also put simion and scrim simulations up on this site about 6 years ago
I do not see why you are using such low ion energies.
As ion energy goes up you go from shallow implant / bounce off to sputtering ( 10keV to 30keV is where most sputtering occurs) to implant at energies above 30keV and yes implant is taking place at lower energies but is surpassed by sputtering.
For most here on this site D and D2 implant depth in fusor shell is probably more important ( and air products O, O2 N2 Ar) notice I did not include N as this is hard to make without a very specialised ion source.
one sim I did not do was using a Pd layer on the shell to store D, D2 which has been postulated to maybe help

but while you are in the seat give it a go
Michael Bretti
Posts: 175
Joined: Tue Aug 01, 2017 12:58 pm
Real name: Michael Bretti

Re: Simulation Data for Ion Interactions with Target Materials from Ion Beam and Plasma Systems

Post by Michael Bretti »

John,

Thank you for your reply and input. I will definitely search the forums to check out your simulation results, I don't think I have seen them yet.

The reason I am using such low energy ions is that this first initial case is not looking at sputtering and implantation due to operating the fusor, but exploring the effects of glow discharge cleaning in the fusor, which usually takes place in the 400-500V range. I posted a paper recently in another topic area looking at the comparison in effects of this cleaning in a tokomak system using argon and hydrogen. I started with nitrogen since it is the closest analog to air and is in the middle of hydrogen and argon in terms of molecular weight. Also, it is not possible to directly simulate compund ions for the flown ions in TRIM such as air or N2, so N was selected as a basic case. Again, from the description above, this was only a preliminary simulation to select the ion energy distribution to be used for the actual full simulation. This simulation also uses input files to randomize parameters within certain bounds to better simulate a diffuse plasma over the standard TRIM default beam.

The next simulation, as also mentioned above, will be of more use here, but this initial case was posted to give an introductory background. Again, hydrogen, nitrogen, and argon will be compared for the reasons mentioned above in the initial post. I do plan on running many other simulations for implantation and sputtering damage during actual fusor run voltages and for beam on target interactions with multiple different materials, but for now this and the next simulation solely looks at the effects of glow discharge cleaning. This is primarily to establish the difference in surface effects between glow discharge cleaning with residual air in the system vs. deuterium, as well as exploring the use of custom TRIM input data.
Michael Bretti
Posts: 175
Joined: Tue Aug 01, 2017 12:58 pm
Real name: Michael Bretti

Re: Simulation Data for Ion Interactions with Target Materials from Ion Beam and Plasma Systems

Post by Michael Bretti »

John,

I finally found the results you did in SRIM for Steven Sesselmann. Very cool to see. SRIM is quite an amazingly powerful tool, and even more amazing that it is free. There are many textbooks even dating back to the 70s that refer to the use of SRIM and TRIM for sputtering and implantation simulations (using the very old versions of the code), so it certainly has a very long history of use in this field.

I had done some preliminary diffuse plasma simulations at 10keV before these runs, but decided to switch focus for now on the low energy glow discharge plasma cleaning. However, one of the things I am interested in seeing and comparing in terms of simulation results is the sputtering yield and damage effects of the typical TRIM beam vs. the widely randomized input I am using. The ability to vary input parameters will also be very useful, for example, looking at wider angle beams, different beam energy distributions, and other plasma sources such as magnetron sputtering heads where I could simulate the race-track sputtering based on positional distribution. I think once I get the low energy stuff completed and posted I will do a comparison with the 10keV case, and then move on to more conventional beam on target loading with complex substrates and various distributions. Since various ions can be flown in a single simulation with custom .DAT input files, I will eventually look at incorporating simulations with numerous ion species (such as air with its various constituents in the proper proportions.) I would also like to see if it is possible to simulate compound targets with distributed water vapor to look at impact-based desorbtion from plasma cleaning, though I do not know if this will work or not yet.

Another very cool thing to see that I don't believe has been explored here is using TRIM to calculate neutron data. The most recent version of SRIM/TRIM has the capability of taking data for photon, electron, and neutron cascades and simulating recoil damages in targets with them. However, this requires the use of an external program such as MCNP to first generate the neutron data deriving the position and recoil statistics for each collision, which can then be input into TRIM for further calculations.

Glad to see that there is another SRIM user well versed in its operation and use to compare notes and bounce ideas off of in the future. I look forward to your input further on the subject as I progress.
John Futter
Posts: 1848
Joined: Wed Apr 21, 2004 10:29 pm
Real name: John Futter
Contact:

Re: Simulation Data for Ion Interactions with Target Materials from Ion Beam and Plasma Systems

Post by John Futter »

michael

We also use Dynamic trim @ work which does the damage cascades much better --but it is extremely processor intensive and it pays not to fly too many ions, as an answer could take days to calculate
Michael Bretti
Posts: 175
Joined: Tue Aug 01, 2017 12:58 pm
Real name: Michael Bretti

Re: Simulation Data for Ion Interactions with Target Materials from Ion Beam and Plasma Systems

Post by Michael Bretti »

A few updates on the simulation progress and previous posts. After a couple of weeks of playing around with various angular distributions and browsing literature on the subject, I finally found the appropriate general angular distributions for ion angles impinging on grounded walls in plasma reactors, as well as energy distributions for DC glow discharge plasmas. Since the previously mentioned preliminary testing utilized several incorrect distributions, and was only a preliminary test to look at varying energies for the input file and comparing the output file results, the data has been removed to avoid confusion and incorrect information on the subject.

The initial simulations were run with uniform angular distributions from -90 to +90 degrees where 0 is normal to the surface. Actual angular distributions based on experimental data however appear to generally follow a Gaussian curve centered around the normal angle of incidence.

For ion energy distributions in DC glow discharge plasmas, experimental data shows that the ion energy distribution generally follows a model presented in the work done by Davis and Vanderslice in 1963 and their resulting distribution equation for ion energy for DC glow discharge plasmas. These corrections will be implemented in my actual simulation runs and presented here when I finish generating and organizing the data, and will be posted when completed. I will also present a comparison between sputtering yields and implantation between narrow ion beam, wider ion beam, and glow discharge plasmas for equivalent voltage potentials.
Michael Bretti
Posts: 175
Joined: Tue Aug 01, 2017 12:58 pm
Real name: Michael Bretti

Re: Simulation Data for Ion Interactions with Target Materials from Ion Beam and Plasma Systems

Post by Michael Bretti »

After about a month of intensive work, I have finally completed the simulations and uploaded all of the data for the DC glow discharge plasma simulations using the advanced features of TRIM. The process has been very long and tedious, with many dozens of preliminary simulations and many hours of research, testing, tweaking, plotting, and curve fitting large amounts of data, but it was well worth the effort. A link to the entire data and write-up page can be found here:

http://appliedionsystems.com/portfolio/ ... ess-steel/

I won't go over the details or complexities here, as these are covered in great detail in the write-up as well as the linked reference papers, but will provide a basic overview of the study. As mentioned prior, the goal of this simulation effort was to utilize custom input files for better simulating the effects of a glow discharge plasma on a stainless steel surface to understand sputtering and implantation with various gases during glow discharge cleaning. The normal TRIM default only allows for mono-energetic ions at a single angle of incidence varied along +/- one atomic width. However, this does not accurately represent more complex phenomenon such as glow discharge plasmas, where there is a distribution for energy as well as impinging angle. For glow discharge plasmas, Davis and Vanderslice published a paper in 1963 detailing the distribution function associated with ion energies for this type of plasma. In 2001, Butz-Jorgensen modified the m parameter (which deals with electric-field linearity) from the original value of 2, to that of 5/3 based on derivations from the Child Collision Law. These parameters are used in the resulting simulation, where a probability density function based on the Davis and Vanderslice model was applied to flown ion energies in the TRIM simulation. For ion angles, experiments in literature observe a Gaussian distribution centered along an angle of direct incidence to the surface with a tight angular spread. These numbers roughly correlate to about +/-80-85% of impinging ions having angles of 10 degrees or less relative to normal incidence to the surface.

Three different gases were explored: hydrogen, nitrogen, and argon. These are some of the most commonly tested gases for plasmas in literature, including others such a helium and neon, and are gases regularly encountered in vacuum plasma systems. In a DC glow discharge plasma, several regions form between the anode and cathode. The area of main interest that ultimately determines resulting ion energies upon the cathode surface is the cathode sheath layer. For a DC glow discharge plasma, the main parameter determining the resulting ion energy distribution is the s/λ parameter, in which s represents the sheath thickness and λ represents the ion collision mean free path, therefore giving the mean number of collisions for an ion traversing through the sheath. Based on the Child Collision law, for DC plasma discharges, s/λ is proportional to s x p, where p is the pressure, and in general, s x p remains constant. Because of this, the mean free path for a given pressure must be balanced by the sheath thickness, which results in the occurrence that variation of pressure does not affect the resulting ion energy distribution. In this case, voltage, system configuration, and other experimental parameters will determine the resulting energy distribution.

For each of the three gases, a total of five s/λ values were simulated, resulting in a total of 15 simulations for this study. s/λ values selected are 5, 15, 25, 45, and 65, representing a wide range of operating conditions. As observed from measurements found in literature, in general, double-ionized ions such as Ar2+ and Ne2+ tend to follow energy distributions with very low s/λ values around 5, where single-ionized ions such as Ar+ and Ne+ have higher measured s/λ. Various parameters such as applied voltage and system topology can affect these numbers. As the s/λ decreases, the energy distribution increases towards a higher percentage of higher energy ions. Distributions become more complex for mixed gases outside of mono-atomic gas cases. Although s/λ values must be determined experimentally for a given setup, based on system pressure and assuming a particular s/λ value, sheath thickness can be derived.

The page in the link above includes a large collection of plots, data, and resulting ion statistics from the simulation. In addition to the data, a full detailed technical write-up is included explaining each of the plots. Source code and instructions are also provided for the Python script that generates the input data. References to all publications used for this study are given, as well as links to PDFs for each of the referenced papers. For anyone interested in the subject of dc glow discharges, and plasma physics in general, I highly recommend reading the Budtz-Jorgensen paper. It is an excellent and very detailed paper exploring the physics and derivations behind this phenomenon.

A few interesting key observations to take away from the simulations:

1.) As expected, for increasing s/λ values, resulting in decreasing probability of higher energy ions, sputtering yields and depths decrease. In addition, for higher atomic weights, sputtering yields increase while penetration depth decreases. Recoils and phonon production also increase as energy increases as well as ion mass.
2.) In the case of hydrogen, no sputtering was observed for all cases of s/λ values for a 500VDC glow discharge plasma.
3.) The bulk stopping losses for hydrogen in the stainless steel target are a result of ionization losses, or losses due to interactions with target atom electrons. For the heavier ions of nitrogen and argon however, losses are dominated by phonon interactions, or bulk vibrations in the target lattice structure due to ion-target atom interactions.
4.) Hydrogen exhibits further travel in the target material with minimal interactions with target atoms, with stopping primarily due to ionization losses as described above.
4.) For higher energy argon ion distributions, losses from phonon recoil interactions dominate total losses.
5.) Back-scattering ions from the surface are significantly higher for hydrogen as opposed to nitrogen and argon. For equivalent energy spreads and ion numbers, a greater number of argon ions will be implanted in the target surface compared with hydrogen.

Now that I have the ability to generate ion data based on custom probability distributions, this will pave the way for running many more advanced ion-target simulations for my fusor, plasma, and ion beam work. As an additional side note, all of the resources for these simulations are freely available for any hobbyist to use themselves, and can be readily found online. While intensive and sound background research is still required, there are a plethora of free tools available at one's disposal for advanced research on a limited budget. In my own case, while I do not have the budget to run the experiments I want for quite a while, and will barely complete my test system for just vacuum conditioning and pumpdown, the lack of funds has turned out to be an incredibly strong motivating factor and powerful asset in that it has forced me to utilize whatever information and tools are available to maximize my knowledge, understanding, and planning in various areas so that when I do finally run experiments, I will be exceedingly well prepared. If you can't build a fusor or similar system, there are always other ways to do research while you work towards realizing physical experiments.
Michael Bretti
Posts: 175
Joined: Tue Aug 01, 2017 12:58 pm
Real name: Michael Bretti

Re: Simulation Data for Ion Interactions with Target Materials from Ion Beam and Plasma Systems

Post by Michael Bretti »

A few additional things to note in terms of DC glow discharge cleaning based on other literature indirectly related to the above simulation study, while on the immediate general topic of vacuum system cleaning and preparation:

1.) Sputtering/surface cleaning can occur in two modes - physical and chemical. The above simulations explore only physical modes of sputtering. However, in various works, different gas mixes can yield enhanced cleaning results from single gases. For example, a common gas employed in glow discharge cleaning is argon. However, various studies have found that mixes of 90% argon/10% hydrogen, or 90% argon/10% oxygen show greatly increased cleaning of the internal chamber surfaces over pure argon. In the case of a mix with oxygen, the oxygen readily combines with carbon-based contaminants on the surface to form molecules such as CO which are readily pumped out. A slight mixture with hydrogen also shows improved effectiveness of surface cleaning due to chemical sputtering methods. It has been shown that 100x lower ion dose is required to remove surface contamination due to carbon from stainless steel when 90/10 mixes of argon/oxygen are used as opposed to pure argon.

2.) Based on work done by A. G. Mathewson at CERN in 1987, despite using relatively low voltages for DC glow discharge plasma cleaning with argon, resulting in shallow penetration depths in the chamber surfaces for high vacuum systems to remove argon from stainless steel, baking at 350C is still required to liberate the implanted argon. Argon cleaning however was shown to be effective for greatly reducing H and CO concentrations and monolayers from the chamber surfaces.

Below are technical papers and studies are referenced above, as well as other studies dealing with vacuum chamber cleaning methods, glow discharge cleaning, and the physics of adsorption and desorbtion of various types of gases from stainless steel and other metals:

Cleaning and Surface Properties.pdf
(263.11 KiB) Downloaded 4402 times
Thermal Outgassing.pdf
(1.71 MiB) Downloaded 1110 times
Post Reply

Return to “Advanced Technical Discussion Area”