Simulating 2D materials

[jmunshi1990]

Hello,

I am trying to simulate EBSD pattern for 2D structures with EMsoft. I have a question about basic geometry.

1. Is there any way to control/change the size of the geometry (such as the thickness, and lateral in-plane dimensions)? Also how do we make sure that the EBSD simulations are being performed in the thickness (vertical) direction?
2. Additionally, does it have any implications on the simulated pattern (theoretically).
3. Our purpose is to compare these with experimental EBSD which will be performed for 2-20 nm thickness (i.e., typically 3-30 layers of MoS2)
4. One more question on top of this, is there any way to simulate 2D heterostructures (for example MoS2 with hBN as substrate)?

I think the all answers to my question lies in the xtal creation step but your elaborations are much appreciated. Thanks and look forward to your response.

Joydeep

[marcdegraef]

Hi Joydeep, 1. you can set the film thickness in the EMMCOpenCL Monte Carlo program simply by limiting the total thickness with the depthmax parameter; that will then determine the maximum integration depth for the dynamical simulations in the EMEBSDmaster program.  So, if you have a 5 nm film, set depthmax to 5.0, and depthstep to 0.5 to have a sufficient number of energy bins. 2. Not sure what "implications" you are referring to here... 3. Typically, the EBSD signal comes from the top 10-20 nm of the sample so you should be able to incorporate that in your simulations. 4. There is no direct way to simulate a two-phase stack of layers. However, if your stack is epitaxial, you can in principle define a long (but low symmetry) unit cell that has both phases in it with the long axis normal to the film.  In your case though, keep in mind that most of the signal will come from Mo and S... EBSD signal is roughly proportional to Z^2 (Rutherford scattering cross section), so if you take two layers of equal thickness, one MoS2 and the other hBN, then the total scattering cross section (ignoring relative atom positions and such) is 42^2 + 2*16^2 = 2,276 for MoS2, and 5^2 + 7^2 = 74 for hBN... so in this case only about 3% of the BSEs will come from the hBN layer.  This of course ignores the actual geometry and beam direction, but I think you will not see much pattern due to the hBN, except perhaps for very thin MoS2 layers on hBN bulk. I would start by computing the master patterns for both phases and then adding them together; in the EMEBSDmaster HDF5 output files, the relative intensities of the two phases will be preserved, so as a first approximation you can add the two patterns in the correct orientation relationship.  The EMEBSDoverlap program will do that for you... I have an illustration of how that works in the following paper: W. Lenthe, L. Germain, M.R. Chini, N. Gey, and M. De Graef, "Spherical indexing of overlap EBSD patterns for orientation-related phases -- application to titanium", Acta Materialia 188:579-590 (2020)  DOI: <https://doi.org/10.1016/j.actamat.2020.02.025> Let me know if this helps... Regards, Marc.

…

On 1/10/22 9:38 PM, Joydeep Munshi wrote: Hello, I am trying to simulate EBSD pattern for 2D structures with EMsoft. I have a question about basic geometry. 1. Is there any way to control/change the size of the geometry (such as the thickness, and lateral in-plane dimensions)? Also how do we make sure that the EBSD simulations are being performed in the thickness (vertical) direction? 2. Additionally, does it have any implications on the simulated pattern (theoretically). 3. Our purpose is to compare these with experimental EBSD which will be performed for 2-20 nm thickness (i.e., typically 3-30 layers of MoS2) 4. One more question on top of this, is there any way to simulate 2D heterostructures (for example MoS2 with hBN as substrate)? I think the all answers to my question lies in the xtal creation step but your elaborations are much appreciated. Thanks and look forward to your response. Joydeep — Reply to this email directly, view it on GitHub <#156>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AB26VWEE2E7QVXDQNNEUXKDUVOJ3LANCNFSM5LVFLGLQ>. Triage notifications on the go with GitHub Mobile for iOS <https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675> or Android <https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub>. You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

-- Marc De Graef Professor Department of Materials Science and Engineering 130 Roberts Engineering Hall Carnegie Mellon University 5000 Forbes Avenue Pittsburgh, PA 15213-3890 USA Ph: (412) 268-8527 Fx: (412) 268-7596 ***@***.*** web:http://materials.cmu.edu/degraef Admin. Asst.: Marygrace Antkowski Phone: (412) 268-7240

[jmunshi1990]

Hello professor Graef,

Thank you very much for your elaborated explanation. For the heterostructure, initially I was not aware of the EMEBSDoverlap and what I did was actually superimpose images with some weights (to the intensity) to EBSD pattern of each phase (Considering there is no interfacial effect/dynamical effect between two phases). Usually in experiments as well hBN phase has a very weak signal and poor SNR compared to say MoS2. So for our machine learning problem, we wanted that to be the case to simulate/mimic real experiments. I will try the overlap code and check the results. Thanks for the suggestions.

Couple of questions related to the sample orientation (as I found these details a bit confusing in the Wiki page as probably I am new to EBSD field):

1. while defining the unit cell, the c direction (mentioned in the commandline input) is always the thickness direction for the sample - is that right? meaning c direction defined through the xtal file is always going to be the back scatter direction. is that right?
2. The last question is about some more clarifications on the euler.txt file at the very end. what we want is to simulate patterns for multiple orientations with different relative orientation between the two phases. for example, keeping hBN and MoS2 layers stacked on top of each other we want to rotate the MoS2 layer relative to hBN layer. How do we vary the euler angles in the file? so if the c direction of the unit cell for both of them are in normal directions (thickness) then we are rotating MoS2 in the a-b plane...
3. also, can we tilt the sample (i.e. rotate the c-axis) by changing the euler angle? As per my understanding if the master simulation considers electrons scattering from the sample keeping c-direction as normal, is it possible to tilt c-axis after the master simulation is done?

Thanks again and please correct me if I am wrong in any case.

Joydeep

[marcdegraef]

Hi Joydeep, see responses below. Regards, Marc.

On 1/11/22 8:52 AM, Joydeep Munshi wrote: Hello professor Graef, Thank you very much for your elaborated explanation. For the heterostructure, initially I was not aware of the EMEBSDoverlap and what I did was actually superimpose images with some weights (to the intensity) to EBSD pattern of each phase (Considering there is no interfacial effect/dynamical effect between two phases). Usually in experiments as well hBN phase has a very weak signal and poor SNR compared to say MoS2. So for our machine learning problem, we wanted that to be the case to simulate/mimic real experiments. I will try the overlap code and check the results. Thanks for the suggestions. Couple of questions related to the sample orientation (as I found these details a bit confusing in the Wiki page as probably I am new to EBSD field): 1. while defining the unit cell, the c direction (mentioned in the commandline input) is always the thickness direction for the sample - is that right? meaning c direction defined through the xtal file is always going to be the back scatter direction. is that right?

No that's not always the case.  In EMsoft, we use the International Tables of Crystallography convention that the z-axis of the cartesian reference frame is normal to the a-b plane, i.e., parallel to the reciprocal c* axis.  The doesn't matter for the Monte Carlo simulations which treat the structure as amorphous, but it does matter for the EMEBSDmaster simulation; the c* axis is always at the center of the master pattern.

1. The last question is about some more clarifications on the euler.txt file at the very end. what we want is to simulate patterns for multiple orientations with different relative orientation between the two phases. for example, keeping hBN and MoS2 layers stacked on top of each other we want to rotate the MoS2 layer relative to hBN layer. How do we vary the euler angles in the file? so if the c direction of the unit cell for both of them are in normal directions (thickness) then we are rotating MoS2 in the a-b plane...

Take a look at the template file for the EMEBSDoverlap program... I think you can do this by varying the orientation relations.

1. also, can we tilt the sample (i.e. rotate the c-axis) by changing the euler angle? As per my understanding if the master simulation considers electrons scattering from the sample keeping c-direction as normal, is it possible to tilt c-axis after the master simulation is done?

The master pattern does not consider a specific sample orientation at all.  Think of the master pattern simulation as the following thought experiment:  take your crystal structure and align it with a Cartesian reference frame; assume that the crystal forms a sphere and that there is a uniform electron source at the center of the sphere.  The electrons channel in all directions through the spherical crystal and then hit a spherical detector (the Kikuchi sphere) surrounding the crystal.  The pattern of intensity on that sphere is the master pattern.  It has nothing to do with sample orientation which is defined afterwards in the EMEBSD program.

…

Thanks again and please correct me if I am wrong in any case. Joydeep — Reply to this email directly, view it on GitHub <#156 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AB26VWCFYZ3SUY735QNVWT3UVQY3XANCNFSM5LVFLGLQ>. Triage notifications on the go with GitHub Mobile for iOS <https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675> or Android <https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub>. You are receiving this because you commented.Message ID: ***@***.***>

-- Marc De Graef Professor Department of Materials Science and Engineering 130 Roberts Engineering Hall Carnegie Mellon University 5000 Forbes Avenue Pittsburgh, PA 15213-3890 USA Ph: (412) 268-8527 Fx: (412) 268-7596 ***@***.*** web:http://materials.cmu.edu/degraef Admin. Asst.: Marygrace Antkowski Phone: (412) 268-7240

[jmunshi1990]

Thank you very much professor.

I have one quick question about high throughput simulations. I am trying to simulate 100s of structures in automated manner whose initial unit cell structures will be extracted from a data repository through python program. Is there any way to use a input file for the crystal structure through the EMmkxtal command. Currently the way it is implemented it is difficult to use a script to run this command. Thanks