Recent Posts

41

General Questions and Answers / Re: Recommended approach to generate an interface between two amorphous materials

« Last post by Anders Blom on October 4, 2025, 01:35 »

Correct, the Builder tools are for crystalline, well-behaved cases where the coincident lattice model makes sense.

There is not really a simple algorithm to create an interface between two random materials. At heart it's a global optimization problem, hard in itself, and any starting point (the configurations you happen to have of the left and right material, respectively) it's just a representative case, and can never be said to be universal, since each time you generate the amorphous sample you will get a new random configuration, and how you choose the surface termination is equally random.

Now, that may have been obvious to you already, but I just wanted to ensure we share the same picture.

As for the approaches you mention, I would definitely try the universal NN-based potentials for this case. They may not be perfect, and often not accurate enough for a reaction barrier or other very detailed calculations, but the time you save from not having to create a new MTP from scratch or even a fine-tuned MACE demo model from compensates for this. And, you can (and probably should) always do a final DFT relaxation - that is, the MLP-generated structure should just be seen as a starting guess for the interface configuration. And, thanks to their speed and readiness, you can try generate dozens or hundred of candidates in very short time.

42

General Questions and Answers / Recommended approach to generate an interface between two amorphous materials

« Last post by evansjc on October 1, 2025, 22:08 »

Hello,
Hi,

I would like to know what the recommended approach is with the most recent update to build an interface between two arbitrary amorphous materials. Is the best approach still to train an MTP that can handle both amorphous materials as well as the interface? Is it better to use a newer neural-network-based universal potential instead of training an MTP? If neural-network-based potentials are recommended, then should they be fined tuned to the specific interfaces we are considering? Also, once the specific potential has been chosen or trained, how do we go about building the interface that will be then be analyzed? Is there a builder tool that is recommended? The builder tool seems to be mostly oriented to creating interfaces between crystals.

I would appreciate any recommendations. Thank you!

43

General Questions and Answers / Re: Question about Defect Diffusion Workflow Template

« Last post by Vaida Arcisauskaite on October 1, 2025, 16:45 »

Hello,

We have looked into the vibrational correction issues in your charged point defect simulations.

- In the Charged Point Defect Analyzer – if you unclick “Include vibrations”, you will be able to see trap levels without vibrational corrections.
- The error is related to the fact that you have negative frequencies (for B reference/elemental material, look for this warning: “There are 973 non-positive eigenmodes”). And if you do have negative frequencies, you can't calculate the vibrational DOS, and therefore you can't work out vibrationally corrected formation energies and trap levels.
-Now the question is why you have got negative frequencies. The geometry optimization of reference material with your MTP didn’t converge, and doing phonon calculations on an under-optimized structure is not a good idea.

- How to solve this? There are two options
1.  When training an MTP, you need to include elemental B material and some displacements into the training data set. Did you use the PBE functional for generating training dataset?
2. Try using a universal MACE NN potential for vibrational corrections and pre-relaxation instead of the MTP and evaluate the results.

Another thing we have noticed, that your NEB was not converged in many cases within 200 steps and these NEB calculations for diffusion take long time (formation energies/trap levels were done within a couple of hours), explaining long simulation time that you mentioned. We will think about this problematic NEB convergence in your case.

Hope, this helps.

Best regards, Vaida

44

General Questions and Answers / Re: Question about Defect Diffusion Workflow Template

« Last post by Vaida Arcisauskaite on September 29, 2025, 21:29 »

Hello,

Thanks a lot for reporting this.
The issue is probably in the custom block "Filter migration pairs based on Pre-relaxation Calculator barrier heights".
If you convert it to a CustomBlock (right click on the block to do this), and then look at the script, the append function is expecting different values, but it gets a list. If you change it to:

for index, _ in indices_and_barriers:
    filtered_defect_migration_pairs_table.append(*defect_pair_table[index])

(note the additional *) then it should work.

We will further investigate this in the QuantumATK team.

We will also look into errors in the vibrational correction and your calculation timings and come back to you.

Best regards, Vaida

45

General Questions and Answers / Best Practices to Calculate Electron Affinity of Oxides

« Last post by evansjc on September 26, 2025, 23:33 »

Hello,
I am trying to follow a procedure similar to the one outlined here (https://docs.quantumatk.com/tutorials/work_function_ag_100/work_function_ag_100.html) to calculate the electron affinity of oxides (like HfO2). However, I am unable to obtain a value that matches the experimental quantity. I generated the slab and set up the boundary conditions as described in the tutorial. I am using hybrid functionals and the LCAO calculator to obtain realistic band gaps. However, the value I am obtaining diverges significantly from the experimental value. I tried increasing the number of k-points and the basis set size but saw no significant difference in the answer. Is there anything that I should be particularly careful with when following this process with a material like HfO2? Are there any best practices for cleaving the surface and setting up ghost atoms? Are there any calculator settings that I should be particularly careful with?

I would appreciate any suggestions. Thanks!

46

General Questions and Answers / Re: Question about Defect Diffusion Workflow Template

« Last post by pshinyeong on September 26, 2025, 11:11 »

Hello ,

I re-ran the calculation with the corrections previously suggested:

Added the chemical potential calculation for the boron atom

Tightened the convergence criteria in the calculator

Linked the calculators as described in the PDF

Despite these changes, the job crashed near the end with the following error:

Code

1_Defect_Diffusion_Workflow/1_Trial_1/250921_003210_4ibppi7p.hdf5' no longer exists.This means no task results will be saved to the new file.
Traceback (most recent call last):
  File "/home/synopsys/quantumatk/X-2025.06/bin/../atkpython/bin/atkpython", line 8, in <module>
    sys.exit(__run_atkpython())
             ^^^^^^^^^^^^^^^^^
  File "zipdir/ATKExecutables/atkwrappers/__init__.py", line 912, in __run_atkpython
  File "./defect_diffusion_MTP_results.py", line 901, in <module>
    filter_migration_pairs_based_on_prerelaxation_calculator_barrier_heights(
  File "./defect_diffusion_MTP_results.py", line 895, in filter_migration_pairs_based_on_prerelaxation_calculator_barrier_heights
    filtered_defect_migration_pairs_table.append(defect_pair_table[index])
  File "zipdir/sergio/HDF5/Table.py", line 2476, in append
  File "zipdir/sergio/HDF5/Table.py", line 1633, in validate
ValueError: The initial_defect column can only contain instances of NamedPointDefect, was list
Abort(1) on node 20 (rank 20 in comm 0): application called MPI_Abort(MPI_COMM_WORLD, 1) - process 20

I’ve attached the link to the relevant output files for your review through message, please alert me if the message was not sent because I have sent the message twice but it didn't show up in the sent box.

Could you please check where my setup or workflow may be incorrect, particularly around the initial_defect field expected to be a NamedPointDefect? Any guidance on resolving this would be appreciated.

Also, this run took a little over three days on 5 nodes (48 cores per node; tasks per node: 6; CPUs per task:; 8 ) before failing. Is this runtime typical for the Defect Diffusion workflow with these settings, or does it suggest a misconfiguration or inefficiency?

Thank you in advance for your help.

47

General Questions and Answers / Re: How to get carrier density in a device?

« Last post by techenthusiast on September 25, 2025, 08:32 »

Especially when processing an I-V curve calculation with bias.

Also, can we Do the analysis of electric field , Current density , EF, Mobility variations for FET Device ?

If then how, Like what will be the efficient / suitable calculator and analysis block

Thanks

48

General Questions and Answers / How to get carrier density in a device?

« Last post by DDLDLL on September 24, 2025, 18:26 »

Especially when processing an I-V curve calculation with bias.

49

General Questions and Answers / Re: HartreeDifferencePotential caculation 2

« Last post by Anders Blom on September 24, 2025, 00:21 »

This is a perfectly normal graph. Those "fluctuations" are the actual variations of the potential with atomic resolution.

50

General Questions and Answers / Re: Recommended Guideline for Creating a Ternary MTP by Combining Binary MTP

« Last post by Anders Blom on September 24, 2025, 00:19 »

Quick check, do you really need Ge-Se-Te or Ge-Sn-Te (commonly, GST)? If it's the latter, we already have such a potential provided in the package.

Otherwise, you are in for a ride, but a potentially very rewarding one!

Your methodology is sort of correct, but you don't fit separately for 1-3, you combine those steps into one. That is, you include a bunch of basic crystal structures for all combinations in one training set (which you can test against the crystal experimental data). Then, instead of your step 4, you use active learning on a set of random alloys (and/or amorphous structures) of Ge-Se-Te to refine the potential for the full system. Ideally, for different stoichiometries! This is similar to the workflow we used for TiSi2 in the tutorial below, just a bit more complex for your case since you have 3 elements.

https://docs.quantumatk.com/tutorials/mtp-training-c-am-TiSi/mtp-training-c-am-TiSi.html