Simulation by qstem & qmb

[ljk]

Hello, I have tried to simulate SrTiO3 (100) & (110) HAADF images by qstem.

I got trouble because the different simulation methods result in different simulated images, (of course, possible..) 

But, the big problem was the visibility of Oxygen atom, which  is usually not visible at HAADF for STO images (experimental referenecs), However, It was visible at my simulated results.

Additionally, I did other ways (the details are on the ppt files and also attached the raw data, cfg and qsc files), then the results were consistent with the experimental results.

One is using cfg files and change the width,height,thickness under box mode also with zone axis. 

The other one is to make the model by qstem model maker at the initial stage and simulate that file, which showed the more consistent images in my case.

When I followed up the tutorials and questions, the former method should be also consistent, but I didn't. 

The other condtiion (scan window, probe arrray, slices, microparameters) are similar and the only diference should be the used crystal models, also estimated by qstem model properties.

I think this problem is critical for my future simulation trying, I tried my best to solve this situation, but couldn't. So, I asked for your comments for what I am doing wrong.

Please refer to the attached ppt files and zip files (raw data - cfg, qsc files are all there).

Thanks

Sincerely

[Christoph Koch]

Hello,

 

the difference in these configurations lies in the Debye-Waller factors you have specified. The file you generated from the CIF file has no Debye-Waller factors at all. This tells you that HAADF-STEM imaging is very sensitive to the Debye-Waller factors of the different Atom species.

 

With best regards,

Christoph Koch.

--
You received this message because you are subscribed to the Google Groups "QSTEM" group.
To unsubscribe from this group and stop receiving emails from it, send an email to qstem+un...@googlegroups.com.
To post to this group, send email to qs...@googlegroups.com.
Visit this group at https://groups.google.com/group/qstem.
For more options, visit https://groups.google.com/d/optout.

[ljk]

Hello, Thank you for your prompt reply. 

I'm regret to be confused more, so I had several questions. 

I attached ppt files and please refer to it. 

Thanks.

2017년 2월 6일 월요일 오후 6시 7분 30초 UTC+9, Christoph Koch 님의 말:

[Christoph Koch]

Hello ?,

 

Let me try to answer your different questions one by one:

 

1.      (Q1) If you generate a .cfg file from a .cif file and want to simulate HAADF-STEM images you should try to search the literature (or the .cif file you have used as input – if the structure stems from X-ray diffraction, then the DW-factors are often fitted and provided as well) for suitable Debye-Waller factors. Otherwise both qstem and qmb will insert default Debye-Waller factors.
Having written this, Ruben Bjorge from Trondheim, Norway and Florian Krause from Bremen, Germany have recently discovered that qstem is currently erroneously dividing the Debye-Waller factor by 3 when doing TDS calculations, so with the current version of qstem, one should specify a temperature that is 3 times higher than the actual specimen temperature (i.e. typically 900 K), IF YOU WANT TO DO A TDS CALCULATION. This is a serious bug, and I will fix it as soon as possible. Setting the temperature to 900 K will rescale all the mean squared atomic displacements by the necessary factor of 3.

 

2.      (Q5) Both, the TDS multislice calculation, as well as the calculation without TDS rely on the very same Debye-Waller factor B. If TDS is turned off, then the atomic potentials will be smoothed by exp(-B*s^2). If TDS is turned on, then this smoothing will not be done, but instead the atoms will be displaced randomly with a displacement amplitude u=sqrt[B/(8pi^2)]. So in either case the value of B is highly relevant. In the kinematic approximation, both treatments should result in exactly the same change in the diffracted intensities – only the diffuse scattering between the Bragg beams will be different (no diffuse scattering, if TDS = off). Since Thermal diffuse scattering occurs predominantly at high scattering angles, this is more important in HAADF-STEM imaging, than in HRTEM or BF-STEM imaging.

 

3.      (Q2,Q3) I do not remember where I have taken the Debye-Waller factors for SrTiO3 from. It was probably some article in the literature. L.M. Peng has also once computed a number of DW factors (see attached paper). But please keep in mind that in each crystal the same atom type may be bonded more or less rigidly, and thus the vibrational amplitudes may also differ.

 

4.      (Q4) The difference between the simulated HAADF-STEM image and the experimental image can be due to several factors (different thickness, tilt, temperature, effective source size, etc.) but it might also be due to the factor 3 that I have mentioned above.

 

With best regards,

Christoph.

[ljk]

Thanks for your answers and It was great help for me to understand the qstem simulation.

I try to simulate more based on your answers and changed TDS temperature as 900K.

Unfortunately, there was no critical difference on HAADF simulation images and then, I had additional two questions.

Please refer to the attached ppt files and for your confirmation if needed, also for stem simulation file.

I'm sorry for repeated questions, but I think I should solve this problem because this is closely related to the core of simulation and for using future q-stem simulation for real samples.

Thanks

Sincerely

J Lee 

2017년 2월 10일 금요일 오후 6시 29분 34초 UTC+9, Christoph Koch 님의 말: