strange contrast

[张辉]

Hello folks,

I was confused by the contrast in the image simulated by QSTEM. Hope you can help me find out what is wrong.  As shown in the first figure, the model comprises Ti, O and C atoms, which are labelled in the figure. However, the average Z number figure is strange and is not consistent with the model as marked by an arrow. What is more, some extra contrast appears at the marked position in the third image (ABF, 0~40 mrad). The xtl and qsc files are attached here. Many thanks for your time and help.

Cheers,

Hui

[Christoph Koch]

Hello Hui,

 

The ‚strange contrast‘ in the projected Z-number may depend on the point spread function (PSF) or number of bins you are using. I suggest to first reduce your PSF to see what is happening, and maybe also increase your number of bins. When confining your structure to a box you have to make sure that there is not accidentally one additional layer of atoms for just a few of the atomic columns. But your BF image indicates that this is not the case. By the way, 0 .. 40 mrad is a BF image, not an ABF image!

 

I also noticed that your probe window is extremely small. You are bound to have aliasing artefacts in your simulations. I guess that increasing the extension of the probe array to at least 10A will remove those artefacts. However, in order to keep the large scattering angles in your simulation you will also have to increase the number of pixels. This will certainly slow down your simulation. If you have a high-end GPU to do the simulation on, you may also try FDES (https://github.com/woutervandenbroek/FDES/tree/light), or another GPU STEM code out there.

 

With best regards,

Christoph.

--
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.

[Christoph Koch]

Hello Hui,

 

one more thing I wanted to add to my first reply: You can reduce the computational cost a lot by computing the STEM image only for a single unit cell and then replicating that unit cell when displaying the results. In your structure, it seems that along the horizontal direction you have included many repetitions of the same motive. For this you have to select an area whose right edge is identical to what is left of its left edge, so that you obtain a smooth transition when replicating the image multiple times along the horizontal direction.

 

Best regards,

Christoph.

 

Von: qs...@googlegroups.com [mailto:qs...@googlegroups.com] Im Auftrag von ??
Gesendet: Samstag, 21. Oktober 2017 03:51
An: QSTEM
Betreff: [QSTEM] strange contrast

 

Hello folks,

--

[张辉]

Hi Prof. Koch,

Thank you so much for your explanation. Indeed, the problem in Z-number is caused by improper PSF value. Really appreciate that you pointed out the wrong collection angle for ABF in my simulation settings, it has been decreased down to 20 mrad. 

Regards,

Hui

[Christoph Koch]

Dear Hui,

 

For actual ABF you want to make the outer angle equal to your convergence angle (i.e. 15 mrad in your case) and the inner angle half that (i.e. 7.5 mrad in your case).

 

With best regards,

Christoph.

--

[张辉]

Dear Prof. Koch,

Thank you so much for your information. According to your previous suggestion, the probe window is set to be slightly larger than 1 nm. The aforementioned strange contrast is absent now. However, the O atoms nearly show no contrast in the simulated haadf image, which is dramatically different from the Z^2 property image, as shown in the attached image. Does it make sense?

Regards,

Hui

[Christoph Koch]

Dear Hui,

 

It is difficult to compare the images on those very different color scales. I suggest that you display them both on the same color scale. You can save the projected Z^2 image in the .img format using the Save data button and then open both, the projected Z^2 and your simulated image using showimage or DM3 (with FRWRtools installed). Only then you can compare properly. However, it is known that  O does not show up in HAADF-STEM images.

 

Best regards,

[张辉]

Dear Prof. Koch,

Thanks for your reply. After adjusting the scale, the contrast of O in the Z^2 and simulated images are still significantly different, as shown in the attached image. But, I believe that O should not show any contrast. 

Regards,

Hui

[Christoph Koch]

It very much depends on what you select as inner angle of your ADF detector. If you, for example, select 70 mrad .. 200 mrad, the O will probably have almost no contrast, especially for thicker specimen.