Humboldt-Universität zu Berlin - Faculty of Mathematics and Natural Sciences - Strukturforschung / Elektronenmikroskopie


This page contains a list of talks regarding image simulation and the use of qstem for various purposes.  This page is not intended as a sole source for TEM/STEM image simulation knowledge.  The interested reader is referred to the existing literature (e.g. the book “Advanced Computing in Electron Microscopy” by Earl Kirkland) for a thoroughly prepared introduction to the topic.

qstem FAQs

  • Is there any way to simulate the STEM images of the solid solution? For example, (A1-xBx)O2.

    Yes, in the config file you simply need to define the A-atoms with an occupancy of 1-x and the B atoms with occupancy x and give them the exact same position. There are now two options how the simulation is run if you include TDS in your simulation (without TDS both options will yield equivalent results):
    a) If you want to simulate the average image, then you simply generate the super cell within qstem itself (either box mode or replication of unit cells). For each TDS run, the atom positions will be distributed between A and B randomly. So each TDS run sees a different configuration, but with the same statistical occupancy of A and B.
    b) If you want to generate a single (random) configuration and only thermally displace the atoms in that, you should generate the super cell in the qstem model builder, and then load the complete super cell into qstem. Qstem will then only wobble the atoms about their positions, but will not re-evaluate the occupancy again.
    More on this can be found in the description of the .cfg file format

  • We have been successful in running the program for pure fcc material like Al or Ag along the [100] zone axis.  However, when we change zone axis to [110]  we seem to get the HRTEM images for a cubic shape particle instead of a thin specimen with parallel surfaces.  What is the correct way to do the simulation along a direction other than the <100> directions?
    In the model section of the qstem GUI you should switch from Ncells mode to Box mode.  You can then specify the size (width, depth, height) of your  model in A.  Qstem will then fill the box that you define in this way with the crystal structure in the specified orientation.  It is up to you to make sure that this box has periodic boundary conditions, if this is what you want.  For a STEM simulation your model will usually be larger than what you need, so you do not need to worry about periodic boundaries.
    Alternatively, you can also use qmb or gbmaker to define more complex unit cells.
  • It seems that Matlab display single pixel coordinates and intensity, by clicking on the desired pixel. Is it possible to export the images in a text format, in order to obtain all pixel coordinates and associated intensity ?
    You can download the FRWRtools plugin for Gatan DigitalMicrograph.  Using the ‘read image’ function in the list of menu items that this plugin provides you can import any .img image or wave function to DigitalMicrograph and process it there (including viewing it in tabular form).  You can also use the Matlab functions I provided for reading the .img format to import these images into Matlab, or write your own function in your favourite software/programming language, using the .img data format specifications.
  • I’m really interested in your S(TEM) simulation software QSTEM. However, I have problem importing “cif” files as starting model for simulation.
    Please take a look at convert2cfg.
  • How does qstem handle the Debye Waller factor? I am confused about the difference between DW factor and B factor in standard CIF files.  The other question is that in the frozen phonon algorism, how is the random displacement related to the DW factor in the cfg file?
    The exact definition of the Debye-Waller factor is DWf=exp(-B*s^2).  But since this is a function of the scattering angle s it is more convenient to just specify B, and this is exactly what qstem does.  I admit, calling it DW-factor is not very accurate, but most people understand what it means.
    If you look at the origin of B it becomes obvious how this relates to the vibration of atoms.  If you select TDS to be included then no Debye-Waller factor will be applied to the atomic potentials, but instead the atoms will be shaken with a 3D Gaussian-distributed position offset u such that B=8*pi^2*<u^2>, where <u^2> is the mean of the squared displacements.
  • I would like to ask you a question regarding the TEM Image Formation program included in the QSTEM package:
    Does the value that one can insert in the field “Amorphous” have any relation to a physical quantity such as the substrate thickness? In my special case this value may alter the resulting image as I try to simulate small clusters on a 3nm carbon substrate, so that sometimes the particle diameter is only littel larger than the substrate thickness. So if, for example a value of “3″ in the amorphous field would correspond to the noise resulting from a 3nm amorphous carbon, I would be able to better adapt this value.
    The amorphous parameter is NOT related to some layer thickness.  All this does is adding random numbers between 0 and (in your example) 3 to the phase of the previously computed exit wave.  This has the effect of producing nicely visible Thon rings in the FFT of the image, but should not be used for the simulation of actual amorphous material.  If you want to simulate amorphous material, I suggest to use gbmaker (instructions can be found in the file ‘Au_Al2O3.gbm’ in the Examples folder).

Specification of QSTEM data formats (link)