Getting More from Your 3D ED data: Dynamical Refinement in Jana2020

This is a written summary of a live webinar presented on November 18, 2026. The recording and resources are available on the recording page.
Presented by:
Dr.rer.nat. Lukáš Palatinus
Head of the Laboratory of electron crystallography
Czech Academy of Sciences in Prague
Connect on LinkedInWebinar summary
This webinar discusses dynamical refinement, which models electron diffraction by accounting for multiple scattering within a crystal. Compared with kinematical refinement, which assumes each measured intensity depends only on the squared structure factor amplitude, this approach can produce a better fit, more accurate structural models, greater sensitivity to weak signals, and absolute structure determination. It also takes more computing time, depends more on crystal quality, and requires additional refinement choices.
JANA 2020 supports dynamical refinement alongside other crystallographic methods. The process uses a model of the crystal structure, its orientation, and its thickness to calculate intensities. Because calculated intensities depend on crystal orientation, symmetry-related reflections are retained individually rather than merged before refinement. Multiple datasets can be refined together against one structural model, but their intensities are kept separate.
A practical workflow begins by importing processed electron diffraction data and a structural model, then estimating crystal thickness by examining how the fit changes with thickness. Refinement can then proceed, with choices such as refining on structure factor amplitude rather than its square to improve robustness against outliers. Other checks include identifying poor diffraction frames, manually removing clear outlier reflections, and reviewing normalized differences to assess whether the weighting of strong reflections needs adjustment.
The examples show that dynamical refinement can be practical for both inorganic and organic structures. Runtime depends on the dataset and chosen settings. Organic structures can be more challenging because they often have lower resolution and poorer crystal quality; absorption corrections are generally useful for heavy atoms, but not for organic compounds. Goodness of fit should be considered alongside R factors and the distribution of normalized differences. Post-refinement merging can produce R factors that are more comparable with conventional kinematical results.
Frequently asked questions
-
Kinematical refinement calculates intensities using the squared structure factor amplitude. Dynamical refinement uses dynamical diffraction theory to account for multiple scattering of electrons within the sample. This often improves the fit and can make structural details more reliable, though it takes longer and is more sensitive to crystal quality.
-
The calculated scattering depends on the distance electrons travel through the crystal. Thickness is therefore part of the intensity calculation. An initial thickness estimate can be obtained by calculating the fit over a range of thicknesses and finding a minimum in the R factor.
-
Their measured intensities can differ because the intensity depends on the crystal’s orientation when each reflection is recorded. Refinement therefore uses the individual observations. Several separate datasets can be refined together against one structural model, while keeping each dataset’s intensities distinct.
-
The workflow described uses processed electron diffraction data and a structural model, such as a CIF file. After importing both, the initial setup includes estimating thickness and choosing refinement settings.
-
Poor diffraction frames can be excluded from the refinement, and clear outlier reflections can be removed through manual culling. The distribution of normalized differences can also help identify whether the strongest reflections are underweighted and whether the weighting scheme should be adjusted.
-
Yes. The technical approach is similar to that used for inorganic structures, and a fully free refinement of an organic structure was demonstrated. Organic crystals can be harder to refine because they often have lower resolution and poorer crystal quality. Absorption corrections are generally not useful for organic compounds.
-
They should be considered together with general refinement statistics, including the distribution of normalized differences. Goodness of fit can be relatively high for inorganic materials; values around 2.5 or below were described as generally acceptable in practice, while values of 4 to 6 were cause for concern. R factors from dynamical and kinematical refinement are not directly comparable because dynamical refinement uses unmerged data.
-
Yes. Dynamical refinement can provide absolute structure determination, which kinematical refinement cannot. It can also achieve very low R factors for exceptionally good crystals, though such results are unusual.
Subscribe to the Crystallography Times newsletter
Stay up to date with single crystal analysis news and upcoming events, learn about researchers in the field, new techniques and products, and explore helpful tips.
Contact Us
Whether you're interested in getting a quote, want a demo, need technical support, or simply have a question, we're here to help.