/GISANS - Advancing data reduction and analysis Workshop

From canSAS
Workshop programme

Summary

Overall Summary

Presentations and discussion were centred on distinct themes that were recognised as being strongly related. Ideas that arose are presented in numbered sections that relate to the organisation of the discussion sessions. These are strongly inter-related and identify areas where further work is needed.

(i) Special issues relating to time-of-flight (ToF) measurements

  • Instrument factors influence measurement strategies and tools are needed to aid this process.
  • Background - effects of inelastic scattering can change background signals that would not be identified/recorded correctly with respect to wavelength by time-of-flight.
  • Resolution is important regarding both wavelength and angle. The latter depends on collimation/beam divergence as well as footprint on the sample and spatial resolution of the detector. The footprint depends on angle of incidence. The footprint gives some spatial and angular resolution projected on the detector.

Full interpretation of data is likely to require extensive information to be available with processed data.

  • Data reduction from raw data will require establishment of appropriate data formats for input to existing modelling and analysis software. The software may also need development to fully exploit the information that is available.
  • ToF and monochromatic mode have different data processing requirements.
    • Spread of wavelength in ToF. This influences scattering strongly.
    • There are possible inelastic and quasi-elastic scattering events that complicate normalisation with respect to the incident spectral distribution.
    • Monochromatic measurements often make assumption of elastic scattering but even for these spectral spread is important and information may be required for analysis.
    • Inelastic scattering mostly impacts on background.

(ii) Multiple scattering

  • Reflection and refraction (optical effects) should be distinguished from multiple scattering
  • There is a need to agree on a common nomenclature
    • “Coherent multiple scattering” can refer to the situation in which successive scattering events are not treated independently but are combined quantum mechanically, i.e., the amplitudes from different scattering paths are added coherently. This use of "coherence" has to be distinguished from others such as simple spin-coherence widely discussed in neutron scattering.
    • Sub-cases of multiple coherent scattering: (i) propagation of the incident wave through stratified media, which is incorporated exactly in the optical formalism treating reflection and refraction; and (ii) scattering from embedded “particles” (additional scattering centers) which is treated as a perturbation within the 1st-order Born approximation. Both (i) and (ii) are within the quantum theory (‘coherent’).
    • “Incoherent multiple scattering” denotes an approximation in which scattering from each center is treated, but the subsequent propagation of scattered waves is represented within optical picture (Gaussian optics) where phase relations between successive scattering events are neglected. This approach is conceptually analogous to the semi-classical treatment of transport phenomena based on the Boltzmann equation. The result corresponds to summing intensities rather than amplitudes of scattered waves.
  • There are several factors that influence multiple scattering: mean free path length & coherence length, sample size / geometry:
    • They have to be identified before each experiment
  • Multiple scattering is often present in background signals. It can be used also to explicitly enhance signal
  • The comparison of X-rays and neutrons is helpful to identify important features:
    • X-rays:
      • There is (mostly) strong absorption that is dominant.
    • Neutrons:
      • Absorption is often weak and therefore, scattering is relative to absorption, of higher intensity.
      • Hydrogen is a particularly strong incoherent scattering nucleus
    • It is important to be careful when comparing models and the theory for GISAXS and GISANS
  • How to measure MS? It would be valuable to develop ways to measure the extent of multiple scattering.
    • Sample geometry is important for the extent of this effect.
  • For multiple scattering it can be useful to include Monte Carlo calculations such as McStas as a tool for GISANS analysis.

(iii) Background handling

  • Background subtraction is dependent on:
    • sample system/geometry/the observable physics.
  • Each system has to be treated in a different way.
  • Useful recent work (F. A. Jung and C. M. Papadakis, 'Strategy to simulate and fit 2D grazing-incidence small-angle X-ray scattering patterns of nanostructured thin films' J. Appl. Cryst. (2023) 56, 1330–1347. https://doi.org/10.1107/S1600576723006520 ) regarding strategy to handle background for fits and simulations to GISAXS patterns.
  • A survey would help to identify the necessary / usual steps used by the community for specific systems.

(iv) Normalization

  • GISANS contrasts with both reflectivity measurements and conventional transmission small-angle scattering.
    • Reflectivity is usually presented as data normalised to the incident beam intensity )i.e. as a simple ratio
    • Small-angle scattering data are usually discussed in terms of the differential scattering cross-section. This depends on the amount of sample in the beam and in transmission measurements of uniform samples requires normalisation by the sample thickness in the beam direction as well as the incident intensity.
  • Full quantitative normalization procedure:
    • measuring GISANS together with NR & Off-specular scattering, to view the whole Q-space.
      • In NR: fit the data+background to the total signal (instead of subtracting!)
  • If a routine or standard procedure is not feasible:
    • Each system has to be classified the way "how to be treated" - Survey needed! (Different normalization procedures at different beamlines and for different classes of systems)
  • General aspects (sample independent):
    • Precise Intensity with wavelength (lambda) normalization of the beamlines has to be performed
    • Good to get a Survey "How is this done at different beamlines currently"?


  • Data reduction:
    • Apply corrections for efficiency etc.,
    • Uncertainty calculation reproducible?
      • Information on footprint, slits,
      • Provide data for further input to uncertainty apart from count rate. These factors can be as large or more significant than counting statistics.
    • Comparison to SAS:
      • As described under normalization (above) there needs to be clarity about what has been done to the data as intensity scaling is necessarily different to transmission SAS (and reflectivity measurements)


(v) Using SANS and other software

  • Discussion considered whether analysis of GISANS and GISAXS could take advantage of software developed for SANS and SAXS or other studies.
    • There are components developed to calculate many different form factors and structure factors that have been used and verified. See e.g. https://marketplace.sasview.org/
    • At many instruments regular SAS software is used at present for data reduction such as correction for detector efficiency.
  • Regarding BornAgain (https://www.bornagainproject.org):
    • Need a set of prototype models (check what is on the BornAgain webpage) - especially for inexperienced users
      • This should include a very detailed description about its physical limitations, directly implemented into BA sending error messages if beyond range.
      • Can we benefit from models in other software packages?
      • In SasView: easy user-insertions of new form-factors - learn from that?
      • Package approach in BornAgain problematic?
    • Further good features to implement: Simulation of (i) intensity with respect to direct beam, (ii) with respect to multiple scattering background, (iii) with respect to general expected background (according to experience).
    • How to get fitting for GISANS working?
  • Inspiration from X-ray programs: e.g. at Brookhaven. There is a catalogue of software published: https://gisaxs.com/index.php/Software . Reflectivity analysis: BoToSim, BoToFit (https://www.ill.eu/for-all-users/instruments/instruments-list/superadam/software ), Magnetic structures: SpinW (https://spinw.org )
    • Using AI?: Requires information (i) published datasets that are similar to the experiments and (ii) what the dataset represents (physically)

(vi) Common data format

  • There are 'Pros' and 'Cons' of separating "reduction" and "analysis", i.e., having an instrument independent analysis without all the original raw information
    • Pro:
      • Datasets are transferable between software.
    • Con:
      • This might lead to 'Throw away information?' that is needed later
      • Have to understand the inputs and what needs to be kept
      • How to calculate uncertainties?
    • Once defined, file formats should not be altered, only extended (e.g. possible in nexus or canSAS XML formats). This should maintain backwards compatibility
      • Strictly one does not "throw away" raw counts after reduction, but keep all information in raw files. However, it is necessary to define what is needed for data analysis so that appropriate data and meta data are provided in reduced files. If suitable / manageable: Ask in Survey?
      • The contents of reduced files should reflect the suggestions and further input with respect to data reduction, normalisation, background, etc.

(vii) Magnetic Scattering

  • Considering ".nxs" files and dimensions: a scan of anything/motor (e.g. temperature, field, sample position, etc.) should be considered a dimension in the "nxs". It should not be restricted to 4 dimensions.
  • On-the-fly polarization analysis: online corrections are good. But highest precision correction should be done at the end of the experiment before the users go home.


Outlook and work to be done

  • Develop small work groups to develop recommendations for (a) common treated data format, and (b) procedures for data reduction, normalisation, and handling background
  • Prepare article for "Neutron News"
  • Need a survey on : (i) Normalization, Background handling, Instrumentation and (ii) Data formats

Detailed Discussion session notes

The presentation slides below were used to introduce the topics for discussion.

POINT A. ToF-GISANS

  • Presentation slides: To come
  • Further Notes from the ILL-Workshop on: /ToF-GISANS

POINT B. Multiple scattering

  • Presentation slides:

File:HF MuSc GISANS.pdf

POINT C. Background handling

  • Presentation slides:

File:Background ideas.pdf

POINT D. Normalization

  • Presentation slides:

File:GISANS-NormalizationDiscussionStarters.pdf

POINT E. Using SANS software

POINT F / Instrumentation & Q/lambda resolution

  • Presentation slides:

File:2026 03 CANSAS Grenoble Resol Fn JFMoulin.pdf File:TopervergGISANS2026ILL1-part1.pdf File:TopervergGISANS2026ILL1-part2.pdf

POINT G. Need good documentation

  • No Introduction talk
  • Further Notes from the ILL-Workshop on: /Documentation

POINT H. Integration of McStas with BornAgain

  • Presentation slides:

File:HF McStas.pdf

  • Further Notes from the ILL-Workshop on: /McStas

POINT I. Get a common data format for GISANS data reduction:

  • Presentation slides:

File:GISANS 2026 File Formats - Brian Maranville.pdf


POINT J. Computer simulations / AI support:

POINT K. Off-specular scattering:

  • Presentation slides:

File:Gutfreund OffSpec new part 1.pdf File:Gutfreund OffSpec new part 2.pdf

POINT L. Magnetic Scattering:

  • Presentation slides:

File:Magnetic GISANS.pdf

POINT M. Data Reduction:

  • Presentation slides:

File:2026 03 CANSAS Grenoble Data Reduction TOF JFMoulin.pdf File:2026 03 12 Monochromatic GISANS reduction v4.pdf

Organizational notes

Retrieved from "https://wiki.cansas.org/index.php?title=/GISANS_-_Advancing_data_reduction_and_analysis_Workshop&oldid=2937"