Avoiding broadened and negative peaks in non-negative matrix factorization: Thermal expansion and background corrections for in-situ diffraction data
Coppedge, Michael
This item is only available for download by members of the University of Illinois community. Students, faculty, and staff at the U of I may log in with your NetID and password to view the item. If you are trying to access an Illinois-restricted dissertation or thesis, you can request a copy through your library's Inter-Library Loan office or purchase a copy directly from ProQuest.
Permalink
https://hdl.handle.net/2142/120324
Description
Title
Avoiding broadened and negative peaks in non-negative matrix factorization: Thermal expansion and background corrections for in-situ diffraction data
Author(s)
Coppedge, Michael
Issue Date
2022-12-19
Director of Research (if dissertation) or Advisor (if thesis)
Shoemaker, Daniel P
Department of Study
Materials Science & Engineerng
Discipline
Materials Science & Engr
Degree Granting Institution
University of Illinois at Urbana-Champaign
Degree Name
M.S.
Degree Level
Thesis
Keyword(s)
X-ray diffraction
non-negative matrix factorization
machine learning
temperature correction
Abstract
Here it is demonstrated with both simulated and experimental in situ X-ray diffraction data that a “correction” of the peak shift due to thermal expansion and an unmodeled, global background subtraction improve the results of non-negative matrix factorization (NMF). NMF is one of a group of matrix decomposition algorithms which has seen increasing use in the automatization of the processing and interpretation of large sets of experimental data. This is due mostly to its non-negativity constraint, which causes it to have a higher likelihood of returning physically meaningful information compared to other decomposition algorithms. The nature of in situ diffraction experiments does not lend itself perfectly to the use of NMF though, as any changes in the constituent phases of the experiment that result in horizontal peak shift – that is, almost any change that would be induced in an in situ experiment – is poorly handled by NMF, hence the motivation for removing thermal peak shift prior to calling NMF. Performing both added steps – thermal correction and unmodeled background subtraction – are computationally inexpensive, completely automatic, and totally general – they can be applied to any set of in situ data. These measures are particularly relevant for high-energy synchrotron data, where the thermal peak shift relative to the peak widths is large – an especially challenging but very common use case for NMF. In all examples studied here it is found that thermal correction and background subtraction aid the NMF algorithm in returning phase components that better represent the underlying data quality and in affording superior phase fraction evolution information. In particular, without temperature correction, NMF returns phase components that have wider peaks than those of the raw data. This is especially egregious for high resolution data, where, as is demonstrated with a simulated experiment, NMF peaks can be up to 75 times wider than the original data.
Use this login method if you
don't
have an
@illinois.edu
email address.
(Oops, I do have one)
IDEALS migrated to a new platform on June 23, 2022. If you created
your account prior to this date, you will have to reset your password
using the forgot-password link below.
