際際滷

際際滷Share a Scribd company logo
A Matlab tool for multivariate analysis of ToF-SIMS datasets
Wednesday, 20 September 2017 1
Gustavo Ferraz Trindade
The Surface Analysis Laboratory, University of Surrey, UK
s i m s M V A
Outline
Wednesday, 20 September 2017 2
Introduction Examples Other techniques
Data analysis methodologies?
Wednesday, 20 September 2017 3
Surface analysis
expertise
Good
instrumentation
Sample knowledge
Sample prep.
Data handling
Data processing
Data visualisation
Good quality data
 File formats
 Import/export
 Memory
management
 Algorithms
 Error analysis
 Software
 Tables
 Plots
 Overlays
INTERPRETATION
Wednesday, 20 September 2017
Very well established at Surrey
Very good literature in SIMS community
NPLs
good practices2
SIA special issue
on MVA1
1February 2009. Volume 41, Issue 2. Pages 75142
2www.npl.co.uk/upload/pdf/chemometrics.pdf
Great information on data processing
and interpretation, not as much on
data handling and visualisation
Data analysis methodologies?
4
Wednesday, 20 September 2017 5
Initial goal of my PhD: Learn and apply
those literature recommendations to data of
industrial samples* typically analysed at
Surreys Surface Analysis Lab.
*paints, adhesives, automotive polymers, additives
Introduction
Wednesday, 20 September 2017 6
Fast forward 3 years
Introduction
Wednesday, 20 September 2017 7
s i m s M V A
Introduction
s i m s M V A
Wednesday, 20 September 2017 8
s i m s M V A is a Matlab-based app for
multivariate analysis of analytical data
(focus on ToF-SIMS)
Developed throughout my PhD,
motivated mainly by the idea of making
MVA quick and accessible to everyone in
the research group
PCA, NMF, k-means of different data
structures (spectra, images, depth
profiles, 3D)
s i m s M V A | Examples
Wednesday, 20 September 2017 9
Three examples from Surreys Surface Analysis Lab. were chosen
to show how simsMVA helped to solve problems or perhaps how
problems helped the development of simsMVA
1) Imaging of
a resin blend
2) Characterisation of
wood growth regions
3) Steel at high
temperature
s i m s M V A | Industrial coating
Wednesday, 20 September 2017 10
1  Imaging of a resin blend
Data from Surreys PhD student Ms Rene Tshulu
Industrial coating is cut using ultra low angle microtomy, exposing/extending interfaces*
DOI 10.1002/sia.1985
*
~ 1000 亮m
s i m s M V A | Large area analysis
Wednesday, 20 September 2017 11
The area covering both interfaces is greater than the
primary ion beam raster size (500x500um2)
Several patches are needed (large area imaging)
Automated acquisition is tricky due to charging and
topography
Solution: create separate datasets and stitch them together
afterwards using s i m s M V A
1  Imaging of a resin blend
s i m s M V A | Images stitching procedure
Wednesday, 20 September 2017 12
.BIF6 files exported from IONTOFs software
Wednesday, 20 September 2017 13
Normalised by total
counts
Poisson scaled MVA sampling 10% of
pixels*
*Low discrepancy subsampling as described in: 10.1002/sia.6042
Images mode
NMF Menu
s i m s M V A | Images tab
Wednesday, 20 September 2017 14
Loadings / NMF spectra
Scores / NMF weights
MVA results tab
Variance / Error
s i m s M V A | MVA results tab
Wednesday, 20 September 2017 15
NMF spectraOverlayed NMF weights
Overlay window
s i m s M V A | Overlay window
Resin 1
Resin 2
Chromium layer
Top surface
Zinc substrate
Wednesday, 20 September 2017 16
Top surface
s i m s M V A
Substrate
Wednesday, 20 September 2017 17
One single map gives a very complete initial description of the
sample, highlighting the two phases of the blend, the top surface
and the metal substrate with a chromate layer
- What if we have more samples?
- Which resin is which in the blend?
- How to discard topographic effects hypothesis?
To expand the analysis even further, we can stitch measurements
of more than one sample together, and also include standards
s i m s M V A | Examples
Wednesday, 20 September 2017 18
s i m s M V A | Data stitching schematic
Factorised columns are folded
back for visualisation
G. F. Trindade et. al. Submitted to Anal. Chem.
Wednesday, 20 September 2017 20
Overlay window
s i m s M V A | Examples
G. F. Trindade et.
al. Submitted to
Anal. Chem.
Cured
resin 2
Cured
resin 1
Uncured
resin 1
Uncured
resin 2
SAMPLE 1
SAMPLE 3
SAMPLE 2
SAMPLE 4
(3) Resin 1
(6) Resin 2
Wednesday, 20 September 2017 21
Surface mass spectra of different wood growth regions*
PLUS standard lignin and cellulose samples
Carefully selected peaks list for MVA (with characteristic peaks of cellulose, lignin and extractives)
DOI: 10.1002/sia.59154
2  Characterisation of wood samples
s i m s M V A | Examples
*
Wednesday, 20 September 2017 22
Overlay windowData table
Poisson scaled
and
Mean centredVariables correlations
can be checked prior
to PCA
Spectra mode
Many variables selectedA few variables selected
PCA
s i m s M V A | Correlation matrix
Ionsofinterest
Ions of interest
Wednesday, 20 September 2017 23
Overlay window
Spectra mode
Loadings
Scores
Variance
Lignin peaks
Cellulose peaks
s i m s M V A | MVA results (Spectra)
Wednesday, 20 September 2017 24
Overlay window
Spectra mode
Biplot shows loadings and
scores in the same plot
K-means finds categories
s i m s M V A | Scores visualisation
Wednesday, 20 September 2017 25
Stainless steel heated up in vacuum using heating/cooling stage of a TOF.SIMS 5 (Iontof)
Secondary ions images acquired periodically to create a data cube
3  Stainless steel at high temperature
s i m s M V A
SAMPLE
HEAT
TOF-SIMS
Crater
with DC
beam
Scratch
using blade
DC beam removes the
oxide layer that regrows
after heating
Wednesday, 20 September 2017 26
Before any MVA, two crucial pre-processing steps
s i m s M V A | Examples
Peak background removal Pixel warping
Wednesday, 20 September 2017 27
Use simsMVA 3D mode to process
s i m s M V A | Examples
sub sampling of
voxels
NMF
Wednesday, 20 September 2017 28
Use simsMVA 3D mode to process
s i m s M V A | Examples
Video button to create a
multivariate film
Wednesday, 20 September 2017 29
s i m s M V A | Examples
Chamber Pressure
Temperature
1 hour
Wednesday, 20 September 2017 30
s i m s M V A | Examples
27Al+
28Si+
52Cr+
55Mn+
Aided by results, it is
possible to go back to
original data and look at
specific ions
Chamber Pressure
Temperature
1 hour
Wednesday, 20 September 2017 31
A few recent examples
s i m s M V A | Examples
Wednesday, 20 September 2017 32
s i m s M V A | Examples
Add M. Baileys
talk reference here
Large area imaging of fingerprints for drug detection
Invited talk yesterday: Melanie Bailey, University of Surrey,
England - "ToF-SIMS for imaging latent fingerprints".
Wednesday, 20 September 2017 33
s i m s M V A | Examples
Depth profile of a multi-layered sample
Wednesday, 20 September 2017 34
Large area imaging of plasma treated polymer
s i m s M V A | Examples
G. F. Trindade et. al. Submitted to SIA ECASIA SPECIAL ISSUE
Wednesday, 20 September 2017 35
Characterisation of coffee
s i m s M V A | Examples
Depth profile of a solar cell
G. F. Trindade et. al.
Submitted to SIA
ECASIA SPECIAL
ISSUE
D. Luo et. al. Submitted to Science
Wednesday, 20 September 2017 36
s i m s M V A can also be used to analyse other
analytical chemistry data
Data matrices can be loaded from Matlab
workspace
s i m s M V A | Other analytical techniques
Wednesday, 20 September 2017 37
s i m s M V A | Other analytical techniques
Al+
Si+
Cr+
Al+
Go back and look at
individual ions
FTIR of polyesters
Wednesday, 20 September 2017 38
s i m s M V A | Other analytical techniques
Al+
Si+
Cr+
Al+
Go back and look at
individual ions
PIXE mapping of a fossil
Wednesday, 20 September 2017 39
s i m s M V A | Other analytical techniques
Al+
Si+
Cr+
Al+
Go back and look at
individual ions
XPS depth profile
s i m s M V A | Find out more
Wednesday, 20 September 2017 40
To know more about simsMVA visit
mvatools.com
- Ms Rene Tshulu
- Mr Jorge Banuls-Ciscar
- Dr. Taraneh B. Moghim
- Dr. Sabrina Tardio
- Dr. Robin Simpson
- Mr Kristof Marcoen
- Mr Min Jang
Acknowledgements
Supervisors
- Prof. John F. Watts
- Dr. Marie-Laure Abel
Sponsors
- CAPES
- Science without borders
- UKSAF
Thank you for your attention!
mvatools.com
simsMVA testers

More Related Content

simsMVA: A Matlab tool for multivariate analysis of ToF-SIMS datasets

  • 1. A Matlab tool for multivariate analysis of ToF-SIMS datasets Wednesday, 20 September 2017 1 Gustavo Ferraz Trindade The Surface Analysis Laboratory, University of Surrey, UK s i m s M V A
  • 2. Outline Wednesday, 20 September 2017 2 Introduction Examples Other techniques
  • 3. Data analysis methodologies? Wednesday, 20 September 2017 3 Surface analysis expertise Good instrumentation Sample knowledge Sample prep. Data handling Data processing Data visualisation Good quality data File formats Import/export Memory management Algorithms Error analysis Software Tables Plots Overlays INTERPRETATION
  • 4. Wednesday, 20 September 2017 Very well established at Surrey Very good literature in SIMS community NPLs good practices2 SIA special issue on MVA1 1February 2009. Volume 41, Issue 2. Pages 75142 2www.npl.co.uk/upload/pdf/chemometrics.pdf Great information on data processing and interpretation, not as much on data handling and visualisation Data analysis methodologies? 4
  • 5. Wednesday, 20 September 2017 5 Initial goal of my PhD: Learn and apply those literature recommendations to data of industrial samples* typically analysed at Surreys Surface Analysis Lab. *paints, adhesives, automotive polymers, additives Introduction
  • 6. Wednesday, 20 September 2017 6 Fast forward 3 years Introduction
  • 7. Wednesday, 20 September 2017 7 s i m s M V A Introduction
  • 8. s i m s M V A Wednesday, 20 September 2017 8 s i m s M V A is a Matlab-based app for multivariate analysis of analytical data (focus on ToF-SIMS) Developed throughout my PhD, motivated mainly by the idea of making MVA quick and accessible to everyone in the research group PCA, NMF, k-means of different data structures (spectra, images, depth profiles, 3D)
  • 9. s i m s M V A | Examples Wednesday, 20 September 2017 9 Three examples from Surreys Surface Analysis Lab. were chosen to show how simsMVA helped to solve problems or perhaps how problems helped the development of simsMVA 1) Imaging of a resin blend 2) Characterisation of wood growth regions 3) Steel at high temperature
  • 10. s i m s M V A | Industrial coating Wednesday, 20 September 2017 10 1 Imaging of a resin blend Data from Surreys PhD student Ms Rene Tshulu Industrial coating is cut using ultra low angle microtomy, exposing/extending interfaces* DOI 10.1002/sia.1985 * ~ 1000 亮m
  • 11. s i m s M V A | Large area analysis Wednesday, 20 September 2017 11 The area covering both interfaces is greater than the primary ion beam raster size (500x500um2) Several patches are needed (large area imaging) Automated acquisition is tricky due to charging and topography Solution: create separate datasets and stitch them together afterwards using s i m s M V A 1 Imaging of a resin blend
  • 12. s i m s M V A | Images stitching procedure Wednesday, 20 September 2017 12 .BIF6 files exported from IONTOFs software
  • 13. Wednesday, 20 September 2017 13 Normalised by total counts Poisson scaled MVA sampling 10% of pixels* *Low discrepancy subsampling as described in: 10.1002/sia.6042 Images mode NMF Menu s i m s M V A | Images tab
  • 14. Wednesday, 20 September 2017 14 Loadings / NMF spectra Scores / NMF weights MVA results tab Variance / Error s i m s M V A | MVA results tab
  • 15. Wednesday, 20 September 2017 15 NMF spectraOverlayed NMF weights Overlay window s i m s M V A | Overlay window Resin 1 Resin 2 Chromium layer Top surface Zinc substrate
  • 16. Wednesday, 20 September 2017 16 Top surface s i m s M V A Substrate
  • 17. Wednesday, 20 September 2017 17 One single map gives a very complete initial description of the sample, highlighting the two phases of the blend, the top surface and the metal substrate with a chromate layer - What if we have more samples? - Which resin is which in the blend? - How to discard topographic effects hypothesis? To expand the analysis even further, we can stitch measurements of more than one sample together, and also include standards s i m s M V A | Examples
  • 18. Wednesday, 20 September 2017 18 s i m s M V A | Data stitching schematic Factorised columns are folded back for visualisation G. F. Trindade et. al. Submitted to Anal. Chem.
  • 19. Wednesday, 20 September 2017 20 Overlay window s i m s M V A | Examples G. F. Trindade et. al. Submitted to Anal. Chem. Cured resin 2 Cured resin 1 Uncured resin 1 Uncured resin 2 SAMPLE 1 SAMPLE 3 SAMPLE 2 SAMPLE 4 (3) Resin 1 (6) Resin 2
  • 20. Wednesday, 20 September 2017 21 Surface mass spectra of different wood growth regions* PLUS standard lignin and cellulose samples Carefully selected peaks list for MVA (with characteristic peaks of cellulose, lignin and extractives) DOI: 10.1002/sia.59154 2 Characterisation of wood samples s i m s M V A | Examples *
  • 21. Wednesday, 20 September 2017 22 Overlay windowData table Poisson scaled and Mean centredVariables correlations can be checked prior to PCA Spectra mode Many variables selectedA few variables selected PCA s i m s M V A | Correlation matrix Ionsofinterest Ions of interest
  • 22. Wednesday, 20 September 2017 23 Overlay window Spectra mode Loadings Scores Variance Lignin peaks Cellulose peaks s i m s M V A | MVA results (Spectra)
  • 23. Wednesday, 20 September 2017 24 Overlay window Spectra mode Biplot shows loadings and scores in the same plot K-means finds categories s i m s M V A | Scores visualisation
  • 24. Wednesday, 20 September 2017 25 Stainless steel heated up in vacuum using heating/cooling stage of a TOF.SIMS 5 (Iontof) Secondary ions images acquired periodically to create a data cube 3 Stainless steel at high temperature s i m s M V A SAMPLE HEAT TOF-SIMS Crater with DC beam Scratch using blade DC beam removes the oxide layer that regrows after heating
  • 25. Wednesday, 20 September 2017 26 Before any MVA, two crucial pre-processing steps s i m s M V A | Examples Peak background removal Pixel warping
  • 26. Wednesday, 20 September 2017 27 Use simsMVA 3D mode to process s i m s M V A | Examples sub sampling of voxels NMF
  • 27. Wednesday, 20 September 2017 28 Use simsMVA 3D mode to process s i m s M V A | Examples Video button to create a multivariate film
  • 28. Wednesday, 20 September 2017 29 s i m s M V A | Examples Chamber Pressure Temperature 1 hour
  • 29. Wednesday, 20 September 2017 30 s i m s M V A | Examples 27Al+ 28Si+ 52Cr+ 55Mn+ Aided by results, it is possible to go back to original data and look at specific ions Chamber Pressure Temperature 1 hour
  • 30. Wednesday, 20 September 2017 31 A few recent examples s i m s M V A | Examples
  • 31. Wednesday, 20 September 2017 32 s i m s M V A | Examples Add M. Baileys talk reference here Large area imaging of fingerprints for drug detection Invited talk yesterday: Melanie Bailey, University of Surrey, England - "ToF-SIMS for imaging latent fingerprints".
  • 32. Wednesday, 20 September 2017 33 s i m s M V A | Examples Depth profile of a multi-layered sample
  • 33. Wednesday, 20 September 2017 34 Large area imaging of plasma treated polymer s i m s M V A | Examples G. F. Trindade et. al. Submitted to SIA ECASIA SPECIAL ISSUE
  • 34. Wednesday, 20 September 2017 35 Characterisation of coffee s i m s M V A | Examples Depth profile of a solar cell G. F. Trindade et. al. Submitted to SIA ECASIA SPECIAL ISSUE D. Luo et. al. Submitted to Science
  • 35. Wednesday, 20 September 2017 36 s i m s M V A can also be used to analyse other analytical chemistry data Data matrices can be loaded from Matlab workspace s i m s M V A | Other analytical techniques
  • 36. Wednesday, 20 September 2017 37 s i m s M V A | Other analytical techniques Al+ Si+ Cr+ Al+ Go back and look at individual ions FTIR of polyesters
  • 37. Wednesday, 20 September 2017 38 s i m s M V A | Other analytical techniques Al+ Si+ Cr+ Al+ Go back and look at individual ions PIXE mapping of a fossil
  • 38. Wednesday, 20 September 2017 39 s i m s M V A | Other analytical techniques Al+ Si+ Cr+ Al+ Go back and look at individual ions XPS depth profile
  • 39. s i m s M V A | Find out more Wednesday, 20 September 2017 40 To know more about simsMVA visit mvatools.com
  • 40. - Ms Rene Tshulu - Mr Jorge Banuls-Ciscar - Dr. Taraneh B. Moghim - Dr. Sabrina Tardio - Dr. Robin Simpson - Mr Kristof Marcoen - Mr Min Jang Acknowledgements Supervisors - Prof. John F. Watts - Dr. Marie-Laure Abel Sponsors - CAPES - Science without borders - UKSAF Thank you for your attention! mvatools.com simsMVA testers

Editor's Notes

  • #3: - Fazer um rapido overview da apresentacao. Quando falar de resultados mencionar que foram feitas analyses SEM, EDX, XPS e SIMS.
  • #4: - Fazer um rapido overview da apresentacao. Quando falar de resultados mencionar que foram feitas analyses SEM, EDX, XPS e SIMS.
  • #5: - Fazer um rapido overview da apresentacao. Quando falar de resultados mencionar que foram feitas analyses SEM, EDX, XPS e SIMS.
  • #6: - Fazer um rapido overview da apresentacao. Quando falar de resultados mencionar que foram feitas analyses SEM, EDX, XPS e SIMS.
  • #7: - Fazer um rapido overview da apresentacao. Quando falar de resultados mencionar que foram feitas analyses SEM, EDX, XPS e SIMS.
  • #8: - Fazer um rapido overview da apresentacao. Quando falar de resultados mencionar que foram feitas analyses SEM, EDX, XPS e SIMS.
  • #9: - Fazer um rapido overview da apresentacao. Quando falar de resultados mencionar que foram feitas analyses SEM, EDX, XPS e SIMS.
  • #10: - Fazer um rapido overview da apresentacao. Quando falar de resultados mencionar que foram feitas analyses SEM, EDX, XPS e SIMS.
  • #11: - Fazer um rapido overview da apresentacao. Quando falar de resultados mencionar que foram feitas analyses SEM, EDX, XPS e SIMS.
  • #12: - Fazer um rapido overview da apresentacao. Quando falar de resultados mencionar que foram feitas analyses SEM, EDX, XPS e SIMS.
  • #13: - Fazer um rapido overview da apresentacao. Quando falar de resultados mencionar que foram feitas analyses SEM, EDX, XPS e SIMS.
  • #14: - Fazer um rapido overview da apresentacao. Quando falar de resultados mencionar que foram feitas analyses SEM, EDX, XPS e SIMS.
  • #15: - Fazer um rapido overview da apresentacao. Quando falar de resultados mencionar que foram feitas analyses SEM, EDX, XPS e SIMS.
  • #16: - Fazer um rapido overview da apresentacao. Quando falar de resultados mencionar que foram feitas analyses SEM, EDX, XPS e SIMS.
  • #17: - Fazer um rapido overview da apresentacao. Quando falar de resultados mencionar que foram feitas analyses SEM, EDX, XPS e SIMS.
  • #18: - Fazer um rapido overview da apresentacao. Quando falar de resultados mencionar que foram feitas analyses SEM, EDX, XPS e SIMS.
  • #19: - Fazer um rapido overview da apresentacao. Quando falar de resultados mencionar que foram feitas analyses SEM, EDX, XPS e SIMS.
  • #20: - Fazer um rapido overview da apresentacao. Quando falar de resultados mencionar que foram feitas analyses SEM, EDX, XPS e SIMS.
  • #21: - Fazer um rapido overview da apresentacao. Quando falar de resultados mencionar que foram feitas analyses SEM, EDX, XPS e SIMS.
  • #22: - Fazer um rapido overview da apresentacao. Quando falar de resultados mencionar que foram feitas analyses SEM, EDX, XPS e SIMS.
  • #23: - Fazer um rapido overview da apresentacao. Quando falar de resultados mencionar que foram feitas analyses SEM, EDX, XPS e SIMS.
  • #24: - Fazer um rapido overview da apresentacao. Quando falar de resultados mencionar que foram feitas analyses SEM, EDX, XPS e SIMS.
  • #25: - Fazer um rapido overview da apresentacao. Quando falar de resultados mencionar que foram feitas analyses SEM, EDX, XPS e SIMS.
  • #26: - Fazer um rapido overview da apresentacao. Quando falar de resultados mencionar que foram feitas analyses SEM, EDX, XPS e SIMS.
  • #27: - Fazer um rapido overview da apresentacao. Quando falar de resultados mencionar que foram feitas analyses SEM, EDX, XPS e SIMS.
  • #28: - Fazer um rapido overview da apresentacao. Quando falar de resultados mencionar que foram feitas analyses SEM, EDX, XPS e SIMS.
  • #29: - Fazer um rapido overview da apresentacao. Quando falar de resultados mencionar que foram feitas analyses SEM, EDX, XPS e SIMS.
  • #30: - Fazer um rapido overview da apresentacao. Quando falar de resultados mencionar que foram feitas analyses SEM, EDX, XPS e SIMS.
  • #31: - Fazer um rapido overview da apresentacao. Quando falar de resultados mencionar que foram feitas analyses SEM, EDX, XPS e SIMS.
  • #32: - Fazer um rapido overview da apresentacao. Quando falar de resultados mencionar que foram feitas analyses SEM, EDX, XPS e SIMS.
  • #33: - Fazer um rapido overview da apresentacao. Quando falar de resultados mencionar que foram feitas analyses SEM, EDX, XPS e SIMS.
  • #34: - Fazer um rapido overview da apresentacao. Quando falar de resultados mencionar que foram feitas analyses SEM, EDX, XPS e SIMS.
  • #35: - Fazer um rapido overview da apresentacao. Quando falar de resultados mencionar que foram feitas analyses SEM, EDX, XPS e SIMS.
  • #36: - Fazer um rapido overview da apresentacao. Quando falar de resultados mencionar que foram feitas analyses SEM, EDX, XPS e SIMS.
  • #37: - Fazer um rapido overview da apresentacao. Quando falar de resultados mencionar que foram feitas analyses SEM, EDX, XPS e SIMS.
  • #38: - Fazer um rapido overview da apresentacao. Quando falar de resultados mencionar que foram feitas analyses SEM, EDX, XPS e SIMS.
  • #39: - Fazer um rapido overview da apresentacao. Quando falar de resultados mencionar que foram feitas analyses SEM, EDX, XPS e SIMS.
  • #40: - Fazer um rapido overview da apresentacao. Quando falar de resultados mencionar que foram feitas analyses SEM, EDX, XPS e SIMS.
  • #41: - Fazer um rapido overview da apresentacao. Quando falar de resultados mencionar que foram feitas analyses SEM, EDX, XPS e SIMS.
  • #42: - Fazer um rapido overview da apresentacao. Quando falar de resultados mencionar que foram feitas analyses SEM, EDX, XPS e SIMS.