Henning Friis Poulsen
NEXMAP
Physics Dept, DTU
Imaging: tomography, 3DXRD, x-ray microscopy
Lecture at KTH, Stockholm Dec 2013
• Why imaging, why 4D
• T omography: the math, recent trends and examples
Short break
• 3DXRD microscopy (grain mapping)
• X-ray microscopy: multiscale physics
T he menu
Diffraction Imaging
Atomic structure within crystals Structure on nm – cm scale
Oil/CO2 storage
Fuel cells
Wind energy
Catalysis
3D imaging of energy materials
5
200 µm
Metal structures
5 µm
Example: Al
Model structure evolution in 4D
200 µm
Metal structures
5 µm
Example: Al
Model structure evolution in 4D
But: Bottum-up not yet feasible
Average properties not enough
Multiscale approach required
Mapping:
Phase
position
morphology
orientation of lattice
stress
plastic strain
dislocation densities
…
200 µm
Materials in
2D 4D
5 µm
Tomography:
the principle
Wilhelm Röntgen The first X-ray, Frau
Röntgen´s left hand
Wilhelm Conrad Röntgen discovered X-ray November 8 in 1895
X-ray radiography
Source
Detector Object
r(x,y,z)?
dsµI
I)ln(
0
Isds
dI)(
I0 I
Absorption contrast tomography
q
Inverse problem.
Solution: reconstruction
Radon Transform
Rf = L f(x,y) ds
Rf (w, t) =
f(x,y) d(t = xcos(w)-ysin(w) ) dxdy
Radon Transform: (x,y) -> (t, w)
Inversion:
1917: Johann Radon: On the determination of functions
from their integrals along certain manifolds.
t w x
y
L
Detector
Sample system:
Insert
Filtered Backprojection
Direct space Fourier Space
|k|
Algebraic Reconstruction Technique
Solve: Ax = b
x: density of each voxel
b: detector pixel intensitites
A: geometry of set-up
Iterative solution:
M
j
kj
M
j
k
jkjk
kk
A
xAb
xx
1
2
11
Slower, but freedom to add constraints and choose projections
xi
bj
Discrete tomography
Binary system
known densities, r1 and
r2
Few projections
Materials science: faster
Medicine: reduced dose
Mathematical tomography, other approaches
• Series expansion
• Expectation maximization
• Linear programming, graph theory
• Difference map algorithms
• Total Variation
• Projection onto convex sets (phase retrieval)
Use of Soduku alorithms…
Discrete values
Combinatorial constraints
Finite nr. of projections
Generalised projections
Solution: Difference map algorithm
by Veit Elser, Cornell (2005)
Use for biological imaing using coherent x-rays
D. Shapiro, P. Thibault, T. Beetz, V. Elser et al.
PNAS (2005), 102, 15343–15346
Reconstruction Segmentation Quantification
Fuzzy connectivity Segmentation
Tomography, the math
Basics : • A.C. Kak, and M. Slaney. Principles of Computerized Tomographic Imaging.
(IEEE Press, New York, 1988).
• G.T. Herman. Image Reconstructions from Projections. The Fundamentals of
Compterized Tomography. (Academic Press, New York).
Mathematical:
• S. Helgason (1980). The Radon Transform. Progress in Math., Vol. 5
(Birkhäuser, Boston, 1980).
Discrete Tomography:
• G.T. Herman and A. Kuba, Eds. Discrete Tomography. (Boston, Birkhäuser,
1988).
Segmentation:
• T.S. Yoo. Insights into Images. (Wellesey, Canada, A.K. Peters, 2004).
4D imaging for materials science
at synchrotrons
Resolution: 0.05 – 2.5 µm
Nr of projections: 1000-2000
Reconstructred volume:
~500x500x500 voxels
Tomography for materials science
at synchrotons in Europe*
Some dedicated sources for parallel beam tomography
60-200 keV: ESRF: ID15
20-60 keV ESRF ID19 + BM5
ESRF NINA
PETRA-III
10-30 keV PSI: TOMCAT
Diamond
Soleil
ALBA
Max-IV ?
……
*The list is not complete and numbers are only approximate
Phase Contrast Tomography
Absorption Phase
0.1
1
10
100
1000
0.1 1 10 100
110100
Energy (keV)
Hard X-raysSoft X-rays
(water window)
Wavelength (Å)
d
Aluminium Smallest detectable hole at 25 keV
in a 4 mm thick sample:
Absorption: 20 µm
Phase: 0.05 µm
Phase/
absorption
Ex.: Holotomography of semisolid Al/Si alloy
800 angular positions multilayer monochromator: total time 40 minutes E = 18 keV
Absorption
100 µm
Holotomography
Al/Si Al
Work by P. Cloetens, ID19, ESRF
3DXRD (grains, Flourescence EXAFS and XANES
orientations)
Powderdiffraction SAXS Coherent diffraction
(phases) (nano-particle)
Tomography using other contrast mechanisms
5 nm resolution, Au nanoparticle:
Schroer et al. PRL (2008) 101, 090801
Bleuet et al. Nature Mat. 7, 468 (2008)
Second
half of
talk J.-D.Grunwaldt et al.
J. of Phys. Chem. B 110, 8674 (2006)
white x-ray
beam
sample
in furnace
tomography detector
monochromatic
x-ray beam
2-D diffraction detector
Laue
monochromotar
Laue
monochromator
Ultrafast tomography and diffraction @ID15, ESRF
EVOLUTION OF MICRO-TOMOGRAPHY AT
ID15A
10ms
50ms
1s
10s
10min
4h
0.01
0.1
1
10
100
1000
10000
100000
1999 2001 2003 2005 2007 2009
Year
t (s
)
Time for 500 frames (s)
1999 2005 2009
100000
100
0.01
4 h
10 min
10 s
1 s
50 ms
10 ms
Courtesy of Honkimäki
Applications
PROPAGATION OF LIQUID FRONT
IN GRANULAR MATERIAL
Resolution:
50 ms
2 µm
Cai, Powell, Yue et al. Appl. Phys. Lett. 83, 1671 (2003).
Ultra-fast x-ray tomography
Fuel-spray studies: 5 s time resolution
Observation of shock-waves
People involved:
Risø: Stefan Poulsen, Erik Lauridsen
NU: Peter Voorhees, Julie Fife,
Anthony Johnson, Larry Aagenson
and Michael Miksis
PSI: Marco Stampanoni
Coarsening in solid/liquid AlCu alloys
Al-26wt%Cu (42% solid)
Synchrotron data
Annealing for 6 hours at 850 C
• Time resolution: 2 min
• Resolution: 1 µm
Al-26wt%Cu (Velocity plot)
Interfacial velocities
Common phenomena involving the break up
of a liquid cylinder into droplets
Al-15wt%Cu (74% solid)
Plateau-Rayleigh instability
Rockwool
M.W. Westneat et al. Science, 299, 558-560 (2003)
Beetle Breath
Tomography in materials science
General (in order of increasing complexity)
• J. Baruchel, J.-Y. Buffiere, E. Maire, P. Merle, G. Peix. X-ray
tomography in materials science. Hermes Science Publications,
Paris, 2000
• W. Reimers, A.R. Pyzalla, A.K. Schreyer, H. Clemens, Eds. Neutrons and
Synchrotron Radiation in Engineering Materials Science Wiley-VCH, 2008
• J. Banhart, Ed.: Advanced tomographic methods in materials research and
engineering, Oxford Uni. Press, 2008.
Conferences
SPIE conference series
X-TOP conference series
The future
X-rays going nano
An x-ray TEM?
Coherence:
Single particles:
Quantum dots:
Fe
Cu
Tomography going Nano
An Xray TEM
Nano-tomography Coherent scattering
Aim:1 nm
Current resolution: 50 nm
Now: parallel beam geometry:
LIMITATION
Resolution: 0.5 m
Sample size: 1 mm
1000-4000
Current resolution: 5 nm
Farfield
Scintillator based detector
• Resolution: ~ 3 µm
• Efficiency: ~ 1%
CCD
Sample
Flourescence
screen
Conventional
Resolution -> 1 µm
Efficiency improvement by factor 5
Waveguiding
-----
Technique originally developed at KTH
Bridge the length scale
Local tomography:
Problem: no unique, exact solution
But: map the singularities correctly
Open field
Local events at DTU
Extra
3D Imaging center
Start: 4 instruments at DTU
Start-up: Xnovo Technologies
Luggage scanning and transport in airports