Talk at the EarthCube End-User Domain Workshop for Rock Deformation and Mineral Physics Research. By Mark Rivers, University of Chicago

1. Future Opportunities and Challenges for
Synchrotron
X-ray Sources in the Earth Sciences
Mark Rivers
University of Chicago

2. GSECARS Overview
• GSECARS is a national user facility for synchrotron
radiation research in geochemistry, mineral physics, and
environmental science.
– Funding: NSF-EAR Instrumentation and Facilities Division
and DOE Geosciences.
– Techniques: high-pressure diffraction (diamond anvil cell
and multi-anvil press), XRF microprobe, microtomography,
surface scattering, x-ray spectroscopy, powder diffraction
– 3 undulator stations, 2 bending magnet stations; 4 stations
operate simultaneously
– Statistics for last 12 months: 466 unique users, 540 user
visits, 283 experiments, 355 beam time requests, 110
publications

3. Personal History of Computed
Microtomography
• 1987 at NSLS
– 1 hour to collect a single slice using first-generation CT
– 1 hour to reconstruct single slice
• 2013 at APS (routine)
– 10 minutes to collect 1040 slices
– 1 minute to reconstruct 1040 slices
• 104-106 improvements
– Brighter x-ray sources
– Detectors: 2-D with high sensitivity and rapid readout
– Computers for fast reconstruction

4. Complete
Reconstruction
Fly through in Z direction
Soil aggregate
6.5 um/pixel
28 keV
Sasha Kravchenko
Michigan State
Studying decay of
organic material
with time

5. Tomography and Computing
Infrastructure
• Where we do NOT need major CI improvements
–
–
–
–
Collection
Reconstruction
Local storage
All of these are “domain independent”
• Extracting scientifically useful information from images
has NOT kept up, we do need help
• Why?
– The information to be extracted is highly specific to the
problem (domain).
– Cottage industry that does not scale to today’s problems

6. Brighter X-ray Beams
• Figure of merit of a
storage ring like the APS,
NSLS-II, or ALS is the
emittance:
• σxσ’x (horizontal)
• σyσ’y (vertical)
• APS emittance today has 3
nm-rad emittance
• Brightness of the x-ray
beams is directly
proportional to the
emittance
APS electron beam profile
today

8. The Future
The “New” APS Upgrade
• BESAC recommended in July 2013
that the US agressively pursue
MBA lattice
• APS has made this new plan for
the already approved upgrade
• Reduce emittance to 60 pm-rad,
50X reduction
• Reduce energy from 7 to 6 GeV
• All new undulators
• 2019 time frame
• ~1 year shutdown
APS electron beam profile
after MBA upgrade

10. 100X Higher Brightness
• 100X more photons in the same
focal spot size we use today
– 100X higher time resolution for
kinetic studies, deformation, etc.
• 100X smaller spot with the same
flux we have today
– Reduce focal spot size in diamond
anvil cell from 3 µm to 30 nm with
no loss in intensity
• Source is nearly diffraction limited
• 100X higher coherent flux
– Coherent x-ray diffraction, photon
correlation spectroscopy (speckle)
experiments have huge gains
XPCS study of deformation

11. Faster Detectors
Microtomography
•
•
•
•
PCO DIMAX HS
2277 frames/sec @ 2000x2000 pixels
5469 frames/sec @1440x1050 pixels
Complete microtomography dataset in 0.1
second
• 18GB/s peak, 600 MB/sec sustained
• Time-resolved tomography: melting,
deformation

12. Faster Detectors
Diffraction
•
•
•
•
•
•
•
•
Dectris Eiger
75 x 75 micron pixels
Single photon counting
Up to 1 MHz/pixel
1030 x 1065 pixels
3,000 frames/sec
3.3 GB/sec
Will stream lossless compressed images at full frame rate,
~600 MB/sec.
• Time-resolved diffraction for fast reactions
• Collect a full crystallography data set in 1 second

13. X-ray diffraction in the diamond anvil cell
Smaller beams, faster detectors
5 micron beam today
50 nm beam (ultra-high pressure,
much more complex)

14. Faster Detectors
X-ray fluorescence mapping
• Hitachi Quad Vortex silicon drift diode
detector
• XIA xMAP Digital Signal Processing
electronics
• 4 elements * 1000 pixels/sec = 4000
spectra/sec
• 16 MB/sec sustained
• 1 Mega pixel map in 20 minutes
• Next generation detector and electronics
reduces this to 5 minutes

15. GSECARS 13-ID-E X-ray Microprobe
XRF Imaging: high spatial resolution (500 nm) with high flux (>10 11 ph/s)
Fe (~70 ppm)
Mn (~70 ppm)
Zn (~100 ppm)
Arabidopsis seed Columbia-0
see Kim et
al., Science,
2009 for
background
7 µm
200 msec
X26A
NSLS
0.7 µm
13 msec
13-ID-E
APS
T. Punshon
and A.
Sivitz,
Dartmouth

16. XRF mapping challenges
•
•
•
•
Map on previous slide has 1 million spectra like this
Currently just map the total counts in each peak (region of interest)
Really need to fit background, deconvolve overlapping peaks
Not practical today

17. Conclusions
• Improvements in x-ray sources and detectors have the potential for
transformative improvements in our science in 5-10 years
– Improved spatial and temporal resolution
– New science by exploiting the coherence of the x-rays
• The problems that Martin Kunz presented yesterday will grow by
several orders of magnitude
• Clear need for major computing infrastructure improvements
– If not there will be mountains of unprocessed data
• These problems go beyond NSF-EAR.
– DOE has major responsibility as operator of the facilities
• However, in many cases we need “domain-specific” solutions because
our problems and our data are often unique