Photon counting using amorphous selenium: Achieving hole dispersion limited count-rate using the Frisch grid detector design

1
Photon counting using amorphous selenium
Achieving hole dispersion limited count-rate
using the Frisch grid detector design

• Karim S. Karim, A. Goldan
• Associate Professor,
• Silicon Thin film Applied Research (STAR) group,
• Dept. of Electrical and Computer Engineering
• University of Waterloo, CANADA
• CMOSET 2009 Workshop (Vancouver)

2
STAR Group
Amir Goldan
3
Digital Medical Imaging
ANALOG
DIGITAL
4
Energy Integrating vs. Photon Counting
5
Making a case for photon counting Dose Efficiency

• Imagers Dose Efficiency considers the
  following factors
• Quality of the x-ray spectrum
• Quantum efficiency
• Energy absorption efficiency
• Scatter rejection efficiency
• Conversion efficiency (of absorbed photons to
  image signal)
• Detector noise(i.e., swank noise, film
  granularity noise, leakage shot noise, variations
  in the conversion gain)
• Readout noise

• Mammography Imager
• Energy-integrating
• Area-imaging
• Smit-Röntgen antiscatter grid
• 35 DQE
• 30kVp Mo spectrum

Q1 Why so low?
Q2 How can it be improved?
M. Lundqvist et al, IEEE Trans. Nucl. Sci. 48(4),
1530-1536 (2001).
6
Scatter, Noise, Conversion Efficiency
7
Memory Artifacts Anatomical Noise

• Memory artifacts in an energy-integrating
  selenium-based imager.
• Photon counting systems are not susceptible to
  low-frequency memory artifacts.

H.G. Chotas et al, Radiology 203, 881- 883(1997).

• Anatomical Noise (or tissue overlap effect)
• - Image2 Overlying tissue (i.e., circle) can
  mask the cancerous lesion (i.e., triangle) from
  being detected.
• Digital Tomosynthesis
• Multiple images are acquired at different
  angles.
• Structures in different planes can be brought
  into focus using shift-and-add of different
  projection images.

J. T. Dobbins III et al, Phys. Med. Biol. 48,
R65R106 (2003).
8
Photon Counting w/ a-Se
Energy 70 keV / Layer Thickness 150 µm
FWHM50

• To guarantee complete charge collection and
  eliminate ballistic deficit
• Shaping time Electron transit time 60µs

• With 10 probability of pileup, selenium
  count-rate (rc) is 500 quanta/s
• Photon count-rate needed for imaging is 50,000
  quanta/s

9
Inequality in Charge Transport

• The material of large-area photoconductors is
  either amorphous or polycrystalline
• Suffer from poor charge transport due to traps in
  the bulk
• For example, the effective carrier mobilities in
  a-Se is
• 0.003 cm2/V.s for e-
• 0.14 cm2/V.s for h

10
Improvements

• Addressing the problem of carrier transport
• Utilizing readout techniques
• Pulse-shape discrimination
• Pulse-risetime compensation/correction
• Modifying the detector structure
• P-type-Intrinsic-N-type (P-I-N) contact structure
• Unipolar Charge-Sensing

11
Shockley-Ramo Theorem

• Shockley-Ramo Theory (1938-1939) Charge
  induction on any electrode by a single electron
  in a vacuum tube is due to electrons motion!
• Qi qVW
• VW Weighting potential
• Theory is valid for carrier motion inside
  semiconductor detectors in the presence of space
  charge.
• G. Cavalleri et al, Nucl. Instr. and Meth. 92,
  137-140 (1971).

12
Unipolar Charge-Sensing

• A method first proposed by O. Frisch in 1944 to
  solve the problem of slow drift and trapping
  effect of positive ions in conventional gas
  detectors
• How? By providing an electrostatic shield near
  the collecting electrode.

13
Kinestatic Imagers

• Kinestatic charge detection (KCD) was proposed
  in 1985 as a compromise solution to
• complexity of nn detectors (i.e., area
  detectors)
• inefficiency of n1 detectors (i.e., scanned-slit
  detectors)
• KCD imager has n1 elements but operates as an
  nm detector (m lt n)
• Nowadays, nn area-detectors are cost-effective
  and available commercially, thanks to two
  technological advances
• availability of cost-effective, large-area a-Si
  TFT readout panels
• reliable coupling of evaporated a-Se
  photoconductors to these TFT panels

F.A. DiBianca and M.D. Barker, Med. Phys. 12,
339-343 (1985).
14
Unipolar Solid-State Area Imagers

• Considering large-area evaporated photoconductors
  (such as amorphous selenium), this research
  proposes unipolar charge-sensing using the Frisch
  grid detector design to
• Enable photon counting operation by improving the
  photon count-rate
• Results can be applied to also improve charge
  collection efficiency, implications of which in
  energy-integrating mode are
• Improved x-ray sensitivity and detective quantum
  efficiency (DQE)
• Improved spatial resolution
• Reduced memory effects

A.H. Goldan et al, Proc. of SPIE 7258, 725816
(2009).
A.H. Goldan and K.S. Karim, Med. Phys. (under
revision).
15
Photon Counting ModeImproved count-rate
1
2
3
1
2
3
1
2
3
rc 500 quanta/s
rc 35,000 quanta/s
rc 10,000,000 quanta/s
16
Food (for thought)

• Realistically speaking, can the unipolar Frisch
  grid design improve the count-rate by 4 orders
  of magnitude?
• Can we implement the Frisch grids very close to
  the collecting pixel electrodes to yield a nearly
  ideal unipolar charge-sensing operation?
• Is there a physical phenomenon that limits the
  count-rate given a nearly-ideal unipolar device?

17
Frisch Grid Fabrication
K.S. Karim, A. Goldan, PCT and US Patent Appl.
No. 12/357,577 (filed on Jan. 2009).
18
Weighting Potential Distribution

• 5 µm wide grid
• 5 µm grid spacing
• 500 nm insulator thickness

19
Food for thought (again)

• Realistically speaking, can the unipolar Frisch
  grid design improve the count-rate by 4 orders
  of magnitude?
• Can we implement the Frisch grids very close to
  the collecting pixel electrodes to yield a nearly
  ideal unipolar charge-sensing operation?
• YES! Thanks to evaporated photoconductors which
  enable grid construction using photolithography.
• Is there a physical phenomenon that limits the
  count-rate given a nearly-ideal unipolar device?

20
Time-of-Flight (TOF)

• TOF transient photoconductivity technique
• Directly measure carrier drift mobility and deep
  trapping time

21
TOF Measurements
3ns Laser Pulse

• The extended decay in the tail is due to the
  dispersion of holes in the drifting packet
• Dispersion is due to
• diffusion
• multiple trapping and release
• Mutual Coulombic repulsion

22
Food for thought (conclusion)

• Realistically speaking, can the unipolar Frisch
  grid design improve the count-rate by 4 orders
  of magnitude?
• Can we implement the Frisch grids very close to
  the collecting pixel electrodes to yield a nearly
  ideal unipolar charge-sensing operation?
• YES! Thanks to evaporated photoconductors which
  enable grid construction using photolithography.
• Is there a physical phenomenon that limits the
  count-rate given a nearly-ideal unipolar device?
• YES! Carrier dispersion limits the count-rate.
  However, a-Se is till capable of achieving a
  count-rate of 1M quanta/s at E10V/µm.

23
Summary

• Many of the limitations associated with todays
  energy-integrating imagers can be alleviated by
  switching to photon-counting.
• However, poor carrier transport in large-area
  evaporated photoconductors has greatly limited
  their photon count rate.
• To circumvent the problem of poor carrier
  transport, this research proposes unipolar
  charge-sensing using the Frisch grid design.
• Results show substantial increase in photon count
  rate using the unipolar Frisch detectors

24
Acknowledgements