Overview of Scientific Imaging using CCD Arrays

1
Overview of Scientific Imagingusing CCD Arrays

• Jaal Ghandhi
• Mechanical Engineering
• Univ. of Wisconsin-Madison

2
Detector Architecture

• Charge-Coupled Device (CCD)
• High quantum efficiency
• Low noise
• High dynamic range
• High uniformity
• Photodiode Array
• CMOS

3
CCD Overview

• Photons incident on silicon form electron hole
  pairs
• Polysilicon mask is used to create a potential
  barrier to isolate the charge in a region of
  space (pixel)
• By modulating the potential the charge can be
  moved with very high efficiency (CTE gt 99.9998)
• Charge is transferred to the output amplifier
  where it is digitized

4
CCD Architecture
Full Frame
Frame Transfer
Interline Transfer
Serial Register
Serial Register
Serial Register
Pixel Array
Masked Storage Array
Pixel Array
Storage Pixels
Active Pixels
Scientific Imaging
PIV Cameras Video-rate Imaging
Video-rate Imaging
5
Microchannel Plate Intensifier

• Gain is controlled by VMCP
• Gating achieved by pulsing VPC
• Intensifier Advantages
• Very short gate times possible (1ns)
• High rejection ratio
• Gain aids in raising signal out of the read-noise
  limited regime
• Intensifier Disadvantages
• Decreased spatial resolution
• Limited dynamic range
• Amplification of noise
• Moderate quantum efficiencies

V
V
V
pc
MCP
ph
n
n
h
h
e
e
-
-
e
-
e
-
n
h
Phosphor
MCP
Photocathode
6
Coupling Intensifier to Camera - ICCD

• Lens coupling not recommended
• Limited f-number
• Alignment
• Fiber coupling

7
Electron Multiplying CCD - EMCCD

• By increasing the clocking voltage in a CCD you
  can create a controlled ionization that generates
  electrons
• The gain factor is small, 1.015?, so it must be
  performed serially
• Low noise amplification

Serial Register
Gain Register
Amplifier
Pixel Array
8
Analysis of SNROptically generated signal
100

• Photons incident on the detector produce
  electrons in a probabilistic manner given by the
  quantum efficiency, ? ?(?)

80
60
40
QE ()
20
300
500
900
700
1100
e2V 47-10 Front-illuminated
?
9
Analysis of SNROptically generated signal
100
Midband coated
UV coated
80
60
Uncoated
40
QE ()
20
FI
300
500
900
700
1100
?
e2V 47-10 Back-illuminated
10
Analysis of SNRThermally generated signal

• Thermal oscillations of the silicon lattice can
  generate electron hole pairs, which is called
  dark charge
• In principle, this can be subtracted from the
  signal
• Cooling is critical!

105
103
Dark Current (e-/pixel/s)
101
10-1
T (?C)
-20
40
0
e2V 47-10 Back-illuminated
11
Analysis of SNRTotal signal

• CA/D counts/e- amplifier gain
• ? - quantum efficiency
• Npp number of photons per pixel
• D dark charge determined by the dark current
  and readout exposure time
• D mean dark charge obtained with no
  illumination
• Since the dark noise is (ideally) repeatable

_
12
Analysis of SNRPhotonic shot noise

• Photon detection in a given area for a given time
  is probabilistic because the photon flux is not
  constant, i.e. the arrival time separation is not
  constant
• Therefore, collecting photons in a given area for
  a fixed time results in an inherent noise called
  shot noise.
• Shot noise is described by Poisson statistics
• Mean ?
• Variance ?
• Result The maximum possible signal-to-noise
  ratio is

Avg SD 2 0 2 0.8
13
Analysis of SNRRead noise

• There is noise introduced to the signal when the
  charge is converted to digital counts in the
  amplifier, termed read noise
• The read noise depends on the frequency (clock
  speed)
• Result slow scan cameras

e2V 47-10 Back-illuminated
14
Analysis of SNRDark noise

• The generation of dark charge is probabilistic in
  nature, and can be described by a Poisson
  distribution
• Subtracting the mean dark charge, D, from a pixel
  results in a residual quantity, D(x,y)-D(x,y),
  which is called dark noise.

_
_
15
Analysis of SNRGain noise

• The signal amplification in ICCDs and EMCCDs
  involves some noise generation.
• ICCD contributes to the shot noise contribution
• EMCCD contributes to shot noise and dark noise
  contributions

16
Analysis of SNR
CCD ICCD EMCCD
Signal
Shot Noise
Dark Noise
Read Noise
Total Noise

• Npp number of signal photons ? - quantum
  efficiency
• G gain factor (e-/e-) F noise factor
• F2 noise factor pc photocathode
• FEMCCD ? 1.3 FICCD ? 1.6 (? ? 2.6)

17
Slow-scan PerformanceTheoretical
18
Intensified vs Slow-scan
19
Slow-Scan PerformanceMeasured

• Apogee AP7 MicroMax

20
Intensified Camera PerformanceMeasured

• PI Max IVRC

21
Camera Selection

• For all applications a slow-scan, deeply cooled,
  back-illuminated CCD is the best choice in terms
  of SNR and image quality, except when
• The signal level is very low, then gain amplifies
  the signal above the read noise EMCCD is best
  option because of superior image quality
• There is strong luminosity and gating is required
  ICCD is required

Scotts note all else being equal, cameras with
big pixels have an advantage
22
Case studyResidual gas measurements in an IC
engine

• MicroMax PI Max