ccd lecture1
TRANSCRIPT
-
8/14/2019 ccd lecture1
1/43
CCDs : Current Developments
Part 1 : Deep Depletion CCDsImproving the red response of CCDs.
Part 2 : Low Light Level CCDs (LLLCCD)A new idea from Marconi (EEV) to reduce or eliminate CCD read-out noise.
-
8/14/2019 ccd lecture1
2/43
Part 1 : Deep Depletion CCDs
Improving the red response of CCDs.
-
8/14/2019 ccd lecture1
3/43
pixel
boun
dary
Charge packetp-type silicon
n-type silicon
SiO2 Insulating layer
Electrode Structure
pixe
l
bou
ndary
incoming
photons
Charge Collection in a CCD.
Photons entering the CCD create electron-hole pairs. The electrons are then attracted towards
the most positive potential in the device where they create charge packets. Each packet
corresponds to one pixel
-
8/14/2019 ccd lecture1
4/43
Electricpotential
Potential along this line
shown in graph above.
Electricpotential
Cross section through a thick frontside illuminated CCD
Deep Depletion CCDs 1.
The electric field structure in a CCD defines to a large degree its Quantum Efficiency (QE). Consider
first a thick frontside illuminated CCD, which has a poor QE.
In this region the electric potential gradient
is fairly low i.e. the electric field is low.
Any photo-electrons created in the region of low electric field stand a much higher chance of
recombination and loss. There is only a weak external field to sweep apart the photo-electron
and the hole it leaves behind.
-
8/14/2019 ccd lecture1
5/43
Electricpotential
Electricpotential
Cross section through a thinned CCD
Deep Depletion CCDs 2.
In a thinned CCD , the field free region is simply etched away.
There is now a high electric field throughout the
full depth of the CCD.
Photo-electrons created anywhere throughout the depth of the device will now be detected.
Photons no longer have to pass through the electrode structure to reach active silicon.
This volume is
etched away
during manufacture
Problem : Thinned CCDs may have good blue
response but they become transparent
at longer wavelengths; the red response
suffers.
Red photons can now pass
right through the CCD.
-
8/14/2019 ccd lecture1
6/43
Electricpotential
Electricpotential
Cross section through a Deep Depletion CCD
Deep Depletion CCDs 3.
Ideally we require all the benefits of a thinned CCD plus an improved red response. The solution is to use a
CCD with an intermediate thickness of about 40m constructed from Hi-Resistivity silicon. The increased
thickness makes the device opaque to red photons. The use of Hi-Resistivity silicon means that there are no field
free regions despite the greater thickness.
There is now a high electric field throughout the full depth of the CCD. CCDs manufactured in this way
are known as Deep depletion CCDs. The name implies that the region of high electric field, also known as
the depletion zone extends deeply into the device.
Red photons are now absorbed inthe thicker bulk of the device.
Problem :Hi resistivity silicon contains much lower
impurity levels than normal. Very few wafer
fabrication factories commonly use this
material and deep depletion CCDs have to
be designed and made to order.
-
8/14/2019 ccd lecture1
7/43
QE Improvements with Deep Depletion CCDs
0
10
20
30
40
50
60
70
80
90
100
300 400 500 600 700 800 900 1000
nm
QE
%
CC1D20 MBE singleAR @320nm
CC1D20 BIV BroadBand AR
EEV12 (StandardThinned)
Marconi DeepDepletion (broad
Band AR)
-
8/14/2019 ccd lecture1
8/43
Deep Depletion CCDs 4.
Thinned Marconi CCD (Current ISIS Blue)
Fringing will also be reduced
CCID20 Deep Depletion CCD
Images illuminated by 900nm filter with 2nm bandpass
Test data courtesy of ESO
-
8/14/2019 ccd lecture1
9/43
-
8/14/2019 ccd lecture1
10/43
Part 2 : Low Light Level CCDs (LLLCCDs)
A new idea from Marconi that creates internal electron gain
in a CCD and reduces read-noise to sub-electron levels.
-
8/14/2019 ccd lecture1
11/43
RAIN (PHOTONS)
BUCKETS (PIXELS)
VERTICAL
CONVEYORBELTS
(CCD COLUMNS)
HORIZONTAL
CONVEYOR BELT
(SERIAL REGISTER)
MEASURING
CYLINDER
(OUTPUT
AMPLIFIER)
CCD Analogy
-
8/14/2019 ccd lecture1
12/43
-
8/14/2019 ccd lecture1
13/43
Edgeof
Silicon
Image Area
Serial Register
Read Out Amplifier
Buswires
Photomicrograph of a corner of an EEV CCD.
-
8/14/2019 ccd lecture1
14/43
pixel
boundary
Charge packetp-type silicon
n-type silicon
SiO2 Insulating layer
Electrode Structure
pixel
bou
ndary
incoming
phot
ons
Charge Collection in a CCD.
Photons entering the CCD create electron-hole pairs. The electrons are then attracted towards
the most positive potential in the device where they create charge packets. Each packet
corresponds to one pixel.
-
8/14/2019 ccd lecture1
15/43
PotentialEnerg
y
Conventional Clocking 1
Surface electrodesCharge packet (photo-electrons)
P-type siliconN-type silicon
Insulating layer
Charge packets occupy potential minimums
-
8/14/2019 ccd lecture1
16/43
PotentialEnerg
y
Conventional Clocking 2
-
8/14/2019 ccd lecture1
17/43
PotentialEnerg
y
Conventional Clocking 3
-
8/14/2019 ccd lecture1
18/43
PotentialEnerg
y
Conventional Clocking 4
-
8/14/2019 ccd lecture1
19/43
PotentialEnerg
y
Conventional Clocking 5
-
8/14/2019 ccd lecture1
20/43
PotentialEnerg
y
Conventional Clocking 6
-
8/14/2019 ccd lecture1
21/43
-
8/14/2019 ccd lecture1
22/43
PotentialEnerg
y
Conventional Clocking 8
-
8/14/2019 ccd lecture1
23/43
PotentialEnerg
y
Conventional Clocking 9
-
8/14/2019 ccd lecture1
24/43
PotentialEnerg
y
Conventional Clocking 10
Charge packets have moved one pixel to the right
-
8/14/2019 ccd lecture1
25/43
Image Area Image Area(Architecture unchanged)
Serial register Serial register{Gain register
On-Chip
Amplifier
On-Chip
Amplifier
The Gain Register can be added to any existing design
LLLCCD Gain Register Architecture
Conventional CCD LLLCCD
-
8/14/2019 ccd lecture1
26/43
PotentialEnerg
y
Multiplication Clocking 1
Gain electrode
In this diagram we see a small section of the gain register
-
8/14/2019 ccd lecture1
27/43
PotentialEnerg
y
Multiplication Clocking 2
PotentialEnerg
y
Gain electrode energised. Charge packets accelerated strongly into deep potential well.
Energetic electrons loose energy through creation of more charge carriers (analogous tomultiplication effects in the dynodes of a photo-multiplier) .
Gain electrode
-
8/14/2019 ccd lecture1
28/43
PotentialEnerg
y
Multiplication Clocking 3
PotentialEnerg
y
Clocking continues but each time the charge packets pass through the gain electrode, further
amplification is produced. Gain per stage is low,
-
8/14/2019 ccd lecture1
29/43
Gain Sensitivity of CCD65
1
10
100
1000
10000
20 25 30 35 40
Clock High Voltage
Gain
Readout Noise of CCD65
0.01
0.1
1
10
100
20 25 30 35 40
Clock High Voltage
Equivalentnoise
electr
onsRMS
The Multiplication Register has a gain strongly dependant on the clock voltage
Multiplication Clocking 4
-
8/14/2019 ccd lecture1
30/43
SNR = Q.I.t.[Q.t.( I +BSKY) +Nr2 ]-0.5
Q = Quantum Efficiency
I = Photons per pixel per second
t = Integration time in secondsBSKY = Sky background in photons per pixel per secondNr = Amplifier (read-out) noise in electrons RMS
Conventional CCD SNR Equation
Noise Equations 1.
Very hard to get Nr < 3e, and then only by slowing down the readout
significantly. At TV frame rates, noise > 50e
Trade-off between readout speed and readout noise
-
8/14/2019 ccd lecture1
31/43
Noise Equations 2.
SNR = Q.I.t.Fn.[Q.t.Fn.( I +BSKY) +(Nr/G)2 ]-0.5
G = Gain of the Gain Register
Fn = Multiplication Noise factor = 0.5
LLLCCD SNR Equation
Readout speed and readout noise are decoupled
With G set sufficiently high,
this term goes to zero, even at
TV frame rates.
Unfortunately, the problem of multiplication noise is introduced
-
8/14/2019 ccd lecture1
32/43
Ideal Histogram, StdDev=Gain x (Mean Illumination in electrons )0.5
Actual Histogram, StdDev=Gain x (Mean Illumination in electrons )0.5 x M
Multiplication Noise 1.
In this example, A flat field image is read out through the multiplication register.
Mean illumination is 16e/pixel. Multiplication register gain =100
Electrons per pixel at output of multiplication register
Probability
Histogram broadened
by multiplication noise
M=1.4
-
8/14/2019 ccd lecture1
33/43
Multiplication Noise 2.
Multiplication noise has the same effect as a reduction of QE by a factor of two.
In high signal environments , LLLCCDs will generally perform worse than
conventional CCDs. They come into their own, however, in low signal, high-speedregimes.
Signal Level
SNR
Conventional CCD
LLLCCD
-
8/14/2019 ccd lecture1
34/43
Offers a way of removing multiplication noise.
Photo-electron
detection threshold
Fast comparator
Photo-electron detection pulses
One
photo-electron
One
photo-electron
Two
photo-electrons
CCD
No
photo-electron
No
photo-electronNo
photo-electron
Co-incidence loss
here
CCD Video waveform
Approx 100ns
Photon Counting 1.
SNR = Q.I.t.[Q.t.( I +BSKY)]-0.5
Noiseless Detector
-
8/14/2019 ccd lecture1
35/43
Photon Counting 2.
If exposure levels are too high, multi-electron events will be counted as single-electron
events, leading to co-incidence losses . This limits the linearity and reduces the effective
QE of the system.
Non-Linearity from Photon-Counting Coincidence Loss
Photo-electron
generation rate Non-Linearity(electrons per pixel per frame) %
0.02 1
0.033 1.6
0.1 5
In the case of a hypothetical 1K x 1K photon counting CCD, the maximum frame rate
would be approximately 10Hz. If we can only accept 5% non-linearity then the maximum
illumination would be approximately 1 photo-electron per pixel per second.
-
8/14/2019 ccd lecture1
36/43
The three operational regimes of LLLCCDs
1) Unity Gain Mode.The CCD operates normally with the SNR dictated by the photon shot noise added in
quadrature with the amplifier read noise. In general a slow readout is required (300KPix/second)
to obtain low read noise (4 electrons would be typical). Higher readout speeds possible but there
will be a trade-off with the read-noise.
2) High Gain Mode.Gain set sufficiently high to make noise in the readout amplifier of the CCD negligible.The drawback is the introduction of Multiplication Noise that reduces the SNR
by a factor of 1.4. Read noise is de-coupled from read-out speed. Very high speed readout
possible, up to 11MPixels per second, although in practice the frame rate will probably be
limited by factors external to the CCD.
3) Photon Counting Mode.Gain is again set high but the video waveform is passed through a comparator. Each trigger
of the comparator is then treated as a single photo-electron of equal weight. Multiplication
noise is thus eliminated. Risk of coincidence losses at higher illumination levels.
Summary.
-
8/14/2019 ccd lecture1
37/43
Possible Application 1.Acquisition Cameras
Performance at CASS of WHT analysed below. The calculated SNR is for a single TV frame (40ms).
It is assumed that the seeing disc of the target star evenly illuminates 28 pixels
(0.6 seeing, 0.1/pixel plate scale). SNR calculated for each pixel of the image.
Assumptions: CCD QE=85%, LLLCCD QE=30%, Image Tube QE =11%
dark of moon, seeing 0.6, 24um pixels (0.1per pixel), 25Hz frame rate
0
0.5
1
1.5
2
2.5
3
3.5
17 18 19 20 21 22
Mv
SNR
Normal CCD
L3CS (LLLCCD)
theoretical limit
Zero-noise image tube
-
8/14/2019 ccd lecture1
38/43
Possible Application 2.Acquisition Cameras
As for the previous slide but instead the exposure time is increased to 10s
0
1
2
3
4
5
6
7
8
9
10
17 18 19 20 21 22
Mv
SNR
Cryocam (standard CCD)
L3CS (LLLCCD)
theoretical limitZero-noise image tube
-
8/14/2019 ccd lecture1
39/43
QE=70%
Amplifier Noise =5e
Background =0.001 photons per pixel per second
Possible Application 3.Photon Counting Faint Object Spectroscopy
LLLCCDs operating in photon counting mode would seem to offer some promise.
The graph below shows the time taken to reach a SNR=3 for various source intensities
0.01
0.1
1
10
0 200 400 600 800 1000
Exposure Time Seconds
Sourceintensityatthedetector
(photonsperpixelp
ersecond)
Thinned LLLCCD with Gain=1000
Thinned LLLCCD +Photon Counting
Conventional CCD
-
8/14/2019 ccd lecture1
40/43
Possible Application 4.Wave Front Sensors
Amplifier Noise=5e
QE= 70%
Algorithm used on the current NAOMI WFS produces reliable centroid
data when totalsignal per sub-aperture exceeds about 60 photons.
0
1
2
3
4
5
6
7
8
9
10
0 10 20 30 40 50 60 70 80 90 100
Photons per pixe l per WFS frame
SNR
Current NAOMI WFS
Thinned LLLCCD With Gain=1000
shot noise limit
-
8/14/2019 ccd lecture1
41/43
CCD65Aimed at TV applicationsas a substitute for image
tube sensors. 576 x 288 pixels.
Thick frontside illuminated,
peak QE of 35%.
20 x 30um pixels
CCD 60128x 128 pixel, thinned, has been built
but still under
development. For possible
application to Wavefront Sensing.
CCD 79,86,87Proposed future devices up to 1K square,
> 10 frames per second readout at
sub-electron noise levels.
Marconi LLLCCD Products 1.
Camera systems based on this
chip available winter 2001
As above
Low Priority for Marconi without
encouragement from the astronomical
community
Would subtend 51 x 39 at WHT CASS
-
8/14/2019 ccd lecture1
42/43
-
8/14/2019 ccd lecture1
43/43
Lecture slides available on the ING web:
http://www.ing.iac.es/~smt/LLLCCD/lllccd.htm