bulk silicon ccds, point spread functions, and photon transfer curves: ccd testing activities at eso...
TRANSCRIPT
Bulk Silicon CCDs, Point Spread Functions, and Photon Transfer Curves:
CCD Testing Activities at ESO
Mark Downing, Dietrich Baade, Sebastian Deiries, (ESO/Instrumentation Division),
Paul Jorden (e2v technologies).
13 Oct 2009 1DfA 2009: Bulk CCD, PSF, &
PTC.
Agenda:• Bulk Silicon CCDs • PSF• Photon Transfer Curves
The CCD Silicon Family
• System designers can now choose the silicon thickness that best suits their application.
13 Oct 2009DfA 2009: Bulk CCD, PSF, &
PTC. 2
• 16 um standard silicon - 100 ohm-cm.
The CCD Silicon Family
• System designers can now choose the silicon thickness that best suits their application.
13 Oct 2009DfA 2009: Bulk CCD, PSF, &
PTC. 3
• 16 um standard silicon - 100 ohm-cm. • 40 um deep depletion - 1500 ohm-cm.
The CCD Silicon Family
• System designers can now choose the silicon thickness that best suits their application.
13 Oct 2009DfA 2009: Bulk CCD, PSF, &
PTC. 4
• 16 um standard silicon - 100 ohm-cm. • 40 um deep depletion - 1500 ohm-cm. • 70 um bulk silicon - > 3000 ohm-cm.
The CCD Silicon Family
• System designers can now choose the silicon thickness that best suits their application.
13 Oct 2009DfA 2009: Bulk CCD, PSF, &
PTC. 5
• 16 um standard silicon - 100 ohm-cm. • 40 um deep depletion - 1500 ohm-cm. • 70 um bulk silicon - > 3000 ohm-cm. • 150 um high-rho - > 3000 ohm-cm.
The CCD Silicon Family
• System designers can now chose the silicon thickness that best suit their application.
13 Oct 2009DfA 2009: Bulk CCD, PSF, &
PTC. 6
• 16 um standard silicon - 100 ohm-cm. • 40 um deep depletion - 1500 ohm-cm. • 70 um bulk silicon - > 3000 ohm-cm. • 150 um high-rho - > 3000 ohm-cm. • 300 um high-rho - > 3000 ohm-cm.
Bulk Silicon CCD
13 Oct 2009DfA 2009: Bulk CCD, PSF, &
PTC. 7
Why the Interest in Bulk Silicon? • Pin and mechanically compatible with existing detector family.• Upgrades are plug and play
→ No rewiring of cryostats,
→ No modification to controllers (to provide high voltages),
→ Standard clock and bias voltages used
→ No re-writing of timing patterns.
• Improve observing efficiency in the “red” without major costs in manpower, controller, instrument down time or schedule risk.
• Objective - to prove performance is as good as current CCDs.• One engineering and one science grade loaned by e2v for test
and evaluation.
13 Oct 2009DfA 2009: Bulk CCD, PSF, &
PTC. 8
Nothing unusual in performance and as good as other family members
13 Oct 2009DfA 2009: Bulk CCD, PSF, &
PTC. 9
Parameter Results(-120 ºC)
Comment
Device Bulk Silicon Science
Serial Number 07382-24-01
Type Number CCD44-82
Extensively used at ESOPixel Size 15μm
Number of Pixels 2048 x 4096
Noise (50 kpix/s) < 2.5 e- rms Gain of 0.6 e-/ADU
Noise (225 kpix/s) < 4 e- rms Gain of 1.6 e-/ADU
Linearity (500e- to 100 ke-)
< ± 0.2%Photon Transfer Curve method - not fully optimized
Dark Current(e-/pixel/hour)
< 0.2Limited by extraneous sources and not CCD
Cosmic hit event rate (events/min/cm²)
3.0
Vertical CTE 0. 9999991 EPER – Extended Pixel Edge ResponseHorizontal CTE 0. 999996
ESO-e2v QE agree
• Good agreement between e2v and ESO results.
• Determining QE depends on knowing gain precisely and care must be taken as:
– calculated gain varies with signal level when using the photon transfer curve (SPIE 2006 Downing et. al.).
• Recommend using binning (2x2 or 4x4) to determine gain.
13 Oct 2009DfA 2009: Bulk CCD, PSF, &
PTC. 10
ESO-e2v QE agree
• Good agreement between e2v and ESO results.
• Determining QE depends on knowing gain precisely and care must be taken as:
– calculated gain varies with signal level when using the photon transfer curve (SPIE 2006 Downing et. al.).
• Recommend using binning (2x2 or 4x4) to determine gain.
13 Oct 2009DfA 2009: Bulk CCD, PSF, &
PTC. 11
ESO-e2v QE agree
• Good agreement between e2v and ESO results.
• Determining QE depends on knowing gain precisely and care must be taken as:
– calculated gain varies with signal level when using the photon transfer curve (SPIE 2006 Downing et. al.).
• Recommend using binning (2x2 or 4x4) to determine gain.
13 Oct 2009DfA 2009: Bulk CCD, PSF, &
PTC. 12
PRNU Measured with 7nm Bandwidth
• PRNU shows very low fringing in the “red”.
13 Oct 2009DfA 2009: Bulk CCD, PSF, &
PTC. 13
• 70 um bulk silicon
PRNU Measured with 7nm Bandwidth
• PRNU shows very low fringing in the “red”.
13 Oct 2009DfA 2009: Bulk CCD, PSF, &
PTC. 14
• 70 um bulk silicon• 40 um deep depletion• 16 um standard silicon
PRNU Measured with 7nm Bandwidth
• PRNU shows very low fringing in the “red”.
13 Oct 2009DfA 2009: Bulk CCD, PSF, &
PTC. 15
• 70 um bulk silicon• 40um deep depletion• 16 um standard silicon
900nm Images
16um Std Si 40um DD 70um Bulk
5 – 95% Histogram Scaling
PRNU Measured with 7nm Bandwidth
• PRNU shows very low fringing in the “red”.• At shorter wavelengths (< 400 nm) PRNU is a little worse
due to thinning and laser annealing.
13 Oct 2009DfA 2009: Bulk CCD, PSF, &
PTC. 16
• 70 um bulk silicon• 40um deep depletion• 16 um standard silicon
350nm Images
16um Std Si 40um DD 70um Bulk
5 – 95% Histogram Scaling
Very acceptable cosmetics at -120 DegC
13 Oct 2009DfA 2009: Bulk CCD, PSF, &
PTC. 17
Blemish Spec. Number
Hot pixels > 100e-/pixel in 1 hour dark 14
Hot Columns > 100 bad pixels in 1 hour dark 0
Dark pixels < 50% response 112
Dark Columns > 100 bad pixels 0
Traps > 200e- 2
Traps
At higher temperatures, hot pixels become a problem.
• Hot pixels are highly temperature dependent.• At < -120 DegC, hot pixels are not a problem and the device is
of excellent scientific grade.• -120 DegC is ESO’s standard operating temperature for all
CCD44-82s at the observatories.• Hot pixels are due to impurities in the silicon.
13 Oct 2009DfA 2009: Bulk CCD, PSF, &
PTC. 18
Temperature(DegC)
Average Dark Current
(e-/pix/hr)
Number of Hot Pixels
(1 hour dark)
Comments
-120 0.2 14
-100 5.5 160500 Average dark current dominated by hot pixels.-80 310 661025
At higher temperatures, hot pixels become a problem.
• Hot pixels are highly temperature dependent.• At < -120 DegC, hot pixels are not a problem and the device is
of excellent scientific grade.• -120 DegC is ESO’s standard operating temperature for all
CCD44-82s at the observatories.• Hot pixels are due to impurities in the silicon.
13 Oct 2009DfA 2009: Bulk CCD, PSF, &
PTC. 19
Temperature(DegC)
Average Dark Current
(e/pix/hr)
Number of Hot Pixels
(1 hour dark)
Comments
-120 0.2 14
-100 5.5 160500 Average dark current dominated by hot pixels.-80 310 661025
-80DegC -100DegC -120DegC
Hot pixels are fixable.
• Note no hot pixels at the edges.• e2v are working on it to have the whole device as good as the
edges.
13 Oct 2009DfA 2009: Bulk CCD, PSF, &
PTC. 20
OverscanPixels
OverscanPixels
OverscanPixels
Edges have no hot pixels
Edges have no hot pixels
CCD Top
CCD Bottom
-80DegC 1 hour dark
Effect of cosmic rays become worse with thicker devices
• The thicker the device, the more chance that a cosmic ray will affect more than one pixel.
– Standard Silicon - mostly single pixel events– Deep Depletion – mix of single and multiple pixel events– Bulk – mostly multiple pixel events
• Expect number of events to scale with thickness.– Deep Depletion Cosmic hit event rate: ~ 1.8 events/min/cm²– Bulk Cosmic hit event rate: ~ 3.0 events/min/cm²– ~ ratio of 70/40
13 Oct 2009DfA 2009: Bulk CCD, PSF, &
PTC. 21
16um Std Si 40um DD 70um Bulk
PSF
13 Oct 2009DfA 2009: Bulk CCD, PSF, &
PTC. 22
PSF depends on CCD design and wavelength
PSF depends on:
1. Thickness of the undepleted region (XUNDEP) at the back of the CCD.
2. The strength of the electric field to draw the electrons into the potential well depends on:
(Vc – Vsub) XTHICK
3. Wavelength and depth of penetration of the photon
Blue photons are in general absorbed nearer the silicon surface and have farther to travel.
While red photons penetrate on the average deeper into the silicon.
13 Oct 2009DfA 2009: Bulk CCD, PSF, &
PTC. 23
400nm
600nm
UV
VIS
900nm
RED
CCDBackside
Cross Section of CCDCross Section of CCD
ElectricField Extent
Collectionphase
CCDFrontside
PotentialWell
Vc
Vsub
XTHICK
XUNDE
P
PSF can be improved by increasing the number of collection phases and the voltage across the CCD
PSF can be improved by :• Increasing the extent (i.e. reduce
undepleted region at back of CCD) and strength of the electric field by:
Increasing the collection phase voltage (Vc).
Decreasing the substrate voltage (Vsub).
Increasing the number of collecting phases (2 for 3 phase device or 3 for 4 phase device).
13 Oct 2009DfA 2009: Bulk CCD, PSF, &
PTC. 24
400nm
600nm
UV
VIS
900nm
RED
CCDBackside
Cross Section of CCDCross Section of CCD
ElectricField Extent
Collectionphase
CCDFrontside
PotentialWell
Vc
Vsub
XTHICK
Virtual Knife Edge
1um SpotSub Window
Pixel
Spot scanning used to measure PSF
13 Oct 2009DfA 2009: Bulk CCD, PSF, &
PTC. 25
70um Bulk CCDGauss Fit PSF FWHM of MIT/LLCCID-20 DD CCD versus
wavelength and collection phase voltage
0.40
0.45
0.50
0.55
0.60
0.65
0.70
0.75
0.80
0.85
0.90
400 500 600 700 800 900
Wavelength (nm)
PS
F F
WH
M (
pix
els
)
Ph=2V
Ph=4V
Ph=6V
Ph=8V
Ph=10V
Ph=12V
40um MIT/LL Hi Rho
Gauss Fit PSF FWHM of e2v CCD44-82 Std Silicon CCD versus wavelength and collection phase voltage
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
400 500 600 700 800 900
Wavelength (nm)
PS
F F
WH
M (
pix
els
) Ph=2V
Ph=4V
Ph=6V
Ph=8V
Ph=10V
Ph=12V
16um e2v Std Si
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
400 500 600 700 800 900
PS
F F
WH
M (p
ixels
)
Wavelength (nm)
Gauss Fit PSF FWHM of e2v CCD44-82 Deep Depletion CCD versus wavelength and collection
phase voltage
Ph=2V
Ph=4V
Ph=6V
Ph=8V
Ph=10V
Ph=12V
40um e2v Deep DepletionPSF Results
VcVc
VcVc
Photon Transfer Curves
13 Oct 2009DfA 2009: Bulk CCD, PSF, &
PTC. 26
Photon Transfer Family of Curves
13 Oct 2009DfA 2009: Bulk CCD, PSF, &
PTC. 27
70um Bulk CCD
16um e2v Std Si
• Increasing the collection phase voltage to improve the PSF impacts the well depth.
40um e2v Deep Depletion
Gain ~ 11e/ADU
16um Standard Silicon is well behaved; text book
13 Oct 2009DfA 2009: Bulk CCD, PSF, &
PTC. 28
70um Bulk CCD
40um E2v Deep Depletion16um e2v Std Si
Gain ~ 11e/ADU
Bloomed Full Well
Surface Full WellBFW=SFW
Vc
Optimum full well = ~ 2V
40 um Deep Depletion starts to show interesting behaviour
13 Oct 2009DfA 2009: Bulk CCD, PSF, &
PTC. 29
70um Bulk CCD
16um E2v Std Si40um e2v Deep Depletion
Gain ~ 11e/ADU
Bloomed Full Well
Surface Full WellBFW=SFW
Vc
Note interesting behaviour at 2-5V.
Interesting behaviour up to where surface full well dominate
• Behaviour enables a larger full well and different optimum full well voltage to standard silicon; 6V versus 2V.
• No explanation yet for the behaviour.
13 Oct 2009DfA 2009: Bulk CCD, PSF, &
PTC. 30
Between 2-4V, starts to bloom at low signal level but then recovers at higher levels.
No blooming
Blooming
Minimal blooming
At 6V, no blooming is observed.
No blooming
Blooming
70 um Bulk is even more pronounced
13 Oct 2009DfA 2009: Bulk CCD, PSF, &
PTC. 31
16um E2v Std Si 40um E2v Deep Depletion
Gain ~ 11e/ADU
70um Bulk CCD
Full Well versus Collection Phase Voltage
• Choose trade between PSF improvement and well depth
13 Oct 2009DfA 2009: Bulk CCD, PSF, &
PTC. 32
70um Bulk CCD
Vc
Vc
Optimum full well = ~ 6V
Conclusion Performance at -120DegC of noise, gain, linearity, cosmetic, dark current,
and CTE is as good as previous e2v CCD44-82s.
Below -120DegC, hot pixels are not a problem.
Bulk delivers better QE in the “red” and much less fringing.
With PSF of ~ 1 pixel, the bulk CCD is very suitable for not too demanding optical designs.
PSF can be improved by increasing collection phase voltage or running at a lower active substrate voltage.
When increasing collection phase voltage, one has to be careful about change in well depth.
As the resistivity of the silicon is increased, the photon transfer curve becomes more interesting and does not agree with the text books.
13 Oct 2009DfA 2009: Bulk CCD, PSF, &
PTC. 33
END
Many thanks
13 Oct 2009 34DfA 2009: Bulk CCD, PSF, &
PTC.