View
32
Download
0
Category
Preview:
Citation preview
OutlineIntroduction
Characterization
EMCCD Photometry
Alejandro FerreroSchool of Physics
University College Dublinalejandro.ferrero@ucd.ie
13/08/2008
Alejandro Ferrero School of Physics University College Dublin alejandro.ferrero@ucd.ieEMCCD Photometry
OutlineIntroduction
Characterization
IntroductionEMCCD vs CCDEMCCD in Photometry
CharacterizationExperimental setupParameters choiceReproducibilityActual GainSaturationLinearitySpectral responsivity and external quantum efficiency
Alejandro Ferrero School of Physics University College Dublin alejandro.ferrero@ucd.ieEMCCD Photometry
OutlineIntroduction
Characterization
EMCCD vs CCDEMCCD in Photometry
EMCCD vs CCD
◮ EMCCD allows higher temporal resolution than CCD.
Alejandro Ferrero School of Physics University College Dublin alejandro.ferrero@ucd.ieEMCCD Photometry
OutlineIntroduction
Characterization
EMCCD vs CCDEMCCD in Photometry
EMCCD vs CCD
◮ EMCCD allows higher temporal resolution than CCD.
◮ Readout noise is negligible in EMCCD.
Alejandro Ferrero School of Physics University College Dublin alejandro.ferrero@ucd.ieEMCCD Photometry
OutlineIntroduction
Characterization
EMCCD vs CCDEMCCD in Photometry
EMCCD vs CCD
◮ EMCCD allows higher temporal resolution than CCD.
◮ Readout noise is negligible in EMCCD.◮ Dynamic range depends on gain. The dynamic range is better
for slow scan. Lower dynamic range than CCD.
Alejandro Ferrero School of Physics University College Dublin alejandro.ferrero@ucd.ieEMCCD Photometry
OutlineIntroduction
Characterization
EMCCD vs CCDEMCCD in Photometry
EMCCD vs CCD
◮ EMCCD allows higher temporal resolution than CCD.
◮ Readout noise is negligible in EMCCD.◮ Dynamic range depends on gain. The dynamic range is better
for slow scan. Lower dynamic range than CCD.◮ A noise factor (
√2) is introduced by the additional
multiplication channel. This factor does not affect readoutnoise, but photon, dark and CIC noises.
Alejandro Ferrero School of Physics University College Dublin alejandro.ferrero@ucd.ieEMCCD Photometry
OutlineIntroduction
Characterization
EMCCD vs CCDEMCCD in Photometry
EMCCD vs CCD
◮ EMCCD allows higher temporal resolution than CCD.
◮ Readout noise is negligible in EMCCD.◮ Dynamic range depends on gain. The dynamic range is better
for slow scan. Lower dynamic range than CCD.◮ A noise factor (
√2) is introduced by the additional
multiplication channel. This factor does not affect readoutnoise, but photon, dark and CIC noises.
◮ Not cooled below −95oC , the limiting noise source is readoutdarkcurrent, which is not important in conventional CCDs.Cooled below −95oC , the limiting noise source is CIC.
Alejandro Ferrero School of Physics University College Dublin alejandro.ferrero@ucd.ieEMCCD Photometry
OutlineIntroduction
Characterization
EMCCD vs CCDEMCCD in Photometry
EMCCD vs CCD
◮ EMCCD allows higher temporal resolution than CCD.
◮ Readout noise is negligible in EMCCD.◮ Dynamic range depends on gain. The dynamic range is better
for slow scan. Lower dynamic range than CCD.◮ A noise factor (
√2) is introduced by the additional
multiplication channel. This factor does not affect readoutnoise, but photon, dark and CIC noises.
◮ Not cooled below −95oC , the limiting noise source is readoutdarkcurrent, which is not important in conventional CCDs.Cooled below −95oC , the limiting noise source is CIC.
◮ When photon noise limited, SNR in EMCCDs is half that inCCDs, because of the noise factor.
Alejandro Ferrero School of Physics University College Dublin alejandro.ferrero@ucd.ieEMCCD Photometry
OutlineIntroduction
Characterization
EMCCD vs CCDEMCCD in Photometry
EMCCD in Photometry
◮ Temporal resolution improves by a factor 1/G , if readout timenegligible with respect to the integration time.
Alejandro Ferrero School of Physics University College Dublin alejandro.ferrero@ucd.ieEMCCD Photometry
OutlineIntroduction
Characterization
EMCCD vs CCDEMCCD in Photometry
EMCCD in Photometry
◮ Temporal resolution improves by a factor 1/G , if readout timenegligible with respect to the integration time.
◮ The advantage disappears when limiting uncertainty is not dueto sensitivity of the detector (readout, dark, charge transfer orspurious charge noises), but to the light level of the source.
Alejandro Ferrero School of Physics University College Dublin alejandro.ferrero@ucd.ieEMCCD Photometry
OutlineIntroduction
Characterization
EMCCD vs CCDEMCCD in Photometry
EMCCD in Photometry
◮ Temporal resolution improves by a factor 1/G , if readout timenegligible with respect to the integration time.
◮ The advantage disappears when limiting uncertainty is not dueto sensitivity of the detector (readout, dark, charge transfer orspurious charge noises), but to the light level of the source.
◮ When source is bright and shot noise higher than the detectornoise; EM unable to overcome the noise and puts anadditional noise in the measurement.
Alejandro Ferrero School of Physics University College Dublin alejandro.ferrero@ucd.ieEMCCD Photometry
OutlineIntroduction
Characterization
EMCCD vs CCDEMCCD in Photometry
EMCCD in Photometry
◮ Temporal resolution improves by a factor 1/G , if readout timenegligible with respect to the integration time.
◮ The advantage disappears when limiting uncertainty is not dueto sensitivity of the detector (readout, dark, charge transfer orspurious charge noises), but to the light level of the source.
◮ When source is bright and shot noise higher than the detectornoise; EM unable to overcome the noise and puts anadditional noise in the measurement.
◮ When the source is faint and shot noise of the backgroundmakes the source undetectable; EM useless, because themultiplication does not distinguish between source andbackground photo-electrons.
Alejandro Ferrero School of Physics University College Dublin alejandro.ferrero@ucd.ieEMCCD Photometry
OutlineIntroduction
Characterization
EMCCD vs CCDEMCCD in Photometry
EMCCD in Photometry
Figure: Different illumination levels in a source. Some pixels are favouredby EM, whereas the SNR of other pixels may decrease due to the EM.
Alejandro Ferrero School of Physics University College Dublin alejandro.ferrero@ucd.ieEMCCD Photometry
OutlineIntroduction
Characterization
EMCCD vs CCDEMCCD in Photometry
EMCCD in Photometry
◮ In any case, the SNR of the brighter sources will be alwayshigher than the SNR of the fainter sources.
Alejandro Ferrero School of Physics University College Dublin alejandro.ferrero@ucd.ieEMCCD Photometry
OutlineIntroduction
Characterization
EMCCD vs CCDEMCCD in Photometry
EMCCD in Photometry
◮ In any case, the SNR of the brighter sources will be alwayshigher than the SNR of the fainter sources.
◮ Therefore, if a chosen gain allows to detect a faint star, itshould be never a problem the detection of brighter sources,although the relative uncertainty will increase a factor
√2 for
these sources.
Alejandro Ferrero School of Physics University College Dublin alejandro.ferrero@ucd.ieEMCCD Photometry
OutlineIntroduction
Characterization
EMCCD vs CCDEMCCD in Photometry
EMCCD in Photometry
◮ SNR is proportional to√
Gtint .
Alejandro Ferrero School of Physics University College Dublin alejandro.ferrero@ucd.ieEMCCD Photometry
OutlineIntroduction
Characterization
EMCCD vs CCDEMCCD in Photometry
EMCCD in Photometry
◮ SNR is proportional to√
Gtint .
◮ The lower limit of tint depends on the photon flux, but aboveall on the CCD, because the accuracy decreases below 0.1 secof integration time. In addition, below tint << treadout
temporal resolution can not be improved.
Alejandro Ferrero School of Physics University College Dublin alejandro.ferrero@ucd.ieEMCCD Photometry
OutlineIntroduction
Characterization
EMCCD vs CCDEMCCD in Photometry
EMCCD in Photometry
◮ SNR is proportional to√
Gtint .
◮ The lower limit of tint depends on the photon flux, but aboveall on the CCD, because the accuracy decreases below 0.1 secof integration time. In addition, below tint << treadout
temporal resolution can not be improved.
◮ The upper limit of G is imposed by the aging. The higher G ,the more the aging. Very high gains decrease the dynamicrange of the detector. Since the sources are distributed in awide range of irradiance levels on the EMCCD, the largestdynamic range gain should be chosen.
Alejandro Ferrero School of Physics University College Dublin alejandro.ferrero@ucd.ieEMCCD Photometry
OutlineIntroduction
Characterization
EMCCD vs CCDEMCCD in Photometry
EMCCD noise
σ2(npe) = σ2
r (npe) + npe + nd (1)
σ2(nae) = σ2
r (nae) + 2G (nae + nad) (2)
σ2(N) = σ2
r (N) + 2GKN (3)
GK =σ2(N) − σ2
r (N)
2N(4)
Alejandro Ferrero School of Physics University College Dublin alejandro.ferrero@ucd.ieEMCCD Photometry
OutlineIntroduction
Characterization
EMCCD vs CCDEMCCD in Photometry
Theoretical effect of EM on a source
Figure: Theoretical effect of EM on a source in the border of thedetection (SNR∼3).
Alejandro Ferrero School of Physics University College Dublin alejandro.ferrero@ucd.ieEMCCD Photometry
OutlineIntroduction
Characterization
EMCCD vs CCDEMCCD in Photometry
EMCCD vs CCD in Photometry
1
10
100
1000
10000
6 8 10 12 14 16 18 20 22
SN
R
Magnitude
CCDEMCCD (G=3.8)
3
Figure: EMCCD (G=3.8) SNR vs conventional mode CCD SNR.
Alejandro Ferrero School of Physics University College Dublin alejandro.ferrero@ucd.ieEMCCD Photometry
OutlineIntroduction
Characterization
Experimental setupParameters choiceReproducibilityActual GainSaturationLinearitySpectral responsivity and external quantum efficiency
Experimental setup
Figure: Radiant source for radiometrical calibration.
Alejandro Ferrero School of Physics University College Dublin alejandro.ferrero@ucd.ieEMCCD Photometry
OutlineIntroduction
Characterization
Experimental setupParameters choiceReproducibilityActual GainSaturationLinearitySpectral responsivity and external quantum efficiency
Andor IXON DU-897 EMCCD: Parameters choice
◮ Integration time > 0.05sec .
◮ Vertical Pixel Shift: Shift Speed 2.2 µs, Vertical ClockVoltage Normal
◮ Horizontal Pixel Shift: Readout rate 1MHz@16bits,Preamplifier Gain × 1.
◮ Temperarature −80oC .
Alejandro Ferrero School of Physics University College Dublin alejandro.ferrero@ucd.ieEMCCD Photometry
OutlineIntroduction
Characterization
Experimental setupParameters choiceReproducibilityActual GainSaturationLinearitySpectral responsivity and external quantum efficiency
Stabilization
600
700
800
900
1000
1100
1200
1300
1400
1500
1600
0 2 4 6 8 10 12
-100
-80
-60
-40
-20
0
Res
pons
e (c
ount
s)
Tem
pera
ture
(C
elsi
us)
Time (minutes)
G=300, tint=0.3 secTemperature
Figure: The response of the EMCCD is clearly temperature dependentand it is stabilized at 0.16% after around 5 minutes.
Alejandro Ferrero School of Physics University College Dublin alejandro.ferrero@ucd.ieEMCCD Photometry
OutlineIntroduction
Characterization
Experimental setupParameters choiceReproducibilityActual GainSaturationLinearitySpectral responsivity and external quantum efficiency
Reproducibility
1560
1580
1600
1620
1640
0 1 2 3 4 5 6 7 8
Res
pons
e (c
ount
s)
Time (minutes)
Conventional mode, tint=1 sec, 800 counts subtractedG=300, tint=0.3 sec
Figure: At G=300 and at conventional mode the stability was 0.1% and0.2%, respectively.
Alejandro Ferrero School of Physics University College Dublin alejandro.ferrero@ucd.ieEMCCD Photometry
OutlineIntroduction
Characterization
Experimental setupParameters choiceReproducibilityActual GainSaturationLinearitySpectral responsivity and external quantum efficiency
Photon Transfer Technique
0
500
1000
1500
2000
2500
3000
3500
0 500 1000 1500 2000 2500 3000
Var
ianc
e (c
ount
s2 )
<N-Nd> (counts)
G=3G=6
G=12G=25G=50
G=100G=200G=300
Figure: Photon transfer technique result.
Alejandro Ferrero School of Physics University College Dublin alejandro.ferrero@ucd.ieEMCCD Photometry
OutlineIntroduction
Characterization
Experimental setupParameters choiceReproducibilityActual GainSaturationLinearitySpectral responsivity and external quantum efficiency
GK product
0.1
1
10
1 10 100 1000 0.1
1
10
GK
(ph
oto-
elec
tron
s/co
unts
)
Rea
dout
noi
se (
phot
o-el
ectr
ons)
Display Gain
Figure: GK product and readout noise as a function of the softwaredisplayed gain.
Alejandro Ferrero School of Physics University College Dublin alejandro.ferrero@ucd.ieEMCCD Photometry
OutlineIntroduction
Characterization
Experimental setupParameters choiceReproducibilityActual GainSaturationLinearitySpectral responsivity and external quantum efficiency
Actual Gain
1
10
100
1000
1 10 100 1000
Act
ual G
ain
Display Gain
Photon-transfer with more than 10 levelsPhoton-transfer with only one level
Signal ratios comparison
Figure: Actual gain as a function of the software displayed gain,assuming that they are identical at G=3.
Alejandro Ferrero School of Physics University College Dublin alejandro.ferrero@ucd.ieEMCCD Photometry
OutlineIntroduction
Characterization
Experimental setupParameters choiceReproducibilityActual GainSaturationLinearitySpectral responsivity and external quantum efficiency
Saturation
0.6
0.8
1
1.2
1.4
100 1000 10000 100000
<N
-Nd>
/tint
(N
orm
aliz
ed)
<N-Nd> (counts)
Conventional modeG=50
G=100G=200G=300
Figure: Saturation of EMCCD at several gains.
Alejandro Ferrero School of Physics University College Dublin alejandro.ferrero@ucd.ieEMCCD Photometry
OutlineIntroduction
Characterization
Experimental setupParameters choiceReproducibilityActual GainSaturationLinearitySpectral responsivity and external quantum efficiency
Linearity
0.97
0.98
0.99
1
1.01
1.02
0 500 1000 1500 2000 2500 3000
Line
arity
fact
or
<N-Nd> (counts)
G=300, level 1G=300, level 2G=300, level 3G=300, level 4
Conventional mode
Figure: EMCCD linearity.
Alejandro Ferrero School of Physics University College Dublin alejandro.ferrero@ucd.ieEMCCD Photometry
OutlineIntroduction
Characterization
Experimental setupParameters choiceReproducibilityActual GainSaturationLinearitySpectral responsivity and external quantum efficiency
Spectral responsivity and external quantum efficiency
0
0.05
0.1
0.15
0.2
0.25
300 400 500 600 700 800 900 1000 1100 0
0.2
0.4
0.6
0.8
1
Res
pons
ivity
(co
unts
/pho
ton)
Ext
erna
l qua
ntum
effi
cien
cy
Wavelength (nm)
Figure: Spectral responsivity and external quantum efficiency.
Alejandro Ferrero School of Physics University College Dublin alejandro.ferrero@ucd.ieEMCCD Photometry
OutlineIntroduction
Characterization
Experimental setupParameters choiceReproducibilityActual GainSaturationLinearitySpectral responsivity and external quantum efficiency
Alejandro Ferrero
School of PhysicsUniversity College Dublinalejandro.ferrero@ucd.ie
Alejandro Ferrero School of Physics University College Dublin alejandro.ferrero@ucd.ieEMCCD Photometry
Recommended