Astronomical Imaging Detectors
Astronomical Imaging Detectors
Juan Estrada 2/21/2012
1
Wednesday, February 22, 2012 Astronomical Imaging Detectors
(mostly optical... and mostly CCDs)
Juan Estrada 2/21/2012
2
Wednesday, February 22, 2012 first thing you should know when thinking about detectors to look into space
the sky is mostly dark!3 Wednesday, February 22, 2012 ... but with the right detector it can be fun! Wednesday, February 22, 2012 To give you an idea:
Naked eye can see stars of magnitude 6 which means about 1000 photons per second.
modern astronomical instruments we are trying to see objects in the sky of magnitude 24, which means about 15 photons per second on a 4 meter telescope (for your eye this would be 5e-5 photons per second).
which brings me to another important point for detectors in astronomy... timescale = seconds (not nanoseconds). 5
Wednesday, February 22, 2012 What are the options for visible and IR? Mainly CMOS and CCDs... will go over CCDs and then discuss the difference with CMOS
6
Wednesday, February 22, 2012 :)/&$;/).2).;&'(*%)<)3=);>.&.+),$&$(&'.+
D+&'E6$2%$(&'.+)(.*&'+5 1FG/&6*&$)6$4.?*% !A))-'5>&)'+&.),$&$(&.6 OF*+&F4 P22'('$+(@ B$&$(&.6)J*&$6'*%/ 1'K)L5H,M$K)3+1GK)1'ND/ 8A))H>*65$)I$+$6*&'.+
P%$(&6'()R'$%,/)'+),$&$(&.6 S.'+&)1;6$*,)RF+(&'.+ (.%%$(&)$%$(&6'(*%)(>*65$ QA))H>*65$)H.%%$(&'.+
HHB HJ71 H>*65$)(.F;%$,H>*65$)(.F;%$, UA))H>*65$ UA))H>*65$E&.E 1.F6($)2.%%.T$6K1.F6($)2.%%.T$6K &6*+/2$6&6*+/2$6 M6*+/2$6 V.%&*5$ HM3DK)B3HM3DK)B3 1$+/'&?'&@ H.+?$6/'.+ WA))H>*65$E&.E J71RPMJ71RPM =*+,.4)*(($//=*+,.4)*(($// D4;%'2'$6D4;%'2'$6 V.%&*5$ WA))1'5+*% .6)2F%%)26*4$)6$*,.6)2F%%)26*4$)6$*, H.+?$6/'.+ M6*+/2$6
a few slides from J. Beletic :A))B'5'&'C*&'.+ 7 #$%$&'()*+,)-../$)0 1('$+&'2'()34*5'+5)1$+/.6/)0 7(&)899" !" Wednesday, February 22, 2012 2009 Nobel Prize in Physics awarded to the inventors of the CCD In 1969, Willard S. Boyle and George E. Smith invented the first successful imaging technology using a digital sensor, a CCD (charge-coupled device). The two researchers came up with the idea in just an hour of brainstorming.
Bell Labs researchers Willard Boyle (left) and George Willard S. Boyle George E. Smith Smith (right) with the charge-coupled device. !"#$#%$&'()%*)%+,-./%%%!"#$#%01(2*$3%450&$(56780()$9:(55%7&;
Wednesday, February 22, 2012 Light into the detector
100
80 1 ! R r o ) Old UCO/Lick y % c (
n y e t
i 60 i c v i i
f light has to get inside the detector t f c e
e
l QE f m for detection. This means that
e 40 u t R
n - a destructive interference has to be u 1 Q 20 example for DECam CCD accommodated for reflections.
0 200 400 600 800 1000 1200 qe_photonics06.eps Wavelength (nm) 9
Figure 7. First results with The Quantum Efficiency Machine (solid circles and error band). 1 R is also shown. Old Wednesday, February 22, 2012 − UCO/Lick measurements are shown for comparison; that CCD had nominally the same antireflective coating but was thicker, so that the IR response is higher.
CCD with nominally the same antireflective (AR) coating are shown by the open circles. Results obtained with the CCD connected as a diode are consistent with the CCD-mode results. Our < 1% error level goal means that most contributions must be far less than this. If a systematic error is one-sided, then the measurement is corrected by half this error so that there is not a systematic bias. Most conceivable sources of error have been investigated and found to be negligible at this level: Filter out-of-bandpass leakage, 2nd order light from the monochromator, light leaks, noise in photodiode current measurements, picoammeter offset, and monochromator wavelength uncertainties. The dark current in a 1 cm2 monitor diode in the integrating sphere is about 50 fA, so that neither light leakage nor PD noise is significant. Intensity and hence exposure times, typically 5–10 s, are set by the requirement that the PD current should 13 be well above noise in the picoammeter, which is less than 10− A. For longer exposures at reduced intensity, the monitor PD in the integrating sphere, calibrated using the standard PD, is used as the reference. A normal scan consists of automatically repeating the following: The monochromator and (if called for) the filter wheel are moved to the new wavelength, after which • there is a 500 s settling time. The motorized input and output slits are adjusted to obtain the desired intensity at the dewar. • A 10 s dark is obtained with the large-aperture shutter closed. • A 10 s exposure is obtained with the shutter open, at the end of which it closes. • Because the PD is coplanar with the CCD in the dewar, there are no geometrical or window-reflection corrections, except for the very tiny radial intensity variation shown in Fig. 4. Fiducial areas are a different problem. So far it has been assumed that the PD area is exactly 1 cm2, but this needs to be checked. Similarly, if the CCD is configured as a photodiode, a mask will be needed to define/check the fiducial area. So far we have simply taken the pixel area as correct, but there may be edge effects. Masks have been obtained but not yet installed. The Uniblitz shutter remains open during the scan described above. The large-aperture shutter on which we presently depend is somewhat problematical. It takes 29 ms to open and 65 ns to close, which means that the center (the CCD fiducial area) and reference PD are illuminated for different times. Since aperture is star-shaped during the process, a correction is difficult to calculate. Better schemes are under discussion, and it will probably be moved to the light-entrance end of the dark box.
107 Charge Generation
Charge generation
The1'%'(.+)::; photoelectric 1'4'%*6)<=>/'(/)2.6)effect makes it possible3?)4*&$6'*%/ for photons with more than 1.1eV to produce electron-hole pairs in Silicon.
!" #$%$&'()*+,)-../$)0 1('$+&'2'()34*5'+5)1$+/.6/)0 7(&)8""9 10
Wednesday, February 22, 2012 Photon Detection
=-5()*(#$#'%5-*(%-(>#(#?'&%#+(15-3(%@# :-*+;'%&-*(")*+ '-*+;'%&-*(>)*+(%-(%@#(A)$#*'#(>)*+ !g !! B !g <)$#*'#(")*+
!"C(D$)*'E('-*.%)*%(FGHG!I8J!K L-;$#M.#'N " C(IH7!S(Q ! F#
T,)%%&'#(3)%'@#+(2*V)W.(F2* V) W.N IR detectors work with the same8HX! principle,8HKU but with crystals that are a lot more !! "#$#%&'()*+(,--.#(/expensive 0'&#*%&1&'(23)4&*4(0#*.-5.(/ than Silicon 6'%(7889 11
Wednesday, February 22, 2012 Spectral Bands :#1&*#+(;<()%3-.=>#5&'(%5)*.3&..&-*(?(+#%#'%-5(3)%#5&)$(=5-=#5%&#.
!"#$%&'()*+,-)./%#*%%*$/
@)A#$#*4%>(B3&'5-*.C
:#%#'%-5 D--$-4<
"#$#%&'()*+(,--.#(/ 0'&#*%&1&'(23)4&*4(0#*.-5.(/ 6'%(7889 !!
12
Wednesday, February 22, 2012 Dark Currentcharge generation of Silicon-based of the unwanted Detectors type
Dark Current of e2v CCDs
1E6 MAXIMUM VALUES 1E5
1E4 1;62*($ <*6=)>;66$+& 1E3
1E2 #;%= 1E1 <*6=)>;66$+& 1E0
1E-1 Electrons/sec/15 micron pixel 1E-2
1E-3 -120 -100 -80 -60 -40 -20 0 20 40 Temperature, °C e2v TECHNOLOGIES so youIn silicon, will dark usually current usua havelly dominated to cool by surface the defectsdetectors... !"#$%!&'()*)+,!- #$%$&'()*+,)-../$)0 1('$+&'2'()34*5'+5)1$+/.6/)0 7(&)899: 13 !"
Wednesday, February 22, 2012 PhotovoltaicCharge Detector Collection Potential Well n-channel CCD L.&*)?$+$)M 345 B =*+)(.%%$(&)$'&A$6 Si P $%$(&6.+/).6)A.%$/
#.6.+),.@$, #.6.+),.@'+5
NA./@A.6.E/),.@$, NA./@A.6.E/),.@'+5
1'%'(.+;)<5=,>$)*+,)3+1?)*6$)@A.&.B.%&*'(),$&$(&.6/C))D%%)E/$)*)@+FGE+(&'.+) &.)By 5$+$6*&$) doping HF2'$%,) the Si '+) &A$)we IF,'6$(&'.+)get a E .2field) $*(A) that @'J$%C) moves ) >A'/) $%$(&6'()the charges 2'$%,) /$@*6*&$/)&A$)$%$(&6.+FA.%$)@*'6/)5$+$6*&$,)?K)*)@A.&.+Cfrom the generation site to the potential well. #$%$&'()*+,)-../$)0 1('$+&'2'()34*5'+5)1$+/.6/)0 7(&)899: !" 14
Wednesday, February 22, 2012 ChargeCCD Transfer Rain bucket in a analogyCCD
#$%$&'()*+,)-../$)0 1('$+&'2'()34*5'+5)1$+/.6/)0 7(&)899! 15 !"
Wednesday, February 22, 2012 2009 Nobel Prize in Physics awarded to the inventors of the CCD In 1969, Willard S. Boyle and George E. Smith invented the first successful imaging technology using a digital sensor, a CCD (charge-coupled device). The two researchers came up with the idea in just an hour of brainstorming. First CCD
Bell Labs researchers Willard Boyle (left) and George Willard S. Boyle George E. Smith Smith (right) with the charge-coupled device. !"#$#%$&'()%*)%+,-./%%%!"#$#%01(2*$3%450&$(56780()$9:(55%7&;
16
Wednesday, February 22, 2012 CCD Architecture Charge Transfer: CCD architecture
17
Wednesday, February 22, 2012
#$%$&'()*+,)-../$)0 1('$+&'2'()34*5'+5)1$+/.6/)0 7(&)899: !" CCD Timing
9.:$4$+&)) .2)(;*65$ '/)<(.=>%$,?
C;*65$ C.=>%$, D$:'($
#$%$&'()*+,)-../$)0 1('$+&'2'()34*5'+5)1$+/.6/)0 7(&)8""! !" 18
Wednesday, February 22, 2012 CCD Timing
9.:$4$+&)) .2)(;*65$ '/)<(.=>%$,?
C;*65$ C.=>%$, D$:'($
#$%$&'()*+,)-../$)0 1('$+&'2'()34*5'+5)1$+/.6/)0 7(&)8""! !" 18
Wednesday, February 22, 2012 !"#$%"$#&'()*&("#+,!(&-"#+%"(+(!.+//(+.0$1"(02(%*+#3&(+14(#&/&+!&(5"(+"(+(/+"&#("5.&(4522$!513("*&(%*+#3&(51"0(+10"*&#( ,5-&/'(65("#+,!(%+1(7&(20$14(&8&1(7&20#&(99:(2+7#5%+"501'(;0<+4+=!(+(2+7#5%+"&4(99:(<5//(*+8&(8&#=(2&<("#+,!'(
>1&"5%(,+#"5%/&!(!$%*(+!(,#0"01!?(&/&%"#01!?(1&$"#01!?(*&+8=(501!?(+14(3+..+(#+=!(%+1(+/!0(3&1&#+"&("#+,!'()*5!( ,#07/&.(5!(.0#&(!&8&#&(20#(!,+%&(.5!!501(%+.&#+!(<&#&(99:!(+#&(.0#&(&-,0!&4("0(#+45+"501'(
9*+#3&(.&+!$#&.&1"@(
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harge#&+40$"(105!&( Measurement5.,/5&!(+(/+#3&#("&/&!%0,&'(
( Reset the output to a known level, sample this level, dump C53$#&(E( the charge into the sense node and check the level again C53$#&(S(!*0
Wednesday, February 22, 2012 signal two integration Charge windows per pixel reference RST transfer 20 Wednesday, February 22, 2012 RG (gate) SW Vref (drain) TG
drain source (VOUT)
21 #$%$&'()*+,)-../$)0 1('$+&'2'()34*5'+5)1$+/.6/)0 7(&)"88! !" Wednesday, February 22, 2012 Alternative to CCDs : Monolithic Monolithic CMOS CMOS ? A monolithic CMOS image sensor combines the photodiode and the readout circuitry in one piece of silicon 0 Photodiode and transistors share the area => less than 100% fill factor 0 Small pixels and large arrays can be produced at low cost => consumer applications (digital cameras, cell phones, etc.) 3T Pixel Reset
SF
PD Select Read Bus
photodiode transistors Micro4T Pixel lenses can be used to improveReset fill factor. Not yet achieving the noise performance TG SF ofPinned CCDs. PD :> +> +> :;/<= Select Read Bus 22
Wednesday, February 22, 2012 #$%$&'()*+,)-../$)0 1('$+&'2'()34*5'+5)1$+/.6/)0 7(&)"889 !" Hybrid Imager Architecture
HYBRIDHybridHybrid CMOSCMOSCMOS InfraredInfrared ImagingImaging SensorsSensors
NOPQRB9 O9:!SO9:!)E'S$%/ B9)4'(6.+)E'S$%)E'&(= NHT'13)/'%'(.+)U3V
Photo courtesy of ;*&<6$)'+&$6(.++$(&)&$(=+'><$? Raytheon Vision Systems @ 7A$6)B!C999C999)'+,'<4) D<4E/)E$6)1$+/.6)F='E) G//$4D%H)I1FGJ) Human ,$4.+/&6*&$, hair @ K::L:M)'+&$6(.++$(&)H'$%, Large, high performance IR arrays !"#"$%&"'()*+,&+'-"&./0. Three Key Technologies
#$%$&'()*+,)-../$)0 1('$+&'2'()34*5'+5)1$+/.6/)0 7(&)899: !" !"many #$%&'()*+,)-$%./001+2)%3)'(/) applications in IR astronomy.425,6/),/'/.'%$)7*8/$ Detectors are not Silicon anymore, readout array 9" :/012+)*+,)3*;$1.*'1%+)%3)'(/) is still silicon. 5<=>)$/*,%?')1+'/2$*'/,).1$.?1') @A=B5C) D" 48;$1,1E*'1%+)%3)'(/),/'/.'%$)7*8/$) '%)'(/)5<=>)A=B5 23 66 Wednesday, February 22, 2012 The charge does not move on CMOS sensors. TheyGeneral have random Architecture access readout of forCMOS-Based all channels. Image Sensors
Bias Generation & DACs Control :-;%&-*)$< Pixel Array & Timing Vertical Scanner
Logic for Row Selection :-;%&-*)$< A/D conversion :-;%&-*)$< Digital Output Horizontal Scanner Analog / Column Buffers Amplification Analog Output
24 "#$#%&'()*+(,--.#(/ 0'&#*%&1&'(23)4&*4(0#*.-5.(/ 6'%(7889 !! Wednesday, February 22, 2012 EMCCDs : CCDs with charge multiplication
This is still very new.
Gain of 10-1000 are possible. In principle it allows for photon counting, which is very nice.
At large signals, the gain fluctuations become larger than the poisson noise, so the S/N drops compared to normal CCDs.
Allows for frame rates of 1 kHz readout.
Recently people have reported some aging effects on the gain.
for the moment small from Hamamatsu lucky imaging 25
Wednesday, February 22, 2012 Dark Energy Camera (DECam)
New wide field imager (3 sq-deg) for the Blanco 4m telescope to be delivered in 2011 in exchange for 30% of the telescope time during 5 years. Being built at FNAL by a large collaboration.
Blanco 4m Telescope Mechanical Interface of Cerro Tololo, Chile CCD DECam Project to the Blanco Readout Filters Shutter
Hexapod
Optical Lenses
Wednesday, February 22, 2012 One night at the Blanco 4m telescope in Chile (R. Smith)
27
Wednesday, February 22, 2012 One night at the Blanco 4m telescope in Chile (R. Smith)
27
Wednesday, February 22, 2012 ...and do this (DES) with this (DECam) Overlap with South Pole Survey Area Telescope Survey 43cm (4000 sq deg)
replace this (mosaicII) Connector region (800 sq deg)
Overlap with SDSS Stripe 82 for calibration (200 sq deg)
Science goals (Dark Energy, z~1): Galaxy Cluster counting to do this from the ground Spatial clustering of galaxies (BAO) we need detectors with high Weak lensing efficiency in the near-IR Supernovae type Ia (secondary survey)
28 Wednesday, February 22, 2012 DECAM OVERVIEW
CCD focal plane is housed in a vacuum vessel (the imager)
C5 LN2 is pumped from the telescope floor to a heat exchanger in the imager: cools the CCDs to -100 C C4 CCD readout electronic crates are mounted to the outside of the Imager and are actively cooled to eliminate thermal plumes C3 C2 Filter changer with 8 filter capacity and shutter fit between lenses C3 and C4.
Hexapod provides focus and lateral alignment capability for the corrector-imager system
C1 Barrel supports the 5 lenses and imager
Marcelle Soares-Santos ! DECam telescope simulator integration tests ! TIPP 2011 ! Chicago, June 13 2011 12 Wednesday, February 22, 2012 DECAM TELESCOPE SIMULATOR AT FERMILAB
Platform for testing Mosaic DECam operations and installation procedures prior to shipping to Chile.
Full system tests, including a mock observing run.
Imager with 24 CCDs, filter changer, shutter, hexapod, LN2 cooling, Telescope Simulator Blanco 4-m Telescope CCD readout crates. this is the ring that you saw at SiDet (Lab-A) 30
Wednesday, February 22, 2012 Marcelle Soares-Santos ! DECam telescope simulator integration tests ! TIPP 2011 ! Chicago, June 13 2011 15 DECam Focal Plane fully assembled
3 sq-deg imager:
62 2kx4k Image CCDs: 520 MPix 8 2kx2k focus, alignment CCDs 4 2kx2k guide CCDs 0.27’’/pixel (15x15 µm)
Imager to start taking data on September 2011. In exchange we get 30% of telescope time for DES during 5 years.
Facility instrument available the rest of the time.
31
Wednesday, February 22, 2012 Requirements for DECam CCDs
Table 1. DECam technical requirements. # description specification These requirements come 1 nonlinearity <1% from the science goals for 2 full well: >130,000 e− DES. Get to z~1. 3 no residual image 4 readout time < 17 sec For this we need detectors 5 dark current <35 e /pix/hour − that get higher QE in the red 6 QE [g, r, i, z]: [60%, 75%, 75%, 65%] 7 QE < 0.5 % per degree K and near-IR. Without 8 read noise <15 electrons degrading the rest of the 9 Charge diffusion σ <7.5µm performance. 10 Cosmetic defects < 0.5 % 11 Crosstalk for two amps. on CCD < 0.001.%
32 Wednesday, February 22, 2012
Figure 3. DECam science package. The CCD is attached to a gold-plated invar foot with two alignment pins for mounting on the focal plane. A high density Airborne23 connector is used for readout.
After each detector is packaged, it is installed in a single CCD testing station where all the DECam technical requirements are verified. An example of a testing station is shown in Fig.4. The staged production testing of a DECam detector takes approximately 2 days. Fermilab has established a CCD testing lab with three fully instrumented testing stations for production testing. The CCD lab has also two additional testing stations, one dedicated to measure the mechanical properties of these CCDs13 and one dedicated to specific studies that are not part of the routine production testing. These stations are often also used to characterize the performance of the custom design DECam electronics.15 The CCD packaging and testing facilities at Fermilab have been able to maintain an average testing rate of 4 detectors a week. This was done early in the project to reduce the risk typically associated with producing such a large number of science grade CCDs. Thanks to this effort, the DECam CCD team has been able to produce all the scientific detectors needed more than 18 months before the scheduled first light for the instrument, and approximately 1 year before they are needed to start building the final focal plane. The details of the production testing facility and procedures are presented in Ref.14
3. QUANTUM EFFICIENCY The main feature distinguishing the DECam detectors from other CCDs is the quantum efficiency in the red and near-infrared. DES imaging will be performed in 4 filters according to the SDSS filter set.10 The requirement Detectors : CCD DECam wafer
Engineering CCDs
Fermilab’s expertise in building silicon trackers has transfered nicely to the design and fabrication of these CCDs (strict mechanical requirements).
+100 built and tested during our R&D stage 33 Wednesday, February 22, 2012 How to get high QE in the red with Si?
104 77 K ) 103
m –100 C = 173 K
! ˚ (
2
10 Surface h t 1 effects g 10
n dominate e
l 300 K
n 100 o i t
p –1 Transparency,
r 10
o interference s
b –2 are issues
A 10
10–3 200 300 400 500 600 700 800 900 1000 1100
Si_abs_T-col.eps Wavelength (nm) Eg ! 1.147 eV Atmospheric "g ! 1081 nm cutoff (silicon bandgap at 150 K)
34
Wednesday, February 22, 2012 Focal Plane Detectors
Science goal for DES: z~1 ~50% of time in z-filter 825-1100nm
LBNL full depletion CCD –250 microns thick (instead of 20 microns) high-z –high resistivity silicon objects –QE> 50% at 1000 nm
IR image of soldering iron with DECam CCDs DECam CCD Thinned CCD
35
Wednesday, February 22, 2012 QE in the DES filters
DECam Existing Imager
36 Wednesday, February 22, 2012 Figure 5. Quantum efficiency for a DECam CCD (black) compared with the QE for the SITE detectors currently installed in the prime focus imager of the Blanco Telescope.
1.5 1.4 1.3 1.2 1.1 1 QE variation 0.9 0.8 0.7 0.6 0.5 500 600 700 800 900 1000 wavelength (nm)
Figure 6. Quantum efficiency relative variations observed in a group of 12 representative DECam CCDs. Stability/Uniformity in QE
The DECam detectors show very uniform QE. DES requires better 10% QE uniformity from CCD to CCD. measured: We also check the dependence with 0.5%/K at 1000nm temperature to establish the requirement for temperature stability and uniformity on our focal plane. requirements for focal plane: > 0.25K stability > 10K uniformity (achieved in prototype) 37 Wednesday, February 22, 2012 Charge diffusion
Holes produced in the back surface have to travel to the collection area. Thicker means more opportunity for diffusion. (fully depleted). Higher QE could get compensated by lower image quality. That is why other detectors are thinner. The 40V applied to the substrate (Vsub) to control diffusion Imaging a diffraction pattern
high Vsub
Diffusion is measured from the analysis of these images {low Vsub 38
Wednesday, February 22, 2012 Diffusion results
Results of the DES devices (blue,red and green) are compared with measurements done at LBNL for a 200 µm SNAP CCD (black). These results also show that the LBNL(2006) scaled to 250 µm devices are fully depleted well before 40 V.
Diffusion is also measured using X-rays from an Fe55 source.
substrate voltage
39
Wednesday, February 22, 2012 X-rays 2
1.7 4.2 4.8 5.9 6.5 10 4
10 3
10 2
10
FIG. 1: Four muons tracks are clearly reconstructed on this 1 image. There is a significant width different between the two 0 1 2 3 4 5 6 7 8 Figure 9. Diffusion measurement using X-rays from an 55Fe source in a 250 µm DECam CCD with pixel size 15 µm x 15 ends of the track. The thicker end corresponds to the region of µm. the track in the back of the CCD, with maximum diffusion. E (keV) The thinner end corresponds to the front of the CCD. The smaller circular hits are diffusion limited hits as expected for nuclear recoils. The line is 100 pixels long and was added to FIG. 2: Spectrum obtained for the reconstructed X-ray hits give a sense of scale. in an 55Fe exposure of a DECam CCD. The arrows mark the direct X-rays from the source: Kα=5.9 keV and Kβ=6.5 keV, their escape lines at 4.2 and 4.8 keV [24], and the Si X-ray at 40 good candidates for a direct dark matter search are their 1.7 keV. The factor 3.64 eV/e is used to convert from charge thickness, which allows the CCD to have significant mass, Wednesday,to ionization February energy. 22, The 2012 feature at 7.6 keV is consistent with and the low electronic readout noise, which permits a pixels that hit by a 5.9 keV and a 1.7 keV X-ray on the same very low energy threshold for the ionization signal pro- exposure. consistent w duced by nuclear recoils. DAMIC uses a single rectangular CCD readout with a two examples of diffusion limited hits. The energy of the Monsoon controller [22]. The detectors have 4.2 million X-ray hit is proportional to the charge collected in all pixels with dimensions 15 µm x 15 µm (active mass of the pixels that belong to this cluster. An example of the 0.5 g), and are read by two amplifiers in parallel. The energy spectrum measured for an 55Fe X-ray exposure in detectors have an output stage with electronic gain of a DECam CCD is shown in Fig. 2. 2.5 µV/e. The signal is digitized after correlated double∼ sampling (CDS) and the noise performance depends on Figure 2 provides a good calibration for X-rays in sili- con. Since the ionization efficiency for nuclear recoils dif- the readout speed. The best noise, σ < 2e− (R.M.S.) per pixel, was achieved with readout times of 50 µs per pixel fers from the ionization efficiency for X-rays, the results [19–21, 23]. are not directly applicable to nuclear recoils expected from DM interactions. The ratio between the ioniza- tion efficiency for nuclear recoils and ionization efficiency for electron recoils is usually referred to as the quench- III. CCD CALIBRATION: X-RAYS AND ing factor (Q). The quenching factor has been measured NUCLEAR RECOILS in Si for recoil energies above 4 keV [27], showing good agreement with the Lindhard model [28, 29]. For recoil X-rays are commonly used in the energy calibration energies less than 4 keV, the quenching factor becomes of CCD detectors [24]. X-rays from 55Fe penetrate only increasingly energy-dependent and no previous measure- 20 µm into the silicon before producing charge pairs ac- ments are available, as shown in Fig.3. In this work, the cording∼ to the well known conversion factor of 3.64 eV/e- Lindhard model [28, 29] is used for converting visible (or [24]. As a result of this process X-rays produce hits in electron equivalent) energy to recoil energy. We test this the detector with size limited by charge diffusion. In a assumption by comparison to neutron data with a known back illuminated DECam CCD the size of the 5.9 keV energy spectrum. X-rays is 7 µm R.M.S. [25]. Using processing tools de- In order to study Q, the DAMIC detector was exposed veloped for∼ astronomical aplications [26], the X-ray hits to a 252Cf neutron source (see Fig.4). Nuclear recoils pro- are identified in the CCD images as reconstructed charge duced by neutrons, similar to those expected from DM clusters with a size limited by diffusion. Fig. 1 shows interactions, give diffusion-limited hits analog to those of Simulated stars
As an additional check we projected “stars” on our detectors. We were able to got what would correspond to a PSF=0.43’’ FWHM for DECam (0.27’’/pixel). This is a, demonstration of good image quality with these CCDs. The CCDs diffusion will NOT be a limiting factor in DECam.
41 Wednesday, February 22, 2012 Figure 11. Left) image produced with a optical projector described in the text. Right) Profile of the image.
Figure 12. Profile near the edge for a flat exposures on a DECam CCD. The red point shows the limit on the usable area defined according to the DES requirements described on the text. Glowing edge . The serial register is divided in two and in this image and − 27,000 e ≈
42
Wednesday, February 22, 2012 Figure 2. Flat exposure with a light level of the serial overscan forthe each image. half is The displayedtape 6 on approximately used the square for center featuresproduced alignment of on during by the along the an CCD.. the rings intermediateflat edges The of field packaging parallel of the step. correction. overscan the resistivity is CCD Also variation displayed correspond on visible on to the are the the silicon the top footprint subtrate. 0.5% of of These variations adhesive features in are the completely signal removed level after a Edge effects on DECam CCDs
Flat field shows additional light collected on the edge pixels. This light comes from the edge of the CCD, from outside the pixel grid.
Figure 11. Left) image produced with a optical projector described in the text. Right) Profile of the image. equipotential and electric field lines
flat field profile a DECam CCD
S. Holland et al 2009 (for a similar CCD) Figure 12. Profile near the edge for a flat exposures on a DECam CCD. The red point shows the limit on the usable area defined accordingregion to the DES not requirements used described for on the DES text. imaging. 43 Wednesday, February 22, 2012
Figure 2. Flat exposure with a light level of 27,000 e−. The serial register is divided in two and in this image and ≈ the serial overscan for each half is displayed on the center of the CCD.. The parallel overscan is displayed on the top of the image. The 6 approximately square features on along the edges of the CCD correspond to the footprint of adhesive tape used for alignment during an intermediate packaging step. Also visible are the 0.5% variations in the signal level produced by the rings of the resistivity variation on the silicon subtrate. These features are completely removed after a flat field correction. Edge effects studies on the sky
Figure 13. Imaging of the Tuc 47 globular cluster produced with a DECam CCD installed on a 1m telescope at CTIO.
Figure 14. Distortions measured in flat field exposures for DECam CCDs (blue curve) compared to the distortions measured by imaging a globular cluster on differentduring the 1m imaginglocations of the Tuc 47 globular of the cluster asCCD discussed in thewe text (blackmeasured points). the distortion due to this effect. Results agree with flat field studies. We understand the issue.
44 Wednesday, February 22, 2012 LH (low light level) 40000 0.1600E+05/ 20 35000 P1 -93.98
30000 P2 187.0
25000 mean(ADU)
20000
15000 Linearity10000 5000
0 0 25 50 75 100 125 150 175 200 exp time(sec) up
0.1 RH vs LH (low light level) The CCDs show good linearity 40000 673.0 / 20 0.08 35000 P1 19.39 up to ~200,000e. 0.06 low light levels x 10 2 30000 P2 0.9511 0.04 2000 25000
mean(ADU) 0.02 1750 non linearity 200000 1500 -0.0215000 mean(ADU) 1250 -0.04 10000 1000 -0.06 5000 750 -0.08 0 5,000e 15,000e 30,000e 500 -0.1 00 255000 10000 50 15000 75 20000 100 25000 125 30000 150 35000 175 40000 200 250 exp time(sec) upup 10 20 30 40 50 60 70 80 90 0.02 exp time(sec) 0.015 Figure 18. Low level nonlinearity measurement. 0.03 0.01 0.02 0.005
mean (e-) 0.01
(LH - RH)/RH 0
non-linearity 0 -0.005
-0.01 -0.01
-0.015 -0.02 -0.02 -0.03 0 5000 10000 15000 20000 25000 30000 35000 40000 0 250 500 750 1000 1250 1500 1750 2000 2250 2500 2 x 10 mean RH mean (e-)non linearity 102 left amp vs rigth amp
Figure 16. Top) Mean signal as a function of exposure time. Bottom) Fractional non-linearity on the upper panel as a toFigure eliminate 19. Non linearities inlamp the relative responsefluctuations of the two signal path of or a DECam CCD. function of mean signal. The 1% technical requirement for DECam is satisfied for signals up to 225,000 e−. shutter problems. (note the scale!) 45 Wednesday, February 22, 2012 Noise
) 45000 2 6248. / 3 Our spec is 15e- at this 40000 PTC @ 4 µsec/pixel the P1 58.23 speed. No problem! 35000 noise is RMS=8.2e-. 30000 P2 0.9272 25000 20000
variance (ADU 15000 10000 5000 0 0 5000 10000 15000 20000 25000 30000 35000 40000 mean (ADU)
Figure 20. Low level photon transfer curve showing a gain of 0.93 ADU/e− and noise of 8.5 e−. The noise for these detectors could served not only as a demonstration of the performance of the CCDs on the sky, but also as a test of the complete signalbe path reduced for DECam on down the mountain. to 2e- The testsRMS were for done using science grade 2kx2k detectors installed in an engineering~30usec/pixel package. Observing time runs (33kpix/sec). of 6 nights were done every season starting on the first semester 2008. The final runs done on March 2010 used the≈ DECam production readout system and the results demonstrated that the system can operate reliably in a telescope environment achieving the same performance seen in the lab. In addition to the system testing of the MCCDTV and the testing performed at the 1m CTIO telescope, the DECam detectors were also tested for long term reliability. An array of 4 2k x 2k CCDs was installed in a vacuum vessel and operated cold continuously for 13 months. The vacuum vessel was kept cold and dark exposures were produced every few hours. An example of a 10-hr dark exposure is shown in Fig.24. The gain, dark current level and readout noise for these detectors were monitored continuously showing stable results. After 13 months 46 Wednesday,,the detectors February were fully 22, characterized 2012 using the standard DECam procedures and the results show no changes with respect to the characterization done before the long operation. These tests were done as part of an R&D effort for the possible application of CCDs in low background experiments.26 The result of this test constitutes a reassuring demonstration of the long-term reliability of the DECam detectors.
8. CONCLUSION The DECam imager is currently being built for installation at the prime focus of the Blanco 4m Telescope at CTIO, Chile. This instrument will be used for the Dark Energy Survey that is scheduled to start operations in the second semester of 2011. The focal plane detectors for the DECam instrument have been produced and characterized and are now waiting installation on the final focal plane. To achieve this goal, the DECam project has established a new CCD packaging and testing facility at Fermilab that has operated continuously in production mode for about 2 years. The DECam CCDs meet all the technical requirement established by the DES project. In many cases the detectors exceed this requirements by a large factor. In this work we presented the main features that make the DECam CCD special because of their thickness : i) quantum efficiency, ii) lateral charge diffusion and iii) edge Pixel full well capacity
0.3
0.2
0.1
0
-0.1
-0.2
-0.3 0 250 500 750 1000 1250 1500 1750 2000 2250 2 x 10 nonlinearity (from variance) mean (e-)
There is a 10% non-linearityFigure 17. Non linearity on measured the variance in the variance at vs 180,000e- mean curve shown (from in Fig.15. the photon transfereffect curve). associated This with the determined shutter or the light our source. pixel Since capacity. both sides are DES read outrequires by independent this signal to be paths, abovethis 130,000e-. will check for any No possible problem! low level defects on one signal amplifier with respect to the other. The results of this measurement are shown in Fig. 19, and show that the two amplifiers follow each other with precision better than 0.1%. To measure the noise, a photon transfer curve at lower light level is produced, shown in Fig.20. This curve is fitted to obtain the linear relation between the variance and the mean. A good measurement of the variance for 47 2 Wednesday, Februaryan image 22, with 2012 no light is obtained, which corresponds to the readout noise for the detector. In this case σ0 = 58 2 ADU , which corresponds to a readout noise of σ0 =7.6 ADU= 8.5e−. The data presented here was collected with a pixel time of 4µsec per pixel which meets the DECam readout time and noise requirements. The noise performance of these detectors has been studied in detail as part of the characterization effort done by the DECam CCD team7.8 The signal is digitized after correlated double sampling (CDS) of the output. Each sample used for the CDS operation is the result of an integration during a time τ. This integration acts as a filter for high frequency noise. The noise measured for a DECam detector as a function of readout time is shown in Fig. 23. The noise observed for pixel readout times larger than 50 µs is σ < 2e− (RMS). At large readout times, τ is one half of the pixel readout time. The detectors have an output stage with a electronic gain of 2.5 µV/e. These results were obtained using a Monsoon15 CCD controller. ∼
7. SYSTEM TESTING The DECam project built a Multi-CCD Test Vessel (MCCDTV) for system testing of the detectors, their readout electronics and the mechanical and thermal properties of the complete system. A photograph of the MCCDTV is shown in Fig.22 and some details of the studies performed using this prototype are discussed in Ref.24 All the technical requirements for the detectors have been satisfied on the MCCDTV at the same level seen in the single CCD test station. A maximum of 28 engineering grade CCDs were installed on the MCCDTV and read out simultaneously. The DEcam detectors have not been extensively used before on astronomical instruments. Some test have been done before for imagers,2 and some similar devices have been installed on spectrographs. As part of certification for the DECam CCDs observations were performed on a 1m telescope at CTIO25 . These tests New opportunities with DECam CCDs
CCDs are readout serially (2 outputs for 8 million pixels). When readout slow, these detectors have a noise below 2e- (RMS). This means an RMS noise of 7.2 eV in ionization energy!
The devices are “massive”,1 gram per CCD. Which means you could easily build ~10 g detector. DECam would is a 70 g detector.
σ = 2e
Interesting for a low threshold DM search. • 7.2 eV noise ➪ low threshod (~0.036 keVee) • 250 μm thick ➪ reasonable mass (a few grams detector) 48
Wednesday, February 22, 2012 muons, electrons and diffusion limited hits. nuclear recoils will produce diffusion limited hits
49
Wednesday, February 22, 2012 DAMIC underground test at FNAL
CCD operated at 350’ underground (MINOS hall)
developed a low background CCD package operated inside a Cu vessel shielded with lead.
50
Wednesday, February 22, 2012 CCD inside a cold 6’’ lead bucket inside cylindrical copper copper box. (IR shield) vaccum dewar
Wednesday, February 22, 2012 Text
52 Wednesday, February 22, 2012 53
Wednesday, February 22, 2012 What comes after DECam for large astronomical surveys?
54
Wednesday, February 22, 2012 Large Synoptic Survey Telescope
55
Wednesday, February 22, 2012 •10,000 square degrees of sky to be covered every three nights on average to magnitude r ~ 24.5 (AB) •The total survey area will include 30,000 deg2 covering the wavelength range 320—1050 nm.
•8.4m (6.7m effective) primary mirror •9.6 deg2 field of view •3.2 Gigapixel camera •Detectors: more, x2 faster than DECam, 1/2 the noise of DECam
56
Wednesday, February 22, 2012 2
57
Wednesday, February 22, 2012 So far we talked about imaging with filters, and loosing most of the spectroscopic information. The big surveys will required followup with spectroscopic instruments (for example to improve redshift estimates)
the current solution is to split the light of each object using a prism...
what about a detector that could measure the energy for each photon in the UV-VIS-IR range?
58
Wednesday, February 22, 2012 Microwave Kinetic Inductance Detectors MKIDs
Each pixel is a superconductor resonator. The resonance frequency changes when you hit the pixel with light. The magnitude of the changes depends on the energy of the photon...
So far arrays of 1000 pixel have been tried for a few nights on a telescope.
Still early developments, but this could change the future of astronomical instruments!
59 Wednesday, February 22, 2012 THANKS!
60 Wednesday, February 22, 2012 System testing: prototype focal plane
We also learned that we CAN get in trouble. Three different events produced an abrupt full-well reduction of some of the detectors on the focal plane.
PTC before and after the incident.
61 Wednesday, February 22, 2012 ... the CCDs have changed!
5.5V 6.0V
7.0V
cube 2/2010 8.5V
could get it back to specs by increasing the V+ clock.
s3-12662 Wednesday, February 22, 2012 exercise in the lab
!"#$%&'( !"#$%&'$()**+,-** .-/01-.2 !"#$%" )*#+ ,$"$(#
-. /+010+- 2,#3$45)6
!"#$%&'(A$'#+B8%)858))& 78',$9:+;4(4<)+=>:+=?@? after gate set to 60V
!"#$% )*#+ ,$"$(#
-. /+010+- 2,#3$45)6=%&>-*-$?0.80*088> 30&'%24$5)6)78$9:4$9;<;
63 A$'#+B8%)858))& 78',$9:+;4(4<)+=>:+=?@? Wednesday, February 22, 2012 what is going on.
If the voltage on the gate is high enough you could move charge into the oxide, and this charge will then shield the silicon from the gate.
By using a higher V+ you recover the performance, compensate for the shielding.
This is how the old memories use to work. So now we are trying to ERASE it with UV...
64
Wednesday, February 22, 2012 Vgate holes accumulate here Vsub potential inside CCD
in normal operating conditions the voltage drop in the gate (E*d) does not change much with Vgate .
If Vgate is too large compared with Vsub. Detector goes into “inversion”, and all the extra voltage drops across the gate. The voltage at the interface gets Vgate pinned pinned. We put the detectors in inversion all the Vsub time to ERASE them. For this we set Vsub=0 and Vgate=8V.
We damaged the detector in our test by going 50V beyond inversion. oxide Unfortunately does not recover with inverted voltage.
Wednesday, February 22, 2012 this happens for all the detectors, here is a plot in Janesick for one normal CCD with opposite polarity.
Wednesday, February 22, 2012 more in the lab
Similar levels of high field in the silicon
) 90000 2 could also be achieved by accumulating too much charge in the CCD. 80000 We now believe that this is what 70000
variance (ADU starting point happened to our detectors in the imager. During our operations we were not 60000 concerned about excessive illumination and this has produced charge migration 50000 into the oxide layer.
40000 Now we are trying to understand the threshold for this effect. 30000 after supersaturation Why are our detectors specially sensitive 20000 to this issue.
10000
0 0 200 400 600 800 1000 1200 1400 2 mean (ADU)x 10
Wednesday, February 22, 2012 measurement of charge injection in s3-126 (now 9/2010)
charge injection is produced when the voltage under the Vog gate is below Vref.
Vref = Vog - Voffset
In this case -12 = 4 -16 Voffset=-16
it moved by 2V.
This effect is similar to what we see on the Vertical clocks! Similar voltage shift everywhere.
Wednesday, February 22, 2012 measurement of charge injection in s3-126 (CCD testing RH, 9/2009)
x 10 2 charge injection is 1400 produced when the voltage under the Vog gate Vref = - 12V is below Vref. 1200 -11 V
MEAN (ADU) -10 V Vref = Vog - Voffset 1000 -9 V In this case -12 = 2 -14 -8 V Voffset=-14 800 -7 V -6 V 600
400
200
0 0 2 4 6 8 10 VOG(V)
Wednesday, February 22, 2012 Observing with DECam CCDs
Detectors have not been use extensively in astronomy. We are also studying them also on the sky. These are also tests of the readout electronics developed for DECam.
(10/2008) sombrero (4/2008) 1m telescope at CTIO
last month completed a new engineering run to understand grounding and filtering at CTIO. Demonstrated that the DECam production electronics meets requirements when used on the mountain.
This is also useful for our technical staff to get familiar with CTIO (people, equipment, environment). M42 (11/2009) 70
Wednesday, February 22, 2012 operations with prototype mosaic
cold electronics (cables/ connectors) + front end crates used in prototype
+ lots of extremely produced a flat focal valuable experience plane. Tested cooling operating a mosaic like mechanical details as this system design. this. support also benefited from prototyping cycles.
71
Wednesday, February 22, 2012 DECam Imager
Prototype imager operated with real imager instrumented this ~50% of detectors instrumented summer operated for ~3 years
72
Wednesday, February 22, 2012 QE stability
Figure 7. QE efficiency variations relative to the nominal operating temperature of 173 K. 73
Wednesday, February 22, 2012