Kyoto's X-ray Astronomical SOI pixel sensor...Kyoto's X-ray Astronomical SOI pixel sensor T.G.Tsuru...
Transcript of Kyoto's X-ray Astronomical SOI pixel sensor...Kyoto's X-ray Astronomical SOI pixel sensor T.G.Tsuru...
Kyoto's X-ray Astronomical SOI pixel sensorT.G.Tsuru ([email protected])
20120314-17_SOIPIX_LBNL_v6.key
Ryu et al. IEEE NSS 2010, Conf. Record 【XRPIX1-CZ -FI】Ryu et al. IEEE TNS 58, 2528 (2011) 【XRPIX1-CZ-FI】Nakashima et al. TIPP 2011 (2011) Submitted 【XRPIX1-FZ-FI】Ryu et al. IEEE NSS 2011, Conf. Record - Trigger【Driven Readout】Nakashima et al. IEEE NSS 2011, Conf. Record 【XRPIX-ADC1】
Arai-sensei’s SOI collaboration + A-R-Tec
+ Bautz et al. (MIT group) for experiment of XRPIX1b-CZ-FI/BI
+ Tsunemi et al. (Osaka group) & Doty (Noqsi) for design of ADC
Ryu (Kyoto) Nakashima (Kyoto) Takeda (KEK)
The standard Imaging Spectrometer of modern X-ray astronomical satellites X-ray CCD
• Fano limited spectroscopy with the readout noise ~3e- (rms).
• Wide and fine imaging with the sensor size of ~20-30mm pixel size of ~30μm□
Non X-ray background of XIS (BI)
Due to high energy particles on orbit.
•Non X-ray background above 10keV is too high to study faint sources.
•The time resolution is too poor (~ sec) to make fast timing obervation of time variable sources.
S26 K. Koyama et al. [Vol. 59,
2.4. Heat Sink and Thermo-Electric Cooler (TEC)
A three-stage thermo-electric cooler (TEC) is used to coolthe CCD to the nominal operating temperature of !90!C. Thecold-end of the TEC is directly connected to the substrateof the CCD, which is mechanically supported by 3 Torlon(polyamide-imide plastic) posts attached to the heat sink. Theheat is transferred through a heat pipe to a radiator panel on thesatellite surface, and is radiated away to space. The radiatorand the heat pipe are designed to cool the base below !40!Cunder the nominal TEC operating conditions. Figure 5 is aphotograph of the inside of base with the frame-store covershield removed. Since the TEC is placed under the CCD, andthe heat pipe is running under the base plate, these are not seenin figure 5.
2.5. Radiation Shield
The performance of any CCD gradually degrades due toradiation damage in orbit. For satellites in low-Earth orbit likeSuzaku, most of the damage is due to large fluxes of chargedparticles in the South Atlantic Anomaly (SAA). The radia-tion damage increases the dark current and the charge transferinefficiency (CTI). The XIS sensor body provides radiationshielding around the CCD. We found from the ASCA SISexperiment that radiation shielding of > 10 g cm"2 equivalentAl thickness is required. The proton flux density at 2 MeV onthe CCD chip through 10gcm"2 of shielding is estimated to be" 2# 103 protonscm"2 MeV"1 d"1 in the Suzaku orbit (sameas the ASCA orbit) at solar minimum.
3. On-Board Data Processing
3.1. XIS Electronics
The XIS control and processing electronics consist ofAE/TCE (analog electronics/TEC control electronics) and DE(digital electronics). The DE is further divided into PPU (pixelprocessing unit) and MPU (main processing unit). Two sets ofAE/TCE are installed in each of two boxes, respectively, calledAE/TCE01 and AE/TCE23. Similarly, 4 PPUs are housedin pairs, and are designated PPU01 and PPU23, respectively.AE/TCE01 and PPU01 jointly take care of XIS 0 and XIS 1,while AE/TCE23 and PPU23 are for XIS 2 and XIS 3. Oneunit of MPU is connected to all the AE/TCEs and PPUs.
The AE/TCE provides the CCD clock signals, controls theCCD temperature, and processes the video signals from theCCD to create the digital data. The clock signals are generatedin the AE with a step of 1/48 pixel cycle (" 0.5µs) accordinga micro-code program, which is uploaded from the ground.The pixel rate is fixed at 24.4µs pixel"1. Therefore one line,consisting of 4 under-clocked pixels, 256 active pixels, and16 over-clocked pixels, is read out in about 6.7 ms. The CCDoutput is sampled with 16-bit precision, but only 12 bits aresent to the PPU. The 12 bits are selected to cover the full energyscale of $ 15 keV in the normal setting of the gain. The fullenergy scale of $ 60 keV can also be selected in the low gainmode.
The AE/TCE controls the TEC (thermo-electric cooler) togenerate a temperature difference of " 50!C relative to thebase, while keeping the CCD chip at !90!C. The AE/TCE
Fig. 5. The CCD and heat sink assembly installed in the base. Thecover shield is removed in this picture.
can also supply reverse current to the TEC, to warm up theCCD chip in orbit. The CCD temperature may be raised a fewtens of !C above that of the base. In practice, the requirement toavoid excessive mechanical stresses due to differential thermalexpansion of the copper heat sink relative to the alumina CCDsubstrate imposes an upper limit on CCD temperature in thismode. For example, with the heat sink at a typical operatingtemperature of !35!C, we have adopted a maximum allow-able CCD temperature of about + 15!C. The CCD temperatureupper limit is higher at higher heat sink temperatures.
The PPU extracts a charge pattern characteristic of X-rays,called an event, after applying various corrections to the digitaldata supplied by the AE/TCE. Extracted event data are sent tothe MPU. Details of the event extraction process are describedin subsection 3.3. The PPU first stores the data from itsAE/TCE in a memory called the pixel RAM. In this process,copied and dummy pixels1 are inserted in order to avoid agap in the event data at segment boundaries, to ensure properevent extraction at segment boundaries and to enable identicalprocessing of all segments of the data. The raw data fromAE/TCE may include a pulse height (PH) offset from the truezero level due to dark current, small light leakage through theOBF, and/or an electric offset. Since the offsets depend onthe CCD position and time, offset corrections are also positionand time dependent. To reduce the computing power and timerequired for such corrections, the offsets are divided into twoparts, dark-level and light-leak. The dark-level is the averageoutput from a pixel with no irradiation of X-rays or chargedparticles. The dark-level is determined for individual pixels,and is up-dated by command only after each SAA passage in1 The data of each CCD segment are transferred through independent lines
from the AE/TCE to the PPU, and are processed in parallel by the sameprocessing scheme in the PPU. For a proper event extraction at thesegment boundary, the data in the two columns of the adjacent CCDsegments must be used. Therefore hard-wired logic is installed to “copy”the two column data in the adjacent CCD segments to the proper locationsin the PPU pixel RAM. These are called as “copied pixels”. In the caseof outer boundaries of segments A and D, such “copied pixels” can not beprepared. Instead, two columns of zero data are prepared in the PPU pixelRAM. These are called “dummy pixels” in the PPU pixel RAM.
Suzaku「すざく」 XIS
“XRPIX” = SOI pixel sensor for future X-ray astronomical satellites
10μsec
Target Specification
Imaging area > 25x25mm2, pixel ~ 30-60μm□ (1” @ F=9m)Energy Band 0.3-40keV with BI, and thick depletion (>300μm) Spectroscopy ΔE < 140eV @ 6keV, Fano limit (<10e-)Timing ~ 10μsecDark Current 1pA/cm2Function Trigger signal & pixel address output, built-in ADCNon X-ray BGD 5e-5 c/s/keV/10x10mm2 at 20keV (1/100 of CCD)
Realize Very Low BGD by Anti-coincidence
non X-ray BGD
(high energy particle)
on-board processor
X-ray
scintillator
Each pixel has its own trigger and analogue readout CMOS circuit.
Fast CMOS (low ρ Si)
Insulator(SiO2)
V_back
Hole
Electron
TimeV_sig
X-ray
BPW
CMOSReadout
CMOSReadout
CMOSReadout
P+
CMOSReadout
Sensor (high ρ,
depleted Si)
Our SOIPIX
Road Map to the Goal
1-CZ
1-FZ
Thick Depletion
Mask
Waf
er
2-FZ-BI
Large Size
Spectroscopy
1b-CZ
Spectroscopy
ADC1
Integration
Goal !1-FZ-BI
Back Illumination
XRPIX1-CZ (X-Ray PIXel detector - CZochralski)
・2.
4 m
m
30.6 µm x 32 Pixel
COL Hit Pattern
1.0 mm
COL Hit Pattern
Row H
it Pattern
COL ADDRCOL Amp
ROW
AD
DR
TRIG_OOR
TRIG_COLTRIG_ROW
A_OUT
◆ CZ (0.7kΩcm), 260μm thick◆ Front Illumination
XRPIX1: Pixel Circuit
STORE
CDS C100fF
Sample C100fF
Sensor
Triggeroutput
Trigger
CDS
Comparator
CDS _VRST VTH
VDD18
PD_VRST
COL_AMP OUT_BUFSF2
SF1
VDD18
GND18
GND18
Sample/Store
Sensor C
Analogoutput
Analog Readout
G=1 G=1
TEST_ECA EOXX
XRPIX1-CZ (0.7kΩcm): Depletion Depth
+ : Experimental results– : Expected value
CZ : 150μm @ VBB=100V
• 17 keV and 8 keV X-rays have different attenuation lengths.• Measure the depletion thickness by observing the ratio
between the counting rates of the two energies X-rays. • The data follow the expectation well.
Nakashima et al., 2011, submitted
Back Bias [V]0 10 20 30 40 50
Dep
letio
n D
epth
[um
]
0
50
100
150
200
250
300
350
60 70 80 90 100 110
Back Bias (V)5 10 15 20 25 30 35 40
Dar
k C
urre
nt (
e/m
s/pi
xel)
-110
1
10
210
310
410w/o CMP
Back Bias (V)5 10 15 20 25 30 35 40
Dar
k C
urre
nt (
e/m
s/pi
xel)
-110
1
10
210
310
410w/o CMP
with CMP
XRPIX1-FZ (7kΩcm): Dark Current from Rough Backside Surface
Nakashima et al., 2011, submitted
Chemical Mechanical Polish (CMP)
Before CMP
Rough backside...
Si wafer
After CMP
Si wafer
Smooth backside !
Polish
Before CMP
Rough backside...
Si wafer
After CMP
Si wafer
Smooth backside !
Polish
Smooth backside
Before CMP
Rough backside...
Si wafer
After CMP
Si wafer
Smooth backside !
Polish
Rough backside
Depletion Layer Reaches the backside at 25V
XRPIX1-FZ (7kΩ): Depletion Depth
Nakashima et al., 2011, submitted
Back Bias [V]0 10 20 30 40 50
Dep
letio
n D
epth
[um
]
0
50
100
150
200
250
300
350
60 70 80 90 100 110Back Bias [V]
0 10 20 30 40 50
Dep
letio
n D
epth
[um
]
0
50
100
150
200
250
300
350
60 70 80 90 100 110
FZ : 250μm @ VBB=30V
• The thickness of the depletion layer of XRPIX1-FZ reaches ~250μm at 30V and stops its growth there.
• The 250μm is nearly equal to the hi-ρ Si thickness (260μm). • Full depletion is achieved at VBB=30V.
CZ : 150μm @ VBB=100V
+ : Experimental results– : Expected value
XRPIX1-FZ,with CMPPchNeXT4 (normalized to 30.6μm□)
Temperature (degree C)-50 -40 -30 -20 -10 0 10 20
Dar
k C
urre
nt (e
/ms/
pixe
l)
-310
-210
-110
1
10
210
310
410
510
610
黒:VBB=30V青:20V赤:10V
3e-2
4e-3PchNeXT4
Dar
k C
urre
nt (
e/m
sec/
30.6μm□
pixe
l)
XRPIX1-FZ-FI (7kΩcm) : Dark (Leak) Current10
From 20110516_OKImeeting_SOI_Dark_v5
Requierment at -40C
= 0.1e/msec/30.6μm□ pix
(depletion = 250μm)
= 2pA/cm2
1/100
Channel40 60 80 100 120 140
Counts
50
100
150
200
250 Red: 109CdBlue: 241Am22.2 keV
FWHM = 1.1 keV
24.9 keV
13.9 keV17.7 keV
Sensitivity = 4.0 μV/e- (including SF, amps)Sensor C = 34 fF
Good Linearity
Energy Resolution
ΔE = 1.1-1.2keV (FWHM) ≫ Fano limitReadout Noise = 100e- (rms)
X-ray Energy [keV]
0 5 10 15 20 25
Ou
tpu
t [A
DU
]
0
20
40
60
80
100
120
140
160
XRPIX1-CZ (T=-50C, VBB=100V)
XRPIX1-CZ : X-ray Spectra in the frame mode
Ryu et al., 2011
• See whether the readout noise of 100e- (rms) is explained by the sum of circuit noises or not.
• Measure the noise of individual circuit element through several DC voltage input points.
XRPIX1-CZ : Readout Noise
Sum = 94e
SF113e
CDS kTC70e
SF213e
COL_AMP31e
OUT_BUFExternal on PCB
53e(100fF ⇒ 51e)
STORE
CDS C100fF
Sample C100fF
CDS _VRST VTH
VDD18
PD_VRST
COL_AMP OUT_BUFSF2
GND18
Sensor C
G=1 G=1
TEST_ECA EOXX
Sensor
CDSSF1
Sample/Store
A
B C DVB
Vout
VC VD
The sum of these noise are almost consistent with the observed readout noise of 100e.
Ryu et al., 2011
Show results from block 2 today
144D_CDS
(400fF)
244M_BPW
(45%)
344D_BPW
(45%)
444D_ORG(BPW100%,CDS=100fF)
XRPIX1 → XRPIX1b
•Block 2, 3•Reduce the area of BPW to
45% to obtain higher gain.• BPW (buried p-well)
suppressing back gate effect dominates the capacitance at the sensor node.
• Block 1• Increase C_CDS from 100fF to
400fF to reduce the reset noise generated at the CDS circuit.
Purpose Improvement of Spectroscopic performance
X-ray Energy [keV]
0 5 10 15 20 25
Ou
tpu
t [A
DU
]
0
20
40
60
80
100
120
140
160
XRPIX1-CZ (T=-50C, VBB=100V)
XRPIX1b-CZ(T=-50C, VBB=100V)
h7Entries 1196Mean 101.9RMS 48.88
/ ndf 2χ 14.99 / 6Constant 3.91± 43.84 Mean 0.2± 152.3 Sigma 0.211± 2.369
[ADU]0 50 100 150 200 250
Cts
0
10
20
30
40
50
60h7
Entries 1196Mean 101.9RMS 48.88
/ ndf 2χ 14.99 / 6Constant 3.91± 43.84 Mean 0.2± 152.3 Sigma 0.211± 2.369
Single Event Histgram (R2)
24.9 keV
Bule: 109Cd
22.2 keVFWHM=810 eV
XRPIX1-CZ→1b-CZ : X-ray Spectra in frame mode
Sensitivity = 4.0 →6.1μV/e-Sensor C = 34 → 23fF
Gain Increased
Better Energy Resolution
ΔE = 1.1-1.2keV → 0.8-1.1 keV(FWHM)Readout Noise = 100e- → 74e- (rms)
x1.5
Second dominating source of C of sensor node.
XRPIX1b-CZ : Single Pixel Readout• In order to study the limit of the spectroscopic performance.• Observe the waveform of analogue output from a single pixel by fixing the
readout address without clocking (Single Pixel Readout like a SSD). • Detect an X-ray as a “step” and measure the pulse height. → X-ray spectrum. • No reset during the measurement → Free from the reset noise• Reduce noises other than the reset noise by introducing LPF.
high_v(100 samples average) - low_v(100 samples average) → LPF with τ=100μs
sample number in one frame0 200 400 600 800 1000
PH [A
DU
]
625
630
635
640
645
650
655
660
sample number in one frame0 200 400 600 800 1000
PH [A
DU
]
627
628
629
630
631
632
633
PH [ADU]0 50 100 150 200
Cts
/ A
DU
1
10
210
310
410
CMS Histgram h1Entries 49668Mean 0.2479RMS 1.017
CMS Histgram
Data Name: Fe_minus30c_sp_Vback_100V_Tscan_480ns_25_25_50000f.root
Cut+Ped+Phy[frm]: 300+500+50000
Hit threhold= 10 ADU, Delay=2 frm, Ave=160 frm
X-ray count: 0.001972 Cts/frame
Readout Noise = 0.330654 ADU
X-ray Hit
PulseHeight
average100 samplesτ=100μsec
h1Entries 398490Mean 0.3396RMS 3.101
/ ndf 2χ 3.97e+05 / 9Constant 1.41± 85.81 Mean 1.41± 54.28 Sigma 11.6984± 0.8022
PH [ADU]0 20 40 60 80 100 120 140
Cts
/ A
DU
0
20
40
60
80
100
120
h1Entries 398490Mean 0.3396RMS 3.101
/ ndf 2χ 3.97e+05 / 9Constant 1.41± 85.81 Mean 1.41± 54.28 Sigma 11.6984± 0.8022
CMS Histgra (R10C10)
FWHM=278eV @8.0 keV
rms=14.6e- 44M-BPW
Cu-Kβ
Cu-Kα
ΔE = 278eV @ 8.0keV (FWHM), readout noise = 14.6e (rms)
average 100 samplesτ=100μsec
Figure 8. Simulated spectra of the supernova remnant E0102-72.3. Top: XIS, for a single BI (red) and single FI (green)sensor. The flight XIS will contain two of each sensor type. Middle: Chandra ACIS-S. Bottom: XMM-Newton EPIC-PN.Note the factor of two larger vertical scale for the EPIC-PN spectrum.
122 Proc. of SPIE Vol. 5501
Black: Chandra BI-CCD
Figure 8. Simulated spectra of the supernova remnant E0102-72.3. Top: XIS, for a single BI (red) and single FI (green)sensor. The flight XIS will contain two of each sensor type. Middle: Chandra ACIS-S. Bottom: XMM-Newton EPIC-PN.Note the factor of two larger vertical scale for the EPIC-PN spectrum.
122 Proc. of SPIE Vol. 5501
Red: Suzaku BI-CCDChemisorption improved CCE.
Green: Suzaku FI-CCD
Back Illumination is Essential but not Easy Task
• Quantum Efficiency• Thin dead layer is essential• Current 1μm → Goal 10nm
(Req. < 100nm including optical block)• Charge Sharing Effect
• Due to diffusion of charge traveling from the back-side to the front-side.
• Reading multiple pixels to collect charge sums their readout noises up.
• Charge Collection Efficiency (CCE)• A part of the charge is lost near the
surface of the back side. It makes a tail-structure in the X-ray spectrum.
• The front side has a circuit-layer of 10μm thickness. (Note: FI-CCD~1μm, BI-CCD~10nm)
• Difficult to detect low-E X-rays with FI. • Difficult to reduce the thickness of the
circuit layer.
Filter Transmission
Filter Transmission http://henke.lbl.gov/tmp/xray9837.html
1 / 1 12/03/14 14:41
Si 1μm
Si 10μm
00.
51.
0Q
E
0.3 1.0 5.0keV
→ BI is essential
XRPIX1b-CZ-FI/BI (100μm): Spectra in Single Pixel Readout (2011.11.22)
Target= Al2O3 , Front-I, Back-I, V_tube=6 kV,V_bais = 100V, Temp=–50ºC, Hit Threshold= 2 ADU, Exposure=400 sec, PIX=R10C10, 2us_sample, 300 µs_ave
h4Entries 128627Mean 1.608RMS 3.417
PH [ADU]-5 0 5 10 15 20 25 30 35 40
Cts
1
10
210
310
410 h4Entries 128627Mean 1.608RMS 3.417
CMS Histgram (offset-corrected)
h4Entries 128627Mean 1.608RMS 3.417
AL-Ka(1.48 keV)
Si Edge (1.84 keV)
RED: BI BLUE: FIh4Entries 128627Mean 2.282RMS 3.155
PH [ADU]0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
Cts
0
100
200
300
400
500
600
700
800
900
1000
1100
1200
h4Entries 128627Mean 2.282RMS 3.155CMS Histgram (offset-corrected)
h4Entries 128627Mean 2.282RMS 3.155
Zoom
O-Ka ??
HV(6kV)
←15 e-rms
←18 e-rms
Cts
DetectionThreshold(2ADU)
502 eVby
Calibration
AL-Ka
VeryPreliminary
• X-ray generator (target = Al, 6kV).
• Al-K + Bremss(+O-K from Al2O3?)
• First detection of X-rays below Si-Edge with SOIPIX (?)
• ΔE (FI) = 157eV, ΔE (BI) = 351eV (if line)
4 DRAFT FOR IEEE TRANSACTIONS ON NUCLEAR SCIENCE (V.2012/02)
B. Crosstalk Measurement
In order to measure the circuit crosstalk between adjacentpixels, we use the Fe K↵ line at 6.4 keV as an input signalwith a constant PH. We acquire the data from 3⇥3 pixels(section IV). The sampling frequency for each data point in thewaveform is 110 kHz. The readout noise is 40 electrons rms.
Fig.6 shows the center-pixel X-ray spectra obtained from aniron primary target; the 6.4 keV line at PH = 43.8±0.02 ADUis well resolved. From the iron spectra of 3⇥3 pixels, thegain non-uniformity among pixels is measured to be 0.08 %(standard deviation); this is good compared to other APSs(e.g.,[4] ).
The filled-red-box region in Fig.6 indicates the half high-energy part of the 6.4 keV gaussian at 3 sigma level, events inwhich should have the holes produced by one 6.4 keV photonfully collected in one pixel, i.e., the center pixel (see [2] formore explanation). We extracted the single-pixel 6.4 keV X-ray events and plot averaged waveforms for the 3⇥3 pixels.As shown in Fig.7, no substantial PH variance in the adjacentpixels is displayed at the moment when the center pixel detectsa 6.4 keV photon, which suggests small crosstalk amongpixels.h_raw_4
Entries 379230
Mean 1.307
RMS 6.535
/ ndf 2! 333.3 / 47
Constant 5.576e+01± 2.685e+04
Mean 0.0017854± 0.0009666
Sigma 0.001± 1.073
PH (ADU)0 10 20 30 40 50 60 70
Coun
ts
1
10
210
310
410
h_raw_4Entries 379230
Mean 1.307
RMS 6.535
/ ndf 2! 333.3 / 47
Constant 5.576e+01± 2.685e+04
Mean 0.0017854± 0.0009666
Sigma 0.001± 1.073
CMS RAW4 Histgramh_1
Entries 13745
Mean 29.04
RMS 11.88
PH (ADU)0 10 20 30 40 50
Coun
ts
0
20
40
60
80
100
120
140
160
h_1Entries 13745
Mean 29.04
RMS 11.88
CMS Split1 Histgramh_2
Entries 2788
Mean 34.35
RMS 13.48
PH (ADU)0 10 20 30 40 50 60 70
Coun
ts
0
5
10
15
20
25
h_2Entries 2788
Mean 34.35
RMS 13.48
CMS Split2 Histgramh_3
Entries 39
Mean 39.01
RMS 14.36
PH (ADU)0 10 20 30 40 50 60 70
Coun
ts
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
h_3Entries 39
Mean 39.01
RMS 14.36
CMS Split3 Histgram
h_4Entries 4
Mean 43.4
RMS 4.836
PH (ADU)0 10 20 30 40 50 60 70
Coun
ts
0
0.2
0.4
0.6
0.8
1
h_4Entries 4
Mean 43.4
RMS 4.836
CMS Split4 Histgramh_5
Entries 0
Mean 0
RMS 0
PH (ADU)0 10 20 30 40 50 60 70
Coun
ts
0
0.2
0.4
0.6
0.8
1
h_5Entries 0
Mean 0
RMS 0
CMS Split5 Histgramh_sum
Entries 16576
Mean 30.1
RMS 12.48
Channel (ADU)0 10 20 30 40 50 60 70
Coun
ts
0
20
40
60
80
100
120
140
160
h_sumEntries 16576
Mean 30.1
RMS 12.48
CMS Sum Histgram Readout Noise = 1.0726 ADUCnt Rate = 0.095914 Cnt/Pix/F (38337/399701)Evt Rate(PH_WIN) = 43.7410 % (16769/38337)Single Event = 81.97 %Double Event = 16.63 %Triple Event = 0.23 %Quad Event = 0.02 %Quint Event = 0.00 %Full Event = 0.01 %Others Event = 1.14 %Event th= 5.0 ADU, Split th = 3.218 ADU
Average Num= 30 samples
Double_Ratio@Pix[1,3,5,7] = 0.15, 0.25, 0.15, 0.18
Fe-Ka: 6.4 keV !Single Pixel Event
h_raw_4Entries 379230
Mean 1.307
RMS 6.535
/ ndf 2 ! 333.3 / 47
Constant 5.576e+01 ± 2.685e+04
Mean 0.0017854 ± 0.0009666
Sigma 0.001 ± 1.073
PH (ADU)010203040506070
Counts
1
10
2 10
3 10
4 10
h_raw_4Entries 379230
Mean 1.307
RMS 6.535
/ ndf 2 ! 333.3 / 47
Constant 5.576e+01 ± 2.685e+04
Mean 0.0017854 ± 0.0009666
Sigma 0.001 ± 1.073
CMS RAW4 Histgramh_1
Entries 13745
Mean 29.04
RMS 11.88
PH (ADU)01020304050
Counts
0
20
40
60
80
100
120
140
160
h_1Entries 13745
Mean 29.04
RMS 11.88
CMS Split1 Histgramh_2
Entries 2788
Mean 34.35
RMS 13.48
PH (ADU)010203040506070
Counts
0
5
10
15
20
25
h_2Entries 2788
Mean 34.35
RMS 13.48
CMS Split2 Histgramh_3
Entries 39
Mean 39.01
RMS 14.36
PH (ADU)010203040506070
Counts
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
h_3Entries 39
Mean 39.01
RMS 14.36
CMS Split3 Histgram
h_4Entries 4
Mean 43.4
RMS 4.836
PH (ADU)010203040506070
Counts
0
0.2
0.4
0.6
0.8
1
h_4Entries 4
Mean 43.4
RMS 4.836
CMS Split4 Histgramh_5
Entries 0
Mean 0
RMS 0
PH (ADU)010203040506070
Counts
0
0.2
0.4
0.6
0.8
1
h_5Entries 0
Mean 0
RMS 0
CMS Split5 Histgramh_sumEntries 16576
Mean 30.1
RMS 12.48
Channel (ADU)010203040506070
Counts
0
20
40
60
80
100
120
140
160
h_sumEntries 16576
Mean 30.1
RMS 12.48
CMS Sum HistgramReadout Noise = 1.0726 ADUCnt Rate = 0.095914 Cnt/Pix/F (38337/399701)Evt Rate(PH_WIN) = 43.7410 % (16769/38337)Single Event = 81.97 %Double Event = 16.63 %Triple Event = 0.23 %Quad Event = 0.02 %Quint Event = 0.00 %Full Event = 0.01 %Others Event = 1.14 %Event th= 5.0 ADU, Split th = 3.218 ADU
Average Num= 30 samples
Double_Ratio@Pix[1,3,5,7] = 0.15, 0.25, 0.15, 0.18
Fe-Ka: 6.4 keV !Single Pixel Event
PH (ADU)
Cou
nts
Fig. 6. X-ray spectra of the Fe primary target with the chip illuminated infront-side. The red region indicates the single-pixel 6.4 keV photon events(see text).
For more quantitative estimation, we compute the PH vari-ance at the time of X-ray jump in the center pixel for theadjacent 8 pixels using the same way calculating the X-rayPH (section IV). The results in shown in Fig.8; the center ofthe PH histogram around zero is equivalent to the crosstalk.The crosstalk in adjacent pixels is 0.2–0.5%, or 0.1–0.25 ADU,or 4–10 electrons in the negative direction with respect to thesignal at the center pixel. These are displayed as the small dip-like structures in Fig.7. The device shows very low crosstalkvalue in comparison with the other APSs (e.g., [2]).
C. Property of Charge Split Event
Charge split event occurs if the holes produced by an X-rayphoton are collected by more than one pixel. In this case, X-ray signals are observed from multiple pixels. The center pixel
Time (! 9 "s)
PH (A
DU
)
Top
Center Pixel
Bottom
Left Right
Fig. 7. Averaged waveform of 3⇥3 pixels for the single-pixel 6.4 keV X-rayevents. The vertical red lines indicated the time when X-ray is detected in thecenter pixel. The error bars of the PH are at one sigma level.
h_raw_win_4
Entries 1286
Mean 43.82
RMS 0.5521
PH (ADU)0 10 20 30 40 50
Cou
nts
0
20
40
60
80
100
120
140
160
180
200
h_raw_win_4
Entries 1286
Mean 43.82
RMS 0.5521
h_raw_win_4
Entries 1286
Mean 43.82
RMS 0.5521
PIX4 Histgram with PH window
-0.25 -0.1(top)
-0.17
--0.07(left)
43-45(center)
-0.05(right)
-0.24 -0.08(bottom)
-0.13
PH (ADU)
200
180
Num
ber o
f eve
nts
PH Offset (ADU)
Fig. 8. PH histogram of 3⇥3 pixels for the single-pixel 6.4 keV X-rayevents. Results of different pixels are shown in different colors.
confining the largest amount of charges has the highest PHamong the adjacent pixels. In general, charge splits depend onthe pixel size, the charge-cloud size when reaching the sensenode, and the location where the photon is absorbed (pixelboundary or center).
In Fig.9, we draw a correlation diagram between the PH ofthe center pixel and the adjacent pixels using the data obtainedfrom the Al primary target. In the charge-split events, adjacentpixels should have positive but smaller PH than the centerpixel. Whereas the PH of adjacent pixels should be ideallyzero in the single-pixel events. We show the expected regionfor the charge-split events and the single-pixel events with thefilled blue and red solid boxes, respectively.
The top diagram in Fig.9 demonstrates that charge splitslarger than the 3 sigma noise level (120 electrons), is absentin FI. On the other hand, in BI (Fig.9 bottom), charge splitsexist in the left, right, top, and bottom pixels, rather than inthe corner pixels. This is not strange considering the relativedistance between sense-nodes among pixels. For example, thesense node of the left pixel is located nearer to the center than
XRPIX1b-CZ-FI/BI : Interpixel Cross-Talk
Preliminary• We select events consisting of the upper half in the Fe-Kα line at 6.4keV to extract single pixel events (no charge sharing).
• See the PH of 8 pixels adjacent to the X-ray hit pixel.
• Interpixel cross-talk is not so high.
4 DRAFT FOR IEEE TRANSACTIONS ON NUCLEAR SCIENCE (V.2012/02)
B. Crosstalk Measurement
In order to measure the circuit crosstalk between adjacentpixels, we use the Fe K↵ line at 6.4 keV as an input signalwith a constant PH. We acquire the data from 3⇥3 pixels(section IV). The sampling frequency for each data point in thewaveform is 110 kHz. The readout noise is 40 electrons rms.
Fig.6 shows the center-pixel X-ray spectra obtained from aniron primary target; the 6.4 keV line at PH = 43.8±0.02 ADUis well resolved. From the iron spectra of 3⇥3 pixels, thegain non-uniformity among pixels is measured to be 0.08 %(standard deviation); this is good compared to other APSs(e.g.,[4] ).
The filled-red-box region in Fig.6 indicates the half high-energy part of the 6.4 keV gaussian at 3 sigma level, events inwhich should have the holes produced by one 6.4 keV photonfully collected in one pixel, i.e., the center pixel (see [2] formore explanation). We extracted the single-pixel 6.4 keV X-ray events and plot averaged waveforms for the 3⇥3 pixels.As shown in Fig.7, no substantial PH variance in the adjacentpixels is displayed at the moment when the center pixel detectsa 6.4 keV photon, which suggests small crosstalk amongpixels.h_raw_4
Entries 379230
Mean 1.307
RMS 6.535
/ ndf 2! 333.3 / 47
Constant 5.576e+01± 2.685e+04
Mean 0.0017854± 0.0009666
Sigma 0.001± 1.073
PH (ADU)0 10 20 30 40 50 60 70
Coun
ts
1
10
210
310
410
h_raw_4Entries 379230
Mean 1.307
RMS 6.535
/ ndf 2! 333.3 / 47
Constant 5.576e+01± 2.685e+04
Mean 0.0017854± 0.0009666
Sigma 0.001± 1.073
CMS RAW4 Histgramh_1
Entries 13745
Mean 29.04
RMS 11.88
PH (ADU)0 10 20 30 40 50
Coun
ts
0
20
40
60
80
100
120
140
160
h_1Entries 13745
Mean 29.04
RMS 11.88
CMS Split1 Histgramh_2
Entries 2788
Mean 34.35
RMS 13.48
PH (ADU)0 10 20 30 40 50 60 70
Coun
ts
0
5
10
15
20
25
h_2Entries 2788
Mean 34.35
RMS 13.48
CMS Split2 Histgramh_3
Entries 39
Mean 39.01
RMS 14.36
PH (ADU)0 10 20 30 40 50 60 70
Coun
ts
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
h_3Entries 39
Mean 39.01
RMS 14.36
CMS Split3 Histgram
h_4Entries 4
Mean 43.4
RMS 4.836
PH (ADU)0 10 20 30 40 50 60 70
Coun
ts
0
0.2
0.4
0.6
0.8
1
h_4Entries 4
Mean 43.4
RMS 4.836
CMS Split4 Histgramh_5
Entries 0
Mean 0
RMS 0
PH (ADU)0 10 20 30 40 50 60 70
Coun
ts
0
0.2
0.4
0.6
0.8
1
h_5Entries 0
Mean 0
RMS 0
CMS Split5 Histgramh_sum
Entries 16576
Mean 30.1
RMS 12.48
Channel (ADU)0 10 20 30 40 50 60 70
Coun
ts
0
20
40
60
80
100
120
140
160
h_sumEntries 16576
Mean 30.1
RMS 12.48
CMS Sum Histgram Readout Noise = 1.0726 ADUCnt Rate = 0.095914 Cnt/Pix/F (38337/399701)Evt Rate(PH_WIN) = 43.7410 % (16769/38337)Single Event = 81.97 %Double Event = 16.63 %Triple Event = 0.23 %Quad Event = 0.02 %Quint Event = 0.00 %Full Event = 0.01 %Others Event = 1.14 %Event th= 5.0 ADU, Split th = 3.218 ADU
Average Num= 30 samples
Double_Ratio@Pix[1,3,5,7] = 0.15, 0.25, 0.15, 0.18
Fe-Ka: 6.4 keV !Single Pixel Event
h_raw_4Entries 379230
Mean 1.307
RMS 6.535
/ ndf 2 ! 333.3 / 47
Constant 5.576e+01 ± 2.685e+04
Mean 0.0017854 ± 0.0009666
Sigma 0.001 ± 1.073
PH (ADU)010203040506070
Counts
1
10
2 10
3 10
4 10
h_raw_4Entries 379230
Mean 1.307
RMS 6.535
/ ndf 2 ! 333.3 / 47
Constant 5.576e+01 ± 2.685e+04
Mean 0.0017854 ± 0.0009666
Sigma 0.001 ± 1.073
CMS RAW4 Histgramh_1
Entries 13745
Mean 29.04
RMS 11.88
PH (ADU)01020304050
Counts
0
20
40
60
80
100
120
140
160
h_1Entries 13745
Mean 29.04
RMS 11.88
CMS Split1 Histgramh_2
Entries 2788
Mean 34.35
RMS 13.48
PH (ADU)010203040506070
Counts
0
5
10
15
20
25
h_2Entries 2788
Mean 34.35
RMS 13.48
CMS Split2 Histgramh_3
Entries 39
Mean 39.01
RMS 14.36
PH (ADU)010203040506070
Counts
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
h_3Entries 39
Mean 39.01
RMS 14.36
CMS Split3 Histgram
h_4Entries 4
Mean 43.4
RMS 4.836
PH (ADU)010203040506070
Counts
0
0.2
0.4
0.6
0.8
1
h_4Entries 4
Mean 43.4
RMS 4.836
CMS Split4 Histgramh_5
Entries 0
Mean 0
RMS 0
PH (ADU)010203040506070
Counts
0
0.2
0.4
0.6
0.8
1
h_5Entries 0
Mean 0
RMS 0
CMS Split5 Histgramh_sumEntries 16576
Mean 30.1
RMS 12.48
Channel (ADU)010203040506070
Counts
0
20
40
60
80
100
120
140
160
h_sumEntries 16576
Mean 30.1
RMS 12.48
CMS Sum HistgramReadout Noise = 1.0726 ADUCnt Rate = 0.095914 Cnt/Pix/F (38337/399701)Evt Rate(PH_WIN) = 43.7410 % (16769/38337)Single Event = 81.97 %Double Event = 16.63 %Triple Event = 0.23 %Quad Event = 0.02 %Quint Event = 0.00 %Full Event = 0.01 %Others Event = 1.14 %Event th= 5.0 ADU, Split th = 3.218 ADU
Average Num= 30 samples
Double_Ratio@Pix[1,3,5,7] = 0.15, 0.25, 0.15, 0.18
Fe-Ka: 6.4 keV !Single Pixel Event
PH (ADU)
Cou
nts
Fig. 6. X-ray spectra of the Fe primary target with the chip illuminated infront-side. The red region indicates the single-pixel 6.4 keV photon events(see text).
For more quantitative estimation, we compute the PH vari-ance at the time of X-ray jump in the center pixel for theadjacent 8 pixels using the same way calculating the X-rayPH (section IV). The results in shown in Fig.8; the center ofthe PH histogram around zero is equivalent to the crosstalk.The crosstalk in adjacent pixels is 0.2–0.5%, or 0.1–0.25 ADU,or 4–10 electrons in the negative direction with respect to thesignal at the center pixel. These are displayed as the small dip-like structures in Fig.7. The device shows very low crosstalkvalue in comparison with the other APSs (e.g., [2]).
C. Property of Charge Split Event
Charge split event occurs if the holes produced by an X-rayphoton are collected by more than one pixel. In this case, X-ray signals are observed from multiple pixels. The center pixel
Time (! 9 "s)
PH (A
DU
)
Top
Center Pixel
Bottom
Left Right
Fig. 7. Averaged waveform of 3⇥3 pixels for the single-pixel 6.4 keV X-rayevents. The vertical red lines indicated the time when X-ray is detected in thecenter pixel. The error bars of the PH are at one sigma level.
h_raw_win_4
Entries 1286
Mean 43.82
RMS 0.5521
PH (ADU)0 10 20 30 40 50
Cou
nts
0
20
40
60
80
100
120
140
160
180
200
h_raw_win_4
Entries 1286
Mean 43.82
RMS 0.5521
h_raw_win_4
Entries 1286
Mean 43.82
RMS 0.5521
PIX4 Histgram with PH window
-0.25 -0.1(top)
-0.17
--0.07(left)
43-45(center)
-0.05(right)
-0.24 -0.08(bottom)
-0.13
PH (ADU)
200
180
Num
ber o
f eve
nts
PH Offset (ADU)
Fig. 8. PH histogram of 3⇥3 pixels for the single-pixel 6.4 keV X-rayevents. Results of different pixels are shown in different colors.
confining the largest amount of charges has the highest PHamong the adjacent pixels. In general, charge splits depend onthe pixel size, the charge-cloud size when reaching the sensenode, and the location where the photon is absorbed (pixelboundary or center).
In Fig.9, we draw a correlation diagram between the PH ofthe center pixel and the adjacent pixels using the data obtainedfrom the Al primary target. In the charge-split events, adjacentpixels should have positive but smaller PH than the centerpixel. Whereas the PH of adjacent pixels should be ideallyzero in the single-pixel events. We show the expected regionfor the charge-split events and the single-pixel events with thefilled blue and red solid boxes, respectively.
The top diagram in Fig.9 demonstrates that charge splitslarger than the 3 sigma noise level (120 electrons), is absentin FI. On the other hand, in BI (Fig.9 bottom), charge splitsexist in the left, right, top, and bottom pixels, rather than inthe corner pixels. This is not strange considering the relativedistance between sense-nodes among pixels. For example, thesense node of the left pixel is located nearer to the center than
hit pixel
8 adjacent pixels
-0.6% -0.2%(upper) -0.4%
-0.2%(left)
1(hit pix)
-0.1%(right)
-0.5% -0.2%(bottom) -0.3%
PH ratio to the hit pixel
bremsstrahlung X-rays
Ryu+ in prep.
XRPIX1b-CZ : Trigger Driven Readout
・
COL Hit Add. Resister
Row H
it Add.
Resister
COL Readout ADDRCOL Amp
ROW
Rea
dout
AD
DR
TRIG_OOR
TRIG_COL
TRIG_ROW
A_OUT
①① ②
④
④
X-ray !
FPGA
ADC⑤
③
XRPIX1b-CZ : Trigger-driven Readout of X-ray Events
h1Entries 1617Mean 58.61RMS 19.57
X-ray PH (ADU)0 20 40 60 80 100 120 140 160
Cou
nts /
AD
U
10
20
30
40
50
60
70h1
Entries 1617Mean 58.61RMS 19.57
h1Entries 487Mean 39.51RMS 10.58
FWHM1.4 keV
8 keV
17.7 keV
13.9 keV
26.4 keV
Data VTH Exposure
Cu+Mo 10 mV 5sec
Am-241 15 mV 20 sec
17.4 keV
X-ray Trigger-driven Spectra
h1Entries 0Mean 0RMS 0
Energy (keV)0 5 10 15 20 25 30
PH (A
DU
)0
10
20
30
40
50
60
70
80
90
100h1
Entries 0Mean 0RMS 0
/ ndf 2! 16.61 / 4
p0 2.477± 15.78
p1 0.1278± 2.715
/ ndf 2! 16.61 / 4
p0 2.477± 15.78
p1 0.1278± 2.715
X-ray Energy Calibration
Output Gain : = 2.7 ADU/keV = 2.4 !V/e-Offset :=15.8 ADU =3.8 mV
Measured Output Gain : = 2.7 ADU/keV = 2.4 !V/e-Offset :=15.8 ADU =3.8 mV
Expected
- Trigger-driven mode basically operates as designed.
- The gain is different from that of the frame mode.- We are now investigating.
Frame mode
Trigger Driven
XRPIX2-CZ : New Device•Large Size, Large Format
• 60μm□ : Single pixel can cover the whole charge cloud to reduce charge sharing effect.
• Some pixels have multi-vias to collect charges efficiently.
•Make the capacitance at the readout node smaller (Area of BPW = 1/4 of XRPIX1).
•Make further increase of C_CDS to reduce the reset noise.
• Submitted Oct. 2011,
• Packaged Chip March. 2012
• I-V seems normal.
1.0mmXRPIX1, 1b
4.0mm
XRPIX2
Summary
•Read Out Noise 100e (frame mode)→ 3e
•BI dead layer 0.3μm (Si) + 0.2μm (Al) → <0.1μm (Req.) 10nm (goal))
•Depletion Layer = 250μm→ 300μm
•Dark Current 100pA/cm2→ 1pA/cm2
•Trigger Driven Readout : Functional but Problem
Thank you.
��
SOI Pixel Project : General View�
Feb. 28, 2011
SOI International Review Meeting
Yasuo Arai, KEK
http://rd.kek.jp/project/soi/�
��
È�;ŔÀ�þýĀÈ;þ¥o�kÿEòúň)`ķĚłĎĵŅġÿėġĤĹ©¨ʼn
T óČĐĬłě&²ńŋŌ,U%ńŌŋ,U%ń¿ÁĐŃĴńĐĬłěIJđŀĤńÔÚÛÞĒĺŅĠģŃĝńģŃĝŅĒŃĤIJĔŅġńàÔÛÌÉÚ×áÌÉÜÙÙńĒŃĥĚĤ�$&²
i�?JńĐĬłě×ÜĀ©¨Ŕ»�ń×ÜV´ÙÞ×Ā©¨Ŕ»�ńĐĬłěs2M£
ńÙÞ×»�ĜŃĝŀĩđŃě
ĐĬńĪĠq´ÙÞ×û«ÄüþČÌÉÞâ)bĈÀt϶ñ÷ĚłġīŅĚ¿ÁÊáßËÉϧcňª�ňN'óČM£ĎT ʼn
)bÎÀt¿ÁĀx¡M£ń¿Á§cĻĪŀ�M£ń¿ÁğĸļŁŅğĽŃM£ń¿ÁĩġīĦħĴ©¨M£ńĖŃĦħĴ¿Ás2M£ń¿Á�FĀÈ婨M£
i�?Jń§cĨŅŀ¯+ňĝĶŅīń¿Á§cĻĪŀ�Lń¿ÁğĸļŁŅğĽŃń)b¿Áª�/TEG©¨ńĖŃĦħĴ¿Ás2
ń¿Á4�ĜŃĝŀĩđŃě
ĐĬłěńŐō&²ûā-�ĀwFĎ�Kñ÷©¨ê¥oðčãÔÓÕ¡�©¨āu{ʼn©¨Ā�ÇüįēİēêDÃʼnĖŃĞĠĽij�LÊÛØőËÿĉċ3³�ĐĬłě©¨�Ď�Lʼn
"�Ā�Çü©¨�lüóČ&²üF�ÿEòúňńėĿĘļľĹĎėġĤķĒĢʼnń)�éĊ3³�Ąû�Lʼnń&²Ā©¨°}ćzBʼn
ÛØőĂĀ5�Ā÷Ć)�3³<ņ.0û3[ŇćT ʼnńÔÓÕĎ~å÷ÓÕ,U%©¨ńÛÜÓÌÓÕÔĀ�Â�Ďs2ńėĤłěĎ{§óČ�Ď�B
�Ā@�ŊĐĬłě&²üÝÖ&²ÿ½óČŒ©¨Ŕª�M£Ā»�œüŒ�a�LœĎ�µüñú3[ñň\`Ā�}iü���0M£Ā�7ÿ®yóČîüʼnÍ
ĐĬłě&²Ā©¨ĐĬłě&²Ā©¨ )bńÀt¿Á§c)bńÀt¿Á§c ĐĬłě©¨�ĀŏŎőĐĬłě©¨�ĀŏŎő
È;G*�Ā)�üþČĕŁĚīłĭĚġûāĴłģħĝńĺĻĿńĐĬłěńu�ńģŃĝĎ¾��ñň
ÈF�ü�ĕĮŀęŅĎ��ðõČD¥êäċĄóʼnğġĩĹĀ6(�ňÀ�Ā÷ĆÿňîčĊĎ
ÐĦħĴ�óČîüê_�ûóêňĐĬłě&²Ā©¨āĪĠĤŀ&²ĀkþÔÓÕ�ê¸ďûèĊôň�ÇĎRù÷
M£�ĀXć�±ñúåĄóʼn
ň�¼Ā�=ü1�Ā÷Ćÿň�OĈ¢��þýĀ|�G*Ď9]�¦óČĻĭĤň pĈ���#Ďh�ń�
CóČğġĩĹþýĀ»�ê¸ďûåĄóʼnģŃĝüĀĒŃĤIJĔŅġüñúňĐŃĴĈÓÕ,U%þýĀĐĬłě&
²êDÃûäċňöĀÈI;�ňÀ��ê¹¥üþùúåĄóʼn
g>�ĕĒĐŅŀĩħĚ
ıĠĮġĀ�`fıĠĮġĀ�`f
URL: http://www.a-r-tec.jpEmail: [email protected]
îĀĉæþ¥¬ÿEçČăìňÑÏÏÐ:Ò^ÿ.0Ā��LdĎćüÿñúň��ĐĬłě&²©¨ń»�Ďi�üóČ
ņgŇĕĒĐŅŀĩħĚĎ©�ñĄñ÷ʼnÐÏ:¼ÿrċňÈ;þÓÕ,U%þýĀĐĬłě&²©¨Āi�ňĚłġīŅ
Ě¿Á§cü¿ÁN'©¨Āi�ň3³�þĐĬłě©¨�Ā�LþýĎ�i�ñúã�5�}iÿ®yñú�
ċĄñ÷ʼn ň.0Ĉ��n½ÿèíČv{0ň��0ň/Y0Ā��~ĀZn�ňÈF�ĦħĴĀ»�ćP.ñň
���0M£ÿć®yûëČĉæÿþċĄñ÷ʼn
�Aüćň.0ü·WñúZM£»�üöĀ3~�ÿ�j�ÿ!ċ�ąň.0ü}iüĀeímĀ@�ćd÷ñú
�ċĄóĀûňĕĒĐŅŀĩħĚĎïHÆå÷øëĄóĉæèÅåå÷ñĄóʼn
ÑÏÐÐ:Ð^ �¤!�@�º 8 � �
A-R-TecAnalog and RF Technologies
ïSQ