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


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


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


/ 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


/ 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


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)


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)


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


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


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


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


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


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


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


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


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


Double_Ratio@Pix[1,3,5,7] = 0.15, 0.25, 0.15, 0.18


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


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


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


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


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


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


Double_Ratio@Pix[1,3,5,7] = 0.15, 0.25, 0.15, 0.18



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


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


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


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


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


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


Double_Ratio@Pix[1,3,5,7] = 0.15, 0.25, 0.15, 0.18


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


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


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

[email protected]

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)bĈÀtĎ¶ñ÷ĚłġīŅĚ¿ÁÊáßËÉĎ§cňª�ňN'óČM£ĎT ŉ

)bÎÀt¿ÁĀx¡M£ń¿Á§cĻĪŀ�M£ń¿ÁğĸļŁŅğĽŃM£ń¿ÁĩġīĦħĴ©¨M£ńĖŃĦħĴ¿Ás2M£ń¿Á�FĀÈå©¨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

Kyoto's X-ray Astronomical SOI pixel sensor...Kyoto's X-ray Astronomical SOI pixel sensor T.G.Tsuru...

Documents

Transcript of Kyoto's X-ray Astronomical SOI pixel sensor...Kyoto's X-ray Astronomical SOI pixel sensor T.G.Tsuru...