ccd response to a be source - university of chicago · 880 900 920 940 960 980 1000 2000 3000 4000...

CCD response to a 124Sb-9Be source

Alvaro E ChavarriaKavli Institute for Cosmological Physics

1

• Photo-neutron calibration principle.

• Nuclear recoil spectrum from data.

• Expected recoil spectrum from simulation.

• Extraction of ionization efficiency.

• Systematics.

• Results.

2

Outline

rRecoil energy / keV

0 0.5 1 1.5 2 2.5 3 3.5

Rate

0

0.5

1

1.5

2

2.5

Simulated recoil spectrum in CCD

3.2 keVr

end-point

124Sb-Be

3

Source24 keV

neutrons from 9Be(γ,n) reaction

82

clearly demonstrates monochromatic neutrons peaking at 23.5 keV in the spectrum.

The tail in the lower energy range in the spectrum is due to neutron moderation by

BeO. The cross-section of 9Be(�, n) at 1.69 MeV is 1.64 mbar from xxxx measurement

in Figure 5.6. It shows that the e�ciency of neutron production with 135.8g of BeO

is 1.47e-4 in the simulations.

Figure 5.4: Setup geometry defined in the MCNP simulations viewed from the side.

The neutrons go through the stainless steel flange with a thinkness of 9/16” after

propagating in the lead. We use a 3He detector to measure the neutron fluxes around

the lead castle and around the CCD chamber to test the neutron propagation in

the lead and the 9Be(�, n) photonuclear cross-section in the simulations, shown in

Figure 5.7. 3He reacts by absorbing thermal neutrons, producing a 1H and a 3H ion.

Its sensitivity to � rays is negligible, and therefore providing a very useful neutron

detector. In order to moderate neutrons from BeSb to thermal neutrons, we put a

polyethylene cylinder 1-inch thick around the 3He detector. Figure 5.8 shows the flux

of neutrons before reaching polyethylene that are actually captured by 3He which

demonstrates that the detector with polyethylene is sensitive to neutrons peaked at

energy ⇠ 23 keV. We put a 6mm cadmium foil around the 3He detector to absorb

thermal neutrons in order to decrease their influences. Natural cadmium contains

12.22 % 113Cd with a cross-section of 104 barn in the thermal energy range. Historical

measurements with the 3He detector show a 10% uncertainty.

Figure 5.7 shows 17 measurement positions with the 3He detector. The spec-

Need to stop γs

• Irradiated at NC State.

• Activity measured with HPGe on September 1st:

4

0.290 ± 0.010 mCi

~5 mCi when data taking started.

89

Figure 5.13: Di↵erent source geometries, including ”Full BeO”, ”Outer BeO”, ”InnerBeO”, ”Outer BeO Inner Al2O3”, and ”Poly”.

1000s n s-1

Replacing BeO parts with Al allows to measure γ background.

124Sb Source

5

CCD setup

Pump Lead

ChamberDAQ

500 μm CCD.8 Mpixel.

15 μm x 15 μm pixels.Set temperature 130 K.

Substrate bias 130 V.

Flex cable

Copper frame

UChicago

Detector

��

Charged particles produce ionization

in CCD bulk.

Charge collected by each pixel on CCD plane is

read out.

3.62 eV for e-h pair.

~2 e- RMS read-out noise.

��

��

��

��

��

��

��

��

��

��

��

��

z

x

y

��

Charge drifted up and held at gates.

z

pixe

l

x

x

y

σxy

σxy 6

7

DM Motivation CCDs Particle detection Quenching DAMIC Near future Summary BACK UP

CCD: readout

1

2

3

7

.. .. ..P2P1 P3P2P1 P33x3 pixels CCD

P2P1 P3 P2P1 P3 P2P1 P3

H2

H1

H3

H2

H1

H3

H2

H1

H3

ampli�er

channelstop

horiz

onta

l reg

iste

r

sens node

channelstop

state

capacitance of the system is set by the SN: C=0.05pF! 3µV/e

10 / 52

CCD readoute

Not triggered

19802000

20202040

20602080

1060

1080

1100

1120

1140

1000

2000

3000

4000

5000

6000

7000

8000

9000

10000image

54321Energy measured by pixel / keV

1980 2000 2020 2040 2060 2080

1060

1080

1100

1120

1140

1000

2000

3000

4000

5000

6000

7000

8000

9000

10000image

1740 1760 1780 1800 1820 1840880

900

920

940

960

980

1000

2000

3000

4000

5000

6000

7000

8000

9000

10000image

Diffusion limitedParticle tracks6 keV front

6 keV back

41804190

42004210

42201280

1290

1300

1310

1320

1330

5 10 15 20 25 30image

Energy measured by pixel / keV30252015105

4180 4190 4200 4210 42201280

1290

1300

1310

1320

1330

5

10

15

20

25

30image

α

e

μ

X-ray?n, WIMP?

Diffusionlimited

50 p

ixel

s

8

Front

Back

9

CCD Performance

>95% of the image good quality.

eeEnergy measured in pixel / eV

0 50 100 150 200

1

10

210

310

410

510

Distribution of pixel values in image

30 ks exposure

blank

6.7 eVee

RMS noise!

10794 images acquired over 126 days. All good.

±6%

SNOLAB

10

Energy / keV0 1 2 3 4 5 6 71

10

210

310

410

510

Fe source spectrum in Chicago chamber55

Mn K! Mn K"

Cr K!Mn Kescape lines

Noise

Si K!Cl K!

Image number4500 5000 5500 6000 6500

ee

mean / keV

αMn K

5.8

5.82

5.84

5.86

5.88

5.9

5.92

5.94

lineα

Stability of Mn K

ee

Resolution RMS / eV

48

50

52

54

56

58

60

62

27 days of 55Fe data

0.1% fluctuations in energy scale1% fluctuations in resolution

55Fe data

53 eV at 5.9 keV from front

x / pixels500 1000 1500 2000 2500 3000 3500 4000

y / pixels

200

400

600

800

1000

1200

1400

1600

1800

2000

Spatial distribution of events

11Energy / keV

0 2 4 6 8 10 12 14 16 18 20

Counts per 50 eV

1

10

210

310

410

spectrum from frontβH 3

Cluster energy / keV0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8

-1 (

10 eV)

-1Si

ngle p

ixel

eve

nts

/ Mp

ix

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5 / ndf 2χ 25.75 / 26

Prob 0.4767A 0.0053± 0.1099 µ 0.0011± 0.5215 σ 0.00104± 0.02148

c 0.0183± 0.3968

H source on CCD3

Source regionExpected noiseBackground region

Backscattered e-?

2 μm gate structure 21 eVee resolution

at 525 eVee

Tritium source

12

Likelihood clusteringN

e

(E)⇥ Gaus(x, y, µx

, µ

y

,�(z))

Number of ionized electrons

Best estimate for mean of energy

depositionLateral spread

Use moving window and fit to a 2D Gaussian

distribution. Register LL of best-fit. Compared to LL of

constant pixel values. Difference between the two LL (ΔLL) allows us to select

for physical events.

60 eV

13

0.01 events per image from noise

Selection cuts / ndf 2χ 152.2 / 58

Prob 2.175e-10

p0 1.284e+06± 1.144e+08

p1 0.002± 1.318

LLΔ-100 -90 -80 -70 -60 -50 -40 -30 -20 -10 0

1

10

210

310

410

510

610

/ ndf 2χ 152.2 / 58

Prob 2.175e-10

p0 1.284e+06± 1.144e+08

p1 0.002± 1.318

LL for events with cdist < 1.5 and 80 eV < E < 250 eVΔ 60 eV

eeMeasured energy / keV

0 0.05 0.1 0.15 0.2 0.25 0.3

Acceptance

0

0.2

0.4

0.6

0.8

1

Acceptance for events in bulk

From events simulated on real images.

50% at 90 eVee

14

eeEnergy / keV

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8

/ pixels

xy

σ

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0

1

2

3

4

5

6

7

8

9

10

= 130 Vsub

source and VγDepth vs. E with

500 μm at 130 V

eeEnergy / keV

4 4.5 5 5.5 6 6.5 7 7.5 8

/ pixels

xy

σ

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0

10

20

30

40

50

60

70

80

90

= 130 Vsub

source and VγDepth vs. E with

Fe KαCr Kα

eeEnergy / keV

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8

/ pixels

xy

σ

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0

20

40

60

80

100

120

140

Depth vs. E for simulated bulk events

Data

Data

Simulation

SimulationSimulated events with diffusion model with

parameters set to data.

Events pasted on a subsample of 1000 acquired images.

eeEnergy / keV

0 0.5 1 1.5 2 2.5 3 3.5 40

50

100

150

200

250

backgroundγMeasured

Compton spectrum model

Sb-Al configuration124Data from

eeEnergy / keV

0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.50

50

100

150

200

250

backgroundγMeasured

Compton spectrum model

Sb-Al configuration124Data from

15

Background spectrum(neutrons off)

K edge

L edge

Backscattered e-?

Increase at low energies correlated with source intensity.Not CCD noise.

eeEnergy /keV0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

eeCounts per 10 eV

0

200

400

600

800

1000

1200

Sb-Be124Observed spectrum with

ray backgroundγ

Data specturmData spectrum

16

Observed spectrum

Normalized to count rate

2-5 keVee.

Uncertainty propagated in

analysis.

17

For a particular run integrate number of 124Sb decays during exposure.

Signal spectrum

eeMeasured energy from ionization / keV

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Counts per bin

0

100

200

300

400

500

600

700

Data spectrum

Spectrum for Nd decays

Background subtracted spectrum

CCDs is fully active and has

very high spatial resolution

This is the single recoil spectrum

18

MCNP gives neutron fluxFull detector geometry modeled in MCNPX 5. 9Be(γ,n) using latest value from Arnold et al1.

Pb elastic scattering cross-sections from A. Robinson2.

1PRC85 0446052PRC89 032801

Neutron energy/keV0 2 4 6 8 10 12 14 16 18 20 22 24

decay

9Flux per 10

0

0.2

0.4

0.6

0.8

1

1.2

1.4

Neutron flux thru CCD from SbBe

Spectrum moderated.Resonances of

nearby materials (mostly Cu) evident.

Incident neutron energy / keV0 5 10 15 20 25 30

/ b

σ

1.7

1.8

1.9

2

2.1

2.2

2.3

2.4

2.5 / ndf 2χ 62.85 / 60

Prob 0.3756

p0 0.01702± 2.122

p1 0.001295±0.01106 −

/ ndf 2χ 62.85 / 60

Prob 0.3756

p0 0.01702± 2.122

p1 0.001295±0.01106 −

σSi elastic scattering nat

19

9. ANGULAR DISTRIBUTIONS

9.1 ELASTIC ANGULAR DISTRIBUTIONS

There is a large amount of neutron elastic scattering data from 40 keV to 15 MeV. Below 40 keV, the angular distribution is assumed to be isotropic, which was observed by Lane et al. (LA61, LA62).

From 40 keV to 800 keV, the distributions were taken from a R-function analysis of the data of Kinney and McConnell (KI76). The angular distribution data of Kinney and McConnell (thinned) were used between 800 keV and 3 MeV. The data at 3.1 MeV are from Popov (P061) and those at 3.5 MeVare from TanT& (TA64). As noted in Section 3.2, from 4 to 9 MeV, angular distributions measured at nearby energies were averaged, resulting in a set of seven distributions at 4.25,4.8,5.25, 5.8,6.6,7.75, and 8.75 MeV. The average distributions were obtained by fitting a Legendre series to each data set and averaging the coefficients to obtain the results used in the evaluation. The data sets which make up each averaged group are given in Table 5. The data at 9.8 MeV are from the work of Obst and Weil (OB73), and at 10.95 MeV from Nellis and Buchanan (NEi72). From 11 to 20 MeV, the results of the optical model calculations from GENOA (PE67) are used to provide the recommended data set.

'

9.2 INELASTIC ANGULAR DISTRIBUTIONS

For the discrete levels 1.779, 4.617, 6.276, 6.879, and 6.8899 MeV of =Si, the angular distributions in ENDFB-VI are a weighted sum of Legendre coefficients from the TNG and direct interaction (DWUCK) calculations. The angular distributions for all other levels of =Si, *'Si, and 30Si were taken from the TNG analyses. The calculated differential (n,n') cross sections for exciting selected low-lying discrete levels are compared with measurements in Figs- 38 through 49. The need for nuclear model analyses and better data can be seen from these figures for in many cases the measurements disagree with each other.

9.3 ANGULAR DISTRIBUTIONS OF NEUTRON-PRODUCTION CROSS SECTIONS

The computed angular distributions of neutron production cross sections �or silicon at an incident energy of 14.5 MeV and for secondary energies of E: = 2.0-3.0,4.0-5.0,6.0-7.0, and 7.0-8.0 MeV are compared with experiments in Fig. 50. Again, discrepancies exist among the measured data sets. The results for E: = 7.0-8.0 MeV correspond to levels in the discrete region and include the sum of TNG and DWUCK calculations. For E',, = 6.0-7.0 MeV, the results are also in the discrete region but are symmetric since there was no direct interaction contribution included and TNG computes the angular distributions for only the compound component of the discrete levels. The computed results lie below the data for the low outgoing energies and will be discussed further in the section on neutron emission below.

21

ENDF-VI

Larson data (±3% systematic)


0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

-1

Count density / eV

0

2

4

6

8

10

12

True recoil spectrum

Simple numerical operation

End-point3.2 keVr

Simulated

Recoil spectrum from flux


0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Counts per bin

0

100

200

300

400

500

600

700

Data spectrum

20

Given an expected recoil spectrum for our setup, we need

to find mapping from Eee to Er that best matches our

measured spectrum.

Expected spectrum is obtained from Monte Carlo.


0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

-1

Count density / eV

0

2

4

6

8

10

12


Both for same Nd

Simulated

Principle


0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Counts per bin

0

100

200

300

400

500

600

700

Data spectrum


0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

-1

Count density / eV

0

2

4

6

8

10

12


21

Eee

Integral down to Eee = N ± σ Integrals up to

N-σ, N, N+σ

Er +σr-σr

σ obtained by adding bin uncertainties in quadrature. ±σr generally asymmetric.

Simulated

Integral solution

22

/ ndf 2χ 132.9 / 151

Prob 0.8524

(0.1) -1f 0.0154± 0.9338

(0.2) -1f 0.019± 1.474

(0.4)) -1f 0.04± 2.32

f'(0) 0.01407± 0.04725

) max

f'(E 1.765± 2.591

) max

f(E 0.530± 1.332

y offset 1.2114± -0.1067

eeIonization energy / keV

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

ee

Counts per 10 eV

0

200

400

600

800

1000

1200 / ndf 2χ 132.9 / 151

Prob 0.8524

(0.1) -1f 0.0154± 0.9338

(0.2) -1f 0.019± 1.474

(0.4)) -1f 0.04± 2.32

f'(0) 0.01407± 0.04725

) max

f'(E 1.765± 2.591

) max

f(E 0.530± 1.332

y offset 1.2114± -0.1067

Cubic spline fit to FullBe data set

Cross-check

For every data set we also do a cubic spline spectral fit.Fit result used to correct for detector resolution.

Fitting and integral solution give same result when they should.

Recoil energy / keV0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2

ee

Ionization energy / keV

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

Fit with cubic spline

Fit with cubic spline

Integral solution

23

89


Source configuration

89


89



0.5 1 1.5 2 2.5 3

rSb decays per eV

124

13

Counts per 10

0

0.5

1

1.5

2

2.5

MCNP simulated spectra

FullBe

InnerBe

InnerAlPoly

InnerAlumina

eeIonization energy / keV

0.1 0.2 0.3 0.4 0.5 0.6

ee

Sb decays per 10 eV

124

13

Counts per 10

0

20

40

60

80

100

120

140CCD data spectra

FullBeInnerBe

InnerAlPoly

InnerAlumina

InnerAl

24


0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2

ee


0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

FullBeInnerBe

InnerAlPoly

InnerAlumina

Results from different source configurations

Source configurationUse to study systematic on MCNP source modeling.

Test neutron propagation with modified neutron spectrum.

Tension between FullBe/InnerBe and

InnerAl/InnerAlumina.

For today take difference as systematic.

25

84

Figure 5.7: Setup of the 3He detector. a) Viewed from above where circles showmeasurement locations. b)3He detector with polyethylene wrapped around.

keV0 5 10 15 20 25 30

dec

ay8

He p

er 1

03

neut

rons

cap

ture

d in

50

100

150

200

250

300

Figure 5.8: Flux of neutrons before reaching polyethylene that are actually capturedby 3He which demonstrates that the detector with polyethylene is sensitive to neutronspeaked at energy ⇠ 23 keV.

3He counter test

Cadmium layer around 3He counter to stop thermals.Thickness of poly chosen as to be most sensitive to

15 - 25 keV neutrons.

3He counter test

26


0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2ee


0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

FullBe

Right

Left

Results for source in different positions

Position

Measured rate / s-1

Predicted rate / s-1

8 14.81 14.90 ± 0.83

9 4.00 4.09 ± 0.23

10 4.01 4.11 ± 0.23

11 0.45 0.46 ± 0.03

12 3.95 4.14 ± 0.23

16 0.70 0.63 ± 0.04

17 0.65 0.58 ± 0.03

6.25% RMS ratio between measurement and simulation. Take as systematic.

Result from moving the neutron source ±3.4 cm consistent with 3He.

Black points include 6.25%

systematic

27


0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2

ee


0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

ee=20eVσFano=0.15,

ee=25eVσFano=0.85,

ee=30eVσFano=1.00,

Results for different energy resolution

Systematics summary• Difference between FullBe and Alumina setup.

• 5.6% from source intensity.

• 6.3% in amplitude from geometry.

• 3% total rate from elastic scattering σ.

Other effects found negligible, e.g. resolution

±15 eVr

extremes


0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2

ee


0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

Ionization efficiency for nuclear recoils in Si

Central valueTotal uncertainty

Systematic

Statistical

28

Results

740±120 eVr at 60 eVee


1 10

ee


-110

1

10

Ionization efficiency in silicon

Be9Sb124UChicago

Antonella (systematic)

Antonella

Gerbier et al (1990)

Lindhard, k=0.15

Lindhard, k=0.05

29

In context

Covered entire DAMIC WIMP search ROI

• Photo-neutron sources offer promising alternative to calibrate detectors at the lowest energies.

• Complementary results by DAMIC collaboration down to DAMIC100 threshold (60 eVee).

• Discrepancy with Lindhard model below 5 keVr in silicon.

30

Conclusions

ccd response to a be source - university of chicago · 880 900 920 940 960 980 1000 2000 3000 4000...

Documents