DANAE - a new experiment for direct dark matter detection using - - PowerPoint PPT Presentation

SLIDE 1

DANAE - a new experiment for 
 direct dark matter detection using 
 RNDR DEPFET detectors

Hexi Shi
 HEPHY ÖAW

11 April 2018 DEPFET workshop 
 Schloss Ringberg

SLIDE 2

Collaboration

• A. Bähr A, J. Ninkovic A, J. Treis A, 

• H. Kluck B,C, J. Schieck B ,C, H. Shi B,



 Max-Planck-Gesellschaft Halbleiterlabor, Germany A, 
 Institut für Hochenergiephysik der Österreichischen Akademie der Wissenschaften, Vienna, Austria B,
 Atominstitut, Technische Universität Wien, Vienna, Austria C


 DANAE (DANAË)
 Direct dArk matter search using DEPFET with repetitive- Non-destructive-readout Application Experiment

OeAW funding for detector technology

“Danae” by G. Klimt

SLIDE 3

3

The project overview

[MeV]

χ

m 10

2

10

3

10 ]

2

[cm

e

σ

44 −

10

43 −

10

42 −

10

41 −

10

40 −

10

39 −

10

38 −

10

37 −

10

36 −

10

35 −

10

y ⋅ , 3 background events, 1.0 kg

• =2e

T

E y ⋅ , 3 background events, 1.0 kg

• =1e

T

E y ⋅ , 3 background events, 3.0 kg

• =2e

T

E y ⋅ , 0 background events, 1.0 kg

• =2e

T

E excluded by Xenon10 excluded by Xenon100

Direct Dark Matter Detection with DEPFET

• minimal reach for nuclear recoil experiments 


about few 100 MeV

• dark matter electron scattering offers 


reach towards MeV dark matter

• measurement of low noise ionisation signal

in low background environment

• RNDR DEPFET sensors developed by

semiconductor laboratory of MPG

• setup for proof-of-principle 


measurement currently prepared

• expect first results early 2019

15

• 1

1 2 3 4 5 6 10 20 30 40 50 60

Counts Pulseheight (# Electrons)

Without laser (Noise peak) Weak laser (poissonian photon distribution) Gauss fit to noise peak

σ = 0,21 e-

more information: arXiv:1706.08666

CRESST nuclear recoil

σ=0.21 e-

from Jochen Schieck “Experimental Dark Matter Search at HEPHY”

EPJ C, 77(12), 279 (2017)

*Repetitive Non-Destructive Readout *

SLIDE 4

4

Dark matter landscape - partly

“WIMP” as a dark matter candidate : 
 


• weakly interacting with matter


<σWIMP・v> ∼ GF2 ・mΧ2 ∼ 1/ΩΧ


• fits the Hubble constant and “relic” 


density of dark matter predicts dark matter WIMP mass between 2 GeV and 120 TeV

1 TeV 1 keV 1 GeV 1 MeV

WIMPs

dominated the direct detection experiments 
 until recently

~ ~

Over 80% of the mass in the universe is invisible dark matter

Credit: NASA / WMAP Science Team

SLIDE 5

5

WIMP direct detection method

light phono electri

nucleus DM e- DM nucleus e- *for silicon image credit R. Essig

Energy deposit in target material in forms of :

• light
• phonon
• electric charge

. 800 km/s

typical DM velocity vχ

look for nuclear recoils from 
 WIMP-nucleus scattering

• Detection limitation :

energy deposit from nucleus recoil ENR ~ 2μχ,N2 vχ/mN

• > for 100 MeV mχ, ENR ~ 1 eV *

plus quenching factors and noise level of the detectors

SLIDE 6

6

DM-nucleus scattering direct search status

arXiv:1711.07692

1 GeV/c2

SLIDE 7

6

DM-nucleus scattering direct search status

arXiv:1711.07692

1 GeV/c2

no evidence for WIMP yet

weak interaction scale

SLIDE 8

6

DM-nucleus scattering direct search status

arXiv:1711.07692

1 GeV/c2 DAMIC CDMS

no evidence for WIMP yet

weak interaction scale

SLIDE 9

7

Dark Sector and Light Dark Matter

1 TeV

WIMPs

1 keV 1 GeV 1 MeV

several sharp “theory” targets

(freeze-out, asymmetric, freeze-in, SIMP , ELDER)

Dark sectors

(DM + new mediators)

DM scattering

Aʹ, Φ

χ

χ

gχ

SM SM DM DM

Dark sector : interaction between DM and standard model particle 
 mediated by a dark photon (one example of mediators) clear predictions from multiple models over wide DM mass region, including 
 keV ~ GeV range

• > comparable observables

in experiments

~ ~

image credit R. Essig

SLIDE 10

8

DM-electron scattering

DM DM e-

nucleus

e-

nucleus

> ∆E

need

vDM . 800 km/s =

⇒

kinematically to overcome binding energy ΔE need

image credit R. Essig

O(100 keV)

JHEP05(2016)046

SLIDE 11

8

DM-electron scattering

DM DM e-

nucleus

e-

nucleus

> ∆E

need

vDM . 800 km/s =

⇒

kinematically to overcome binding energy ΔE need

image credit R. Essig

transferred energy:

ΔΕ

(for outer shell electron)

typical recoil energy

bound e- does not have definite momentum, typical momentum transfer is set by e- not by DM.

ΔEe ~ 4 eV

O(100 keV)

JHEP05(2016)046

SLIDE 12

9

Target materials for electron recoils

Target Type Examples Eth mχ 
 threshold Status Timescale

Noble liquids Xe, Ar, He ~ 10 eV ~ 5 MeV

Done w data; improvements possible

existing Semi- conductors Ge, Si ~ 1 eV ~ 200 keV

(Eth ~ 40 eV SuperCDMS, DAMIC)
 Eth ~ 1eV SENSEI, DEPFET R&D

~ 1-2 years Scintillators GaAs, NaI, CsI, … ~ 1 eV ~ 200 keV

R&D required

≲ 5 years Supferfluid He ~ 1 eV ~ 1 MeV

R&D required
 unknown background

≲ 5 years

Super- conductor Al ~ 1 meV ~ 1 keV

R&D required
 unknown background

~ 10 - 15 years

arXiv:1608.08632

SLIDE 13

10

Application of Silicon detector DAMIC

nucleus recoil CCD, with physics results Readout noise determines threshold of ~ 11 e-

Ω μ

• z

x y

σ σ

Physics Procedia 61 (2015) 21 – 33

DAMIC 1 GeV/c2 (or ~ 40 eV)

SLIDE 14

10

Application of Silicon detector DAMIC

nucleus recoil CCD, with physics results Readout noise determines threshold of ~ 11 e-

Ω μ

• z

x y

σ σ

Physics Procedia 61 (2015) 21 – 33

For O(MeV) DM-electron scattering, required threshold : O(e- )
 Sub-electron noise level necessary DAMIC 1 GeV/c2 (or ~ 40 eV)

SLIDE 15

Skipper CCD for SENSEI

PRL 119(1) 131802 (2017)

from SENSEI homepage 11

Readout noise : 1 sample : 3.55 e- rms 4k samples : 0.068 e- rms @ 140 K expected dark current 
 (from DAMIC CCD) : 
 < 10-3 e-/pix/day DAMIC CCD with repetitive readout

SLIDE 16

DM-electron cross section

12

SENSEI first result from a surface run

]

• Charge [e

1 2 3 4 5 6

Entries

• 1

10 1 10

2

10

3

10

4

10

Exposure: 0.019 gram-days

arXiv:1804.00088v1

Active mass : 0.071 grams 427 minutes exposure (0.33 g-hr) above sea level 220 m single read noise : ~ 4 e- effective noise : ~ 0.14 e- (800 repetitions) dark current : ~1.14 e-/pixel/day assume all events DM induced

• > conservative limit
• χ

σ

α

• |
• χ []

σ[]

=(α/)

• =
• |
• |

= |

• |

=

• |
• |

= |

• |
• ultralight dark photon mediator

from SENSEI homepage Freeze-In arXiv:1804.00088v1 liquid Xenon

DM-electron cross section

SLIDE 17

DM-electron cross section

12

SENSEI first result from a surface run

]

• Charge [e

1 2 3 4 5 6

Entries

• 1

10 1 10

2

10

3

10

4

10

Exposure: 0.019 gram-days

arXiv:1804.00088v1

Active mass : 0.071 grams 427 minutes exposure (0.33 g-hr) above sea level 220 m single read noise : ~ 4 e- effective noise : ~ 0.14 e- (800 repetitions) dark current : ~1.14 e-/pixel/day assume all events DM induced

• > conservative limit
• χ

σ

α

• |
• χ []

σ[]

=(α/)

• =
• |
• |

= |

• |

=

• |
• |

= |

• |
• ultralight dark photon mediator

from SENSEI homepage Freeze-In arXiv:1804.00088v1

• χ

σ

α

• ||=||
• liquid Xenon

DM-electron cross section

SLIDE 18

13

DEPFET with RNDR

• DM

structure of a basic DEPFET cell : a “subpixel” structure of RNDR DEPFET “super-pixel” RNDR : repetitive non-destructive readout fully-depleted n-Si

EPJ C, 77(12), 279 (2017) EPJ C, 77(12), 279 (2017)

SLIDE 19

14

RNDR

• RNDR readout

read N times effective noise : σeff = σ/(√N)

EPJ C, 77(12), 279 (2017)

SLIDE 20

14

RNDR

• RNDR readout

read 1 : noise σ read N times effective noise : σeff = σ/(√N)

EPJ C, 77(12), 279 (2017)

SLIDE 21

14

RNDR

• RNDR readout

read 1 : noise σ transfer gate open read N times effective noise : σeff = σ/(√N)

EPJ C, 77(12), 279 (2017)

SLIDE 22

14

RNDR

• RNDR readout

read 1 : noise σ transfer gate open read N times effective noise : σeff = σ/(√N) read 2 : noise σ

EPJ C, 77(12), 279 (2017)

SLIDE 23

14

RNDR

• RNDR readout

read 1 : noise σ transfer gate open read N times effective noise : σeff = σ/(√N) clear charges read 2 : noise σ : repeat N times

independent
 measurements

EPJ C, 77(12), 279 (2017)

SLIDE 24

15

DEPFET RNDR single pixel performance

EPJ C, 77(12), 279 (2017)

confirmed the 1/ √N decrease of σeff minimal noise level limited by leakage current at 230 K (-40 ℃)

SLIDE 25

15

DEPFET RNDR single pixel performance

EPJ C, 77(12), 279 (2017)

confirmed the 1/ √N decrease of σeff minimal noise level limited by leakage current at 230 K (-40 ℃) estimated temperature dependence, to be testified in measurement

EPJ C, 77(12), 279 (2017)

• 50 ℃

ideal case

SLIDE 26

15

DEPFET RNDR single pixel performance

EPJ C, 77(12), 279 (2017)

confirmed the 1/ √N decrease of σeff minimal noise level limited by leakage current at 230 K (-40 ℃) new architecture with “blind-gate” possibility of reducing leakage current during readout estimated temperature dependence, to be testified in measurement

EPJ C, 77(12), 279 (2017)

• 50 ℃

ideal case

SLIDE 27

16

DEPFET RNDR single pixel performance

EPJ C, 77(12), 279 (2017)

single pixel RNDR DEPFET effective noise : 0.2 e− RMS at 200 K

SLIDE 28

17

A comparison with skipper CCD

Type Pixel format
 [μm] prototype mass

• perating

temp dark current readout time (1sample) readout noise (optimal) skipper CCD 15 x 15 x 200 0.071 g 140 K ~1.14 
 e-/pix/day 10 μs/pix/ amplifier 0.068 
 e-rms/pix RNDR DEPFET 75 x 75 x 450 0.024 g ≲ 200 K < 1 
 e-/pix/day 4 μs/
 64 pix 0.2 
 e-rms/pix similar concepts of non-destructive readout, compatible performance; different architecture, different systematics;

• > good complementary from experimental point of view

SLIDE 29

18

DANAE proof-of-principle measurement

matrix readout

• ptimization for operational/readout parameters

temperature dependence of leakage current proto-type : 75 um x 75 um x 450 um single pixel, 64 x 64 matrix sensitive volume 0.024 g

At HLL : In Vienna:

low background environment measurement

• r surface measurement with veto


 MC simulation for background budget

• Expect to have operating

matrix by the second half 


• f 2018

Detector prototype at HLL-MPG courtesy of J. Treis

SLIDE 30

19

DANAE preparation status

SLIDE 31

20

DANAE test setup - design image Assembly at HLL

10 cm Stirling-cycle cryocooler Vacuum chamber

SLIDE 32

Setup at HLL

Vacuum and cooling tests done in March 2018 cooling pad reached 150 K

21

SLIDE 33

22

Cooling & shielding layout

top-out top-inner bot-out window top out top inner

• uter shielding : support structure

inner shielding : cooling contact

SLIDE 34

23

Cooling of the detector and electronics PEEK columns thermal insulators

Readout 
 board @ RT detector matrix flex PCB

SLIDE 35

24

Readout electronics Switcher-S VERITAS

64x2 channel analog multiplexer

switcher id W N E

function

Gate 1 & 2 Gate common clear & transfer gate Voltage [V]

• 2.5 ~ + 5
• 0.5 ~ +20
• 0.5 ~ + 20/25

Detector matrix

• VERITAS 2 ASIC in the AMS 0.35 μm CMOS 3.3 V technology 

• 64 analog readout channels able to process in parallel the 


signals coming from 64 DEPFET devices. Front-end ASICS for the 64x64 matrix with interface to Switcher-S, VERITAS

ADC

FADC type digitizer

Readout board

SLIDE 36

25

Readout electronics designs

Pitch adaptors detector matrix flexible PCB 7.5 cm Readout board

SLIDE 37

26

Monte Carlo simulation with Geant4

• to have a guideline of particle tracks and hit pattern,

prepare the library of analysis routine;

• for future design of VETO counters and calibration layout.
• geometry of the setup from vacuum parts 3D model;
• primitive geometrical shape for DEPFET :

75 um x 75 um x 450 um bulk pixel, 64 by 64 matrix;

• to check response to X-rays, ambient gamma, cosmic charged particles, 


and neutrons;

SLIDE 38

[MeV]

χ

m 10

2

10

3

10 ]

2

[cm

e

σ

44 −

10

43 −

10

42 −

10

41 −

10

40 −

10

39 −

10

38 −

10

37 −

10

36 −

10

35 −

10

y ⋅ , 6 background events, 1.0 kg

• =2e

T

E y ⋅ , 6 background events, 0.0009 kg

• =2e

T

E y ⋅ , 6 background events, 3.0 kg

• =2e

T

E y ⋅ , 0 background events, 1.0 kg

• =2e

T

E excluded by Xenon10 excluded by Xenon100

P r e l i m i n a r y

plot from Jochen Schieck

physics run goal 0.9 g*y

• > 40 matrices

~1 g sensitive
 volume

Physics run perspective

• Expect preliminary results from the prototype setup (0.024 g

sensitive volume) in late 2018

• physics run with significant result requires more matrices

27

DM-electron cross section

SLIDE 39

28

Future tasks & topics

• readout electronics, DAQ in preparation;
• readout test, leakage current test;
• calibration regime and configuration;
• simulation for background;
• design for further shielding - passive and active.

Summary

• sub e- ENC low noise semiconductor detector provides the

possibility to detect the energy deposit from sub-GeV DM- electron recoil;

• DANAE prototype for test-of-principle measurement with single

matrix in preparation;

• one of the first generation experiments using non-destructive

repetitive readout method.

SLIDE 40

Let us put DANAE into the landscape 


• f direct DM detection experiments!!

SLIDE 41

30

Hit pattern simulation with Geant4

Results from DAMIC experiment

• to have a guideline of particle tracks and hit pattern,

prepare the library of analysis routine;

• for future design of VETO counters and calibration layout.

Physics Procedia 61 (2015) 21 – 33

SLIDE 42

31

Detector Structures

Macropixel sensors

• Prototypes
• 64 x 64 pixels (500 µm & 300 µm)
• 3.2 x 3.2 cm²
• high framerate (4 kHz)
• near Fano-Limited energy resolution

Johannes Treis / Halbleiterlabor der MPG

SLIDE 43

32

t p t a

3 2 2 1 2 2

2 1 2 A I q A C a A C g kT ENC

L tot f tot m

+ + =

thermal noise 1/f noise leakage current

Noise analysis: multi-parameter fit » extraction of

• total capacitance

Ctot

• 1/f noise coeff.

af

• leakage current

IL independent measurement of

• transconductance gm

(180 … 250 µS) Ai = filter constants (Gaussian 6th order) τ = shaping time constant q = electron charge α = 2/3 for FET in saturation

Spectroscopic detectors

for good resolution & high count rate capability the total capacitance must be minimised!!

m L tot

• pt

g I C q kT A A τ 3 2 2

2 1 3

=

Johannes Treis / Halbleiterlabor der MPG

Johannes Treis / Halbleiterlabor der MPG

Small pixel device layouts

• Global Clear variant
• Blinds also used for clearing
• 36 x 36 m

m2

SLIDE 44

Detector Structures – Matrix Devices

readout sequence

Correlated double sampling: 1st measurement: signal + baseline clear: removal of signal charges 2nd measurement: baseline difference = signal complete clear is mandatory!

matrix operation

horizontal supply lines, row selection vertical signal lines 1 active row, other pixels integrating

• ption to speed up (1)

readout parallelisation 2 x readout channels, 2 active rows

Johannes Treis / Halbleiterlabor der MPG

DEPFET CDS circle

SLIDE 45

CCD (skipper) readout

CCD: readout

4 3rd Berkeley Workshop on the Direct Detection of Dark Matter December 6, 2016

Lowering the noise: Skipper CCD

Main difference: the Skipper CCD allows multiple sampling of the same pixel without corrupting the charge packet. The final pixel value is the average of the samples Pixel value = 1

NΣN i (pixel sample)i

6 3rd Berkeley Workshop on the Direct Detection of Dark Matter December 6, 2016

Lowering the noise: Skipper CCD

Main difference: the Skipper CCD allows multiple sampling of the same pixel without corrupting the charge packet. The final pixel value is the average of the samples Pixel value = 1

NΣN i (pixel sample)i

low frequency noise Regular CCD Skipper CCD pedestal signal high frequency noise pixel charge measurement

6 3rd Berkeley Workshop on the Direct Detection of Dark Matter December 6, 2016

CCD: readout

1 2 3 7

.. .. ..

P2 P1 P3 P2 P1 P3

3x3 pixels CCD

P2 P1 P3 P2 P1 P3 P2 P1 P3 H2 H1 H3 H2 H1 H3 H2 H1 H3

amplifier channel stop horizontal register sens node channel stop

state

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

2 3rd Berkeley Workshop on the Direct Detection of Dark Matter December 6, 2016

Javier Tiffenberg

SW : summing-well gate OG : output gate RG : reset gate VR : V_ref

SLIDE 46

Skipper CCD

• The “skipper” allow multiple

readouts of the charge in each pixel.

• Floating gate output

instead of floating diffusion output used in regular CCDs.

• The charge can be

moved back and forth between

• Each readout integration

time is kept short to make 1/f noise negligible.

• A noise reduction of

1/sqrt(N) is achieved for N reads.

• The total readout time per

pixel increases linearly with N.