Ultra Low Background Front End Electronics

for the Majorana Demonstrator

Paul Barton

Applied Nuclear Physics Program, Nuclear Science Division

Lawrence Berkeley National Laboratory

14 May 2019

Conference on Science at SURF

P. Barton – 2CoSSURF

Semiconductor Detector Lab Activities at LBNL

Low Mass Front End (LMFE) for MajoranaLAr LMFE for LEGEND 0νββLDRD for LEGEND ASIC

CPG CdZnTe Module w/ depth sensingSpherical Coded Aperture / Compton Imager

HPGe double-sided strip detectorsfor gamma ray imaging

LBNL Scientific CCD for MeV Photon Beam Diagnostics

P. Barton – 3CoSSURF

Neutrinoless Double Beta Decay

in 76Ge

Is the neutrino it’s own antiparticle?

P. Barton – 4CoSSURF

Double-Beta Decay

(A,Z)

Nuclear

(A,Z+2)

2νββ-decay

Process

W−

W− e−

ν̄e ν̄e

e−

(A,Z)

(A,Z+2)

0νββ-decay

(“vanilla model”)

W−

e−

e−

Nuclear ν̄Process ν

W −

• 2νββ-decay: SM allowed and observed.• 0νββ-decay: Violates lepton number conservation (a requirement of

some neutrino mass models). Not yet observed.

• For 76Ge (Q = 2039 keV):

T2ν1/2

∼ 1021 y → Rare process.

T0ν1/2

> 1026 y → Search needs ultra-low background, large mass, and long

counting time. Good energy resolution helps distinguish from 2νββ-decay.

1/20ν 28e.g. T ∼ 10 y would give signal of ∼ 0.5

counts/ton/yr

P. Barton – 5CoSSURF

Ton-Scale Neutrinoless Double Beta Decay

(Out of 4 Ranked Recommendations)

P. Barton – 6CoSSURF

Next Generation 76Ge: LEGEND

Large Enriched Germanium Experiment for Neutrinoless ββ Decay

Mission: “The collaboration aims to develop a phased, Ge-76 based

double-beta decay experimental program with discovery potential at a half-life beyond

1028 years, using existing resources as appropriate to expedite physics results.”

Select best technologies, based on what has been learned from GERDA and the

Majorana Demonstrator, as well as contributions from other groups and experiments.

First phase:

• (Up to) 200 kg.

• Modification of existing

GERDA infrastructure at

LNGS.

• BG goal (×5 lower)

0.6 c/(FWHM·t·y)

• Start by 2021.

Subsequent stages:

• 1000 kg (staged).

• Timeline connected to U.S. DOE

down select process.

• BG goal (×30 lower)

0.1 c/(FWHM·t·y)

• Location: TBD.

Required depth (Ge-77m) under

investigation.

P. Barton – 7CoSSURF

The MAJORANA Demonstrator

Requires very low background front end electronics! (< 1 count in ROI per ton, per year!)

P. Barton – 8CoSSURF

Coaxial

Vb

C ~ 20 pF

Small Electrode

C ~ 1 pF

Vb

Detector Geometries

P. Barton – 9CoSSURF

Germanium Point Contact Detector

HV Bias

P. Luke, “Low Capacitance Large Volume Shaped-field Germanium Detector,” IEEE TNS 36 (1989) 926.

Point ContactElectrode

(~mm)

• Originally proposed as dark matter detector

• Discrimination of multi-site events

• Low capacitance → low noise

• Lower trapping from p-type (than n-type),hence the LBNL P-Type Point Contact (PPC) detector

HPGe

similar rise time, different drift timesMulti-Site EventIdentification

Lithium drifted contact(hole barrier)

P. Barton – 10CoSSURF

Resistive Feedback Pulsed Reset

Pro

- Simple

Con

- Johnson noise

- Excess noise

- Resistor mass/radioactivity

- Energy-rate limitation

- Pole-zero difficulties

Pro

- No feedback resistor

Con

- Reset crosstalk (multi-channel system)

- Dead time during reset

- Added input capacitance

- Added complexity

a-Ge

resistor

Increase Rf

Det. Det.

Preamplifier Configuration

P. Barton – 11CoSSURF

Resistive Feedback JFET Front End

Junction Field Effect Transistor (JFET) – a Voltage Controlled Resistor (tens of ohms)

HVgate

drain

source

feedback

pulser

Front End ElectronicsPPC

Moxtek JFET

example operating point

VdsVgs

P. Barton – 12CoSSURF

voltage current

1/f

Voltage (Series) Noise

ENC ~ Vn,FET (CFET + CDet + CF + Cstray + …)

Vn, FET ~ CFET-1/2

→ Min. noise: CFET ~ (CDet + CF + Cstray)

Current (Parallel) Noise

ENC ~ (2qIL + 4kT/Rp)1/2

IL – full shot-noise leakage current

1/f Noise

Lossy dielectric, trapping-detrapping, ...

Det. shaper

Electronic Noise & Equivalent Noise Charge (ENC)

P. Barton – 13CoSSURF

Mini-PPC

Low Mass Front End Test Cryostat

LMFE Test with Mini-PPC

Optimization of JFET operating temperature with thermally isolating glass PCB

P. Barton – 14CoSSURF

NoDetector

(preamp only)

LMFE Minimum Noise

P. Barton – 15CoSSURF

LMFE Minimum Noise

preamp & detector

P. Barton – 16CoSSURF

Am-241

Mini-PPC Spectral Performance

P. Barton – 17CoSSURF

BEGe Noise with Moxtek (MX11) JFET

150 eV-FWHM limited primarily by capacitance of crystal

…but OK for some large physics experiments →

~1 kg

P. Barton – 18CoSSURF

LBNL Low Mass Front End (for MAJORANA)

Front End Preamplifier

Cry

ost

at

P. Barton – 19CoSSURF

LBNL Low Mass Front End

Output

Front End PreamplifierLow Mass Front End

Cry

ost

at

200 µm thick Fused Silica

SputteredTi / Au Traces200A / 4000A

7 mm

~80 K

~200 K

• JFET experiences Joule heating from drain current

• Board thermally separated from LN temperatures

P. Barton – 20CoSSURF

LBNL Low Mass Front End

Output

Front End PreamplifierLow Mass Front End

Cry

ost

atEpoxiedMOXTEKMX-11RC Al(1% Si)

1 mil Wire Bonds

TRA-DUCT 2902 Silver EpoxyRoom temp cure, low outgassing0.001 ohm-cmCTE = 0.6e-6 cm/cm/K3.0 W/m/K

MOXTEK JFETMX-11RC (reduced capacitance)designed for x-ray detectorsCgs = 0.7 pFgm = 8en

2 = 1.5 nV/√Hz

DrainSource

Gate

GD

S

P. Barton – 21CoSSURF

LBNL Low Mass Front End

Front End PreamplifierLow Mass Front End

Cry

ost

atCp = 45 fF

Cf = 176 fF

3D Electrostatic SimulationSet voltages, Look for charge

Potential Electric Field

A

Capacitance = V-1∫Surface charge density

1

0 0 0 0

Feedback

F

P

Pulser

A

A

P. Barton – 22CoSSURF

LBNL Low Mass Front End

Output

Front End PreamplifierLow Mass Front End

Cry

ost

at

Sputtered Germanium, LBNL-B774000A24 hr pumpdown, 2 mTorr, 120 W, 90 A/min

a-Ge at low temperatures = low noise

Rf

P. Barton – 23CoSSURF

LBNL Low Mass Front End

Output

Front End PreamplifierLow Mass Front End

Cry

ost

at

Simple cascode design a la F. Goulding (1967)

To Buffer /Drive Stage

P. Barton – 24CoSSURF

Low Mass (~Radio-Pure) Materials

Ge Al(Si)

Ti / Au Epoxy FEP / CuSi

SiO2

Ultrasonically DrilledHoles (Bullen, Inc.)for MAJORANA

Preliminary Assay results courtesy J. Loach

Budget < 4.1 cnt/ROI-ton-yr

Goal = 3 cnt/ROI-ton-yr

715 nBq / LMFE → 0.3 cnt/ROI-ton-yr

P. Barton – 25CoSSURF

Majorana LMFE Design Layout

Ti / Au = 11.76 mm2 (x 200A / 4000A)

Ge = 0.71 mm2 (x 4000A)

1. Revision2. Lot3. Wafer4. Board

1/20th mil (1.27 µm) Laser Transparency

Layout 10 boards per 2 inch wafer

P. Barton – 26CoSSURF

Majorana LMFE Fabrication

P. Barton – 27CoSSURF

Majorana LMFE Fabrication

P. Barton – 28CoSSURF

Majorana LMFE Fabrication

P. Barton – 29CoSSURF

Majorana LMFE Fabrication

P. Barton – 30CoSSURF

↳ T & humidity control↳ dry boxes to store producHon parts under

LN boil-‐off atmosphere

↳ ionizer + HEPA filter bench

LMFE Production Facility

Class-100 clean room (with ESD flooring) @ LBNL

P. Barton – 31CoSSURF

Loading Electroformed Copper Clips

P. Barton – 32CoSSURF

mechanical QA electrical QA

↳ pressure corresponding to 650g applied on board

↳ full signal path + preamp test↳ check baseline

↳ pulser check of 1st and 2nd stages

Production: QA

P. Barton – 33CoSSURF

Majorana LMFE Assembly at SURF

P. Barton – 34CoSSURF

MJD Readout Block Diagram

a

HV LV

Detectorx5

Front-Endx5

Preamplifierx5

Controller Card

Mother Board 1

Mother Board 2

Digitizer Cardsx2

DAQ

DAQ Slow Control

interlock

cryostat

Detectorx5

Front-Endx5

DAC

ADC ADC

DAC DAC

DAC

FPGA

SPI interface

1st stage

2nd stage

Vds

pulsers

Power + digitizer Back-end electronics Front-end electronics

Low-noise LVpower supply

Controller card

Mother board

Front-end board

Preamp.

WARM COLD

P. Barton – 36CoSSURF

The MAJORANA Demonstrator (prelude to ton-scale)

Located at the 4850’ levelof Sanford Underground Research Facilityin South Dakota

Low background passive Pb and Cu shieldwith active muon veto

Ultraclean electroformedcopper cryostat

30 kg of87% enriched 76Ge

10 kg ofnatural Ge

P. Barton – 37CoSSURF

Noise performance

P. Barton – 38CoSSURF

Spectroscopic performance

Energy Resolution (FWHM)

2.4 keV FWHM at Q=2039 keV

P. Barton – 39CoSSURF

Silica substrate (radio-pure, electrical, thermal, mechanical)

1-100 GΩ Amorphous-Ge Rf (radio-pure, low excess noise)

200 eV-FWHM noise with large PPC detectors (Majorana)

150 eV-FWHM noise with BEGe detector

85 eV-FWHM noise with mini-PPC detector (55 eV w/o detector)

Low Mass Front End

Can we lower the noise?

Thank You…

This work was funded by: DOE {NP, NNSA {DNN R&D, NSSC}}

I gratefully acknowledge the support of my colleagues: Kai Vetter, Paul Luke, Mark Amman, Alan Poon, et al.

P. Barton – 41CoSSURF

Equivalent Noise Charge

222 2 2 2 24 4

4 28

dn na f d e d na

m F

Ce kT kTENC Q e A C q I i

g R

= = + + + + +

100:140

1.2 pF

3.0 pA

16 G

d

d

F

T

C

I

R

=

=

=

=

Voltage (series) Current (parallel)1/fsimple CR-RC Shaper

e2/8

T

gm

ena2

Cd2, Cd

2

τ, τ

Af

Id

T

Rf

ina2

from shaping filter (affects V/I differently)

JFET temperature

JFET transconductance (~gain)

preamp / amp voltage noise

detector (and other) capacitance

shaping time constant

fabrication-dependent factor

detector leakage current

feedback resistor temperature

feedback resistance

preamp / amp current noise

▪

▪

▪

▪

▪

▪

Front-End Related

P. Barton – 42CoSSURF

The Shaping Amplifier and Ideal Filters

CRRC

CR RC4

Semi-Gaussian Shaper: (tau * s) / (1 + tau * s)n

e.g. Canberra 2026 Spectroscopy Amplifier

Shaping Time

Preamp Out

Filter

Amp Out

Judicious choice of shaping filter can(de)emphasize Series vs. Parallel noise

less series noise

Frequency (Hz)

Time (s)

Vo

ltag

e (m

V)

Mag

nit

ud

e

P. Barton – 43CoSSURF

JFET Heating Affects Feedback Resistance

Preamp Output

36.8 %

(time after preamp turn-on)

Convert decay time into Rf by:

tau = Rf * Cf,

where Cf is known and constant

176 fF → 8-17 GΩ ( ~20 K range)

P. Barton – 44CoSSURF

JFET Heating Fights Feedback Resistor

1 mA

7 mA

(12 electrons-rms)

2009 – Mini PPC

P. Barton – 45CoSSURF

MOSFET Alternative to JFET

Investigating cold (<77 K) CMOS options:

XGLab CUBE ASICEntire Preamp-on-a-chipOriginally for SDDFunctional at 4 K

Pulsed-Reset0.75 x 0.75 mm2

3.4 el.rms alone

4.4 electrons-rmsfor Silicon Drift Detector

would be 24 eV-FWHM in Ge,

for a safe threshold of 70 eV!

P. Barton – 46CoSSURF

Cold CMOS Schematic

Trade feedback resistorfor Reset transistor

Preamp-on-a-chip

HV pulser for C-V

CMOS improves below LN2

Cin

Voltage 1 / f Current

P. Barton – 47CoSSURF

Microphonics during averaged 0.609 Hz pulser (100 K)

Microphonics from LN2 Boiling

And mechanical cryocoolers have significantly more vibrations

P. Barton – 48CoSSURF

Ultra-Low Vibration GM-Cycle Cryostat

Test HPGe and electronics at 10 – 100 K with GM CryocoolerNational Nuclear Security Consortium (NSSC) Supported

Additional vibration isolation used in optical systems for nm-scale vibration

P. Barton – 49CoSSURF

V_IO

V_SSS

OUT

I_BiasRST

GNDPulser

CUBE ASIC Printed Circuit Board

Test Pulse Capacitor

CP = 10.3 fF simulated

= 9.3 fF measured

PCB: Rogers 4000 – Low Dielectric Loss (low leakage current)

Power supply filters (RC surface mount) for Vio and Vsss

Minimal test pulse capacitor

3D Electrostatic Model

P. Barton – 50CoSSURF

From Pin to Wirebond

Existing Mini-PPC Dimple Lapped Off

1. Reduce capacitance

2. Streamline assembly

3. Improve reliability

4. Reduce microphonics

5. Minimize masswirebonding by Mark Amman

1. Wet etch to remove damage

2. Sputter Al on Li

3. Sputter a-Si on top

4. Evaporate 0.75 mm Al forpoint contact

5. Wirebond

Reprocessing Steps

Advantages

P. Barton – 51CoSSURF

Detector Characteristics

Detector Capacitance = 0.26 pF

Leakage Current = 20 fF (at 30 K)

2 mil Al

1 mil Al (1% Si)

P. Barton – 52CoSSURF

Be Window

241Am

Spectral Performance

Front View

P. Barton – 53CoSSURF

Ultra-Low Noise Ge (Neutrino) Detector Collaboration – ULGeN

First: Mechanically Cooled, Wirebonded PPC HPGe, with CMOS Front End

Atmospheric Pressure He GasProvides ultra-low vibration thermal link using standard GM cycle (10 – 80 K)→ Eliminates all vibrations

Ultra-Low CapacitanceSmaller point contact (0.26 pF)enabled by wire bonding→ Ultra-Low Electronic Noise

Preamp-on-a-ChipCMOS ASIC for SDD4 electrons-rms noise→ Better than JFET

at low temperatures

Low temperature and ultra-low capacitance of CMOS and Ge.Result: lowest noise HPGe detector: 39 eV-FWHM at 43 K.

P. Barton – 54CoSSURF

A Common Theme

1. the ASIC!

2. die attach (e.g. epoxy)

3. substrate

4. traces

5. bond wires

6. bypass capacitor(s) + passives

7. passives’ attachment

8. connector

ASIC Front End Elements

Remember:• Electrical• Mechanical• Thermal• (Radiopurity)

P. Barton – 55CoSSURF

LBNL internal LDRD funding for a more suitable preamplifier ASIC

1. Optimized for operation in liquid argon

2. Optimized for large mass 1.5 kg HPGe detectors (1-3 pF?)

3. Single power supply (reducing the cabling needs)

4. Integrated feedback resistor (reduce external a-Ge)

5. Built-in power supply capacitor and filter (eliminate external cap)

6. Differential output for >10 m cable to DAQ

Coupled with silicon MEMS capacitors for off-chip PS decoupling

Silicon (radiopure) PCB, with wirebonds to detector

Next Steps for L-1000

Thank You…

This work was funded by: DOE {NP, NNSA {DNN R&D, NSSC}}

I gratefully acknowledge the support of my colleagues: Kai Vetter, Paul Luke, Mark Amman, Alan Poon, et al.

P. Barton – 57CoSSURF

A tale of two experiments

Majorana Demonstrator (SURF)

Detectors (29.7 kg enr76Ge) in 2 vacuum

cryostats.

Surrounded by passive shield. Ultra-

clean materials used.

GERDA (LNGS)

• Detectors (37.6 kg enr76Ge) in

liquid argon (LAr).

• LAr an active shield: Backgrounds

tagged with scintillation light.

detector arrays have demonstrated the lowest

backgrounds and best energy resolution of all 0νββ

experiment technologies.

P. Barton – 58CoSSURF

LEGEND Discovery Potential

10−3 10−2 10−1 10 102 1031Exposure [ton-years]

1024

1025

2610

1027

1028

1029

1030

T1

/2 3

DS

[years

]

IO mmin range

Background free

0.1 counts/FWHM-t-y

1.0 count/FWHM-t-y

10 counts/FWHM-t-y

76Ge (88% enr.)

MJD/GERDA

LEG

EN

D-

200

LEG

END

-1000

• Reduction of backgrounds is crucial to achieve good discovery potential.

• Majorana and GERDA technologies combined with new R&D to accomplish this.

P. Barton – 59CoSSURF

Broad LEGEND Methodology:

Use best radiopurity elements from Majorana

Use LAr shielding from GERDA

Ongoing Tasks

Test LMFE in LAr

Modify a-Ge resistor on LMFE

Choose JFET (MX11,SF291)

Modify Preamp (GERDA | Majorana)

Current Front End Steps for L-200

High Mass Front End surrogate w/ modified GERDA preamp

P. Barton – 60CoSSURF

LEGEND-200

• Initial 200 kg phase permits early science with a world-leading experiment.

• Exposure 1 ton·yr, sensitivity > 1027 yr

• Keeps young people involved and maintains skilled workers.

• Reduces risk for a future ton-scale phase. Reuse of GERDA infrastructure and

• Majorana/GERDA detectors (60 kg) permits possible data taking in 2021-2022.

• Room in GERDA cryostat for 200 kg of enriched detectors.

Jordan Myslik (LBNL) July 7,201860/ 14

water

(Ø 10m,

lock

Ge detector array

LAr veto

cryostat

(Ø 4m, 64m

PMT of muon

floor of clean

roof of clean

plastic muon

glove box

Required reduction in background for LEGEND-200 has already been demonstrated as feasible in the Majorana Demonstrator, GERDA, and dedicated test stands.