Background Reduction in Cryogenic Detectors

Dan Bauer, Fermilab

LRT2004, Sudbury, December 13, 2004

Detector

Shielding

Veto

U/Th/K/Rn

,nU/Th/K/Rn

x

RockRock

n

LRT 2004 Dan Bauer

Cryogenic Dark Matter Search - CDMS

•Dark Matter Search Goal is direct detection of a few

WIMPS/year• Signature is nuclear recoil with

E<100 KeV

•Cryogenic Cool very pure Ge and Si crystals

to < 50 mK

•Active Background Rejection Detect both heat (phonons) and

charge• Nuclear recoils produce less

charge for the same heat as electron recoils

•Deep Underground (Soudan)• Fewer cosmic rays to produce

neutrons • Neutrons produce nuclear recoils

Detector Tower

DilutionRefrigerator

Shield/Muon Veto

Electronics and Data Acquisition

•Shielding (Pb, polyethylene, Cu)Reduce backgrounds from radioactivityActive scintillator veto against cosmic rays

LRT 2004 Dan Bauer

CDMS Background Rejection Strategy

Detector Rejection of Backgrounds

y

x

Phon

on ti

min

g

Y = Charge/phonons

Y =

Cha

rge/

phon

ons

Erecoil (keV)

gamma cal.

Charge yield: , Phonon timing: surface events ()

Multiple-scatters: n(also Si vs Ge rates) Position information: locate discrete sources

LRT 2004 Dan Bauer

CDMS Background Reduction Strategy

Layered shielding (reduce , , neutrons)~1 cm Cu walls of cold volume (cleanest material)Thin “mu-metal” magnetic shield (for SQUIDs)10 cm inner polyethylene (further neutron moderation)22.5 cm Pb, inner 5 cm is “ancient” (low in 210Pb)40 cm outer polyethylene (main neutron moderator)All materials near detectors screened for U/Th/K

Active Veto (reject events associated with cosmics)Hermetic, 2” thick plastic scintillator veto wrapped around shieldReject residual cosmic-ray induced eventsInformation stored as time history before detector triggersExpect > 99.99% efficiency for all , > 99% for interacting MC indicates > 60% efficiency for -induced showers from rock

LRT 2004 Dan Bauer

The Radon Problem• Radon levels high, vary seasonally at Soudan (200-700 Bq/m^3)

Decays include energetic gammas which can penetrate to detectors, and eject betas from Compton scatters (‘ejectrons’)

Need to displace Radon from region inside Pb shield Six purge tubes along stem shield penetrations

• Purge gas is medical grade breathing air ‘aged’ in metal cylinders for at least 2 weeks to allow decay of 90% of 222Rn

Radon variation at Soudan

0

100

200

300

400

500

600

700

800

Jun-01 Jan-02 Jul-02 Feb-03 Aug-03 Mar-04 Oct-04 Apr-05

LRT 2004 Dan Bauer

Measured Gamma Backgrounds•Typically “bulk” events

High ionization yield in detector bulk

Rejection 99.9999% at 70% nuclear recoil efficiency

•Sources Residual contamination in the

Pb, polyethylene and copper Environmental radon

• Three event classes Compton scatters from nearby

passive materials have low solid-angle for hitting detectors

Compton scatters from nearest neighbor can be vetoed

Dominant component is 1 in ~30000 gammas interacting in dead layer: expect <0.1 events in CDMSII (after timing cuts)

• Comparison of data and MC: Gammas from U/Th/K in Pb, Poly, Cu at

assayed level Radon between purged volume & Pb

• Fit concentration to data in summed spectra• 35 Bq/m3 compared with ambient ~500 Bq/m3

Fair agreement but actual radon level may be slightly lower based on:

• 609 keV 214-Bi line lower in data• 1765 keV 214-Bi line agrees

L. Baudis, UFL

U/Th/K: ~1/4 total rate

Radon: fit to data

LRT 2004 Dan Bauer

Measured Beta Backgrounds

• Typically surface events: rejected at 99.4% in present analysis

Timing 97% Ionization yield 80%

• Sources Residual contamination on detector and

nearby surfaces: “intrinsic” betas Soft x-rays Pb-210, K-40, C-14 primary focus

• Identification in situ direct counting

• Correlate with gammas and alphas surface science techniques

• Auger, SIMS, RBS+PIXE

• Rates Observe ~0.4/det/day on inner detectors Expect ~7 Events in CDMS-II for present

analysis and rate Modest improvements will keep us background free

• Important to ID and characterize these backgrounds for CDMSII

Robust leakage estimates• Convolve source spectrum in Monte

Carlo to model charge collection• Confirm with calibration/TF data

— Charge side— Phonon side

Depth (um)

Cha

rge

Eff

icie

ncy

J.-P. Thompson, Brown

LRT 2004 Dan Bauer

Sources of residual beta background

• Pb-210 — from airborne radon daughters Could be dominant source — further analysis needed Complex decay chain with numerous alphas and betas expect and observe

roughly equal numbers• Detailed simulations to check relative detection efficiency in progress

char

ge

Recoil Energy (keV) Recoil Energy (keV)

Eve

nts

J. Cooley-Sekula, Stanford

LRT 2004 Dan Bauer

Sources of residual beta background

• K-40 — from natural potassium Direct upper limit less than half observed rate 1460 keV gamma: lack of observed photopeak or compton edge sets

upper limit of 0.15 betas/det/day RBS+PIXE surface probe for natK and assumption that 40K is in standard

cosmogenic abundance limits rate to 0.04 betas/det/day

• C-14 — from natural carbon Auger spectroscopy and RBS indicate 2-3 monolayers of “adventitious”

carbon 0.3 betas/det/day to 156-keV endpoint 0.05 betas/det/day in 15-45

keV

• Work is ongoing Complete Pb-210 analysis Broaden scope to more possible isotopes Just beginning use of new technique: ICP-MS

• Inductively coupled plasma mass spectroscopy• Antimony found on test wafer - normalization not known yet

R. Schnee, D. Grant, Case; P. Cushman, A. Reisetter, U Minn

LRT 2004 Dan Bauer

Reduction of EM Backgrounds

• Reduce beta contamination via active screening/cleaning Observed alpha rate indicates dominated by 210Pb on detectors

• Improved radon purge should help, if this is correct Materials surface analysis (PIXE/RBS/SIMS/Auger) (in progress)

• Try to pinpoint source(s) of beta contamination Developing multiwire proportional chamber or cloud chamber as

dedicated alpha/beta screener (Tom Shutt talk)• Necessary for 17 beta emitters that have no screenable

gammas/alphas

• Reduce photon background via improved shielding Active (inexpensive) ionization

“endcap” detectors to shield against betas, identify multiple-scatters

Add inner ‘clean’ Pb shielding Improved gamma screening (Rick

Gaitskell talk)

LRT 2004 Dan Bauer

Neutron Backgrounds

• Predictions based on neutron propagation from rock and shield, normalized to Soudan muon flux

Expected <0.05 unvetoed neutrons in first data set - none observed Expected 1.9 vetoed neutrons - none observed (agrees at 85% CL)

• Should see ~ 5 vetoed neutrons in second data set Will allow normalization of Monte Carlos Observe one muon-coincident multiple-scatter nuclear recoil so far

• Ongoing work to refine estimates Direct measure of muon flux from veto Throw primary muon spectrum in Fluka + Geant4

• Hadron production• Correlations of particles from same parent muon• Simulate vetoed fraction of externally produced events• Predict 60% of “punch through” (>50 MeV) are vetoed by outer scintillator

Expect <0.2 unvetoed neutrons in full CDMS-II exposure• Will reach ‘natural’ neutron background limit at Soudan in a few years

S. Kamat, R. Hennings-Yeomans, Case; A. Reisetter, U Minn; J. Sander, H. Nelson, UCSB

LRT 2004 Dan Bauer

Neutron Reduction Strategies

Depth (meters water equivalent)

Mu

on

Flu

x (

m-2s-

1)

Super CDMS @ SNOLABAvoid the problem by reducing muon flux by

500x

CDMS II @ SoudanCould add inner neutron veto

LRT 2004 Dan Bauer

CDMS GoalMaintain Zero Background as MT increases

CDMS II Goal 1998

Tower 1: Fall 03Expected CDMSII end 2005

Expected Tower 1+2 Summer 04

Zero background 58% efficiency

Blue points illustrate random fluctuation from experiment to experiment

04/04/14

Currently 45% Z 2,3,5 > 10keV90% CL upper limit 0.005

Improvement linear until background events appearThen degrades as √MT until systematics dominate