Bubble Chambers for Dark Matter Detection

Andrew Sonnenschein Fermilab

SLAC Advanced Instrumentation Seminar, March 5, 2008

Cartoon of a Galaxy

Invisible material of unknowncomposition

Stars and bright gas

>85% of total mass

“dark matter halo”

~ 200 kpc

Dark Matter From SupersymmetryStandard

ModelMSSM Higgs

Superpartnerflavor states

Superpartner mass states

u,d,s,c,t,b squarks

e, µ, τ sleptons

νe,νµ,ντ sneutrinos

W± charged winos χ±2

H± charged higgsinos χ±1

γ photino χ04

Z0 zino χ03

h h higgsino χ02

H higgsino χ01

A0

G gravitino

Charginos

Neutralinos

mixing

Lightest neutralino is dark matter?

Mass: 30-1000 GeV

Based on simple assumptions:• Particles are gravitationally bound to halo, with Maxwellian

velocity distribution (Vrms=220 Km/s) and local density 0.3 GeV/cm3

• WIMPs are heavy particles, 10 GeV< MWIMP< 1 TeV.• Nuclear scattering can efficiently transfer energy to a nucleus, since Mnucleus~Mwimp.

The signal will be a nuclear recoil, with energy ~10 keV

• Scattering is non-relativistic.

• Shape of spectrum does not depend on particle physics inputs.

• Amplitude of spectrum depends onunknown supersymmetry parametersand some astrophysical uncertainties.

Spectrum of WIMPs in a Detector on Earth

Germanium detector

Even

ts/k

eV-K

g-da

y

Energy of Nuclear Recoil [keV]

Even

t Rat

e [e

vent

s/keV

-Kg-

day]

Energy [keV] WIMP Mass [GeV/c2]

WIM

P-nu

cleo

n Cr

oss S

ectio

n [c

m2 ]

Excluded RegionLargest compatible Signal for 1 TeV

What Determines Sensitivity?

Assume halo density 0.3 GeV/cm3

• Construction of sensitivity plot is illustrated with data from an early experiment.

• Environmental radioactivity limits sensitivity.

Backgrounds from Radioactivity and Cosmic Rays

( figure from Brodzinski et al, Journal of Radioanalytical and Nuclear Chemistry, 193 (1) 1995 pp. 61-70)

Gammas & betasFrom primordial, cosmogenic, and manmadenuclei: (not an exhaustive list!)

238U, 232Th + daughters (incl. 222Rn)40K, 14C 85Kr, 137Cs, 3H - nuclear tests68Ge, 60Co - cosmogenic in detector setups

Cosmic Rays (p, π , µ, e…)Can be reduced by going underground.The µ’s penetrate to great depth.

NeutronsFrom µ spallation or (α, n) reactionsin rocks, with alphas from U/Th chains. Canbe shielded with moderator at low energies.

• A long history of successful attempts toreduce by choosing special materials andshielding.

40 keV Ar in 40 Torr Ar 13 keV e- in 40 Torr Ar

• WIMPs interact with the nucleus, while most backgrounds are due to electron scattering by gamma and beta rays.

• The resulting spatial distributions of energy and charge are very different-- this is fundamental physical basis of most discrimination techniques.

(Figures from DRIFT collaboration)

Discriminating Against Backgrounds

X (mm) X (mm)

Y (m

m)

Y (m

m)

Observables Which Could be Used to Separate WIMP-nucleus Events from Backgrounds

• Spatial distribution of charge deposited in a low-pressure drift chamber.

• Ratio of ionization to deposited energy in a cryogenic detector (CDMS).

• Pulse shape discrimination in scintillating materials (organic & inorganic).

• Ratio of ionization to scintillation in liquid noble gases.

• Ratio of scintillation to deposited energy in a cryogenic detector.

• Annual modulation due to motion of Earth around the Sun.

• Daily modulation in direction of ion tracks in a low-pressure drift chamber, due to rotation of the Earth.

• Efficiency for bubble formation in superheated liquids (bubble chambers).

Recoil Energy [keV]

Char

ge E

nerg

y [k

eV] Co-60 γ rays.

Cf-252 neutrons.

Background Discrimination With CDMS Cryogenic Detectors

+

γ Rays,neutrons,

&WIMPs

Semiconductor

-+

- -- -

- +++

phonons

electrons& holes

Temperature sensor

E

Charge collector

CDMS-I

CDMS-II Detectors

Direct detection of WIMP signalNuclei recoil by elastic scatteringRead both phonon signal (4 channels) and ionization

signal (inner and outer electrode)1cm thick, 7.8 cm diameter250g Germanium, 100g Silicon

Phonon signalFrom a quadrant

Charge signalFrom inner electrode

D

C

A

B

Phonon signal from one quadrant

CDMS-II results from DM2008, Marina Del Rey, Feb. 23, 2008

Why Bubble Chambers?1. Large target masses would be possible.

• Multi ton chambers were built in the 50’s- 80’s.

2. An exciting menu of available target nuclei. No liquid that has been tested seriously has failed to work as

a bubble chamber liquid (Glaser, 1960).

• Most common: Hydrogen, Propane

• But also “Heavy Liquids”: Xe, Ne, CF3Br, CH3I, and CCl2F2.

1. Good targets for both spin- dependent and spin-independent scattering.

• Possible to “swap” liquids to check suspicious signals.

• Low energy thresholds are easily obtained for nuclear recoils.

• < 10 keV easy to achieve according to standard nucleation theory.

• Backgrounds due to environmental gamma and beta activity can be suppressed by running at low pressure.

Why Do Liquids Not Always Boil When They Pass into the “Vapor” Part of the Phase Diagram?

• In the “vapor” region, the equilibrium state is a vapor.

• But liquids have surface tension, so there is an energy cost to create a bubble.

• This energy barrier may be greater than kt.

a metastable (“superheated”) liquid state may continue to exist for some time.

• The liquid will boil violently once the energy barrier to the vapor phase is overcome.

Bubble Nucleation by Radiation (Seitz, “Thermal Spike Model”, 1957)

Rc

Pvapor

Pexternalparticle

Surface tension σ

• Pressure inside bubble is equilibrium vapor pressure.• At critical radius Rc surface tension balances pressure.

• Bubbles bigger than the critical radius Rc will grow, while smaller bubbles will shrink to zero.• Boiling occurs when energy loss of throughgoing particle is enough to produce a bubble with radius > Rc

Rc

Ec

Radius

Work

�

RC = 2σ

vaporP − externalP

dE/dX Discrimination in Bubble Chambers• Energy to make bubble must be deposited inside critical radius, which depends on vapor pressure.• Since heavy particles have higher dE/dx, operating conditions can be chosen so that chamber is triggered by heavy nucleus recoils, but not electrons.

dE/dX Nucleation Threshold

1

10

100

1000

10000

0 2 4 6 8 10 12 14 16 18 20

Vapor Pressure - Operating Pressure [bar]

dE/

dX

[keV

/m

icro

n]

nuclear recoils

Electron recoils

The Problem with Classical Bubble Chambers• Classical bubble chambers were only sensitive for a few milliseconds

per cycle.

Example: SLAC 1-meter hydrogen chamber.

Ballam and Watt, 1977.

~ 10 msec

recompression decompression

Bubble Nucleation in Cracks

Liquid 0.1 µm

Solid

nucleation sites• Trapped gas volumes in surface imperfections are now known to be the primary source of nucleation.

• Most construction materials have rough surfaces at scales below 1 mm, but some materials, like glass, much better than others.

• Historically, problem was overcome for high energy physics experiments by rapid cycling of chamber in sync with a pulsed beam. Bubbling at walls was tolerated because of finite speed of bubble growth.

• A few small glass chambers (~10 ml) were built in the 50’s and 60’s, with sensitive times ~1 minute.

Ways to preserve superheated state:

• Elimination of porous surfaces in contact with superheated liquid.

• Precision cleaning to eliminate particulates.

• Vacuum degassing.

• … a few other tricks borrowed from chemical engineers

High-Stability Bubble Chamber, Waters et al., 1969.

IEEE Trans. Nuc. Sci. 16(1) 398-401 (1969)

Anti-coincidence counters

Quartz inner vessel (5-15 ml)

Heat exchanger

Pressure sensor

Expansion bellows Compressed gas line

filling volume control

Low temp bath inlet

Lucitecontainer

Cold volume

Hot volume

Plot of event rate vs. “superheat pressure” (= vapor pressure - operating pressure)

electrons

protons

α plateau

(psi)

Prototype Dark Matter Detector (2003)

3-way valve

Propylene glycol bufferliquid prevents evaporation of superheated liquid.

Acoustic sensor

pressuresensor

filter

gas

glycol

gas

Exhaust to Room

Compressed Air @ 140 psi

air

liquid

glycol

Quartz pressure vessel

Glass dewar with heat-exchange fluid

Piston

Camera(1 of 2)

Possible Target Liquids

Mass Fractions DensityBoilingPoint

@ 1 atmComments

CF3Br

8% C (Z=6)38% F (Z=9)54% Br (Z=35)

1.5 g/cc -58 °C

Good for spin-dependent and spin- independent couplings.

CF3I

6% C 29% F 65% I (Z=53)

2.1 g/cc -23 °C

Spin-dependent andspin-independent Non- ozone depleting

C3F8

19% C 81% F 1.4 g/

cc -37 °C Spin- dependent only.

Xe 100% Xe (Z=54) 3.0 g/cc -108 °C

Possible use in hybridscintillating bubble chamber.

High Speed Bubble Chamber Movie1000 frames/ second241Am-Be neutron source

High Speed Bubble Chamber Movie1000 frames/ second241Am-Be neutron source

Bad surface Good surface

Neutron Multiple Scattering

• Multiple bubbles are present in approximately 4% of events in our “background” data set.

• These events can only be caused by multiple neutron scattering, since uniform size of bubbles implies simultaneous nucleation at multiple sites.

• Events such as this can be used to measure neutron backgrounds in- situ while searching for recoils due to WIMPs.

Wilson Fellows: Andrew Sonnenschein

Staff Scientists: Peter Cooper Mike Crisler Martin Hu Erik Ramberg Bob Tschirhart

PrincipalInvestigator:JuanCollar (spokesperson)

GraduateStudents:NathanRileyMatthewSzydagis

Undergraduates:LukeGoetzkeHannesSchimmelpfennig

KICPFellows:BrianOdom

The COUPP Collaboration

University of Chicago National Science FoundationKavli Institute for Cosmological PhysicsDepartment of Energy

Funding

Fermi National Accelerator Laboratory

Principal Investigator: Ilan Levine

Undergraduates: Earl Neeley Tina Marie Shepherd

Engineers: Ed Behnke

Indiana University South Bend

test site~300 m.w.e.

at Fermilab

Neutron Flux Underground

Heusser, 1995.

NuMi Tunnel

U. Chicago Sub-basement lab

Dominant neutron sources:

• Direct cosmic ray neutrons.

• Neutrons from radioactivity in rock.

• Neutrons made by muons passing through high-Z materials around the detector.

Design Concept for Large Chambers• Central design issue is how to avoid metal contact with superheated liquid.

• Fabrication of large quartz or glass pressure vessels is not practical, but industrial capability exists for thin-walled vessels up to ~ 1 m3 in volume.

Thin- walled quartz bell jar

Steel pressure vessel

Pressure balancing bellows

SUPERHEATEDL IQU ID

VIEWPORT

VIEWPORT VIEWPORT

HEAT EXCHANGE FLUID

PISTON

Buffer fluid

Installation of 1 Liter Chamber At Fermilab• Prototypes design features required for chambers up to 1000 liters

160 msec of Video Buffer (20 msec/frame)

Muon Track @ 160 psi Superheat Pressure

Energy Calibration with Neutrons

Seitzbubble

nucleation model

Threshold

Gamma Calibration

Subtracted image

projection of subtracted image onto horizontal axis

projection with noise subtraction

Starting image Zoom In

Image Analysis Procedure

Spatial Distribution of Single BubblesWall Events: not a background, but they reduce our live time due to the need to decompress afterwards, prohibitive for larger chambers. ~ 300/day

Bulk events: indistinguishable from WIMP interactions on an event-by-event basis.

~ 20- 100 events/day

Alpha Particle Backgrounds• Alpha decay produces monoenergetic, low energy nuclear recoils.

For example, consider 210Po->206Pb:

206PbαEα = 5.407 MeV ER= 101 keV

• The recoiling nucleus will nucleate a bubble in any chamber that is sensitive to the lower energy (~10 keV) recoils expected from WIMP scattering.

• The 238U and 232Th decay series include many alpha emitters, including radon (222Rn) and its daughters.

• Radon is highly soluble in bubble chamber liquids.

• Solar neutrino experiments (Borexino, Kamland, SNO) have demonstrated feasibility of reduction to ~1 event per day in scintillator and water-- about 2 orders of magnitude lower rates than seen in current-generation dark matter experiments.

Table of Isotopes

Iodine

Rare earth group

Uranium and Thorium Decay Chains

Fluorine

222Rn

prot

ons

neutrons

betas

Bromine

226 RadiumT1/2=1600 y

222 RadonT1/2=3.8 d

218

PoloniumT1/2=3.1 m

214 LeadT1/2= 27 m

214

BismouthT1/2= 20 m

Present at significant levels in most natural and man-made materials

Noble gas- highly diffusive, universally present in air, water

Red=alphadecay

Blue=betadecay 214Poloniu

mT1/2=0.2 ms

210 LeadT1/2= 22 y

210

BismouthT1/2= 5 d

210

PoloniumT1/2= 138 d 206 Lead

Stable

Long half-life, implanted in surfaces by 100 keV recoil from decay of parent

Hard to clean off!

Attracted to surfaces by electrostatic charge

Radium Decay Chain: Dominant Source of Environmental α’s

Radium Decay Chain: Dominant Source of Environmental α’s

226 RadiumT1/2=1600 y

222 RadonT1/2=3.8 d

218

PoloniumT1/2=3.1 m

214 LeadT1/2= 27 m

214

BismouthT1/2= 20 m

Present at significant levels in most natural and man-made materials

Noble gas- highly diffusive, universally present in air, water

Red:alphadecay

Blue:betadecay

214Polonium

T1/2=0.2 ms

210 LeadT1/2= 22 y

210

BismouthT1/2= 5 d

210

PoloniumT1/2= 138 d 206 Lead

Stable

Long half-life, implanted in surfaces by 100 keV recoil from decay of parent

Hard to clean off!

WALL?

BULK

BULK

O-rings

metalbellows

quartz vessel

water volume

sensitive volume(CF3I)

VitonrubberO-ring

radon

Radon Emanation from Viton Rubber Seals

• Borexino collaboration has measured Viton to release radon at a rate of 15 atoms/cm2-day

=> Our O-rings should each emit 50 decays/day of radon

• Expect a total of 300 alpha decays per day in inner vessel, accounting for 2 O-rings and 3 alpha emitters in the decay chain.

• We observe only a fraction of this decay rate in the sensitive volume (20- 100 per day). Decays in water are undetectable.

• Variation in counting rate believed due to changes in efficiency of radon transport with temperature and operating conditions.

New Low-Background Vessel

Water

CF3I

• Teflon-coated inconel seals replace Viton.Radon emanation <1% of previous design

• Quartz etched with HF acid by vendor to remove implanted radon daughters.

• All parts cleaned and assembled in Accelerator Division RF cavity cleaning facility.

• Very high purity, well-studied water from Sudbury Neutrino Observatory.

• Bellows made using non-thoriated welding electrodes.

New Run of 1-liter Chamber

•NewrunstartedJuly30,2007.•Someradonintroducedduringfilling,nowdecayedtoequilibrium.

Muon Veto System • An array of counters is used to tag events with an associated muon.

• This system will extend sensitivity to below 1 event/kg-day at Fermilab site.

Muon and Beam Coincident Events

NUMI Veto

Acoustic

Pressure

-3000 -2000 -1000 0 1000 2000 3000 Time (microseconds)

• There are ~ 10 events a day coincident with NuMi neutrino beam and muon veto hit.

Neutrinos -> Muons -> Neutrons-> nuclear recoils

Precision Cleaning at Fermilab• Accelerator Division has superb cleaning facilities developed for RF cavities: Class 10 assembly area, large ultrasound machines, high pressure rinse machine.• Using SNO- recommended special detergent formulation to remove non-implanted fraction of 210Pb and 210Po surface contamination. • High pressure rinse to remove dust -- Goal is 1 microgram per liter for 0.01 bubbles/day.

Inner vessel high pressurerinse test

Data from 2006 Run• Datafrompressurescanattwotemperatures.• Fittoalphas+WIMPs

Solid lines:Expected WIMP

response for σSD(p)=3 pb

Radon background

Energy Threshold In KeV

Results from 2006 Run• Wehavecompetitivesensitivityforspin-dependentscattering,despitehighradonbackground

• Nowpublished,Science,319:933-936(2008).Spin-dependent Spin-independent

COUPP - 60 Kg Bubble Chamber

Engineering LayoutSolid Model Assembly Drawing

112”

84” 12”

60 Kg Chamber

20 Kg Chamber

Camera array

UC HEP seminar J.I. Collar Feb. 25 UC HEP seminar J.I. Collar Feb. 25

2 weeks ago, 20 kg installation

16 of 17

Summary:

• • First results from 2-kg chamber now published.

Very competitive spin-dependent sensitivity.

• Radon problem seems to be solved.

• 20 Kg and 60 Kg chambers in advanced stages of construction, will come on line in next year.

• Approaching limits of our shallow underground site.