Search for Cosmic Dark Matter at CDMS

Laura Baudis

Stanford University

SLAC Topical Conference, August 16, 02

What is our Universe made of?

We live in a very interesting Universe!

matter 1/3

energy 2/3

ordinary matter : 5%

cold dark matter : 30%

cosmological constant : 65%

apparently flat universe : T = M + = 1

data from 26 experiments(including VSA and CBI)

Evidence for a flat Universe

large angular scales: ~ 1/l

Tegmark & Zaldarriaga 2002

€

Ωo ≈1

l ≈200/ Ω0

CMB+ IRAS PCSz power spectrum+ Hubble param. prior h=0.7

dm = 0.24+/-0.06b = 0.04+/-0.02

wang,tegmark,zaldarriaga,astro-ph/0105091

m = 0.29 +/- 0.04b = 0.04 +/- 0.02

Galaxy clustering: the 2dF galaxy redshift survey > 160 000 galaxies

percival et al., ,astro-ph/0105252: likelihood surfacesfor the best fit linear power spectrum

The matter budget

Cold dark matter candidates

axions strong CP, m~10-5eVWIMPs SUSY, m~ GeV

were NOT invented to solve the dark matter problem!

exotic: primordial black holes superheavy dark matter (WIMP/SIMPzillas)...

Weakly interacting massive particles

long lived or stable particles left over from the BB

€

Ωχ ∝1

σAv

σA ≈σweak

coincidence or big hint???

actual abundance:annihilation <

equilibrium abundance:

l l~ ~

Searching for WIMPs

WIMP searches fall into 3 main categories

WIMP production at accelerators

Indirect detection via :• ’s & cosmic rays due to WIMP annihilation in the halo• ‘s from WIMP annihilation in the Solar/Earth cores

Direct detection experiments• Recoils due to WIMPs elastically scattering from nuclei

0

measure the energy deposited by the recoiling nucleusin a terrestrial, low background detector

v/c 10-3

ER

0

What do we need to know?

Particle physics: a good candidate MW, el

Astrophysics: density of WIMPs in the halo velocity distribution f(v)

Nuclear physics: nuclear form factor F(q)

Detector physics: energy threshold ET, resolution ionization efficiency, discrimination,...

Input from several fields required to estimate the event rates in a direct detection experiment :

Favourite WIMP candidate: neutralino

if SUSY exists and R-parity (-1)3(B-L)+2S is conserved

=> LSP is stable: potential DM candidate!

€

χ =α˜ γ +β˜ Z +γ˜ H 10 +δ ˜ H 2

0

in general: mixture of photino, zino and higgsinos

prediction of masses, scattering cross sectionselastic nucleus cross section dominated by SI part

Study the low energy SUSY theories which arise from GUT, supergravity or string theories, reduce > 100 MSSM parameters to 5-7...

When masses and couplings fixed: calculate the WIMP-nucleus cross section (from -quark cross section, QCD, nuclear physics...)

Theorists survey a large set of models with masses and couplings within a plausible range; impose laboratory and relic density constraintsplots of elastic scattering cross sections versus neutralino mass

in general: : 10-5 and 10-11 pb

sensitivity of current experiments: ~ 10-6 pb

WIMP nucleus cross section

Local dark matter density

measured rotation curve of Milky Way: flat out to 50 kpc

vcirc ~ 220 km/s; assumption: halo is spherical

0 ~ 0.3 - 0.5 GeV/cm3, v ~ 220 km/s, MB-distrib.

R

220 km/s

R

Differential rate for WIMP elastic scattering

€

dR

dER=

σ 0ρ 0

π v0mχmr2F 2(Q)T(Q)

F 2(Q) =3 j1(qR1)

qR1

⎡

⎣ ⎢

⎤

⎦ ⎥

2

exp − qs( )2

( ),T(Q) = exp −vmin2 /v0

2( ) :MB

vmin =ERmN2mr

2;v0 = 220km /s

σ 0 =mrmr− p

⎛

⎝ ⎜ ⎜

⎞

⎠ ⎟ ⎟

2

A2σ χ − p

mr =mχmNmχ + mN

;mr− p =mχmp

mχ + mp

spin-independent interaction

form factor velocity distribution

An example....

m ~ 100 GeV, 0=10-40 -10-44 cm2

0= 0.4 GeV/cm3, v = 220 km/s

rate of scattering from Ge (mN ~73 GeV)

R < 1 event/kg keV d

ER max= 2mr2v2/mN

< 250 keV

Detectors:

low threshold

low background

large masses

100 GeV WIMP

Ge

Comparing target nuclei

Selected Target Nuclei

Nucleus A

Ionization/

scintillationPhonons

Si 28 .3 1

Ge 73 .3 1

I 127 0.08 --

Xe 131 .2 --

Elastic scattering rates

~ mN4 favors heavier nuclei

but - form factor suppression

favors lighter nuclei

integrated visible rate:

threshold as low as possible!

The CDMS experiment

CDMS collaboration

Case Western Reserve UniversityD.S. Akerib,D. Driscoll, S. Kamat,

T.A. Perera, R.W. Schnee, G.WangFermi National Accelerator Laboratory

M.B. Crisler, R. Dixon, D. Holmgren

Lawrence Berkeley National LabR.J. McDonald, R.R. RossA. Smith

Nat’l Institute of Standards & Tech.J. Martinis

Princeton UniversityT. Shutt

Santa Clara UniversityB.A. Young

Stanford University L. Baudis, P.L. Brink,

B. Cabrera, C. Chang, T. SaabUniversity of California, Berkeley

S. Armel, V. Mandic, P. Meunier, W. Rau, B. Sadoulet

University of California, Santa BarbaraD.A. Bauer, R. Bunker, D.O. Caldwell, C. Maloney, H. Nelson, J. Sander, S. Yellin

University of Colorado at DenverM. E. Huber

Brown UniversityR.J. Gaitskell, J.P. Thomson

Ultra-pure Si and Ge crystals:

1cm thick; 7.5cm diameter.

measure phonons and ionization

signals after an interaction

discrimination between nuclear

and electron recoils

nuclear recoils: WIMPs, n

electron recoils: e

CDMS detectors

gamma source

neutron source

Ionization threshold

ZIP: advanced athermal phonon detectors

1

Qinner

Qouter

A

B

D

C

Rbias

Ibias

SQUID array Phonon D

Rfeedback

Vqbias

Z - dependent ionization and phonon detectorssuperconducting thin films of W/Al

Phonon signalInteraction creates THz (~ 4meV) phonons

Phonons propagate to SC Al-fins on the

surface, break Cooper pairs and create

quasiparticles

Quasiparticles diffuse in 10 s through

the Al-fins and are trapped in the W

transition-edge sensors (TES)

where they release their binding

energy to the W electrons

The electron system T is raised

increased R

The TES is voltage biased and operated

in the Electro-Thermal Feedback (ETF)

mode,PJ = VB2/R:

when R increases, I decreases

Current change is measured by SQUIDs

Al

Si or Ge

W

qp-trapAl Collector

W TESqp diffusion

R

T

Electro-Thermal Feedback

Ionization signal

An interaction breaks up the electron-hole pairs in the crystal

An electric field through the crystal separates the electrons and holes

The charge is collected by electrodes on the surface of the crystal

Two charge channels:

Main electrode: a disk in the center of the crystal surface.

Second electrode: a ring at the edge of the crystal surface.

TES sideShared Event Fully contained Qin event

Vbias

Readout

electrons

holes+ ++

+--

--

+ +++ --

--

Position Measurement

Ionization signal is ~ instantaneous: measure of time

Speed of sound in Si (Ge) crystal of ~ 1 (0.5) cm/s results in measurable

delays between the pulses of the 4 phonon channels

=> (x,y) position of the interaction (x= delay(A) - delay(D), y = delay(B) - delay(A))

A

B C

D

many-hole collimated, large surface Cd-109 source

Particle identificationDifferent types of interactions in crystal => different signatures in the phonon and charge signals

gammas

neutrons

betas

Risetime of the phonon pulses => information about the depth of the event position => discriminate against betas (surface events)

surf

ace

bu

lk

neutrons

gammas

electrons

How do we identify neutrons?

WIMPs: Ge has ~6x higher interaction rate per kg than Si

neutrons: Si has ~2x higher interaction rate per kg than Ge

WIMPS 40 GeV neutrons

1. use Si and Ge detectors:

2. look at multiple scattered events: measure the neutron background

Active Muon Veto

Inner Pb shieldPolyethylene

Pb Shield

Stanford Underground Facility17 mwe of rockhadronic component: >1000 muon flux: ~5

Active Scintillator Muon Vetomuon veto >99.9% efficientrejects “internal” neutrons produced by muons within shield

Low-Background Environment15 cm Pb shield: flux>100025 cm PE: -induced n-flux >100radiopure cold volume (10 kg)43 kg additional internal (ancient) lead shielding11 kg internal PE shielding

Current location: SUF

detectorsmuon veto

Stanford Underground Facility

CDMS operations since fall of 1996, Runs 13-21

Data acquisition system Icebox and shielding

SUF icebox and towerSQUET card

Tower

The Towerprovides support for the ZIPdetectors and their low T readout electronics

spans various thermal stages in the fridge (4K --> 10mK)

Phonon readout (SQUIDs) @ 600 mKIonization readout (FETs) @ 4K6 ZIP detectors stacked vertically, 2 mm separation between each pair.

Current CDMS run @ SUF: since July 01

First physics run with 6 ZIP tower

and last run @ SUF: > 100 livedays

6 new ZIP detectors

(4x 250 g Ge, 2x 100 g Si)

lower n background due to internal PE

and better discrimination against

Goal:

Improved physics results: 3x better WIMP sensitivity

Performance test of Soudan Tower 1

(-> Soudan end 2002)

GeGeGe

GeSi

Si

SQUID cards

FET cards

4K

0.6 K

0.06 K0.02 K

Baseline performance

Observed noise spectra for all 4 phonon channels and both charge channels of one Si detector

As expected,70-100 nV/Hz

As expected,10 pA/Hz

Ionization signal

• rise ~ 1s, fall = 40s

•Resolution < 1 keV FWHM

Phonon signal

• rise ~ 10s, fall ~ 400s

• Resolution ~ 400 eV FWHM

Ge ZIP

Recoil Energy Threshold

Sensitivities of direct detection searches depend strongly on Eth

Threshold energy depends, in turn, on the detector energy resolution

ZIP detectors able to achieve 5 keV (recoil energy) energy threshold

Ge ZIPs Si ZIPs

•137Cs calibration source provides a gamma line at 662 keV

•71Ga decay provides an x-ray line at 10.36 keV in the Ge detectors

Lines/features to calibrate the charge and phonon responses

•137Cs calibration source provides a gamma line at 662 keV

•73mGe provides a gamma line at 66.7 keV in the Ge detectors

Calibrating the Energy Scale

•73mGe provides a gamma line at 66.7 keV in the Ge detectors

•71Ga decay provides an x-ray line at 10.36 keV in the Ge detectors

= 10.4 keV = 0.34 keV

2D Plots with gammas and neutrons

Response of the detectors to calibration gamma and neutron sources

Si ZIP Ge ZIP

Y plots with gammas & neutrons

Discrimination parameter Y (=Eionization/Erecoil) black lines: +/- 2 bounds of the electron/nuclear recoil band

Si ZIP Ge ZIP

Effectiveness of the discrimination parameter

Histogram of yield parameter as a function of energy for gammas

>99.99% of ‘s rejected between 5-100 keV (@ 90% CL)

> 99.8% rejection in 5-10 keV bin

No gamma events leaked into the nuclear recoil band : CL limited by size of calibration data set

Vertical black lines: +/- 2 bounds of the nuclear recoil band

Ge ZIP

Background data

Yield plots for background data from the current run (muon coincident!):gamma background band clearly visiblemuon coincident neutrons populate the nuclear recoil band

Si ZIP Ge ZIP

Gamma background ratesGamma background rate ~ 1 evt/keV/kg/day in Ge and ~ 3 evt/keV/kg/day in SiWith a discrimination ability of > 99.8% the gamma background is reduced to < 2 x 10-3 evt/keV/kg/day

Si ZIP Ge ZIP

Current and projected limits

Current CDMS SUF Limit(PRL 84, 2000)

Projected CDMS SUF Limit

Projected CDMS Soudan Limit

Edelweiss 2002

CDMS 2000 + Edelweiss 2002already rule out DAMA!

Location: 2000 mwe

muon flux suppressed by factor 104

neutron flux suppressed by ~ 300

expected sensitivity 1 year ~ 0.07 evt/kg/day

final expected sensitivity ~ 0.01 evt/kg/day

Identical Icebox: 7 towers each with 3 Ge & 3 Si ZIPs (Mtot > 7kg)

CDMSII @ Soudan: end 2002

CDMS experimental enclosure

CDMS II icebox and shield

Goal of CDMS @ Soudan

CDMS II: 0.01 events/kg/d, (MT)eff=230kg.d

CRESST II

CDMS II

MSSM models

DAMA3 claim

‘g-2’ constraint

New results with improved stats (hep-ex/0208001)~ 2.6 SM deviation:M < 450-650 GeV < 10-10 pb(Baltz&Gondolo, astro-ph/0207673)

WIMPs are excellent candidates for dark matter their discovery would have deep implications for

cosmology and particle physics !

CDMS well suited to search for WIMPs

1999 results + 2002 EDELWEISS results: incompatible with DAMA

(independent of halo model, Copi & Krauss 2002)

last run @SUF started July01-July02, > 100 livedays (48 kg d)

6 ZIP tower, excellent background discrimination: final results soon!

sensitivity @ SUF limited by external n-background

fridge is being commissioned at Soudan mine

first tower in Soudan: late 2002

reach 100 times better sensitivity ~1 event / kg / year

Conclusions

‘The constitution of the universe may be set in first place among all natural things that can be known. For coming before all others in grandeur by reason of its universal content, it must also stand above them all in nobility as their rule and standard.’

Galileo Galilei, Dialogue

more slides.....

CDMS Spectrum and DAMA Predictions

Observed CDMS Nuclear Recoils versus expected WIMP Spectra

WIMP spectra corrected for CDMS eff. function

DAMA NaI/1-4 (3)

DAMA NaI/1-4 (3 CL) contour lowest cross-section(40 GeV, 2.3 10-42 cm2)

DAMA NaI/1-4 best-fit WIMP(52 GeV, 7.2 10-42 cm2)

~2 event/kgGe/day

Exp

ecte

d W

IMP

s D

etec

ted

Recoil Energy (keV)

CorrectionsLarge Tc non-uniformity can

distort the phonon signals

Use position and risetime

information to correct for this

non-uniformity

Z1 in Run2: 2-step correction:

1) First correct the phonon energy based on risetime - for good detectors, the yield (ratio of charge and phonon signals) is not correlated with the risetime (i.e. the blob is vertical)

2) Then correct the phonon energy based on event position

Not corrected

Corrected

(s)

Ch

ar g

e E

ne

rgy

(ke

V)

Phonon energy (keV) Phonon Energy (keV)

Annual Modulation of Rate & Spectrum

galactic center

v0

Sun 230 km/s

Earth 30 km/s (15 km/s in galactic plane)

log

dN

/dE

reco

ilErecoil

June

Dec

~5% effect

€

Combining earth and solar system motion around galaxy:

T Q( ) =π v0

4veerf

vmin + vev0

⎛

⎝ ⎜

⎞

⎠ ⎟− erf

vmin − vev0

⎛

⎝ ⎜

⎞

⎠ ⎟

⎡

⎣ ⎢

⎤

⎦ ⎥

where ve = v0 1.05 + 0.07cos2π t − t p( )

1 yr

⎛

⎝ ⎜ ⎜

⎞

⎠ ⎟ ⎟

⎡

⎣ ⎢ ⎢

⎤

⎦ ⎥ ⎥

tp = June 2 ± 1.3 days

June

Dec.

WIMP Isothermal Halo (assume no co-rotation) v0~ 230 km/s

ZIP Detector 380 mm Al fins

60 m

wid

e

Phonon Sensor Design

Crystals: 7.6cm diam, 1cm thick.The films deposited are:• Amorphous Si (40 nm)• Al (300 nm on phonon side)• W for traps (35 nm on phonon side)• W, again, for TESThere are 4 phonon sensors, each covering a quadrant of the crystal. Each sensor is split into 37 cell.Each TES is 1 m wide and 250 m long and it is connected to Al fins.

Ionization Readout

The expected noise level after this amplification is 70-100 nV/Hz, implying that the lowest observable signal is ~1.5 keV.

Ionization signal read-out: based on FETs

An electric field is maintained through the crystal.

Charge created in an interaction is effectively collected on the feedback capacitor and then dissipated in the feedback resistor.Fall-time: RFCF time constant (~40 s).

Rise-time: determined by the amplifier (~1 s).

The noise level of the readout system is given by:

Dominant source of noise: the FET itself

FET operated at 150 K to minimizes its noise

(to ~0.5 nV/Hz at ~50 kHz).

Phonon Readout

Phonon read-out based on SQUIDs.

Phonon sensor is voltage biased.

Interaction changes sensor resistance, which

changes current through input coil of the SQUID.

Current signal in feedback coil: larger by the

turns ratio (10)

Pulse shape determined by the detector physics

The noise level is given by:

SQUID: operated at 600 mK, with responsivity

r = 100 (RTFb).

SQUID and amplifier noise contributions suppressed

The dominant voltage noise source is the shunt resistor

For typical Rs= 100 m at T = 35 mK, we expect the noise level to be at 10 pA/Hz.

Rsh = 20 m

Rs

Rp

Relevant Processes

The phonon behavior is governed by

two processes, both strongly

dependent on frequency:

Anharmonic decay with rate D ~ f-5

Isotopic scattering with I ~ f-4

These two processes define the

“quasidiffusion” of phonons.

An interaction creates phonons at

Debye energy (13.4 THz for Si) - they

are almost still and they decay VERY

quickly (within a few us) to phonons of

energy ~1THz, at which the mean free

path is ~1cm (Si) - i.e. they are ballistic

0 10 20 30 40 50 60 70 800

20

40

60

80

100

120

140

Concentration of Fe [ppm]

Adjustment of W Tc for Optimization

Target Tc

W

substrate

50 keV Feimplantation

Tc measurements at Stanfordin KelvinOx 15

AG limit1st order RKKY

B. A. Young, et al, JAP 86, 6975 (1999)

ZIP detector fabrication: CIS, Stanford

Al/W Grid

60% Area Coverage

37 - 5 mmSquares 888 X 1 µm

tungsten TESin parallel

Aluminum Collector Fins

8 Traps

Superconducting Transition Edge Sensors

Steep Resistive Superconducting Transition

Voltage bias is intrinsically stable

T

• W Tc ~ 70 mK• 10-90% <1 mK

Rshunt

Ibias

W ETF-TES

SQUIDArray

α =dR

dT

R

Tunitless measureof transition width

The Joule heating produced by bias

PJ =V B2

R⇒ PJ ↓whR ↑

iαlwhαfouiα

PJ =IB2 R ⇒ PJ ↑whR ↑

whihiiiiαllyuαl

R