Towards Direct Detection SignalsTowards Direct Detection SignalsCDMS, DAMA, andCDMS, DAMA, and
Constraining the WIMP mass and cross sectionConstraining the WIMP mass and cross section
Richard Schnee
Syracuse University
TASI
June 24, 2009
TASI 2009 DM Experiment Richard Schnee
Q inner
Q outer
A
B
D
C
Rbias
I bias
SQUID array Phonon D
Rfeedback
Vqbias
1 µ tungsten
380µ x 60µ aluminum fins
•250 g Ge or 100 g Si crystal
•1 cm thick x 7.5 cm diameter
•Photolithographic patterning
! Thousands of W transition-edgesensors
! Fractional area coverage
•Collect athermal phonons
Measure ionization in low-field
(~volts/cm) with segmented
contacts to allow rejection of
events near outer edge
Z-sensitive
Ionization and
Phonon-mediated
detector
Qouter
Qinner
@50 mK
CDMS II: Ionization & Athermal Phonons
TASI 2009 DM Experiment Richard Schnee
CDMS ZIP Detector Phonon Sensors
Al
quasiparticle
trapAl Collector W
Transition-Edge
Sensor
Si or Ge
quasiparticle
diffusion
phonons
• Measurement of athermal
phonon signals allows position
imaging, additional
discrimination based on timing
superconducting
normal
W Transition-Edge Sensor:
sensitive thermometer
T (mK)Tc ~ 80mK
RT
ES
(!
)
4
3
2
1
~ 10mK
Richard Schnee TASI 2009 DM ExperimentPage 4
CDMS II Background RejectionCDMS II Background Rejection
• Effect is evenlarger forotherwisetroublesomesurfaceelectron-recoils
•Set all data-selection cuts“blind” to yieldabout 0.6expectedleakage eventsin WIMP-searchdata
photons from 133Ba source
(30x WIMP-search
background!)
Neutrons from 252Cf
source
133Ba surface events
•Nuclear recoils have lower ionization yield and slowerphonon signals than electron-recoil backgrounds
TASI 2009 DM Experiment Richard Schnee
CDMS WIMP-search Data
Photons
Surface events
Nuclear recoil region
J. Filippini, UCB
PRL 102, 011301,
arXiv:0802.3530
•2006-2007data
•3.75 kg Ge
•400 kg daysexposure(before cuts)
•No WIMPcandidates
•3x data inhand, resultsby end ofsummer
TASI 2009 DM Experiment Richard Schnee
• Research and development towards 300x mass scale-up (GeODM)
! Critical issue is manufacturability (R&D in progress)
! 20x Larger detectors exploiting (inexpensive) dislocation-free Ge crystals
! “Smarter” readout to simplify infrastructure, reduce heatload on cryogenics
! means in hand to stay
background-free at 15x
exposure from current
results
• Improved analysis
• 2.5x Thicker detectors
• Reduced exposure to radon
daughters
• New channel
layout
CDMS Future: SuperCDMS & GeODM
• CDMS background-free, best proven discrimination of backgrounds
• ~4x mass scale up underway
6”
TASI 2009 DM Experiment Richard Schnee
~100 evt/ton/year
CDMSXENON10
CRESST
ZEPLIN II
EDELWEISS
ZEPLIN III
DAMA
Current WIMP Limits
90% CL upper limits on spin-independent interaction assuming
standard halo
• Excludes regions of
supersymmetry
parameter space
under some
frameworks
! Kim et al. 2004 in
light blue (MSSM)
• Starting to probe
Constrained MSSMTrotta et al. 2008 in
blue (68% CL) & green
(95% CL)
TASI 2009 DM Experiment Richard Schnee
Neutrino Limiting Background
• Ton-scale dark matter experiments will detect 100s - 1000s of 8B
neutrinos from sun coherently scattering with nucleus below 5-10 keV
! Directional detectors can reject/subtract
• A 100-ton detector would have background from atmospheric neutrinos
! Limits ultimate detector size L. Strigari, 2009
Richard Schnee TASI 2009 DM ExperimentPage 9
Current WIMP Detection? DAMA/LIBRACurrent WIMP Detection? DAMA/LIBRA
• If WIMPs exist, expect annualmodulation (Drukier, Freese, &
Spergel 1986)
• DAMA/LIBRA do not distinguishbetween WIMP signal andbackground directly, but inferWIMPs from annual modulation
0
25
50
75
100
125
-0.5 -0.1 0.3 0.7 1.1 1.5
Background
WIMP Signal
95
97
99
101
103
105
-0.5 -0.1 0.3 0.7 1.1 1.5
±2%
JuneJune
Dec Dec
JuneJune
Dec DecDec Dec
in lowest-energy single-scatter interactions, assumingbackgrounds don’t modulate:
December
June
log
Richard Schnee TASI 2009 DM ExperimentPage 10
NaI
NaI
NaI
NaI
PM
T
PM
T
DAMA ApparatusDAMA Apparatus
• 250-kg NaI scintillator array• 25 crystals, each 9.7 kg, 10 cm x 10 cm x 25.4 cm
• Each crystal viewed by 2 PMTs through suprasil-B
lightguides, 5.5-7.5 photoelectrons/keVee
• 0.82 ton-years of exposure!
• Located at Gran Sasso
Underground Lab (4000 mwe)
+ Photon and Neutron shielding
• Detectors separated by lots of Cu
• No surrounding scintillator veto
• Extreme efforts taken to avoid
contamination• Etching followed by HP N2
atmosphere
• Installation in N2 atmosphere usingbreathing apparatus
Ultraclean copperAncient leadpolyethylene
Richard Schnee TASI 2009 DM ExperimentPage 11
DAMA Annual Rate Variation
modulation amplitude is only significant at low energy
arXiv:0804.2738
arXiv:0804.2741
Sum of residuals after subtracting time-averaged rates
in each energy bin in each detector
Consistent with a variety of models (fewer when DC rate spectrum also
considered)Richard Schnee TASI 2009 DM ExperimentPage 12
DC SpectrumDC Spectrum
• Average spectrum measures background level
• Impressive raw background rate: ~0.1/kg/keVr/day
• QF in NaI is 11, so 1 keVr ~ 0.09 keVee, keVee is what is
usually plotted
efficiency corrected
S. Golwala
Richard Schnee TASI 2009 DM ExperimentPage 13
DC SpectrumDC Spectrum
• Compare background spectra between DAMA/NaI and
DAMA/LIBRA
• Continuum background ~2X lower than DAMA/NaI" any explanation must not depend on absolute level of continuum
background
• Slight but statistically significant increase in bgnd ~2-2.5 keV
bgnd has gone up in 2-2.5 keV region!?
efficiency correctedupper: DAMA/NaIlower: DAMA/LIBRA
S. GolwalaRichard Schnee TASI 2009 DM ExperimentPage 14
MultiplesMultiples
• No modulation
seen in multiples
• Folded with
modulation
phase
• open circles:singles
• filled triangles:multiples
• Prime evidence that
signal is not a
simple modulation of
background
S. Golwala
Richard Schnee TASI 2009 DM ExperimentPage 15
Other Modulation Experiments
!"#$%#&&'()*$+,'-
KIMS (Yangyang Lab, Korea)Published SI Limit from 4 x 8.7 kg CsI(Tl) crystals (3409 kg-d)
12 detectors (104.4 kg) now installed
pulse shape discriminationliquid scint veto, 1 dru bkgd
ANAIS (Canfranc, Spain) Tests on one 10.7 kg of NaI (Tl)
viewed by 2 PMTs with acrylic light guides Bkgd: 232Th in Cu welds, 40K, ! internal contamination
Eventual array of 10 selected crystalsstored underground since 1988.
pulse shape discrimination, ancient lead shield
Richard Schnee TASI 2009 DM ExperimentPage 16
•Could some background be causing the annual modulation?Such a background would have to fulfill the annual modulationcharacteristics of a standard WIMP:
• Rate = cos(t)
• 1 year period
• Known phase
• Low energies only
• Single hits only
• Amplitude < 7% at maximum
• Consistent signal between NaI/LIBRA and differentdetectors
Possibilities Other Than DM?Possibilities Other Than DM?
Many differences between summer/winter
Most likely to be affected by systematic effect
Not very powerful test
Richard Schnee TASI 2009 DM ExperimentPage 17
Consistency of Consistency of DAMA/NaI DAMA/NaI & DAMA/LIBRA& DAMA/LIBRA
• Funny math in comparing 2-6 keV between expts:
• “difference is 0.008±0.004, only 2#”; actually, it is 0.0093±0.0037,
2.5#, 1.2% chance
• Differential rates: 5-6 keV is 0.0169±0.0056 in DAMA/NaI, 0.0010±0.0031 in DAMA/LIBRA, also 2.5# different
S. GolwalaRichard Schnee TASI 2009 DM ExperimentPage 18
Potassium 3.2 keV PeakPotassium 3.2 keV Peak
• 40K chainelectron capture to
excited state of 40Ar
10.66%
excited state of 40Ar
decays by 1460.85 keV
gamma emission
daughter Ar
atom is missing
K-shell electron now,
so 3.2 keV emerges
as Auger and x-rays when
an atomic electron is capture
and the shells rearrange
S. Golwala
Richard Schnee TASI 2009 DM ExperimentPage 19
The Potassium 3.2 keV PeakThe Potassium 3.2 keV Peak
• Event topology
• 3.2 keV appears in same detector as parent/daughter nucleus
• 1461 keV can deposit all or some of its energy in parent detector, all or some in
other detectors, all or some in copper
• Many kinds of events will be seen. Two interesting examples:
• 1461 keV deposits full energy in another detector. Identified as double-scatter
event with 1461 keV in one detector, 3.2 keV in another. Used to measure
natural K contamination = 13 ppb.
• 1461 keV deposits all energy in Cu. Event appears as 3.2 keV single-scatter.
Inferred contribution to single-scatter DC rate is < 0.1 /( kg keV day).
• Unless rate is grossly wrong,
implies need ~50% percent
modulation in detection or in K in
detector (due to change in
humidity?) - surprisingly high.
• Would like to see check of 1461-
3 keV doubles for modulation
S. GolwalaRichard Schnee TASI 2009 DM ExperimentPage 20
Muon-Induced Muon-Induced 3 keV Peak?3 keV Peak?
• Modulation of muon rate
• DAMA claim that fast neutrons cannot be produced at largeenough flux for this modulation to appear as a ratemodulation in DAMA
• What if muon spallation in NaI detectors creates metastable
isotope that decays with emission of 3 keV? (Spencer Klein)
• Lifetime must be > 500 !s trigger holdoff time
• Beam test of NaI could see such a line
• MACRO measured
annual modulation
of muon rate,
phase is correct to
explain DAMA
signal
Richard Schnee TASI 2009 DM ExperimentPage 21
Possible Modulation of Noise Cut EffectivenessPossible Modulation of Noise Cut Effectiveness
• Cut on sharpness of pulse used to remove “PMT noise” events.
• How do they know what the tail of the PMT noise distribution looks like,
and that this tail does not modulate?
• Dark current in PMTs exponentially temperature dependent; not clear
that temperature stability measurements are good enough.
• Total number of noise events
doesn’t modulate
• Efficiency of cut for particle
interactions does not
modulate
• Poorly motivated but cannot
be ruled out
Richard Schnee TASI 2009 DM ExperimentPage 22
CDMSXENON10
WARP
ZEPLIN II
EDELWEISS
Not aNot a Standard WIMPStandard WIMPSpin-independent interactions
• “Standard”(theoreticallyfavored, >40GeV, spin-independentinteraction on I )excluded
Richard Schnee TASI 2009 DM ExperimentPage 23
CDMSXENON10
Not aNot a Standard WIMPStandard WIMPSpin-independent interactions
• “Standard”(theoreticallyfavored, >40GeV, spin-independentinteraction on I )excluded
• Low-massWIMP with SIinteraction onNa also barelyexcluded
CDMS Si
CoG
EN
TRichard Schnee TASI 2009 DM ExperimentPage 24
CRESST I
NAIAD
Super-K
PICASSO
COUPP
XENON10
Not aNot a Standard WIMPStandard WIMPSpin-dependent on proton
• To be WIMP, need atleast 2 of
• Low-mass WIMPscattering on Nanot I
• Spin-dependentinteraction (mostlyon proton)
• Non-standardWIMP halo
• Or
• Ion Channeling
No experimentalevidence
• Non-standardinteraction (~10suggestions onarXiv)
What could we learn fromWhat could we learn from directdirect
detection of detection of WIMPsWIMPs??
Let’s start with the WIMP mass.
Richard Schnee TASI 2009 DM ExperimentPage 26
Direct Detection Mass Determination BasicsDirect Detection Mass Determination Basics
•Measure number of WIMP recoils and their recoil energies• Assume for what follows that detector cannot determine
directionality (so no diurnal modulation), not enoughstatistics/stability to search for annual modulation
•Energy spectrum results from assumed WIMP velocitydistribution + elastic scattering kinematics
• Standard assumption is Maxwellian (isothermal halo)
• See Lewin & Smith, Astropart. Phys. 6, 87 (1996) for review ofdetails.
• Recoil energy spectrum is roughly exponential e-aE with coefficient
So a $ M-2 for M << MT
a = constant for M >> MT
!
a"
1+M
T
M
#
$ %
&
' (
2
v0
2
Mass of target nucleus
WIMP Mass
Characteristic galactic velocity
Richard Schnee TASI 2009 DM ExperimentPage 27
Dependence of Spectrum on MassDependence of Spectrum on Mass
Maximum energy due
to assumed galactic
escape velocity
Ge
!
a"
1+M
T
M
#
$ %
&
' (
2
v0
2
• Larger WIMP mass M
yields harder energy
spectrum (but small
effect for large
masses)
Richard Schnee TASI 2009 DM ExperimentPage 28
Dependence of Spectrum on Mass, Dependence of Spectrum on Mass, vv00
Ge 270 km/s
170 km/s
• Larger WIMP mass M
yields harder energy
spectrum (but small
effect for large
masses)
• Uncertainty on v0
translates into
uncertainty on M
• For small M,
!
a"
1+M
T
M
#
$ %
&
' (
2
v0
2
!
"M
M="v
0
v0
Richard Schnee TASI 2009 DM ExperimentPage 29
Dependence of Spectrum on Halo ModelDependence of Spectrum on Halo Model
•Dark mattervelocity distributionis not well known
•Different halomodels result indifferent energyspectra
•Double-edgesword
• At first, willincreaseuncertainty on M
• Ultimately, mayallow inferenceof which halomodel is right
Isothermal, triaxial, Evans
halo models with various
parameters
Ge M = 60 GeV/c2
Craig Copi (Case Western)Richard Schnee TASI 2009 DM ExperimentPage 30
Dependence of Spectrum on Interaction(s)Dependence of Spectrum on Interaction(s)
• Different nuclear formfactors
• spin-independent
• spin-dependent on p
• spin-dependent on n
• With a single targetisotope, distinguishonly by difference inenergy spectrum
• Unless assume SDinteractions areinsignificant
• Detections withseveral detectorisotopes would makeproblem much easier.
• Spectral shapes are very similar for low WIMP masses
• Differences for high WIMP masses may confuse mass determination forsmall statistics Jeff Filippini (UC Berkeley)
Richard Schnee TASI 2009 DM ExperimentPage 31
Best-case ScenarioBest-case Scenario
• Sweep all uncertainties and confusion under the rug• Assume SI interaction only, with well-known velocity distribution
• If dominated by experimental statistical uncertainties, how
well do we measure WIMP mass & WIMP-nucleon cross
section for given # events detected?
Richard Schnee TASI 2009 DM ExperimentPage 32
Mass Determination with Mass Determination with GeGe, , XeXe, , SiSi
David Jackson, Rick Gaitskell (Brown U.), RWS
•Same exposure for
each
• 10 keV threshold
for Ge, Si
• 16 keV threshold
for Xe
• Ge gives slightly
better precision
than Xe, followed
by Si
•50 GeV WIMP typically
measured to range 25-
160 GeV with 10 events
in Ge
•Get only lower limits
for more massive WIMP
No upper
limit
Richard Schnee TASI 2009 DM ExperimentPage 33
Mass Determination with Mass Determination with GeGe, , XeXe, , SiSi
David Jackson, Rick Gaitskell (Brown U.), RWS
•Same exposure for
each
• 10 keV threshold
for Ge, Si
• 16 keV threshold
for Xe
• Ge gives slightly
better precision
than Xe, followed
by Si
•50 GeV WIMP typically
measured to range 43-
70 GeV with 100 events
in Ge
Richard Schnee TASI 2009 DM ExperimentPage 34
Mass Determination with Mass Determination with GeGe, , XeXe, , SiSi
David Jackson, Rick Gaitskell (Brown U.), RWS
•Same exposure for
each
• 10 keV threshold
for Ge, Si
• 16 keV threshold
for Xe
• Ge gives slightly
better precision
than Xe, followed
by Si
•50 GeV WIMP typically
measured to ±10% with
1000 events in Ge
Richard Schnee TASI 2009 DM ExperimentPage 35
Mass and Cross Section DeterminationMass and Cross Section Determination
•Uncertainties on mass can translate into uncertainties onWIMP-nucleon cross section #
• All under the assumption we know the WIMP local mass density
60 GeV 120 GeV 250 GeV
10 events expected (30% Poisson uncertainty on #)
1#
90%
99%
10-6
10-7
10-8Cro
ss S
ectio
n (
pb)
David Jackson, Rick Gaitskell (Brown U.), RWSRichard Schnee TASI 2009 DM ExperimentPage 36
Mass and Cross Section DeterminationMass and Cross Section Determination
60 GeV 120 GeV 250 GeV
100 events expected (10% Poisson uncertainty on #)
1#
90%
99%
10-6
10-7
10-8
Cro
ss S
ectio
n (
pb)
David Jackson, Rick Gaitskell (Brown U.), RWS
•Uncertainties on mass can translate into uncertainties onWIMP-nucleon cross section #
• All under the assumption we know the WIMP local mass density
Richard Schnee TASI 2009 DM ExperimentPage 37
Mass and Cross Section DeterminationMass and Cross Section Determination
1000 events expected (3% Poisson uncertainty on #)
60 GeV
120 GeV 250 GeV
1#
90%
99%
David Jackson, Rick Gaitskell (Brown U.), RWSRichard Schnee TASI 2009 DM ExperimentPage 38
Direct Detection and LHCDirect Detection and LHC
• For many models, LHCwill constrain WIMP massto 10%
• However, it is difficult tomeasure WIMPproperties well
• Given WIMP mass fromLHC, direct detection candetermine WIMP-nucleoncross section much betterthan LHC alone• Statistical uncertainties
simply Poisson based onnumber of WIMPs detected
SuperCDMS
15000 kg-d
(region of %
coanihilation)
• Presumes same particle for both, WIMP comprises entire halo
• If relax second assumption, direct detection provides lower limiton WIMP-nucleon cross section
Baltz, Battaglia, Peskin &
Wizansky hep-ph/0602187
Richard Schnee TASI 2009 DM ExperimentPage 39
ConclusionsConclusions
• Determination of WIMP mass is much better for low WIMP
masses
• Systematic uncertainties on WIMP mass due to halo are of
order ± 20% - ±50% (should be explored in more detail)
• Better measurements of WIMP mass from colliders may be
combined with information from direct detection to better
constrain WIMP-nucleon cross section (and hence particle-
physics parameters)
250 GeV/c269 GeV55 GeV1000
100 GeV/c2101 GeV45 GeV100
50 GeV/c2none30 GeV10
Upper
99% Limit
Lower
99% Limit
Events
Detected
No upper limit
for WIMPs
with M >
For 60 GeV WIMP