results from the cryogenic dark matter search using a 2 analysis joel sander december 2007
DESCRIPTION
Clusters Mass of luminous matter estimated from cluster luminosity & mass-light ratios Total cluster mass is measured by: – Virial method – X-ray – Gravitational Lensing Coma Cluster - Xray T y(w) =8(10)keV, R vir = 1.8h -1 Mpc 27.1 o 28.6 o o o Gravitational pressure balanced by gas pressure: Assuming n % r -2 : = 1.4x10 15 h -1 M o In agreement with the virial est. p=nk B T.TRANSCRIPT
Results from the Cryogenic Dark Matter Search Using a 2 Analysis
Joel SanderDecember 2007
First Evidence for Dark Matter• Fritz Zwicky’s 1933 observation of the Coma cluster using the virial method
• Measured the radial velocity of 8 galaxies
• Implied gravitational force 400 times greater than expected from luminosity
• Accurate value of the Hubble constant would have changed discrepancy to a factor of 50
• Believed there was unobserved matter “dunkle kalte Materie” or dark cold matter
Mvir = 22vvir/GN
Andromeda
• In 1970, Rubin showed the rotational vr(r) = const. for the Andromeda galaxy
Clusters• Mass of luminous matter estimated from cluster luminosity & mass-light ratios
• Total cluster mass is measured by:
– Virial method
– X-ray
– Gravitational Lensing
Coma Cluster - Xray
Ty(w)=8(10)keV, Rvir = 1.8h-1Mpc
27.1o
28.6o
195.4o 194.1o
Gravitational pressure balanced by gas pressure:
Assuming n r-2:
= 1.4x1015h-1Mo
In agreement with the virial est.
p=nkBT
.
Cosmic Microwave Background
• Light from decoupling of photons and baryons in the early (z~103) nearly isotropic and homogeneous universe
• CMB Photons travel to us from the equidistant surface of last scattering
• Observations described by a blackbody spectrum with T0 = 2.725K
• Initial small fluctuations lead to potential wells, causing small temperature fluctuations
• Size and intensity of the fluctuations are sensitive to early state of the universe
Roughly speaking, ~ 2/l
WMAP only assuming CDM model
Universe Budget: Mostly Unidentified
• Data from Supernovae, CMB, and Clusters in agreement!
• Significant dark matter required.
• But what kind of dark matter?
WMAP, SDSS, and Supernova
m
b
k
Weak-Scale Structure Formation
← Growth with no dark matter
← Current
Size of fluctuation
R – time of decoupling
Weakly Interacting Massive Particles (WIMPs) provide an explanation for the observed structure
Dark Matter Around Here?
rrGMV )(total
r
rrM x )(
Is it possible to detect WIMP interactions?
• Energy deposited, Er
• Rate– Cross section– Flux on detector
• Backgrounds• Assumptions
– Local WIMP density W = 0.3 GeV/c2
– Local galactic velocity: v = 220 km/s = 0.7x10-3c
WIMP Interactions
v/c = 0.7 10-3
Direct Detection: Recoil Experiment
WIMP
We use Germanium, A=73;others: Si, I, Xe, W
v/c = 0.7 10-3
ER 2 v2/mGe
402 (0.7 10-3)2/ 73 10 keV x-ray energy! Easy!
Sun moves with V~ 220km/sec through the Dark Matter halo
Donald H. Perkins, 1987
What is the Rate of Interactions?
Rate = 5 x 10-9 [kg day]-1 Ouch!
R = N = (8x1024 atoms/kg) (6x109cm-2day-1) = 5x1034[cm2] recoils/kg/day
The Cross SectionCalculating the transition rate from Fermi’s golden rule and equating with v gives:
It is useful to express relative to the cross section on a single nucleon, 1
… assuming 12 = 1 and
F2(q2) = 1, we have:r
2A2
= r2A2 r
2A21r
2 A2
[cm2]
For massive WIMPs, cross section proportional to A4!
10-36 [cm2] … 100GeV WIMP on Ge
R = 0.05 recoils / kg / day … still low, but possible to detect
The Form Factor
• The form factor accounts for the structure of the nucleus
• It is given by the Fourier transform of the nuclear wave function
• F(q=0) = 1.
• Helm Form Factor approximates nucleus as:
0: core with constant density and radius, R 1.2A1/3fm
1: surface with thickness of 1fm, 1 e-r2/2s2, s 0.9fm
0 and 1 normalized:
Rate of Main Background
Rate about 20 / (kg-day) !
Strategies: shield and distinguish electron recoils from nuclear recoils
Shield it!
40K: 7x104 /day
1m
Backgrounds: Like an Onion
Peel back one background, and you find another background!
Background Method of Removal• Gammas Shield
Reject – Analysis• Neutrons Shield
Veto• Surface Events Reject - Analysis
Will discuss now.
Will discuss later after describing the detectors.
Shielding Gammas
Low Activity Lead
µ-metal (with copper inside)
Ancient lead
23cm
Use passive shielding to reduce rate toGe:~1312 events / kg / daySi: ~5485 events / kg / day(from 15-45keV)
• Lead and Copper for gammas• Inner lead layer from Roman shipwreck
Ionization
Inner and Outer electrode to reject events near edge
Z-sensitive Ionization Phonon Detector
Q inner
Q outer
D
C
A
B
R sh
I bias
SQUID arrayPhonon A
R feedback
Vqbias
Z-dependentIonization and Phonon-mediated
zy
x
@50 mK
• 250g Ge (100g Si)
• 1 cm thick x 7.6 cm dia.
• Inner/Outer electrodes
• Four phonon sensors
• X-Y-Z Information
• Photolithographic patterning
Phonon Sensors
Phonon Sensors
R0
R
TT0
380 m
60
m
Al
Al Collector W Transition Edge Sensor
(TES)
Ge or Siphonons
Cooper Pair
Sensors held in equilibrium between Normal and Super Conducting. Highly sensitive to small energy deposit. Fast signal. SQUID Readout
Excellent Energy, Position Resolution
Am241 : 14, 18, 20, 26, 60 kev
Cd109 + Al foil : 22 kevCd109 :
22 kevi.c. electr 63, 84 KeV
Detector Calibration at Berkeley
Energy Resolution
Detector Mean (keV) (keV) Mean (keV) (keV)T1Z1 10.58 1.01 10.40 0.44 (1.34)
T1Z2 10.37 0.50 10.30 0.30 (0.78)
T1Z3 10.43 0.32 10.91 0.26 (0.61)
T1Z5 10.43 0.37 10.34 0.32 (0.74)
T2Z3 10.29 0.44 10.28 0.29 (0.73)
T2Z5 10.17 0.39 10.05 0.61 (1.28)
Ionization Phonon
Use the 10.36keV line from 71Ge decay (100% EC) to examine the energy resolution
T1Z2 - Ionization
dN/d
E
T1Z2 - Phonon
dN/d
E
Detector Width (keV)T1Z2 5.6
T1Z5 4.2
T2Z3 2.9
T2Z5 NA
67keV line
CDMS Towers of Detectors6 Detectors Per Tower
– 0.25kg (0.10kg) per Ge (Si) detector– Livetime of 74.58 days– 2 Tower results published– Will describe simultaneous, parallel, blind
analysis– neutron and (open & closed ) calibration data
Each tower holds 6 ZIPs
ZIP 1 (Ge)ZIP 2 (Ge)ZIP 3 (Si)ZIP 4 (Ge)ZIP 5 (Si)ZIP 6 (Ge)
Tower 1
Cold Electronics
Tower 2 Tower 1
Detection: Signal and Background
0 (calibrate: neutron)
710-4
NucleusRecoils
dense energy depositionPoor Ionization Efficiency
Signal
Er
0.3
ElectronRecoils
Background
Sparse Energy DepositionExcellent Ionization Eff.
Er
Recoil Differences giveParticle Identification
Phonon Energy is the true measure of the recoil energy
Shutt et al., 1992
Nuclear recoils (induced by a neutron source)
Electron recoils (induced by a source)
IonizationPhonons
=1 (bkgd)1/3 (sig)
Excellent Primary () Background Rejection
Yie
ld
Ioni
zatio
n / P
hono
n E
nerg
y Calibration Data
The Analysis
List of Cuts:o Data Quality Cuts
• Reject Obviously Bad Data
• Higher Standard for Potential WIMP interactions
o Analysis Regime
o Fiducial Volume Cut
o Ionization Threshold Cut
o Single Scatter Cut
o Nuclear Recoil Band Cut
o Veto Anti-coincidence Cut
o Timing Outlier Cut
o Surface Event Rejection Cut
Will discuss with rejection of gamma bkg. (Now)
Will discuss with neutron bkg.
Will discuss with surface event bkg.
Rejecting Bad Events• Bad Detectors, T1Z1, T1Z6, and T2Z1
5 (4) remaining good Ge (Si) detectors
• Data taken after a power outage
• Spike in event rate after a pump trip
• Data taken during cryogen transfer
• Failed Kolmogorov-Smirnov (KS) - test
• Other (wrong Q-bias, missing global trigger, etc)
• T1Z1 (Ge): high Tc -> uneven detector response and increased number of background events (T1Z1 was first detector fabricated)
• T1Z6 (Si): known contamination with 14C
• T2Z1 (Si): earlier test data & in situ calibration data indicate poor signal-to-noise in 2 phonon channels
List of Cuts:o Data Quality Cuts
• Reject Obviously Bad Data
• Higher Standard for Potential WIMP interactions
o Analysis Regime
o Fiducial Volume Cut
o Ionization Threshold Cut
o Single Scatter Cut
o Nuclear Recoil Band Cut
o Veto Anti-coincidence Cut
o Timing Outlier Cut
o Surface Event Rejection Cut
WIMP-search data: 5.23 million potential interactions
Higher Standard for Potential WIMPs
WIMP-search data: 4.54 million potential interactions
List of Cuts:o Data Quality Cuts
• Reject Obviously Bad Data
• Higher Standard for Potential WIMP interactions
o Analysis Regime
o Fiducial Volume Cut
o Ionization Threshold Cut
o Single Scatter Cut
o Nuclear Recoil Band Cut
o Veto Anti-coincidence Cut
o Timing Outlier Cut
o Surface Event Rejection Cut
Y-delay (s)
Ioni
zatio
n E
nerg
y
T2Z5
• Events with abnormal ionization pulses are removed
• Events with phonon activity preceding the event are removed
• Events with any abnormally negative (6) phonon pulse are removed
• Events within a region of abnormal ionization energy collection are removed
Analysis Regime: Threshold• Need to determine the low recoil energy analysis threshold (7keV, 20keV for T1Z4)
• WIMP spectrum falls exponentially (100keV)
• Don’t want mis-estimation of the ionization energy to allow electron recoils fake nuclear recoils
• Fit ionization pulses with characteristic white noise
• Fit ionization pulses using identical 2 minimization as is used for data. (Large pulses are harder to misfit.)
• Phonon recoil energy threshold chosen such that ~0.1 electron recoils misfit as nuclear recoils expected to be above threshold in WIMP-search data
pr = pt – Q 1549 events from pr=3keV to 3.5keV misfit:
An ER with Q=3.5keV (pr=3.5keV), misfit to Q=1.75keV: pr=5.25keV, 1/400
Q=5keV (pr=5keV), misfit to Q=2.5keV: pr=7.5keV 0/20k
Y=1/3, nuclear recoil
WIMP-search data: 51.4 thousand
calibration data: 526 thousand potential interactions
List of Cuts:o Data Quality Cuts
• Reject Obviously Bad Data
• Higher Standard for Potential WIMP interactions
o Analysis Regime
o Fiducial Volume Cut
o Ionization Threshold Cut
o Single Scatter Cut
o Nuclear Recoil Band Cut
o Veto Anti-coincidence Cut
o Timing Outlier Cut
o Surface Event Rejection Cut
Fiducial Volume Cut• Events near the outer surface can have uncollected ionization (and phonon) energy.
• Require events’ ionization energy to be contained within the inner electrode by requiring ionization energy of outer electrode to be consistent with noise
• About 82% of detector area covered by the inner electrode
Qi
Qo
calibration data: 327 thousand potential interactions
List of Cuts:o Data Quality Cuts
• Reject Obviously Bad Data
• Higher Standard for Potential WIMP interactions
o Analysis Regime
o Fiducial Volume Cut
o Ionization Threshold Cut
o Single Scatter Cut
o Nuclear Recoil Band Cut
o Veto Anti-coincidence Cut
o Timing Outlier Cut
o Surface Event Rejection Cut
• Rejected
• Accepted
Ionization Threshold
calibration data: 326 thousand potential interactions
List of Cuts:o Data Quality Cuts
• Reject Obviously Bad Data
• Higher Standard for Potential WIMP interactions
o Analysis Regime
o Fiducial Volume Cut
o Ionization Threshold Cut
o Single Scatter Cut
o Nuclear Recoil Band Cut
o Veto Anti-coincidence Cut
o Timing Outlier Cut
o Surface Event Rejection Cut
3.85
Cou
nt
Gammas
Nuclear recoils
Zero charge
• Designed to require potential signal events to have an ionization signal
• ~4 above mean noise event to reject zero-charge events
• <0.01 zero charge events expected above threshold (~80 zero charge events between 5-100keV)
Single Scatter Cut
calibration data: 94 thousand potential interactions
List of Cuts:o Data Quality Cuts
• Reject Obviously Bad Data
• Higher Standard for Potential WIMP interactions
o Analysis Regime
o Fiducial Volume Cut
o Ionization Threshold Cut
o Single Scatter Cut
o Nuclear Recoil Band Cut
o Veto Anti-coincidence Cut
o Timing Outlier Cut
o Surface Event Rejection Cut
• WIMPs will not scatter in multiple detectors while backgrounds can
• Rejects events with total phonon energy more than 6 away from the mean in any other detector
Accepted Rejected
Yield Plots – Nuclear Recoil Band
• Yield Plots standard way of showing data
• bands and nuclear recoil bands defined using 133Ba and 252Cf calibration data
Yie
ld
Ioni
zatio
n / P
hono
n E
nerg
y Calibration Data
Gamma-induced electron recoil background peeled away. What about the neutron-induced nuclear recoil background?
calibration data: 59 potential interactions
List of Cuts:o Data Quality Cuts
• Reject Obviously Bad Data
• Higher Standard for Potential WIMP interactions
o Analysis Regime
o Fiducial Volume Cut
o Ionization Threshold Cut
o Single Scatter Cut
o Nuclear Recoil Band Cut
o Veto Anti-coincidence Cut
o Timing Outlier Cut
o Surface Event Rejection Cut
Cosmic Muon Induced Neutron Background
Limited earlier result at shallow (~15mwe)…moved to a deep mine
Where?Soudan, Minnesota
Enter here
Depth of 689m (2341 ft.)
underground (2090 mwe)
Muon flux 50,000 times less than flux at the surface
Depth (meters water equivalent)
Kamioka (Japan)
0 2000 4000 6000 8000 10000
3210
-1-2-3-4-5-6
-7-8
Soudan
Log 1
0(Muo
n Fl
ux) (
m-2s-1
)
Polyethylene
41 cm14
Use passive shielding to reduce /Neutrons• Lead and Copper for photon• Polyethylene for low-energy neutron
Passive Neutron Shielding
Muon Veto
Neutron-induced nuclear recoil background peeled away. What about the surface event background?
calibration data: 57 potential interactions
• Surround detectors with active muon veto• Reject Veto-coincident events• 1 veto coincident multiple scatter nuclear recoil observed
List of Cuts:o Data Quality Cuts
• Reject Obviously Bad Data
• Higher Standard for Potential WIMP interactions
o Analysis Regime
o Fiducial Volume Cut
o Ionization Threshold Cut
o Single Scatter Cut
o Nuclear Recoil Band Cut
o Veto Anti-coincidence Cut
o Timing Outlier Cut
o Surface Event Rejection Cut
Pre-unblinding neutron background estimate:Ge: 0.06 eventsSi: 0.05 events
Where Surface Events Come From• There are three classes of surface events:
1) Ambient gammas that Compton-scatter off electrons in a detector sometimes scatter in the first few microns of a detector. An electron (ejectron) that is scattered near the surface can be ejected from the detector towards an adjacent detector.
2) Electrons Compton-scattered near the detector surface that do not escape the detector.
3) Electrons from beta-decay of radioactive contaminants on detector surfaces. Decays from ambient radon implant 210Pb less than a micron into the surface of a detector.
• These classes do not contribute equally to surface events in WIMP-search data and gamma calibration data.
• Most surface events in calibration data are from classes 1) and 2).
• Most surface events in WIMP-search data are from class 3).
Why Surface Events are Dangerous
↑ Open Gamma Calibration Data
Surface events:
• have incomplete collection of ionization energy
• have abnormally low yield causing them to droop down from the gamma band and potentially into the nuclear recoil band
• are a clear background – most/all of the 57 “gamma” remaining events are surface events
• must be rejected in analysis, but how?
Parameters Used to Reject Surface EventsThere are 3 parameters are used to reject surface events:
A 2 analysis uses these 3 correlated parameters to distinguish between surface events and nuclear recoils. Neutron-induced nuclear recoils from neutron calibration data are used as a surrogate for WIMP-induced nuclear recoils.
1) Phonon Delay – how long after the ionization pulse until the largest phonon pulse reaches 20% of its maximum height.
2) Phonon Risetime – how long it takes for the largest phonon pulse to rise from 10% to 40% of its maximum height.
3) Energy Partition – the ratio of energy deposited in the primary phonon sensor to the energy deposited in the opposite sensor
Nuclear recoils
Surface events
A 2 Analysis
• Calculate the 2 deviation from the nuclear recoil hypothesis, n2, and the
surface event hypothesis, b2
• The values of the 3 surface event rejection parameters have distributions of for nuclear recoils and surface events with a mean, :
• Calculate a covariance matrix expressing the correlations between the surface event rejection parameters:
• The value of b2 and n
2 for the kth event is given by:
i – index indicating nuclear recoil hypothesis (i=n) or surface event hypothesis (i=b)
j – index indicating which surface event rejection parameter
2 Distributions of EventsTo pass the surface event rejection cut an event must meet 2 requirements:
• Must be consistent with the nuclear recoil hypothesis, n
2 < 10.
Nuclear Recoils passing:
0.670.01(Ge) / 0.790.01 (Si)
Surface Events rejected:
0.340.01 (Ge) / 0.560.02 (Si)
• A parameter d2 characterizes whether an event better fits the surface event hypothesis or the nuclear recoil hypothesis.
Will choose a value of d2 using what criteria?
b2 – n
2
Cou
nt
Coadded Germanium
Discovery PotentialSet Ge value of d
2 for 3 hint of a WIMP signal with 3 observed events
Si value of d2 set by eye
o
Table of maximum allowed backgrounds.
No = nr x Ni
m = 1.25 kg
L = 74.6 days
= # of target nuclei / kg = 103NA/A
T = SF between (target nucleus) and (target nucleon)
Sensitivity and Discovery Potential
• Chose a surface event rejection cut with 3 discovery potential for 3 observed events. (B < 0.27 events.)
• Guessed that the neutron background would be 0.1 events. (obviously over estimated the neutron bkg.) (B < 0.17 events, d2 = 12)
• Did not have the plot of sensitivity vs. B at the time, but we knew the cuts were near the optimum.
• Defined stricter cut with B < 0.08 events (d2 = 18).
• Made stricter cut be the primary surface event rejection cut.
b2 – n
2
Cou
ntCoadded Germanium
Silicon Detectors
• Values of d2 chosen
under time pressure
by eye
before deciding to use Si detectors to set a limit
• Notice that newer Tower 2 silicon detectors have much lower surface event rate!
Legend:Nuclear recoils
Surface events
Surface events passing consistency cut (n
2 < 10).
Analysis Efficiency
Testing On Closed Gamma Calib. Data
• Expect 255 events (134 events for the stricter cut) to pass the cut.
• Observe 35 events (26) to pass the cut, 1.9 (3.6) > than expectation.
• What happened?
• Probably due to drift in the experiment.
• Data was split every 2 hours (open then closed) while the noise environment changes in about 1hr.
• The noise template was derived from first 500 events and better describes the open gamma calib. noise environment.
• Resplit the gamma calib. data every other event, found the resplit subsets to be statistically consistent.
Expected Surface Event Background• Background measured from WIMP-search sidebands (“wide beta” – 3 NR band single scatters) and the closed gamma calibration data (all “wide beta” events)
– Gamma calibration: better stats
– WIMP-search: surface event sample in situ, surface events from same source <- better
• The number of surface events passing the cut are scaled to expected number of 2 nuclear recoil band single scatters in WIMP-search data
• Scalings determined using data prior WIMP-search run
Closed
WIMP-Search:
Estimate from Closed calibration
Ge Stricter 0.16+0.08 (10-100keV)
Ge Looser 0.21+0.11 (10-100keV)
-0.07
-0.09
Estimate from WIMP-Search Sidebands
Estimation of Systematics
64020T2Z5
11000T2Z3
55000T1Z5
00000T1Z3
10100T1Z2
TotalFiducial VolumeShifted T2Z5 CliffYield (Upper)Detector
=13 x S4 = 0.4 events
Stricter Cut:
108020T2Z5
22000T2Z3
76100T1Z5
22000T1Z3
10100T1Z2
TotalFiducial VolumeShifted T2Z5 CliffYield (Upper)Detector
=22 x S4 = 0.6 events
Looser Cut:
Estimate from WIMP-Search Sidebands
What Expectations?
Assuming cross section of 10-42 cm2 (normalized to a single nucleon) and assuming standard (simplistic) halo parameters
The Ge Result I
• Pass all cuts except the surface event rejection cut
o Also pass looser cut
o Also pass stricter cut
WIMP-Search WIMP-Search
Closed & neutron calib.
Closed & neutron calib.
• neutrons from calibration data
• “wide beta” surface events
• “wide beta” surface events passing the looser surface event rejection cut
The Ge Result II
• Pass all cuts except the surface event rejection cut
o Also pass looser cut
o Also pass stricter cut
• neutrons from calibration data
• “wide beta” surface events
• “wide beta” surface events passing the looser surface event rejection cut
The Ge Result III
• Pass all cuts except the surface event rejection cut
o Also pass looser cut
o Also pass stricter cut
• neutrons from calibration data
• “wide beta” surface events
• “wide beta” surface events passing the looser surface event rejection cut
• near miss event
No WIMP-induced nuclear recoils observed!
Set a limit.
The Si Result I
• Pass all cuts except the surface event rejection cut
o Also pass looser cut
o Also pass stricter cut
• neutrons from calibration data
• “wide beta” surface events
• “wide beta” surface events passing the looser surface event rejection cut
The Si Result II
• Pass all cuts except the surface event rejection cut
o Also pass looser cut
o Also pass stricter cut
• neutrons from calibration data
• “wide beta” surface events
• “wide beta” surface events passing the looser surface event rejection cut
• 1 event within 2 nuclear recoil band One potential WIMP-induced nuclear recoil observed,
consistent with background. Set a limit.
A LimitLegend:--- Prior result at Soudan
This Ge result (-.-. looser)
Combined result
This Si result
Edelweiss (2005)
DAMA low mass allowed region (90% conf.) (Gondolo & Gelmini 2005)
An MSSM region (Kim et. al. 2002)
What’s Next?
• 5 tower run underway, data currently being analyzed
• Detectors with lower rates of surface event
• Better analysis techniques
• Leading to:
?
The CDMS Collaboration I
Remote Data Acquisition