Experimental Direct Detection of Dark Matter: Present and...

ExperimentalDirect Detection of Dark Matter:

Present and FutureSunil Golwala

CaltechDaMaSC III2014/04/17

DaMaSC 2014/04/17 Sunil Golwala

Overview

Dark Matter CandidatesOld paradigms and new paradigms for dark matter

Techniques to Search for Dark MatterModel-independent paradigms

Dark Matter Direct Detection SignaturesOld paradigms and new paradigms for the signatures

Experimental Innovations

2


Dark Matter Candidates

What do you need?Generated in early universe

Non-relativistic at the time structure formation beganTo explain connection between CMB power spectrum and LSS today

Weakly- or non-interacting with normal matterTo avoid prior detection

Cold or warm todayTo avoid washing out small-scale structure today

The old paradigmFavored because natural from particle physics perspective:

sterile neutrinos (Kusenko -- this morning), axions, weakly interacting massive particles (WIMPs) esp. SUSY WIMPs

Reasonably well motivated but unpopular or too esotericsuperWIMPs (gravitinos, axinos, etc.), light scalars from string theory,

non-perturbative field configurations (Q-balls), ...

Ad hoc: developed to solve DM problem rather than by other particle physicsprimordial black holes, superheavy dark matter (WIMPzillas, strangelets, quark nuggets), ...

3


Axions

4

G. Raffelt

CosmologyCosmology

CosmicCosmicStringString

In spite of small mass, axionsIn spite of small mass, axionsare bornare born nonnon--relativisticallyrelativistically((““nonnon--thermal relicsthermal relics””))

Cold dark matterCold dark mattercandidatecandidatemmaa ~ 1~ 1--10001000 µµeVeV

Search for Axion Dark MatterSearch for Axion Dark Matter

SS

NN

γγaa

BBextext

Microwave resonatorMicrowave resonator(1 GHz(1 GHz == 44 µµeV)eV)

PrimakoffPrimakoffconversionconversion

ParticleParticle--Physics MotivationPhysics Motivation

CP conservation in QCD byCP conservation in QCD byPecceiPeccei--Quinn mechanismQuinn mechanism

For fFor faa >>>> ffππ axions areaxions are ““invisibleinvisible””and very lightand very light

AxionsAxions aa ~~ ππ00

mmππffππ ≅≅ mmaaffaa

γγ

γγ

aa

Solar and Stellar AxionsSolar and Stellar Axions

Axions thermally produced in stars,Axions thermally produced in stars,e.g. by Primakoffe.g. by Primakoff productionproduction

•• Limits from avoiding excessiveLimits from avoiding excessiveenergy drainenergy drain

•• Search for solar axions (CAST)Search for solar axions (CAST)

aaγγ

CosmologyCosmology





SS

NN

γγaa

BBextext








γγ

γγ

aa





aaγγ

CosmologyCosmology





SS

NN

γγaa

BBextext








γγ

γγ

aa





aaγγ

4

Hence

�2a(k0)

�2R(k0)

=H2

I

⇡2�2R(k0)✓2i f

2a

=↵0(k0)

1� ↵0(k0), (29)

where the axion adiabaticity ↵0(k0) is constrained by theWMAP 5-year data [1] to

↵0 < 0.072 (at 95% CL). (30)

Using the value of �2R(k0) in Eq. (2) and the BICEP2

result for HI in Eq. (7), this bound can be rephrased as

✓i fa,12 > 2.8⇥ 1011. (31)

Combined with Eq. (24), this leads to the bounds

✓i <

✓⌦c h

2

0.0051

◆2 ✓108 GeV

2.4HI

◆3

= 1.65⇥ 10�19, (32)

fa > 9⇥ 1037 GeV. (33)

The latter is much larger than the Planck scale and there-fore Scenario B is excluded.

Scenario A extends in the region fa < HI/2⇡, whichfor the BICEP2 value of HI corresponds to

fa < 0.96⇥ 1014 GeV, ma > 0.06 µeV. (34)

In this scenario, the axion energy density does not dependon the tensor-to-scalar ratio r. The preferred PQ scaleand mass for CDM axions depend on the contribution↵dec from decays of axionic strings and walls. We findthem to be

fa = (8.7± 0.2)⇥ 1010 GeV (↵dec + 1)�6/7, (35)

ma = (71± 2)µeV (↵dec + 1)6/7. (36)

Since ⌦ah2 ⌦mis

a h2 ⌦ch2, the numerical coe�-

cients also represent a cosmological upper limit on faand lower limit on ma.

P.G. was supported in part by NSF grant PHY-1068111. Within a day of this note, Refs. [41, 42] ap-peared on the same topic.

⇤ Electronic address: [email protected]† Electronic address: [email protected]

[1] E. Komatsu et al. [WMAP Collaboration], Astrophys. J.Suppl. 180, 330 (2009).

[2] P. A. R. Ade et al. [Planck Collaboration],arXiv:1303.5076.

[3] S. Weinberg, Phys. Rev. Lett. 40, 223 (1978).

[4] F. Wilczek, Phys. Rev. Lett. 40, 279 (1978).[5] R. D. Peccei, H. R. Quinn, Phys. Rev. Lett. 38, 1440

(1977).[6] J. E. Kim, Phys. Rev. Lett. 43, 103 (1979).[7] M. A. Shifman, A. I. Vainshtein, V. I. Zakharov, Nucl.

Phys. B 166, 493 (1980).

[8] M. Dine, W. Fischler, M. Srednicki, Phys. Lett. B 104,199 (1981).

[9] A. P. Zhitnitskii, Sov. J. Nucl. Phys. 31, 260 (1980).[10] G. G. Ra↵elt, Lect. Notes Phys. 741, 51 (2008).[11] J. Preskill, M. Wise, F. Wilczek, Phys. Lett. B 120, 127

(1983).[12] L. Abbott, P. Sikivie, Phys. Lett. B 120, 133 (1983).[13] M. Dine, W. Fishler, Phys. Lett. B 120, 137 (1983).[14] F. W. Stecker, Q. Shafi, Phys. Rev. Lett. 50, 928 (1983).[15] M. S. Turner, F. Wilczek, Phys. Rev. Lett. 66, 5 (1991).[16] D. H. Lyth, E. D. Stewart, Phys. Rev. D 46, 532 (1992).[17] M. Beltran, J. Garcia-Bellido, J. Lesgourgues, Phys. Rev.

D 75 103507 (2007).[18] M. P. Hertzberg, M. Tegmark, F. Wilczek, Phys. Rev. D

78, 083507 (2008).[19] P. Fox, A. Pierce, S. Thomas, arXiv:hep-th/0409059v1.[20] P. Sikivie, Lect. Notes Phys. 741, 19-50 (2008),

arXiv:astro-ph/0610440v2.[21] L. Visinelli, P. Gondolo, Phys. Rev. D 80, 035024 (2009).[22] L. Visinelli, P. Gondolo, Phys. Rev. D 81, 063508 (2010).[23] P. A. R. Ade et al. [BICEP2 Collaboration],

arXiv:1403.3985.[24] D. J. Gross, R. D. Pisarski, L. G. Ya↵e, Rev. Mod. Phys.

53, 43 (1981).[25] E. W. Kolb, M S. Turner, The Early Universe, Addison-

Wesley (1990).[26] M. S. Turner, Phys. Rev. D 33, 889 (1986).[27] D. H. Lyth, Phys. Rev. D 45, 3394 (1992).[28] K. Strobl, T. J. Weiler, Phys. Rev. D 50, 7690 (1994).[29] K. J. Bae, J. Huh, J. E. Kim, JCAP 9, 005 (2008).[30] T. Kobayashi, R. Kurematsu and F. Takahashi, JCAP

1309, 032 (2013).[31] N. D. Birrell, P. C. W. Davies, Quantum Field Theory in

Curved Space-Time, Cambridge University Press (1982).[32] D. Harari, P. Sikivie, Phys. Lett. B 195, 361 (1987).[33] C. Hagmann, S. Chang, P.Sikivie, Nucl. Phys. B Proc.

Suppl. 72, 81 (1999).[34] R. Davis, Phys. Rev. D 32, 3172 (1985); Phys. Lett. B

180, 225 (1986).[35] R. A. Battye, E. P. S. Shellard, Nucl. Phys. B 423, 260

(1994); Phys. Rev. Lett. 76, 2203 (1996).[36] T. Hiramatsu, M. Kawasaki, K. I. Saikawa and

T. Sekiguchi, Phys. Rev. D 85, 105020 (2012) [Erratum-ibid. D 86, 089902 (2012)].

[37] P. Sikivie, Phys. Rev. Lett. 51, 1415 (1983); ibid. 52, 695(1984).

[38] S. Asztalos et al., Phys. Rev. D 64 092003 (2001); ibid.69, 011101(R) (2004).

[39] L. D. Du↵y et al., Phys. Rev. D 74, 012006 (2006).[40] M. Shibata et al., J. Low Temp. Phys. 151, 1043 (2008).[41] T. Higaki, K. S. Jeong and F. Takahashi,

arXiv:1403.4186.[42] D. J. E. Marsh, D. Grin, R. Hlozek and P. G. Ferreira,

arXiv:1403.4216.

BICEP2+ constrains axion mass:Visinelli and Gondolo (arXiv: 1403.4594):

αdec = (mass density due to decays of axionic topological defects)/(mass density due to initial vacuum misalignment) ~ 0.1-200 (?)


Cosmologically interesting:provides appropriate ΩDM,ma = 1 µeV to 1 meV maybe ~100 µeV

Axions: DefinitivelyTestable

Microwave cavity conversion1 GHz = 4 µeV: use

high-Q tunable cavity in high B field; when f0 = ma, excess power

Detection: RF amplifier + Fourier transform power spectrum, (excited Rydberg atom photodetection)

Can cover ~1 µeV to 100 µeV; cavities become too small > 100 µeV

Good prospects for covering full QCDaxion range (KSVZ to DFSZ) up to100 µeV with near-term dev’ts, perhaps to higher mass

5

DirectDirectsearchsearch

OverclosureOverclosure

TeleTelescopescopeExperimentsExperiments

Globular clustersGlobular clusters(a(a--γγ--coupling)coupling)

Too manyToo manyeventsevents

Too muchToo muchenergy lossenergy loss

SN 1987A (aSN 1987A (a--NN--coupling)coupling)

101033 101066 101099 10101212 [[GeVGeV]] ffaa

eVeVkeVkeV meVmeV µµeVeVmmaa

Hot dark matter limitsHot dark matter limits(a(a--ππ--coupling)coupling)

CASTCAST

CosmologicallyInteresting


after G. Raffelt26 CONTENTS

Figure 3. The landscape of axion searches. The vertical axis is the axion’s coupling to two photons. Thehorizontal axis is the axion’s mass. The diagonal lines are the expected range in coupling for the QCD axion.The allowed QCD axion window is approximately between 1 µeV and 1 meV. Dark matter QCD axions arein the approximate mass range 1 µeV to 100 µeV, with the bounds having considerable uncertainties. Alsoshown are upper limits from SN1987A (also white dwarfs) and HB stars (the red giant bound). Sensitivitiesof various technologies are also shown (“Laser”, etc.). The QCD (PQ) dark-matter axions will be exploredwith high sensitivity in the next decade by RF-cavity experiments. The solar experiments (CAST andIAXO) have sensitivity a a large part of the non-PQ search space and the upper end of the QCD axionwindow. Of course, there could be surprises in both mass and couplings.

4.3 Black holes

Primordial black holes (PBH) constitute a viable dark matter candidate. This is probably the only possibleform of dark matter that is not made up of exotic new elementary particles or forms of matter. Black Holes(BH) have been contemplated as Dark Matter candidates since at least 1970s [183, 184, 185]. In fact, one of

Community Planning Study: Snowmass 2013

CSS CF3 working group


Cosmologically interesting:provides appropriate ΩDM,ma = 1 µeV to 1 meV maybe ~100 µeV

Axions: DefinitivelyTestable

Microwave cavity conversion1 GHz = 4 µeV: use

high-Q tunable cavity in high B field; when f0 = ma, excess power

Detection: RF amplifier + Fourier transform power spectrum, (excited Rydberg atom photodetection)

Can cover ~1 µeV to 100 µeV; cavities become too small > 100 µeV

Good prospects for covering full QCDaxion range (KSVZ to DFSZ) up to100 µeV with near-term dev’ts, perhaps to higher mass

5

DirectDirectsearchsearch

OverclosureOverclosure

TeleTelescopescopeExperimentsExperiments

Globular clustersGlobular clusters(a(a--γγ--coupling)coupling)

Too manyToo manyeventsevents

Too muchToo muchenergy lossenergy loss

SN 1987A (aSN 1987A (a--NN--coupling)coupling)

101033 101066 101099 10101212 [[GeVGeV]] ffaa

eVeVkeVkeV meVmeV µµeVeVmmaa

Hot dark matter limitsHot dark matter limits(a(a--ππ--coupling)coupling)

CASTCAST



after G. Raffelt

Lawrence Livermore National Laboratory LLNL-PRES-650592 25

ADMX & ADMX-HF Targets.

ADMX-HF year 1 target (4-6 GHz @ KSVZ)


A WIMP χ is like a massive neutrino: produced when T >> mχ via pair annihilation/creation. Reaction maintains thermal equilibrium.

If interaction rates high enough, comoving density drops as exp(−mχ / T) as T drops below mδ : annihilation continues, production becomes suppressed.

But, weakly interacting → will “freeze out” before total annihilation if

i.e., if annihilation too slow to keepup with Hubble expansion

Leaves a relic abundance:

for mχ = O(100 GeV)→ if mχ and σann determined bynew weak-scale physics, then Ωχ is O(1)

The Classic WIMP Scenario

6

freeze out

canonical Kolb and Turnerfreeze-out plot

H > �ann � n�

��ann v�

��

�H0

100 km/s/Mpc

�� 10�27

��ann v� fr

cm3 s�1


119

005100010010

1000

2000

2.7e−09

3.2e−09

3.7e−09

3.7e−09

3.7e−09

3.7e−09

4.4e−09

4.4e−09

114

119

119

122

122

124

124

124

005100010010

1000

2000

mh = 126 GeV

127

125

124

122.5

m0 (

GeV

)

m1/2 (GeV)

tan = 30, A0 = 2.5m0, > 0Assume CMSSM:

Constrained MinimalSupersymmetric Standard Model

Very narrow blue strips: LSP relic density matchesDM density

Green: BR(b → s γ) too large

Pink: gµ − 2 deviation from SMexplained by SUSY

Purple line: lower limit on parameterspace due to absence of missing transverse energy events at LHC

Green line: BR(Bs → µ+µ−) provideslower limit on parameters space

Black lines: Various Higgs mass values

Changes in tan β and A0 affect mh ⇒ reduced compatibility with relic density

Very limited parameter space where LSP relic density can match DM density, complies with excluded regions, and provides acceptable Higgs mass. Cannot explain gµ − 2.Can release assumptions about SUSY (e.g. non-universal Higgs mass) at cost in elegance.

LHC Tests of Constrained Minimal Supersymmetry

7

disallowed because LSP is charged

disallowed because LSP ischarged

J. El

lis (

2013

)

joint allowed region

(blue)

mh = 125.9 ± 0.4 GeV


Shifting Paradigms: Beyond Supersymmetric Dark Matter

8

mSUGRA

R-parity

Conserving

Supersymmetry

pMSSM

R-parity

violating

Gravitino DM

MSSMNMSSM

Dirac

DM

Extra Dimensions

UED DM

Warped Extra

Dimensions

Little Higgs

T-odd DM

5d

6d

Axion-like Particles

QCD Axions

Axion DM

Sterile Neutrinos

Light

Force Carriers

Dark Photon

Asymmetric DM

RS DM

Warm DM

?

Hidden

Sector DM

WIMPless DM

Littlest Higgs

Self-Interacting

DM

Q-balls

T Tait

Solitonic DM

Quark

Nuggets

Techni-

baryons

Dynamical

DM


Shifting Paradigms: Beyond Supersymmetric Dark Matter

8

mSUGRA

R-parity

Conserving

Supersymmetry

pMSSM

R-parity

violating

Gravitino DM

MSSMNMSSM

Dirac

DM

Extra Dimensions

UED DM

Warped Extra

Dimensions

Little Higgs

T-odd DM

5d

6d

Axion-like Particles

QCD Axions

Axion DM

Sterile Neutrinos

Light

Force Carriers

Dark Photon

Asymmetric DM

RS DM

Warm DM

?

Hidden

Sector DM

WIMPless DM

Littlest Higgs

Self-Interacting

DM

Q-balls

T Tait

Solitonic DM

Quark

Nuggets

Techni-

baryons

Dynamical

DM

previousslide


or, perhaps: A Phenomenological Approach to DM

All possible interactions with χ need to be mapped out experimentallyusing all the tools we have available...

9

Quarks

W

LeptonsχGluons

Photons

Z

Higgs

LHC

Direct Scattering

Gamma Rays

Neutrinos

ILC?

LEP

Anti-matter

Tait


Interaction of Dark Matter with Normal Matter

The old paradigmIn NR limit, all interactions reduce to

spin-independent or spin-dependent couplings of WIMP to quarks

Coherently sum over quarks in nucleon and nucleons in nucleus to obtain coupling proportional to A2 or J2

Large A and large J provide best sensitivity

Billiard ball scattering of WIMP with nucleus: search constrains σSI and/or combinations of σSD,p and σSD,n

Scattering with nuclei much higher rate than scattering with electrons: signature of WIMPs is nuclear recoils

Form factor describes breakdown of coherence: momentum transfer probes structure of larger nuclear at lower ER than lighter nuclei

10

χ

�

Er

E� =12m�v2

Er

E�=

4mNm�

(mN + m�)2cos2 �

v� � vgalactic � 0.001c

I/Xe

Ge

Si


The new paradigmNR limit not ok for nucleons!

Fitzpatrick, Haxton, Katz, Lubbers, Xu 2013:Generic effective theoryNeed to consider

much larger set of couplings to nucleons (8)

Orbital angular momentum of nucleons can be important

Changes natureof coherence

Surprising patterns

Why not violate isospin?

Interaction of Dark Matter with Normal Matter

11

H. Nelson

1.25¥ 105 1.80¥ 105

1.17¥ 106

3.13¥ 106 3.23¥ 106

»Mp»

F Na Ge I Xe

500000

1.0¥ 106

1.5¥ 106

2.0¥ 106

2.5¥ 106

3.0¥ 106

M

1.51¥ 105 2.11¥ 105

1.83¥ 106

5.69¥ 1066.11¥ 106

»Mn»

F Na Ge I Xe

1¥ 106

2¥ 106

3¥ 106

4¥ 106

5¥ 106

6¥ 106

M

2.26¥ 103

340.

0.0241252.

0.153

»Sp'»

F Na Ge I Xe

500

1000

1500

2000

S'

0.775 2.31

46.3

10.2

122.

»Sn'»

F Na Ge I Xe

20

40

60

80

100

120

S'

1.14¥ 103

203.

0.00573

181.

0.0619

»Sp''»

F Na Ge I Xe

200

400

600

800

1000

S''

0.330 0.987

36.2

8.57

125.

»Sn''»

F Na Ge I Xe

20

40

60

80

100

120

S''

40.9

378.

0.770

718.

19.2

»Dp»

F Na Ge I Xe

100

200

300

400

500

600

700

D

29.646.6

306.

117.

155.

»Dn»

F Na Ge I Xe

50

100

150

200

250

300

D

64.0 2.41¥ 103

6.01¥ 104

2.63¥ 105

1.52¥ 105»Fp''»

F Na Ge I Xe

50000

100000

150000

200000

250000

F''

417. 3.97¥ 1032.37¥ 104

5.20¥ 104

2.37¥ 105

»Fn''»

F Na Ge I Xe

50000

100000

150000

200000

F''

Figure 1: Size of integrated form factorsR

100MeV

0

1

2

qdqF (q2) for di↵erent nuclear responses,

weighted by the natural abundances of isotopes.

24

�L

�L · �s

proton neutron


The Dark Matter “Beam” and Recoil Energy Spectrum

The old paradigmMaxwell-Boltzmann

halo

vc = 220 km/sσv = 270 km/s

vesc = 544 km/sflux × s-wave scattering→ exponentialrecoil energy spectrum

The new paradigm (c.f. first half of afternoon)Deviations from Maxwell-Boltzmann (Green 2011):

excess particles at low speeds lower, flatter peakcircular velocity does match most likely speed: vc/v0 ~ 0.85

Imperfect relaxationClumpiness, spikes in phase space due to tidal streams

Dark disk?

12

WIMP flux~105/cm2/sec

s-wave scattering

v2c

r= G

M(r)

r2

⌧1

2

m⇤ v2

�=

⌧1

2

G Mgal m⇤r⇤

�

pg = n k Tg

dpg

dr=

G Mgal ⇢g

r2

H > �ann ⇠n�

h�ann vi

⌦� h2 ⇡ 10

�27

h�ann vifr

cm

3s

�1

dR

dER/ n� �n�

E0exp

✓�ER

E0

◆A2 F 2

(ER)

E0 =

2 m2� mN

(m� + mN )

2 v20 ⇡

mN

10

6⇠ 50 keV

E�1�

E� exp

✓�E�

kT

◆

v2c

r= G

M(r)

r2

⌧1

2

m⇤ v2

�=

⌧1

2

G Mgal m⇤r⇤

�

pg = n k Tg

dpg

dr=

G Mgal ⇢g

r2

H > �ann ⇠n�

h�ann vi

⌦� h2 ⇡ 10

�27

h�ann vifr

cm

3s

�1

dR

dER/ n� �n�

E0exp

✓�ER

E0

◆A2 F 2

(ER)

E0 =

2 m2� mN

(m� + mN )

2 v20 ⇡

mN

10

6⇠ 50 keV

E�1�

E� exp

✓�E�

kT

◆

Recoil energy

Log

(eve

nt ra

te)

〈E〉 ~ 30 keVfor 100 GeV

on Ge


Signal Characteristics

Event-by-event characteristics:Nuclear recoils ~ keV to tens of keV

Single-scatter

Statistical properties: modulationby WIMP beam kinematicsAnnual modulation:

few % addition/subtraction from Sun’s velocity

Diurnal modulation: O(unity) variation in recoil

direction with time of day

13

Nuclear recoils

Annual flux modulation

Diurnal direction modulation

No multiplicity

EVENT-BY-EVENT

STATISTICAL


Backgrounds

Particle types:α, β, γ, n

Source:radiogenic cosmogenic

14

14 WIMP Dark Matter Direct Detection

!Figure 1-11. Background sources and shielding in a typical direct detection experiment

neutrinos form a fundamental lower bound on the cross section for background-free WIMP detection [43].Next generation experiments will have sensitivity within an order of magnitude of the neutrino signal formost of the mass range, and will actually detect the 8B solar neutrino signal.

Finally, another method to deal with backgrounds is to exploit the fact that the Earth is moving through thedark matter that surrounds our galaxy, yielding a “WIMP wind” that appears to come from the constellationCygnus. This should, in principle, create a small “annual modulation” in the detected WIMP rates, as wellas a somewhat larger daily modulation, as shown in Fig. 1-12. However, if such e�ects were detected in an

Figure 1-12. Schematic of the possible sources of annual modulation (left) and daily modulation (right)e�ects if WIMPs are detected in direct detection experiments

experiment, there would still have to be a convincing demonstration that there are no such modulations inbackground sources.


Depth (mwe)

Log 1

0(M

uon

Flux

) (/

m2 /

s)

Parent location:in the target

on the surfaceon nearby surfacesin surrounding materials


Backgrounds

Particle types:α, β, γ, n

Source:radiogenic cosmogenic

14

14 WIMP Dark Matter Direct Detection

!Figure 1-11. Background sources and shielding in a typical direct detection experiment

neutrinos form a fundamental lower bound on the cross section for background-free WIMP detection [43].Next generation experiments will have sensitivity within an order of magnitude of the neutrino signal formost of the mass range, and will actually detect the 8B solar neutrino signal.

Finally, another method to deal with backgrounds is to exploit the fact that the Earth is moving through thedark matter that surrounds our galaxy, yielding a “WIMP wind” that appears to come from the constellationCygnus. This should, in principle, create a small “annual modulation” in the detected WIMP rates, as wellas a somewhat larger daily modulation, as shown in Fig. 1-12. However, if such e�ects were detected in an

Figure 1-12. Schematic of the possible sources of annual modulation (left) and daily modulation (right)e�ects if WIMPs are detected in direct detection experiments

experiment, there would still have to be a convincing demonstration that there are no such modulations inbackground sources.


Depth (mwe)

Log 1

0(M

uon

Flux

) (/

m2 /

s)

Parent location:in the target

on the surfaceon nearby surfacesin surrounding materials

+ eventually, the ultimate background:

coherent nuclear scattering of

solar, atmospheric, and diffuse supernova

bgnd neutrinos.Irreducible!


Nuclear Recoil Discrimination

χ

v/c ≈ 10-3

NuclearRecoils

Dense Energy Deposition

γ

ElectronRecoils

Signal Background

Neutrons same, but

σ ≈ 1020

higher; must reduce/moderate

Alphas also have highenergy deposition densities

v/c ≈ 0.3

Sparse Energy Deposition

Er

Er

Density/Sparsity: Basis of Discrimination

15


Discrimination Techniques

Need sensitivity to energy deposition characteristics(density, energy) to discriminate nuclear recoils (NRs),

electron recoils (ERs), and alphas

16

ionization scintillation

phonons

2-phase LXe and LAr

sub-KelvinGe and Si

sub-Kelvin scintillatingcrystals (CaWO4,

etc.)

pulse rise timesingle-phase LAr, LNe

bubblechambers

bubbles nucleationdiscriminates NRs and ERs

alphas discriminated in bubble chambers using acoustic or

scintillation signal


Where We Are, Where We Are Going

17

1 10 100 1000 10410�5010�4910�4810�4710�4610�4510�4410�4310�4210�4110�4010�3910�3810�37

10�1410�1310�1210�1110�1010�910�810�710�610�510�410�310�210�1

WIMP Mass �GeV⇤c2⇥

WIMP�nucleoncrosssection�cm2 ⇥

WIMP�nucleoncrosssection�pb⇥

8BNeutrinos

Atmospheric and DSNB Neutrinos

Xenon100 (2012)CRESST

CoGeNT(2012)

CDMS Si(2013)

DAMA

DAMA

CRESST

SuperCDMS Si7BeNeutrinos

SuperCDM

S Si HV

SuperCDMS Si HV

SuperCDMS Si SuperCDMS Ge

SuperCD

MS Ge HV SuperCDMS Ge HV

LUX (2013)

DarkSide G2LZ

Xenon1TSuperCDMS Ge

COHERENT NEUTRIN O SCATTERING CO

HERENT NEUTRI NO SCATTERING COHERENT NEUTRINO SCATTERING

CDMSlite (2013)

DAMIC (2012)

PICO250-C3F8

LUX 300-day

PICO250-CF3I

Figure 1: A compilation of spin-independent WIMP-nucleon cross-section limits (solid curves), regions of interest(ROI) for possible dark matter signals (closed contours) and projections (dot and dashed curves) for US-led directdetection experiments that are proposed to operate over the next decade. The region above the solid curves repre-sents the excluded parameter space (note the tension with the ROIs). Also shown is an approximate band wherecoherent scattering of 8B solar neutrinos, atmospheric neutrinos, and di�use supernova neutrinos with target nucleiwill limit the sensitivity of direct detection experiments to dark matter particles [45]. SuperCDMS SNOLAB pro-jected sensitivity (for five years of operation) is shown separately for the Ge and Si payloads, and for the standardand high-voltage (CDMSlite) detectors. The black rectangles delineate the new parameter space to be covered bySuperCDMS SNOLAB, and the shaded regions indicate areas to be tested by Ge, Si, and Xe targets. The detectorperformance requirements and predicted backgrounds used to generated these curves are described in Sections 3.3and 4.5, respectively. Figure adapted from the SNOWMASS report [46]. Recent theoretical work [47] has emphasizedthat the spin-independent framework, while it serves as a useful way to track experimental progress, represents onlya subset of the possible interactions of dark matter with nuclei, with relative sensitivities of di�erent target materialsvarying by factors of 10 or more compared to those shown in this plot.

3.1.2 ComplementarityIf another direct detection technology detects a possible dark matter signal, SuperCDMS SNOLABwould provide a confirmation using two di�erent targets and a significantly di�erent technique withvery low backgrounds. Additionally, the detection with a di�erent target material greatly enhancesthe information we would obtain about the dark matter particle. As shown in [48, 49], analyzingsignals from multiple targets improves the accuracy of determining the dark matter particle’s massand cross section, in part because astrophysical and nuclear uncertainties may otherwise providedegeneracies in fits to the data from a single target type. Furthermore, data from only one elementaltarget cannot distinguish between the di�erent possible dark matter couplings, even in the tradi-tional framework involving only spin-independent (SI) and spin-dependent (SD) couplings. Multipletargets with di�erent sensitivity to di�erent couplings are required to disentangle degeneracies anddiscriminate among di�erent dark matter models. Targets with di�erent ratios of protons to neu-

4

spin

-inde

pend

ent,

assu

min

g A

2 sca

ling

spin

-inde

pend

ent,

assu

min

g A

2 sca

ling


Innovation in Techniques: SuperCDMS

SuperCDMS: 1990s:

phonons + ionizationdiscriminate NRs from ERsat low bias (few V)

2000s: phonon rise time

discriminates surface events from bulk events

2010s: sophisticated electrode

structure discriminates surface events from bulk events (EDELWEISS also)

double-sided phonon sensorpromises phonon asymmetrydiscrimination

measure ionization only using phonons with high field: new sensitivity to low mass

18

Ionization measurement

Inner electrode(85%)

Outerelectrode(15%)

Two ionization channels:! Inner fiducial volume! Outer electrode where field lines are not uniform

!!

"#$%$&'()*+$

%,#$%$-.-/0'(+-

e- and h+ drift to surfaces in 3 or 4 V/cm applied field

FET amp

phonon sensors on this surface



SuperCDMS: 1990s:








18

• bulk electron recoils (gamma source)• bulk nuclear recoils (neutron source)X surface electron recoils (NND selection)



SuperCDMS: 1990s:








18

Ionization measurement

Inner electrode(85%)

Outerelectrode(15%)

Two ionization channels:! Inner fiducial volume! Outer electrode where field lines are not uniform

!!

"#$%$&'()*+$

%,#$%$-.-/0'(+-

e- and h+ drift to surfaces in 3 or 4 V/cm applied field

FET amp

surface events suffer poor ionization collection

surface events suffer poor ionization collection



SuperCDMS: 1990s:








18

surface event

nuclear recoil

rising edge slope



SuperCDMS: 1990s:








18

With this E-field configuration, a recoil event in the bulk of the detector will have e� transportto the narrow positively biased electrode instrumented with charge amplifiers, while the h+ aretransported to the identically instrumented negatively biased electrode on the opposite face, leadingto a symmetric ionization signal for bulk events. For events near a face, the carriers follow thelarge transverse E-fields that run between the interleaved electrodes on the same face leading toasymmetric ionization collection signals. For radial fiducialization, we partition and separatelyreadout the ionization instrumented electrodes on each face into 2 concentric toroids (for a total of4 ionization channels per detector).

X [mm]

Z [m

m]

!2 !1 0 1 2

0

1

2

Figure 3: Left: Phonon and ionization sensor layout for the iZIP detector deployed at Soudan. The two faces areinstrumented with interleaved ionization (40 µm wide) and phonon sensors with �1,mm pitch. The phonon sensorchannel patterns diagrammed in Fig. 2 are visible. Right: Magnified cross section view of electric field lines (red)and equipotential contours (blue) near the bottom face (Z = 0 mm) of a SuperCDMS iZIP detector. The ionizationelectrode lines (yellow) are narrower than the athermal phonon collection sensors (green).

As a side benefit, the large E-fields parallel to the surface of the detector limit charge trappingto the point that very few electron-recoil surface events are indistinguishable from nuclear recoilseven without z fiducialization.

Ionization Detection Through Luke-Neganov GainDuring charge transport, Luke-Neganov phonons are created with a total energy (in eV) equal

to the voltage the carriers travel through. These phonons are measured by our athermal phononsensors, in addition to the intrinsic recoil phonons. In our standard iZIP running mode, we operatewith the smallest bias voltage required to achieve a good measurement of the ionization chargeso that the true phonon recoil signal is not overwhelmed by Luke-Neganov phonons, which woulddegrade the discrimination of electronic and nuclear recoils via ionization yield.

Nuclear recoils from very low-mass WIMPs (M� . 3 GeV/c2) are su⌅ciently small that therelatively uniform (versus energy) electron-recoil background has a reduced overlap with the WIMPsignal region. As a result, it is advantageous to exchange the background discrimination providedby ionization yield for the improved sensitivity to low energy recoils achieved by increasing thevoltage bias across the detector [71]. The limiting bias voltage is determined by the level at whichthe phonon noise due to leakage current across the detector dominates the sensor noise. In ourfirst tests at Soudan, the maximum E-field was found to be 27V/cm, which corresponds to 90 Vacross a 33.3 mm thick detector. To set the scale of what can be achieved, the low Tc Si prototypementioned above would have a signal 2� above the noise floor for a single e�/h+ pair (�15 eV fornuclear recoils).

Interestingly, the relatively uniform electron-recoil background is reduced because the distri-bution of electron recoil energy is “stretched” to higher energies due to electron recoils producingsignificantly more ionization than a nuclear recoil of the same energy. This results in a factor of �5suppression of the electron-recoil background in the low-energy WIMP signal region.

Since we are now e�ectively measuring the ionization liberated by a nuclear recoil rather thanthe nuclear-recoil phonon energy directly, we are sensitive to the precise ionization calibration scalefor nuclear recoils in this high voltage operational mode, similar to the noble liquid dark matter

10

With this E-field configuration, a recoil event in the bulk of the detector will have e� transportto the narrow positively biased electrode instrumented with charge amplifiers, while the h+ aretransported to the identically instrumented negatively biased electrode on the opposite face, leadingto a symmetric ionization signal for bulk events. For events near a face, the carriers follow thelarge transverse E-fields that run between the interleaved electrodes on the same face leading toasymmetric ionization collection signals. For radial fiducialization, we partition and separatelyreadout the ionization instrumented electrodes on each face into 2 concentric toroids (for a total of4 ionization channels per detector).

X [mm]

Z [m

m]

!2 !1 0 1 2

0

1

2

Figure 3: Left: Phonon and ionization sensor layout for the iZIP detector deployed at Soudan. The two faces areinstrumented with interleaved ionization (40 µm wide) and phonon sensors with �1,mm pitch. The phonon sensorchannel patterns diagrammed in Fig. 2 are visible. Right: Magnified cross section view of electric field lines (red)and equipotential contours (blue) near the bottom face (Z = 0 mm) of a SuperCDMS iZIP detector. The ionizationelectrode lines (yellow) are narrower than the athermal phonon collection sensors (green).

As a side benefit, the large E-fields parallel to the surface of the detector limit charge trappingto the point that very few electron-recoil surface events are indistinguishable from nuclear recoilseven without z fiducialization.

Ionization Detection Through Luke-Neganov GainDuring charge transport, Luke-Neganov phonons are created with a total energy (in eV) equal

to the voltage the carriers travel through. These phonons are measured by our athermal phononsensors, in addition to the intrinsic recoil phonons. In our standard iZIP running mode, we operatewith the smallest bias voltage required to achieve a good measurement of the ionization chargeso that the true phonon recoil signal is not overwhelmed by Luke-Neganov phonons, which woulddegrade the discrimination of electronic and nuclear recoils via ionization yield.

Nuclear recoils from very low-mass WIMPs (M� . 3 GeV/c2) are su⌅ciently small that therelatively uniform (versus energy) electron-recoil background has a reduced overlap with the WIMPsignal region. As a result, it is advantageous to exchange the background discrimination providedby ionization yield for the improved sensitivity to low energy recoils achieved by increasing thevoltage bias across the detector [71]. The limiting bias voltage is determined by the level at whichthe phonon noise due to leakage current across the detector dominates the sensor noise. In ourfirst tests at Soudan, the maximum E-field was found to be 27V/cm, which corresponds to 90 Vacross a 33.3 mm thick detector. To set the scale of what can be achieved, the low Tc Si prototypementioned above would have a signal 2� above the noise floor for a single e�/h+ pair (�15 eV fornuclear recoils).

Interestingly, the relatively uniform electron-recoil background is reduced because the distri-bution of electron recoil energy is “stretched” to higher energies due to electron recoils producingsignificantly more ionization than a nuclear recoil of the same energy. This results in a factor of �5suppression of the electron-recoil background in the low-energy WIMP signal region.

Since we are now e�ectively measuring the ionization liberated by a nuclear recoil rather thanthe nuclear-recoil phonon energy directly, we are sensitive to the precise ionization calibration scalefor nuclear recoils in this high voltage operational mode, similar to the noble liquid dark matter

10

0V +Vb 0V +Vb 0V

e− h+

e−

h+



SuperCDMS: 1990s:








18

experiments. Significantly, in the case of Ge, the ionization scale has been precisely measureddown to nuclear recoil energies of 254 eVr [75] using a novel technique involving capture of thermalneutrons onto 72Ge.4.1.2 Detector Performance at SoudanThe performance of 76 mm diameter iZIP detectors was first studied at the UC Berkeley (UCB)test facility from 2009�2011. The UCB studies yielded extremely promising background rejection,but were limited by cosmogenic neutron backgrounds in the dark matter signal region. Backgroundrejection for events in the dark matter signal region can be measured directly for detectors operateddeep underground. Since 2012, this has been carried out using fifteen iZIP detectors at Soudan.Two of these detectors were installed with a 210Pb source. This isotope, along with its daughters,is one of the primary sources of surface events for SuperCDMS. The deployed sources produceelectrons by beta decay between 10–100 keVnr at a rate of ⇤70 events/hour/source. As seen inFig. 4 (center), these electrons populate a region of reduced ionization yield, which lies betweenthe electron-recoil (yield ⇤1) and nuclear-recoil bands. In addition to electrons, the recoiling 206Pbnucleus from the 210Po alpha decay is also seen with an ionization yield of ⇤0.2. These events occurat a rate ⇤ 25% that of the electrons, with some of the events lying in the nuclear recoil yield bandat low recoil energies. Both the betas and the 206Pb events are distinguished from events in thebulk of the detectors by their asymmetric ionization response. A similar technology has also beendeployed by EDELWEISS [76]. Events in the outer radial regions of the detector, which also su�erfrom reduced yield, are removed by comparing the ionization collected in the outer guard electrodeto that collected on the inner electrode.

Figure 4: (left) Shown are the symmetric charge events (blue dots) in the interior of the crystal, and the eventsthat fail the symmetric charge cut (red dots), including surface events from betas, gammas and lead nuclei incidenton side 1 from the source. (center) Same data, but in the ionization-yield versus phonon recoil-energy plane with±2� ionization-yield range of neutrons indicated (area within green lines). The hyperbolic black line is the ionizationthreshold (1.6 keVee - ‘ee’ for electron equivalent); the vertical black line is the recoil energy threshold (8 keVr).Electrons from 210Pb (below ⇥60 keVr) and 210Bi (mostly above 60 keVr) are distinctly separated from 206Pb recoils(low yield, below ⇥110 keVr). A low-yield outlier (blue with black circle), which is outside the signal region but justsatisfies the charge symmetry requirement is easily removed with a loose phonon symmetry cut (see Fig. 5 right).(right) In addition to the data in left & center, this panel also shows nuclear recoils from a 252Cf neutron source(green). As bulk events, these exhibit a symmetric response between side 1 and 2 like the bulk electron recoils athigher yield, and are thus nicely separated from charge-asymmetric surface events.

Over 2500 live-hours, containing 182,180 betas and 206Pb recoils, have been analyzed from theSoudan run (Fig. 4). A nuclear recoil signal region was defined by the 2� band around the meanyield measured for nuclear recoils (using a 252Cf neutron source). After application of ionization-signal fiducial-volume cuts, no surface events were found in the nuclear recoil signal region above arecoil energy of 8 keVnr. This fiducialization yields a spectrum averaged passage fraction of ⇤50%in the energy range of 8�115 keVr for an ⇤60 GeV/c2 WIMP. The upper limit to the surface eventrejection is < 1.26 ⇥ 10�5 at 90% C.L. This analysis is an update to the work presented in [68].

11



19

1 10 100 1000 10410�5010�4910�4810�4710�4610�4510�4410�4310�4210�4110�4010�3910�3810�37

10�1410�1310�1210�1110�1010�910�810�710�610�510�410�310�210�1




8BNeutrinos



CoGeNT(2012)

CDMS Si(2013)

DAMA

DAMA

CRESST


SuperCDM

S Si HV

SuperCDMS Si HV


SuperCD


LUX (2013)

DarkSide G2LZ

Xenon1TSuperCDMS Ge



CDMSlite (2013)

DAMIC (2012)

PICO250-C3F8

LUX 300-day

PICO250-CF3I



4

spin

-inde

pend

ent,

assu

min

g A

2 sca

ling

spin

-inde

pend

ent,

assu

min

g A

2 sca

ling



19

1 10 100 1000 10410�5010�4910�4810�4710�4610�4510�4410�4310�4210�4110�4010�3910�3810�37

10�1410�1310�1210�1110�1010�910�810�710�610�510�410�310�210�1




8BNeutrinos



CoGeNT(2012)

CDMS Si(2013)

DAMA

DAMA

CRESST


SuperCDM

S Si HV

SuperCDMS Si HV


SuperCD


LUX (2013)

DarkSide G2LZ

Xenon1TSuperCDMS Ge



CDMSlite (2013)

DAMIC (2012)

PICO250-C3F8

LUX 300-day

PICO250-CF3I



4

spin

-inde

pend

ent,

assu

min

g A

2 sca

ling

spin

-inde

pend

ent,

assu

min

g A

2 sca

ling

SuperCDMS 2014



19

1 10 100 1000 10410�5010�4910�4810�4710�4610�4510�4410�4310�4210�4110�4010�3910�3810�37

10�1410�1310�1210�1110�1010�910�810�710�610�510�410�310�210�1




8BNeutrinos



CoGeNT(2012)

CDMS Si(2013)

DAMA

DAMA

CRESST


SuperCDM

S Si HV

SuperCDMS Si HV


SuperCD


LUX (2013)

DarkSide G2LZ

Xenon1TSuperCDMS Ge



CDMSlite (2013)

DAMIC (2012)

PICO250-C3F8

LUX 300-day

PICO250-CF3I



4

spin

-inde

pend

ent,

assu

min

g A

2 sca

ling

spin

-inde

pend

ent,

assu

min

g A

2 sca

ling

SuperCDMS 2014

ionization asymmetryprovides gains hereand above

phonon asymmetry provides gains here

HV operation provides gains here


Innovation in Techniques: 2-Phase Liquid Nobles

20

Rick Gaitskell (Brown) / Dan McKinsey (Yale)LUX Dark Matter Experiment / Sanford Lab

Two-phase Xenon WIMP Detectors

Z position from S1 – S2 timingX-Y positions from S2 light pattern

Excellent 3D imaging (~mm resolution) - eliminates edge events - rejects multiple scatters

Gamma ray, neutron backgroundsreduced by self-shielding

Reject gammas, betas by charge (S2) to light (S1) ratio. Expect > 99.5% rejection.

77

2-Phase Liquid NoblesMultiple realizations ~ 2000

scintillation/ionization (S1/S2) discriminates NRs from ERs in LXe, LAr

scintillation (S1) rise time discriminates NRs from ERs in LAr (and LNe)

LXe has no worrisome isotopes and is highly purifiableprimarily Kr, Rn, and e-attaching

impurities to be worried about

Around 2005Self-shielding could make up for limited

ER rejection (99%-99.9%) of LXeLight collection key to LXe low-mass sensitivityUnderground Ar could provide LAr low in 39Ar beta decay

Very successful program thanks to these innovations:LXe: XENON100, LUX have best limits at high mass; XENON1T to commission this yearLAr: DArkSide 50 recently completed atmospheric Ar commissioning runMulti-ton experiments proposed

Single-phase (S1 only) LAr close to starting to take data (MiniCLEAN, DEAP-3600)



20

Rick Gaitskell (Brown) / Dan McKinsey (Yale)LUX Dark Matter Experiment / Sanford Lab

Electron Recoil and Nuclear Recoil Bands

23

Tritium provides very high statistics electron recoil calibration (200 events/phe)Neutron calibration is consistent with NEST + simulations

Gray contours indicate constant energies using a S1-S2 combined energy scale

23












20

LUX Dark Matter Experiment / Sanford Lab Rick Gaitskell (Brown) / Dan McKinsey (Yale)

And it continues to get quieter - Xe Cosmogenic Activity cools (rate in 44 days)

28

log 1

0 cou

nts/

keV/

kg/d

ay












20










S1 [PE]60 80 100 120 140 160 180 200

F90

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1total_s1_corr_f90_after_lsv_cuts_hist

Entries 2.119474e+07

Mean x 131.2

Mean y 0.3062

RMS x 40.52

RMS y 0.05257

0

100

200

300

400

500

600

700total_s1_corr_f90_after_lsv_cuts_hist

Entries 2.119474e+07

Mean x 131.2

Mean y 0.3062

RMS x 40.52

RMS y 0.05257

Background free exposure of 280 kg·day

80%65%

50%

90%

DArkSide-50 atmospheric Ar run: equivalent to 2.6 yr,discrimination from pulse shape only



20










[GeV]rM21010 310

]2 [cm

m

-4510

-4410

-4310

-4210

Experimental limitsDarkSide50 - 3 yDarkSide50 - 2.6 yXenon100LUXpMSSM (post LHC)

DS-50 projected sensitivity (90% C.L.)

Projected sensitivity evaluated assuming: • the measured PSD performance; • no rejection from S2/S1; • fiducialization along z axis-only; • zero neutron-induced events; • NR quenching and F90 acceptance

curves from SCENE @ 200V/cm

Present systematics on NR Quenching and F90 NR acceptance curves cause a ~10% variation of the projected sensitivity around 100 GeV/c2

.

Threshold 35 keVR

Fiducial mass 44.1 kg LY=8.0 PE/keVee @ null field NR Quenching from SCENE F90 NR acceptance function of ER



21

1 10 100 1000 10410�5010�4910�4810�4710�4610�4510�4410�4310�4210�4110�4010�3910�3810�37

10�1410�1310�1210�1110�1010�910�810�710�610�510�410�310�210�1




8BNeutrinos



CoGeNT(2012)

CDMS Si(2013)

DAMA

DAMA

CRESST


SuperCDM

S Si HV

SuperCDMS Si HV


SuperCD


LUX (2013)

DarkSide G2LZ

Xenon1TSuperCDMS Ge



CDMSlite (2013)

DAMIC (2012)

PICO250-C3F8

LUX 300-day

PICO250-CF3I



4

spin

-inde

pend

ent,

assu

min

g A

2 sca

ling

spin

-inde

pend

ent,

assu

min

g A

2 sca

ling

improved radiopurity,use of self-shielding

light collection


Innovation in Techniques: Bubble Chambers

Bubble ChambersClassic Seitz bubble theory gave incredible

discrimination against ERs

But: alphas from Rn contamination→ acoustic measurements discriminate alphas

Higher threshold, poorer F (and C) recoil efficiency than desired in CF3I→ develop C3F8

No energy informationDevelop bubble chambers with scintillating materials

22

photographic images appear in sharp contrast to a retrore-flective background and are identified in the image analysisalgorithm as clusters of pixels that have changed signifi-cantly between consecutive frames. Reconstruction of thedata from two stereo views provided the spatial coordinatesof the bubble to a typical accuracy of a few millimeters,depending on the proximity of the bubble to the cameras.

The pressure rise analysis was based on data from an ac-coupled fast pressure transducer [14] which was sampled at10 kHz for 160 msec around the onset of a nucleation.Empirically, the rate of pressure rise was well fit by asimple quadratic time dependence for bubbles formed inthe bulk of the target fluid. The quadratic coefficient of thefit was found to be proportional to the number of bubbles inthe event, and the quality of the fit was uniform over thevolume of the experiment except near the boundaries.Because bubble growth is affected by the proximity ofthe bubble to the quartz vessel walls or the CF3I waterinterface, the quality of the quadratic fit deteriorated rap-idly for bubbles near a boundary. The sensitivity of thebubble growth to the proximity of a boundary was studiedusing calibration neutron events where it was found that amodest cut on the chi-square reliably identified events thatwere near the vessel walls or the CF3I water interface. Thepressure growth chi-square cut effectively provided a fidu-cial volume definition that was uniform around the perime-ter of the chamber and performed somewhat better for thispurpose than the stereo reconstruction of the camera im-ages. The pressure growth fit was therefore used to providethe formal fiducial volume cut for the experiment.

The third and final element of event reconstruction wasthe evaluation of the acoustic signals and classification ofevent types. The acoustic transducer signals were digitizedwith a 2.5-MHz sampling rate and recorded for 40 msec foreach event. The signals were filtered using a single-polehigh-pass filter with a cutoff at 500 Hz, and a low-pass anti-aliasing filter cutting off at 600 kHz. The preevent baselinefor each of the acoustic signals was examined to determinethe time of bubble formation, t0. A fast Fourier transformwas constructed for the times t0 ! 1 msec< t < t0 þ9 msec. The sound of bubble nucleation showed a broademission distinctly above background noise up to a fre-quency of 250 kHz. The acoustic signature for a singlerecoiling nucleus was calibrated by studying events initi-ated by neutron sources. The acoustic power was observedto vary slightly with the position of the bubble within thechamber, and the position dependence was found to varywith frequency. To account for the position and frequencydependence, the acoustic signal was analyzed separatelyin four frequency bands (1.5–12, 12–35, 35–150, or150–250 kHz) which were separately corrected for spatialdependence and normalized. The acoustic event discrimi-nation was based on a single acoustic parameter AP [8]which is a frequency weighted acoustic power densityintegral, corrected for sensor gain and bubble position:

AP ¼ AðTÞX

j

Gj

X

n

Cnð ~xÞXfnmax

fnmin

f& psdjf; (1)

where AðTÞ is an overall temperature dependent scalefactor, Gj is the gain of acoustic transducer j, Cnð ~xÞ isthe correction factor for the bubble position dependence infrequency bin n, ~x is the position of the bubble, f isfrequency, fmin and fmax are the boundaries of the fre-quency band, and psdjf is the power spectral density for

the bin with center frequency f for sensor j. The AP wasscaled to have a value of unity at the peak observed in itsdistribution for nuclear recoils induced by neutron sourcesas shown in Fig. 2. The clear separation seen between thealpha peak and the single nuclear recoil peak in Fig. 2illustrates the power of the acoustic discrimination toeliminate alpha emitter contamination as a source of back-ground for the experiment.All data have been subject to a set of data quality cuts

including the requirement that the chamber expand suc-cessfully to the desired operating pressure and be stable forgreater than 30 sec prior to the event. Other quality cutseliminate events with acoustic noise prior to the event andevents in which the video trigger failed to capture theinitiation of the bubble. The fiducial volume, determinedby analyzing the acceptance of the pressure growth fit cutfor events initiated with a neutron source, is 92:1' 1:8%,equivalent to removing the outer 2 mm of the liquidvolume. This fiducial volume was consistent within statis-tical errors over all neutron calibration data. The overallefficiency for all data quality and fiducial volume cuts is82:5' 1:9%, independent of operating temperature. The

−1 0 1 2 3ln(AP)

Cou

nts/

0.05

AP

AmBe data sampleWIMP search data

20 recoil− like events

2474alphas

Acousticcut, 96%acceptance

0

20

40

60

80

100

120

FIG. 2 (color online). Data from a 553 kg-day WIMP search,shown as a distribution in lnðAPÞ as the solid red line. Twentysingle nuclear recoil event candidates and 2474 alpha eventswere observed. The dash-dotted blue histogram shows the iden-tical analysis for data taken in the presence of an AmBe neutronsource. We define an acoustic cut of 0:7<AP< 1:3 to selectnuclear recoils with an acceptance of 95.8% as determined by theAmBe calibration.

FIRST DARK MATTER SEARCH RESULTS FROM A 4-KG . . . PHYSICAL REVIEW D 86, 052001 (2012)

052001-3

Understanding+low+efficiency+from+fluorine+recoils+

TAUP%September%9th,%2013% Russell%Neilson,%University%of%Chicago% 22%

SRIM%simulation%of%15keV%F%recoils.%%

track length/bubble nucleation radius

F, C recoils in CF3I generally deposit energy over too large a

distance to nucleate bubbles. Better in C4F10 (and C3F8).

PICO%2L:)a)low%mass)WIMP)search)

Preliminary*results*suggest*the*same*10210*gamma*rejection*is*possible*with*C3F8,*and*at*a*lower*nuclear*recoil*threshold.*A*lower*threshold*extends*the*sensitivity*to*lower*mass*WIMPs.*

TAUP%September%9th,%2013% 14%Russell%Neilson,%University%of%Chicago%

Bubble%nucleation%probability%from%gamma%interactions%in%C3F8%and%CF3I%

Excellent discrimination against ERs, but at lower

threshold for C3F8 and CF3I

ER B

ubbl

e N

ucle

atio

n Pr

obab

ility

acoustic parameter


Innovation in Techniques: Bubble Chambers

23

1 10 100 1000 10410�5010�4910�4810�4710�4610�4510�4410�4310�4210�4110�4010�3910�3810�37

10�1410�1310�1210�1110�1010�910�810�710�610�510�410�310�210�1




8BNeutrinos



CoGeNT(2012)

CDMS Si(2013)

DAMA

DAMA

CRESST


SuperCDM

S Si HV

SuperCDMS Si HV


SuperCD


LUX (2013)

DarkSide G2LZ

Xenon1TSuperCDMS Ge



CDMSlite (2013)

DAMIC (2012)

PICO250-C3F8

LUX 300-day

PICO250-CF3I



4

spin

-inde

pend

ent,

assu

min

g A

2 sca

ling

spin

-inde

pend

ent,

assu

min

g A

2 sca

ling

gains from acousticdiscrimination

gains from acoustic discriminationand from use of C3F8


Innovation in Techniques: Addressing DAMA

DAMA annual modulation asore point for communityHuge statistical significance

No other existing expt uses Na and I

New efforts to test underway!DM-ICE: NaI with different systematics

southern hemisphere, situated inside IceCube

also a movable copy: run in N and Soperation in ice demo’d, ice v. cleanworking on reducing contaminations

SABRE: NaI with reduced backgroundsBetter source powder for NaILower radioactivity photomultipliersBetter light collectionLower radioactivity housingsSurrounded in liquid scintillator to reject backgrounds (esp. 3 keV 40K escape peak)

24

Dark Matter Searches Rick Gaitskell, Brown University, LUX / DOE

•Would like to see the Multiples presented as a ratio to the underlying rate, rather than an absolute. ◆Also add the error bars for multiples

!55 140226_DM2014_Gaitskell_Review_v05 - February 28, 2014

UCLA'Dark'Matter:'February'2014'Walter'C.'Pettus' 8'

Dominant'LE'features:'•  3'keV'40K'Auger'e4/X4rays'•  (crystal)'

•  ~14'keV'238U4chain'X4rays'•  (copper'encapsulation'surface)'

•  46.5'keV'210Pb'gamma'•  (crystal)'

•  65.3'keV'125I'gamma+X4ray'•  (crystal,'cosmogenic)'

DAMA achieved bgnd

DM-ICE

UCLA'Dark'Matter:'February'2014'Walter'C.'Pettus' 8'

Dominant'LE'features:'•  3'keV'40K'Auger'e4/X4rays'•  (crystal)'

•  ~14'keV'238U4chain'X4rays'•  (copper'encapsulation'surface)'

•  46.5'keV'210Pb'gamma'•  (crystal)'

•  65.3'keV'125I'gamma+X4ray'•  (crystal,'cosmogenic)'


Expected SABRE Background

19

* This spectrum was made using NaI powder radioactivity; crystal can be better.

* External background is estimated to be relatively small compared to internal.








24












24




]2 [GeV/cΧm10 210

]2 [c

mSI

σ

-4210

-4110

-4010

-3910 = 1 0N = 2 0N = 5 0N

Threshold = 2.0 keVee

]2 [GeV/cΧm10 210

]2 [c

mSI

σ

-4210

-4110

-4010

-3910 = 0.5 0N = 1 0N = 2 0N

Threshold = 4.0 keVee

Figure 3: The curves show the sensitivity of hypothetical 500 kg-year exposures with varying total event rates (in cpd/kg/keVee). Shown aretwo energy threshold scenarios. The left plot shows sensitivities with a 2 keVee experimental threshold. The right plot shows sensitivity with a4 keVee threshold. The gray regions show the 90% (dark) and 99.7% (light) DAMA/LIBRA allowed regions for interactions with Na (masses of⇠10 GeV/c2) and I (masses of ⇠100 GeV/c2). DAMA/LIBRA allowed regions are calculated without channeling.

for the time dependent rate shown in equation 3 remainsunchanged.

Experimental data are typically displayed using thecalibrated electron-recoil equivalent energy (keVee) be-cause NaI cannot provide event-by-event discrimina-tion between nuclear and electron recoils. Working inthe same units, we present results in two intervals inelectron-recoil equivalent energy: [2, 6] keVee and [4, 6]keVee. To convert the energy ranges defined in the ex-perimentally measured units into recoil energy for theappropriate target nucleus, we use q = 0.3 keVee/keVfor Na recoils and q = 0.09 keVee/keV for I recoils,where EkeVee = qErecoil and q is the quenching factortaken from [39]. The energy threshold of 2 keVee ischosen based on the analysis thresholds and trigger ef-ficiencies achieved by previous experiments such asDAMA/LIBRA [10] and NAIAD [17]. The higher4 keVee threshold provides a comparison to a slightlymore pessimistic scenario. The upper bound is 6 keVee.Above this, the rate for recoils, in a standard halo model,drops well below the considered background rates of0.5 � 5.0 cpd/kg/keVee. In order to compare to DAMAdata, we define the constant rate of events, N0, which isthe sum of the time-independent dark matter component(R0) and the time-independent background.

The sensitivity is calculated by fitting equation 3 to asimulated dataset containing no modulation signal in it(null hypothesis). The simulated data are created by ran-domly drawing events from a Poisson distribution withmean given by N0. We report on several trials with N0

ranging from 0.5 to 5 cpd/kg/keVee. The value N0 = 0.5corresponds to an optimistic goal of achieving a betterbackground rate compared to DAMA/LIBRA (N0 ⇡ 1).The highest rate (N0 = 5) is characteristic of the back-ground levels achieved by NAIAD [17]. To obtain arepresentative statistical uncertainty, the total number ofthrown events corresponds to the average number ex-pected for a 250 kg exposure over two years, integratedover the experimental energy range. For a given value ofN0, the simulated experiment is performed for the twoenergy intervals listed above.

A simple chi-square fit allows S 0 and S m to floatwhile keeping the frequency, !, and the phase, tc, fixed.Under the assumption of a null result, the 90% C.L. up-per limit to S m is calculated from the error returned onthe fit. The resulting sensitivities, in the form of 90%C.L. upper limits to the spin independent cross-section,are shown in Figure 3. For comparison, the calculatedDAMA/LIBRA 90% and 99.7% allowed regions, with-out channeling [40], are also displayed on the figures.

The results demonstrate that low energy thresholds of⇠2 keVee, or lower, are critical for obtaining adequatesensitivity to constrain the DAMA/LIBRA allowed re-gions. Figure 3 shows that with a 2 keVee threshold,an experiment that has 5 times the DAMA event rate(and hence ⇠5 times greater background) provides sig-nificant constraints with two years of exposure. Con-sidering the sensitivity curves for the higher thresholdscenario, it is clear that control of background becomesmore important. With a 4 keVee energy threshold, one

6

bgnd in counts/kg/keV/day:

, 500 kg-yr (2 yrs)


Innovation in Techniques: Directional Detection

Demonstratorscontinue to makegood progress

DMTPC:Improved range

reconstructionprovides betterhead-tail sensitivity, critical for directional signal

Scaling up to 1 m3, running 1L prototype at WIPP

DRIFT“Minority carriers”

have different speed,provides t0 andrejection of surfacebackgrounds

Deploying DRIFT IIetoward DRIFT III

25

Reconstruction Improvements

Fitting improves reconstructionof range, energy, angle andsense.

Clear way to make cuts

Possible to calculate relativelikelihoods of each sense

Cosmin Deaconu (MIT/LNS) DMTPC UCLA 2014 9 / 18

Reconstruction Improvements

Fitting improves reconstructionof range, energy, angle andsense.

Clear way to make cuts

Possible to calculate relativelikelihoods of each sense


5 mm = 210 keVr013303-4 Daniel P. Snowden-Ifft Rev. Sci. Instrum. 85, 013303 (2014)

0 1 2 3 4 5 6

0.00

00.

001

0.00

20.

003

0.00

40.

005

Time (ms)

Cur

rent

(ar

bitra

ry u

nits

)

FIG. 5. A 210Po alpha particle event in the detector filled with 40 Torr CS2and 0.94 ± 0.01 Torr of O2 in drift field of 293 V/cm at T = 23.4 ◦C. Asdiscussed in the text the rapid collection of charge at the beginning of theevent provides a t0 to fiducialize the alpha particle. The inferred currents onthe 4 monitored lines (spaced at 4 mm) were offset by 0.001 for presentationpurposes.

separation from the I peak decreases with distance to theMWPC, as expected,

!tS,P =!

1µI

− 1µS,P

"!P

E

"z, (3)

where z is the distance between the ionization and the MPWC.For the earliest ionization, the minority peaks are completelyhidden within the I peak. The large spike in ionization att = 0 is caused by the alpha particle traveling through theMWPC and the rapid accumulation of charge in the high-field, 1.1 cm, grid-anode gap. This distinct, rapid collectionof charge fiducializes the alpha particle; the ionization att = 0 was created at z = 0. The arrival time of an I peakrelative to this t0 provides an accurate, triggered measure ofthe average z of the ionization segment, z0. On the other handthe separation between the S and P peaks and the I peak pro-vides another, trigger-less, estimation of the average z of theionization segment, zMP through Eq. (3). Figure 6 shows therelation between z0 and zMP.

For this analysis the position of all the peaks were mea-sured by eye so for ionization segments near the detector nomeasurement was possible. A sophisticated peak-finding al-gorithm would certainly improve this result. For this resultthe agreement between the two measurements is quite goodz >∼ 7.5 cm. The importance of this result is that for a sin-gle ionization segment created, for instance by a neutron orWIMP recoil in DRIFT, where no trigger is possible, i.e.,where the time of ionization creation, t0, is unknown, a mea-surement of zMP is still possible. For DRIFT this will allowthe removal of its only known backgrounds, events from theMWPC and central cathode, without undue loss of effectivevolume.

0 5 10 15

05

1015

z0 (cm)

z MP (

cm)

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

FIG. 6. The distance, z, between an alpha created ionization segment andthe MWPC measured in 2 ways. In the first the time difference between thearrival of the I peak and a t0, discussed in the text, provides one measure of z,z0. In the second the time difference between either the S peak or the P peakand the I peak, Eq. (3) and the values in Table I provide a second measure ofz, zMP. z0 vs zMP is shown on the plot with the S peak data shown with opencircles and the P peak data shown with filled circles. The vertical and hori-zontal lines at 15.24 cm show the z dimension of the detector. At least one ofthe measured alphas hit the cathode while over one of the measurement lines.The line drawn with a slope of 1 shows the strong agreement between thesetwo ways of measuring z. The measurement of zMP suffers from lower reso-lution. Assuming the resolution of z0 is much greater than zMP the inferredRMS for the measurement of z from the S peak is σ S = 0.67 cm and for theP peak σP = 0.33 cm, the difference due simply to the larger separation forthe P peak. It also suffers from the inability to measure events close to theMWPC. However, the measurement of z using the minority peaks has thedistinct advantage of not requiring a trigger.

CONCLUSION

In conclusion, the addition of O2 to gas mixtures in-cluding CS2 and in the presence of ionization produces atleast 2 distinct species of drifting anions in addition to thenormal, drifting CS2 anion. The various measurements pre-sented above on these other anions may help to identify them.Aside from their inherent interest, the existence of multipleionization-created anions is already proving to have enormouspractical importance offering, as it does, a new trigger-lessmode of operation for drift chambers.

ACKNOWLEDGMENTS

This work is supported by the National Science Founda-tion. The author is very grateful to Dr. J. L. Roberts and tothe DRIFT collaboration and, in particular, Alex Murphy foruseful discussions and comments in the development of thiswork.

1G. F. Knoll, Radiation Detection and Measurement (John Wiley & Sons,New York, 1989).

2C. J. Martoff et al., Nucl. Instrum. Methods Phys. Res. A 440, 355(2000).

3T. Ohnuki, C. J. Martoff, and D. Snowden-Ifft, Nucl. Instrum. MethodsPhys. Res. A 463, 142 (2001).

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to IP:131.215.220.164 On: Thu, 17 Apr 2014 21:14:44

013303-2 Daniel P. Snowden-Ifft Rev. Sci. Instrum. 85, 013303 (2014)

The main practical benefit of the minority peaks is, anal-ogous to the fiducialization of earthquakes locations using Sand P waves, trigger-less fiducialization of events in the gas.The DRIFT dark matter experiment, for instance, is heavilylimited in its search for dark matter recoils by radon progenyrecoils from its central cathode.5 Without a trigger DRIFTwas forced to use the root-mean-square (RMS) in time of theobserved events as a crude method of fiducialization but thisentailed a factor of ∼25 reduction in its detection efficiency.Were it not for their small size the ability to fiducialize eventswith the minority peaks would have a huge impact on DRIFT.Trigger-less operation could benefit many other drift chamberexperiments.

PEAK ENHANCEMENT

Fortunately, the addition of a small amount of O2 to thegas mixture has been found to greatly increase the size of theminority peaks, as presented in Fig. 1.

It will be shown below that the I carriers in mixtures withsmall additions of O2 are the same as the I carriers observed inprevious experiments. Only 3 minority peaks have been iden-tified thus far. The appearance of the D peak is inconsistentfrom run to run so this paper will focus solely on the proper-ties of the S and P peaks.

It is now thought that the appearance of the miniscule Sand P minority peaks in Ref. 4 was due to a small concen-tration of O2 in the gas due to outgassing. Previous mobil-ity measurements have been performed with gas mixtures ofCS2 and CF4, Ar, Xe, CH4, He, Ne, and CO2

2–4, 6, 8 and therehave been no reports of significant minority peak formation inthose gas mixtures. That they appear with the addition of O2

is a clue, a convenient means of studying them and a practicalmeans for taking advantage of them. The purpose of this paper

FIG. 1. The arrival time distribution of negative ions after a 15.24 cm driftin a 273 V/cm drift field in a mixture of 30 Torr CS2, 10 Torr CF4, and1 Torr O2. Following earthquake fidudialization and spectroscopic notationthe minority peaks are labeled as shown.

is to characterize the behavior of the S and P minority carri-ers to begin an understanding of their identity and formationmechanism.

PEAK HEIGHT

Since the widths of the peaks are identical the heightserves as a convenient proxy for the total amount of post-avalanche charge in the peak. The first experiment for thispaper to study the charge distribution of the minority peaksused a UV spot positioned on the cathode over one of thewires at a fixed drift field but with varying concentrations ofO2. Data were taken first with a base gas of 40 Torr CS2. O2

was then added to the gas mix in ∼0.2 Torr increments up to∼1.0 Torr added O2. The experiment was then repeated with a30–10 Torr CS2–CF4 gas mixture as the base. The increase ingas pressure and changing gas mixture changed the gas gainon the wire. For this reason the ratio of the heights of the mi-nority peaks relative to the I peak is displayed in Fig. 2.

As Fig. 2 makes clear the formation of the S and P mi-nority carriers requires the presence of O2. The linear de-pendence shown in Fig. 2 suggests a formation mechanisminvolving only a single O2 molecule, consistent with theBlock-Bradbury attachment mechanism. The percentagesshown in Fig. 2 for the P peak are identical for the differentgas mixtures, with different CS2 partial pressures, but differ-ent for the S peaks suggesting that the formation mechanismsfor the S and P carriers are different.

LIFETIME

The S carriers appear to have a finite lifetime. In this ex-periment a UV spot was, again, positioned on the cathode over

0.0 0.2 0.4 0.6 0.8 1.0 1.2

010

2030

4050

6070

O2 partial pressure (Torr)

S a

nd P

hei

ghts

rel

ativ

e to

I he

ight

(%

)

S

P

FIG. 2. The size of the S (square data points) and P (diamond data points)minority peaks relative to the size of the I peak for varying partial pressuresof O2 added to either 40 Torr CS2, shown in brown, or 30–10 Torr CS2–CF4,shown in green. Both measurements were done with a 15.24 cm drift distance.The CS2 measurements were done in a drift field of 299 V/cm while the CS2–CF4 measurements were done in a field of 314 V/cm. A linear fit is shownfor each data set.


z true [cm]

z fro

m m

inor

ity c

arrie

rs [c

m]

DRIFT pulseshowing minority

carrier drift










25

Scaling up to a m3 fiducial volume

Vacuum chamber arriving in twomonths (diameter: 5 feet, length: 7feet)

50 times more fiducial volume thancurrent largest prototype (150 g at 30Torr)

New amplification region design (triplemesh) allows imaging two TPC’s fromone side

4 TPC’s, 8 cameras (4 for each pair ofback to back TPC’s)









0 1 2 3 4 5 6

0.00

00.

001

0.00

20.

003

0.00

40.

005

Time (ms)

Cur

rent

(ar

bitra

ry u

nits

)

FIG. 5. A 210Po alpha particle event in the detector filled with 40 Torr CS2and 0.94 ± 0.01 Torr of O2 in drift field of 293 V/cm at T = 23.4 ◦C. Asdiscussed in the text the rapid collection of charge at the beginning of theevent provides a t0 to fiducialize the alpha particle. The inferred currents onthe 4 monitored lines (spaced at 4 mm) were offset by 0.001 for presentationpurposes.

separation from the I peak decreases with distance to theMWPC, as expected,

!tS,P =!

1µI

− 1µS,P

"!P

E

"z, (3)

where z is the distance between the ionization and the MPWC.For the earliest ionization, the minority peaks are completelyhidden within the I peak. The large spike in ionization att = 0 is caused by the alpha particle traveling through theMWPC and the rapid accumulation of charge in the high-field, 1.1 cm, grid-anode gap. This distinct, rapid collectionof charge fiducializes the alpha particle; the ionization att = 0 was created at z = 0. The arrival time of an I peakrelative to this t0 provides an accurate, triggered measure ofthe average z of the ionization segment, z0. On the other handthe separation between the S and P peaks and the I peak pro-vides another, trigger-less, estimation of the average z of theionization segment, zMP through Eq. (3). Figure 6 shows therelation between z0 and zMP.

For this analysis the position of all the peaks were mea-sured by eye so for ionization segments near the detector nomeasurement was possible. A sophisticated peak-finding al-gorithm would certainly improve this result. For this resultthe agreement between the two measurements is quite goodz >∼ 7.5 cm. The importance of this result is that for a sin-gle ionization segment created, for instance by a neutron orWIMP recoil in DRIFT, where no trigger is possible, i.e.,where the time of ionization creation, t0, is unknown, a mea-surement of zMP is still possible. For DRIFT this will allowthe removal of its only known backgrounds, events from theMWPC and central cathode, without undue loss of effectivevolume.

0 5 10 15

05

1015

z0 (cm)

z MP (

cm)

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

FIG. 6. The distance, z, between an alpha created ionization segment andthe MWPC measured in 2 ways. In the first the time difference between thearrival of the I peak and a t0, discussed in the text, provides one measure of z,z0. In the second the time difference between either the S peak or the P peakand the I peak, Eq. (3) and the values in Table I provide a second measure ofz, zMP. z0 vs zMP is shown on the plot with the S peak data shown with opencircles and the P peak data shown with filled circles. The vertical and hori-zontal lines at 15.24 cm show the z dimension of the detector. At least one ofthe measured alphas hit the cathode while over one of the measurement lines.The line drawn with a slope of 1 shows the strong agreement between thesetwo ways of measuring z. The measurement of zMP suffers from lower reso-lution. Assuming the resolution of z0 is much greater than zMP the inferredRMS for the measurement of z from the S peak is σ S = 0.67 cm and for theP peak σP = 0.33 cm, the difference due simply to the larger separation forthe P peak. It also suffers from the inability to measure events close to theMWPC. However, the measurement of z using the minority peaks has thedistinct advantage of not requiring a trigger.

CONCLUSION

In conclusion, the addition of O2 to gas mixtures in-cluding CS2 and in the presence of ionization produces atleast 2 distinct species of drifting anions in addition to thenormal, drifting CS2 anion. The various measurements pre-sented above on these other anions may help to identify them.Aside from their inherent interest, the existence of multipleionization-created anions is already proving to have enormouspractical importance offering, as it does, a new trigger-lessmode of operation for drift chambers.

ACKNOWLEDGMENTS

This work is supported by the National Science Founda-tion. The author is very grateful to Dr. J. L. Roberts and tothe DRIFT collaboration and, in particular, Alex Murphy foruseful discussions and comments in the development of thiswork.

1G. F. Knoll, Radiation Detection and Measurement (John Wiley & Sons,New York, 1989).

2C. J. Martoff et al., Nucl. Instrum. Methods Phys. Res. A 440, 355(2000).

3T. Ohnuki, C. J. Martoff, and D. Snowden-Ifft, Nucl. Instrum. MethodsPhys. Res. A 463, 142 (2001).



The main practical benefit of the minority peaks is, anal-ogous to the fiducialization of earthquakes locations using Sand P waves, trigger-less fiducialization of events in the gas.The DRIFT dark matter experiment, for instance, is heavilylimited in its search for dark matter recoils by radon progenyrecoils from its central cathode.5 Without a trigger DRIFTwas forced to use the root-mean-square (RMS) in time of theobserved events as a crude method of fiducialization but thisentailed a factor of ∼25 reduction in its detection efficiency.Were it not for their small size the ability to fiducialize eventswith the minority peaks would have a huge impact on DRIFT.Trigger-less operation could benefit many other drift chamberexperiments.

PEAK ENHANCEMENT

Fortunately, the addition of a small amount of O2 to thegas mixture has been found to greatly increase the size of theminority peaks, as presented in Fig. 1.

It will be shown below that the I carriers in mixtures withsmall additions of O2 are the same as the I carriers observed inprevious experiments. Only 3 minority peaks have been iden-tified thus far. The appearance of the D peak is inconsistentfrom run to run so this paper will focus solely on the proper-ties of the S and P peaks.

It is now thought that the appearance of the miniscule Sand P minority peaks in Ref. 4 was due to a small concen-tration of O2 in the gas due to outgassing. Previous mobil-ity measurements have been performed with gas mixtures ofCS2 and CF4, Ar, Xe, CH4, He, Ne, and CO2

2–4, 6, 8 and therehave been no reports of significant minority peak formation inthose gas mixtures. That they appear with the addition of O2

is a clue, a convenient means of studying them and a practicalmeans for taking advantage of them. The purpose of this paper

FIG. 1. The arrival time distribution of negative ions after a 15.24 cm driftin a 273 V/cm drift field in a mixture of 30 Torr CS2, 10 Torr CF4, and1 Torr O2. Following earthquake fidudialization and spectroscopic notationthe minority peaks are labeled as shown.

is to characterize the behavior of the S and P minority carri-ers to begin an understanding of their identity and formationmechanism.

PEAK HEIGHT

Since the widths of the peaks are identical the heightserves as a convenient proxy for the total amount of post-avalanche charge in the peak. The first experiment for thispaper to study the charge distribution of the minority peaksused a UV spot positioned on the cathode over one of thewires at a fixed drift field but with varying concentrations ofO2. Data were taken first with a base gas of 40 Torr CS2. O2

was then added to the gas mix in ∼0.2 Torr increments up to∼1.0 Torr added O2. The experiment was then repeated with a30–10 Torr CS2–CF4 gas mixture as the base. The increase ingas pressure and changing gas mixture changed the gas gainon the wire. For this reason the ratio of the heights of the mi-nority peaks relative to the I peak is displayed in Fig. 2.

As Fig. 2 makes clear the formation of the S and P mi-nority carriers requires the presence of O2. The linear de-pendence shown in Fig. 2 suggests a formation mechanisminvolving only a single O2 molecule, consistent with theBlock-Bradbury attachment mechanism. The percentagesshown in Fig. 2 for the P peak are identical for the differentgas mixtures, with different CS2 partial pressures, but differ-ent for the S peaks suggesting that the formation mechanismsfor the S and P carriers are different.

LIFETIME

The S carriers appear to have a finite lifetime. In this ex-periment a UV spot was, again, positioned on the cathode over

0.0 0.2 0.4 0.6 0.8 1.0 1.2

010

2030

4050

6070

O2 partial pressure (Torr)

S a

nd P

hei

ghts

rel

ativ

e to

I he

ight

(%

)

S

P

FIG. 2. The size of the S (square data points) and P (diamond data points)minority peaks relative to the size of the I peak for varying partial pressuresof O2 added to either 40 Torr CS2, shown in brown, or 30–10 Torr CS2–CF4,shown in green. Both measurements were done with a 15.24 cm drift distance.The CS2 measurements were done in a drift field of 299 V/cm while the CS2–CF4 measurements were done in a field of 314 V/cm. A linear fit is shownfor each data set.


z true [cm]

z fro

m m

inor

ity c

arrie

rs [c

m]

DRIFT pulseshowing minority

carrier drift










25













• Preference for CH-based material

• Unit cell composed = 8 m3

• 250 of 4 kg modules gives 1 ton would fit into a 500m tunnel at Boulby

DRIFT III modularity

• Modular design, 3 unit cells give 4 kg target, 24 m3

Friday, February 28, 14


A View to the Future

We would like to see:Accelerator production via multiple consistent channels

Indirect detection in multiple channels with consistent parametersDirect detection in multiple targetsEventually, direct detection with

recoiling particle directionality

These will tell us:The couplings of dark matter to a

variety of normal matter particlesThe local density and velocity structure

of the dark matter haloThe dark matter abundance globally

in the halo of our galaxy and innearby galaxies

Is what we detect enough?

A lot of work to do, and will require broad interactions (between people as well as particles)

26

Quarks

W

LeptonsχGluons

Photons

Z

Higgs

LHC

Direct Scattering

Gamma Rays

Neutrinos

ILC?

LEP

Anti-matter

Tait

Experimental Direct Detection of Dark Matter: Present and...

Documents

Transcript of Experimental Direct Detection of Dark Matter: Present and...