The Large Scale U.S. Dark-Matter Axion Search

Darin Kinion (UCB/LLNL)Les Houches

January 31, 2002

Talk Outline

• Brief review of the axion

• The Sikivie microwave cavity technique

• The U.S. large-scale experiment

• R&D on SQUID amplifiers

• Summary

Brief summary of the Axion

• The Axion is a light pseudoscalar resulting from the PQ-mechanism to enforce strong-CP conservation

• fa, the SSB scale of PQ-symmetry, is the one important parameter in the theory

P01545-kvb-u-004

1012 GeVfa

( (•

Mass and Couplings

Axion Mass ‘Window’

ma ~ 6 µeV 5 µeVma

αgγπfa

( (

Cosmological Abundance

Ωa ~ 7/6

gaγγ = ; gγ =0.97 KSVZ–0.36 DFSZ

(Vacuum misalignmentmechanism)

10–(5 to 6) eV < ma < 10–(2 to 3) eV

(Overclosure) (SN1987a)

With lower end of window preferred if ΩCDM ~ 1

How to detect dark-matter axions (Sikivie, 1983)

Primakoff Conversion Resonant Conversion: hν = mac2[1 + O(β2)]

Signal

a

Virtualphoton

Singlerealphoton

B0

∆νν

Power

Frequency (GHz)

P01545-kvb-u-005

Psig ~ (5×10–22W) •B

7.6T

sn

V220l

gγ0.97

ρa

0.45 GeV/cm3

ma

3 µeV( ((( (

( ( ( (( ((• •

• •

•

Dicke’s Radiometer Eqn. → Integration Time

t∆ν

PsigkTS

•

•

= ; TS = T + TN

Present exp’t: T ~ TN ~ 1.5 K

dνdt

Scaling Laws

dνdt

1T2

1TS

For fixed model g2

This is a narrow-band experiment. There is no other way to get the required sensitivity!

For fixed scan rate

B4V2∝ g2 ∝ B2V –1~ 10–6

2 2

γ

S

Layout of the Axion Experiment

3.6

met

ers

Microwave Cavity

Superconducting magnet

Cavity LHe reservoir

Dielectric tuning rod

Metal tuning rod

Amplifiers

Vacuum Pump

Magnet LHe reservoir

Cryostat vessel

Tuning mechanism

Stepping motors

Cavity vacuum chamber

1.3K J-T refrigerator

Magnet support

Axion Detector Electronics

Bo

125 Hz Bin FFT

0.02 Hz Bin FFT

35 kHZ AF

5 kHZ AF

RF10.7 MHz IF

NETWORK ANALYZER

T,R

DRTD

MAGNET

CAVITY & TUNING RODS

HFET AMP.

DIR. COUPLER

1.5 K

SWEEP GENERATOR

SWITCH SETTINGS: D = Data Taking T = Transmission Measurement R = Reflection Measurement

Cs Clock 10 MHz Timebase

L.O. #1 L.O. #2

L.O. #3

I.R. MIXER

CRYSTAL FILTER

CRYSTAL FILTER

MIXER #2

MIXER #3

DISK

ma

Power

freq

∆ΕΕ - 10 -6

∆ΕΕ - 10 -17

Expected Signal

Brief outline of analysis

P01545-kvb-u-008

Data, with Theoretical Curve(Gaussian noise through receiver and analysis)

S/N > 4 for Thermalized KSVZ Axion

• Each frequency appears in >45 subspectra

• Weighted and co-added to produce spectrum

• 800,000 bins (125 Hz)/100 MHz

→ 6535 candidates > 2.25 √6 σ (95% C.L.)→ Rescan all to same sensitivity→ 23 candidates (Net 90% C.L.)→ Each examined: radio peaks

For a persistent peak, the ultimatetest is to turn off the magnet!700

10–23

10–22

10–21

720

KSVZ signal power

125 Hz

Noise in:750 Hz (∆E/E = 10-6)

740Frequency (MHz)

Po

wer

/125

Hz

(W)

760 780 800

–6 –4 –2 00

1

2

3

4

5

Power excess (standard deviations)2 4 6

Lo

g10

(co

un

t)

Sample data and candidates

Po

wer

Po

wer

772.11 772.12 772.13 772.14 772.15 772.16 772.17

Frequency (MHz)

• High-resolution data analyzed similarly• Also looked for ‘coincidences’ between high and medium resolution data

Environmental Statistical

Subspectra

Total

P01545-kvb-u-009

775.72 775.73

Frequency (MHz)775.75775.74 775.76 775.77

Subspectra

Total

Signal maximizes off-resonance:Radio peak

Signal distributed over many sub-spectra: a good threshold candidate (but did not persistin rescan)

New twist — ultrafine structure to axion signal?

p00080-kvb-u-009

• Main axion signal expected to be Maxwellian with ∆E/E ~ 10–6

• Sikivie et al have recently predicted non-thermalized fine-structure with ∆E/E~ 10–17!

— ‘Last-infall’ lines could contain several percent of total signal

— May greatly improve our sensitivity

• We have instrumented independent parallel Fast Fourier Transform streams for both medium (10–7) and high (10–11) resolution spectra Maxion

Frequency(energy)

∆E/E ~ 10–6

∆E/E ~ 10–17

Expected axion spectrum

Cosmic axion exclusion plot

P01672-kvb-u-005

10–6

10–14

10–16

10–18

10–20

10–5

Axion mass (eV)

* C. Hagmann et al.; Phys Rev. Lett. 80, 2043 (1998)

Co

up

ling

co

nst

ant

mas

s

KSVZ

Currently runningwith 4-cavity array

Phase II Upgrade DSFZ

UF RBF

HEMT’sTS ~3 K

SQUID’sTS ~300 mK

GeV

–2(

((

(eV

2

2

Completed(unpublished)

To coverin Phase Iupgrade

PRL* 3/98

1-cavity 4-cavity

• Phase I Upgrade: SQUIDs at 1.3 K will allow us run at KSVZ 4 times faster than with HEMTS

• Phase II Upgrade: SQUIDs at 200 mK will give us sensitivity to DFSZ axions even if they only constitute 50% of the halo

Multiple Cavities

50 cm.

λd - 10 - 100 m.

f010 ∝ R−1

• Exploring higher mass (frequency) range requires smaller cavities

• Placing a single, smaller cavity inside the magnet volume would be inefficient.

• One solution is to fill the magnet volume with multiple cavities.

• Since the axion signal is coherent over the magnet volume, the power from the individual cavities can be combined in phase.

• Four cavities with R = 20 cm will be used to cover the frequency range of 800 - 2000 MHz.

Multiple Cavities/Piezoelectric Tuning

Power Combining Test

810.2810.1810.0809.9809.8Frequency (MHz)

Measured Combined in Phase Simple Sum

POWER COMBINER

AMPLIFIER

MINOR PORT

MAJOR PORT CAVITY &

TUNING ROD

NETWORK ANALYZER

POWER DIVIDER

SWEEP GENERATOR

810.2810.1810.0809.9809.8Frequency (MHz)

Measured Combined in Phase Simple Sum

Tra

nsm

itted

Pow

erT

rans

mitt

ed P

ower

The precision of the piezos

P01545-kvb-u-012

Stepping Motors and Gears Piezos — Step #0

Piezos —Step #200 (w/o Feedback)

809.970 809.990 810.010 810.030Frequency (MHz)

810.360 810.380 810.400 810.420 810.440Frequency (MHz)

30.0 60.00.0 90.0 120.0 150.0 180.0Frequency Shift (Hz)

∆f = 1.99 kHzσ = 0.46

0 1.0 2.0 3.0 4.0 5.0Frequency Shift (kHz)

Piezos

Steppingmotors

∆f = 94.4 Hzσ = 13.6

Piezo Motors

First Four-Cavity Exclusion Plot

10-19

10-18

10-17

g aγγ

/ m

a2 [G

ev-2

/ eV

2 ]

812.62812.60812.58812.56812.54

Frequency (MHz)

3.364083.36392 3.36424

ma (µeV)

KSVZ

R&D towards 100 µeV (25 GHz) — Resonators

P01545-kvb-u-034

• Single cavity TM010: — Number∝ f3 in fixed volume – To minimize TE, TEM intruders• Periodic Post Resonators can have

very high TM010 frequencies — Number ∝ f – Height ∝ f-1 to minimize

mode crossings – Tuned by global shift of

alternate posts – Stacked as pans• Modeling begun; first warm prototype

being designed

We believe this is a promising avenue to the next decade in mass

TM010 Electric Fieldfor 96-post Array

Noise Temperature of the First Amplification Stage

• The system noise temperature is the sum of the physical temperature of the cavity and the noise temperature of the first amplification stage.

NcavS TTT +=

1.3 K for Phase I

VB

TS2ysensitivitPower ∝

ratescan Fixed

2

24

rateScan ST

VB∝

ysensitivitpower Fixed

• The present experiment uses GaAs HFET amplifiers supplied by NRAO with noise temperatures from 1.7 to 4.0 K.

The dc Superconducting Quantum Interference Device

• DC SQUID— two resistively shunted Josephson junctions on a superconducting ring

I

VΦ

Ib

nΦo

(n+1/2)Φo

∆V

I

V

V

0 1 2 Φo

Φ

δV

• Current-voltage (I-V) characteristic modulated by magnetic flux :— period one flux quantum o = h/2e = 2 x 10-15 Tm2

δΦ

Square Washer dc SQUID

washercounter-electrode

shunt

superconducting input coil

20 µm

1000 µmJosephson Junction

Shunt

What the devices look like

20

15

10

5

0

Gai

n (d

B)

1000800600400200Frequency (MHz)

Gain Measurements

• Gain of 11-turn microstrip SQUID with the washer grounded

22 Ω

180 Ω180 Ω56 Ω 56 Ω

220 Ω

IΦ

IΒ

20 dB, 50 Ω Attenuator

4 dB, 50 Ω Attenuator

From Generator To Spectrum Analyser

600

400

200ν res (

MH

z)

604020Coil Length (mm)

30

25

20

15

10

Gai

n (d

B)

700600500400300200100Frequency (MHz)

71 mm33 mm

15 mm

7 mmGai

n (d

B) ν r

es (

MH

z)

Gain vs. Coil Length

Gain of 6-turn Microstrip SQUID

20

18

16

14

12

10

Gai

n (d

B)

18001600140012001000Frequency (MHz)

Varactor Tuning of Microstrip SQUID

56 Ω 56 Ω

220 Ω

IΦ20 dB, 50 Ω Attenuator

Varactor 1 - 9 pF

for 20 - 0 VSQUID

Input Output

30

25

20

15

10

5

Gai

n (d

B)

220200180160140120100Frequency (MHz)

-1 V 0 1 3 4 6 9 22 V2

NoVaractor

Gai

n (d

B)

Noise Temperature Measurements

Τi

Pout

kBTSGSGA

slope: kBGSGA

T1 T2 T3

AP

NBASNiBout GTkGGTTkP ++= )(

IΦ

IΒ

50 Ω

Vacuum can

TiHeater

Thermometer Pout

K 4.2Tbath ≤K 8

dB 17

≈

≈P

N

A

T

G

K 80

dB 40

K 295 @ Amp.

≈

≈P

N

A

T

G

Noise Temperature of 31-turn SQUID with Cooled HEMT Postamplifier

22 Ω

180 Ω56 Ω

220 Ω

IΦ

-VGS VDSIΒ

~ 1 pF

2.0

1.5

1.0

0.5

0.0Noi

se T

empe

ratu

re (

K)

270260250240230Frequency (MHz)

Tbath = 1.9 K

K 25.0min =NT

200gain power SQUID ≈GK 8ier postamplif HEMT cooled of re temperatuNoise ≈P

NT

K 04.0/ ≈GT PN

Postamplifier Noise

-ART

)( 2AGGTTT PNN

sysN =+=

NP

N GTT << require Thus,

NT

PNT

SQUID Postamplifier

IΦ,1

IΒ,1

8 dB, 50 Ω Attenuator

IΦ,2

IΒ,2

Vin Vout

50 ΩSQUID 1 SQUID 2

35

30

25

20

15

Gai

n (d

B)

420400380360340Frequency (MHz)

Noise Temperature at 519 MHz vs. Bath Temperature: Resonant Source

2

4

68

100

2

4

68

1000N

oise

tem

pera

ture

(m

K)

102 3 4 5 6 7

1002 3 4 5 6 7

Temperature (mK)

TQ

-75

-74

-73

521520519518517Frequency (MHz)

Noi

se p

ower

(dB

)

Noise Energy at 140 kHz vs. Bath Temperature

56789

1

2

3

4

56789

10S Φ

(140

kH

z)/2

L (

h)

102 3 4 5 6 7 8 9

1002 3 4 5 6 7 8 9

1000Temperature (mK)

Summary

• The axion is well-motivated in particle physics and a very credible dark-matter candidate

• The parameter space (mass,coupling) is bounded and present experiments have already scanned well into this region

• Near-quantum-limited SQUID amplifiers are an enabling technology for a truly definitive search

• R&D effort underway to extend the search into the second decade in frequency