Searching for Dark Matter Axions with ADMX-HF - …Searching for Dark Matter Axions with ADMX-HF...
Transcript of Searching for Dark Matter Axions with ADMX-HF - …Searching for Dark Matter Axions with ADMX-HF...
Searching for Dark Matter Axions with ADMX-HF(The Axion Dark Matter eXperiment – High Frequency)
Ben BrubakerYale University
February 18, 2016
UCLA
Ben Brubaker (Yale) ADMX-HF UCLA Dark Matter 2016 1 / 26
Outline
Motivation and challenges for high-frequency haloscopes
Overview of ADMX-HF design
Progress towards quantum-limited noise performance
Status of first production data run
Near-term upgrade plans and R&D overview
Ben Brubaker (Yale) ADMX-HF UCLA Dark Matter 2016 2 / 26
Axion Parameter Space
Parameter spacemostly unexplored
ADMX-HF: bothpathfinder forhigh-mass regionand innovationtestbed
2 × KSVZ coveragefor 4 − 8 GHz(16 − 33 µeV) in 3years with currenttechnology
Ben Brubaker (Yale) ADMX-HF UCLA Dark Matter 2016 3 / 26
Axion Parameter Space
Parameter spacemostly unexplored
ADMX-HF: bothpathfinder forhigh-mass regionand innovationtestbed
2 × KSVZ coveragefor 4 − 8 GHz(16 − 33 µeV) in 3years with currenttechnology
Ben Brubaker (Yale) ADMX-HF UCLA Dark Matter 2016 3 / 26
Haloscopes at High Frequencies
Challenges
At constant coupling,dνdt∼ ν−14/3
for resonator geometries used in axion searches to date
Largely due to small volume of high-frequency resonators
Standard Quantum Limit (SQL): kTS ≥ hν for linear amplifiers
Motivation
∼ 20 µeV axions provide closure density in simplest models
Cryogenics much simpler at 5 cm scale than 50 cm scale
Josephson parametric amplifiers (JPAs): tunable amplifiers in the2-12 GHz range which can approach quantum noise limitsBen Brubaker (Yale) ADMX-HF UCLA Dark Matter 2016 4 / 26
ADMX-HF Collaboration
Yale University (host)Steve Lamoreaux, Ling Zhong, Ben Brubaker, Sid Cahn
UC BerkeleyKarl Van Bibber, Maria Simanovskaia, Samantha Lewis,Jaben Root, Kelly Backes, Al Kenany
Lawrence Livermore National LabGianpaolo Carosi, Tim Shokair
CU Boulder/JILAKonrad Lehnert, Maxime Malnou, Dan Palken
Ben Brubaker (Yale) ADMX-HF UCLA Dark Matter 2016 5 / 26
ADMX-HF Design
9 T superconducting solenoid fromCryomagnetics
VeriCold dilution refrigeratoroperated at T ∼ 120 mK
Copper cavity with Q ∼ 20,000,tunable from 3.5 to 5.85 GHz
JPA: noise near SQL, tunable from4.4 to 6.5 GHz
Sensitive to P ∼ 2.5 × 10−23 W(≈ 1 5 keV WIMP event/year!)
Ben Brubaker (Yale) ADMX-HF UCLA Dark Matter 2016 6 / 26
Cavity and Motion Control
Qc ∼ 20,000, tunablefrom 3.5 to 5.85 GHz
Tuning via rotation ofoff-axis Cu rod
Kevlar lines forcryogenic motioncontrol: no heat load at100 mK
Linear drives fordielectric fine tuningand antenna insertion
Ben Brubaker (Yale) ADMX-HF UCLA Dark Matter 2016 7 / 26
Josephson Parametric Amplifier
An LC circuit with nonlinear SQUIDinductance⇒ parametric gain with a strongpump tone near resonance.
Added noise is just thermal noise of the “idlermode” from opposite side of pump
Apply DC magnetic flux to tune from 4.4 to6.5 GHz with ∼ 20 dB gain
Bucking coil, Pb/Nb/Cryoperm shields, andpassive NbTi coils for ∼ 108 net reduction offield on JPA
Ben Brubaker (Yale) ADMX-HF UCLA Dark Matter 2016 8 / 26
Noise calibration principle
kTS = hν(
1ehν/kT − 1
+12
+ NA
)
Linear detection: ≥ 1/2 photon at the input of any linear amplifier,because quadrature amplitudes don’t commute with Hamiltonian.
The Standard Quantum Limit: A phase-insensitive linear amplifiermust add noise NA ≥ 1/2, because quadrature amplitudes don’tcommute with each other.
Measure NA using blackbody source at known temperature (theY-factor method) – includes JPA added noise, HEMT added noiseand loss before JPA.
Y =PHot
PCold=
GH [NH + NA (NH)]
GC [NC + NA (NC)]
Ben Brubaker (Yale) ADMX-HF UCLA Dark Matter 2016 9 / 26
Noise calibration principle
kTS = hν(
1ehν/kT − 1
+12
+ NA
)
Linear detection: ≥ 1/2 photon at the input of any linear amplifier,because quadrature amplitudes don’t commute with Hamiltonian.
The Standard Quantum Limit: A phase-insensitive linear amplifiermust add noise NA ≥ 1/2, because quadrature amplitudes don’tcommute with each other.
Measure NA using blackbody source at known temperature (theY-factor method) – includes JPA added noise, HEMT added noiseand loss before JPA.
Y =PHot
PCold=
GH [NH + NA (NH)]
GC [NC + NA (NC)]
Ben Brubaker (Yale) ADMX-HF UCLA Dark Matter 2016 9 / 26
Noise calibration principle
kTS = hν(
1ehν/kT − 1
+12
+ NA
)
Linear detection: ≥ 1/2 photon at the input of any linear amplifier,because quadrature amplitudes don’t commute with Hamiltonian.
The Standard Quantum Limit: A phase-insensitive linear amplifiermust add noise NA ≥ 1/2, because quadrature amplitudes don’tcommute with each other.
Measure NA using blackbody source at known temperature (theY-factor method) – includes JPA added noise, HEMT added noiseand loss before JPA.
Y =PHot
PCold=
GH [NH + NA (NH)]
GC [NC + NA (NC)]
Ben Brubaker (Yale) ADMX-HF UCLA Dark Matter 2016 9 / 26
Noise calibration results
Design goal:TS ≈ 1.5hν ≈ 400 mK
We measure NA ≈ 1.5⇒ TS ≈ 575 mK offresonance
Total noise increases toTS ≈ 4hν ≈ 1.1K onresonance
Off-resonance noise consistent with 20% thermal contribution,3 − 7 K HEMT noise, 1 − 2 dB loss before JPA
Temperature- and gain-dependence of resonant noise bumpimplicates thermal link to tuning rod
Ben Brubaker (Yale) ADMX-HF UCLA Dark Matter 2016 10 / 26
Noise calibration results
Design goal:TS ≈ 1.5hν ≈ 400 mK
We measure NA ≈ 1.5⇒ TS ≈ 575 mK offresonance
Total noise increases toTS ≈ 4hν ≈ 1.1K onresonance
Off-resonance noise consistent with 20% thermal contribution,3 − 7 K HEMT noise, 1 − 2 dB loss before JPA
Temperature- and gain-dependence of resonant noise bumpimplicates thermal link to tuning rod
Ben Brubaker (Yale) ADMX-HF UCLA Dark Matter 2016 10 / 26
Noise calibration results
Design goal:TS ≈ 1.5hν ≈ 400 mK
We measure NA ≈ 1.5⇒ TS ≈ 575 mK offresonance
Total noise increases toTS ≈ 4hν ≈ 1.1K onresonance
Off-resonance noise consistent with 20% thermal contribution,3 − 7 K HEMT noise, 1 − 2 dB loss before JPA
Temperature- and gain-dependence of resonant noise bumpimplicates thermal link to tuning rod
Ben Brubaker (Yale) ADMX-HF UCLA Dark Matter 2016 10 / 26
Noise calibration results
Design goal:TS ≈ 1.5hν ≈ 400 mK
We measure NA ≈ 1.5⇒ TS ≈ 575 mK offresonance
Total noise increases toTS ≈ 4hν ≈ 1.1K onresonance
Off-resonance noise consistent with 20% thermal contribution,3 − 7 K HEMT noise, 1 − 2 dB loss before JPA
Temperature- and gain-dependence of resonant noise bumpimplicates thermal link to tuning rod
Ben Brubaker (Yale) ADMX-HF UCLA Dark Matter 2016 10 / 26
Timeline
4/2012 − 6/2014:Design/construction
7/2014 − 1/2016:Integration/commissioning:
I Eliminated vibrationallycoupled JPA gain fluctuations
I Eliminated JPA/cavityinterference
I Eliminated IF spikes in spectraI Added analog flux feedbackI Implemented blind injection of
synthetic axion signals
1/26/2016: Production data runstarted!
Ben Brubaker (Yale) ADMX-HF UCLA Dark Matter 2016 11 / 26
Timeline
4/2012 − 6/2014:Design/construction
7/2014 − 1/2016:Integration/commissioning:
I Eliminated vibrationallycoupled JPA gain fluctuations
I Eliminated JPA/cavityinterference
I Eliminated IF spikes in spectraI Added analog flux feedbackI Implemented blind injection of
synthetic axion signals
1/26/2016: Production data runstarted!
Ben Brubaker (Yale) ADMX-HF UCLA Dark Matter 2016 11 / 26
Timeline
4/2012 − 6/2014:Design/construction
7/2014 − 1/2016:Integration/commissioning:
I Eliminated vibrationallycoupled JPA gain fluctuations
I Eliminated JPA/cavityinterference
I Eliminated IF spikes in spectraI Added analog flux feedbackI Implemented blind injection of
synthetic axion signals
1/26/2016: Production data runstarted!
Ben Brubaker (Yale) ADMX-HF UCLA Dark Matter 2016 11 / 26
Timeline
4/2012 − 6/2014:Design/construction
7/2014 − 1/2016:Integration/commissioning:
I Eliminated vibrationallycoupled JPA gain fluctuations
I Eliminated JPA/cavityinterference
I Eliminated IF spikes in spectraI Added analog flux feedbackI Implemented blind injection of
synthetic axion signals
1/26/2016: Production data runstarted!
Ben Brubaker (Yale) ADMX-HF UCLA Dark Matter 2016 11 / 26
Timeline
4/2012 − 6/2014:Design/construction
7/2014 − 1/2016:Integration/commissioning:
I Eliminated vibrationallycoupled JPA gain fluctuations
I Eliminated JPA/cavityinterference
I Eliminated IF spikes in spectraI Added analog flux feedbackI Implemented blind injection of
synthetic axion signals
1/26/2016: Production data runstarted!
Ben Brubaker (Yale) ADMX-HF UCLA Dark Matter 2016 11 / 26
Status of First Data Run
All parameters at design specs except TS
3 full scans, 1 month each, with ∼ 70% live time
We will cover 5.7 − 5.8 GHz at 2.5 × KSVZ sensitivity
Ben Brubaker (Yale) ADMX-HF UCLA Dark Matter 2016 12 / 26
Preliminary Analysis
Noise is Gaussian out to at least 3 σ and at least 5.5 hours of averaging
Ben Brubaker (Yale) ADMX-HF UCLA Dark Matter 2016 13 / 26
Preliminary Analysis
Noise is Gaussian out to at least 3 σ and at least 5.5 hours of averaging
Ben Brubaker (Yale) ADMX-HF UCLA Dark Matter 2016 13 / 26
Preliminary Analysis
Ben Brubaker (Yale) ADMX-HF UCLA Dark Matter 2016 13 / 26
What’s Next?
R&D for next-generation searches:I Boost Qc : thin-film
superconducting cavities (UCB)I Evade SQL: preparing cavities in
squeezed states (CU)I Avoid V suppression: photonic
band gap cavities (UBC)
JPA/cavity fabrication to extendfrequency range
Improve thermal link to rod: reduceTS by ∼ 500 mK
Transfer experiment to newBlueFors dil fridge: more stable,reduced vibrations⇒ colder
Ben Brubaker (Yale) ADMX-HF UCLA Dark Matter 2016 14 / 26
Thank you!
We gratefully acknowledge support from the National ScienceFoundation, the Heising-Simons Foundation,
and the Department of Energy.
Ben Brubaker (Yale) ADMX-HF UCLA Dark Matter 2016 15 / 26
Extra Slides
Ben Brubaker (Yale) ADMX-HF UCLA Dark Matter 2016 16 / 26
Hybrid Superconducting Cavity R&D
Superconducting cavities: high Q, but notin high B fields.
Need cavity to withstand high static fields,not high RF power as in acceleratorapplications.
Thin film type-II superconductors (NbTiN,NbN, MgB2): high (Bc2)‖, no vortexdissipation if field is parallel and d . ξ
With appropriate coatings on barrel andcopper endcaps, we can increase Q by ∼the aspect ratio of the cavity (∼ 6×).
Plasma deposition system installed andcoating prototypes at UCB.
Ben Brubaker (Yale) ADMX-HF UCLA Dark Matter 2016 17 / 26
Squeezed states for axion detection
JPAs can operate in a mode where theyamplify one signal quadrature andsqueeze the other: no SQL
If we align the squeezed quadrature ofone JPA with the amplified quadrature ofanother, no 1/2 photon from lineardetection either: kTS � hν!
Cavity must be overcoupled; squeezedstate injected in reflection. Works due tofinite axion coherence time ∼ 200 µs.
Eliminating loss before JPA is a challenge.
Work on a prototype squeezed-statereceiver is underway at CU, perhapsready as early as late 2016.
Ben Brubaker (Yale) ADMX-HF UCLA Dark Matter 2016 18 / 26
Photonic band gap cavities
Isolate a single mode using a defect in an open periodic lattice ofmetal and/or dielectric rods
Much higher volume at a given frequency than conventionalcylindrical cavity; well-defined TM010 mode
Challenge is making them tunable
HPC simulations of resonator designs and tolerances at UCB
Ben Brubaker (Yale) ADMX-HF UCLA Dark Matter 2016 19 / 26
Bolometric axion detection
Thermal noise from whole cavity band,but no standard quantum limit.
δPlin
δPbol∼
√Qc
Qaehν/kT > 1 above 10 GHz.*
Perhaps even at lower frequencies if we can improve Qc .
Mature detector technology in AMO (Rydberg atoms), quantumcomputing (cavity QED), CMB (Transition Edge Sensors).
*: S. K. Lamoreaux et al., Phys. Rev. D 88, 035020 (2013).
Ben Brubaker (Yale) ADMX-HF UCLA Dark Matter 2016 20 / 26
Haloscope Signal Power
P ∼ g2aγγ (ρa/ma) B2QcVCnm`
(Matrix element)2
Axion number density
Virtual photon number density
Resonant enhancement of axion-photon conversion
Effective volume occupied by cavity mode
Form factor Cnm` =(∫
dV Enm` · B)2/(V
∫dV E2
nm`
)⇒ best for low-order TM modes: L ∼ ν−1
P ∼ 2.5 × 10−23 W (≈ 1 5 keV WIMP event/year!)
Ben Brubaker (Yale) ADMX-HF UCLA Dark Matter 2016 21 / 26
Haloscope Noise
Thermal noise in cavity bandwidth kBT ∆νc ≈ 10−18 W.
The Dicke radiometer equation:
SNR =P
kTS
√t
∆νa
Noise temperature TS includes blackbody and amplifier noise.
Linear detection:Qa ≈ 50Qc ⇒ ∆νa � ∆νc .
RMS noise decreases as 1/√
t .
Ben Brubaker (Yale) ADMX-HF UCLA Dark Matter 2016 22 / 26
Microwave Layout
3 paths for injection intofridge: transmission,reflection, JPA pump.
Cryo microwave switch(Radiall) and terminator atstill plate for Y-factormeasurement.
Second-stage amplifier:LNF LNC4_8A: TN ≈ 4 K.
Ben Brubaker (Yale) ADMX-HF UCLA Dark Matter 2016 23 / 26
Microwave Layout
GaGe ADC: 14 bits, 2 GS memory, 25 MS/s sampling.VNA for cavity and JPA measurements.
Ben Brubaker (Yale) ADMX-HF UCLA Dark Matter 2016 23 / 26
DAQ procedure
Noise is mixed down to MHzand digitized at 25 MS/s fort ∼ 15 min.
In-situ computation andaveraging of power spectrawith 100 Hz resolution.
Step resonance by ∼ ∆νc/4and repeat O
(104
)times.
Data rate ∼ 20 GB/100 MHz(500 TB/100 MHz to save fulltime series data).
Ben Brubaker (Yale) ADMX-HF UCLA Dark Matter 2016 24 / 26
Cavity Tuning
Ben Brubaker (Yale) ADMX-HF UCLA Dark Matter 2016 25 / 26
JPA Tuning
-0.4 -0.3 -0.2 -0.1 0.0 0.1
4.5
5.0
5.5
6.0
6.5
JPA current HmAL
JPA
freq
uenc
yHGH
zL
In 5 T field (blue) and no field (red)
Ben Brubaker (Yale) ADMX-HF UCLA Dark Matter 2016 26 / 26
JPA Tuning
Ben Brubaker (Yale) ADMX-HF UCLA Dark Matter 2016 26 / 26