Title: Physics of Microwave Kinetic Inductance Detectors MKIDs
1Physics of Microwave Kinetic Inductance Detectors
(MKIDs)
- PhD Candidate Jiansong Gao
- Advisor Jonas Zmuidzinas
- Dec 9, 2005
2Outline
- Introduction to MKIDs
- Device Physics
- Overview
- Kinetic Inductance
- Excess Phase Noise
- Future Research Plan
- Theory Experiment
- What has been solved or understood whats not
3MKIDs Pair-breaking Superconducting Detectors
- Analogous to semiconductor detectors
4Principle of Operation
- We use transmission line resonator to readout
change in Zs
- Superconductor LmLki(nqp(hn))
anim.avi
CPW coplanar waveguide
5Advantages
- Simple fabrication
- Powerful multiplexing capability
6Applications
- Optical/UV/X-ray detector array
- Dark matter detection (phonon detector)
7MKIDs Physics Overview
- - QP generation-recombination noise
- HEMT amplifier noise
Excess Phase Noise !
8Calculation of Kinetic Inductance
- Superconducting transmission line
9I Surface Impedance Zs and Penetration Depth leff
R Popel
- Local Limit
- lltltx0, lltltlL ? q1gtgt1
- Extreme Anomalous Limit
- lLltltx0, lLltltl ? q12 ltlt1
- K(q), s1, s2 generally involve very complicated
integrals and has to be solved numerically. Under
the condition TltltTc and hwltltkbTc, it has
analytical expression
10Numerical results I
- Effective penetration depth
11Numerical results II
Parameters Tc, lL, v0, l, D0/kTc taken from last
page.
12II Kinetic Inductance Fraction (a)
- Shwarz-Christoffel Mapping (Conformal Mapping)
z-plane
w-plane
- Kinetic Inductance Fraction (a) by Matlab
e.g. Al CPW resonator, s3mm, g2mm, t200nm,
er12
Lm 0.3185m0, C 18.95 e0, g 4.3e005,
leff51.4nm, Lki0.0221m0
13Experimental Determination of a
- Direct determination of a from f0
- e.g., if fsc0 or fm0 has an relative error of 1,
the relative error in a will be 98 for a 2,
- Determination of a from df0(T)
Doesnt require theoretical model on leff
- 5 different geometry
- one geometry has large a which can be determined
directly from f0 - other 4 geometries indirectly from df0(T) data
14200nm Thick Al Film Result
Calculation agrees well with the experimental
results.
af directly from resonance frequency ar from
the temperature curve and the ratio of a atheo
theoretical calculation from conformal mapping
and M-B theory
1520nm Thin Al Film Result
T(K)
T(K)
leff goes up dramatically for thin film
16Unsolved, Future Work
- Theoretical calculation of kinetic inductance for
thin film
- solve M-B non-local equation in 2D
- Measurement of different thickness and other metal
- tested 20nm, 40nm, 200nm, 320nm
- other metal Nb, Re
17Noise
- On resonance excess phase noise
18Observations of Excess Phase Noise I
- QP generation-recombination noise
The noise source has to be inside the resonator!
19Observations of Excess Phase Noise II
slope -0.5
Frequency noise
Internal power
Frequency noise scales with internal power
(field) by power law
20Observations of Excess Phase Noise III
- Metal and Substrate Dependant
- In general,
- lower noise on sapphire than on Si
Noise is related with substrate.
- However,
- Nb on Si as good as Al on sapphire
Nb
21Observations of Excess Phase Noise IV
Noise and dissipation decrease as goes to the
wider geometry.
22Preliminary Two Level System Model
23A Qualitative Picture of Phase Noise from TLS
Q
I
24Quantum Mechanical Treatment of Resonant
Fluorescence
- Hamiltonian (Schrodinger picture)
-TLS Quantum Mechanics, -E Field Classical -b,
b Phonon bath
- Master Equation (Interaction picture)
-g decay rate of upper level, Wgtgtg -Dw atomic
detuning
- Bloch Equation (Interaction picture)
25Resonant Fluorescence
- Two-time Correlation Function and Quantum
Regression Theory
Y dipole radiation pattern
- Spectrum of Fluorescent Light
Strong driving Wgtgtg Mallow triplet
26Squeezed Spectrum Noise Blob
covariance matrix
Two channel spectrum analyzer
power spectrum matrix
v1
v2
SII
SQQ
27Why phase direction?
- Direction of the noise blob
2q
- Noise blob rotated by Argab
- Side peak LP filtered out by resonator
resonator bandwidth 10KHz to 100KHz ltlt Rabi
frequency
28Prediction of the model
wTLS
wdriving
Assume wTLS has a uniform distribution, results
from single TLS need to be integrate over d
a
r
29Unsolved, Future WorkI Experiment
- Complete information of excess phase noise
- Where is the TLS noise source? In volume or near
surface?
- surface of metal
- surface of gap
- metal-substrate interface
- thin layer into the substrate
- bulk substrate
noise v.s. geometry
30Unsolved, Future WorkII Theory
- Complete the Model
- temperature dependence
- explain df0
- full QM model
- nonlinear effect bistability, hysteresis
- many TLSs
- various parameters in the model
- g, wTLS, g, W
31Future WorkResearch Goal
- Put up a complete picture of the device physics
- Establish general formulations to calculate and
predict responsivity and noise - Propose an optimal design of the most sensitive
device
32Summary of Questions Need to be Answered
33Thank you !
34Observations of Excess Phase Noise V
- Drive at nearby frequency
35Readout Power Saturation
- Can we increase the readout power to suppress the
noise?
The resonator become nonlinear at high readout
power.
36Modeling the Nonlinear Resonator
T312
Assume f0 (Pint), Q (Pint) monotonously decrease
with power.
f
Pint
low power - normal
high power - distorted
very high power - discontinuous
37Observations of Readout Power Saturation
- All resonator of same geometry saturate at the
same internal field
- Superconductor dependence
- Nb resonators can be readout at 30dB more
microwave power than Al resonators - Substrate dependence
- Al on Sapphire resonators can be readout at 5dB
more microwave power than Al on Si resonators - Geometry dependence
- Resonators with small geometry (larger alpha) are
easier to saturate than large geometry (small
alpha)
38Unsolved, Future Work
- What is the mechanism of this saturation?
- Magnetic field, critical current
- explains superconductor, geometry dependence
- substrate heating
- explains substrate dependence
- Theoretical calculation of saturation power