When Every Photon Matters: OpticalUV Imaging Spectrophotometers

1
When Every Photon MattersOptical/UV Imaging
Spectrophotometers

• Benjamin Mazin, June, 2008

The Caltech/JPL MKID team Caltech Jonas
Zmuidzinas, Sunil Golwala, Tasos Vayonakis,
Jiansong Gao, David Moore, Omid Noozorian, Jack
Sayers JPL Bruce Bumble, Peter Day, Rick LeDuc,
Ben Mazin
2
A New View on the Universe

• CCDs are powerful tools, but have some weaknesses
• Poor time resolution
• Energy resolution requires dispersive optics or
  filters
• Read noise, dark current, and cosmic rays degrade
  performance
• Low Temperature Detectors (LTDs) can measure the
  energy and arrival time of every incoming photon
• Microsecond time resolution
• Energy resolution RE/?E20-50 or higher, wide
  bandwidth
• No read noise or dark current, near perfect
  cosmic ray rejection
• Limited count rate
• Large arrays are difficult to make
• Arrays of near-IR/optical/UV LTDs allow us to
  address scientific questions that are difficult
  or impossible with conventional focal planes

3
Faint, Rapidly Variable Compact Objects

• Optical counterparts are seen for a handful of
  pulsars
• North Crab, PSR B065614, Geminga, PSR B192910,
  etc.
• South PSR B0540-69, Vela, PSR B1055-52, etc.
• Optical pulsations have been seen in a subset of
  these objects
• The source of the optical/UV emission from
  pulsars is unknown
• Polar Cap Model
• Outer Gap Model
• And more
• Determining the source of the optical/UV emission
  may help us understand pulsar emission at other
  wavelengths and explore one of the most energetic
  and interesting natural physics labs

4
Imaging Spectrophotometer forCompact Object
Science

• Many of the difficulties in the CCD-based pulsar
  camera could be overcome by using a
  photon-counting, energy-resolving focal plane
• spectral changes
• number of peaks
• separation between peaks
• lag between optical, gamma-ray, and radio peaks
• ratio of the flux in the peaks
• polarization
• Many other interesting observations of compact
  objects (LMXBs, Am Hercs, AXPs, SGRs, etc.)

Crab Pulsar Romani et al. 1999
5
Planet Finding

• Coronagraphs from the near-IR to optical are one
  of the best ways to directly image terrestrial
  planets around nearby stars
• Suppress the starlight by a factor of 10 billion
  and image a planet as close in as the second or
  third Airy ring
• Harder than it sounds, and it sounds pretty hard!
• Ground-based (TMT) Space-based (TPF-C)

6
Imaging Spectrophotometer for Planet Finding

• Coronagraphs are limited by speckles from
  scattered and diffracted light
• Energy-resolving focal planes could increase
  sensitivity by a factor of up to 100 (!)
• Lower mirror and coronagraph tolerances
• Removes requirement of a separate spectrograph
• Gives the spectra of all planets in the dark box

Energy Resolved Photon Counting at Each Pixel
7
Simulation of Speckle Suppression

• Simulation from Soummer Remi and Ben Oppenheimer
  at AMNH

8
General Astrophysics inthe Near-IR/Optical/UV

• Integral Field Units are becoming popular
  instruments for diverse general astrophysical
  problems
• Even low energy resolution (RE/?E 20-50)
  allows determination of redshifts
• Galaxy age/metallicity/extinction variations with
  morphology
• Larger fields of view allows survey science
• Strong and Weak Lensing
• Microlensing
• Gamma-ray bursts afterglows
• UV missions
• Order sorter for Echelle spectrograph
• Much more!

Redshift determination of simulated 300-800 nm
observations of spiral galaxies, R55, 1 hour on
Keck (Mazin and Brunner, AJ 2000)
9
General Astrophysics at Other Wavelengths

• Basic technology behind MKIDs applicable from the
  millimeter to the X-ray
• CMB Polarization
• Submillimeter Galaxies
• Iron Lines from Black Hole Accretion Disks
• Also good for some particle detection and
  industrial applications
• Cryogenic Dark Matter Search (CDMS)
• Neutrino mass through Rhenium beta decay (MARE)
• X-ray microanalysis

10
Microwave Kinetic Inductance Detectors

• Microwave Kinetic Inductance Detectors (MKIDs)
  are pair-breaking superconducting detectors
• Superconductivity is caused by bound pairs of
  electrons known as Cooper Pairs. Free electrons
  not bound in pairs are known as quasiparticles.
• There is a gap energy separating the ground state
  of Cooper Pairs from the quasiparticles. This is
  know as the gap of the superconductor, and is
  lt1 meV. (Panel a)
• Energy is required to accelerate or decelerate
  Cooper pairs in a superconductor, leading to a
  extra inductance known as kinetic inductance.
• Changing the population of Cooper pairs by
  breaking them with photons or increasing the
  temperature changes the kinetic inductance.
• In an MKID we use a superconductor as the
  inductor in a high quality factory (Q104106) LC
  resonant circuit. (Panel b)

11
Microwave Kinetic Inductance Detectors

• Breaking Cooper Pairs increases the kinetic
  inductance and surface resistance
• The resonance shifts to lower frequency and gets
  shallower
• This can be observed as a change in the phase and
  amplitude of a probe signal centered on the low
  temperature resonant frequency
• A resonant circuit can be designed to have nearly
  perfect transmission off resonance and reflect
  power on resonance (Panel c)
• Built in frequency domain multiplexing!
• For more details, see

Nature, vol. 425, pp. 817-821, 2003Mazins PhD
Thesis, Caltech, 2004App. Phys. Lett., 89,
222507, 2006
12
Microwave Kinetic Inductance Detectors

• We chose to pursue coplanar waveguide (CPW)
  transmission line MKIDs
• Quarter wavelength long, one end open, the other
  shorted to ground
• Simple, one layer design
• No lossy thin-film dielectrics
• Elbow coupler allow accurate control of Q
• Position dependent response
• hybrid architecture all Nb except the last 100
  microns of center strip, which is usually Al, Re,
  or Ti

13
Microwave Kinetic Inductance Detectors

• Transmission line and lumped element
  implementations possible
• Excite from a common feedline
• Lithographically set each resonator to a
  different resonant frequency (by changing the
  length, L, or C)
• Dump quasiparticles into shorted end for maximum
  responsivity
• Excite with a comb on frequencies generated at
  room temperature
• Use a wide bandwidth HEMT to amplify the signal
  from all resonators
• Use room temperature electronics to sort out the
  signals

14
Phase and Amplitude

• We can represent the transmission past the
  resonator, S21, in the complex plane
• We reference phase and amplitude changes to the
  center of the lowest T resonance loop

O
Points on the resonance loop are equally spaced
in frequency
15
Quasiparticle Trapping

• Quasiparticles that diffuse from a higher gap
  metal into a lower gap metal quickly emit a
  phonon and become trapped
• Decouples photons absorption from energy
  detection
• Allows large absorbers with small, sensitive
  detectors

16
MKID Strip Detectors

• A linear array of strips provides a pixel array
• ABSORPTION
• Each photon breaks many Cooper pairs in the
  absorber.
• DIFFUSION
• Quasiparticles diffuse about the absorber,
  reaching the ends in microseconds.
• TRAPPING
• MKIDs made of a material with lower
  superconducting gap energy are attached to the
  ends of the strip the quasiparticles emit
  phonons and collapse to the new gap energy,
  becoming trapped in the MKID.
• DETECTION
• Photon energy is determined by summing the
  response from the MKIDs on both sides of the
  absorber
• RESOLVE X-POSITION
• The relative ratio of signals at the two sensors
  provides the position along the strip, so the
  energy resolving power determines the strip
  aspect ratio and pixel size.
• MULTIPLEXING
• All resonators are coupled to a single drive
  line.

17
MKID Testbed at Caltech

• Kelvinox 25 outfitted for MKID measurements

18
Optical/UV/X-ray MKID Arrays

• Strips can be stacked to form a high fill factor
  array

19
X-ray Strip Detector Results

• Working X-ray strip detector demonstrated in 2006
• 600 nm Ta absorber
• 200 nm Al MKID
• Illuminated with 55Fe X-rays
• ?E 62 eV at 6 keV
• Data from a poor quality sapphire substrate,
  should be able to do significantly better
• Mazin et al., App. Phys. Lett., 89, 222507 (2006)

20
Ta/Al Optical/UV Strip Detector

• 4 Layer Device on Sapphire
• 80 nm epi-Ta
• 40 nm SiO2 Ta protect
• 20 nm Al resonator
• 100 nm Nb resonator body
• 1 micron center strip!

21
Optical Strip Detector Results

• Optical strip detector demonstrated in late 2007
• ?E 40 nm at 250 nm
• Working now on refining array fabrication to
  improve energy resolution
• Arrays of many thousands of pixels are feasible
  now

22
New Ta/Al Strip Detector

• 150 nm Ta, 50 nm SiO2 protect, 50 Al

23
New Array Frequency Noise

• New optical/UV array frequency noise (40nm Al)

24
Excellent Frequency Accuracy

• 0.8 MHz resonator to resonator frequency jitter!

25
Al/Ti Strip Detectors

• New hybrid array design (200 nm Al/100 nm Ti on
  Sapphire)

26
Titanium Hybrid MKIDs

• Titanium hybrids have very long penetration
  depths and quasiparticle lifetimes, making
  extremely sensitive detectors.

NEP 4 x 10-19 W Hz-0.5 At 100 Hz!
27
Microwave Kinetic Inductance Detectors

• IQ readout of amplitude and phase using homodyne
  mixing
• Standard microwave technique with both analog and
  digital implementations

28
Software Radio Readout

• Software Defined Radio (SDR) Overview
• Generate frequency comb and upconvert to
  frequency of interest
• Pass through MKID and amplify
• Downconvert and Digitize
• Extract signals in a powerful FPGA

29
Software Radio Signal Generation

• Generate an comb of frequencies at baseband with
  a fast D/A
• Modulation possible
• Waveform buffer size important
• No phase jumps!

f1
f3
f2
30
Up and Down Conversion

• Shift the comb to microwave frequencies with
  mixer (SSB, IQ modulator, etc.)
• Using the same synthesizer for up and down
  conversion removes the synthesizer phase noise
  contribution (to first order)
• Pass the comb through the detector
• Amplify the signal with a low noise amplifier
  (HEMT, SQUID Amp, etc.)
• Amplifier 1 dB compression can limit the number
  of probe signals
• Depends on detector readout power, number of
  detectors
• Weinreb HEMTs can handle -40 dBm
• Carrier suppression can help with this
• Downconvert back to baseband

31
Software Radio Demodulation

• Sample with a fast A/D
• A/D dynamic range important!
• Figure out how to get the amplitude and phase
  modulation imprinted by the detectors out
• Lots of possibilities
• Direct digital downconverters on dedicated IC
• Direct digital downconverters on FPGA
• Polyphase filterbank or FFT on FPGA
• Hybrid FFT/DDC solutions
• Careful of dynamic range, output sample rates,
  phase skips, etc.

32
SDR Hardware

• Innovative Integration x5-400m board
• Dual 400 MHz 16-bit D/A
• Dual 400 MHz 14-bit A/D
• Xilinx Virtex 5 (SX95T) FPGA
• Complete 144 resonator, 400 MHz bandwidth readout
  on a PMC module!
• 75/resonator, 7.5/optical pixel (due to strip
  multiplexing)
• Other boards (Pentek, CASPER) are also
  possibilities

33
SDR Demonstration

• We have demonstrated SDR readouts in the lab and
  at the telescope

Mazin et al., Proceedings of LTD-11, 559 (2006)
34
MKID Camera for Palomar 200

• Lens coupled 20x64 pixel array in cryogen-free
  ADR, 0.33 pixels
• 350 nm to 850 nm simultaneous bandwidth
• Energy resolution of 20 at 400 nm
• On-the-sky QE of 30-40
• Maximum count rate/pixel 200 cts/sec due to
  strip multiplexing
• Only for faint objects (V gt 20)
• Dark time only
• All digital room temperature readout
• Possibility of future operation with PALM-3000 AO
  system
• Preliminary design is for Coudé focus
• Camera Team Ben Mazin, Bruce Bumble, Sunil
  Golwala, David Moore
• Collaborators J. Zmuidzinas, C. Martin, P. Day,
  R. LeDuc

35
Multicolor Submillimeter MKIDs

• Antenna coupled multi-color submillimeter
  detectors

36
4x4 Pixel Antenna-Coupled MKIDs
Fabricated and tested prototype 4x4
antenna-coupled MKID array for 200-400 GHz.
Nb pixel ground plane and microstrip
resonators
Al feedline and resonators
feedline
3.3 mm
20 mm
Lithographic lumped-element bandpass
filters (multicolor pixel)
37
4x4 Pixel Antenna-Coupled MKID photos
38
Frequency Response and Beam Maps
Band1
Band2
39
DemoCam

• We brought a submillimeter MKID camera
  (DemoCam) with a SDR readout to the Caltech
  Submillimeter Observatory (CSO) in April, 2007

40
DemoCam

• We detected an object outside of the solar
  system!
• Plan to return this summer with a 6x6 pixel, 4
  color array with 144 channel SDR readout

41
The Future Microstrip MKIDs

• Funding started in late 07 from ROSES-APRA2 grant
• Goal is to develop megapixel optical/UV imagers
  using microstrip MKIDs
• Thin dielectrics (5-20 nm) yield high kinetic
  inductance fraction, short resonators
• Can be made on top of absorbers and
  back-illuminated for high fill factors in 2-d
• Use amplitude readout to avoid excess phase noise
  problems
• High Q thin-film dielectrics needed, develop in
  conjunction with Martinis at UCSB, Pappas at NIST

Martinis et al. (2005)