Optical and NearInfrared Detectors

1
Optical and Near-Infrared Detectors

• Astronomy Postgraduate Course
• Alastair Basden

2
Course outline

• Facilities Overview (URLs etc.)? (Richard)?
• Opt/IR Detectors (Alastair)?
• Imagers (Jürgen Schmoll)?
• Spectroscopy (Jurgen)?
• Adaptive Optics (Tim Morris)?
• Advanced spectrograph systems (Jurgen)?
• Optical interferometry (Alastair)?
• Advanced AO systems (Tim)?
• Space Instrumentation (Jurgen)?
• Tour of NetPark (Richard)?

3
Contents

• Older Detectors
• photography
• intensified detectors
• Current detectors
• CCDs
• near-IR arrays
• CMOS
• Possible Future Detectors

4
Detector Books

• The detector bible is Janesicks Scientific
  Charged-Coupled Devices 58
• Latest Janesick book (2007) looks interesting
  Photon Transfer 38
• Also, one by Ian McLean probably cheaper,
  though a few years old now

5
Older technologies - photography

• Advantages
• Stability
• storage capacity (109 pixels/UKST plate)?

(Iout/Iin)?

• Disadvantages
• low quantum efficiency (QE) 1-2 even when
  hypersensitized
• non-linear, non-reproducible response
• small dynamic range
• non-digital
• not instant

6
Older detector technologies

• Photomultiplier (PM)
• light-sensitive photocathode emits electrons -
  accelerated and multiplied by dynode chain -
  collected as large current pulse at anode.
• Advantages
• linear response
• fairly good QE (20 max. for S20) but rather
  narrow band (UV,V)?
• large dynamic range
• Disadvantages
• poor QE in red and IR (GaAs much better than
  S20)?
• only one pixel - not an imaging system
• Signals cannot be too close in time

7
PM multiplication
Photo-electron
Large output signal
8
PMT tubes

• Several PM tubes

9
Older detector technologies

• Image Photon Counting System (IPCS)?
• 4-stage intensifier each with S20 photocathode
  and phosphor. Electrons are guided by axial
  magnetic field and accelerated by large electric
  field. Final output to TV camera using event
  centring.
• Advantages
• blue QE peaking at 20
• large detector, 30mm diameter and 10002 pixels
• Real-time observations
• READOUT NOISE essentially zero
• Disadvantages
• Fragile irreparable damage by over exposure
• stability problems (flop)?
• Saturation not suitable for use at high photon
  rates
• Image distortion (non uniformity of magnetic
  field)?
• Poor QE in red and IR

10
IPCS-2

• Improves dynamic range and eliminates centring
  distortion
• Uses micro-channel plate (MCP) intensifier to
  obtain gain of 106
• MCPs are used in other photon counting systems,
  e.g., ESO MAMA detector uses curved-channel MCP
• HST Faint Obj Camera

MCP
phosphor
Photocathode
Chevron shaped tubes with metal coating
glass
1 kV potential
11
Current technologies APDs

• Avalanche PhotoDiode (APD)?
• Semiconductor analogue to PM tubes
• Internal gain due to impact ionisation
• high QE achromatic (many wavelength) photon
  counter
• so far, it has been possible to build small
  arrays (separated pixels)?
• Array sizes of up to 256x256 pixels
• A common previous approach was to used
  fibre-coupled, up to 85 pixels
• Non-linear response above a certain signal level
  (due to detector dead time)?

12
CCD

• Charged Coupled Device.
• Si microcircuit with 5-30? pixels
• standard detector for optical wavelengths
• (?lt 1?m)?
• First used for astronomy in about 1980
• essentially an array of MOS capacitors
• bias is controlled by sets of metal strips
  (electrodes) deposited on a thin Si02 layer
  providing insulation from the bulk Si.
• Modern CCDs are thinned and illuminated from the
  backside (back illuminated) rather than through
  the electrodes. This improves blue sensitivity.

13
Important point...

• CCD and electronics (usually termed controller or
  camera) may be from separate manufacturers...
• You may buy both separately, or buy a camera that
  has a CCD inside it (but that may be made by a
  different manufacturer)?

14
CCD readout

• Absorbed photons produce electron-hole pairs
• Electrons gather in potential wells beneath the
  electrodes
• At the end of the exposure the charge is
  transferred out of the chip

lt100 x 100 pixels for specialist applications,
e.g., AO wavefront sensing gt4000x2000 pixels for
long-integration imaging when large images are
required (ie larger fov)?
Vertical charge transfer of all columns together
Output amplifier - reads one pixel at a
time 500ns-50?s per pixel Slow readout
10kHz Fast readout 10MHz
Amplifier and digitisation
Horizontal charge transfer of serial
register (yellow) only
15
CCD readout improvements

• 10MHz pixel rate takes a long time for a large
  device (1s for a 10MPixel digital camera)?
• Improve this by
• Reading faster but this increases the noise
• Using multiple output ports on the CCD
  (quadrants) but this increases flat field
  problems
• Make the CCD smaller (!)?
• Read only part of the CCD

16
CCD transfer architecture
17
CCD charge well formation(simple surface channel
configuration - now superceded)?
18
CCD3-phaseCharge Transfer
19
Drive pulses

• Also called clocks
• Must be jitter free (otherwise image can't be
  calibrated) requiring very precise timings
• Must be finely tunable to provide enable optimum
  CCD performance

20
CCD 3-phase electrode deposition
21
Phases

• 2 phase and 4 phase operation also fairly
  standard, though not as common.
• 2 phase requires special doping of the substrate,
  and the resulting CCD can usually be clocked
  (read out) at higher rates (less demanding
  electronics).

22
CCD buried channel design
Avoids trapped charges in surface defects (image
smearing)?
23
CCD detail and output circuit
VQ/C
Effective storage Capacitor
Storage capacitor recharged before each pixel
arrives
24
CCDs - key characterstics

• Electron-hole pairs generated at room
  temperatures.
• Gives rise to a dark current/noise.
• Operate around 150K for most astronomy
• Readout noise
• uncertainty in amount of charge ? 2-3 e- rms (2-3
  photons)?
• Much larger at faster readout rates
• Charge transfer efficiency
• normally extremely high but local defects
  possible
• Cosmetic defects
• shorts to ground column defects
• charge injection bright columns
• charge traps tadpoles
• flatfield noise pixel/pixel sensitivity
  variations 1
• Cosmic ray events
• several 100 electrons in small spots
• 730 in 2000 secs for 1024x1024 Tektronics (aka
  SiTE)?

25
CCDs - key characterstics

• Full-well capacity
• Determines exposure time and light level
  combination
• Longer exposure times will lead to saturation
• Gain
• Determines number of photons per ADU
  (Astronomical data unit)?
• Readout rate
• Determines time taken to read out image after
  exposure
• Parallel transfer rate
• Rate at which rows are shifted during readout
• Smearing if not shuttered

26
CCDs

• Advantages
• excellent QE (can be gt90 over good ? range)?
• high dynamic range (up to 400,000 full well)?
• no damage from over exposure
• photon arrival rate unimportant
• excellent linearity
• excellent stability
• Digital images

27
Problems with CCDs

• Disadvantages
• Still not enough pixels mosaics being built
• cannot see image build up (destructive readout)?
• temporal resolution costs readout noise
• Non-uniform surfaces
• Dead pixels
• Not photon counting

28
Photon Transfer Curve

• Photon arrival follows Poisson distribution,
  where meanvariance
• Photon transfer curve is obtained from spatial
  (or temporal) statistics of a set of images,
  illuminated with constant flat calibration
  source
• To plot a photon transfer curve, it is common to
  take 2 images per exposure time (to eliminate
  sensitivity differences), starting from 0 sec and
  increasing, walking through the dynamic range of
  the detector
• The inverse gain is obtained from the photon
  transfer curve (shown in next slide), which gives
  the of electrons per digital count on each pixel

29
Photon Transfer Curve
mean(img1) mean(img2)?
Slope is inverse gain
variance(img1 - img2)?
30
Photon Transfer Curve in Action
Non-linear CCD
Normal CCD
Photon transfer curve is not only useful to
obtain the gain of the detector readout system
( electrons per digital count), but also to
evidence problems within the dynamic range of the
pixel. This example shows a non-linearity seen in
Gemini South GMOS instrument, caused by a wrong
setup in bias voltages
31
CMOS devices

• Complementary Metal Oxide Semiconductors (CMOS)
  detectors
• provide advantages over CCD image sensors
  particularly in speed and low power consumption
• Low fill factor (low QE)
• though now up to 100
• Typically higher readout noise

32
CMOS

• Individual readout for each pixel
• More electronics on the CMOS device
• Not charge coupled
• Usually cheaper than CCDs
• Not commonly used for astronomy (but becoming
  more common)?
• e.g. Mobile phones

33
CMOS

• Simple electronics
• Easier to design astronomical instruments
• Only standard 3.3V needed
• Timing non-critical
• Quality and size improving rapidly

34
Intensified CCDs (ICCDs)?

• Image intensifier placed between the source and
  CCD
• Amplify signals above the readout noise
• BUT
• Signals must be very faint (coincidence losses)?

35

• Electron multiplying CCDs (EMCCDs)?
• (Low Light Level CCDs or L3CCDs)?
• E2V Technologies and Texas Instruments
• An internal gain mechanism boosts the CCD output
  signal to such an extent that electronic noise in
  the camera is eliminated. Detection of single
  photons now possible.
• gt10MHz readout rate at lt1e noise (photon
  counting).
• gt90 Quantum Efficiency.
• Technology can be applied to any image format.
• Can approach ideal detector performance.
• Entirely solid state (robust)?

36
(No Transcript)
37
EMCCD operation

• Extension of the serial register termed the
  multiplication register
• 3 (4) phase register, one clock operated at much
  higher voltage
• Leads to electron multiplication
• Small probability (p0.01-0.02) at each stage
• Typically 500 stages leads to large overall gain
• (1p)500 gt 1-10000
• Gain can be changed by simply changing a voltage

38
EMCCD operation

• Many such stages...

Large output signal (even for only one photon
input)? is now well above the readout noise level
when a large gain is used.
39
EMCCD usage

• Becoming more widely used in astronomy
• For real-time applications
• High speed readout
• Small sizes
• High framerate

40
The CCD60 from E2V.
1 electron in
Multiplication registers
1000 electrons out
Output Amp.
41
EMCCD problems

• High voltage clock waveform required (most CCD
  controllers need adaption)?
• Stochastic multiplication process introduces
  additional source of noise (multiplication noise)
  effectively halves QE
• Clock induced charge now a problem

42
EMCCD output distribution
Significant overlap in the outputs for
different photon inputs leads to information
loss extra form of noise
43
IR detectors

• Similar to CMOS technology at longer wavelengths
• Each pixel has individual readout

44
IR atmospheric transmission
Absorption is dominated by atmospheric CO2 and H2O
45
Detail of IR atmospheric transmissiom
Mauna Kea for typical H2O vapour level
J H K L M
46
IR atmospheric windows
Kshort 2.0-2.3, K1.95-2.30, L3.5-4.1
47
IR background20 emissive telescope _at_ 275K
Sky and Telescope (dominant) Thermal
Atmospheric OH
48
IR thermal background
At ? 0.43?m, msky 24 (dark of moon)? At ?
2.2?m, msky 13.5 At ? 10?m, msky -3.0
(negative!)?
Sky
Each warm telescope or instrumental surface adds
a fraction of its blackbody radiation Plank
function to the beam. The fraction is the
emissivity 1- reflectivity. The Plank function
gives the absolute temperature T and wavelength
dependence
Wien displacement law gives peak of distribution
49
IR background subtraction
Wobbling secondary mirror switches between source
and reference position on sky 10 - 20
Hz Required for 3.5?m
Remain in each nod position for mins then
switch - avoid systematic sky gradient effects
50
IR array construction
NOT charge coupled pixel charge does not move
so can be read out non-destructively BUT no
on-chip binning. Each pixel has own FET which
acts as a buffer
51
Example IR array detail(SBRC 62x58 InSb)?
52
Example IR array layoutSBRC 256x256 InSb(other
types vary extensively)?
53
IR array properties
54
Possible Future Detectors
6x6 STJ 25 micron pixels
UV QE

• Superconducting Tunnel Junction (STJ)?
• http//astro.estec.esa.nl/SA-general/Research/Dete
  ctors_and_optics/detectors_home.html
• lt6x6 pix (18x50 prototype), T lt 1K
• high QE 200nm to 2 microns (60 - 70)?
• PROPORTIONAL DETECTOR (spectroscopic imager with
  wavelength resolution 10nm) A 3D camera!
• Photon counting

55
Superconducting Tunnel Junctions
STJ's consist of two thin films of
superconducting metal (niobium, tantalum,
aluminium etc) sandwiching thin insulating layer.
Cooled below the superconductor's critical
temperature (typically lt 1 K), single photons
perturb device's equilibrium.
Applied magnetic field suppresses Josephson
current and the bias voltage leads to extraction
of charge proportional to energy of perturbing
photon.
Spectral resolution on order of 1-10 in near UV
Response of a tantalum-based STJ to single-photon
illumination at 500 nm.
56
ESTEC Prototype 18x50 STJ array(not all pixels
connected)?
57
Possible Future Detectors III
Rockwell Hybrid Visible Silicon Imager The
High-performance CMOS Alternative to CCDs
HyViSi uses an FPA-type multiplexer with an
optical Silicon imager combines best of IR and
CCD detectors. Currently in 640x480 format, 2048
x 2048 under development High QE from 200-1050nm
58
CCD controllers

• Complex electronics
• Capable of providing user-defined waveforms to
  control CCD readout
• User-defined voltages
• Usable with multiple CCDs simultaneously

59
e.g. SDSU controller
60
LuckyCam CCD controller