04/20/23 1

Observational Astrophysics I

detectors and Calibrations

2

NIR detectors

NIR detectors are similar toCCDs

Special non-silicon layer is usedto generate photoelectrons: HgCdTe (Hawaii) and InSb (Indium Antimonide, “insbe”, Aladdin) are sensitive between 0.9 and 25 microns.

Silicon electronics is welldeveloped, therefore weuse hybrid systems

Working temperatures: 30-60 K

04/20/23

3

Thermal IfraRed detectors

Raw frame Reduced frame

04/20/23

4

Thermal IR

HgCdTe (“mercad”) arrays depending on the exact structure are sensitive in 1-17 micron range.

Detector needsto be cooled downto 5-10 K

Main problem isthermal emission:

04/20/23

5

Fighting thermal background

Cooling the wholeinstrument

Taking short exposures Chopping

and noddingthe telescope

Non-destructive readout04/20/23

6

Non-destructive readout We can measure accumulated charges in each

pixel without dumping the charges This can be done several times before the dark

current of detector catches up with the shot noise of the signal

Instead of using eachindividual frame wemeasure how chargesgrow (linear regression)

Typically we can make16-64 readout beforethe array must be reset Dark current

Readout & Shot noise

04/20/23

7

New integrating detectors

High-resistivity fully depleated CCDs with ≈0 readout noise!

Courtesy Lawrence Berkeley National Lab

04/20/23

8

High-resistivity CCDs

The first 2k2k results:

• Read-out at 10 MHz with readout noise of 0.2 e-

• QE at 950 nm > 80%

• Excellent charge transfer efficiency

• At 1 MHz can be also used as a PCD device

Courtesy Lawrence Berkeley National Lab

04/20/23

9

CMOS detectors The idea is borrowed from the IR

detectors

The integrating part is made out of silicon

CMOS multiplexor allows non-destructive readout, partial readout etc.

04/20/23

10

Superconducting devices

Superconducting Tunnel Junction (STJ - or Josephson junction) combines high QE, huge spectral range (from 100 nm to 3000 nm) and (some) spectral resolution

04/20/23

11

PCD

Photomultiplier

Multi-anodemicrochannelarray (MAMA)

04/20/23

12

PCD properties Noise sources: shot noise and dark current No readout noise (since there is no ADC) Cosmic rays are minor concern – detector

of choice for many space missions Limited dynamic range (why?) Linearity problem Can easily be tuned to any spectral range,

no need for thinning or other risky operations

Maximum QE is about 50% (why?) MAMA allows reading 2D frames

04/20/23

13

Comparison

CCDs Large dynamic range Large QE Extremely linear Large sizes (4k4k) Sensitivity drops

sharply in the blue and the red

Readout noise Cosmic rays Cooling

PCD Digital output in real time No readout noise Insensitive to cosmic rays No need for deep cooling Much easier to make and

therefore much cheaper Small dynamic range Small QE High voltages

04/20/23

14

Operation of astronomical detectors

Space: Test detectors as much as

possible and as many as possible Think of high radiation

background and large temperature variation

Think of detector aging Think of cooling (active and

passive) Automate calibration procedures Store all original calibration data

in case you want to go back

04/20/23

15

Operation of astronomical detectors

Ground: Think of detector orientation Think of cooling side effects

(flexure) Recycle calibration procedures Data flow and data reduction

04/20/23

16

Calibrations

Goals: convert data from detector coordinates to physical coordinates and remove detector signatures as much as possible

Bias: 0 second exposure(s) with shutter closedEstimate of electronic signal offset for log amplifier

Darks: variable length exposures with shutter closedEstimate of dark current rate

Flat fields: short exposures with homogenous illumination and open shutterEstimate of relative pixel sensitivity

Calibrated source exposuresEstimate of QE

04/20/23

17

Calibrations should not add noise!

Take a sequence of bias frames (or dark frames) Combine them rejecting cosmic rays, replacing

cosmetic defects and increasing S/N ratio (master bias)

Take a sequence of flat fields Combine them (master flat) and normalized the

flat Subtract master bias from master flat and

science frames Divide science frames by master flat

Therefore we:

04/20/23

18

CCD example: Bias

04/20/23

19

Flat field

Fragment of a master flat field

04/20/23