Introduction to Optical Detectors: Plates, PMTs and CCDs Matt A. Wood Florida Institute of Technology Dept of Physics and Space Sciences

Overview ● Photographic Plates ● The Photoelectric Effect ● Photomultiplier Tube Basics ● CCD Basics: Structure and Operation ● Quantum Efficiency ● Binning ● System Gain ● Noise Sources ● Optimal Data and Calibration Images ● Data Reduction Basics

Photographic Plates ● Used historically ● Wide FOV, high resolution ● But terrible Quantum Efficiency (QE ~ 1%) – QE = (#photons detected) / (#photons incident) x 100 – QE = 100% means you count every photon that hits detector. ● Non linear behavior, so difficult to get good magnitudes using plates ● Photographic plates no longer used at major observatories

The Photoelectric Effect Photon with energy greater than the work function of the material can free an electron h = W + KE max –No emission for frequencies below c = W/h –Current proportional to light intensity above c –Current proportional to frequency above c –Einstein Nobel work, established photon nature of light In a Photomultiplier Tube (PMT), photon ejects electron, starts cascade (see diagram later) In a Charge Coupled Device (CCD) photon creates electron-hole pair. Electrons attracted to buried electrode

Photomultiplier Tube Basics ● “Electron multiplier phototube” = “photomultiplier tube” ● Basic need: single electrons released via photoelectric effect can’t be measured, so stack a series of plates, and let electrons cascade – Photon releases electron at cathode – A dynode is placed close, and with a potential difference of ~100V. When electron strikes the dynode, 2-3 electrons are released. – Stack several dynodes, and finally detect pulse of ~10 6 electrons at the Anode

PMT Applications Historically, preferred to photographic plates for measuring magnitudes High time-resolution astronomy (still beats CCDs in this area) Describe 3-star photometer …

Sources of Noise ● Dark Current / Thermal noise ● Random pulse sizes ● Cosmic Rays ● Magnetic Fields (use  -metal) ● Also, aging from vacuum leakage, bright illumination ● Keep in light-tight enclosure

Photoelectric Effect in Semiconductors Atomic Energy levels perturbed if nearby atoms Split to 2 levels if 2 atoms N levels if N atoms -> BAND structure

Photoelectric Effect in Semiconductors N levels if N atoms -> BAND structure In metal, outermost electrons are in the valence band – free to conduct charge If valence level filled – insulator Require vacant sublevels, “dope” with impurities to make semiconductor –n-type: current carried by e- –p-type: current carried by “holes”

Photoelectric Effect in Semiconductors In semiconductor, there is only a small gap between valence and conduction band Thermal motions, photon absorption can excite e- to conduction band. Once in conduction band, the e- can move through the semiconductor, for example towards an electrode with +charge

CCD Basics ● Physical Structure ● Transferring Charges ● Binning (Figures from Apogee ccd.com website)

CCD Basics: Single Pixel Basic Structure of a single pixel Electrode insulated from semiconductor via thin oxide layer. +Voltage attracts e-, repels holes In effect a radiation- driven capacitor Figures from Astrophysical Techniques by Kitchin

Basic Structure: Array Array of pixels with insulators between (high p-type doping) Each develops charge proportional to illumination intensity Now just need to read it out

Multiple Electrodes & Charge Transfer If charge is under B, and all A-D are at +10V, then charge will diffuse to be equal under all 4 If A and C kept at +2V, though, then charge remains under B This allows us to transfer charge Charge transfer efficiency %

Multiple Electrodes & Charge Transfer

3 Phase CCD

3 Phase CCD: Operation During Exposure, photogate at +10V, gate at +2V To readout, gate at +10V as well as central electrode in each pixel Lower photogate to +2V Allow transfer, then ramp gate to +2V (charges now over central electrodes) Can start next exposure while clocking out charge in CCD array

2-Phase CCDs Are Most Common Front Illuminated vs. Back Illuminated Noise: Dark current Sensitivity Variations (pixel-to-pixel) Cosmic Rays Bias counts from electrons pulled to electrodes even in a zero-second exposure Requires only a single clock, but requires buried electrodes

Quantum Efficiency ● QE = (#photons detected) / (# photons Incident) The closer to 100% the better! Detector absolutely needs to be linear for you to do photometry Note: Different sensitivities at different wavelengths, so must calibrate through each filter. Also, must integrate longer for same S/N in regions with lower QE.

System Gain ● Our SiTE 502AB chip has a well depth of over 300,000 electrons ● The output node has a 16-bit A/D converter, so can represent numbers from 0 to 65,535. ● Gain is 6.1 e-/ ADU, so approx 50,000 ADUs if 300,000 electrons (sometimes called inverse gain)

Sources of Noise Readout Noise: Imperfect repeatability when charge read through A/D converter, and other unwanted counts from electronics Dark current: Thermal motions of atoms bump electrons into valence band – lower temp for lower dark current

More Noise ● Shot noise: if Poisson statistics (which photon arrival times obey): Signal proportional to countsS  C Noise proportional to sqrt(counts) N   C So S/N ~  C So to get S/N of 100:1 -> need 10,000 photons (I usually aim for 10k ADUs for sky flats, target star, etc., if possible)

Practical Aspects ● Bias Frames: Take early evening, and/or at end of night. I take 30 bias frames and median filter to produce a Master Bias. ● Dark Frames: CCD temp must be same as data frames. Exposure must be at least as long as longest data frame – longer is ok. (e.g., 20 5-min exposures). Bias subtract and Median Filter to remove cosmic rays -> Master Dark ● Sky Flats: Take as many as possible, but at least 3/filter. Shoot for 10k-20k ADU per pixel (definitely <30k/pixel). If using filters, go UBVRI – do less sensitive wavelengths first when sky is brighter. Either Tel drive off, or dither b/n frames. Bias subtract, Dark Correct, then Median filter weighted by counts for Master Flats. Note these are normalized to unity.

Data Reduction Steps ● Download raw frames ● Make master Bias frame (IRAF: zerocorrection) ● Apply Bias correction to dark frames and make master Dark (IRAF: darkcorrection) ● Apply Bias and Dark corrections to sky flats and make master Flat(s) (IRAF: flatcorrection) ● Apply Bias/Dark/Flat corrections to data frames (IRAF: ccdproc) ● That's it – now you're ready to extract photometry or do other image analysis!