CCDs for Big BOSS Chris Bebek LBNL 20
CCDs for Big. BOSS Chris Bebek LBNL 20 November 2009
Baseline Functional Block Diagram LBNL e 2 v (LBNL? ) BOSS model
QE “Blue” QE “Visible” and “Red” CCD QE
Fully depleted, back-illuminated CCD LBNL CCD concept 1) Fabricate a conventional CCD on a thick, high -resistivity silicon substrate 200 -250 µm typical 675 µm in special cases 2) Use a substrate bias voltage to fully deplete the substrate of mobile charge carriers 3) The thickness results in high near-infrared quantum efficiency and greatly reduced fringing 4) The fully depleted operation results in the ability to control the spatial resolution via the thickness and the substrate bias voltage
Quantum efficiency Note abs. length approaching 1 mm Note temp dependence
Spatial resolution: Effect of substrate voltage Electric Field VSUB = 5 V y Depleted Undepleted Depletion edge Point source illumination Measured charge distribution Each square represents 1 pixel
Spatial resolution: Effect of substrate voltage Electric Field y Depleted Depletion edge Point source illumination VSUB = 115 V
Thick CCD – crazy talk • Motivation - Introducing Hg. Cd. Te into the detector mix is – Expensive – Thermally complicated – Induces early aging onset • Can we ameliorate the pain of eliminating the NIR with a stretch of a CCD design? • Let’s consider a 500 µm thick CCD – Look red end QE – Estimate of the dark current – PSF impact – Stray light – Cosmics / radioactivity background • Study plan
Si Absorption Length
QE Temperature Dependence Hg. Cd. Te QE 500 µm thick Note: CCD photons can be more valuable than Hg. Cd. Te ones depending on readnoise of the latter I’m going to use 175 K for what follows.
Sample QE Temperature Trends 100 90 70 60 50 250 µm, 50 nm ITO, 100 nm Si. O 2 40 30 20 100 120 140 160 1050 T (K) Wavelength (nm) @ 25% QE QE @ 1000 nm 80 180 200 1045 1040 1035 1030 1025 1020 1015 1010 1005 1000 120 140 160 T (K) 180 200
Dark Current for Thick Device 100000. 0 Dark Current (e-/hr/pixel) 10000. 0 Data for 250 µm device 1000. 0 100. 0 Predict <10 e/hr @ 175 K 500 mm device. 10. 0 1. 0 0. 1 120 130 140 150 160 170 180 T (K) 190 200 210 220 230
Optics • What happens to the tip of the f-cone focused on surface of thick silicon when the optics is fast? • Large index of refraction of silicon will help a lot. Enough? I use a constant in what follows.
Experiment - Ray Generation • • • Set temperature to 175 K. Scan wavelength. Toss incidence angle for f/2. 5 beam. Generate photons, propagate, and convert at exponential sampled conversion depth. Drift each photon the remaining distance to the pixel plane with a random diffusioninduced offset. • Weight the photon increment to a pixel by f-cone radius of the photon. • Note – this is a photon-as-particle approach, not as a wave.
n=1 n = 3. 6 Note re diffusion: I have assumed that a 500 µm device can be over-depleted to the same extent as a 200 µm device – we know we can do this with the right silicon. A photon that converts at the optical surface will then have a 10 µm rms diffusion constant. This scales down linearly with the conversion height.
f/2. 5 radius weighted 150 µm fiber pixel in pixel plane
f/2. 5 convoluted pixel We’ve thrown some energy outside the PSF. I think there is room to reduce diffusion to partly address this. in pixel plane
3 -phase, overlapping triple polysilicon gate electrode pixel n(Si) ~ 3. 6 n(Si. O 2) ~ 1. 5 What happens to this photon? Poly 2 Poly 3 Poly 1
What happens when you go deep 1 µm wavelength ~2 µm diameter pin hole projected onto 250 µm thick CCD. 60 x 60 pixels N. Mostek
2048 x 4096 5 µm rms diffusion, 250 µm thick 1800 s mountain top cosmic rays 375 x 700 region
1800 sec lab dark – dominated by radioactivity Don’t let this happen. Shield.
Integral threshold plots Thickness variation
What next • Have identified Ti. N high-r wafers (10. 6 -14. 2 k. W-cm) SNAP CCDs – 650 mm will deplete at 100 V. • Backside processing at MSL is possible with existing equipment, unmodified. • Thin and process two wafers to ~600 µm thickness. • Measurements required – QE versus T – high side wavelength cutoff sensitivity – Dark current versus T – bulk contribution important for thick device – Pin hole projector versus T versus Vdepletion • Green light baseline – diffusion constant • >1 mm light – coupling/reflection from frontside.
- Slides: 23