Single Electron per Pixel Counting with Fully Depleted Charge Coupled Devices

Transcription

1 Single Electron per Pixel Counting with Fully Depleted Charge Coupled Devices Miguel Sofo Haro 1,2,3 1 Instituto Balseiro, Universidad Nacional de Cuyo, Argentina. 2 Consejo Nacional de Investigaciones Científicas y Técnicas (CONICET), Argentina. 3 Fermi National Accelerator Laboratory (Fermilab), Batavia, IL, United States IEEE Nuclear Science Symposium and Medical Imaging Conference 1 / Miguel Sofo Haro Single Electron per Pixel Counting with Fully Depleted Charge Coupled Devices 1/20 20

2 CCD semiconductor detector 200 µm thick Fully-Depleted 4 Mpixels cm2 readout noise of e rms per pixel. 2 Miguel Sofo Haro Single Electron per Pixel Counting with Fully Depleted Charge Coupled Devices /20 2/20

3 Charge Coupled Devices (CCDs) The charge collected by each pixel is shifted up to the sense node, where it is converted to a voltage through a small capacitance. C SN < 0.05pf S V/e > 3µV/e low readout noise p channel vertical gates E substrate voltage With fully depleted thick CCDs it is possible to achieve 95% of QE in the infrared range. 675 µm, 6 6 cm 2 detector have a mass of 5.2 g The low noise and considerable mass has motivated their application in low energy threshold particle experiments. Two examples are CONNIE (Coherent Neutrino Nucleous Interaction Experiment) and DAMIC (Dark Matter in CCDs). 3 / Miguel Sofo Haro Single Electron per Pixel Counting with Fully Depleted Charge Coupled Devices 3/20 20

4 Electronic readout noise (CCDs) RG VR MR M1 VDD T ON MR T Reset level ΔT Front-end T Signal level Q M1 Vn CSN CCD FE Vn G DAQ Vn DAQ RL σ 2 [µv 2 ] OPTIMAL T Integration time [µs] σ W 2 2 σ 1/f 2 σ 2 1/f σ =σw+σ1/f+σ1/f 2 T excellent for removing high frequency noise but sensitive to low frequencies 1/f impose a minimum noise. R&D started at Fermilab (PI: Javier Tiffenberg) to reduce the readout noise with a skipper CCD. 4 / Miguel Sofo Haro Single Electron per Pixel Counting with Fully Depleted Charge Coupled Devices 4/20 20

5 Skipper CCD: Multiple sampling In the time domain In the frequency domain Noise PSD, [nv/sqrt(hz)] Frequency response amplitude frequency [khz] Averaging multiple samples remove high and low frequency noise. (a) 1/f noise (b) T=55 mus 1 sample. (c) T=5.5 mus 5 samples. (b) and (c) have the same readout time but with (b) the effective noise bandwidth is reduced (G. Moroni et al. s x) 5 / Miguel Sofo Haro Single Electron per Pixel Counting with Fully Depleted Charge Coupled Devices 5/20 20

6 Skipper CCD output stage circuit and operation To take multiple samples of the charge packet in a non destructive way, a floating gate amplifier was introduced in the output circuit of the CCD. Proposed in 1990 by Janesick et al. (doi: / ) 6 / Miguel Sofo Haro Single Electron per Pixel Counting with Fully Depleted Charge Coupled Devices 6/20 20

7 The sensor Designed at the Microsystems laboratory of LBNL 4k 1k (4M) pixels of µm2 (total area of cm2 ) 200 µm thick high resistivity silicon fully depleted with 40V. Each quadrant was designed with a different SN area U1 L1 U2 L2 SN area 15 4 µm µm µm µm2 7 Miguel Sofo Haro Single Electron per Pixel Counting with Fully Depleted Charge Coupled Devices /20 7/20

8 The instrument The system integration was done at Fermilab Operated in vacumm and below 130 kelvin to reduce dark current Custom cold electronics Optimization of operation parameters 8 Miguel Sofo Haro Single Electron per Pixel Counting with Fully Depleted Charge Coupled Devices /20 8/20

9 Readout electronics Signal chain: Skipper CCD Line Driver Flex Vdd Cable (50cm) R1 Preamplifier R2 Coax Cable Readout System LMPS 10K 50Ω Monsoon RL <140ºK 290ºK Vacuum chamber 9 / Miguel Sofo Haro Single Electron per Pixel Counting with Fully Depleted Charge Coupled Devices 9/20 20

10 Sequencer and image processing Counts per bin First sample Second sample th Pixel Sample Value (ADU) The sequencer: timing voltages of the clock signals, the bias of the FG and image processing was developed to achieve the same gain (equal amount of charge) in all the samples 10 / Miguel Sofo Haro Single Electron per Pixel Counting with Fully Depleted Charge Coupled Devices 10/20 20

11 The result: pixel charge distribution 11 / Miguel Sofo Haro Single Electron per Pixel Counting with Fully Depleted Charge Coupled Devices 11/20 20

12 The result: pixel charge distribution 11 / Miguel Sofo Haro Single Electron per Pixel Counting with Fully Depleted Charge Coupled Devices 11/20 20

13 The result: pixel charge distribution 11 / Miguel Sofo Haro Single Electron per Pixel Counting with Fully Depleted Charge Coupled Devices 11/20 20

14 The result: pixel charge distribution It is possible to distinguish pixels with 0,1,2,3... electrons. 11 / Miguel Sofo Haro Single Electron per Pixel Counting with Fully Depleted Charge Coupled Devices 11/20 20

15 The result: minimum noise Low range Number of Pixel hlow Entries χ 2 / ndf / 22 Constant 1397 ± 15.8 Mean ± Sigma ± High range Number of Pixel hhigh Entries χ 2 / ndf / 21 Constant 23.3 ± 1.9 Mean 777 ± 0.0 Sigma ± Pixel Charge (e ) Pixel Charge (e ) With 4000 samples per pixel a readout noise of e rms/pix was achieved over a wide dinamic range The probability that the charge per pixel is misetimated by >0.5 e is p / Miguel Sofo Haro Single Electron per Pixel Counting with Fully Depleted Charge Coupled Devices 12/20 20

16 The result: minimum noise in the image 4.48 e rms/pix (1 sample) 13 / Miguel Sofo Haro Single Electron per Pixel Counting with Fully Depleted Charge Coupled Devices 13/20 20

17 The result: minimum noise in the image 4.48 e rms/pix (1 sample) e rms/pix (4000 sample) After averaging the 4000 samples it is possible to identify only one pixel with only one electron in the image Single-electron threshold equal to the silicon band gap (1.1 ev) single-photon sensitivity in the optical and near-infrared regime. 13 / Miguel Sofo Haro Single Electron per Pixel Counting with Fully Depleted Charge Coupled Devices 13/20 20

18 Image example 14 / Miguel Sofo Haro Single Electron per Pixel Counting with Fully Depleted Charge Coupled Devices 14/20 20

19 Image example 14 / Miguel Sofo Haro Single Electron per Pixel Counting with Fully Depleted Charge Coupled Devices 14/20 20

20 Image example 14 / Miguel Sofo Haro Single Electron per Pixel Counting with Fully Depleted Charge Coupled Devices 14/20 20

21 Image example 14 / Miguel Sofo Haro Single Electron per Pixel Counting with Fully Depleted Charge Coupled Devices 14/20 20

22 Readout speed The noise falls with the N uncorrelated samples. For a noise of 0.1 e rms/pix 1200 samples readout time of 12 ms/pix (10 µs/pixel/sample). 15 / Miguel Sofo Haro Single Electron per Pixel Counting with Fully Depleted Charge Coupled Devices 15/20 20

23 Linearity Exposure to light: 16 / Miguel Sofo Haro Single Electron per Pixel Counting with Fully Depleted Charge Coupled Devices 16/20 20

24 Linearity Exposure to a 55 Fe source: The distance between peaks keeps constant, showing the extremely linearity of the device. 17 / Miguel Sofo Haro Single Electron per Pixel Counting with Fully Depleted Charge Coupled Devices 17/20 20

25 Application of the skipper CCD: the SENSEI experiment Its aim is to look for dark matter signal in the electron-recoil channel. (made by Rouven Essig, Tomer Volansky and Tien-Tien Yu) 0.2 grams version was installed at MINOS (Fermilab) and it is currently taking science data. 18 Miguel Sofo Haro Single Electron per Pixel Counting with Fully Depleted Charge Coupled Devices /20 18/20

26 Conclusions The sensor readout electronics was optimized to get a low and white readout noise per sample. The sequencer, clock and bias voltages were set to properly operate the skipper output stage and the FG amplifier. An ultra low readout noise of e per pixel was achieved over millions of pixels on a stable, large-area detector. For the first time, the skipper CCD technology was successfully demonstrated, and single electron per pixel counting was possible. The detector has other potential applications like: exoplanets searches high resolution spectroscopy small size neutrinos detector for nuclear reactor monitoring A 10 grams version (20 skipper CCDs) of the SENSEI experiment is currently under development. 19 / Miguel Sofo Haro Single Electron per Pixel Counting with Fully Depleted Charge Coupled Devices 19/20 20

27 THANK YOU VERY MUCH 20 / Miguel Sofo Haro Single Electron per Pixel Counting with Fully Depleted Charge Coupled Devices 20/20 20

28 Neutrino event rate (ANGRA reactor Brasil - G. Moroni et al. PhysRevD ) 20 / Miguel Sofo Haro Single Electron per Pixel Counting with Fully Depleted Charge Coupled Devices 20/20 20

29 6 6 cm 2 CCD 20 / Miguel Sofo Haro Single Electron per Pixel Counting with Fully Depleted Charge Coupled Devices 20/20 20