Andrew Blue 24 th February 2007 Characterisation of Vanilla - a Novel Active Pixel Sensor for Radiation Detection Andrew Blue* 1, D Maneuski 1, A. Laing 1, V O'Shea 1, R Bates 1, S Bonhiek 2, A Costas 2, R Turchetta 3, A Clark 3 1 – Dept. Physics & Astronomy, University of Glasgow, 2 – Dept Med Physics UCL, 3 - CCLRC 11 th Vienna Conference on Instrumentation

Andrew Blue 24 th February 2007 Outline CCD vs. APS (Active Pixel Sensors) MI 3 - Multi Integrated Intelligent Imaging Vanilla APS Characterisation + Optimisation PTC & Spec. Response Stability Variation of Integration Time Linearity of A+D ADCs ROI Medical Applications UV Imaging (Etching) X-Ray Imaging –Scint + 3D DNA tagging Future work + Conclusions

Andrew Blue 24 th February 2007 CCDs Charge passed along columns Electron to Voltage conversion done at end CTI (Charge Transfer Inefficiency) from “bucket spillage” CCDs most popular imaging devices for decades Low noise High Sensitivity

Andrew Blue 24 th February 2007 Active Pixel Sensors Photon to Voltage conversion done within pixel Integrated electronics in circuit to suit applications (eg discriminator, flags) Low power consumption Cost falling with popularity of CMOS technology

Andrew Blue 24 th February 2007 MI 3 Project MI 3 = Multi-dimensional Integrated Intelligent Imaging Main goals are: To significantly extend the effective spectral response of APS’s from high- energy gammas and ionising particles to the infra-red (including the increasingly important soft X-ray/EUV regions) To develop on-chip "intelligence" down to the pixel level, through adaptive signal processing/pattern recognition, to extent the limits of delectability and applicability, and mitigate the problems of data overload. To provide a continuing responsive export core to meet the future imaging challenge within the UK science base; to ensure through building upon existing links with industry the future availability and exploitation of APS devices and systems. M-I 3 is supported by an RC-UK Basic Technology Programme 4-year £4.4 (€6.5) Million grant.

Andrew Blue 24 th February 2007 Who's Who Council for the Central Laboratory of the Research Councils (Instrumentation, Space Science and Technology Departments) University of Sheffield (Department of Electrical and Electronic Engineering) University of Glasgow (Particle Physics Experimental Group) University of Liverpool (Liverpool Semiconductor Detector Centre) University of York (Applied Electromagnetics and Electron Optics Research Group) Brunel University (Imaging for Space and Applications) Institute for Cancer Research (Royal Marsden Hospital) MRC (Laboratory of Molecular Biology, Cambridge) University College, London (Radiation Physics) University of Liverpool (Lab. for Environmental Gene Regulation) University of Surrey (Centre for Vision, Speech and Signal Processing)

Andrew Blue 24 th February 2007 Medical imaging High resolution autoradiography High Resolution Gamma Imager Low dose mammography Application Areas Space science Astronomy ‑ Adaptive Optics Solar Physics Space Science X ‑ ray sensor Electron Microscopy X-ray Photoelectric Electron Microscopy (XPEEM) Cryo-electron Microscopy of large molecules Synchrotron Radiation Applications Spectroscopy Diffraction UV/IR Electrons Soft X-ray Materials science X-ray Photon Coherent Scattering (XPCS) Lensless / Diffraction Imaging Bioarray imaging

Andrew Blue 24 th February 2007 MI3 Sensors Startracker – Test Device OPIC - On Pixel Intelligent CMOS HDR – High Dynamic Range Vanilla – ROI, Flushed Reset, A/D readout LAS – Large Area Sensor eLaNor - Low Noise Sensor

Andrew Blue 24 th February 2007 Vanilla 520x520 (25  m) pixels. 12-bit digital output for full frame mode also be readout in analogue mode. Readout rate 100 FPS for full frame 20 kHz (in analogue mode) at 10 bits ROI Comprises a standard three transistor pixel with diode. 3 possible reset modes - soft, hard and flushed. The sensor has be designed to operate with a readout noise of <25e - & full well capacity of ~100,000 e - Region of Interest (ROI) readout for up to 6 ROI (6x6 pixels/ROI), Pad layout allows for the butting of sensors on two sides.

Andrew Blue 24 th February 2007 Vanilla DAQ System System centred around a Memec Virtex-II Pro  20FF1152 FPGA development board Generates the require control signals for the target device Equipped with an optical transceiver to enable upload of the image data to a host PC at Gigabit speeds. Once uploaded to the host PC the image data is transferred efficiently to Labview via dedicated C++ middleware.

Andrew Blue 24 th February 2007 Photon Transfer Curve Photon Transfer Curve – logarithmic plot of rms noise of sensor vs signal mean under uniform illumination at various intensities Camera Gain Constant Read Noise – Independent of S => Const with illumination Shot Noise – N  S 1/2 (Poisson Statistics) => Gradient of 1/2 FPN – Fixed Pattern Noise.  S => Gradient of 1 FWC – Full Well Capacity. Max e - pixel can detect => Max. point on slope Quantum Efficiency

Andrew Blue 24 th February 2007 Photon Transfer Curve Illuminations achieved using super-bright, narrow bandwidth LED Coupled with diffuser produced uniform illumination (<1% dev.) Measurements made using standard settings for shot noise limited performances 30 Frames at each illumination stage PTC measurement using Hard Reset in Analog Mode

Andrew Blue 24 th February 2007 Performance Parameters ValueUnits Camera Gain10.9electrons/DN Read Noise51electrons Full Well (ADC saturation)~ 4 x 10 4 electrons Shot noise limited SNR47dB Dynamic Range70dB

Andrew Blue 24 th February 2007 Spectral Response Experiment Setup an experiment for Spectral analysis 1) Deuterium + Halogen Source 2) Automated Monochromator 3) Vanilla 4) Keithley 237 (misc. detectors)

Andrew Blue 24 th February 2007 Spectral Response Optical power of source measured using a calibrated Hamamatsu photodiode Measured from nm Fibre Optic incident on Sensor Removal of pedestal Integration of ADC units over exposed area

Andrew Blue 24 th February 2007 Spectral Response Interactive 520nm – 58% Max 600nm = 69%

Andrew Blue 24 th February 2007 Flushed Reset Soft reset results in a lowered reset noise i.e. (KTC/2) 1/2 instead of (KTC) 1/2 However, frames taken utilising soft reset are affected by image lag, where the current image is affected by the previous frame. Using hard reset, image lag is overcome, but results in full (KTC) 1/2 noise and reduced full well capacity. By using a hard reset followed by a soft reset, it is possible to get the best of both reset methods. This is known as flushed reset.

Andrew Blue 24 th February 2007 Variation of Integration Time Measurement of the effect of increasing integration time Measurement of dark fields at varying integration times AFTER pedestal subtraction for all 3 reset modes Increase in dark signal should be due to leakage curren and indpt of reset mode Measurements show Increase in leakage current with integration time for each reset mode Variation in leakage dependant on reset mode. Flushed reset shows least rise of all modes

Andrew Blue 24 th February 2007 Stability Tests Stability of sensor tested 50 frames acquired every 30 min for 72 hours Frames taken in light tight box Mean of frames plotted against time Leakage Current in Si devices doubles every 7K 15% variation = 1.1K

Andrew Blue 24 th February 2007 Linearity of ADCs Linearity of Digital ADC measured Plotted the Signal AFTER pedestal subtraction for increasing illuminations of LED Optical power of LED measured using calibrated photodiode and Keithley 237 unit Non-linearity of ADCs measured using least squared fitting method Non Linearity of ~0.98%

Andrew Blue 24 th February 2007 Region Of Interest 12 bit digital output for full frame mode. Region of Interest (ROI) readout for up to 6 ROI (6 x 6 pixels/ROI). 20kHz analogue readout at 10 bits resolution for ROI. M42 Camera lens attached to sensor, images taken using varying ROIs Max Measured frame rate – 24,395fps for 6x6 region 520x520 – 4fps 200x200 – 28fps50x50 – 432fps

Andrew Blue 24 th February 2007 UV Imaging Eliminate Dead Layers - UV photon and low-energy  ’s (from 3 H for example) absorbed in the thin inactive top layer. - To gain sensitivity the oxide layers overlaying the photodiode sites on the front surface can be removed (yields less than 100% fill factor) P EPILAYER P SUBSTRATE OXIDE Enhanced UV imaging essential for Proteomics Synchrotron Applications Plasma Physics

Andrew Blue 24 th February D Structures High-Z structures - High-energy radiation (e.g., x-rays, gammas) benefit from high-Z material - Can now be fabricated in polycrystalline form - These layers can be directly bonded on to the APS surface - Eliminates expensive and relatively unreliable bump-bonding technology. Transforming the unused thick Si substrate - 3D detectors - Etched vias filled with x-ray scintillators - Involves techniques to 1) Prevent active surface damage 2) Eliminate risks of high applied potentials 3) Develop compatible scintillators and filling procedures. P EPILAYER P SUBSTRATE OXIDE METAL SCINTILLATOR

Andrew Blue 24 th February 2007 Dual Energy Contrast Enhanced Mammography Currently evaluating Vanilla for the assessment of breast tumour angiogenesis. Vanilla’s on-chip functionality can be developed with properties to suit this application; such as fast readout and region of interest read out. Increased uptake of radiographic media that is strongly related with malignant tumours takes place within 2 minutes thus fast read out is required in order to accurately measure the rate of uptake. Additionally, ROI read out is required in order to reduce unnecessary dose to tissues that have normal uptake of radiographic media. Dual energy contrast enhanced mammography makes use of tumour angiogenesis, which causes cancer to take up contrast agent faster and to a greater extend than do normal tissue or benign masses

Andrew Blue 24 th February 2007 DNA tagging Spotted oligonucleotides, biotin labelled Added Streptavidin-Horse Radish Peroxidise (HRP) Washed Added Chemoluminescence reagent Achieved detection of high level of chemoluminescence signal After 5 min Level of white spots ~ 410 Average level ~ 260 Dejan Karadaglić University of Liverpool School of Biological Sciences

Andrew Blue 24 th February 2007 Future Work + Conclusions Conclusions An APS, Vanilla, characterised through Spec Response + PTC Measurements Novel features of ROI, Flushed reset and A+D readout tested Outline of future medical demonstrators for Vanilla Future Work Finish Optimisation of Digital Mode Construct MIPS experiment for B detection Show demonstrations in –Plasma physics –Medical –Em applications

Andrew Blue 24 th February 2007 LCFI The Linear Collider Flavour Identification (LCFI) Collaboration consists of particle physicists from five British academic institutions. The collaboration is pursuing an ambitious research and development programme to develop a pixel-based vertex detector for heavy flavour identification at the International Linear Collider. Vertex Detector for the ILC Inner layer pixels: 100  13 mm 2, 2500(V)  650(H) pixels per end; Outer layers: 2 sensors with size 125  22 mm 2, 6250(V)  1100(H) pixels; 120 detectors, 8  10 8 pixels (20 μm square) in total; doses up to ~10 11 – n eq /cm 2 What Readout speed is needed?