First results from the HEPAPS4 Active Pixel Sensor

Outline of the talkActive Pixel Sensors an introductionThe HEPAPS4 sensorBasic characterisationNoise components and reset behaviourPhoton Transfer CurveLinearityDark currentFirst test beam plots (very preliminary)Summary

Lars Eklund, INSTR08

Active Pixel Sensors IntroductionSilicon sensor technology based on industry standard CMOS processesMonolithic: Sensing volume and amplification implemented in the same silicon substrateSimilar use case as CCDsPhotonic applications:Biological/medical applications, HPDs, digital cameras, Envisaged HEP applicationsLinear collider vertex detector (~109 channels)Linear collider calorimeter (Si/W, ~1012 channels)


Active Pixel Sensors Principle of OperationFunctional descriptionPhoto diode: n-well in the p-type epilayer of the silicon Charge collection:e-h pairs from ionising radiationDiffusion of charge in epi-layerCollected by the diode by the built-in field in the pn-junctionIn-pixel circuitry built in p-well.Collected charge changes the potential on the source follower gate VG = QPD/CPDGate voltage changes the transconductancePixel selected by the select MOSOutput voltage = VDD-gds*IBiasSimplest design of APS: 3MOS pixel Photo diode Reset MOS (switch) Select MOS (switch) Source follower MOS


Active Pixel Sensor - Cartoonpixel size 10-100 m* Typical values found in different designs


Cartoon array of active pixels


(non-judgemental) pros and consOnly uses the epi-layer as active volumeQuite small signal (compared to hybrid pixel devices)Can be thinned to minimise materialBut S/N is what countsIntrinsically very low noiseFight other noise componentsRadiation hardnessRelies on diffusion for charge collection (no bias voltage)No charge shifting (as in e.g. CCDs)Compared to LHC style sensorsRelatively slow read-out speedVery low power consumption (saves services!)CostSi sensors relatively expensive per m2Standard CMOS processSaves integration costs


The HEPAPS4 large area sensor for HEP applicationsFourth in series designed at RALSelected most promising design in HEPAPS2Basic parameters15x15 m2 pixel size384x1024 pixels20 m epi-layer1 MIP = 1600 e- spread over several pixelsThree different design variants:Results presented here are from the single diode design


HEPAPS4 operating principle3MOS APS pixel as described previouslyLoop over all rows, select one at a time.Sample the signal on to a capacitor for each columnLoop over columns, read out the value through four independent output driversReset the rowRolling shutter: Continuously cycle through pixels to read-out and reset


Noise componentsFixed Pattern NoiseGain VariationPedestal variation280 ADC or ~1400e- for HEPAPS4Removed by subtracting:Pedestal frame from dark measurementTwo adjacent frames

Reset noiseAs described on next slideCan be removed by Correlated Double SamplingSample the reset value before charge collectionCan be done in H/W or (partially) offlineDark CurrentLeakage current in the photo diodeDepends on the Average offset = IleakShot Noise = sqrt(Ileak)Common mode noiseCan be partially subtractedRead noiseThe fundamental noise of the chip


Pixel Reset BehaviourPixels are reset by asserting a signal on the Reset SwitchThe charge collecting node is set to the reset voltage VRSTSoft Reset: VRST = VDDLess reset noise Hard Reset: VRST < VDDLess image lagDecreases dynamic range

Plot shows HEPAPS4 reset from fully saturated with soft reset operation (VRST=VDD).


HEPAPS4 as imaging sensorAPS sensor also used in digital camerasHEPAPS4 becomes a B/W 400 kpix cameraPlots show visually the effects of pedestal subtraction


Photon Transfer Curve (PTC) Basic PrinciplePlot signal variance vs. meanRead noise will dominate for low intensities, but:Conceptual PTC curveShine light on the sensor and increase the illumination gradually


Photon Transfer Curve - ResultsRead out a region of 200x200 pixelsSubtract pedestal frameCalculate mean signal per pixelSubtract two adjacent frames:Removes fixed pattern noiseCalculate variance per pixelAssume constant gain 1000-5000 ADCSlope gives gain = 5.1 e-/ADCNoise floor not shown, since it is dominated by systems noiseSaturation starts after 6000 ADCAverage variance vs. average mean for 40000 pixelsExpected value from design:7.8 e-/ADC


Photon Transfer Curve Gain DistributionsSame analysis for 40 000 pixels800 frames to calculate mean and varianceFit slope for each pixelGain = 5.1Gain RMS = 0.51# pixels with gain:< 3 e-/ADC: 41pix> 8 e-/ADC: 90 pix


LinearityNon-linearity arises fromChange in sense node capacitance when chargedNon-linearity of source followerMeasurement method:Constant, low illuminationContinuous acquisition of framesNo reset between framesRelative gain is the derivative of this curve, normalised to the gain at 0 ADCRelative gain > 90%0-1640 ADC0-8360 e-Relative gain > 80%0-3780 ADC0-19300 e-


Dark CurrentDark current is leakage current in the photo diodeMethod:Readout two consecutive frames without reset between the framesSubtract the two frames to remove pedestal and fixed pattern noiseVary the integration timeLinear fit to the curveSlope 530 ADC/s = 2700 e-/sPixel size 15x15 m2Idark = 0.2 nA/cm2


First Test Beam PlotsTest beam at DESY: 6GeV electronsCombined with ISIS pixel sensorConfiguration of sensor slightly different from photonic measurements in the labAnalysis in progressTwo muse-bouche plots:Hit map: accumulation of reconstructed clustersLandau: Cluster charge in ADC values


SummaryActive Pixel Sensors are an attractive technology for certain applications in HEPThe principle of operationHEPAPS4 a large area APS designed for HEP applicationsFirst results from the characterisation:Gain 5.1 e-/ADC with an RMS of 0.5 e-/ADC over 40k pixels90% of gain up to 8400 e-80% of gain up to 19300 e-Dark current 2700 e-/s per pixelTwo preliminary plots from the beam test


Documents

First results from the HEPAPS4 Active Pixel Sensor