- Open Research Online
Contents
1. P Nn oo R fon S B Howell Handbook of CCD Astronomy Cambridge University Press 2006 pp 94 95 m o e SS N J R Srour et al Review of Displacement Damage Effects in Silicon Devices IEEE Trans Nucl Sci 50 2003 653 A Refregier et al Summary of DUNE Mission Concept Proc SPIE 7010 2008 701018 D J Hall Euclid Monte Carlo Charge Transfer Model Open_Euclid_TR_09 01 2010 o o 10 D J Hall Comparison and Applicability to the Euclid Mission of the ESA CDM and the OU Monte Carlo modelling of CCD Radiation Damage OPEN EUCLID TR 20 V1 2011 11 A Short A Charge Distortion Model for Euclid EUCLID TN ESA AS 003_0 20102. BY IOP PUBLISHING FOR SISSA RECEIVED October 26 2011 ACCEPTED December 19 2011 PUBLISHED January 13 2012 THE 9 INTERNATIONAL CONFERENCE ON POSITION SENSITIVE DETECTORS 12 16 SEPTEMBER 2011 ABERYSTWYTH U K Modelling charge storage in Euclid CCD structures A S Clarke D J Hall A Holland and D Burt e2v centre for electronic imaging PSSRI The Open University England body Technologies Plc Chelmsford England E mail a s clarke open ac uk ABSTRACT The primary aim of ESA s proposed Euclid mission is to observe the distribution of galaxies and galaxy clusters enabling the mapping of the dark architecture of the universe 1 This requires a high performance detector designed to endure a harsh radiation environment The e2v CCD204 image sensor was redesigned for use on the Euclid mission 2 The re sulting e2v CCD273 has a narrower serial register electrode and transfer channel compared to its predecessor causing a reduction in the size of charge packets stored thus reducing the number of traps encountered by the signal electrons during charge transfer and improving the serial Charge Transfer Efficiency CTE under irradiation 3 The proposed Euclid CCD has been modelled using the Silvaco TCAD software 4 to test preliminary calculations for the Full Well Capacity FWC and the channel potential of the device and provide indications of the volume occupied by varying signals These results are esse
3. CCD273 Pixel FWC electrons 180k 240k 212k Peno Voltage 6 9 7 1 Other measureable parameters such as channel potential cno and Full Well Capacity FWC are taken to give a good indication of the model s general accuracy Verifying the signal volume model is more difficult but can be done through CTE characterisation where a change in CTE performance between the CCD273 and the CCD204 can be compared against the change in charge packet volume in the models of the two devices However this only offers an indirect indica tion of the model accuracy and is dependent on several assumptions for trap interaction Further verification will come from the effectiveness of the radiation damage models to which this work contributes 9 The FWC is a soft parameter because its value can vary depending on the limits set due to the fringing fields and measurement method so a value range is given The experimental results are obtained from the first batch of manufactured devices shown in table 1 The values in the table show a rough agreement between the Silvaco models and the first test devices Benchmarking work is ongoing 2 5 Signal volume theory Current trapping models make assumptions about charge packet volume to model the charge trans fer in CCDs The various models can be described by the interaction volume function given in an ESA technical note written by A Short 11 shown in equation 2 1 and graphically
4. The same error can be seen on the 10 electrons zm plot when the signal level drops below 10 electrons The values of the lower density cut off measurements are most reliable at these low signal levels At small signal the gradient of the signal volume plot appears to flatten before the small signal errors mentioned previously begin to appear This changing gradient is not included in the simple volume function of equation 2 1 but might be expected according to some signal volume models 10 The generally linear trend of the plots in figure 3 at larger signals gives rise to simple functions which can describe the increasing charge packet volume in relation to the signal size this enables the charge trapping models to be as realistic as possible However the reducing gradient at small signals as evident for 1 electron m3 diverges from that predicted in theory so an extra variable is necessary in the signal volume function 4 Implications for charge transfer The functions derived from the signal volume plots are in the form Volume yS a 4 1 Where S is the signal size B is the fitting parameter as already described describes the changing gradient at small signal and y is a scaling factor which depends on the cut off density 9 Using the different cut off densities to define the edges of the charge packet allows the separa tion of trap time constants according to SRH theory 9 and enables trapping volumes to be defined where o
5. measure its dimensions figure 1b which allows an approximation of the charge packet volume It is necessary to take different charge density cut off values to define the edge of the charge packet because of the gradually reducing edge density These values effectively define the trapping volumes as charge trapping is dependent on the charge density in the vicinity of a trap 9 There fore areas where the charge density of the signal is sufficient for capture can be approximated by a function produced from the signal volume model Each of the charge density cut off values used to define the edge of the charge packet are shown in figure 1b 1 electron um is equivalent to 10 electrons cm 10 electrons um is equivalent to 10 3 electrons cm and 100 electrons um is equivalent to 10 electrons cm For reference 10 electrons cm is the carrier concentration of intrinsic silicon 2 4 Benchmarking The models need to be verified and benchmarked to ensure they offer a realistic representation of the Euclid device The aim of the models is to gain functions which describe the charge packet volume over a range of signal levels for use in the radiation damage studies However the vol ume values obtained cannot be directly verified since charge packet volumes cannot be directly observed in the test devices Table 1 Comparison of model values against preliminary test results Silvaco Model Experimental Results CCD273 Pixel
6. e detector cause displacement damage in the silicon lattice where Si atoms in the path of the particle are pushed away from their lattice position leaving behind a vacancy These vacancies are unstable and are able to migrate through the device until they become stable Phosphorous atoms are able to stabilise these vacancies but the combination of Si vacancy and P atom creates an electron trap known as a PV centre 7 Phosphorus is the element used predominantly to dope the buried channel in n channel devices as such PV centres are likely to be located randomly throughout the buried channel Traps located in the buried channel will be able to trap charge from passing charge packets thus reducing CTE 2 1 Emission Traps with emission times much larger than the readout speed may release trapped charge after the image has been read out of the device thus reducing the signal levels Traps with shorter emission times comparable to the transfer time from pixel to pixel may release trapped charge from one charge packet into the next distorting the PSF of a point source such as an astronomical object This effect is critical in the Euclid mission where the weak lensing survey depends on the accurate measurement of distortion of galaxy shapes and positions 1 2 2 Capture probability The radiation damage models calculate the probability of capture using Shockley Read Hall the ory 9 The only traps affecting charge transfer are those in the vicin
7. ed These gen erally follow the trend expected for increasing charge packet volume described by the interaction volume function 9 However the signal volume plots deviate from the expected trend at smaller signals creating the need for an extra variable in the function At high signal levels Ne gt 1 x 10 electrons the difference in trapping volumes observed between the CCD273 and the CCD204 is significantly lower than the expected 2 5 times factor This will limit the expected improvement in CTE in this signal range However at small signals the volume difference between charge packets in the two devices is approximately 2 5 times larger in the CCD204 This should allow the expected 2 5 times improvement in CTE performance in this signal range where it is needed to meet the Euclid mission objectives These findings will be verified during future CTE characterisation of the devices Acknowledgments With thanks to the staff at The Open University and at e2v Technologies who have helped and contributed to this work References 1 ESA Euclid Science Requirements DEM SA Dc 0001_1_0_Euclid_SciRD_2008 2008 S Bowring e2v technologies plc CCD273 Design Report EUCV E2V RP 002 2010 w N J Pickel et al Radiation Effects on Photonic Imagers A Historical Perspective IEEE Trans Nucl Sci 50 2003 12 Silvaco Inc ATLAS User s Manual 2010 R Laureijs Euclid Science Requirements DEM SA Dc 00001 2008
8. in figure 2 b Vc Ne Co 2 1 Ve awe aa Where Vc is the volume of the charge packet V is the volume of the charge packet at full well capacity Ne is the signal size FWC is the signal size at full well capacity and B is simply a fitting parameter which describes how the charge packet changes with signal size One of three assumptions about the charge packet might be made 10 e A charge packet will always fill the volume available to it only increasing its charge density when charge is added density driven model B 0 e Charge density will remain constant and only the volume which the charge packet occupies will increase as charge is added volume driven model B 1 e Alternatively somewhere in between the two extremes where both charge density and charge packet volume change as charge is added 0 lt B lt 1 This final assumption is currently thought to be the most physically realistic overall but there may be some deviation from the trend at small signal levels 0 T r 0 0 2 0 4 Ne 0 6 0 8 1 FWC Figure 2 Plot of the signal volume function 11 showing the density driven model 6 0 volume driven model 1 and the more realistic assumption 0 lt B lt 1 normalised against FWC It is possible to observe and measure the charge packets in the Silvaco device simulations in three dimensions figure 1 These charge packets do not have a well defined edge but gradually reduce in charge dens
9. ity of the charge packet mak ing the signal volume models essential for a more realistic radiation damage model The signal volume plots are produced using the Silvaco models of the detector device structures with charge A Device Electrodes 100e um 10e um 1e um Oxide 0 Distance Microns Electron Concentration cm 3 Depth into the Device Microns Kamia Electron Conc cm3 in tree ara a 18 2 22 24 26 28 3 32 34 9 8 7 6 5 4 3 2 Distance across the device Microns Distance from the Surface of the CCD Microns Figure 1 a shows a 2D cross section of the charge packet in the Euclid pixel b is the 1D plot of electron concentration across the charge packet shown in a indicated by the dashed line induced allowing the observation of the charge packet size location and density Plotting the vol ume occupied by the charge packet against the signal size gives rise to functions which describe the signal volume relation It is these functions which are used in the radiation damage models 2 3 Signal volume in Silvaco Models Images of the charge packet in 2 dimensions can be taken from the 3D device models as in fig ure la The charge packets have a high density core equivalent to the maximum doping density for the buried channel This reduces gradually at the edges of the charge packet until it becomes negli gible A 1D plot of charge density can be taken across the charge packet to
10. ity until the density becomes negligible When measuring the charge packet dimensions in the models it is necessary to take a cut off such that the charge packet can be mea sured and the volume calculated for each cut off density In this way plots of volume against signal size can be produced figure 3 Measurements of charge packet volume are used to derive an empir ical formula for use in the radiation damage models to describe the changing volume with signal The volume calculations are static measurements but Shockley Read Hall theory shows that the probability of capture is dependent on charge density in the vicinity of the trap capture cross section this varies for different traps and dwell time of the charge packet under each CCD phase The functions derived from the signal volume plots can be related to the capture time constant de pending on the trap species and the charge density to calculate capture probability in the radiation damage models for different dwell times which relate to the CCD clocking schemes 9 3 Signal volume characterisation The signal volume plot produced as shown in figure 3 gives the image area pixel left and readout register element right for the CCD273 It is worth noting that the theoretical plot in figure 2 is normalised against FWC and FWC volume but the FWC of the two structures in figure 3 are different hence the adjacent figures are plotted over a different range of signal sizes These signal v
11. k energy map of the universe 5 8 The detector used on the proposed mission will be formed from an array of CCD273s 2 These devices were designed and are currently being manufactured by e2v U K The detec tor was specifically designed for Euclid based on an older device but with changes designed to improve CTE under irradiation The design changes were necessary due to the harsh operating environment the estimated end of life 10 MeV proton fluence is around 6 x 10 protons cm in the worst case where the satellite orbit is at L2 and to provide the high accuracy required for a good weak lensing survey Radiation damage causes electron traps to form in solid state imaging devices If the traps are located in the buried channel these traps can capture electrons from passing charge packets during transfer Traps with fast emission times may release the captured charge into an adjacent packet distorting or smearing the image PSF 6 It is necessary to reduce the effects of radiation damage on the images captured such that distortion due to weak lensing can be separated from distortion caused by radiation damage The aim of this work is to aid the calibration of the radiation damage models and also to aid in the post processing of images to remove the effects of radiation damage The radiation damage models require knowledge of electron distributions in the device structure such that the probability of trapping can be calculated
12. more accurately This study details the electron distributions obtained from three dimensional models of the Eu clid device structures to calculate charge packet volume produced using the commercially available Silvaco ATLAS software 4 The aim is to enable the calibration of the radiation damage mod els through empirically derived formulae which approximate charge packet volume based on the observations of the charge packet in the device simulations The device models are benchmarked using measurable parameters against values measured in test devices to ensure consistency However charge packet volume cannot be directly measured in the test devices and so it has to be inferred from CTE characterisation results The change in CTE performance between the two devices CCD273 and CCD204 is related to the volume occupied by charge packets and will indicate the general accuracy of the signal volume models 2 Radiation damage Charge is read out of CCDs by transferring the charge stored in each pixel in parallel along the transfer channels until it reaches the register and then transferred along the register in series until it reaches the readout node where it is output from the device These charge packets can undergo thousands of transfers until they reach the output node depending on the size of the CCD and their starting location in it so it is essential that the CTE is optimized to preserve image fidelity High energy particles incident on th
13. ntial for the realisation of the mission objectives and for radiation damage studies with the aim of producing empirically derived formulae to approximate signal volume characteristics in the devices These formulae will be used in the radiation damage charge trapping models The Silvaco simulations have been tested against real devices to compare the experimental measurements to those predicted in the models Using these results the implications of this study on the Euclid mission can be investigated in more detail KEYWORDS Detector modelling and simulations II electric fields charge transport multiplica tion and induction pulse formation electron emission etc Solid state detectors Corresponding author 2012 IOP Publishing Ltd and SISSA doi 10 1088 1748 0221 7 01 C01058 Contents 1 Introduction 1 2 Radiation damage 2 2 1 Emission 2 2 2 Capture probability 2 2 3 Signal volume in Silvaco Models 3 2 4 Benchmarking 3 2 5 Signal volume theory 4 3 Signal volume characterisation 5 4 Implications for charge transfer 6 5 Comparison of the CCD273 vs CCD204 7 6 Conclusion 8 1 Introduction The aim of the Euclid mission is to create detailed wide field images of the universe in order to cal culate the weak lensing effect that dark energy masses will have on the light from distant galaxies This will allow the mass of the intervening dark energy to be calculated without knowledge of its composition to produce a dar
14. olume distributions are shown over a logarithmic scale but the plots follow a curve similar to that shown in figure 1 when the fitting parameter B lies between 0 and 1 i e both charge density and volume change as charge is added to the packet The trapping volume diverges at small signal levels across the different cut off densities be cause the electrons which make up the charge packet disperse across the volume available in the 100 100 10 4 F n gt m be o ab Volume um 3 1 m m gt gt gt Volume um 3 e 0 1 4 1e um 01 s 1e um 0 01 E 10e um E m 10e um 100e pm 100e um 0 001 7 7 1 1 0 017 j T T T T j 1E 00 1E 01 1E 02 1E 03 1E 04 1E 05 1E 00 1E 01 1E 02 1E 03 1E 04 1E 05 1E 06 Charge electrons Charge electrons Figure 3 The signal volume plots for a the pixel and b the readout register structures of the CCD273 in logarithmic scale device reducing the overall charge density and making the low density volumes bigger At larger signals close to the FWC the difference in trapping volumes across the cut off values becomes much less pronounced as the charge packet becomes more densely populated across the whole available volume With very small signals less than 100 electrons the volume measurement for the cut off density defined as 100 electrons um 1014 electrons cm7 becomes less reliable due to the low signal level
15. the middle of the charge packet is not representative of the overall charge packet size The plot shows that the difference in the charge packet volume at small signals Ne lt 1000 is approximately 2 5 times larger in the CCD204 as expected However as signal levels increase the difference in volume occupied between the devices reduces to a factor of around 1 5 at high signal levels This will reduce the expected improvements of CTE at high signals and offers a way in which the devices can be tested to confirm the suitability of the models by measuring the improvement in CTE between the devices over a signal range 6 Conclusion The aim of these device models is to produce simple functions to approximate the interaction trap ping volumes of charge packets in the CCD273 pixel and register structures These functions will be used to help develop more realistic radiation damage models The radiation damage models reproduce the effects which radiation damage has on images with the aim of eventually post pro cessing captured images to remove the effects of radiation damage This information can only be approximated from device models as charge packets cannot be observed in test devices Device parameters such as FWC and the channel potential y 9 closely match those mea sured in the first production batch of the CCD273 verifying the general accuracy of the model and giving confidence in the signal volume models which cannot be directly confirm
16. ty iversi Open Research Online The Open University s repository of research publications and other research outputs The Open Un Modelling charge storage in Euclid CCD structures Journal Article How to cite Clarke A S Hall D J Holland A and Burt D 2012 Modelling charge storage in Euclid CCD structures Journal of Instrumentation 7 C0105 For guidance on citations see 2012 IOP Publishing Ltd and SISSA Version Version of Record Link s to article on publisher s website http dx doi org doi 10 1088 1748 0221 7 01 C01058 ttp iopscience iop org 1748 0221 7 01 C01058 Copyright and Moral Rights for the articles on this site are retained by the individual authors and or other copy right owners For more information on Open Research Online s data on reuse of materials please consult the policies page oro open ac uk IOP scien ce iopscience iop org Home Search Collections Journals About Contactus My lOPscience Modelling charge storage in Euclid CCD structures This content has been downloaded from IOPscience Please scroll down to see the full text 2012 JINST 7 C01058 http iopscience iop org 1748 0221 7 01 C01058 View the table of contents for this issue or go to the journal homepage for more Download details IP Address 137 108 145 39 This content was downloaded on 15 11 2013 at 15 31 Please note that terms and conditions apply inst PUBLISHED
17. utside this volume the charge density is not high enough to trap electrons in the transfer 2 8 2 6 2 4 2 2 2 1 8 1 6 1 4 1le um 1 2 10e um _ 1 T T T 1E 00 1E 01 1E 02 1E 03 1E 04 1E 05 1E 06 Charge electrons Relative Volume CCD204 CCD273 Figure 4 Ratio of the volumes occupied by varying charge packet sizes in the CCD204 readout register when compared to the CCD273 read out register period Deriving the functions allows the charge trapping to be modelled as closely as possible to the physical system 5 Comparison of the CCD273 vs CCD204 The CCD273 is redesigned from the older CCD204 to improve operation under irradiation which is achieved by reducing the width of the transfer channel in the serial readout register from 50 um to 20 um It is assumed that this 2 5 times reduction in channel width will improve serial CTE under irradiation by roughly the same factor by reducing the volume in which the charge packets stored After simulating the charge packet volume in both device structures and extracting the fitting pa rameters it is possible to plot a comparison of the volume occupied by the same charge packet across the two structures The difference in volume occupied in each device is shown in figure 4 The lower density cut off values are plotted to offer a more realistic impression of the volume difference between charge packets in the two devices as the high density area in
Download Pdf Manuals
Related Search