Journal of Instrumentation

3D monolithically stacked CMOS active pixelsensor detectors for particle tracking applicationsTo cite this article D Passeri et al 2012 JINST 7 C08008

View the article online for updates and enhancements

Related content3D monolithically stacked CMOS ActivePixel Sensors for particle position anddirection measurementsL Servoli D Passeri A Morozzi et al

-

A two-tier monolithically stacked CMOSActive Pixel Sensor to measure chargedparticle directionD Passeri L Servoli S Meroli et al

-

Vertically integrated circuit development atFermilab for detectorsR Yarema G Deptuch J Hoff et al

-

Recent citations3D monolithically stacked CMOS ActivePixel Sensors for particle position anddirection measurementsL Servoli et al

-

A two-tier monolithically stacked CMOSActive Pixel Sensor to measure chargedparticle directionD Passeri et al

-

This content was downloaded from IP address 11714612371 on 12102021 at 1302

2012 JINST 7 C08008

PUBLISHED BY IOP PUBLISHING FOR SISSA MEDIALAB

RECEIVED June 6 2012ACCEPTED July 19 2012

PUBLISHED August 28 2012

WIT2012 mdash WORKSHOP ON INTELLIGENT TRACKERS3ndash5 MAY 2012INFN PISA ITALY

3D monolithically stacked CMOS active pixel sensordetectors for particle tracking applications

D Passeriab1 L Servolib S Merolib D Magalottib P Placidiab and A Marrasc

aUniversity of Perugia Dipartimento di Ingegneria Elettronica e dellrsquoInformazioneVia G Duranti 93 06125 Perugia Italy

bIstituto Nazionale di Fisica Nucleare (INFN) Sezione di Perugiavia Pascoli 1 06100 Perugia Italy

cDeutsches Elektronen-SynchrotronNotkestraszlige 85 22607 Hamburg Germany

E-mail danielepasseridieiunipgit

ABSTRACT In this work we propose an innovative approach to particle tracking based on CMOSActive Pixel Sensors layers monolithically integrated in an all-in-one chip featuring multiplestacked fully functional detector layers capable to provide momentum measurement (particle im-pact point and direction) within a single detector This will results in a very low material detectorthus dramatically reducing multiple scattering issues To this purpose we rely on the capabilitiesof the CMOS vertical scale integration (3D IC) technology A first chip prototype has been fabri-cated within a multi-project run using a 130 nm CMOS CharteredTezzaron technology featuringtwo layers bonded face-to-face Tests have been carried out on full 3D structures providing thefunctionalities of both tiers To this purpose laser scans have been carried out using highly fo-cussed spot size obtaining coincidence responses of the two layers Tests have been made as wellwith X-ray sources in order to calibrate the response of the sensor Encouraging results have beenfound fostering the suitability of both the adopted 3D-IC vertical scale fabrication technology andthe proposed approach for particle tracking applications

KEYWORDS Detector design and construction technologies and materials Particle tracking detec-tors (Solid-state detectors) Detector alignment and calibration methods (lasers sources particle-beams) VLSI circuits

1Corresponding author

ccopy 2012 IOP Publishing Ltd and Sissa Medialab srl doi1010881748-0221708C08008

2012 JINST 7 C08008

Contents

1 Introduction 1

2 The two-layer 3D CMOS APS detector 1

3 Electrical and functional characterization 231 The optical workbench 232 Signal to noise evaluation X-rays analyses 3

4 Conclusions 7

1 Introduction

Typical tracking systems for particle trajectory reconstruction in High Energy Physics experimentsare usually based on different separated sensing layers featuring pixels andor strips sensitive el-ements In this work we propose an innovative approach to particle tracking based on CMOSActive Pixel Sensors (APS) layers monolithically integrated in a all-in-one chip featuring multi-ple stacked fully functional detector layers capable to provide momentum measurement (particleimpact point and direction) within a single detector This will results in a very low material detectorthus dramatically reducing multiple scattering issues To this purpose we rely on the capabilitiesof the CMOS vertical scale integration (3D IC) technology In particular instead of using differ-ent tiers of the stacked 3D structure for heterogeneous integration (namely by devoting differenttiers to the sensing layer and to the analog and digital circuitry) identical fully-functional CMOSAPS matrix detectors including both sensing area and controlsignal elaboration circuitry couldbe stacked in a monolithic device by means of Through Silicon Vias (TSV) connections

The information coming from thinned multiple stacked layers could be usefully exploited toextend the detection capability of the monolithic sensor In principle such a detector would be ca-pable of giving accurate estimation not only of the impact point of a ionizing particle (with spatialresolution in the microm range) as well as of its incidence angle (with angular precision around 1) [1]A single detector allowing particle momentum measurement therefore could be built at the sametime being a low material detector multiple scattering effects are expected to be negligible with re-spect to conventional structures since incoming particles have to cross only few microm of bulk silicon

2 The two-layer 3D CMOS APS detector

A first chip prototype has been fabricated within a multi-project run using a 130 nm CMOS 3DCharteredTezzaron technology [2 3] featuring two layers bonded face-to-face The top (outer)tier has been thinned down to less than 10 microm while the bottom (inner) tier has not been modi-fied with respect to the standard planar (2D) realization The two tiers host two almost identical(ie mirrored) layouts featuring several corresponding test structures namely single pixels as wellas different matrices eg featuring 5times5 and 16times16 pixels (figure 1) Each pixel is based on the

ndash 1 ndash

2012 JINST 7 C08008

Figure 1 Schematic cross-section of the two tiers (sketch not to scale) Figure 2 Large (top) andsmall (bottom) pixel layouts

standard three transistors (3T) active pixel architecture featuring 10 micromtimes10 microm size with differentsensitive element (photodiode) layout In particular a small sensitive area with small capacitance(aiming at maximizing the charge to voltage conversion gain) and a large sensitive area (aiming atmaximizing the fill-factor of the detector) (figure 2) The structures of both tiers can be read out inparallel by means of dedicated (separated) output bond pads located at the backside of the top tier

3 Electrical and functional characterization

31 The optical workbench

For electrical and functional characterization a suitable read-out set-up has been devised and fab-ricated In particular an advanced optical workbench with IR UV and VISible laser heads withmicro-focusing (spot size below 2 microm) and micro-positioning (scan step 021 microm) capabilities hasbeen used (figure 3a) It allows up to four sensors parallel read-out for track reconstruction andspatial resolution analysis as well as 2D scans for surface mapping (figure 3b) [4]

Coincidence responses to a focused IR spot of corresponding outer and inner matrices havebeen obtained The coincidence responses can be clearly observed by translating the spot alongthe matrix (ie along the chip surface) and looking at the different responding clusters of pixels(figure 4) The focussing capabilities allow the stimulation of a single pixel on the top back-sideilluminated matrix while a wider and spread-out response has been obtained from the bottommatrix due to the scattering effect of the metal layers between the two chips and to the shieldingeffects of the bond-point metal-6 octagons whose dimensions (34 micromtimes34 microm) are of the sameorder of the pixel size Misalignments of about 12 microm in one direction and of about 20 microm on theorthogonal one have been demonstrated (figure 5) These findings have been confirmed by deeperinvestigations carried out by means of computed tomography (CT) scans at DESY Hamburg In

ndash 2 ndash

2012 JINST 7 C08008

Figure 3 (a) The optical workbench visible IR and UV wavelengths can beused 5 translational and 1 rotational stages allow very precise spot focusingand pointing

Figure 3 (b) Detail of the laserobjective stimulating the 3D chiphosted in a dedicated PCB (RAPS04daughter)

particular figure 6 shows a detail of the pad regions From the dimensions of the pads (120 microm)the estimated misalignment which is of the order of 10 of the pad size is about 15 microm ingood agreement with the previous findings It should be emphasized however that the significantmisalignment between the two tiers does not prevent the communication from the bottom tier tothe top tier The implemented circuits actually features TSV interconnections only at the chipperiphery (eg at the pad level) The huge number of TSVs in these regions even if the two tiersare tiltedshifted still guarantees the electrical connections between the pads

Surface scans can be carried out as well in order to evaluate the response of the pixel as afunction of the spot positions The micro-focusing and micro-positioning capabilities allow verydeep investigation of the point spread response of the matrix The response to a back-side illu-mination with visible light (531 nm) is reported in figure 7 Clear peak responses correspondingto the sensitive regions of the pixels can be observed On the other hand a broader response hasbeen obtained with front-side illumination the effects of the metal layer tend to spread out the laserstimuli Moreover the shielding effect of octagonal bondpoints can be observed as well in figure 8the superimposition of the actual layout with the response of a 3times3 pixel sub-set is reported

32 Signal to noise evaluation X-rays analyses

A more quantitative characterization has been carried out by looking at the single pixel analogueresponse and the digitized pixel response The responses of the small and the large test pixelshave been compared with the responses of the corresponding pixels located within the matrices(figures 9 and 10) thus allowing the estimation of the conversion factor between ADC counts and

ndash 3 ndash

2012 JINST 7 C08008

Figure 4 Coincidence responsesto a IR laser of the 16times16 outer(top) and inner (bottom) matrices

Figure 5 Differences betweenpeak response coordinates of thetop and bottom matrices

Figure 6 Computed tomography(CT) of the stacked die

Figure 7 Back-side illumination no metal-shielding effects regular pattern response to a X-Ylaser scan

Figure 8 Front-side illumination effects ofmetal-layers and octagonal bondpoints on the re-sponse to a X-Y laser scan

millivolts Actually voltage drops of about 180 mV and 160 mV have been obtained for the largeand small pixel layouts respectively This in turn leads to a conversion factor of 043 mV per ADCand 030 mV per ADC for the large and small pixel layouts

A comprehensive noise analysis has been carried out as well by considering the overall con-tribution of temporal and spatial noise Very similar noise figures have been obtained for cor-responding pixels of outer (top) and inner (bottom) tiers Small pixel architecture resulted in ahigher noise as expected due to the kTC reset noise contribution which is significantly greaterfor the small pixel (due to the lower sensitive region capacitance) with respect to the large pixel(figures 11 and 12)

ndash 4 ndash

2012 JINST 7 C08008

Figure 9 Dark response (discharge) of a large lay-out pixel - analog vs digital read-out comparison

Figure 10 Dark response (discharge) of a small lay-out pixel - analog vs digital read-out comparison

Figure 11 Overall (spatial and temporal) large pixelnoise contributions outer (top) and inner (bottom)tier

Figure 12 Overall (spatial and temporal) smallpixel noise contributions outer (top) and inner (bot-tom) tier

An Amptex Mini-X X-ray source has been used for sensor calibration purposes with photonsof different energies In particular Fe and Cu targets have been used in fluorescence mode A goodlinearity of the sensor response has been found at different photon energies from the expectedenergy peak deposition of Fe and Cu targets (64 keV and 81 keV respectively) the number of ehpairs generated within the Si substrate can be estimated The peak positions on the cluster signaldistributions of figures 13 and 14 correspond to almost complete charge collection therefore sen-sor calibration can be carried out allowing an estimation of the conversion gain of around 22 ehpairs per ADC for the outer tier and 25 eh pairs per ADC for the inner tier (figures 15 and 16)Eventually after calibration the overall (spatial and temporal) measured noise was around 41eThis suggests a reasonable SN for a Minimum Ionizing Particle equivalent signal assuming aslower limit (imposed by the upper thinned tier) an equivalent collection depth of less than ten

ndash 5 ndash

2012 JINST 7 C08008

Figure 13 16times16 small pixel matrices X-rays re-sponses (Fe target) cluster signal distributions forouter (top) and inner (bottom) tier

Figure 14 16times16 small pixel matrices X-rays re-sponses (Cu target) cluster signal distributions forouter (top) and inner (bottom) tier

Figure 15 16times16 small pixel matrices calibrationresponses (ADC) to different X-ray photons (outertier)

Figure 16 16times16 small pixel matrices calibrationresponses (ADC) to different X-ray photons (innertier)

microm and therefore an equivalent signal of about 500 eh pairs This assuming a complete chargecollection within such small volumes leads to a SN of about 12 for both inner and outer small pho-todiodes matrices From this point of view at least for X-ray photons no particular worries aboutsignal coming from top (thinned) tier with respect to signal coming from bottom tier have beenexperienced On the other hand a small layout pixel behaves better than a large layout pixel evenif small sensitive area features slightly worse noise the smaller capacitance allows significantlybetter charge-to-voltage conversion gain

ndash 6 ndash

2012 JINST 7 C08008

4 Conclusions

A first functional characterization of 3D monolithically stacked Active Pixel Sensors layers fabri-cated in Chartered Tezzaron 130nm 3D technology for particle tracking purposes has been carriedout Good electrical contacts between bottom and top tiers have been verified provided that con-tacts are located only at the array periphery (eg between pads) and adopting redundant bondpointsscheme Both tiers are fully functional different test structures and matrix structures (5times5 16times16small vs large pixel diode) have been characterized Clear coincidence responses between bottomand top matrices have been obtained with laser stimuli an X-Y misalignment between top and bot-tom tiers between 10 microm and 20 microm (depending on the direction) has been estimated Noise anal-ysis and X-rays calibrations with Fe and Cu fluorescence have been accomplished encouragingresults have been found in terms of SN ratio fostering the suitability of the adopted 3D-IC verticalscale fabrication technology and of the proposed approach for particle tracking applications

References

[1] D Passeri L Servoli and S Meroli Analysis of 3D stacked fully functional CMOS Active Pixel Sensordetectors 2009 JINST 4 P04009

[2] 3DIC Consortium http3dicfnalgov

[3] P Garrou C Bower and P Ramm Handbook of 3D Integration Wiley-VCH (2008)

[4] D Passeri A Marras P Placidi M Petasecca and L Servoli A laser test system for characterizingCMOS active pixel sensors Nucl Instrum Meth A 565 (2006) 144

ndash 7 ndash

2012 JINST 7 C08008

Figure 1 Schematic cross-section of the two tiers (sketch not to scale) Figure 2 Large (top) andsmall (bottom) pixel layouts

standard three transistors (3T) active pixel architecture featuring 10 micromtimes10 microm size with differentsensitive element (photodiode) layout In particular a small sensitive area with small capacitance(aiming at maximizing the charge to voltage conversion gain) and a large sensitive area (aiming atmaximizing the fill-factor of the detector) (figure 2) The structures of both tiers can be read out inparallel by means of dedicated (separated) output bond pads located at the backside of the top tier

3 Electrical and functional characterization

31 The optical workbench

For electrical and functional characterization a suitable read-out set-up has been devised and fab-ricated In particular an advanced optical workbench with IR UV and VISible laser heads withmicro-focusing (spot size below 2 microm) and micro-positioning (scan step 021 microm) capabilities hasbeen used (figure 3a) It allows up to four sensors parallel read-out for track reconstruction andspatial resolution analysis as well as 2D scans for surface mapping (figure 3b) [4]

Coincidence responses to a focused IR spot of corresponding outer and inner matrices havebeen obtained The coincidence responses can be clearly observed by translating the spot alongthe matrix (ie along the chip surface) and looking at the different responding clusters of pixels(figure 4) The focussing capabilities allow the stimulation of a single pixel on the top back-sideilluminated matrix while a wider and spread-out response has been obtained from the bottommatrix due to the scattering effect of the metal layers between the two chips and to the shieldingeffects of the bond-point metal-6 octagons whose dimensions (34 micromtimes34 microm) are of the sameorder of the pixel size Misalignments of about 12 microm in one direction and of about 20 microm on theorthogonal one have been demonstrated (figure 5) These findings have been confirmed by deeperinvestigations carried out by means of computed tomography (CT) scans at DESY Hamburg In

ndash 2 ndash

2012 JINST 7 C08008

Figure 3 (a) The optical workbench visible IR and UV wavelengths can beused 5 translational and 1 rotational stages allow very precise spot focusingand pointing

Figure 3 (b) Detail of the laserobjective stimulating the 3D chiphosted in a dedicated PCB (RAPS04daughter)

particular figure 6 shows a detail of the pad regions From the dimensions of the pads (120 microm)the estimated misalignment which is of the order of 10 of the pad size is about 15 microm ingood agreement with the previous findings It should be emphasized however that the significantmisalignment between the two tiers does not prevent the communication from the bottom tier tothe top tier The implemented circuits actually features TSV interconnections only at the chipperiphery (eg at the pad level) The huge number of TSVs in these regions even if the two tiersare tiltedshifted still guarantees the electrical connections between the pads

Surface scans can be carried out as well in order to evaluate the response of the pixel as afunction of the spot positions The micro-focusing and micro-positioning capabilities allow verydeep investigation of the point spread response of the matrix The response to a back-side illu-mination with visible light (531 nm) is reported in figure 7 Clear peak responses correspondingto the sensitive regions of the pixels can be observed On the other hand a broader response hasbeen obtained with front-side illumination the effects of the metal layer tend to spread out the laserstimuli Moreover the shielding effect of octagonal bondpoints can be observed as well in figure 8the superimposition of the actual layout with the response of a 3times3 pixel sub-set is reported

32 Signal to noise evaluation X-rays analyses

A more quantitative characterization has been carried out by looking at the single pixel analogueresponse and the digitized pixel response The responses of the small and the large test pixelshave been compared with the responses of the corresponding pixels located within the matrices(figures 9 and 10) thus allowing the estimation of the conversion factor between ADC counts and

ndash 3 ndash

2012 JINST 7 C08008

Figure 4 Coincidence responsesto a IR laser of the 16times16 outer(top) and inner (bottom) matrices

Figure 5 Differences betweenpeak response coordinates of thetop and bottom matrices

Figure 6 Computed tomography(CT) of the stacked die

Figure 7 Back-side illumination no metal-shielding effects regular pattern response to a X-Ylaser scan

Figure 8 Front-side illumination effects ofmetal-layers and octagonal bondpoints on the re-sponse to a X-Y laser scan

millivolts Actually voltage drops of about 180 mV and 160 mV have been obtained for the largeand small pixel layouts respectively This in turn leads to a conversion factor of 043 mV per ADCand 030 mV per ADC for the large and small pixel layouts

A comprehensive noise analysis has been carried out as well by considering the overall con-tribution of temporal and spatial noise Very similar noise figures have been obtained for cor-responding pixels of outer (top) and inner (bottom) tiers Small pixel architecture resulted in ahigher noise as expected due to the kTC reset noise contribution which is significantly greaterfor the small pixel (due to the lower sensitive region capacitance) with respect to the large pixel(figures 11 and 12)

ndash 4 ndash

2012 JINST 7 C08008

Figure 9 Dark response (discharge) of a large lay-out pixel - analog vs digital read-out comparison

Figure 10 Dark response (discharge) of a small lay-out pixel - analog vs digital read-out comparison

Figure 11 Overall (spatial and temporal) large pixelnoise contributions outer (top) and inner (bottom)tier

Figure 12 Overall (spatial and temporal) smallpixel noise contributions outer (top) and inner (bot-tom) tier

An Amptex Mini-X X-ray source has been used for sensor calibration purposes with photonsof different energies In particular Fe and Cu targets have been used in fluorescence mode A goodlinearity of the sensor response has been found at different photon energies from the expectedenergy peak deposition of Fe and Cu targets (64 keV and 81 keV respectively) the number of ehpairs generated within the Si substrate can be estimated The peak positions on the cluster signaldistributions of figures 13 and 14 correspond to almost complete charge collection therefore sen-sor calibration can be carried out allowing an estimation of the conversion gain of around 22 ehpairs per ADC for the outer tier and 25 eh pairs per ADC for the inner tier (figures 15 and 16)Eventually after calibration the overall (spatial and temporal) measured noise was around 41eThis suggests a reasonable SN for a Minimum Ionizing Particle equivalent signal assuming aslower limit (imposed by the upper thinned tier) an equivalent collection depth of less than ten

ndash 5 ndash

2012 JINST 7 C08008

Figure 13 16times16 small pixel matrices X-rays re-sponses (Fe target) cluster signal distributions forouter (top) and inner (bottom) tier

Figure 14 16times16 small pixel matrices X-rays re-sponses (Cu target) cluster signal distributions forouter (top) and inner (bottom) tier

Figure 15 16times16 small pixel matrices calibrationresponses (ADC) to different X-ray photons (outertier)

Figure 16 16times16 small pixel matrices calibrationresponses (ADC) to different X-ray photons (innertier)

microm and therefore an equivalent signal of about 500 eh pairs This assuming a complete chargecollection within such small volumes leads to a SN of about 12 for both inner and outer small pho-todiodes matrices From this point of view at least for X-ray photons no particular worries aboutsignal coming from top (thinned) tier with respect to signal coming from bottom tier have beenexperienced On the other hand a small layout pixel behaves better than a large layout pixel evenif small sensitive area features slightly worse noise the smaller capacitance allows significantlybetter charge-to-voltage conversion gain

ndash 6 ndash

2012 JINST 7 C08008

4 Conclusions

A first functional characterization of 3D monolithically stacked Active Pixel Sensors layers fabri-cated in Chartered Tezzaron 130nm 3D technology for particle tracking purposes has been carriedout Good electrical contacts between bottom and top tiers have been verified provided that con-tacts are located only at the array periphery (eg between pads) and adopting redundant bondpointsscheme Both tiers are fully functional different test structures and matrix structures (5times5 16times16small vs large pixel diode) have been characterized Clear coincidence responses between bottomand top matrices have been obtained with laser stimuli an X-Y misalignment between top and bot-tom tiers between 10 microm and 20 microm (depending on the direction) has been estimated Noise anal-ysis and X-rays calibrations with Fe and Cu fluorescence have been accomplished encouragingresults have been found in terms of SN ratio fostering the suitability of the adopted 3D-IC verticalscale fabrication technology and of the proposed approach for particle tracking applications

References

[1] D Passeri L Servoli and S Meroli Analysis of 3D stacked fully functional CMOS Active Pixel Sensordetectors 2009 JINST 4 P04009

[2] 3DIC Consortium http3dicfnalgov

[3] P Garrou C Bower and P Ramm Handbook of 3D Integration Wiley-VCH (2008)

[4] D Passeri A Marras P Placidi M Petasecca and L Servoli A laser test system for characterizingCMOS active pixel sensors Nucl Instrum Meth A 565 (2006) 144

ndash 7 ndash

2012 JINST 7 C08008

Figure 3 (a) The optical workbench visible IR and UV wavelengths can beused 5 translational and 1 rotational stages allow very precise spot focusingand pointing

Figure 3 (b) Detail of the laserobjective stimulating the 3D chiphosted in a dedicated PCB (RAPS04daughter)

particular figure 6 shows a detail of the pad regions From the dimensions of the pads (120 microm)the estimated misalignment which is of the order of 10 of the pad size is about 15 microm ingood agreement with the previous findings It should be emphasized however that the significantmisalignment between the two tiers does not prevent the communication from the bottom tier tothe top tier The implemented circuits actually features TSV interconnections only at the chipperiphery (eg at the pad level) The huge number of TSVs in these regions even if the two tiersare tiltedshifted still guarantees the electrical connections between the pads

Surface scans can be carried out as well in order to evaluate the response of the pixel as afunction of the spot positions The micro-focusing and micro-positioning capabilities allow verydeep investigation of the point spread response of the matrix The response to a back-side illu-mination with visible light (531 nm) is reported in figure 7 Clear peak responses correspondingto the sensitive regions of the pixels can be observed On the other hand a broader response hasbeen obtained with front-side illumination the effects of the metal layer tend to spread out the laserstimuli Moreover the shielding effect of octagonal bondpoints can be observed as well in figure 8the superimposition of the actual layout with the response of a 3times3 pixel sub-set is reported

32 Signal to noise evaluation X-rays analyses

A more quantitative characterization has been carried out by looking at the single pixel analogueresponse and the digitized pixel response The responses of the small and the large test pixelshave been compared with the responses of the corresponding pixels located within the matrices(figures 9 and 10) thus allowing the estimation of the conversion factor between ADC counts and

ndash 3 ndash

2012 JINST 7 C08008

Figure 4 Coincidence responsesto a IR laser of the 16times16 outer(top) and inner (bottom) matrices

Figure 5 Differences betweenpeak response coordinates of thetop and bottom matrices

Figure 6 Computed tomography(CT) of the stacked die

Figure 7 Back-side illumination no metal-shielding effects regular pattern response to a X-Ylaser scan

Figure 8 Front-side illumination effects ofmetal-layers and octagonal bondpoints on the re-sponse to a X-Y laser scan

millivolts Actually voltage drops of about 180 mV and 160 mV have been obtained for the largeand small pixel layouts respectively This in turn leads to a conversion factor of 043 mV per ADCand 030 mV per ADC for the large and small pixel layouts

A comprehensive noise analysis has been carried out as well by considering the overall con-tribution of temporal and spatial noise Very similar noise figures have been obtained for cor-responding pixels of outer (top) and inner (bottom) tiers Small pixel architecture resulted in ahigher noise as expected due to the kTC reset noise contribution which is significantly greaterfor the small pixel (due to the lower sensitive region capacitance) with respect to the large pixel(figures 11 and 12)

ndash 4 ndash

2012 JINST 7 C08008

Figure 9 Dark response (discharge) of a large lay-out pixel - analog vs digital read-out comparison

Figure 10 Dark response (discharge) of a small lay-out pixel - analog vs digital read-out comparison

Figure 11 Overall (spatial and temporal) large pixelnoise contributions outer (top) and inner (bottom)tier

Figure 12 Overall (spatial and temporal) smallpixel noise contributions outer (top) and inner (bot-tom) tier

An Amptex Mini-X X-ray source has been used for sensor calibration purposes with photonsof different energies In particular Fe and Cu targets have been used in fluorescence mode A goodlinearity of the sensor response has been found at different photon energies from the expectedenergy peak deposition of Fe and Cu targets (64 keV and 81 keV respectively) the number of ehpairs generated within the Si substrate can be estimated The peak positions on the cluster signaldistributions of figures 13 and 14 correspond to almost complete charge collection therefore sen-sor calibration can be carried out allowing an estimation of the conversion gain of around 22 ehpairs per ADC for the outer tier and 25 eh pairs per ADC for the inner tier (figures 15 and 16)Eventually after calibration the overall (spatial and temporal) measured noise was around 41eThis suggests a reasonable SN for a Minimum Ionizing Particle equivalent signal assuming aslower limit (imposed by the upper thinned tier) an equivalent collection depth of less than ten

ndash 5 ndash

2012 JINST 7 C08008

Figure 13 16times16 small pixel matrices X-rays re-sponses (Fe target) cluster signal distributions forouter (top) and inner (bottom) tier

Figure 14 16times16 small pixel matrices X-rays re-sponses (Cu target) cluster signal distributions forouter (top) and inner (bottom) tier

Figure 15 16times16 small pixel matrices calibrationresponses (ADC) to different X-ray photons (outertier)

Figure 16 16times16 small pixel matrices calibrationresponses (ADC) to different X-ray photons (innertier)

microm and therefore an equivalent signal of about 500 eh pairs This assuming a complete chargecollection within such small volumes leads to a SN of about 12 for both inner and outer small pho-todiodes matrices From this point of view at least for X-ray photons no particular worries aboutsignal coming from top (thinned) tier with respect to signal coming from bottom tier have beenexperienced On the other hand a small layout pixel behaves better than a large layout pixel evenif small sensitive area features slightly worse noise the smaller capacitance allows significantlybetter charge-to-voltage conversion gain

ndash 6 ndash

2012 JINST 7 C08008

4 Conclusions

A first functional characterization of 3D monolithically stacked Active Pixel Sensors layers fabri-cated in Chartered Tezzaron 130nm 3D technology for particle tracking purposes has been carriedout Good electrical contacts between bottom and top tiers have been verified provided that con-tacts are located only at the array periphery (eg between pads) and adopting redundant bondpointsscheme Both tiers are fully functional different test structures and matrix structures (5times5 16times16small vs large pixel diode) have been characterized Clear coincidence responses between bottomand top matrices have been obtained with laser stimuli an X-Y misalignment between top and bot-tom tiers between 10 microm and 20 microm (depending on the direction) has been estimated Noise anal-ysis and X-rays calibrations with Fe and Cu fluorescence have been accomplished encouragingresults have been found in terms of SN ratio fostering the suitability of the adopted 3D-IC verticalscale fabrication technology and of the proposed approach for particle tracking applications

References

[1] D Passeri L Servoli and S Meroli Analysis of 3D stacked fully functional CMOS Active Pixel Sensordetectors 2009 JINST 4 P04009

[2] 3DIC Consortium http3dicfnalgov

[3] P Garrou C Bower and P Ramm Handbook of 3D Integration Wiley-VCH (2008)

[4] D Passeri A Marras P Placidi M Petasecca and L Servoli A laser test system for characterizingCMOS active pixel sensors Nucl Instrum Meth A 565 (2006) 144

ndash 7 ndash

2012 JINST 7 C08008

Figure 4 Coincidence responsesto a IR laser of the 16times16 outer(top) and inner (bottom) matrices

Figure 5 Differences betweenpeak response coordinates of thetop and bottom matrices

Figure 6 Computed tomography(CT) of the stacked die

Figure 7 Back-side illumination no metal-shielding effects regular pattern response to a X-Ylaser scan

Figure 8 Front-side illumination effects ofmetal-layers and octagonal bondpoints on the re-sponse to a X-Y laser scan

millivolts Actually voltage drops of about 180 mV and 160 mV have been obtained for the largeand small pixel layouts respectively This in turn leads to a conversion factor of 043 mV per ADCand 030 mV per ADC for the large and small pixel layouts

A comprehensive noise analysis has been carried out as well by considering the overall con-tribution of temporal and spatial noise Very similar noise figures have been obtained for cor-responding pixels of outer (top) and inner (bottom) tiers Small pixel architecture resulted in ahigher noise as expected due to the kTC reset noise contribution which is significantly greaterfor the small pixel (due to the lower sensitive region capacitance) with respect to the large pixel(figures 11 and 12)

ndash 4 ndash

2012 JINST 7 C08008

Figure 9 Dark response (discharge) of a large lay-out pixel - analog vs digital read-out comparison

Figure 10 Dark response (discharge) of a small lay-out pixel - analog vs digital read-out comparison

Figure 11 Overall (spatial and temporal) large pixelnoise contributions outer (top) and inner (bottom)tier

Figure 12 Overall (spatial and temporal) smallpixel noise contributions outer (top) and inner (bot-tom) tier

An Amptex Mini-X X-ray source has been used for sensor calibration purposes with photonsof different energies In particular Fe and Cu targets have been used in fluorescence mode A goodlinearity of the sensor response has been found at different photon energies from the expectedenergy peak deposition of Fe and Cu targets (64 keV and 81 keV respectively) the number of ehpairs generated within the Si substrate can be estimated The peak positions on the cluster signaldistributions of figures 13 and 14 correspond to almost complete charge collection therefore sen-sor calibration can be carried out allowing an estimation of the conversion gain of around 22 ehpairs per ADC for the outer tier and 25 eh pairs per ADC for the inner tier (figures 15 and 16)Eventually after calibration the overall (spatial and temporal) measured noise was around 41eThis suggests a reasonable SN for a Minimum Ionizing Particle equivalent signal assuming aslower limit (imposed by the upper thinned tier) an equivalent collection depth of less than ten

ndash 5 ndash

2012 JINST 7 C08008

Figure 13 16times16 small pixel matrices X-rays re-sponses (Fe target) cluster signal distributions forouter (top) and inner (bottom) tier

Figure 14 16times16 small pixel matrices X-rays re-sponses (Cu target) cluster signal distributions forouter (top) and inner (bottom) tier

Figure 15 16times16 small pixel matrices calibrationresponses (ADC) to different X-ray photons (outertier)

Figure 16 16times16 small pixel matrices calibrationresponses (ADC) to different X-ray photons (innertier)

microm and therefore an equivalent signal of about 500 eh pairs This assuming a complete chargecollection within such small volumes leads to a SN of about 12 for both inner and outer small pho-todiodes matrices From this point of view at least for X-ray photons no particular worries aboutsignal coming from top (thinned) tier with respect to signal coming from bottom tier have beenexperienced On the other hand a small layout pixel behaves better than a large layout pixel evenif small sensitive area features slightly worse noise the smaller capacitance allows significantlybetter charge-to-voltage conversion gain

ndash 6 ndash

2012 JINST 7 C08008

4 Conclusions

A first functional characterization of 3D monolithically stacked Active Pixel Sensors layers fabri-cated in Chartered Tezzaron 130nm 3D technology for particle tracking purposes has been carriedout Good electrical contacts between bottom and top tiers have been verified provided that con-tacts are located only at the array periphery (eg between pads) and adopting redundant bondpointsscheme Both tiers are fully functional different test structures and matrix structures (5times5 16times16small vs large pixel diode) have been characterized Clear coincidence responses between bottomand top matrices have been obtained with laser stimuli an X-Y misalignment between top and bot-tom tiers between 10 microm and 20 microm (depending on the direction) has been estimated Noise anal-ysis and X-rays calibrations with Fe and Cu fluorescence have been accomplished encouragingresults have been found in terms of SN ratio fostering the suitability of the adopted 3D-IC verticalscale fabrication technology and of the proposed approach for particle tracking applications

References

[1] D Passeri L Servoli and S Meroli Analysis of 3D stacked fully functional CMOS Active Pixel Sensordetectors 2009 JINST 4 P04009

[2] 3DIC Consortium http3dicfnalgov

[3] P Garrou C Bower and P Ramm Handbook of 3D Integration Wiley-VCH (2008)

[4] D Passeri A Marras P Placidi M Petasecca and L Servoli A laser test system for characterizingCMOS active pixel sensors Nucl Instrum Meth A 565 (2006) 144

ndash 7 ndash

2012 JINST 7 C08008

Figure 9 Dark response (discharge) of a large lay-out pixel - analog vs digital read-out comparison

Figure 10 Dark response (discharge) of a small lay-out pixel - analog vs digital read-out comparison

Figure 11 Overall (spatial and temporal) large pixelnoise contributions outer (top) and inner (bottom)tier

Figure 12 Overall (spatial and temporal) smallpixel noise contributions outer (top) and inner (bot-tom) tier

An Amptex Mini-X X-ray source has been used for sensor calibration purposes with photonsof different energies In particular Fe and Cu targets have been used in fluorescence mode A goodlinearity of the sensor response has been found at different photon energies from the expectedenergy peak deposition of Fe and Cu targets (64 keV and 81 keV respectively) the number of ehpairs generated within the Si substrate can be estimated The peak positions on the cluster signaldistributions of figures 13 and 14 correspond to almost complete charge collection therefore sen-sor calibration can be carried out allowing an estimation of the conversion gain of around 22 ehpairs per ADC for the outer tier and 25 eh pairs per ADC for the inner tier (figures 15 and 16)Eventually after calibration the overall (spatial and temporal) measured noise was around 41eThis suggests a reasonable SN for a Minimum Ionizing Particle equivalent signal assuming aslower limit (imposed by the upper thinned tier) an equivalent collection depth of less than ten

ndash 5 ndash

2012 JINST 7 C08008

Figure 13 16times16 small pixel matrices X-rays re-sponses (Fe target) cluster signal distributions forouter (top) and inner (bottom) tier

Figure 14 16times16 small pixel matrices X-rays re-sponses (Cu target) cluster signal distributions forouter (top) and inner (bottom) tier

Figure 15 16times16 small pixel matrices calibrationresponses (ADC) to different X-ray photons (outertier)

Figure 16 16times16 small pixel matrices calibrationresponses (ADC) to different X-ray photons (innertier)

microm and therefore an equivalent signal of about 500 eh pairs This assuming a complete chargecollection within such small volumes leads to a SN of about 12 for both inner and outer small pho-todiodes matrices From this point of view at least for X-ray photons no particular worries aboutsignal coming from top (thinned) tier with respect to signal coming from bottom tier have beenexperienced On the other hand a small layout pixel behaves better than a large layout pixel evenif small sensitive area features slightly worse noise the smaller capacitance allows significantlybetter charge-to-voltage conversion gain

ndash 6 ndash

2012 JINST 7 C08008

4 Conclusions

A first functional characterization of 3D monolithically stacked Active Pixel Sensors layers fabri-cated in Chartered Tezzaron 130nm 3D technology for particle tracking purposes has been carriedout Good electrical contacts between bottom and top tiers have been verified provided that con-tacts are located only at the array periphery (eg between pads) and adopting redundant bondpointsscheme Both tiers are fully functional different test structures and matrix structures (5times5 16times16small vs large pixel diode) have been characterized Clear coincidence responses between bottomand top matrices have been obtained with laser stimuli an X-Y misalignment between top and bot-tom tiers between 10 microm and 20 microm (depending on the direction) has been estimated Noise anal-ysis and X-rays calibrations with Fe and Cu fluorescence have been accomplished encouragingresults have been found in terms of SN ratio fostering the suitability of the adopted 3D-IC verticalscale fabrication technology and of the proposed approach for particle tracking applications

References

[1] D Passeri L Servoli and S Meroli Analysis of 3D stacked fully functional CMOS Active Pixel Sensordetectors 2009 JINST 4 P04009

[2] 3DIC Consortium http3dicfnalgov

[3] P Garrou C Bower and P Ramm Handbook of 3D Integration Wiley-VCH (2008)

[4] D Passeri A Marras P Placidi M Petasecca and L Servoli A laser test system for characterizingCMOS active pixel sensors Nucl Instrum Meth A 565 (2006) 144

ndash 7 ndash

2012 JINST 7 C08008

Figure 13 16times16 small pixel matrices X-rays re-sponses (Fe target) cluster signal distributions forouter (top) and inner (bottom) tier

Figure 14 16times16 small pixel matrices X-rays re-sponses (Cu target) cluster signal distributions forouter (top) and inner (bottom) tier

Figure 15 16times16 small pixel matrices calibrationresponses (ADC) to different X-ray photons (outertier)

Figure 16 16times16 small pixel matrices calibrationresponses (ADC) to different X-ray photons (innertier)

microm and therefore an equivalent signal of about 500 eh pairs This assuming a complete chargecollection within such small volumes leads to a SN of about 12 for both inner and outer small pho-todiodes matrices From this point of view at least for X-ray photons no particular worries aboutsignal coming from top (thinned) tier with respect to signal coming from bottom tier have beenexperienced On the other hand a small layout pixel behaves better than a large layout pixel evenif small sensitive area features slightly worse noise the smaller capacitance allows significantlybetter charge-to-voltage conversion gain

ndash 6 ndash

2012 JINST 7 C08008

4 Conclusions

A first functional characterization of 3D monolithically stacked Active Pixel Sensors layers fabri-cated in Chartered Tezzaron 130nm 3D technology for particle tracking purposes has been carriedout Good electrical contacts between bottom and top tiers have been verified provided that con-tacts are located only at the array periphery (eg between pads) and adopting redundant bondpointsscheme Both tiers are fully functional different test structures and matrix structures (5times5 16times16small vs large pixel diode) have been characterized Clear coincidence responses between bottomand top matrices have been obtained with laser stimuli an X-Y misalignment between top and bot-tom tiers between 10 microm and 20 microm (depending on the direction) has been estimated Noise anal-ysis and X-rays calibrations with Fe and Cu fluorescence have been accomplished encouragingresults have been found in terms of SN ratio fostering the suitability of the adopted 3D-IC verticalscale fabrication technology and of the proposed approach for particle tracking applications

References

[1] D Passeri L Servoli and S Meroli Analysis of 3D stacked fully functional CMOS Active Pixel Sensordetectors 2009 JINST 4 P04009

[2] 3DIC Consortium http3dicfnalgov

[3] P Garrou C Bower and P Ramm Handbook of 3D Integration Wiley-VCH (2008)

[4] D Passeri A Marras P Placidi M Petasecca and L Servoli A laser test system for characterizingCMOS active pixel sensors Nucl Instrum Meth A 565 (2006) 144

ndash 7 ndash

2012 JINST 7 C08008

4 Conclusions

A first functional characterization of 3D monolithically stacked Active Pixel Sensors layers fabri-cated in Chartered Tezzaron 130nm 3D technology for particle tracking purposes has been carriedout Good electrical contacts between bottom and top tiers have been verified provided that con-tacts are located only at the array periphery (eg between pads) and adopting redundant bondpointsscheme Both tiers are fully functional different test structures and matrix structures (5times5 16times16small vs large pixel diode) have been characterized Clear coincidence responses between bottomand top matrices have been obtained with laser stimuli an X-Y misalignment between top and bot-tom tiers between 10 microm and 20 microm (depending on the direction) has been estimated Noise anal-ysis and X-rays calibrations with Fe and Cu fluorescence have been accomplished encouragingresults have been found in terms of SN ratio fostering the suitability of the adopted 3D-IC verticalscale fabrication technology and of the proposed approach for particle tracking applications

References

[1] D Passeri L Servoli and S Meroli Analysis of 3D stacked fully functional CMOS Active Pixel Sensordetectors 2009 JINST 4 P04009

[2] 3DIC Consortium http3dicfnalgov

[3] P Garrou C Bower and P Ramm Handbook of 3D Integration Wiley-VCH (2008)

[4] D Passeri A Marras P Placidi M Petasecca and L Servoli A laser test system for characterizingCMOS active pixel sensors Nucl Instrum Meth A 565 (2006) 144

ndash 7 ndash