Howard MatisPixel A High Resolution Vertex Tracker for the STAR Experiment using Active Pixel Sensors and Recent work using APS Sensors F. Bieser, R. Gareus, L. Greiner, J. King, J. Levesque, H.S. Matis, M. Oldenburg, H.G. Ritter, F. Retiere, A. Rose, K. Schweda, A. Shabetai, E. Sichtermann, J.H. Thomas, H. Wieman, Lawrence Berkeley National Laboratory S. Kleinfelder, S. Li, University of California, Irvine H. Bichsel, University of Washington F. Bieser, R. Gareus, L. Greiner, J. King, J. Levesque, H.S. Matis, M. Oldenburg, H.G. Ritter, F. Retiere, A. Rose, K. Schweda, A. Shabetai, E. Sichtermann, J.H. Thomas, H. Wieman, Lawrence Berkeley National Laboratory S. Kleinfelder, S. Li, University of California, Irvine H. Bichsel, University of Washington

Howard MatisPixel physics motivation for a thin vertex detector  Study initial properties of a nuclear collision  u, d, s quarks gain mass become thermalized  Final state effects  Measures later/cooler times of the collision  d, b quarks produced at early time  Intrinsic mass  Measure of early collision  Study initial properties of a nuclear collision  u, d, s quarks gain mass become thermalized  Final state effects  Measures later/cooler times of the collision  d, b quarks produced at early time  Intrinsic mass  Measure of early collision deconfinement Phase and Chiral transitions u-, d-quarks and ‘bound-states’ gain mass PART I - DETCTOR  Need to measure particles above 0.5 GeV/c  High collision density - more than 2000 tracks  Measures secondary particles >100 µm from collision point  Need to measure particles above 0.5 GeV/c  High collision density - more than 2000 tracks  Measures secondary particles >100 µm from collision point

Howard MatisPixel detector requirements  Study D 0 measurement  Multiple scattering in beam pipe sets fundamental limits  “Dream” Detector  Thickness 240 µm Si equivalent  Position resolution 8 µm  Study D 0 measurement  Multiple scattering in beam pipe sets fundamental limits  “Dream” Detector  Thickness 240 µm Si equivalent  Position resolution 8 µm

Howard MatisPixel star micro vertex detector  Two layers  1.5 cm radius  4.5 cm radius  24 ladders  2 cm  20 cm each  < 0.3% X 0  ~ 100 Mega Pixels  Two layers  1.5 cm radius  4.5 cm radius  24 ladders  2 cm  20 cm each  < 0.3% X 0  ~ 100 Mega Pixels

Howard MatisPixel close-up view

Howard MatisPixel sensor Efficiency for min ionization98% Accidental rate< 100 /cm 2 Position resolution < 10  m Pixel dimension 30  m  30  m Detector chip active area 19.2 mm  19.2 mm Detector chip pixel array 640  640 Sensor under development at IReS First prototype made using 0.25 µm process by TSMC Second version in production using 0.35 µm by AMS

Howard MatisPixel ladder  10 thinned APS detectors  Top of a matching row of thinned readout chips  Three-layer aluminum Kapton cable  Silicon cable structure is bonded to a carbon composite v, closing the beam to make a rigid structure  Wire bonding to the cable  10 thinned APS detectors  Top of a matching row of thinned readout chips  Three-layer aluminum Kapton cable  Silicon cable structure is bonded to a carbon composite v, closing the beam to make a rigid structure  Wire bonding to the cable

Howard MatisPixel ladder 2 carrier candidates – X 0 =0.11 % Top layer = 50 µm CFC Middle layer = 3.2 mm RVC Bottom layer = 50 µm CFC Outer shell = 100 µm CFC (carbon fiber composite) Fill = RVC (reticulated vitreous carbon foam)

Howard MatisPixel ladder prototypes  Mechanical Prototype with 4 MIMOSA-5 detectors glued to the Kapton cable assembly. Tested for  Vibration  Stiffness A prototype cable (Cu) has been designed, constructed and tested. Prototype ladder using thinned 50 µm MIMOSA-5 detectors. Currently under test with DAQ  Mechanical Prototype with 4 MIMOSA-5 detectors glued to the Kapton cable assembly. Tested for  Vibration  Stiffness A prototype cable (Cu) has been designed, constructed and tested. Prototype ladder using thinned 50 µm MIMOSA-5 detectors. Currently under test with DAQ

Howard MatisPixel heavy flavor tracker (hft) parameters Total number of pixels 98  10 6 Number of pixels per chip 640 x 640 Pixel Readout rate 100 ns Readout time per frame 4 ms Dynamic range of the ADC 10 bits Raw data from one sensor using a 10 bit ADC 1 Gb/s Fixed pattern noise 2000 e  Noise after Correlated Double Sampling 10 e  Maximum signal 900 e  Dynamic range after Correlated Double Sampling 8 bits Total power consumption 90 W

Howard MatisPixel mechanical requirements  Geometry  Maintain position resolution of ~ 10 µm  Low mass / radiation length (X 0 ~ 0.3% / layer)  Coverage of -1 <  < 1  Function Easy to calibrate Easy to align Easy to remove, repair and replace electronics (ladders will need to have a local survey) Fit easily into the existing detector and infrastructure at STAR  Geometry  Maintain position resolution of ~ 10 µm  Low mass / radiation length (X 0 ~ 0.3% / layer)  Coverage of -1 <  < 1  Function Easy to calibrate Easy to align Easy to remove, repair and replace electronics (ladders will need to have a local survey) Fit easily into the existing detector and infrastructure at STAR

Howard MatisPixel conceptual mechanical design Mounted to SVT cone Slides in and out on one end Ladders moves as beam pipe diameter increases

Howard MatisPixel kinematic support structure Support bolts unto STAR Green structure provides stable support for the ladder Three point kinematic mounts assure accurate positioning Can move detector in and out with reproducibility

Howard MatisPixel studies with scanning electron microscope 12 µm  Access to keV scanning electron microscope  Thought needed to punch through 2-3 µm  Believed could detect these electrons PART II - APS RESEARCH 30 keV electrons

Howard MatisPixel cross sectional view (Tilt at 52 0 ) Pt Layer Top of IC Artifact due to charge Epi-layer Top coating

Howard MatisPixel element analysis Al Ti W Pt Si O Ga

Howard MatisPixel kev electrons do not penetrate to the epilayer

Howard MatisPixel can detect “electrons” with reasonable accuracy  Can see microscope  Measuring Bremsstrahlung  Maximum intensity ~3000  /frame  Evaluate charge sharing of cell  Evaluate position resolution algorithms  Best  Can see microscope  Measuring Bremsstrahlung  Maximum intensity ~3000  /frame  Evaluate charge sharing of cell  Evaluate position resolution algorithms  Best µm

Howard MatisPixel track efficiency is critical with noise level  Monte Carlo study two different algorithms with MIMOSA 5  Look for seed pixels  Smooth data and then look for seed pixels  Real pedestal data with imbedded electron spectrum  Efficiency algorithm dependent  Algorithm choice dependent on noise  Monte Carlo study two different algorithms with MIMOSA 5  Look for seed pixels  Smooth data and then look for seed pixels  Real pedestal data with imbedded electron spectrum  Efficiency algorithm dependent  Algorithm choice dependent on noise

Howard MatisPixel how much signal do you get out of an aps sensor?  Calculations show that energy loss in thin materials much less than thicker  Bichsel & Saxon, Phys. Rev. A 11, 1286 (1975).  Observed in aluminum  Perez & Sevely, Phys. Rev. A 16, 1061 (1977).  Calculations show that energy loss in thin materials much less than thicker  Bichsel & Saxon, Phys. Rev. A 11, 1286 (1975).  Observed in aluminum  Perez & Sevely, Phys. Rev. A 16, 1061 (1977). Energy Deposited - eV Landau Bichsel & Saxon 0.76 µm Al 1 MeV e -

Howard MatisPixel study at lbnl advanced light source  Study 1.5 GeV/c electrons  Calculated expected energy  Use Bichsel formalism  0.25 µm TSMC  8 µm epitaxial layer  Need to shift theory by 1.5 for good agreement  Study 1.5 GeV/c electrons  Calculated expected energy  Use Bichsel formalism  0.25 µm TSMC  8 µm epitaxial layer  Need to shift theory by 1.5 for good agreement

Howard MatisPixel some checks  Epitaxial (epi) layer 8 µm (error perhaps 1 µm)  Use Bichsel formalism on 8.5 µm aluminum data  1.66 keV scales to 1.43 keV silicon (most probable)  Bichsel predicts  1.43 keV  Total systematic error %  Cannot explain 50% excess  Epitaxial (epi) layer 8 µm (error perhaps 1 µm)  Use Bichsel formalism on 8.5 µm aluminum data  1.66 keV scales to 1.43 keV silicon (most probable)  Bichsel predicts  1.43 keV  Total systematic error %  Cannot explain 50% excess epitaxial layer

Howard MatisPixel a hypothesis  Extra charge equivalent to 4 µm  Electrons could be coming from upper p- well and p ++ substrate  Check with Mimosa-5 data (AMS 0.6 µm)  Most Probable e -  Bichsel e -  Equivalent to extra 4.7 µm over nominal 14 µm  Extra charge equivalent to 4 µm  Electrons could be coming from upper p- well and p ++ substrate  Check with Mimosa-5 data (AMS 0.6 µm)  Most Probable e -  Bichsel e -  Equivalent to extra 4.7 µm over nominal 14 µm

Howard MatisPixel scaling of cell size  UCI design a multi-spacing chip  5 µm, 10 µm, 20 µm and 30 µm  All cell sizes on one chip  Minimize systematic errors  Charge sharing very similar  Can see small absorption of charge in epitaxial layer  Good scaling  UCI design a multi-spacing chip  5 µm, 10 µm, 20 µm and 30 µm  All cell sizes on one chip  Minimize systematic errors  Charge sharing very similar  Can see small absorption of charge in epitaxial layer  Good scaling Linear Scale Log Scale

Howard MatisPixel summary  Proposal for a vertex detector with APS technology  Awaiting funding  Transmission scanning microscopes can be used to probe sensors  Software algorithms important to get high hit reconstruction - choice very sensitive to absolute noise  Cell scales from 5 to 30 µm  More charge then expected coming from APS  Proposal for a vertex detector with APS technology  Awaiting funding  Transmission scanning microscopes can be used to probe sensors  Software algorithms important to get high hit reconstruction - choice very sensitive to absolute noise  Cell scales from 5 to 30 µm  More charge then expected coming from APS

Howard MatisPixel A Heavy Flavor Tracker for STAR Z. Xu Brookhaven National Laboratory Y. Chen, S. Kleinfelder, A. Koohi, S. Li University of California, Irvine H. Huang, A. Tai University of California, Los Angeles V. Kushpil, M. Sumbera Nuclear Physics Institute AS CR C. Colledani, W. Dulinski, A. Himmi, C. Hu, A. Shabetai, M. Szelezniak, I. Valin, M. Winter Institut de Recherches Subatomique, Strasbourg M. Miller, B. Surrow, G. Van Nieuwenhuizen Massachusetts Institute of Technology F. Bieser, R. Gareus, L. Greiner, F. Lesser, H.S. Matis, M. Oldenburg, H.G. Ritter, L. Pierpoint, F. Retiere, A. Rose, K. Schweda, E. Sichtermann, J.H. Thomas, H. Wieman, E. Yamamoto Lawrence Berkeley National Laboratory I. Kotov Ohio State University

Howard MatisPixel end

Howard MatisPixel backup slides

Howard MatisPixel precision tie points coupling the hft system to the star support cone

Howard MatisPixel thin beam pipe  Central beryllium region  14.5 mm radius  10   beam size  500 µm thick walls  Outer region  30 mm radius aluminum  Exoskeleton caries load  Central beryllium region  14.5 mm radius  10   beam size  500 µm thick walls  Outer region  30 mm radius aluminum  Exoskeleton caries load

Howard MatisPixel end view showing the hft ladders between spokes of the inner beam pipe support

Howard MatisPixel data flow and processing stages in the readout chip  Each stage can be bypassed to allow raw or partially unprocessed data to be routed to the DAQ  The first stage is a CDS preprocessor which is followed by pedestal subtraction and a pixel masking filter  Further processing allows us to sum up the value of 1, 4 or 9 pixels before a threshold cut is applied.  The last stage includes zero suppression and transcoding to hit positions.  Each stage can be bypassed to allow raw or partially unprocessed data to be routed to the DAQ  The first stage is a CDS preprocessor which is followed by pedestal subtraction and a pixel masking filter  Further processing allows us to sum up the value of 1, 4 or 9 pixels before a threshold cut is applied.  The last stage includes zero suppression and transcoding to hit positions.

Howard MatisPixel readout layout  sketch of the readout-topology on a detector ladder  one of ten APS and the corresponding readout chip layout.  sketch of the readout-topology on a detector ladder  one of ten APS and the corresponding readout chip layout.

Howard MatisPixel rdo asic ADC – 10 bit ADC for signals from sensor chip CDS – Chip will perform correlated double sampling High speed LVDS output Configuration, control, clock, synch functions ADC – 10 bit ADC for signals from sensor chip CDS – Chip will perform correlated double sampling High speed LVDS output Configuration, control, clock, synch functions Both chips thinned to 50 µm thickness. X 0 = % each

Howard MatisPixel daq Figure5: ladders can be combined to one optical link.

Howard MatisPixel hit loading Au+Au Luminosity 1  cm -2 s -1 dN/d  (min bias) 170 Min bias cross section 10 barns Interaction diamond size, σ 30 cm Outer LayerInner Layer Radius 5 cm 1.5 cm Hit Flux 4.3 kHz/cm 2 18 kHz/cm 2 Hit Density 4 ms Integration 17/cm 2 72/cm 2 Projected Tracking Window Area 0.6 mm mm 2 Probability of Tracking Window Pileup 10 % HFT Hit Resolving Area mm 2 Probability of HFT Pileup 0.14% 0.58%

Howard MatisPixel Comparison with mimosa-5 ParameterDetector MIMOSA-5 Detection efficiency  – 40 C ~ 99% ≤ 20  C resolution< 10 µm ~ 2 µm pixel pitch)30 µm 17 µm Read-out time4 – 10 ms 24 ms (  20 ms possible) Ionizing radiation tolerance 2.6 kRad/yr  100 kRad Fluence tolerance2  n eq /cm 2 ≤ n eq /cm 2 Power dissipation  100 mW/cm 2 ~ 10 mW/cm 2 Chip size ~2  2 cm  1.7 cm 2 Chip thickness  50  m 120  m

Howard MatisPixel material budget Material Material Thickness (µm of Si) % X 0 Beryllium beam pipe500 µm of Be MIMOSA detector Adhesive RDO chip Adhesive Cable assembly Adhesive Carbon fiber / RVC beam Total for the ladder components

Howard MatisPixel using an aps as a camera