Functionality Test of Bare Modules for the Phase I Upgrade of CMS - - PowerPoint PPT Presentation
Functionality Test of Bare Modules for the Phase I Upgrade of CMS Pixel Detector Presentation of Master Thesis, 23.10.2015 Bojan Hiti INSTITUT F UR EXPERIMENTELLE KERNPHYSIK (EKP), KIT www.kit.edu KIT University of the State of
INSTITUT F ¨ UR EXPERIMENTELLE KERNPHYSIK (EKP), KIT
Presentation of Master Thesis, 23.10.2015 Bojan Hiti
KIT – University of the State of Baden-Wuerttemberg and National Laboratory of the Helmholtz Association
www.kit.edu
Motivation for the Phase I Upgrade Structure of pixel modules and bump bonding technology Bare module production at KIT and motivation for the functionality test Bare module probe station setup Test routines on bare modules Evaluation of test reliability Production yields and conclusion
Bojan Hiti – Functionality Test of Bare Modules for the Phase I Upgrade of CMS Pixel Detector October 23, 2015 2/17
LHC Run 2:
13 TeV center of mass energy for pp 2 × 1034 cm−2s−1 instantaneous luminosity (doubled design luminosity)
Increased luminosity challenging for the CMS pixel detector
Particle rate up to 600 MHz cm−2 (Run 1: 115 MHz cm−2) Significant data loss with the current pixel detector due to buffer
For this reason a new pixel detector will be installed in CMS during the Phase I Upgrade (2016/2017) and will be used until Phase II
Bojan Hiti – Functionality Test of Bare Modules for the Phase I Upgrade of CMS Pixel Detector October 23, 2015 3/17
Detector changes in the Phase I Upgrade:
New readout chip (ROC) Additional layer 4 Layer 1 closer to the interaction point
→ reduced beam pipe diameter
Power supplies moved outside of the active volume
Institutes involved in Phase I: PSI (CH), CERN, INFN (I), DESY (D), KIT & RWTH Aachen (D) KIT will produce one half of the modules required for barrel layer 4
Bojan Hiti – Functionality Test of Bare Modules for the Phase I Upgrade of CMS Pixel Detector October 23, 2015 4/17
Smallest building block of the pixel detector 350 modules for Phase-I-Upgrade will be produced at KIT (4/day)
Bojan Hiti – Functionality Test of Bare Modules for the Phase I Upgrade of CMS Pixel Detector October 23, 2015 5/17
Smallest building block of the pixel detector 350 modules for Phase-I-Upgrade will be produced at KIT (4/day) Intermediate step in module assembly:
Bare module: Sensor and 16 ROCs
Bojan Hiti – Functionality Test of Bare Modules for the Phase I Upgrade of CMS Pixel Detector October 23, 2015 5/17
Hybrid detector design: sensor and ROCs manufactured on separate chips Electrical interconnections on pixel level required → bump-bonding (BB) Bump-bonding: vertical interconnections via solder balls (size ∼ 30 µm) 4160 solder bumps per ROC 66 560 bumps per module 99 % good BB connections per ROC for class A module required
Bare Module: REWORK of defect ROCs still possible Early discovery of bonding issues
Bojan Hiti – Functionality Test of Bare Modules for the Phase I Upgrade of CMS Pixel Detector October 23, 2015 6/17
Process taking place in a class ISO 10000 clean room
Bojan Hiti – Functionality Test of Bare Modules for the Phase I Upgrade of CMS Pixel Detector October 23, 2015 7/17
Process taking place in a class ISO 10000 clean room Process steps:
Cleaning of electrical components
Bojan Hiti – Functionality Test of Bare Modules for the Phase I Upgrade of CMS Pixel Detector October 23, 2015 7/17
Process taking place in a class ISO 10000 clean room Process steps:
Cleaning of electrical components Optical inspection of sensors and bumped ROCs
Bojan Hiti – Functionality Test of Bare Modules for the Phase I Upgrade of CMS Pixel Detector October 23, 2015 7/17
Process taking place in a class ISO 10000 clean room Process steps:
Cleaning of electrical components Optical inspection of sensors and bumped ROCs Flip-chip bonding
Bojan Hiti – Functionality Test of Bare Modules for the Phase I Upgrade of CMS Pixel Detector October 23, 2015 7/17
Process taking place in a class ISO 10000 clean room Process steps:
Cleaning of electrical components Optical inspection of sensors and bumped ROCs Flip-chip bonding Bare module test Eventual reworking
Bojan Hiti – Functionality Test of Bare Modules for the Phase I Upgrade of CMS Pixel Detector October 23, 2015 7/17
Process taking place in a class ISO 10000 clean room Process steps:
Cleaning of electrical components Optical inspection of sensors and bumped ROCs Flip-chip bonding Bare module test Eventual reworking Reflow soldering
Bojan Hiti – Functionality Test of Bare Modules for the Phase I Upgrade of CMS Pixel Detector October 23, 2015 7/17
Process taking place in a class ISO 10000 clean room Process steps:
Cleaning of electrical components Optical inspection of sensors and bumped ROCs Flip-chip bonding Bare module test Eventual reworking Reflow soldering
Bojan Hiti – Functionality Test of Bare Modules for the Phase I Upgrade of CMS Pixel Detector October 23, 2015 7/17
Electrical test of bare modules based on injection of internal calibration signals in the ROC During the test, a temporary electrical connection to wire-bond pads
Needle Card Custom readout board (DTB) – ROC powering and readout via needle card connection Additionally: −150 V bias voltage applied → bias needle
Bojan Hiti – Functionality Test of Bare Modules for the Phase I Upgrade of CMS Pixel Detector October 23, 2015 8/17
Bojan Hiti – Functionality Test of Bare Modules for the Phase I Upgrade of CMS Pixel Detector October 23, 2015 9/17
Probe station for the electrical test of bare modules built from scratch NI LabWindows based control software (GUI) Pixel eXpert Analysis Readout (pXar) – software for running electrical tests Driving to test positions and execution of electrical test automated Module positioning via pattern recognition Operator input required for the confirmation of steps and evaluation
Bojan Hiti – Functionality Test of Bare Modules for the Phase I Upgrade of CMS Pixel Detector October 23, 2015 10/17
Tests conducted on a bare module:
I-V curve – discover sensors with increased leakage current Pretest – set ROC currents and configure internal calibration signals Pixel alive – discover dead pixel unit cells Bump bond test - find missing bump bonds
Bare modules classified in three quality classes: A (best), B and C (worst), depending on the test outcome Compared to other production centers, KIT bumps have larger size
→ Direct application of the standard bump bond test not reliable → A custom bump bond test had to be developed → Another bump bond developed in parallel at DESY, used at KIT for
cross checking
Bojan Hiti – Functionality Test of Bare Modules for the Phase I Upgrade of CMS Pixel Detector October 23, 2015 11/17
Pixel Alive: check responsiveness of pixel unit cells
10 calibration signals injected directly into the preamplifier Pixels with no hits identified as dead
hits
N
1 2 3 4 5 6 7 8 9 10
PixelAlive
column 10 20 30 40 50 row 20 40 60 80
PixelAlive Bojan Hiti – Functionality Test of Bare Modules for the Phase I Upgrade of CMS Pixel Detector October 23, 2015 12/17
Pixel Alive: check responsiveness of pixel unit cells BB-test step 1: Set optimal amplifier threshold on ROC level
Run a noise scan (no signal injection) Vary threshold level and count noisy pixels Minimal threshold with less than 5 noisy pixels selected
vthrcomp
100 105 110 115 120 125 130
noisy pixels
10 0 noisy pixels selected threshold 8 noisy pixels 102 103 Low threshold
Bojan Hiti – Functionality Test of Bare Modules for the Phase I Upgrade of CMS Pixel Detector October 23, 2015 12/17
Pixel Alive: check responsiveness of pixel unit cells BB-test step 1: Set optimal amplifier threshold on ROC level BB-test step 2: 100 calibration signals injected via air gap capacitance into the sensor and collected through the bump-bond
Measure pixel efficiency – fraction of received triggers (cut 50 %)
Efficiency [%]
10 20 30 40 50 60 70 80 90 100 Efficiency Map CalDel = 150 column 10 20 30 40 50 row 20 40 60 80 Efficiency Map CalDel = 150
Bojan Hiti – Functionality Test of Bare Modules for the Phase I Upgrade of CMS Pixel Detector October 23, 2015 12/17
Also ROCs with a weak needle contact, which have to be retested
Insert Module + Alignment I-V curve Test 16 ROCs
Manually test ”bad” ROCs Fill database entries
5 min 5 min 16 × 4 min 5 min per ROC or more 5 min
Bojan Hiti – Functionality Test of Bare Modules for the Phase I Upgrade of CMS Pixel Detector October 23, 2015 13/17
X-ray test: independent verification of the bump bond test Signals generated by full absorption of X-ray photons Only possible on a fully built module (no more reworking)
Bojan Hiti – Functionality Test of Bare Modules for the Phase I Upgrade of CMS Pixel Detector October 23, 2015 14/17
32 full modules tested by X-ray Maximal number of missing bumps on a full module: 20 (Reminder: 41 missing bumps per ROC still allowed for grade A) Excellent bump bonding yield
defects
N 2 4 6 8 10 12 14
modules
N 1 2 3 4 5 6 7 8
X-ray test (cut 3): pixel defects X-ray test (cut 3): pixel defects
Bojan Hiti – Functionality Test of Bare Modules for the Phase I Upgrade of CMS Pixel Detector October 23, 2015 15/17
Bump bond test reliability estimated by comparison to the X-ray test:
Unidentified X-ray defects – appeared functional on the bare module Overestimated X-ray defects – falsely declared as defect
defects
N 1 2 3 4 5
modules
N 5 10 15 20 25 UndetectedIX-rayIdefects
DESYIBBItest KITIBBItest
=I0.06
DESY
b =I0.08
KIT
b
defects
N 2 4 6 8 10
modules
N 2 4 6 8 10 12 14 16 18
DESYrBBrtest KITrBBrtest
=r0.24
DESY
a =r0.42
KIT
a
Overestimatedrdefects
Bojan Hiti – Functionality Test of Bare Modules for the Phase I Upgrade of CMS Pixel Detector October 23, 2015 16/17
CMS pixel detector will be replaced with a new one at the end of 2016 Monitoring of bare modules quality important during the production process Bare module probe station was developed and can test up to four bare modules per day Bump bond test developed and works with a sufficient reliability By September 2015 48/350 production bare modules tested: 45 A, 1 B, 2 C, 10 had to be reworked
Bojan Hiti – Functionality Test of Bare Modules for the Phase I Upgrade of CMS Pixel Detector October 23, 2015 17/17
Bojan Hiti – Functionality Test of Bare Modules for the Phase I Upgrade of CMS Pixel Detector October 23, 2015 1/0
Bojan Hiti – Functionality Test of Bare Modules for the Phase I Upgrade of CMS Pixel Detector October 23, 2015 2/0
Pixel Unit Cell
VwllPr VwllSh comparator VthrComp trimming Vtrim sample & hold DC token pixel row address 9 bit hold delay VhldDel PH buffer 80× row address buffer 80×9 bit write read FIFO serializer sdout– sdout+ Double column address + – PHOffset PHscale Vcomp_ADC 8 bit 80 MHz SAR ADC
Controller & interface block Double column periphery
Trim bits reset bump pad VCal CalDel Vcomp Vsh sensor sensor calib calib VIColOr VIbias_bus Vana Vana top metal pad time stamp buffer 24×8 bit clock trigger WBC token in token
160 MHz PLL clock 23 bit 6 bit 4 bit programmable DAC Vdig CtrlReg
Bojan Hiti – Functionality Test of Bare Modules for the Phase I Upgrade of CMS Pixel Detector October 23, 2015 3/0
Module positioned with 4-axis positioning stage with precision 10 µm ROC0 and ROC7 positions obtained by pattern recognition, others
Bojan Hiti – Functionality Test of Bare Modules for the Phase I Upgrade of CMS Pixel Detector October 23, 2015 4/0
Bojan Hiti – Functionality Test of Bare Modules for the Phase I Upgrade of CMS Pixel Detector October 23, 2015 5/0
Time synchronization of the internal charge injection with the 25 ns clock (”Tornado plot”) 2-D scan of threshold/injection delay in a single pixel Optimal working point selected
noise threshold
CalDel 50 100 150 200 250 VthrComp 50 100 150 200 250
0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
50 DAC units working point VthrComp CalDel low threshold
time signal low thr Δt high thr
t1 t2
Bojan Hiti – Functionality Test of Bare Modules for the Phase I Upgrade of CMS Pixel Detector October 23, 2015 6/0
Charge injection pulses not simultaneously fired in entire ROC → Scan different delays to find the best value (CalDel Scan) Best efficiency at any CalDel used as criteria for each pixel
column 10 20 30 40 50 row 10 20 30 40 50 60 70
10 20 30 40 50 60 70 80 90 100
column 10 20 30 40 50 row 10 20 30 40 50 60 70
10 20 30 40 50 60 70 80 90 100
column 10 20 30 40 50 row 10 20 30 40 50 60 70
10 20 30 40 50 60 70 80 90 100 BBeffMapBest_C0_(V0) 100 100 100
Efficiency_map_CalDel_=_110 Efficiency_map_CalDel_=_150 Efficiency_map_CalDel_=_90
Bojan Hiti – Functionality Test of Bare Modules for the Phase I Upgrade of CMS Pixel Detector October 23, 2015 7/0
Bojan Hiti – Functionality Test of Bare Modules for the Phase I Upgrade of CMS Pixel Detector October 23, 2015 8/0
Based on measuring the threshold range of each pixel, for which it detects calibration signals Pixels with missing bumps have a smaller plateau width – discrimination possible
50 100 150 200 2 4 6 8 10
threshold too high
Ntrig
Single pixel plateau
plateau width saturation by noise
VthrComp 250 20 40 60 80 100 1 10
2
10 Plateau width distribution
3 missing bumps
plateau width cut 35
Npix
Bojan Hiti – Functionality Test of Bare Modules for the Phase I Upgrade of CMS Pixel Detector October 23, 2015 9/0
Bias voltage needs to be filtered to ensure low noise in the signal Low pass RC filter placed on the chuck
Bojan Hiti – Functionality Test of Bare Modules for the Phase I Upgrade of CMS Pixel Detector October 23, 2015 10/0
Bojan Hiti – Functionality Test of Bare Modules for the Phase I Upgrade of CMS Pixel Detector October 23, 2015 11/0
Full Module M4509 Missidentified BB: 24