Studies of Silicon Photomultipliers for the CMS HCAL Upgrade Yu. - PowerPoint PPT Presentation

Studies of Silicon Photomultipliers for the CMS HCAL Upgrade Yu. Musienko 1,2 , A. Heering 2 , A. Karneyev 1 , V. Postoev 1 R. Ruchti 2 , M. Wayne 2 1 INR RAS, Moscow 2 University of Notre Dame, Notre Dame 1

Outline • The CMS Hadron calorimeter (HCAL) at the LHC • Motivation for the photodetector upgrade • SiPM requirements for the CMS HCAL Upgrade • SiPM R&D goals and proposed solutions • HPK and KETEK SiPM performances (developed for the CMS HCAL project) • 175 preproduction array results • Summary for the preproduction array measurements • Conclusion 2

CMS Hadron Calorimeter (HCAL) HB, HE, HO similar technology: scintillator tiles with Y11 WLS fiber readout, brass (steel for HO) absorber. HPD was selected as the CMS HCAL photodetector. The CMS HCAL photodetector upgrade was proposed after several years of successful operation of the HPDs at the LHC. 3

Motivation for the HB/HE photo- detector upgrade 1. SiPMs have better quantum efficiency, higher gain, and better immunity to magnetic fields than HPDs. Since SiPMs operate at relatively low voltages, they do not produce large pulses from high voltage breakdown that mimic energetic showers like HPDs do. These features of the SiPMs together with their low cost and compact size compared to HPDs enable several major changes to the HCAL. 2. Implementation of depth segmentation which has advantages in coping with higher luminosities and compensating for radiation damage to the scintillators. This is made possible by the use of SiPMs. 3. Use of timing to clean up backgrounds, made possible by the extra gain and better signal-to- noise of the SiPMs. 4

Main CMS HCAL HB/HE SiPM requirements Area: ~Ø3 mm • PDE(515 nm): > 15% • Operating voltage: <90 V • Gain: <700 000 • ENF: <1.3 • Optical X-talk between cells: <20% • Temperature coefficient: <5%/ ° C • Dynamic range: > 20 000 “effective” cells/SiPM • Cell recovery time: <10 ns • Dark current (T=24 ° C, after 2*10 12 n/cm 2 ): <1000 µ A • Fractional Gain*PDE (after 2*10 12 n/cm 2 ): >65% • Neutron sensitivity: low • HO estimated neutron fluence: <10 11 n/cm 2 HE estimated neutron fluence: ~10 11 n/cm 2 HB estimated neutron fluence: ~7*10 11 n/cm 2 5

> 5 years of R&D to develop SiPMs for the CMS HCAL Phase I Upgrade Why it was difficult? High neutron fluences  high dark noise  large size cells (we need them for high PDE!!) are permanently fired  V-VB approaches “0”  significant drop of the SiPM PDE and gain  SiPM has low PDE, gain and it is useless as a photodetector for the calorimetry… To achieve the goal we performed an optimization of the SiPM structure: Small cell size (<15 µ m)  smaller dark noise generation rate (to avoid cell blocking effects); • Fast cell recovery (<10ns)  1/(dark count rate)<<cell recovery time  small PDE*Gain • losses Improve SiPM’s geometric factor  High PDE (>15%)  better S/N ratio after irradiation • “Thick” epitaxial layer and deep p-n junction  better PDE for green Y11 light  Small gain • (700 000)  less dark current after irradiation small “parasitic” (parallel to R q ) capacitance  smaller gain  smaller X-talk&afterpulsing  • smaller dark current and smaller noise after irradiation SiPM electric field engineering  smaller dark noise generation rate, faster noise reduction • with temperature Many different SiPM structures were developed during >5 years of R&D performed by the CMS SiPM group and commercial companies (CPTA, Zecoteck, Hamamatsu, KETEK, FBK …) Below we report the results achieved with the Hamamatsu and the KETEK SiPMs 6

Progress in PDE for the 15 µ m cell pitch HPK and KETEK SiPMs (2011-2014) Significant improvement of PDE for the HPK and KETEK developers during 2011-2014 R&D (most of the results were presented at NDIP-14 conference, see talks of A.Heering and Y. Musienko) In June 2015 Hamamtsu was selected by the CMS collaboration as a vendor for the HE HCAL after testing of 175 preproduction arrays 7

Preproduction Hamamatsu arrays Delivered as scheduled on April 1, 2015 175 eight-channel arrays – 1400 channels 70% of arrays with 2.8 mm devices, complement are 3.3 mm Entire order of 175 arrays in ceramic packages protected by 100 micron thick glass windows 8

IV curves for all 1400 SiPMs All 1400 channels operational IV curve – no light Dark current within spec A few channels show high dark current below operating voltage – may remove with additional spec at production stage 2.8 mm – 984 channels 3.3 mm – 416 channels IV curve – LED illumination Good uniformity in V b and at operating voltage V op = (V b + 3 volts) After calibration, spread at operating voltage will give us spread in PDE 9

VB measurements Spread in V b 123 arrays, 984 channels All 2.8 mm diameter devices Overall spread < 1.0 volt, RMS will be smaller Spread in V b 52 arrays, 416 channels All 3.3 mm diameter devices Overall spread < 1.0 volt, RMS will be smaller 10 1/IdI/dV method is used (see talk Y. Musienko et al., NDIP-2014)

Gain*PDE uniformity for 2.8 and 3.3 mm SiPMs 960 SiPMs (Ø2.8 mm) 320 SiPMs (Ø3.3 mm) 35 80 70 30 Mean= 1.00 Mean = 1.00 RMS = 1.13 % RMS = 0.96 % 60 25 50 Frequency Frequency 20 40 15 30 10 20 5 10 0 0 0.95 1 1.05 1.1 0.95 1 1.05 Current(dVB=3 V)/Cal. Coeff. Current (dVB=3 V)/Cal. Coeff. Excellent Gain*PDE uniformity: RMS~1% 11

Gain vs V-VB (T=22 ° C) 600 500 2.8 mm dia. SiPM 3.3 mm dia. SiPM 400 Gain, 10 3 300 200 100 0 0 1 2 3 4 5 V-VB [V] Gain is 350k at V - VB= 4 volts, meets the specification 12

Photon Detection Efficiency vs V-VB 50 2.8 mm dia. SiPM 40 3.3 mm dia. SiPM PDE(515 nm) [%] 30 20 10 0 0 1 2 3 4 5 6 7 V-VB [V] PDE ~ 30% at dVB = 3 Volts, 35% at dVB = 4 volts, exceeds the specification PDE the same within errors for 2.8 mm and 3.3 mm devices 13

PDE – Spectral response T=25 C 45 40 dVB=4.0 V 35 30 PDE [%] 25 20 15 10 5 0 350 400 450 500 550 600 650 700 750 800 Wavelength [nm] Glass widow with special filter was designed by Hamamatsu for the CMS SiPM arrays to cut UV light which can be produced by muons and hadrons in plastic fibers 14

Optical Cross Talk 45 40 2.8 mm dia. SiPM 35 3.3 mm dia. SiPM 30 X-talk [%] 25 x-talk from one micro-pixel to 20 another within a single SiPM 15 10 < 20% at V - V b = 4 volts, meets 5 the specification 0 0 1 2 3 4 5 6 7 V-VB [V] Note: x-talk from one device to its neighbor is too small to measure 15

C vs V measurements 2.8&3.3 mm dia. HPK arrays Ch1 1.0E-09 Ch2 Ch3 Ch4 8.0E-10 Capacitance [F] Ch5 Ch6 6.0E-10 Ch7 Ch8 Ch9 4.0E-10 Ch10 Ch11 Ch12 2.0E-10 Ch13 Ch14 0.0E+00 Ch15 Ch16 0 20 40 60 80 Bias [V] Two arrays plotted Good uniformity within array Similar behavior for 2.8 mm and 3.3 mm devices Capacitance at operating voltage well within spec 16

IV dependences (forward bias) 3.3 mm dia. SiPMs 2.8 mm dia. SiPMs 2.5E-02 2.5E-02 ch1 ch1 2.0E-02 2.0E-02 ch2 ch2 ch3 ch3 Current [A] Current [A] ch4 1.5E-02 1.5E-02 ch4 ch5 ch5 ch6 ch6 1.0E-02 1.0E-02 ch7 ch7 ch8 ch8 5.0E-03 5.0E-03 0.0E+00 0.0E+00 0 1 2 0 1 2 Bias [V] Bias [V] Used to measure forward resistances  quenching resistances 17

Forward resistance  R q ~ 900 kOhm (small spread <4% RMS) 18

Recovery time 2.8 mm dia. SiPMs 0.05 -0.05 ch1 Amplitude [V] ch2 ch3 -0.15 ch4 ch5 Response to laser pulse for: -0.25 ch6 ch7 2.8 mm devices (above) ch8 -0.35 3.3 mm devices (below) 2.0E-08 4.0E-08 6.0E-08 8.0E-08 1.0E-07 Time [s] Recovery time is ~ 7-8 nsec for both, meets specification 3.3 mm dia. SiPMs 0.05 -0.05 ch1 Amplitude [V] ch2 ch3 -0.15 ch4 ch5 ch6 -0.25 ch7 ch8 -0.35 2.0E-08 4.0E-08 6.0E-08 8.0E-08 1.0E-07 Time [s] 19

Accelerated aging and thermal cycling tests • Total of 80 channels were operated nonstop at 70 o C for a 4 week period • 24 channels were subjected to temperature cycling between -15 o C and 50 o C under high humidity – initially with a one hour cycle time, then fast cycling every 15 minutes • Devices were monitored continuously during these tests • Before/after measurements of IV curves to assess damage 20

Stability at high temperature 10 arrays, 80 channels in continuous operation at 70 o C for 4 weeks V b was increased after two days, correct for temperature effect Stable dark current, no runaways, no failing channels First 2 weeks of data are shown at this slide. 21

IV curves for dark current after 4 weeks at 70 ° C IV curves for dark current, before (top) and after (bottom) four weeks of continuous operation at 70 o C 80 channels plotted no failures, no increase in dark current (actually a bit quieter!) 22

Temperature cycling (and humidity) tests Temperature cycling between -15 o C  +50 o C, one hour per cycle, 200 cycles Humidity relatively low and stable 3 arrays, 24 channels tested, all channels monitored Repeated with fast cycles (15 min), higher humidity – devices still stable 23

IV curves for dark current before/after thermal cycling (with high humidity) tests IV curves for dark current, before (top) and after (bottom) three days of fast thermal cycling with high humidity 8 channels plotted no discernible effect observed 24