Calorimeter Backgrounds Occupancy Issues Physics impact of occupancy

Calorimeter Backgrounds • • • Occupancy Issues Physics impact of occupancy Types of background Radiation damage A few comments on alternatives to Cs. I(Tl) 2021/12/16 steve playfer 1

Calorimeter Occupancies in Ba. Bar • Raw Digis > 0. 75 Me. V but can be isolated. These are important for DAQ issues. There are ~400/6580 of these from electronics noise (6%). • Default Digis > 1 Me. V with a neighbour > 5 Me. V are the input to clustering algorithms. – When the default digi occupancy exceeds 2000/6580 crystals (30%) the p 0 -finding efficiency decreases rapidly (empirical result from trickle injection data) • Clusters have a seed digi > 10 Me. V and additional neighbouring digis > 1 Me. V. – There are ~8 of these in multihadron physics events. – Bhabha events give clusters in endcap and backward barrel which scale with luminosity 2021/12/16 steve playfer 2

p 0 signals with trickle injection <1 ms after injection >15 ms after injection p 0 efficiency reduced by x 2 due to high multiplicity after trickle injection 2021/12/16 steve playfer 3

Calorimeter occupancy (digis) in Physics Data Forward E n d c a p 2021/12/16 A typical 45 minute run Backward B a r r e l steve playfer 4

Ba. Bar calorimeter occupancies extrapolated to 2007 Raw Digis/3000 events (random triggers) Clusters/3000 events HER 2021/12/16 Lumi LER HER steve playfer Total LER 5 Lumi Total

Comments on Occupancy • Ba. Bar extrapolations are dominated by the luminosity term • Strong forward endcap peak from (radiative) Bhabhas • Significant peak in last few rows of backward barrel (also Bhabhas) • Remainder of barrel occupancy is from low energy clusters (< 100 Me. V) 2021/12/16 steve playfer 6

Ways of dealing with occupancy • Increase digi (cluster) energy thresholds (costs efficiency and resolution) • Make tighter cuts on digi timing (limited to ~250 ns by preamp shaping time) • Develop digital filtering of waveforms to resolve overlapping digis Subject of talk by Tsukamoto… 2021/12/16 steve playfer 7

Cs. I(Tl) Radiation Damage • Light yield reduces with time due to formation of colour centres in crystals – Equivalent to increase in threshold and/or electronic noise • Depends quite a lot on quality of crystals – Now have plenty of data from Ba. Bar/BELLE • Change is mostly at the front of the crystals – Affects uniformity of crystals response, but studies suggest this is not yet a problem. • No evidence of significant recovery from annealing during long down periods. 2021/12/16 steve playfer 8

Radioactive source results (6 Me. V photons) After initial steep drop, there is now a shallow logarithmic dependence on the integrated luminosity Barrel Modules 1 = forward 7 = backward 2021/12/16 steve playfer Endcap rings 1 = inner 8 = outer 9

The steps are due to redefinitions between run periods This agrees well with the source measurements 2021/12/16 steve playfer 10

Radiation Dose Measured by Rad. FETs at front of calorimeter 4. 1 k. Rad/ab-1 in Endcap 2. 5 k. Rad/ab-1 in Barrel Slope going down (trickle injection & lower backgrounds) 2021/12/16 Integrated luminosity fb-1 steve playfer 11

Evidence for neutrons hitting Diodes Some annealing during down time (200 crystals) (700 crystals) 2021/12/16 steve playfer This is not a problem for calorimeter operation! 12

Calorimeter survivability • Radiation damage reduces light output of crystals (proportional to integral luminosity) – Endcap limit of 20 k. Rad reached after 5 ab-1 – Barrel limit reached after 10 -20 ab-1 • Occupancy increases with luminosity and eventually limits ability to find p 0 s – 30% occupancy in Endcap and backward Barrel in 2007 – How much can we reduce the luminosity term? – How much can we improve occupancy with digital filtering and tighter timing cuts? • Endcap should be replaced in any upgrade • Barrel replacement is a major effort and cost! 2021/12/16 steve playfer 13

Minimal upgrade solution • Replace Endcap crystals with pure Cs. I • Fast component has decay time 28 ns • Light yield lower than Cs. I(Tl) by x 20 (and shifted from 565 nm to 320 nm) • Readout has been demonstrated using APDs • Expected gain of x 2 in effect of radiation damage – There is not much data on this! • Cheap and easy to implement! Subject of next talk … 2021/12/16 steve playfer 14

k. Te. V experience with pure Cs. I Similar to Cs. I(Tl) Caveats: PMT readout Very large crystals Crystal quality? 2021/12/16 steve playfer 15

Conclusions • Occupancy limit of 30% is reached at about: – Endcap 3 x 1034 (integral 2 ab-1) – Barrel 3 x 1035 (integral 20 ab-1) • Radiation damage limits are similar • Digital filtering and raised thresholds may give a bit more headroom (but not a lot) • Endcap replacement will need a fast radiation hard technology – Is pure Cs. I sufficiently radiation hard? – Are other alternatives such as Liquid Xenon or LSO/LYSO crystals viable and cost effective? Subject of talk by Hitlin… 2021/12/16 steve playfer 16

BACK UP SLIDES FOR DISCUSSION 2021/12/16 steve playfer 17

Two-Stage Model to reach 50 ab-1 Year 2004 2006 2008 2010 2012 2014 2016 2018 2021/12/16 Peak Lumi 1. 2 x 1034 2. 0 x 1034 3. 0 x 1034 1. 0 x 1035 2. 0 x 1035 5. 0 x 1035 7. 0 x 1035 1. 0 x 1036 Int. Lumi 240 fb-1 570 fb-1 1. 2 ab-1 3. 0 ab-1 7. 5 ab-1 13 ab-1 30 ab-1 50 ab-1 steve playfer Now 1 -3 x 104 LER Upgrade 1 st Stage 1 -2 x 1035 HER Upgrade 2 nd Stage 5 -10 x 1035 18

Radiation Dose Year 2004 2006 2008 2010 2012 2014 2016 2018 Endcap 1. 0 k. Rad 1. 5 -3 k. Rad 2 -8 k. Rad 3 -24 k. Rad 7 -50 k. Rad 10 -80 k. Rad 25 -200 k. Rad 50 -400 k. Rad Barrel 0. 6 k. Rad 0. 8 -2 k. Rad ENDCAP Upgrade 1 -5 k. Rad 2 -15 k. Rad 5 -35 k. Rad 6 -50 k. Rad BARREL Upgrade 15 -120 k. Rad 30 -250 k. Rad Can run Barrel 2 -4 years longer than Endcap. 2021/12/16 Lowest Scenario = No Lumi term Highest Scenario = 2 x Lumi term steve playfer Is this compatible with 2 -stage model? 19

Light Loss Projections Year 2004 2006 2008 2010 2012 2014 2016 2018 Endcap 0. 82 -0. 89 0. 77 -0. 86 0. 72 -0. 82 0. 66 -0. 78 0. 60 -0. 74 0. 53 -0. 69 0. 45 -0. 62 0. 35 -0. 55 Barrel 0. 91 -0. 96 0. 86 -0. 94 0. 82 -0. 91 0. 76 -0. 87 0. 68 -0. 82 0. 61 -0. 78 0. 55 -0. 73 0. 49 -0. 66 Change from Log Linear extrapolation to one that matches existing data. ENDCAP Upgrade BARREL Upgrade Extrapolation beyond 2012 is very uncertain! (No measurements of Cs. I(Tl) radiation damage in large crystals beyond 50 k. Rad) 2021/12/16 steve playfer 20

Occupancy Projections Usual caveat about extrapolating current backgrounds a long way and doing nothing to reduce the luminosity term Integrated Luminosity Raw Digis (2008) 1 ab-1 1800 (10^35) 10 ab-1 (7500) Default Digis 270 1500 (7000) Barrel Clusters 4. 5 20 (100) Endcap Clusters 4. 0 (20) X 2021/12/16 steve playfer (10^36) 100 ab-1 X 21

Pure Cs. I crystals • Fast component has decay time 28 ns which is x 30 faster than Cs. I(Tl). Solves occupancy problem. • Light yield is lower than Cs. I(Tl) by x 20 (and shifted from 565 nm to 320 nm) – Readout has been demonstrated using APDs – Resolution could be comparable to Cs. I(Tl) • We think there is a gain of at least x 2 in radiation hardness (based on one set of measurements and vague claims from manufacturers!) • No change to geometry of calorimeter • Cost is ~$4/cc which is x 2 more than Cs. I(Tl) 2021/12/16 steve playfer 22

Hybrid crystal solution (Bill W. ) • Replace innermost few radiation lengths of Endcap with LSO crystals (rest replaced with pure Cs. I) • LSO(Ce) - Lutetium Oxyorthosilicate has decay time 40 ns • Light yield similar to Cs. I(Tl) but shifted from 565 nm to 420 nm • Smaller radiation length 1. 1 cm and Moliere radius 2. 3 cm (Cs. I 1. 9 cm and 3. 8 cm) allow an increase in granularity (240 crystals per ring) or a decrease in radius of the calorimeter. • Readout has been demonstrated using APDs • Radiation hard up to 100 MRad! • Very expensive - $50/cc. Only one manufacturer in the world. Major application is in PET scanners. Interesting idea. Solves radiation damage and occupancy. How much LSO can we afford? Can this solution be extended to the barrel? 2021/12/16 steve playfer 23

LSO (or LYSO) Crystals Lutetium (+Yttrium) Oxy. Orthosilicate • Fast light output in 40 ns. Solves occupancy problem. • Smaller radiation length 1. 15 cm (Cs. I 1. 86 cm) and Moliere radius 2. 3 cm (Cs. I 3. 8 cm) • Believed to be radiation hard to 100 MRad! • Light output is 50% (60%) of Cs. I(Tl), but shifted to 420 nm from 550 nm. • Again use APDs to read them out. • LYSO has slightly more light output than LSO, and may be easier to obtain commercially (3 -4 suppliers instead of only one) • Currently the cost is ~$50/cc!! 2021/12/16 steve playfer 24

Xenon calorimeter (David Hitlin) • Light output is within ~20 ns. • Radiation length is 2. 9 cm – Need all of radial space between 700 and 1350 mm for cryostat, Liquid Xe and readout. • Moliere radius 5. 7 cm. – Need sampling along shower depth for overlaps. • Light yield is similar to Cs. I(Tl) but at 175 nm. – Use wavelength shifters and readout by APDs • Radiation hardness is not an issue • Cost of Liquid Xe is $2. 5/cc – Total cost $20 M + readout and mechanics? 2021/12/16 steve playfer 25