Yu. Guz

LHCb HCAL PMT problems

initial report

1

Yu. Guz

LHCb HCAL & ECAL

2

PMTs of type HAMAMATSU R7899-20 are used in both HCAL and ECAL. The detector granularities are shown below: 6016 PMTs are used in ECAL and 1488 in HCAL.Total of 8000 PMTs R7899-20 were purchased from HAMAMATSU in 2004.

Since then, the PMTs installed in HCAL experience strange (and serious) problems, while in ECAL they behave as expected.

HCAL ECAL

Yu. Guz

LHCb HCAL & ECAL: PMT working conditions

3

Each PMT is equipped with an individual Cockcroft-Walton (CW) HV supply mounted on the PMT pinsThe HV value is controlled by an analog input on the CW PCB. CW: same schematics but different PCB and different way of mountingPMT mounting: see next slide

•HCAL: gains range from 104 (centre) to 2·105 (periphery)

•HV ranges from 900 to 1300 V•Working anode current (at LHC stable running at luminosity of 2·1032 cm-2s-1): from 0.5 µA (periphery) to 15 µA (centre), 1.5 µA average

•ECAL: gains range from 103 (centre) to 2·104 (periphery)

•HV ranges from 600 to 900 V•Working anode current (at LHC stable running at luminosity of 2·1032 cm-2s-1): from 0.2 µA (periphery) to 6 µA (centre), 1 µA average

Yu. Guz

LHCb HCAL & ECAL: PMT usage

4

ECALMechanical stresses are possible in HCAL PMTs: • the PMT is soldered to PCB

through a hard plastic spacer, the pins can remain stressed after soldering

• the PMT entrance window is in tight contact with a polystyrene light mixer (pressed to its surface with the plastic nut shown on the photo)

The ECAL mounting design does not have such features – less mechanical stress

HCAL

CW PCB PMT

nut spacer

magneticshielding

PMT/CWassembly

HCAL: The PMT bulbs are covered by a black paint (of course except entrance window), for light tightness

Yu. Guz

Significant degradation of PMT parameters during LHC Run I (timescale of ~years)! ~15% of HCAL PMTs affected! not the case for ECAL! ~150 degraded PMTs dismounted from HCAL, can be used for studies.Three kinds of problems:• Significant (positive) rate effect, up to +5..30 % (initial selection: within

±2%).The rate effect is measured as a ratio of LED signals before and after the collisions start; the anode current changes from ~20-100 nA to 1-20 μA

• Dark current in many PMTs (increasing with time: from 0..2 nA to few μA)• Degradation of gain: factor of 2-10 per year

HCAL: summary of problems with PMTs

5

Common features: None of the effects is correlated to occupancy or radiation dose (distance to the

beam); the affected PMTs are randomly distributed over the HCAL surface The presence of the three types of effect is correlated, but not 100%

Yu. Guz

Features of PMT dark current (DC): “ohmic” behaviour, i.e. (roughly) proiportional to HV, therefore is not

connected to the photocathode emission (PMT gain is ~HV6.5). Such PMTs show 200..1000 MOhm resistance between anode and

some dynode(s) (!!!) Once the non-zero DC appears, it gradually increases all the time the

PMT is installed on HCAL (even during LHC shutdowns, without HV) On statistics of several PMTs with DC, which were dismounted from

HCAL in 2011 and 2012: the DC stops increasing after dismounting the PMT from HCAL

HCAL: PMT dark current

6

Yu. Guz

HCAL dark current @ PHYSICS HV, February 2013

7

I>100 nA: 42 PMTsI>50 nA: 77 PMTsI>10 nA: 147 PMTs

Distribution of dark currents in the 1488 HCAL PMTs (left); the dark current values w.r.t PMT positions on the detector (right)

HCAL dark current @ PHYSICS HV, November 2013

8Yu. Guz

Generally, the dark current increased (during shutdown! No beam, no HV). New cells appeared with dark current > 10 nA; in particular, in one PMT out of those which were replaced last time (2011/2012). (in 2012 the dark current was <1 nA, rate effect -0.7%).

I>100 nA: 58 PMTsI>50 nA: 100 PMTsI>10 nA: 187 PMTs

Yu. Guz

HCAL: PMT dark current: location

9

One PMT (large dark current, conductivity of ~1 GOhm between anode and a dynode) was opened, in order to try to understand the location of conductive substance:- no conductivity between pins over the bulb surface the black paint is not a

culprit; - large (and unstable) conductivity between pins over the dynode support.

Yu. Guz

PMT rate effect (reminder: working anode current 0.5-20 mkA): LED monitoring system is running continuously during the data taking,

the LEDs flash at “empty” bunch crossings The rate effect of each individual PMT can be measured as variations

of LED response before and after the start of collisions in LHC The LED amplitude increases slowly after the collisions start (few

hours). The magnitude of such variation gradually increases in the affected

PMTs: e.g., in November it is larger than in May

HCAL: PMT rate effect

10

Yu. Guz

HCAL rate effect Nov 2012

11

139 PMTs have >5%180 PMTs have >4%

Significant positive rate effect in >10% of HCAL PMTs !!!A slow effect (hours), must be caused by PMTs, rather than electronics (e.g. CW)

an example

The rate effect is very large in some PMTs!

t0ref

An example of time dependence of the amplitude of the calibration signal (top left).ref – right before the collisions startt0 – 6 hours later

The distribution of rate effect values (bottom left) The r.e. values w.r.t. PMT positions on the detector

Yu. Guz

cf: ECAL rate effect, same period

12

The rate effect in ECAL is negligible compared to that in HCAL.

The average is slightly negative (-0.6%)

reft0

Yu. Guz

HCAL PMTs: sensitivity degradation over 2012

13

Significant sensitivity degradation (up to factor of 16) occurred in 2012 for many PMTs with high rate effect (apart from high occupancy area).

LHC run I: high rate effect is accompanied by sizeable sensitivity degradation (info available from detector calibrations)

No significant sensitivity degradation in ECAL (next slide).

Yu. Guz 14

No significant sensitivity degradation in ECAL during the same period (Apr-2012 / Dec-2012). Of course we have to select the lower part of ECAL, to avoid zones with clear fiber radiation damage. There are 1024 PMTs in this selection, comparable to the total number in HCAL (1488). The values are much more relaxed (0.9..1.4).

cf: sensitivity degradation in ECAL (Apr-Aug 2012)

Yu. Guz

HCAL PMTs: possible reason of degradation ?

15

A hypothesis: loss of vacuum via micro cracks in the HCAL PMTs. Supported by different design of the HCAL PMT/CW – unlike in ECAL, mechanical tension to PMT pins may occur (a hard spacer between PMT and PCB, see slide 4) micro cracks in glass. Not the case for ECAL.

An experiment: test with He (proposed by V. Rykalin) – compare life time of a good and a bad PMT in He.This test was performed: the result is ambiguous (next page)

Yu. Guz

HCAL PMTs: He test

16

Two PMTs, good and bad, in a tight box flushed with He. Continuous monitoring with LED pulse; the LED itself was monitored by a PIN photodiode (HAMAMATSU S1223-01)

The bad PMT: dark current ~ 50 nA, rate effect ~ 20%The good one: dark current <5 nA, rate effect -1%

Yu. Guz

HCAL PMTs: He test

17

The LED signal reduced at similar speed in both PMTs: by factor of ~2 after 35 days

the dark current did not changeHowever the mechanical situation was not completely similar to that at HCAL: the PMT entrance window was not touching a polystyrene light mixer

good PMT

bad PMT