Recent Experiences with Aging in RPC Systems - desy.de · Recent Experiences with Aging in RPC Systems Daniel Marlow Princeton University Workshop on Aging in Gaseous Detectors Hamburg,

Recent Experiences with Agingin RPC Systems

Daniel MarlowPrinceton University

Workshop on Agingin Gaseous DetectorsHamburg, Germany

October 5, 2001

DESY Aging Workshop, October 5, 2001 2

Talk Outline

• A provocation!• RPC basics• Experience in Belle• Experience in BaBar• Conventional Aging (LHC)• Conclusions and Summary


A Provocation: the Promise

Mid 1990s: RPCs are thought to be a robust,economical, and proven technology ideally suited tolarge-area detection systems. Both Belle and BaBaradopt them for their muon systems.


. . . A Provocation: The Reality

October 2000: A BaBar review committee writes:“The IFR was identified by BaBar as an area ofmajor concern. There is substantial degradation ofthe efficiency of the RPCs. If this continues, it islikely that muon identification will be fatallyimpaired by the end of 2002 . . . “

September 1998: Belle is forced to switch off itsRPC system in the face of a rapid deterioration indetection efficiency. The system is quickly dying.


Notes

• In this talk I will attempt to describe what wentwrong.

• I will spend more time on Belle since that storymay not be as well known and since there willbe a separate presentation by BaBar in the nexttalk.

• I will comment briefly on the prospects forLHC, but once again there will be separate talksfrom ATLAS and CMS.


The Magnet

Belle

Both Belle and BaBarhave instrumented theirflux returns.Requirements:

• Low rates

• Modest pos. res.

• Good efficiency

• Large area:

• low cost

• easy assembly

• simple electronics RPCs


RPC Principles of Operation

A passing charged particle induces an avalanche,which develops into a spark. The discharge isquenched when all of the locally ( )available charge is consumed.

2cm 1.0≈r

Spacers

Signal pickup (x) India Ink

Glass plates 8 kV

Signal pickup (y) India Ink

+++++++++++++++_ _ _ _ _ _ _ _ _ _ _

+++ +++++_ _ _ _ _ _ _The discharged area recharges slowly through

the high-resistivity glass plates.

Before

After


Plateau Curve

2 mm gap RPCs plateau at afairly high voltage.

By using both positive andnegative supplies one canavoid the complications of asingle-ended 7.5 kV HVPS.

Note the slight falloff inefficiency well above theplateau. This effect is realand typical in glass RPCs.


Principles of Operation: I vs V CurveGlass RPCs have a distinctiveand readily understandablecurrent versus voltagerelationship.

•Low voltage•

•

•High voltage•

•

∞≈gapR

0gap ≈R

spacerRdIdV =

glassRdIdV =


RedundantDesign

One very fortunateaspect of the Belledesign is that each set ofpickup strips (x & y)“sees” two gas gaps.This has really helped tosave our bacon.


System Performance

Cosmic ray plateaucurves. Thebenefits of havingredundant layersare evident.


Cosmic Ray Timing Spectra

The 32 us range of theTDCs pulls in a lot of noisehits, but most of the signalevents are contained in anarrow ( ) rangearound prompt. ns 100±

ns 100±

The timing errors aredominated by cable lengthvariations (no correctionshave been made).


A Major Problem DevelopsThe first signs of troubleshowed up shortly afterinstallation and lookedsomething like the plotto the right. The currentfrom a chamber would“suddenly” show adramatic increase.

Given that there is a“pedestal” currentresulting from thespacers, the true dark-current increase was infact substantial.


. . . A Major Problem Develops

Many RPCs showeda dramatic loss ofefficiency. Thesituation became sobad that we decidedto switch off thesystem (whichturned out to be alucky guess!).


. . . A Major Problem Develops

• The problem started to show up almost immediately uponinstallation in June of 1998.

• The number of high-current RPCs increased steadily overthe summer. The failure rate increased dramatically in lateAugust, forcing us to shut down the system pending a betterunderstanding.

• High current is a serious problem in glass RPCs (see nextslide).


. . . A Major Problem Develops . . .

High dark currents induce asignificant IR voltage dropacross the glass plates,which lowers the voltageacross the gap, causing thechamber to slide off theefficiency plateau.

Increasing the appliedvoltage doesn’t help since itmerely results in increaseddark current.

The is what I like to call theclassic “RPC Death Spiral”.


A Major Problem Develops . . .

The expectedcorrelation betweendark current andefficiency loss isreadily apparent.


A Solution Emerges . . .

• After several weeks of study we determined thatthe problem was due to high levels of water vaporin the gas.

• Although we were aware of this problem (evenmore severe problems occurred in the early days ofour R&D when some of our collaborators usedurethane gas tubing, which is highly permeable),we had incorrectly assumed that water would notpermeate our polyethylene (Polyflow) tubing.


. . . A Solution Emerges• In fact we were susceptible to water contamination forthe following reasons:

•Low flow rates

•Long (5~12 m) runs of plastic tubing.

•Hot and humid weather during the Japanese rainyseason.

• Using published data for polyethylene, we were able toaccount for the ~2000 ppm concentrations of water vaporthat we observed.



We replaced the longruns of polyethylenewith copper (~5 km inall!) and flowed gas atthe highest possible rate(~one volume changeper shift).

Slowly, but surely, theRPCs began to dry out.



We were greatlyrelieved to see anaccompanying dropin the dark currents!



Although some residual damage remains, the situation is muchimproved.


Efficiency

++ Ψ→ KJB / candidate

A simple test of systemperformance comes fromlooking for missing hitsalong muon tracks.

The three-muon eventto the left shows that theRPC system is workingwell.


Where does all that water come from?

The RPCs continuedto dry for over threemonths. All told anamount of waterequivalent to ~100monolayers wasremoved.


Where Does All That Water Come From?

• Naively one would believe that the surfaces couldhold only ~three molecular monolayers.

• But that is true only for a smooth surface.• As we will see, the surfaces of chambers quickly

become any but smooth.


STM Picture of a Virgin Glass Surface

m 10 µ m 10 µ

z (Å)


Surface of “Good’’ Anode

m 10 µ m 10 µ

z (Å)


Surface of “Bad’’ Anode

m 10 µm 10 µ

z (Å)

We believe that the surfaceswere etched by HF acidformed from the water andthe Freon (R134A) in thegas. The trapped waterprobably affects the surfacesconductivity of the glass.


Belle Efficiency History: Barrel

The barrel systemhas been stablesince “dry out.”Small problemsare masked by theredundant readoutarrangement.

Oct. 1999

Note suppressed zero!


Belle Efficiency History: Endcap

The endcapsystem has seensome degradationover the past twoyears. This maybe the result ofhigh rates in thelayers that areexposed to thebeam.


Belle Efficiency History: Endcap

The situation improveswhen the outer layers,which are exposed tobeam backgrounds areremoved from theaverages. Theavailable data cannotunambiguouslydistinguish betweenaging and rate effects.


BaBar

The bakelite RPCsin BaBar haveexhibited seriousefficiency loss.Attempts at alasting antidotehave not yet beensuccessful.


BaBarNewer RPCs seemto being faringbetter.

⇒ The problems thatafflicted BaBar alsoappear to beepisodic (poor QCduring constructionand/or carelesshandling duringearly operation).

New

Original

See talk by D. Piccollo.


What About Conventional Aging?With the exception ofsome modules thatsee large beambackgrounds, theintegrated doses at theB Factories are small.

That will not be thecase at the LHC.

Here conventionalaging studies areneeded.

G. Pugliese(CMS)

London Sept.1999


What About Conventional Aging?

G. Pugliese (CMS)

London Sept. 1999Streamer mode, Qtotal= 0.05 /cm2/gap


Summary and Conclusions

• The two largest deployments of RPCs have notbeen unqualified successes:

• The Belle RPCs came close to disaster, butseem to be OK now.

• The BaBar RPCs have suffered seriousdamage that appears to be irreparable.

• The problems are not classic aging problems andprobably would not have shown up in standardaccelerated aging tests.


. . . Summary and Conclusions• Data from LHC R&D aging tests indicate thatunder the right circumstances, RPCs can withstandlarge integrated doses.

• Many of the good things that were said aboutRPCs are true, but there are no miracles.

• Unpredictable surface effects make RPCsparticularly finicky (c.f. CsI photocathodes,MSGCs).


. . . Summary and Conclusions

• As with most detector technologies it isimportant to:

• maintain the high QC standards duringmanufacture.

• treat RPCs as the delicate instruments thatthey are during operation.

• If this is done, RPCs will probably work OK, butif I had to do it over again I would take a long hardlook at scintillating strips.

Documents

Recent Experiences with Aging in RPC Systems - desy.de · Recent Experiences with Aging in RPC Systems Daniel Marlow Princeton University Workshop on Aging in Gaseous Detectors Hamburg,