HAL Id: in2p3-00551856
http://hal.in2p3.fr/in2p3-00551856
Submitted on 11 Apr 2011
HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.
The ATLAS Liquid Argon calorimeter alignment in time for exotic search
L. Aperio Bella
To cite this version:
L. Aperio Bella. The ATLAS Liquid Argon calorimeter alignment in time for exotic search. Journées
Jeunes Chercheurs 2010, Nov 2010, Angers, France. pp.91-94. �in2p3-00551856�
Ludovia APERIO BELLA
Laboratoire d' Anney-le-Vieux de Physique des
Partiules
Résumé
Le alorimétre ArgonLiquide(LAr) del'expériene
ATLAS doit pourvoir à la reonstrution d'énergie
des les partiules qui sont produites au ollisionneur
LHC. L'alignementautemps destouteslesellulesdu
alorimétreestessentielpourgarantireunebonneréso-
lution autemps,eparamètrejoueunrleimportante
dansbeauoupdesthéoriesaudelàdeModeléStandard
quiprédisentl'existenedespartiulesproduiteaveun
tempsdeviemesurabledansledéteteur.Danserap-
port on se propose de dérire l'alignement en temps
desleellulesduLArqueaétéfaitaveledonnesdes
ollisionsprotonprotondansl'année2010.
1.1 Introdution
The Liquid Argon Calorimeter (LAr) system is an
importantpartoftheATLASdetetor[1℄.Itallowsto
trigger, to identify and to measure eletrons, photons
and jets of energies ranging from a fewGeV to a few
TeV. To insure optimal energy reonstrution all the
alorimeter readouthannelsshould bewellalignedin
time. Previousstudiesresultedin thealignmentupto
afew ns [2℄[3℄. Test-beamstudieshaveshown thatit
is possible to ahieve time alignment on the level of
100
ps[4℄.Amorerenedtimemeasurementofthealorimeter
signal ould have several appliations in the ATLAS
environment.
It anhelp in the identiation of longlived parti-
les and in themeasurementof theirdeaytime. One
typialexampleours in theGaugeMediated Super-
symmetriModelsinwhihtheneutralinotravelsasig-
niantdistanebeforedeayingintoaninvisiblegrav-
itino and a non-pointing and out-of-time photon [5℄.
Thisfeatureanbeusedtomeasuretheneutralinolife-
timeusingtiminginformationfromtheeletromagneti
alorimeterofthereonstrutedphotondiretion.Ade-
layed luster timeis expetedbeauseof the negative
valueof
β =
vc< 1
of themassiveneutralinosand the longertrajetorythroughthedetetor.Timing information ould also be used in searhes
for exoti partiles with anomalously high ionization
energylossestodistinguishthemfromStandardModel
bakgrounds[6℄.
Finallyathighluminositythetimeinformationmay
alsobeusedtorejeteventswithlargepile-upativity
from previousbunh rossings.
This note desribes the LAr hannel timing align-
ment as it was performed during the 2010 proton-
proton data-takingoftheLHC.
1.2 The ATLAS Liquid Ar-
gon alorimeter and its signal
read-out system
Figure1.1Cut-awayviewoftheATLASLiquidArgon
alorimeter
As shown in Fig. 1.1, The ATLAS LAr alorimeter
onsistsoffoursubdetetors:theeletromagnetibar-
rel(EMBarrel),theeletromagnetiendap(EMEC),
thehadroniendapandtheforwardalorimeter.
The subdetetors over an
η
-range of up to|η| = 4.9
whereη
isthe pseudo-rapiditydened likeη = − ln(tan(
θ2))
.Inadditiontothesubdivisionalongη
forthe EMBarreland alongφ
fortheEMEC, theyare separatedintothree dierentlongitudinal setions
orlayers(Fig.1.2).Thefrontlayerisnelysegmented
intostripsalong
η
andφ
,respetively.Themiddleand baklayersareoarser inη
but ner inφ
forthe EMBarrelandvieversafortheEMEC.Furthermoreinthe
entralregion(
|η| < 1.8
)theeletromagnetialorime- ters areomplementedbypresamplers(PS).TheLAralorimeterisasamplingalorimetermade
of liquid argon as the ative material and lead, op-
perortungstenasthepassiveabsorber.Whenharged
partilesrosstheliquidargongapbetweeneletrodes
andabsorbers,theyionizeit.Undertheinueneofthe
eletrield, theionizationeletronsdrifttowardsthe
eletrodeinduingaurrent.
At the hannel level, the treatment of the analog
∆ϕ = 0.0245
∆ η = 0.025 37.5mm∆ η = 0.0/8 = 4.69031 mmm
∆ϕ=0.0245x4 36.8mmx
Trigger Tower
∆ϕ = 0.0982
∆ η = 0.1
16X0
4.3X0
2X0
1500 mm
470 m m
η ϕ
η = 0
Stri p cel l s i n L ay er 1
Square cel l s i n L ay er 2 1.7X0
Cells in Layer 3
∆ϕ×∆η = 0.0245× 0.05
Cells in PS
∆ϕ×∆η = 0.025 × 0.1
Trigger Tower
=147.3mm4
Figure1.2Theaordionandsamplingstrutureofthe
EMBarrelalorimetershowingthediereneingranularity
betweenthedierentsamplings.Itslongitudinalsegmenta-
tionallowsustolookattheshowerdevelopment[1℄
signal is performed in the front-end eletronis. On
the front-end boards (FEB) loated in the Front-End
Cratesoutsidetheryostatthesignalisrstamplied
by a preamplier. In order to aommodate the large
dynamirangeofpulse(fromtensofMeVtoafewTeV)
and minimize thetotalnoisefrom the eletronis,the
signalis shaped andsplit into three linearsaleswith
a typial multipliative fatorsof
1
,9.2
and92
alledlow,medium andhighgain.
Foragivenhannel thesethreesignalsarestoredin
an analogpipeline [7℄ untilthetriggerdeision.When
thedeisionismadethesignalseitherfromeahgainor
fromthemostsuitedoneaordingtoaseletionbased
on theamplitudeof thesignalwith themedium gain,
are digitized by a 12 bit Analog-to-Digital Converter
(ADC)in vesamplesspaedby
25
ns.ToeahFront-End Crateisassoiatedaratewhih
ontainstheread-outdriver.Thereonstrutionofthe
timeandenergyofthephysissignalforeahhannel,
performedintheread-outdriverbytheOnlineDiglital
SignalProessor,isbasedonanOptimalFiltering(OF)
algorithm [8℄.
1.2.1 The Optimal Filtering Method
TheOptimalFilteringmethod providesoptimalval-
ues of timeand energyfor eah given hannel in eah
givenevent.Thismethodrequiresthattheexatshape
oftheexpetedphysissignalaftertheeletroniread-
outhainisknown.Toobtainthisshapefromtheorig-
inal ionization pulse atransfer funtion is needed for
every hannel. This funtion is obtained from a ded-
iated alibration runs in whih both the input and
outputpulse arewell-known.
TheOptimalFilteringCoeients(OFCs)
a
iandb
iare omputed per hannel from the predited physis
pulse-shapeintimeusingthefollowingformulas:
A = X
5i=1
a
is
i (1.1)Aτ = X
5i=1
b
is
i (1.2)where
A
isthephysispulseamplitudeinADCounts,τ
isthepeak-timeshiftbetweenthereferenepulseandthe physispulse and
s
i are theADC samples. Theseequationsareonlyvalidifthevaluesof
τ
aresmall.Forperfetly timed detetor and in-time partiles
τ
mustbe lose to zero, while the larger values indiate the
need for better timing or the presene of out-of-time
partiles in the event.If ADC samples are dominated
by noise, Eq. 1.2 does not give reliable results. Than
τ
is omputedonly forpulses whihhaveatleast oneADC sample largerthan fourtimes thenoiseRMS in
that readouthannel.
As
τ
mightnotbesmallforsomehannels,espeiallyin thebeginingof data taking, there are 8sets of the
OFCs with 3ns phase-shift between then used in the
physisrunningand24setswith1nsphase-shiftusedin
the alibrationruns toonstrutthe referenephysis
pulse-shape 1
. Figure1.3 showsthedierene between
apreditedsignalshapeandarealionizationsignalin
the seond layerof the EM Barrel,given by aosmi
muon.
time (ns) 0 100 200 300 400 500 600 700 800
ADC counts
-200 0 200 400 600 800 1000 1200
-0.04 -0.02 0 0.02 0.04 Data
(Data-Prediction)/Max(Data) Prediction
MIDDLE LAYER EM BARREL 2008 ATLAS cosmic muons
ATLAS Preliminary
Graph
Graph
Graph
Figure 1.3 A typialpulsefor thedierentEMBarrel
middlelayers. Theblue dotsorrespondto thepredition,
theredonestothedataandthegreenpointsorrespondto
thediereneinperentagebetweendataandpredition.
With thepresentsettings (8 phaseswith 3ns shift)
the setsarehosenin awaythatthefourth setof the
OFCsorrespondsonaveragetothesignalpeakbeing
in thethirdsample.
Atthe beginning oftheLHC data taking,the OFC
sets loaded in the LAr Online Diglital Signal Proes-
sor arehosenaordingtothetiminginformationex-
trated from the alibration data and the omplete
1. The predited pulse is dened with a time dierene of
25ns
24 ns between two onseutive samples and isbinned in 24
delaystepsspaedby∼1 .04ns
knowledge of the readout path [2℄. When the physis
pulse is reorded
A
andτ
an be alulated from theEqs.(1.1)and(1.2).If
τ
obtainedislarge,theOFpro-edure isperformediterativelyusingsets oftheOFCs
with appropriatephasesuntil thevaluesof
τ
obtainedarelessthan502ps.
Notethatatimemisalignmentof
∼ 5
nsbetweenthephysispulseandthereferenepulseinduesabiason
thereonstrutedenergyof
0.5%
[9℄.1.3 Timing analysis
Tohoose good physispulses for the timing align-
mentweuseseletionasoutlinedbelow.
Only hannels in the high gain and in whih are
properlyalibrated and for whih the OFCs iteration
onvergesareused in thefollowinganalysis.A
χ
2-likequalityfator
Q
,estimatedbyomparingthemeasuredpulseshapewiththerefereneshapeanbeusedtose-
let onlygoodpulses.
Toalulate onlyLArhannelsthatmeasure anen-
ergy above1GeV are seleted for this study. A mea-
surementabovethisthresholdisextremelyunlikelyto
betheresultofeletroninoisealone.Figure1.4shows
this utinthetimevsenergydistributionfor thefour
dierentlayersintheEMBarrel.Thisutisneessary
beauseforagivenhannelinaollisioneventthetime
spreadwithtorespettotheenergymeasuredisaon-
volution of the notalignment in time of the hannels
andtheintrinsitimespreadofthehannelduetothe
energydependeneofthetimeresolution(
σ
τ=
Ea⊗c
).With this energy ut for eah layer the time spread
within aFEB islessthan
∼ 3
ns.ofcTime [ns]
-100 -80 -60 -40 -20 0 20 40 60 80 100
Energy [MeV]
0 2000 4000 6000 8000 10000 12000 14000 16000 18000 20000
1 10 102 ATLASWork in Progress
EM Barrel PS
ofcTime [ns]
-100 -80 -60 -40 -20 0 20 40 60 80 100
Energy [MeV]
0 2000 4000 6000 8000 10000 12000 14000 16000 18000 20000
1 10 102 103 ATLASWork in Progress
EM Barrel strips
ofcTime [ns]
-100 -80 -60 -40 -20 0 20 40 60 80 100
Energy [MeV]
0 2000 4000 6000 8000 10000 12000 14000 16000 18000 20000
1 10 102 ATLASWork in Progress
EM Barrel middle
ofcTime [ns]
-100 -80 -60 -40 -20 0 20 40 60 80 100
Energy [MeV]
0 2000 4000 6000 8000 10000 12000 14000 16000 18000 20000
10-1 1 10 ATLASWork in Progress
EM Barrel back
Figure 1.4 The distributionof time vsenergyfor the
fourdierentlayersintheEMBarrel.
Any distortion ofthe signalshapean also produe
a bias of the time reonstrution,for that reason an-
other important eet whih ould ompromise the
timealignmentoftheLArhannelsistheeletriross-
talk.Theeletroniross-talkisaphenomenainwhih
neighboring ells. This signal loss in the neighboring
ellsauseadistortionin thepulseshape.
Toavoidthatthiseetouldbiasthetimemeasure-
mentweonsideredforeaheventthehighestenergeti
hannelineahlayer(
E
1),andwedidnottakeintoa-ounttherst twoneighboringells(
E
2).Toinreasethestatistisweonsideredalsothethehighestenergy
ellaftertheseondneighborsifitsenergy
E
3 >E
1.Toalignintimethe
≈ 180
kellsoftheLAralorime-ter twoadjustmentsareused:
a global hardware alignment at the level of the
FEB timeoset
ahannelbyhanneladjustmentofthesetofOp-
timalFiltering Coeients(OFC).
The latterhasaniteminimal orretionof
∼ 500
psbeausethereisurrentlynoextrapolationbetweentwo
onurrentpointsin thereferenepulsewhihare
1.04
nsapart.
1.4 Timing alignement results
BytheendofApril2010weolleted
∼ 0.3
nb−1ofollision data whih gaveenough statististo haveat
least two eventsin
94, 4%
of the hannels. With thatdata wealulatedtheFEBtime osetastheaverage
of themeantimeof eah hannel forasingleFEB for
eah LAr subdetetor. This orretion was applied to
alignthealorimetertimeatthehardwarelevel.
At theend of July 2010 with aintegrated luminos-
ityof
∼ 1.1
nb−1 weimplemented ahannel byhan- nel adjustment having97, 3%
of hannels on. For thehannelbyhannelorretionwetakethesinglehan-
nel averagetimeand wesubtrat theglobalrun time,
thisavoidsadependenyofthisorretionontheshift
ausedbytheLHClokdriftfoundinMay2010data.
FEB time offset [ns]
-10 -5 0 5 10
Number of FEBs per 0.5 ns
1 10 10
210
3s = 7 TeV collisions
Work in Progress
EM BARREL
RMS before correction = 1.0 ns RMS after correction = 0.4 ns ATLAS
Figure 1.5ForEMBarrel:theblakhistogramshows
the FEB time osets before the orretion, and the red
histogramshowsthesamedistributionaftertheorretion
wereapplied.
Weappliedthesinglehannelorretionasashiftto
thereferenepulse shape.BeausetheOFCsetsome
in units of 1.04ns, time orretion preisionis driven
bythisbinningonstraint.
EndCap_tch_TOT_FEB_slot_6_pfx
Entries 128
Mean 64.06
Mean y 0.0002555
RMS 36.94
RMS y 0.3419
channel
0 20 40 60 80 100 120
ofcTime [ns]
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
EndCap_tch_TOT_FEB_slot_6_pfx
Entries 128
Mean 64.06
Mean y 0.0002555
RMS 36.94
RMS y 0.3419
ATLASWork in Progress
RMS 0.3419
ofcTime [ns]
-15 -10 -5 0 5 10 15
# channel Slot6
1 10 102
103
RMS 0.3419 ATLASWork in Progress
Figure1.6ForEMEndap:intheupperplotonaverage
the timeversusthe hannelnumberinall theFEB inthe
middlelayerwithaφsimmetryintheatualtimesituation;
inthe bottomplotthe hannel timespreadwithin aFEB
inthemiddlelayer,intheatualtimesituation.
The eet of the timing adjustment at the level of
theFEBs isshownin Figure1.5. For example,forthe
EM Barrel of the LAr the blak histogram shows the
FEB time osets before the orretion, and the red
histogram shows the same distribution after the im-
plementation of both the orretions (FEBs time o-
set and the hannel by hannel one) with Otober-
November2010data orresponding to aluminosityof
∼ 8
pb−1.WegetasigniantimprovementbeausealltheFEBsare wellalignedwithagoodRMSof
0.4
ns.Figure1.6showsthedistributionofthehannelstime
spread versus the hannel number. From the results
shownhereitislearthatwereahatimespreadofthe
singlehannelsinside aFEBanthe orderof
∼ 400
nsforalmostallthehannels.
1.5 Outlook
Inthispaperwehaveshowntheresultsofourwork
on the time alignment of the hannels of the LAr
alorimeteruptoafewhundredpslevel.Timingmea-
surement beomes more preise with higher statistis
aquired. It is not possible to ombine runs from the
wholeyear2010forthismeasurementbeausetheLHC
lokdriftof
2
nspermonthintroduesanon-negligible shiftinourtimesale.Thuswehaveusedonlythehigh-eststatistirunsavailable.In2011thedatawillbea-
quired more eiently thus this problem is removed.
We will implement a new round of FEBs orretions
a new set of hannel by hannel orretions with the
early2011data.Thenwewill onentrateonstudying
small systemati eets, suh asa shift in the beam-
spot position, the analog pipeline storing delays and
the implementation of extrapolation proedure in the
OF phase ode whih would allow themodiationof
the preisionofthehannelby hannel orretionim-
plementation.Thosemodiationwouldn'tallowusto
reahultimatetimingalignmentonthelevelof
100
ps.Aknowledgements
ThisworkisbeenpossiblethanksNiolasBergerwho
has produedthentuples usedfor thisanalysis,Paolo
Iengo and Maro Del Mastro for the implementation
of the time orretion whih we have produed and
IsabelleWingerter-Seez and Tetiana Hryn'ova for the
usefuldisussions.
Référenes
[1℄ ATLASCollaboration.ATLASCalorimeterPerfor-
mane.Tehnialreport,1997
[2℄ Conf.ATLAS-COM-LARG-2009-001T.Guillemin,
G. Perrot, L. Ionomidou-Fayard, G. Unal, H.
Wilkens Time alignment of the ATLAS liquid ar-
gonalorimeters.
[3℄ Conf. ATLAS-COM-LARG-2009-048 A. Penson
and K.Leonhardt Timing Alignment of the entire
ATLASLiquidArgonCalorimeterwithosmirays.
[4℄ M. Aharrouhe et al. Time resolution of the AT-
LAS barrel liquid argon eletromagneti alorime-
ter.Nul. Instr. and Meth.A: Aelerators,Spe-
trometers, Detetors and Assoiated Equipment,
597(2-3):178-188,2008.
[5℄ TheATLASCollaboration, Supersymmetry Signa-
tures with High-pT Photons or Long-Lived Heavy
Partiles.ATL-PHYS-PUB-2009-069.
[6℄ The ATLAS Collaboration. Searh for Massive
Long-lived Highly Ionising Partiles with the AT-
LASDetetorattheLHC arXiv:1102.0459v2[hep-
ex℄
[7℄ D.Breton,etal.,HAMAC,arad-hardhighdynami
range analog memory for ATLAS alorimetry, in :
Craow2000,EletronisforLHCExperiments,pp.
203 − 207
.[8℄ W.E. Cleland, E.G. Stern, Nul. Instr. and Meth.
A338(1994)467.
[9℄ Conf. ATL-COM-LARG-2010-008 L. Courneyea;
D. Dannheim; M. Delmastro; S. Elles; M.
Gouighri; L.Ionomidou-Fayard; I.Koletsou; W.
Lampl; Z. Liang; L.Marh; P. Strizene; F. Tar-
rade; R. Ueno; M. Vinter; Z. Weng; H. Zhang
Computation and validation of the eletroni al-
ibration onstants for the ATLAS Liquid Argon
Calorimeters