Results from Pion Calibration Runs for the H 1 Liquid Argon Calorimeter and Comparisons with Simulations H 1 Calorimeter Group

Home , H1 (particle detector)

NUCLEAR INSTRUMENTS Nuclear Instruments and Methods in Physics Research A 336 (1993) 499-509 & METHODS North-Holland IN PHYSICS RESEARCH Section A

Results from pion calibration runs for the H 1 liquid argon calorimeter and comparisons with simulations H 1 Calorimeter Group

B. Andrieu °, J . Bân', E. Birrelet °, H. Bergstein a, G. Bernardi °, M. Besançon', E. Binder h, H. Blume m, f, K. Borras V. Boudry °, F. Brasse h, W. Braunschweig a, V. Brisson n, A.J. Campbell s, T. Carli °, M. Colombo f, Ch. Coutures', G. Cozzika `, M. David', B. Delcourt °, L. DelBuono P, M. Devel n, P. Dingus °, A. Drescher f, J. Duboc P, O. Diinger', R. Ebbinghaus f, S. Egli ", N.N . Ellis c, J . Feltesse `, t, h, m, h, Y. Feng P, F. Ferrarotto s, W. Flauger M. Flieser m, K. Gamerdinger J. Gayler L. Godfrey e, m, L. Goerlich d, M. Goldberg P, R. Grdssler b, T. Greenshaw', H. Greif M. Haguenauer °, L. Hijduk d, O. Himon P, P. Hartz f, V. Haustein', R. Hiydar n, W. Hildesheim P, N. Huot P, M .-A . Jabiol 1, n, n, m, e, A. Jacholkowska M. Jaffre H. Jung b, F. Just °, C. Kiesling Th . Kirchhoff h, F. Kole d,t, m, e, V. Korbel h, M. Korn f, W. Krasny J.P. Kubenka H. Kiister h, J. Kurzhbfer f, B. Kuznik n, R. Lander t, m, J .-F . Laporte `, U. Lenhirdt f, P. Loch h, D. Lüers J. Marks s, J. Martyniak d , T. Merz h, B. Niroska', A. Nau h , H.K. Nguyen P, F. Niebergall', H. Oberlack', U. Obrock f, F. Ould-Saada', n, m, C. Pascaud H.B. Pyo h, K. Rauschnabel f, P. Ribarics M. Rietz b, Ch. Royon `, V. Rusinov t, b, m, a, h, N. Sahlmann E. Sânchez P. Schacht m, P. Schleper W. von Schlippe k, C. Schmidt D. Schmidt', t,f, V. Shekelyan H. Shooshtari s, Y. Sirois °, P. Stiroba q, M. Steenbock', H. Steiner P, B. Stella', a, m, U. Straumann ", J. Turnau d, J. Tutas L. Urban C. Vallee P, M. Vecko v,°, P. Verrecchia `, G. Villet t, n,r, E. Vogel a, A. Wagener b, D. Wegener f, A. Wegner', H.-P . Wellisch m, T.P. Yiou P, J. 2âcek Ch. Zeitnitz h and F. Zoiner n a I. Physikalisches Institut der RWTH, Aachen, Germany b III. Physikalisches Institut der RWTH, Aachen, Germany School of Physics and Space Research, University of Birmingham, Birmingham, UK ** d Institutefor Nuclear Physics, Cracow, Poland e Physics Department and IIRPA, University of California, Divas, CA, USA $ f Institut für Physik, Universität Dortmund, Dortmund, Germany s Department of Physics and Astronomy, University of Glasgow, Glasgow, UK ** h DESY, Hamburg, Germany ' II Institut für Experimentalphysik, Universität Hamburg, Hamburg, Germany J Institute of Experimental Physics, Slovak Academy of Sciences, Kosice, Slovakia k Queen Mary and Westfield College, London, UK ** e Institute for Theoretical and Experimental Physics, Moscow, Russian Federation m Max-Planck-Institut für Physik, München, Germany n LAL, Université Paris-Sud, IN2P3-CNRS, Orsay, France ° LPNHE, Ecole Polytechmque,IN2P3-CNRS, Palaiseau, France P LPNHE, Universit& Paris VI and VII, IN2P3-CNRS, Paris, France q Institute of Physics, Czech Academy of Sciences, Praha, Czech Republic Nuclear Center, Charles University, Praha, Czech Republic s INFN Roma and Dipartimento di Fisica, Universita "La Sapienzia", Roma, Italy t DAPNIA, Centre d Etudes de Saclay, Gif-sur- Yvette, France * Fachbereich Physik, Bergische Universität Gesamthochschule Wuppertal, Wuppertal, Germany Physik-Institut der Universität Zürich, Ziirich, Switzerland $$

Received 15 April 1993 and in revised form 30 June 1993

We present results on calibration runs performed with pions at the CERN SPS for different modules of the H 1 liquid argon calorimeter which consists of an electromagnetic section with lead absorbers and a hadronic section with steel absorbers . The data cover an energy range from 3.7 to 205 GeV . Detailed comparisons of the data and simulation with GHEISHA 8 in the framework of GEANT 3.14 are presented . The measured pion induced shower profiles are well described by the simulation. The total signal ofpions on an energy scale determined from electron measurements is reproduced to better than 3% in various module configurations. After application of weighting functions, determined from Monte Carlo data and needed to achieve compensation, the reconstructed measured energies agree with simulation to about 3%. The energies of hadronic showers are reconstructed with a resolution of about 50%/~E ® 2%. This result is achieved by inclusion of signals from an iron streamer tube tail catcher behind the liquid argon stacks .

500 Hl Calorimeter Group I Results from pion calibration runs for Hl

1 . Introduction in section 4. In section 5 we discuss the energy recon- During the years 1989 and 1990 several calibration struction first on the e.m. scale defined by electron runs with different modules of the H 1 liquid argon calibration and then on the hadronic energy scale. (LAO calorimeter were performed at the H6 beam Event selection criteria are given in section 6 . Experi- at the CERN SPS. The main goals ofthese runs were mental data and simulation are compared in section 7 to provide an energy calibration for electrons [ 1 ] in for several calorimeter modules in various test config- the H 1 detector [2 ], to study electron-hadron separa- urations. Results are presented for the total energy on tion [3 ] and to determine energy calibration functions the e.m . scale and after full reconstruction of weighted for hadrons andjets. hadronic energy, the achieved energy resolutions, the In previous papers [4,5 ] we reported on the perfor- effective electron-to-pion signal ratio, the longitudinal mance ofprototype LAr test modules . Here we present and lateral shower profiles and on signal distributions results achieved with final modules actually used in the in single channels . H 1 experiment at HERA- The emphasis is put on the comparison of pion signals from calibration measure- 2. Hl liquid argon calorimeter ments with corresponding simulations. This includes The H1 LAr calorimeter at HERA covers an an- results for the fully reconstructed pion energy obtained gular range of 4° ~ 9 ;~ 154°, where 0 is measured with the H 1 standard energy reconstruction code as with respect to the proton axis at HERA (fig. 1) . The used in the actual experiment, running on experimen- calorimeter consists of 8 wheels which are divided in tal and simulated signals from test beam events with- azimuth into 8 modules (octants) in the barrel and out any further change . into 2 modules in the forward region . The modules The Hl calorimeter is non-compensating (i.e. the are divided into an inner, e.m. lead/LAr stack (EMC) signals from electrons are on average higher than those and an outer, hadronic steel/LAr stack (HAC) . The from pions of the same energy), but its high segmen- depth of the EMC varies between 30 radiation lengths in forward 20X0 tation allows to distinguish electromagnetic (e.m. ) (X0) the and in the central barrel part corresponding 1 .4 1 .0 and hadronic shower components and to reconstruct to and interaction lengths (A) the hadronic energies by applying weighting func- for hadrons . The total depth for hadrons including the tions (as discussed first in ref. [6] ) to the measured EMC is in the forward region of IF and FB/OF 6 and and .1 in barrel (CB) . signals . The determination of such functions mainly 8A respectively near 5 the region The calorimeter highly segmented in both the from experimental data has already been studied by is e.m. and hadronic sections of the modules with a total the H1 Calorimeter Group [4,5,7-9] .The quality of of around 45 000 geometric cells in a quasi projective the Monte Carlo predictions for pion test data in the geometry (fig. 1) . The EMC has a 3 or 4 fold longi- H 1 calorimeters [ 10-14 ] now allows for extensive use tudinal segmentation while the HAC has 4 to 6 longi- of simulations for the determination ofthe parameters ofthese functions [ 111 . In this paper we first compare in various module configurations simulation and data for the mean total signal and other quantities char- FB ('B acterizing the details of the shower development . We _.. nom G.111111~~ . then study the deviations on the total hadronic energy "..~ 11111 o weighting AME reconstructed after having applied the same ma-1=2 a-mi-M-jaa.,lm1! 11! _ :. .. 1111 NI !i functions to experimental and simulated data. e~_ °-sE k~ The outline ofthe paper is as follows: In sections 2 and 3 we briefly review the main features of the H 1 test LAr calorimeter and the experimental setup in the [rit-ction Point beam including the iron streamer tube tail catcher . The Monte Carlo simulation procedure is described ____: 111ÎÎ11Î1Î " 1111111111 1IÏ1111ÏI1IIiii0 4- M M 11 ~ ~."0 ElmElm~ t Deceased. 1 . Supported by the Bundesministerium für Forschung rin and Technologie, Germany, under contract numbers 6AC17P, 6AC47P, 6DO57I, 6HH27I, 6MP17I, and Fig . 1 . Schematic view of the wheel and cell structure of the 6WT87P, respectively, HI liquid argon calorimeter with inner forward (IF) and Supported by the UK Science and Engineering Research outer forward (OF) wheels, the forward and central barrel Council. wheels (FB and CB respectively) and the backward barrel t Supported in part by USDOE grant e.m . wheel (BBE). The lines from the interaction point indi- DEF603 91 ER40674 . cate the directions of flight of the particles in the test beam tt Supported by the Swiss National Science Foundation . setup for the different calibration modules discussed here .

HI Calorimeter Group /Resultsfrom pion calibration runsfor Hl 50 1

tudinal segments. The lateral cell size in the forward region corresponds to about 1 Moli6re radius in the Iron e.m. and 1A in the hadronic sections; the cells in the barrel region are typically twice this size. Beam The H 1 LAr calorimeter is described in much more detail in refs. [ 2,15 ] . Vertically Moveable 'Cable Finger Counter 2 (width 3 cm) 3. Experimental setup program with electrons, pions and MWPC2 (25x25 ein') The calibration - ® ® ' Hole Counter 100 x 20 CMZ, hole ~5 6 em muons was performed in the H6 beam [ 16 ] at the CERN SPS. A survey of the setup at the test beam ®I® is shown in fig. 2. The good homogeneity and stability of LAr calorimeters allows to calibrate single rep- liiiiiiiiiiiiiiiiiiiiillillillillillI resentative module configurations rather than each individual module of each calorimeter wheel. 3.1. Liquid argon calorimeter configurations In total 8 characteristic module configurations of MWE the H 1 LAr calorimeter (section 2) were tested in sep- arate runs. In most cases the same modules were used as those later installed in the H1 cryostat at HERA. The configurations considered for this report are the inner forward calorimeter (IF, 0 .: 10'), parts of the forward barrel and outer forward calorimeter (FB/OF, 0 25'), a section of the forward barrel alone (FB, 0 34'), and a part of the central barrel (CB, 0 101 °) region. The wheel BBE is ofEMC type only and it is not used for the results reported here. The modules were mounted into the cryostat such that the impact angle of the beam corresponds to the angle 0 of a particle coming from the vertex of the e-p collisions at HERA (fig. 1) . The cryostat could house two com- plete H 1 modules with the exception of IF, where an extra module ofhalfthe transverse size, but otherwise Fig. 2. Schematic view on the setup in the test area at the H6 identical, was built for the purpose ofcalibration. beam (dimensions in cm) . The purity of the LAr was constantly monitored by probes identical to those in H1, positioned in the cryo- The charge calibration of the readout system is stat close to the calorimeter modules and consisting of performed via capacitors of 47 pF (selected to within a LAr ionisation chamber where the cathode is coated f 1 %) which are charged by voltage pulses of known with a 207Bi source. For more details see ref. [ 15 ] . amplitude . Two systems are available : one uses calibration capacitors in the liquid argon directly at the 3.2. Electronics and calibration stacks (cold calibration), the other at the preampli- The front end electronics and calibration system fier level (warm calibration) . The latter was used as was as close as possible to the present H 1 system at a backup system only. The observed relation of ADC HERA (described in more detail in refs. [15,17]) . The signal lines response vs. injected charge for each channel is rep- of each calorimeter cell are fed out resented by a polynomial fit of third order which is of the cryostat into an analog signal processing chain consisting ofa warm charge sensitive preamplifier used in the off line data analysis. The overall precision and of the calibration was 0.5%. The stability depended a shaping amplifier yielding a bipolar signal (peak at 2.4 tus) which is strobed hold mainly on temperature variations which made it nec- into a sample and unit essary to calibrate at least once in 24 hours. upon trigger. The signals feed the analog readout system after multiplexing by a factor 128. Two different 3.3. Iron streamer tube calorimeter gains are used to extend the dynamic range of a 12 bit A warm iron-gascounter calorimeter (also called ADC to 14bit for halfofthe 2048 calorimeterchannels "tail catcher") with a depth of4.5R located behind the available at the test. The ADC data were read out by LAr modules allows for the measurement of hadronic a fast CAB processor as described in ref. [4 ] . No zero shower tails and punch through pions. It consists of suppression nor data corrections were applied online. 10 iron plates (area 2.7 x 2.7 m2, thickness 7.5 cm)

502 HI Calorimeter Group /Results from pion calibration runs for HI

interleaved with 10 streamer tube layers. In addition, per simulation step dx in the liquid argon is corrected three layers are mounted in front as well as behind the by Birks' Law [23] : iron structure . The detector planes are equipped with 2) dE/dx read-out pads (typical size 30 x 30 cm or strips . The dE dx = ( 1+kBdx digital read out of the wires (pitch 1 cm) and of the orthogonal strip electrodes (width 2 cm) serve for pat- where dE' is the corrected deposited energy. A factor tern recognition . The analog read out of the pad tow- kB = 0.005 g MeV - ' cm-2 was assumed (compare ers (two longitudinal segments of 5 pads/front-tower ref. [24] ) . and 6 pads/back- tower, respectively) gives the energy Each event is overlaid by an experimental empty measurement . The calorimeter has been calibrated in event from a random trigger to include the effects of a stand-alone mode in the same beam line as the LAr electronic noise (section 5.2) . calorimeter, using incident pions and muons. It is lin- For physics analyses H1 uses a faster simulation ear up to 40 GeV, the resolution amounts to Q/E _ with parametrized e.m. showers [25] and hadron 100%/ E (GeV) (for details see refs . [18,19]) . terminators [ 14 ] . The results for the simulation of pion beam data are practically identical to those of the 3.4. Beam detailed simulation discussed here [ 14,26 ] . The setup of the H6 beam line [ 16 ] , the trigger elements and the beam counters are practically the same 5. Energy reconstruction in the liquid argon as in previous 1986-88 test runs [4] . The main beam calorimeter defining elements are two Cherenkov counters used The energy reconstruction in the H1 LAr calorime- to select either electrons or pions and a set of multi- ter involves several levels of charge corrections for ex- wire proportional chambers (MWPC's) to measure perimental data [ 1,15 ] andchannel selection and clus- the horizontal and vertical beam positions. Typical ter finding algorithms for data and Monte Carlo. The measured values for the horizontal and vertical beam charges and the energies deposited in the LAr for data widths are Q :z~ 0.8 cm for the lower beam energies and simulation respectively are first reconstructed cell around 5-10 GeV and a ;:z~ 0.3 cm for 80 GeV; the mo- by cell on an electromagnetic (e .m.) energy scale (see mentum spread is 0.5% . The systematic uncertainty below) . This procedure is independent of the particle ofthe beam energy scale is 15%/p (GeV) ® 0.5% #1 . type and is the same fortest beam electrons or pions, or The phase space of the particles is mainly defined by for jets at HERA . On this energy scale the experimental two narrow scintillation counters limiting the accep- and simulated calorimeter response are directly com- 2. tance for the beam to an area of 3 x 3 cm More de- parable . In addition, all further energy corrections and tailed descriptions of the setups for the discussed run applied weighting functions are exactly the same for periods can be found in refs . [ 1,11,13,19,20 ] measured and simulated signals . Inthe following we will briefly describe the most im- 4. Simulations portant steps of the energy reconstruction for hadrons; The calibration measurements have been simu- further details follow in ref. [2] . lated in the GEANT 3.14 [ 21 ] framework generating hadronic showers by GHEISHA 8 [22 ] . Electrons and 5.1. The electromagnetic energy scale photons produced in the hadronic showers are simu- The transformation of signals in the calorimeter to lated with the default e.m. shower code of GEANT. the e.m. energy scale is performed for experimental The simulation program contains a detailed de- scription ofthe LAr calorimeter geometry and the gap Table 1 and absorber structure which is identical to the stan- The electron calibration constants used for experimental (cexp) and for simulated signals for different wheels dard H 1 detector simulation . In addition it includes (c,,,,,) of the H 1 LAr calorimeter . isjust the inverse ofthe sam- line, the cryostat and inactive csi, elements of the beam pling fraction of electrons and therefore dimensionless material in front and behind the calibration mod- central position of the beam, the spatial ules . Also the Calorimeter EMC HAC widths perpendicular to the beam axis and the known wheel name cex p cexp csim momentum spread (section 3.4) are simulated . cg,n, The thresholds for the tracking of particles in the [GeV/pC] [GeV/pC] simulation were set to 1 MeV for electrons/positrons, hadrons and to 0.2 MeV for photons . CB 3.58 12 .68 7.58 27.14 Recombination effects were taken into account in FB 3.41 12 .76 6.70 25.23 the calculation of the signal . The energy dE deposited IF 3.57 12 .73 7.23 26.82 OF - - 6.70 24.93 11 N. Doble, private communication (1991) .

HI Calorimeter Group / Results from pion calibration runsfor HI 503

data by applying an electron calibration constant to rection has to be applied to the signal obtained on the the charge signal on the cell level. This constant has e.m. scale. been determined for the EMC sections by calibration The aim is to equalize the response to the e.m. and measurements with electrons [ 1 ] . Corrections for pure hadronic components of a hadronic shower and, dead materials in front of the calorimeter or analysis therefore, to suppress the influence ofthe large fluctua- cuts have been applied using detailed Monte Carlo tions in the hadronic shower composition ontherecon- simulation. For the hadronic sections, where no direct structed energy. Thetechnique exploits the fact that lo- electron calibration measurements are yet available, cal energy deposits of high density are mainly of elec- the e.m. calibration constants are derived from the tromagnetic origin while the hadronic component is ones measured in the e.m. sections, correcting for the much more spread out. Thus, in a well segmented calo- differences in the sampling fractions by detailed elec- rimeterthe amount of energy deposited in the cells can tron shower simulation using GEANT and EGS 4 [27] be used for statistical separation of e.m. and hadronic (compare ref. [28] ) . energy depositions for which different correction fac- In case of simulation the energies deposited in the tors are needed. LAr layers are converted by multiplication with the ef- Signal weighting techniques to get the proper fective inverse sampling fraction for simulated elec- hadronic energy scale have already been studied for trons. single pions in test beams [4-6] . The reconstruction The calibration constants used are given in table 1 . of e-p events at HERA, however, requires a good en- The present uncertainty is ;~ 3%. The main error source ergy measurement also forjets in a wide energy range. is due to impurities in the LAr and the determination Results ofa first study of the application of weighting ofthe charge collection efficiency. For more details on to signals from simulated jets have been presented the definition of the e.m. scale and the determination in refs. [7,8] . In the present approach the weighting of the electron calibration constants see refs. [ 1,29 ] . functions are derived by simulation ofjets in the full H l detector thereby making use in the reconstruc- 5.2. Noise suppression tion of clustering of cells as described below. In the The typical electronic noise of a single channel is remaining part ofthis chapter the signals are assumed about 3.5 x 10° electrons, equivalent to 20 (40) MeV to be already reconstructed on the e.m. energy scale on the e.m. energy scale, in the EMC (HAC) sections. (section 5 .1 ) . A channel selection based on the energy signal on this scale is performed to reduce the possible noise contri- 5.3 .1. Clustering butions to the total signal. The selection is identical for All cells passing the noise suppressing step (sec- experimental and simulated events. To the latter real- tion 5.2) are subject to clustering which is based on istic electronic noise was added by overlaying random the correlation between signals in neighbouring cells. trigger events. The following cuts are also applied in The aim is the formation ofgroups ofcells with signals the general Hl reconstruction code [2]: corresponding to particle showers. The algorithms A certain channel i contributes to the signal of an used are tuned such that the cells containing energy event only ifits absolute energy signal I Eô J on the e.m. depositions from an individual e.m. shower, initiated scale passes the condition : by a photon or electron in the jet, are most probably ~EO'J > 4EQ or (JEô1 > 2EQ and Eô > 4EQ) , (1) collected into one cluster. Hadronic clusters, on the other hand, with their large spatial fluctuations are in where EQ, EQ are the energy equivalents of the elec- general split into several clusters [30] . tronic noise of channels i, j, and j is a channel adja- Each cluster ofcells with total energy above 1 GeV cent in space to channel i. The EQ are standard de- is classified to be either of e.m. or hadronic nature. viations of the noise distributions on the e.m. energy The former are identified using characteristics like scale obtained by random events. Note that E,, > 0 al- the fraction ofcluster energy deposited in EMC (con- ways, while the energy signal Eô might be positive or tainment in EMC), in the first layer of EMC (early negative, in the latter case representing a pure noise shower development), and in the four most energetic contribution. Thus only weak thresholds are applied cells ofa cluster (compactness), see ref. [3] . Clusters around a significantsignal seed and statistical compen- above 1 GeV which are not recognized as of e.m. type sation ofpositive and negative noise contributions is are called hadronic. Below 1 GeV clusters with sig- achieved. nificance E(Eô/E,) z > 8 and developing deeply inside the calorimeter are called hadronic as well. 5.3. Hadronic energy reconstruction Since in the H 1 LAr calorimeter the response to 5.3 .2. Weightingfunction application hadrons is typically 30% smaller than that for elec- Electromagnetic clusters are regarded as being cal- trons of the same incident energy, an additional cor- ibrated on the e.m. energy scale while signals in cells

504 HI Calorimeter Group / Results from pion calibration runsfor HI

of hadronic objects have to be weighted. The latter are For those results where the iron streamer tube tail formed by cells which are not included into an e.m. catcher is included, we veto against punch through by cluster and located in a tube around a hadronic cluster requiring to observe no digital hits in the last 3 layers (r < 50 cm in HAC and r < 25 cm in EMC) . ofthe tail catcher. They are located behind its last iron The functions used to calculate the weights applied plate, i.e. behind 10.5 interaction lengths in total. to signals in these cells are qualitatively the same as At lower energies, Ebeam < 7 GeV, the muon con- introduced in refs. [ 7,8 ], the weighted energy E,e, in a tamination ofthe test beam can not be neglected either. cell i is calculated by: Here the tail catcher and parts of the veto system behind the tail catcher are usedto eliminatethese muons. a, (-aEô/V' ) } Eâ, (2) E,'ec = {ao + exp These selections lead ofcourse to a suppression ofsuch where Eô is the energy on the e.m. scale in this cell, V` events in the data sample, where the pions are passing the volume ofthe cell, and ao, a, and a are the param- the LAr calorimeter with a late or no inelastic interac- eters of the weighting function. tion. The parameters ao, a, and a have to be determined 7. Experimental results and comparisons with for each calorimeter type (EMC, HAC). Two ap- simulations proaches have been studied to determine the weight- 7.1. Total energy signals in different calorimeter ing parameters as functions of quantities which are modules directly measurable inside the calorimeter alone. One Total energies are calculated by summing over the possibility is to use simulated single pion response to cells of the EMC and HAC sections of the LAr calori- determine the weighting parameters in different calo- meter, where each single channel has to pass the noise rimeter wheels as functions of the signal ofa hadronic cut given in eq. (1) . Electronic noise is also included cluster [ 11 ] . The other approach is to use the response in case ofsimulation (section 5 .2) . Results on the e.m. to the hadrons in simulated jets to determine the pa- scale (Eo ) and hadronic energy scale (E~e. ) are shown rameters as functions of the reconstructed jet energy, in fig. 3 for experimental and simulated pions at the calculated iteratively inside cones of 10° opening an- lowest energies available in the IF and FB/OF calori- gle, as seen from the vertex. The results of the latter meter modules. The distributions agree quite well, be- approach are applied in this paper. At low energies, sides somewhat larger tails towards high energies in below 7 GeV, the ansatz in eq. (2) is replaced by sim- case ofsimulation. Examples for pions at higher beam ple multiplicative factors corresponding to effective energies in various calorimeter regions are shown in e/7t ratios in EMC and HAC. In the region 7-10 GeV fig. 4. The tails of the distributions to lower energies both methods contribute to the correction in order are due to pions with late or no inelastic interaction to get a smooth transition from the simple correction in the calorimeter . They are already at incident ener- factors to the weighting according eq. (2). gies as low as 30 GeV more enhanced in the CB mod- We mention again, that there is no difference in the weighted energy reconstruction for jets and single hadrons or for simulated and experimental data, the same functions and parameters are used in all cases. The reconstructed energies for single pions shown in section 7 are obtained by running the above described standard H1 reconstruction algorithm -including the filtering of e.m. clusters - on the experimental and simulated signals. 6. Event selection Pions are selected by Cherenkov counters in the beam line (section 3) . For beam energies Ebeam >_ 10 GeV all events are accepted if their total signal Eo on the e.m. energy scale fulfills Eo ~ (2-4) GeV, depend- ing on the depth ofthe actual module. This suppresses contributions from possible beam muons in case ofex- 0 2 4 6 0 2 4 6 8 perimental data; the cut is, however, applied to simu- 1:' [Gey] E [(1cV[ lated data as well. No additional cut is applied against Fig. 3. Distributions of the energy on the electromagnetic events with longitudinal energy leakage to avoid in scale Eo (o simulations, histograms experiment) and ofthe the calibration at high incident energies a bias to- reconstructed energy E.e. after weighting (e simulations, wards hadronic showers with a large electromagnetic shaded histograms experiment) for the lowest available pion content. energies at the test beam.

HI Calorimeter Group / Resultsfrom pion calibration runs for HI 505

00 400 800 1200 1800 E [GeV] ~7 [~1 7J T le 1 dN rI 1 lI IF NJE 12 1 0.16 0.16 0 15 ~ 80 GeV . . . 0.14 0.14

0.12 0.12 0.1 0.1 0 10

0.08 0.08

0.06 0.06

0 .04 0.04 0.05 0 .02 0.02

0 0 40 80 0 100 E [GeV] E [( ;oV] 000 0 0 500 100.0 1500 2000 2500 Fig. 4. Distributions of the energy on the electromagnetic E [GeV] scale Ep (o simulations, histograms experiment) and of the reconstructed energy Ere, after weighting (e simulations, Fig. 6. Energy reconstruction for pions at 120 and 205 GeV shaded histograms experiment) at various incident energies for LAr module IF with tail catcher (hatched histogram) in different calorimeter modules. and IF only (o) . Solid line: fitted Gaussian distribution .

o FB/OF o IF 9 CB

25 50 75 100 125 150 175 200

Ebea .a [GeV]

10 '~ o FB/OF o IF e CB 5 ......

0 -_._._ ._._ ._____ ._._ ._x_._._ ._ ._._-._ ._._ 4 0 . . . . . ---- . . . .0 ...... _ . . . ._ ......

0 25 50 75 100 125 150 175 200

Ebea,n [GeV] im P Fig. 5. The relative difference between simulated (EÔ ) and experimental (Eo' ) pion signals on the electromagnetic energy scale (top) and after weighting (Eëm,E,P) (bottom), in three different calorimeter regions.

506 HI Calorimeter Group /Resultsfrom pion calibration runsfor HI

ule which is shorter (total depth d Pz~ 5R) than other (T(E)1E[%o] modules such as FB (d ~ 8A) . 40 based on fitted Gaussian distributions, for simu- 35

lated and experimental pions as function ofthe beam 30 -~ energy given in fig. The maximal is 5. deviations are 25 n . energy < 3% on the e.m and < 5% on the hadronic 1 0 scale. They are mainly systematic. A comparison of 15 the results the different agreement of stacks shows 0 better than ±2%. If the results of the different stacks 5 are averaged at a given beam energy, the deviations of simulation from experimental data are in the or- 25 50 75 100 L 125 150 175 - 200 225 der of 3%. The reconstruction code applied to simulated events reproduces, under the condition of re- energy Fig. 7. Measured energy resolutionforpions in the IF module alistic electronic noise and analysis cuts, the (o) as function ofthe beam energy. The reconstruction at the deposited in the calorimeter to better than ±2%. Fur- higher energies includes signals from the tail catcher (o) . ther studies of the calibration with the full detector at HERA are in progress [2 ]. The above results are based on the LAr calorimeter to be possible [ 31 ] . only. The inclusion ofthe tail catcher (section 3.3) is The measured energy resolution as a function ofthe essential at high energies. This is visible in fig. 6 which beam energy is shown in fig. 7. The data are fully re- shows the reconstruction at pion energies of 120 and constructed on the hadronic energy scale (section 5.3) . 205 GeV with the LAr calorimeter alone (6A) and to- The signals from the tail catcher, which at HERA is in- gether with the tail catcher (10.51 in total) . tegrated into the instrumented iron muon system [2], 80 7.2. Hadromc energy resolution have been included at beam energies Ebeam > GeV The energy resolution for pions is determined by to recover energies leaking out ofthe liquid argon calo- fitting a Gaussian function to the distribution of fully rimeter. The curve results from a fit to the data points, using the resolution function reconstructed energies Erec, as for example shown in fig. 4, thereby excluding the low energy tails. The a (E) a2 b2 simulated distributions generally show larger fluctu- _ + + c2 ations than the experimental data. This effect can be E Tb -4 Ebeam observed in all the different calorimeter modules. Ta- ble 2 gives the direct comparison on the e.m. scale for with a stochastic term al v/E-b a,,,, a noise term b/Ebeam the IF stack and shows that the resolution for simula- and a constant term c. This fit yields tion is on average worse by 7% than the measured one. a = These systematic effects are under study and some (50.7±0.7)% GeV, improvement of GHEISHA at high energies appears b = (945±32) MeV, c = (1 .6±0.1)%, Table 2 Standard deviations of Gaussian distributions fitted to the which can be regarded as typical values for the whole total energy on the e.m. scale in the IF module for data and calorimeter . As c - (e/7r -1) [32], the relatively small simulation with GHEISHA 8 (statistical errors) constant term in eq. (3) shows the good compensation achieved with the weighting functions. Electromagnetic scale Ebeam 7.3. Effective electron-to-pion signal ratio in the liquid [GeV] ae,,p [GeV] uim [GeV] argon calorimeter The energy Eo reconstructed on the e.m. scale is a 5 0.93±0.02 1 .01±0.02 very good estimate ofthe energy Edep deposited in the 10 1.39 ± 0.03 1 .39 ± 0.02 calorimeter in case ofelectrons, however it is on aver- 30 3.25 ± 0.04 3.46 ± 0.04 age smaller than Edep orEbeam in case ofincidentpions. 50 5.08 ± 0.07 5.48 ± 0.09 The electron-to-pion signal ratio on the e.m. scale can be expressed by: 80 7.80±0.11 8.37±0.14 120 11 .06±0.16 11 .94±0.35 e - (Edep) . (4) 170 15.63±0.28 17.19±0.47 (Eo) 205 18.42 ± 0.23 19.73 ± 1 .10 As Edep is known exactly only for Monte Carlo events, we have calculated an effective (e/n)err for different

HI Calorimeter Group /Resultsfrom pion calibration runs for HI 507

different dependence on Ebeam than the deeper IF and Eb_/ (Eo) (Ed, y) / (Eo) FB/OF. This is due to longitudinal energy leakage, which occurs in case of the CB module already signif- ,7 ;H L Hs/o1 o

13 t 13 dependence of (e/n)esr.

12 The quantity e/n, as given in eq. (4), has also been calculated correcting for leakage by simulation, for which the deposited energy is known event by event 40 80 120 0 40 80 120 (fig. 8b) . The resulting e/n is falling with increasing Ebea, [GeV] isb-, [GeV] particle energy, as expected, and is very similar for all the studied modules. The curves in fig. 8b are fits with Fig. 8. The effective electron-to-pion signal ratio calculated functions falling linearly with increasing log Ebeam. as given in eq. (5) for experimental pions, as function ofthe This behaviour corresponds to a rise of the intrin- energy in calorimeter regions (a). Pic- beam Ebeam different sic electromagnetic fraction fem - log Ebeam of the ture (b) shows e/n corrected by Monte Carlo for the effect hadronic [5,33] of limited acceptance and longitudinal energy leakage. The shower . curves show results from fits, as discussed in the text. 7.4. Shower profiles Longitudinal and transverse shower profiles have calorimeter wheels and pion energies Ebeam been measured at different beam energiesand are com- e _ Ebeam pared with Monte Carlo simulation. The IF and CB n etr (Eo) calorimeter modules are especially well suited for this study because of the approximately perpendicular im- This quantity overestimates the e/n ratio, since it is pact direction of the beam particles into these stacks. affected by energy losses due to material in front of Fig. 9 shows the average relative energy deposit, the calorimeters, limited lateral acceptance of the cal- measured on the e.m. energy scale and normalized to ibration modules and, especially at high beam ener- the depth of each longitudinal segment, as a function gies, by longitudinal energy leakage. Measured val- of the calorimeter depth for pions at different beam ues of (e/7r )eff are shown in fig. 8a. Simulations yield energies in the IF and CB modules. In general the similar results for (e/n)esr on the level of agreement simulation follows quite well the experimental data. observed in fig. 5 . The shorter CB module shows a Some discrepancies are visible in the first segments of the hadronic stacks. I/ (E0) dEo/dt [I/Al - IF IF IF CB CB CB 0.6 5 GeV 30 GeV 80 GeV 10 GeV 30 GeV 80 GeV

0.5 10 OO 0.4

0 0 0.3

0.2

0 0 2 4 6 0 2 4 6 0 2 4 6 0 2 4 6 0 2 4 6 0 2 4 6 t [A] t [A] t [a] t [a] t [A] t [A] EMC ~2 HAJ Fig. 9. The longitudinal shower profile for experimental (e) and simulated (shaded histograms) pions at three different beam energies in IF and CB. Shown is the average relative energy deposition dEo/ (EO), as measured on the e.m. scale per unit absorption length A, as a function ofthe calorimeter depth t.

508 HI Calorimeter Group /Results from pion calibration runs for H1

1/ (Ea) dEo/dr[1/(0 .1 A)]

Fig. 10. The transverse shower profiles for pions at different energies in the IF module, experiment (9) and simulation (shaded histograms). The average relative energy deposition dEp/ (EO), measured on the e.m. scale and integratedover thewhole shower depth, is shown as function of the distance r from the shower axis.

The transverse profiles for experimental and simu- ofthe energies Eô over several orders of magnitude in lated pions at 5, 30 and 80 GeV in the IF module are probability. This is important if, as in the present en- compared in fig. 10. The average relative energy loss, ergy reconstruction scheme for the H1 LAr calorime- integrated over the whole shower depth, is shown as ter, the determination ofweighting functions is based functionofthe perpendicular distance fromthe shower on simulation of hadrons or jets. The weights applied axis, which is measured for each event by determin- in the hadronic energy reconstruction (section 5 .3) to ing the impact point at the front face of the calorime- the signal energies Eô are functions of these energies ter and the center ofgravity ofthe shower. The lateral themselves. acceptance ofthe module is limited to about 2A. 8. Conclusions 7.5. Signals in single channels Detailed comparisons of signals from pions in H 1 In fig. 11 we compare fluctuations of energies Eô liquid argon calorimeter modules and simulation in of individual cells averaged overthe FB/OF calorime- the LEANT/GHEISHA frame have been presented ter modules at pion beam energies of 3.7, 30 and 80 on the electromagnetic energy scale which is deter- GeV. The simulation follows very well thedistribution mined from electron measurements. The total signal

~ V dr 7T-Gev ~-âF ~i1~v~

- 1 - 1 10 10

-2 -2 10 10

-3 -3 10 10

0 1 2 0 10 20 E, [GeV] E,' [LeV] Fig. 11 . The distributions of signals Eô in single cells for experimental (e) and simulated (shaded histograms) pions in the FB/OF region, at different beam energies.

HI Calorimeter Group /Resultsfrom pion calibration runsforH1 509

for pions is described within an uncertainty of < 3% [9] J. Marks, DESY F21-90-01 (1990), Thesis, University for beam energies between 3.7 and 205 GeV. The lon- of Hamburg (1989) . gitudinal and lateral shower spreads are well repro- [10] J. Gayler, in: MC91, Workshop of Detector and Event duced by the simulation. The same is true for energy Simulation, eds. K. Bos and B van Eijk, NIKHEF-H, fluctuations in single cells. Amsterdam (1991) p. 312. Loch, Thesis, University of Hamburg (1992) . Full reconstruction of the energy deposited by sin- ,,1, P. [12] J. 2a~ek et al., H1 Internal report, Hl-04/92-220 gle pions in the calorimeter can be achieved by signal (1992) ; and weighting functions determined from simulated jets. E. Sânchez, H1 Internal report, H1-07/92-232 (1992) . Here the deviations of experimental and simulated [ 131 M. Flieser, MPI-Phe/92-08, Diploma Thesis, Technical energies are about 3%. An energy resolution of about University of Miinchen (1992) . 50%/ V/-E (GeV) with a constant term below 2% is [14] M. Rudowicz, MPI-PhE/92-14, Thesis, University of achieved. Hamburg, 1992. [ 151 H 1 Calorimeter Group, B. Andrieu et al., DESY Acknowledgement preprint 93-078 (1993), to be published in Nucl. Instr. We appreciate the big efforts of all the technical and Meth. collaborators who constructed and maintained the [16]P. Coet, CERN/EBS note 85-14 (1985) ; CERN calorimeters . The support of the CERN staff operat- preprint, SPS/EPB/PC (1981) . ing the SPS, the H6 beam line and the computer facil- [ 17] H1 Collaboration, C. Berger et al., DESY-report PRC 86-02, Hamburg (1986) . ities is gratefully acknowledged . We thank all funding [ 181 H 1 Muon Group, H. Bergstein et al., H 1 internal report, for financial support. Finally we would like agencies H1-10/91-197 (1991),andibid, H1-O1 /93-261 (1993) . to thank H. Fesefeldt for useful discussions. [ 19] H. Bergstein, Thesis, RWTH Aachen (1993) . [201 J.-F. Laporte, Thesis, University ofParis-Sud (1991) ; References R. Haydar, Thesis, University ofParis-Sud (1991) ; and [I] H1 Calorimeter Group, Calibration of the H 1 liquid R. Grässler, Diploma Thesis, RWTH Aachen (1991) . argon calorimeters by electrons, in preparation. [21 ] R. Brun et al., CERN DD/EE/84-1 (1987) . [2] H1 Collaboration, The Hl-detector HERA, DESY [22] H. Fesefeldt, PITHA-Report 85-02, RWTH Aachen preprint 93-103 (1993), to be submitted to Nucl. Instr. (1985) . and Meth. [23] J.B. Birks, Proc. Phys. Soc. 64 (1951) 874. [ 3 ] H 1 Calorimeter Group, B. Andrieu et al., Proc. 3rd Int. [24] C.W. Fabjan et al., Nucl. Instr. and Meth. 141 (1977) Conf. on Advanced Technol. and Part. Phys., Como, 61; and Italy, June 1992, to be publishedin Nucl. Phys. B (Proc. J. Brau and T.A. Gabriel, Nucl. Instr. and Meth. A 238 Suppl. ). (1985) 489. [4] H1 Calorimeter Group, W. Braunschweig et al., Nucl. [25] S. Peters, MPI-PhE/92-13, Thesis, University of Instr. and Meth. A 265 (1988) 419; Hamburg (1992) . H 1 Calorimeter Group, W. Braunschweig et al., Nucl. [26] M. Kuhlen, in: Proc. 26th Int. Conf. on High Energy Instr. and Meth. A 275 (1989) 246. Physics, Dallas, 1992, ed. J.R. Sanford, vol. 2, p. 1787. [ 5 ] H 1 Calorimeter Group, W. Braunschweig et al., DESY [27] R.L. Ford and W.R. Nelson, SLAC-Report-210; and preprint 89-022 (1989) . SLAC preprint, EUN-SLAC-0 . [6] J.P. Dishaw et al., Phys. Lett. B 85 (1979) 142; [28] R. Tamoschat, Diploma Thesis, University of Dort- J.P. Dishaw, SLAC-Report 216 (1979) ; and mund (1992) . CDHS Collaboration, H. Abramowicz etal.,Nucl. Instr. [29] J. Gayler, H. Kiister and P. Loch, H1 Internal report, and Meth. 180 (1981) 429. Hl-04/91-171 (1991) . [7] H 1 Calorimeter Group, H. Jung et al., contributed [30 ] L. Gdrlich and H.P. Wellisch, H 1 Internal report, H1- paper to the XXV Int. Conf. on High Energy Physics, 12/91-204 (1991) . Singapore (1990) . [31 ] A. Wagener, PITHA-Report 92/37, Diploma Thesis, [8] H. Greif, MPI-PAE/Exp.El. 229 (1990), Thesis, RWTH Aachen (1992) . Technical University of Miinchen (1990) . [321 R. Wigmans, Nucl. Instr. and Meth. A 259 (1987) 389. [33] C.W. Fabjan, CERN preprint, EP/85-54 (1985) .