ARTICLE IN PRESS

0168-9002/$ - se

doi:10.1016/j.ni

�Correspondi

E-mail addre1On leave fr

Romania.2Also at St.P3Now at ITC

Nuclear Instruments and Methods in Physics Research A 539 (2005) 250–261

www.elsevier.com/locate/nima

Results from beam tests of large area silicon drift detectors

E. Crescioa,�, M. Bondilab,1, P. Cerelloa, P. Giubellinoa, A. Kolozhvaria,2,S. Kouchpilc, G. Mazzaa, L.M. Montanoa,d, D. Nouaisa, S. Pianoe, C. Piemontee,3,A. Rashevskye, L. Riccatia, A. Rivettia, F. Toselloa, A. Vacchie, R. Wheadona

aTorino University, INFN—Sezione di Torino, Via Pietro Giuria 1, Torino 10125, ItalybJyvaskyla University, Finland

cNPI, Rez, Czech RepublicdCINVESTAV, Mexico City, Mexico

eINFN Sezione di Trieste, Italy

Received 27 September 2004; accepted 7 October 2004

Available online 23 December 2004

Abstract

Silicon drift detectors with an active area of 7:0 � 7:5 cm2 will equip the two middle layers of the Inner Tracking

System of the ALICE experiment. The performance of several prototypes was studied during beam tests carried out at

the CERN SPS facility. In this paper, the results obtained from the data taken during August 2000 will be presented.

The spatial resolution was studied in the case of tracks perpendicular to the detector and in the case of inclined tracks.

Results on the signal analysis and of the double-track resolution are also shown.

r 2004 Elsevier B.V. All rights reserved.

PACS: 29.40.Wk; 29.40.Gx; 07.05.Kf

Keywords: Silicon drift detectors; Beam test; Resolution

1. Introduction

The two middle layers of the Inner TrackingSystem (ITS) of the ALICE experiment [1] will be

e front matter r 2004 Elsevier B.V. All rights reserve

ma.2004.10.006

ng author.

ss: crescio@to.infn.it (E. Crescio).

om Institute of Spatial Sciences, Bucharest,

etersburg University, St. Petersburg, Russia.

-irst, Trento, Italy.

equipped with Silicon Drift Detectors (SDDs)[2–4], which offer the high granularity requiredby the high particle density predicted in heavy-ionreactions and provide the energy-loss measure-ment needed for particle identification. Variousprototypes have been tested with particles since1997 at the CERN PS and SPS in dedicated beamtests, to evaluate the detector performance inactual experimental conditions. The results dis-cussed in this paper were obtained from the

d.

ARTICLE IN PRESS

E. Crescio et al. / Nuclear Instruments and Methods in Physics Research A 539 (2005) 250–261 251

analysis of the beam test data taken at the SPSduring August 2000. In Sections 2 and 3, the experi-mental details and the first steps of the analysisprocedure are described. The study of the detectorspatial resolution for perpendicular and inclinedtracks and the two-track resolution are discussedin Section 4.

2. Experimental details

2.1. The detectors

During the August 2000 beam test, one proto-type (named ALICE-D2) with characteristics closeto those of the final design for ALICE [5] wastested. The detector, produced by CanberraSemiconductors on a 300mm thick 500 NeutronTransmutation Doped (NTD) wafer with a resis-tivity of 3KO cm; has an active area of 7:02 �

7:53 cm2: The active area is split into two adjacent35 mm long drift regions, each equipped with 256collecting anodes (294mm pitch) and with inte-grated voltage dividers which deplete the detectorand generate the drift field [5]. Since the driftvelocity is very sensitive to the detector tempera-ture variation (v / T�2:4) the monitoring of thisquantity is performed by means of three rows of 33implanted point-like MOS charge injectors [6,7].Besides the D2 prototype, two prototypes from aprevious design, named D1B (with only smalldifferences with respect to the final one), were usedto provide extra tracking information, whereas themain aim of the test was to study the performanceof the D2 prototype.

2.2. Beam test setup

The beam test setup presents the same basicgeometrical structure of the beam tests performedsince 1997 [8]. The detectors under test were placedon the beam line within a telescope made of siliconstrip detectors, which was used for the trackreconstruction. The telescope consisted of fivepairs of single-sided silicon strip detectors withalternating orientations ðx; yÞ: Each microstripdetector covered an area of about 20 � 20 mm2;with the strip pitch of 50mm: The telescope was

aligned with two pairs of scintillating counters,placed at both ends of the setup. The coincidenceof a signal in all the counters was used as beamtrigger. The overlapping region between each pairof scintillators covered an area of about 20 �

20mm2: Since the SDD area was larger than thebeam spot and the coverage of the scintillatingcounters, the SDDs were mounted on a micro-metric XY-table in order to perform positionscans. The SDD support could also be rotated totake data with tracks at different incidence angles.For the study of the two-track resolution, a 9 mmthick iron target was placed upstream at a distanceof 8 cm from the D2 prototype. The test wasperformed using a pion beam at 140GeV=c:

2.3. Readout electronics

The readout electronics included the OLA [9]front-end chip, specifically designed for SDDs, avoltage amplifier/driver and a Flash A/D coverter(FADC). OLA is a 32-channel low-noise, bipolarVLSI chip. Each OLA channel consists of acharge-sensitive preamplifier, a semi-Gaussianshaping amplifier and a symmetrical line driver.The peaking time is 55 ns for a d-like input signaland the Equivalent Noise Charge in laboratoryconditions was measured to be �230e� at zerodetector capacitance. The OLA was followed by avoltage amplifier, originally developed for gas driftchambers [10], driving a 30 m long twisted-paircable connected to the flash-ADC which sampledthe signal at 40 MHz. The FADC system was theDL350 [11], developed by Laben and Struck. It isbased on 8-bit resolution FADC chips having aninput circuitry that expands the dynamic range to10 bits with a non-linear transfer function.

2.4. Data sets

Most of the data (3.2M events) were taken withsingle tracks with perpendicular incidence on theSDD. For reason of geometrical acceptance andacquisition speed, it was chosen to connect 72anodes from each detector to the FADCs.Whereas for the D1B detectors the same groupof 72 anodes was maintained for all data-taking,the connections to the D2 detector were changed

ARTICLE IN PRESS

E. Crescio et al. / Nuclear Instruments and Methods in Physics Research A 539 (2005) 250–261252

during the position scans in order to ensure thatthe performance of the complete half-detectorcould be measured. The nominal drift field was667 V/cm for the ALICE-D2 detector and for oneD1B, and 583 V/cm for the other D1B prototype.

Some data (600k events) were also taken withinclined tracks in the plane formed by the SDDanode axis and the SDD normal axis. In ALICE,the maximum value possible for the angle betweenthe beam axis and SDD normal axis is 451 forprimary tracks. The inclination angles chosen forthe test were 22.5 and 361, since the mechanicalsupport did not allow larger angles. Another set ofdata (800k events) were taken with the target inorder to collect multitracks events to study thedouble-track separation capabilities of the SDD.

3. Corrections

3.1. Correction for doping inhomogeneities

Since 1999, systematic errors in the determina-tion of the particle impact point position wereobserved [12] and attributed to inhomogeneities ofthe dopant concentration. In SDDs, dopinginhomogeneities affect the performance of thedetector, since they induce variations in thepotential distribution causing a parasitic electricfield. The component of the parasitic electric fieldparallel to the drift direction locally changes thedrift velocity and causes systematic errors in themeasurement of the coordinate along the driftaxis. The component of the parasitic electric fieldperpendicular to the drift direction induces devia-tions of the electron trajectories leading tosystematic errors in the anode coordinate. Toobtain an optimal resolution, it is necessary tocorrect the measured coordinates for these sys-tematic deviations. For the ALICE-D2 prototype,it was possible to map the systematic deviationsalong both the drift and the anode directions forthe full half-detector tested. In Fig. 1, thesystematic deviations of the measured anodiccoordinate (top) and of the measured driftcoordinate (bottom) with respect to the referenceposition given by the microstrip telescope isrepresented as a function of the anode coordinate

(x) and the drift distance (y). The value of thesystematic deviation for each bin of the histogram,corresponding to an area of 0:5 � 0:3mm2; wasobtained by considering the centroid of theresidual distribution between the SDD and thereference coordinates. The empty areas in Fig. 1(a)correspond to noisy or non-working channels andwere not taken into account during the analysis. InFig. 1(b), the horizontal line at a drift distance of�17mm; corresponding to a sharp change of theresidual distribution for all the anodes is due to ashort-circuit between two adjacent biasing cath-odes at that drift distance. The circular structurescentered on the middle of the SDD are attributedto the radial dependence of the dopant concentra-tion fluctuations [13]. The systematic deviationsreach values up to 400mm: In Fig. 2, as anexample, a projection on the y-axis of the previoushistograms, representing the systematic deviationsas a function of the drift distance for a slice aroundthe anode no. 147 ðx ¼ 44mmÞ; is shown for boththe coordinates. This anode corresponds to thefirst anode closest to the central region after thegroup of the discarded channels mentioned above.

These maps were used in the analysis to correctthe coordinates measured by the SDD and thusoptimise their spatial precision. From the values ofthe systematic deviations, it is possible to estimatethe values of the components of the parasiticelectric field. It can be demonstrated that thecomponents of the parasitic electric field paralleland perpendicular to the drift direction can beexpressed by

Ek ¼ EdriftdDy=dy

1 � dDy=dyð1Þ

E? ¼ EdriftdDx=dy

1 � dDy=dyð2Þ

where dDy=dy and dDx=dy represent, respectively,the derivatives of the systematic deviations alongthe drift direction and along the anode directionwith respect to the drift coordinate. Edrift repre-sents the drift field generated by the biasingcathodes. The maps shown in Fig. 3 were obtainedwith Edrift ¼ 667V=cm: In Fig. 3(b), the effect ofthe short-circuit at the drift distance of 17 mm isclearly visible. In Fig. 4, a projection on the y-axis

ARTICLE IN PRESS

Fig. 1. Systematic deviations (represented by the grey scale in mm) of the measured anode coordinate (a) and of the measured drift

coordinate (b) with respect to the reference impact point, as a function of the anode coordinate ðxÞ and the drift distance ðyÞ:

E. Crescio et al. / Nuclear Instruments and Methods in Physics Research A 539 (2005) 250–261 253

from the previous two-dimensional histograms isshown for both the components of the parasiticelectric field. The plots represent the variation ofthe perpendicular (top) and parallel (bottom)components of the parasitic electric field as afunction of the drift distance for a slice around theanode no. 147 at x ¼ 44mm: The parasitic electricfield reaches values up to �50V=cm:

3.2. Correction for temperature variations

Silicon drift detectors are very sensitive totemperature fluctuations, which induce carriermobility, and therefore drift velocity, variationsof about 1%/K at room temperature. In theALICE SDDs, the current flowing through theintegrated voltage divider causes heat dissipation,creating temperature gradients in the sensitiveregion of the detector. Variations of the ambient

temperature during operation also affect thetemperature of the detector and thus the driftvelocity. To monitor the drift velocity variations,the ALICE SDDs are equipped with MOS chargeinjectors [6,7]. For the analysis presented in thispaper, the drift velocity variations caused bytemperature fluctuations were corrected using thereference position given by the microstrip tele-scope.

4. Results

4.1. Cluster size

The electron cloud generated by an ionizingparticle in the SDD undergoes a diffusion whiledrifting to the collection anodes. After thedigitisation of the anode signals, the cloud is

ARTICLE IN PRESS

0 5 10 15 20 25 30 35-400

-300

-200

-100

0

100

200

300

400

anode coordinatem

)µ

Sys

tem

atic

dev

iati

on

s (

Drift distance (mm)

0 5 10 15 20 25 30 35-400

-300

-200

-100

0

100

200

300

400

drift coordinate

m)

µS

yste

mat

ic d

evia

tio

ns

(

Drift distance (mm)

Fig. 2. Systematic deviations from the value predicted by the microstrip telescope of the anode (top) and drift (bottom) coordinate as a

function of the drift distance in the region around the anode coordinate x ¼ 44mm:

E. Crescio et al. / Nuclear Instruments and Methods in Physics Research A 539 (2005) 250–261254

represented by a two-dimensional set of amplitudevalues, called a ‘‘cluster’’. A measurement of thecluster size was performed in both the anode andthe drift directions. The time evolution of theamplifier output signal results from the convolu-tion of the anodic current and the time response ofthe amplifier. The anodic current shape is, to agood approximation, similar to the electroncollection time distribution. The nominal valueof the OLA shaping time is 55 ns for a zero inputcapacitance, but it was larger and not preciselyknown in our experimental conditions, since thedetector capacitance, expected to be about 300 fF[9], was not measured. This uncertainty did notallow the precise determination of the size of theelectron cloud along the drift axis. However, sincethe shaping time of the OLA was constant duringall the data taking and of the same order ofmagnitude of the anodic signal duration, aqualitative comparison of this duration is possibleby calculating the r.m.s. cluster size along the driftdirection, that is the r.m.s. of the anode signalsalong the time axis. Fig. 5 shows the distributionof the r.m.s. of the signal as a function of the drifttime for the various values of the incident angle.

Calculating the ratios between the most probablevalue (m.p.v.) of the signal r.m.s. distribution at01 and at different inclination angles, no signifi-cant difference is observed, confirming the inde-pendence of the cluster size along the drift axisfrom the inclination angle. Fig. 6 shows therelative amounts of clusters collected by one, twoor three anodes as a function of the drift time forinclination angles of 01, 22.51 and 361. Thenumber of clusters collected by more than threeanodes is less than 1%. As expected, the number ofmulti-anode clusters increases with the inclinationangle of the tracks. The number of single-anodeclusters in the anode region decreases with theinclination angle changing from 11% of the totalnumber of clusters for 01 to 4% for 361. Thenumber of three-anode clusters increases with thedrift time and the inclination angle from 20% for01 to 41% for 361 of the total amount of clusters.

4.2. Charge

Since the OLA amplifier has a linear response,for each channel the integral of the output signal isexpected to be proportional to the charge collected

ARTICLE IN PRESS

Fig. 3. Maps of the components of the parasitic electric field, caused by dopant concentration fluctuations of the silicon wafer,

perpendicular (a) and parallel (b) to the drift direction. The values of the field are given by the grey scale in V/cm and are represented as

a function of the anode coordinate ðxÞ and the drift distance ðyÞ:

E. Crescio et al. / Nuclear Instruments and Methods in Physics Research A 539 (2005) 250–261 255

by the anode. Under the condition that thethreshold is sufficiently low compared to themaximum amplitude of the signal, that the signalhas no undershoot, and that the sampling period issmall enough compared to the signal shaping time,the sum of the linearized amplitudes of the clusteris proportional to the total collected charge. In thefollowing, this sum will be called charge. In Fig. 7(top) the most probable value (m.p.v.) of thecluster charge versus the drift distance areshown for the perpendicular tracks and inclinedtracks for each group of the analysed runs. As onecan notice, the collected charge decreases as afunction of the drift distance. A charge collectioninefficiency was already observed in this detectoron the test bench in the laboratory. The necessityto understand the origin of the phenomenon wasone of the reasons to choose this detector forthe beam test. The reason for the collection

inefficiency is attributed to electron trappingcentres in the bulk silicon, introduced by impu-rities during detector fabrication. A cross-checkwith the data from the production cycle allowed topin down the reason for the contamination andcorrect the problem in the subsequent detectorproduction.

The ratio between the m.p.v. for perpendiculartracks and the m.p.v. for the inclined tracks has,apart from the experimental fluctuations, a con-stant value versus the drift time consistent with thevalue of the cosine of the inclination angle, thatcorresponds to the increase in the track length insilicon. The obtained value for the ratio is 0:805 �

0:011 for the inclination angle of 361 and 0:917 �

0:009 for the angle of 22.51. Due to this increase inthe cluster size along the anode axis, an improve-ment of the resolution along the anode axis whenthe angle increases is expected.

ARTICLE IN PRESS

0 5 10 15 20 25 30 35-50-40-30-20-10

01020304050

Drift distance (mm)

Par

asit

ic E

per

p(V

/cm

)

0 5 10 15 20 25 30 35-50-40-30-20-10

01020304050

short-circuit

Drift distance (mm)

Par

asit

ic E

par

alle

l(V

/cm

)

Fig. 4. Components of the parasitic electric field induced by the doping inhomogeneities of the silicon wafer as a function of the drift

distance. The components perpendicular (top) and parallel (bottom) to the drift direction are averaged within a slice centered at the

anode coordinate x ¼ 44mm:

E. Crescio et al. / Nuclear Instruments and Methods in Physics Research A 539 (2005) 250–261256

4.3. Position resolution

The detector position resolution was evaluatedconsidering the difference between the clusterposition measured by the SDD and the impactpoint coordinate reconstructed with the microstriptelescope. The resolution is defined as the r.m.s. ofthe residual distribution. The data were correctedfor the doping inhomogeneities and, for theresolution along the drift axis, for the temperaturevariations. For the data taken with perpendiculartracks, all the anodes of a half-detector were readout and the resolution in different regions, alongboth the anode and the drift direction, wasstudied. For the analysis of two-track resolutionand for the data with inclined tracks a smallerregion was studied.

4.3.1. Perpendicular tracks

The resolution along both the drift axis and theanode axis was evaluated on eight separate regionsof 32 anodes each. Fig. 8(a) shows the resultsobtained for the resolution along the drift axis,after the corrections for the doping inhomogene-

ities and the temperature variations. The blackcircles represent the average of the values of theresolution obtained in the separate regions. Thegrey area represents the difference of the values ofthe resolution in the different regions. The limitsare given by the worst and the best values. Theaverage value of the resolution is �35mm: Fig. 8(b)shows the spatial resolution along the anode axis.At short drift distances, the resolution isaffected by the intrinsic uncertainty due to thecharge cloud being smaller than the anode pitch.At a drift distance of �15mm; the average r.m.s.of the residual distribution reaches its minimumvalue of �15mm: At the longest drift distances, theresolution deteriorates with respect to the mini-mum value because of the reduction of thesignal-to-noise ratio ðS=NÞ due to the signalspreading over more anodes due to diffusion andthe charge loss. S=N can be defined approxi-mately as Q=ðs

ffiffiffin

pÞ where Q is the integral of

the signal (proportional to the charge), s is thenoise r.m.s. and n is the average number ofanodes which collect the signal. The difference inS=N between zero and the maximum drift distance

ARTICLE IN PRESS

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

s)µDrift time (

0 0.5 1 1.5 2 2.5 3 3.5 4

RM

S (

x25

ns)

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

at 0 degrees, anodes 0-60

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

s)µDrift time (

0 0.5 1 1.5 2 2.5 3 3.5 4

RM

S (

x25n

s)

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

at 0 degrees, anodes 60-110

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

s)µDrift time (

0 0.5 1 1.5 2 2.5 3 3.5 4

RM

S (

x25n

s)

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

at 22 degrees, anodes 0-60

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

s)µDrift time (

0 0.5 1 1.5 2 2.5 3 3.5 4

RM

S (

x25n

s)

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

at 36 degrees, anodes 60-110

Fig. 5. Distributions of the cluster size r.m.s. along the drift direction as a function of the drift time and for different anode regions and

angles.

E. Crescio et al. / Nuclear Instruments and Methods in Physics Research A 539 (2005) 250–261 257

can be estimated:

k ¼ð1 � aÞQ0=ðs

ffiffiffiffiffiffiffi2:5

pÞ

Q0=ðsffiffiffiffiffiffiffi1:1

pÞ

¼ ð1 � aÞffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi1:1=2:5

p¼ 0:4

(3)

where a represents the charge loss and s the noiser.m.s. This value corresponds to a variation ofS=N of about 60%.

The average resolution is of the order of�25mm: The differences in the values of theresolution for the different regions are due todifferent gain stability of the electronic channels.The regions with worse resolution correspond toelectronic channels which have the gain less stablein time than the others.

4.3.2. Spatial resolution with inclined tracks

The spatial resolution with inclined tracks ispresented only along the anode axis, where an

improvement, with respect to the one withperpendicular tracks, is expected. A modestimprovement of the spatial resolution along thedrift direction is expected from the higher value ofthe S=N ratio. However, the S=N variationobtained by changing the incidence angle by anangle y with respect to the perpendicular is1= cos y which is about 20% for an angle of 351.In Fig. 8(a), we can see that a variation of S=N upto 60% produce a change of just 6 mm on theresolution. On the other hand, simulations made inthe AliRoot framework, the ALICE simulationtool, have also indicated that a negligible influenceis expected by increasing the S=N ratio of 20%[14]. In Fig. 9, the comparison between the spatialresolution at 01 and at different inclination anglesas a function of the drift distance is shown. Fig.9(a) shows the comparison between the measuredresolution for perpendicular tracks and for the

ARTICLE IN PRESS

s)µDrift time (0 0.5 1 1.5 2 2.5 3 3.5 4

s)µDrift time (

0 0.5 1 1.5 2 2.5 3 3.5 4

s)µDrift time (0 0.5 1 1.5 2 2.5 3 3.5 4

%

0102030405060708090

100

%

0102030405060708090

100

%

0102030405060708090

100

one-anode clusterstwo-anode clustersthree-anode clusters

at 0 degrees

at 22 degrees

at 36 degrees

(a)

(b)

(c)

Fig. 6. Percentage of the events in which a cluster is collected by one, two or three anodes as a function of the drift time for inclination

angles of 01 (a), 22.51 (b) and 361 (c).

E. Crescio et al. / Nuclear Instruments and Methods in Physics Research A 539 (2005) 250–261258

inclination of 22.51 in the SDD region betweenanodes 0 and 60. The comparison between theresolution for perpendicular tracks and for theinclination of 361 is illustrated in Fig. 9(b), forthe SDD region between anodes 60 and 110. It canbe observed that in the vicinity of the SDDanodes, the values of the anodic resolutionfor inclined tracks become better with respect to

those for perpendicular tracks. This behaviouris caused by the decrease of the fraction ofnarrow clusters in the inclined track events.For longer drift distances, the values of theresolution are very similar to those for perpendi-cular tracks. Except at short drift distances,the average value of the resolution is betterthan 30mm:

ARTICLE IN PRESS

Drift distance (mm)

0 5 10 15 20 25 30

m)

µR

eso

luti

on

(

0

10

20

30

40

50

60

70Drift axis

Drift distance (mm)0 5 10 15 20 25 30

m)

µR

eso

luti

on

(

0

10

20

30

40

50

60

70Anode axis

(a)

(b)

Fig. 8. Spatial resolution along the drift direction (a) and along

the anode direction (b) as a function of the drift distance. The

black circles represent the average of the values of resolution

measured in different anode regions, the limits of the grey areas

represent the worst and the best values.

sµDrift time,

0 0.5 1 1.5 2 2.5 3 3.5 4

Ch

arg

e

0

200

400

600

800

1000

1200

1400

1600

at 0degreesat 22degreesat 36degrees

sµDrift time,

0 0.5 1 1.5 2 2.5 3 3.5 4

Rat

io

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

Charge_0/Charge_36

Charge_0/Charge_22

(a)

(b)

Fig. 7. (a) Most probable value of the charge collected (in

arbitrary units) as a function of the drift time for inclined tracks

and for perpendicular tracks. (b) Ratio of the most probable

value of the charge as a function of the drift time. The lines

indicate the values of cos 22:5� and cos 36�:

E. Crescio et al. / Nuclear Instruments and Methods in Physics Research A 539 (2005) 250–261 259

4.3.3. Two track resolution

The double-track resolution was studied for theD2 detector. The detector was positioned at adistance of 8 cm from the iron target and a specificselection of data were applied. The events with twoand only two tracks of the secondary particles,reliably reconstructed by the rear microstriptelescope, were selected. A minimisation procedureusing the single track data were applied for theseparation of the overlapping SDD clusters. Bothof these reconstructed tracks had to pass throughthe SDD area selected for the analysis. If the SDDreconstructed exactly two hits within a compatibledistance from the tracks, the event was counted asresolved. If there was only one SDD reconstructedhit, the event was unresolved. Otherwise, the eventwas discarded due to inefficiency of the analysisprocedure. The fraction of resolved two-track

events as a function of distance between tracks isshown in Fig. 10. The 50% point, where half of thetrack pairs are resolved, is reached after aseparation value of 450mm; while 90% is reachedat 800mm:

5. Conclusions

An extensive study of the performance of asilicon drift detector with characteristics close tothe final SDD designed for the ALICE experimentwas carried out using the OLA front-end ASIC.The characteristics of the detector were studied in

ARTICLE IN PRESS

Fig. 9. Spatial resolution along the anode axis as a function of the drift distance: comparison between the resolution obtained with

perpendicular tracks and that obtained with tracks inclined at 22.51 (a) and at 361 (b).

0 200 400 600 800 1000 1200 14000 200 400 600 800 1000 1200 1400

0

0.2

0.4

0.6

0.8

1

m)µDistance between tracks (

reso

lved

eve

nts

/to

tal n

um

ber

of

even

ts

Fig. 10. Fraction of resolved two-tracks events as a function of

distance between tracks.

E. Crescio et al. / Nuclear Instruments and Methods in Physics Research A 539 (2005) 250–261260

the case of perpendicular tracks and with inclinedtracks. The analysis showed the presence ofsystematic effects due to doping inhomogeneities,

which affect the measurement of the coordinatesleading to systematic deviations up to 400mm:Applying a correction for this effect, the averagevalues of the spatial resolution in the case ofperpendicular tracks are �25mm along the anodeaxis and �35 mm along the drift axis. As expected,a significant improvement of the resolution alongthe anode axis was observed for the data takenwith inclined tracks for small drift distances(p10mm). The double-track resolution was alsomeasured and an average value of 450 mm wasobtained.

Acknowledgements

L.M. Montano wishes to thank Conacyt forfinancial support.

ARTICLE IN PRESS

E. Crescio et al. / Nuclear Instruments and Methods in Physics Research A 539 (2005) 250–261 261

References

[1] ALICE Collaboration, CERN/LHCC 99-12.

[2] E. Gatti, P. Rehak, Nucl. Instr. and Meth. A 225 (1984)

608.

[3] S. Beole, et al., Nucl. Instr. and Meth. A 377 (1996) 393.

[4] S. Beole, et al., Il Nuovo Cim. A 109 (9) (1996).

[5] A. Rashevsky, et al., Nucl. Instr. and Meth. A 461 (2001)

133.

[6] V. Bonvicini, et al., Il Nuovo Cim. A 112 (1999)

137.

[7] V. Bonvicini, et al., Nucl. Instr. and Meth. A 439 (2000)

476.

[8] V. Bonvicini, et al., Nucl. Instr. and Meth. A 459 (2001)

494.

[9] W. Dabrowski, et al., Nucl. Phys. B 44 (1995) 637.

[10] A. Adamo, et al., Sov. J. Nucl. Phys. 55 (1992) 806.

[11] F. Balestra, et al., Nucl. Instr. and Meth. A 323 (1992) 523.

[12] D. Nouais, et al., Nucl. Instr. and Meth. A 461 (2001) 222.

[13] W. von Ammon, H. Herzer, Nucl. Instr. and Meth. 226

(1984) 94.

[14] E. Lopez Torres, P. Cerello, ALICE-INT-2001-35.