ARTICLE IN PRESS

0168-9002/$ - se

doi:10.1016/j.ni

�CorrespondE-mail addr

Nuclear Instruments and Methods in Physics Research A 570 (2007) 276–280

www.elsevier.com/locate/nima

The GLAST LAT tracker construction and test

F. Bellia,�, L. Andreanellia, F. Angelinib, R. Bagaglib, M. Bagnic, L. Baldinib, G. Barbiellinid,F. Bellardib, R. Bellazzinib, A. Brezb, M. Brigidae, G. Busettof, A. Caliandroe, M. Ceccantib,

C. Cecchig, A. De Angelish, C. Favuzzie, M. Fazzib, G. Fogliac, M. Frailish, P. Fuscoe,F. Garganoe, D. Gasparrinig, S. Germanig, R. Giannitrapanih, N. Gigliettoe, F. Giordanoe,

M. Kussb, L. Latronicob, F. Longod, F. Loparcoe, P. Lubranog, B. Marangellie,M. Marchettia, M.M. Massaib, G. Mazzengaa, M.N. Mazziottae, N. Menonc, M. Minoria,M. Minutib, N. Mirizzie, M. Mongellie, A. Morsellia, N. Omodeib, M. Pepeg, M. Perchiazzie,

M. Pesce-Rollinsb, P. Picozzaa, S. Raino’e, R. Randof, E. Rapposellic, M. Razzanob,A. Sacchettie, N. Sagginic, G. Scolierig, C. Sgro’b, G. Spandreb, P. Spinellie, A. Tenzeb,

A. Troianielloc

aINFN-Tor Vergata and University of Roma Tor Vergata, Via della Ricerca Scientifica 1, 00133 Roma, ItalybINFN-Pisa and University of Pisa, Via Buonarroti 2, 56100 Pisa, Italy

cStellar Solutions Inc. and INFN-PisadINFN-Trieste and University of Trieste, Via Valerio 2, 34127 Trieste, Italy

eINFN-Bari, Via E. Orabona 4, 70126 Bari, ItalyfINFN-Padova and University of Padova, Via Marzolo 8, 35131 Padova, Italy

gINFN-Perugia, Via A. Pascoli, 06154 Perugia, ItalyhINFN-Trieste and University of Udine, Via delle Scienze 208, 33100 Udine, Italy

Available online 18 October 2006

Abstract

GLAST is a next generation high-energy gamma-ray observatory designed for making observations of celestial gamma-ray sources in

the energy band extending from 10KeV to more than 300GeV. Respect to the previous instrument EGRET, GLAST will have a higher

effective area (six times more), higher field of view, energy range and resolution, providing an unprecedented advance in sensitivity

(a factor 30 or more). The main scientific goals are the study of all gamma-ray sources such as blazars, gamma-ray bursts, supernova

remnants, pulsars, diffuse radiation, and unidentified high-energy sources. The construction and test of the Large Area Telescope (LAT)

tracker, has been a great effort during the past years, involving tens of people from many Italian INFN sections and industrial partners.

Environmental and performance tests of the hardware, detectors and reading electronics, have been carried on during all the steps of the

LAT construction. The resulting LAT performance are better than the ones required by the original science proposal, demonstrating the

quality of the italian group effort. In this article we summarize the LAT construction and test workflow, presenting its main results.

r 2006 Elsevier B.V. All rights reserved.

PACS: 07.85.-m; 07.87.+v; 52.70.La

Keywords: Gamma-ray; Satellite; Silicon detectors; Tracker

e front matter r 2006 Elsevier B.V. All rights reserved.

ma.2006.09.062

ing author.

ess: Francesco.Belli@roma2.infn.it (F. Belli).

1. The GLAST scientific program

GLAST (Gamma-Ray Large Area Telescope) [1] is anext generation experiment for the observation of cosmic-gamma rays in the energy range from 10KeV to more than

ARTICLE IN PRESSF. Belli et al. / Nuclear Instruments and Methods in Physics Research A 570 (2007) 276–280 277

300GeV. This will permit the study of gamma-ray sourcessuch as blazars, gamma-ray bursts, supernova remnants,pulsars, diffuse radiation and unidentified sources in anuncovered energy range with respect to previous experi-ments, such as EGRET [2] and with an overlap at higherenergies with ground experiments (Fig. 1). Furthermore, incomparison with EGRET, GLAST will have betterperformance in terms of effective area, angular resolution,field of view and dead time (Fig. 2), providing animprovement factor of 30 or more in sensitivity, and bettercapability to study transient phenomena [3].

Fig. 1. GLAST energy range.

Fig. 2. GLAST LAT perform

The GLAST construction has involved many worldwidepartners from United States, Italy, Japan, France andSweden, both from research institutions and from indus-tries. The construction of the Large Area Telescope (LAT)Tracker modules [4] performed in Italy and finished lastSeptember, after years of efforts with excellent results,demonstrating a great coordination work between re-searchers and industrial partners.

2. The LAT tracker construction and test workflow

Modularity is the main concept in the tracker design.This allows the necessary redundancy to guarantee a safeoperability in space environment and reproducibility in theconstruction workflow. Furthermore, severe acceptance,functional, mechanical and environmental tests are indis-pensable during all the instrument construction steps.The LAT (Fig. 3) is a 4� 4 array of 16 towers, each one

composed by a tracker, in which the photon conversiontakes place, followed by a calorimeter; the whole issurrounded by a segmented anticoincidence shield to allowthe charged particle signal rejection.The tracker (composed by 16 towers, for a total

detecting surface of 83m2), has been built with almost10,000 single sided, 8:95� 8:95 cm microstrip siliconsensors, 228mm pitch, bonded in groups of four to forma ladder; four ladders constitute a tracker tower plane.These planes are stacked with their strips alternativelyoriented at 90� to form an xy coordinates grid: a tower iscomposed by 19 panels for a total of 18 of such xy planes.

ance compared to Egret.

ARTICLE IN PRESS

Fig. 3. The GLAST-LAT instrument.

Depletion voltage

MeasuredExpected

MeasuredExpected

250

200

150

100

50

0

250

300

350

400

450

200

550

600

650

700

750

500

850

900

95010

00800

150

100500

20018016014012010080604020

0

0 10 20 30 40 50 60 70 80 90 110

120

130

140

150

160

170

180

200

190

100

Depletion voltage (v)

Leakage current @ 150 V

Leakage current (nA)

(a)

(b)

Fig. 4. Ladder measurement results.

SSD's

Bias-Circuit

StructuralTray Panel

ConverterFoils

Bias-Circuit

SSD's TMCM

Fig. 5. Exploded view of a tray.

F. Belli et al. / Nuclear Instruments and Methods in Physics Research A 570 (2007) 276–280278

The planes are interleaved with tungsten converter foils, fora total radiation length of 1.5 X 0. The tracker will have atotal of 880,000 reading channels and will measure photondirection, using any three consecutive xy plane triggersignal, with a hit efficiency 499% and a noise occupancyo�10�5 for each channel. The whole LAT payload will beabout 3000 kg, for a total consumption of about 650W.

The low consumption of silicon sensors (vendor:Hamamatsu Photonics) is provided by their low leakagecurrent (o500 nA@150V; o200 nA averaged on any 100SSDs), low depletion voltage ðo120VÞ and high break-down voltage ð4175VÞ. Hamamatsu produced and tested11,500 of such sensors. After delivery they have been testedby INFN both for geometrical and electrical characteriza-tion: only a fraction of 0.5% from the total of SSDs havebeen rejected, the remaining ones showing an excellentagreement with the features declared by the vendor, inaverage already better than the aggressive requiredspecifications. Good mechanical alignment is the firstimportant feature to assure tracker precision, and theexcellent silicon wafer slicing has been the first step toassure a good final result.

In order to assembly a ladder, four SSDs are glued, theirstrips bonded and these junction zones encapsulated,obtaining a strips 40 cm long detector. This assembly wasperformed by two high-tech italian companies (G&AEngineering and Mipot), in close collaboration with INFNpersonnel. Mechanical measures of the ladders alignmentshowed no differences from the single SSD 2mm tolerance.Electrical measures on bulk capacitance, depletion voltageand leakage current of the 2850 produced ladders areshown in Fig. 4, with a final rejection rate of �1%, and aglobal result of the measured quantities in excellentagreement with what expected from the component SSDs.The slight difference in the expected and measured leakagecurrent distribution is due to differences in temperatureand humidity of the testing environments. Single stripcurrent and capacitance tests have been performed on anyladder, with a final result of 0.016% bad channels, causedby bonding or probing.

The trays, the mechanical supports for the sensors, arecomposite panels (see Fig. 5): an Al honeycomb core issurrounded by carbon–carbon close-outs, on which theread-out electronics (tracker multi chip modules, TMCMs,

from SLAC) are mounted, while the top and bottom sides,where the sensors are glued, are carbon fiber sheets withkapton bias circuits, interleaved in the bottom side by theconverter foils.Trays manufacturer was the Italian company Plyform,

specialized in the assembly of composite materials. Thegluing of their wide face-sheets needed great care andstructural verification tests were performed along anyassembly step. Opto-acoustic (Electronic Speckle PatternInterferometry, ESPI) tests were performed at INFN site

ARTICLE IN PRESS

100

99.5

99ency

(%

)

0

0.5

1 anne

ls (

%)

F. Belli et al. / Nuclear Instruments and Methods in Physics Research A 570 (2007) 276–280 279

after the carbon fiber sheets gluing, to detect honeycombskin debonding; before mounting electronics and SSDs, athermal–vacuum test (24-h vacuum at 60 �C, performed atINFN site) on tray panels detected trays with badly gluedkapton circuits (this problem was resolved using etched/primed W tiles). The tower bottom trays have specialflexures for mounting them on the integration grid togetherwith the calorimeter modules, as well as to permit theirthermal exchange inside the satellite. This flexures willsuffer particular mechanical stress, especially during thesatellite launch: for this reason bottom trays were subjectto static tests to demonstrate the static strength andstability of the bonding of each flexure. All parts andassembly phases have been dimensionally checked with aCoordinate Measuring Machine (CMM). The laddersdisplacement ðX=Y=ZÞ on trays has an RMS �20 mm.After read-out electronics were mounted, tray functionaltests were performed in order to verify leakage current,single channel gain and noise (to detect disconnectedchannels), along with cosmic rays data taking. Theresulting average leakage current was �100 nA=SSD,showing that the SSD performance were preserved allalong the assembly phases. After completion of assemblyand test at G&A, each tray went through four thermalcycles from �15 to 45 �C (at INFN site), then tray groupwere stacked in parallel to reproduce a tower configuration(Fig. 6 on the left). Functional tests with cosmic rays,monitored with an external trigger setup (three scintillatorplanes) verified the self-trigger capability of the detectorplanes.

To build a tower (Fig. 6 on the right) trays were slit inposition with a crane and stacked into the assembly jig (topto bottom and alternatively rotated by 90�). Each trayposition was referred to the assembly jig (absolutereference frame) and displacements in X ;Y or Z wereaveraged out and did not sum. Tower sides are equipped

Fig. 6. On the left: stacked trays reproducing a tower configuration. On

the right: an assembled tower.

with two redundant cables, each connecting to a customTower readout Electronics Module (TEM). The completetowers were measured under the INFN CMM, and eachtower was well aligned and inside the stay-clear zone(1.5mm/side) defined to interface the towers to the satellitegrid. The alignment did not change after the vibration testsand shipment steps.Cosmic rays data taking permitted also towers analysis

with offline reconstructed events, showing all towers layerefficiencies 499:5% (inefficiencies due to missing chan-nels), and typical layer displacement and rotation, respec-tively, of 90mm and 0.2mrad (see Figs. 7 and 8).Environmental tests of the complete towers were

performed in Alenia Spazio facilities in Rome, undersupervision of INFN personnel. Vibration tests (sinusoidaland random vibrations, plus low level sweeps for resonancefrequency search) showed first resonance frequencies with-in specifications, with no significant changes throughoutthe test. During thermal–vacuum tests (four completecycles, �15 �=þ 45 �C, 10�5 torr), tower functionalities andperformance were continuously monitored for about oneweek test duration. Study of fine effects (gain and noise offront-end chip) together with statistics of bad channels,alignment and detection efficiency, as a function of thetemperature, showed a stable behaviour. Only four failureson a total of 306 trays (580 detector planes) were detected(tower 1,7,8,14). These towers were partially disassembledand fixed.All the workflow steps have been recorded into a custom

electronic database developed by INFN, to permit the

99.5Effi

ci

98

1.5

Bad

ch

2A B 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

LAT TKR performance

Efficiency Bad chans fraction

Fig. 7. Tower efficiencies.

80Silicon layer horizontal shift

70

60

50

40

30

20

10

0-140 -120 -100-80 -60 -40 -20 0

µm20 40 60 80 100 120 140

16 towers

Fig. 8. Tower plane horizontal shift.

ARTICLE IN PRESS

Fig. 9. LAT final performance.

F. Belli et al. / Nuclear Instruments and Methods in Physics Research A 570 (2007) 276–280280

traceability and managing (analysis, quality control andnon-conformance reports) of all the production and testdata.

The final performance of the LAT are summarized inFig. 9, compared to the ones required by the originalScience Document [5].

3. Conclusions

The completion of the LAT tracker in Italy marked amilestone in the space detector construction, obtaining thelargest (83m2 of silicon microstrip detector), high efficient,low noise detector of this kind ever built so far. The sameperformance homogeneity hold among all the trackerconstruction steps has been crucial for a smooth integra-tion with the other subsystems and a good calibration of

the LAT. Furthermore, the good coordination, skill andefforts of all the involved collaborating institutions andgroups led to final performance better than the originallyrequired ones, with undoubted positive impact on theGLAST (whose launch is scheduled for August 2007) finalperformance.

References

[1] W. Atwood, et al., Nucl. Instr. and Meth. A 342 (1994) 302.

[2] P. Nolan, et al., IEEE Trans. Nucl. Sci. NS-39 (1992) 993.

[3] A. Morselli, Nucl. Instr. and Meth. A 530 (2004) 158.

[4] R. Bellazzini, et al., Nucl. Phys. B 113B (2002) 303.

[5] E. Bloom, G. Godfrey, S. Ritz, Proposal for the Gamma-ray Large

Area Telescope, SLAC-R-22, February 1998.