Table Of Content

The ATLAS Pixel Detector Jörn Grosse-Knettera, 4 0 on behalf of the ATLAS Pixel collaboration 0 2 a Physikalisches Institut, Universita¨t Bonn, Nussallee 12, D-53115 Bonn, Germany n a J 4 Abstract 1 ] The ATLAS Pixel Detector is the innermost layer of the ATLAS tracking system t e andwillcontributesignificantly totheATLAStrack andvertex reconstruction.The d - detector consists of identical sensor-chip-hybrid modules, arranged in three barrels s in the centre and three disks on either side for the forward region. n i . The position of the Pixel Detector near the interaction point requires excellent s c radiation hardness, mechanical and thermal robustness, good long-term stability, i s all combined with a low material budget. The detector layout, results from final y prototyping and the status of production are presented. h p [ Key words: silicon detector, pixels, LHC 1 PACS: 06.60.Mr, 29.40.Gx v 8 6 0 1 Introduction into three barrel layers in its centre, 1 one of them around the beam pipe 0 4 (r = 5 cm), and three disks on either 0 The ATLAS Inner Detector [1] is side for the forward direction. With / s designed for precision tracking of a total length of approx. 1.3 m this c i chargedparticleswith40MHzbunch results in a three-hit system for par- s y crossing identification. It combines ticles with |η| < 2.5. h tracking straw tubes in the outer p : transition-radiation tracker (TRT) v i and microstrip detectors of the semi- X conductor tracker (SCT) in the mid- r a dle with the Pixel Detector, the cru- cial part for vertex reconstruction, The main components are ap- as the innermost component. prox. 1700 identical sensor-chip- hybrid modules, corresponding to a The Pixel Detector [2] is subdivided 7 total of 8 · 10 pixels. The modules Email address: have to be radiation hard to an AT- [email protected](Jo¨rn LAS life time dose of 50 MRad or 15 Grosse-Knetter). 10 neutron-equivalent. Preprint submitted to Elsevier Science 2 February 2008 2 Module Layout coupled second stage and a differen- tial discriminator. The threshold of the discriminator ranges up to 1 fC, Apixel module consists ofa single n- itsnominalvaluebeing0.5fC.When 2 on-n silicon sensor, approx. 2×6 cm a hit is detected by the discrimi- in size. The sensor is subdivided into nator the pixel address is provided 47,268pixelswhichareconnectedin- together with the time over thresh- dividually to 16 front-end(FE) chips old (ToT) information which allows via “bumps” [3]. These chips are reconstruction of the charge seen by connected to a module-control chip the preamplifier. (MCC)[4]mountedonakapton-flex- hybrid glued onto the back-side of the sensor. The MCC communicates 3 Module Prototypes with the off-detector electronics via opto-links, whereas power is fed into the chips via cables connected to the Several ten prototype modules have flex-hybrid. been built with a first generation of radiation-hard chips in 0.25µm- To provide a high space-point reso- technology. In order to assure full lution of approx. 12µm in azimuth functionality of the modules in the (rφ), and approx. 90µm parallel to later experiment, measurements at the LHC beam (z), the sensor is sub- the production sites, after irradia- dividedinto41,984“standard”pixels tion, and in a test beam are per- of 50 µm in rφ by 400 µm in z, and formed. 5284 “long” pixels of 50 × 600 µm2. Thelongpixelsarenecessarytocover the gaps between adjacent front-end 3.1 Laboratory measurements chips. The module has 46,080 read- out channels, which is smaller than the number of pixels because there is Animportanttestthatallowsalarge a 200 µm gap in between FE chips rangeof in-laboratorymeasurements on opposite sides of the module, and is the threshold scan. Signals are togetfullcoveragethelasteight pix- createdwithon-modulechargeinjec- els at the gap must be connected to tionand scanning the number of hits onlyfourchannels(“ganged”pixels). versus the so injected charge yields Thus on 5% of the surface the infor- the physical value of the threshold of mationhasatwo-foldambiguitythat the discriminator and the equivalent will be resolved off-line. noise charge as seen by the preamplifier. A set of such scans is used to The FE chips [3] contain 2880 in- reduce the threshold dispersion by dividual charge sensitive analogue adjusting a DAC-parameter individ- circuits with a digital read-out that ually for each channel. The resulting operates at 40 MHz clock. The ana- threshold and noise after threshold logue part consists of a high-gain, tuningisshowninfigure1.Typically fast preamplifier followed by a DC- approx. 100 e threshold dispersion 2 Threshold distribution CCoonnssttaanntt 88999922 Noise distribution Noise distribution long 894567000000000000000000 MSMSiieeggaammnnaa 77 334422..77448866 212313005055500000000000000 CMSCMSiiooeeggnnaammssnnaatt aa nn tt 1111 3377773377..3399..554488 246713580000000000000000 CMSCMSiiooeeggnnaammssnnaatt aa nn tt 77112299880000..1155....552233 23000000 100 150 200 250 300 350 4N0o0ise4 5(e0) 100 150 200 250 300 3504N0o0ise4 5(e0) 1000 Noise distribution ganged Noise distribution betw. ganged e)Thre5s0h0o0ld sca2t0te0r0 plo2t500 3000 3500 4000 45T0h0res5h0o0l0d (e) 11024628000000 CMSCMSiiooeeggnnaammssnnaatt aa nn tt 441122441188..991155....558833 21111100462462880000000000 CMSCMSiiooeeggnnaammssnnaatt aa nn tt 11222288000066..4411....779966 Threshold ( 34234005550000000000 Noise (e)1N0o03434i5500s10000e50 sc20a0tte2r5 p0lo30t0 3504N0o0ise4 5(e0) 100 150200 250 300 3504N0o0ise4 5(e0) 2000 225000 0 10"0C00hannel"2 0=0 r0o0w+160*3c0o0l0u0mn+288400*0c0h0ip 1150000 5000 10000 15000 20000"C2h50a0n0nel"3 =0 0r0o0w+136500*c0o0lum4n00+020880*4c5h0i0p0 Fig.1. Distributionsofthreshold(left)andnoise(right)ofamoduleafterindividual threshold tuning. across a module and a noise value approx. the dose expected after 10 of below 200 e for standard pixels is years of ATLAS operation. The ra- achieved, as is needed for good per- diation damage is monitored reading formance. In a similar fashion, the the leakage current individually for cross-talk is measured to a few per each pixel. The single event upset centforstandard50×400µm2pixels. rate is measured during irradiation. A measurement of the timewalk, The threshold dispersion and the i.e. the variation in the time when noise after irradiation as shown in the discriminator input goes above figure 2 is only modestly increased threshold, is an issue since hits with and still well in agreement with re- a low deposited charge have an ar- quirements for operation in ATLAS. rival time later than those with high charges, inparticular for gangedpix- els. 3.3 Test beam measurements Data taken when illuminating the sensor with a radioactive source al- Tests have been performed in the lowsin-laboratorydetectionofdefec- beam line of the SPS at CERN using tive channels. The source-spectrum 180 GeV/c hadrons. The setup con- reconstructedfromtheToT-readings sists of a beam telescope for the po- is in agreement with expectations. sition measurement [5], trigger scin- tillators for timing measurement to 36 ps, and up to four pixel modules. 3.2 Irradiation The number of defective channels is −3 observed to less than 10 and for standard 50 × 400 µm2 pixels the efficiency for normal incidence parti- cles is 99.57±0.15%.The timewalk is Some of the prototype modules have measured to values similar to those beenirradiatedtoadoseof50MRad, from lab tests (see above). 3 Fig.2. Distributionofthreshold(left)andnoise(right)ofamoduleafterirradiation with 50 MRad. Modules irradiated as described walk and too little SEU tolerance in section 3.2 have been tested in have been addressed and solved in a the beam line and the bias voltage newdesignofFEandMCC chips [3]. needed for full depletion is measured to be between 500 and 600 V, see figure 3. The deposited charge is measured via the ToT readings and no significant changes in the unifor- mity w.r.t. unirradiated modules are observed. 4 Off-detector electronics 3.4 Next generation of chips The off-detector readout electronics is designed to process data at a rate The results of the prototype mod- of up to 100 kHz level-1 triggers. ules show that the chips fulfil largely The main data-processing compo- the requirements of electrical perfor- nent is the “read-out driver” (ROD), mance and radiation hardness and of which first prototypes have been that the production process has a built to pixel specifications and are high yield. Problems such as time- beingevaluated.Thefirst-stepevent- building and error flagging is done via FPGAs. The communication to the rest of the DAQ-system is run through a 1.6 Gbit/s opto-link. The communication to modules, online monitoring and calibration runs are performed with SRAMs and DSPs; their programming is ongoing and modules have already been config- Fig. 3. Depletion depth measured in a ured and operated successfully with testbeamasafunctionofbiasvoltage. a ROD. 4 5 System aspects Noise ratio distribution CCoonnssttaanntt 33330000 MMeeaann 00..99995599 3000 SSiiggmmaa 00..00667733 2500 2000 1500 5.1 Support structures 1000 500 0.6 0.8 1 1.2 1.4 1.6 1.8 Noise ratio Noise ratio scatter plot Ttohegumareacnhtaeneicgsooodf ptohseitisoynsatlemstahbails- Noise ratio111...468 1.2 ity of the modules during operation 1 0.8 while the amount of material has 0.6 0 5000 10000 15000 20000 25000 30000 35000 40000 45000 "Channel" = row+160*column+2880*chip to be kept to a minimum. At the same time it has to provide cooling Fig. 4. Ratio of noise on one module to remove the heat load from the comparing between simultaneous oper- modules and maintain the sensors at ation with two other modules and indi- a temperature of -6◦C to keep the vidual operation. radiation damage low. 6 Conclusions Barrel-modulesaregluedto“staves”, long, flat carbon-structures with Prototype modules built with a first cooling pipes embedded. The staves generation of radiation hard chips are mounted inside halfshells, which show largely satisfying performance themselvesareassembledintoframes inlaboratory-test,intestbeamstud- to form the barrel system. ies and after irradiation. Remaining problems have been solved in a new Thedisksareassembledfromcarbon- generation of chips which is now sectors with embedded cooling cov- ready for production. ering 1/12 of a wheel. The modules areglueddirectlytoeithersideofthe Work on the off-detector electron- disk sectors. ics and the support structures have beengoingoninparallelandarewell on track. First systemtest results are promising. 5.2 Systemtests References First mini-systemtests have been performedwithsixmodulesonadisk sectorandthreemodulesonabarrel- [1] Technical stave. The noise behaviour on the Design Reportof theATLAS Inner disks or staves shows no significant Detector,CERN/LHCC/97-16and differences compared to similar mea- CERN/LHCC/97-17 (1997). surements with the same modules [2] Technical Design Report of the individually, see figure 4. Larger sys- ATLAS Pixel Detector, temtests are already in preparation CERN/LHCC/98-13 (1998). and will include realistic powering and read-out. [3] F. Hu¨gging, this proceedings. 5 [4] R. Beccherle et al., Nuclear Instr. Meth. A 492, 117 (2002). [5] J. Treis et al., Nuclear Instr.Meth. A 490, 112 (2002). 6