1. Pixel 2000 Workshop Christian Grah grah@whep.uni-wuppertal.de University of Wuppertal www.atlas.uni-wuppertal.de June 2000, Genova O. Bäsken K.H.Becks P.Gerlach Ch. Grah O.Ehrmann M.Töpper J.Wolf Pixel Detector Module using MCM-D Technology for the B-layer of the ATLAS Pixel Detector

2. Pixel 2000, Genova Ch.Grah University of Wuppertal 2 Overview äThe concept of building modules in MCM- D technology äMCM-D modules for the ATLAS Pixel Detector äMeasurements on Prototypes âLab measurements on full scale module and single chip devices âTestbeam measurements on single chip devices äConclusion and Outlook

3. Pixel 2000, Genova Ch.Grah University of Wuppertal 3 ATLAS Pixel Detector ä2200 modules ä2.2 m 2 active Si ä1 x 10 8 channels

4. Pixel 2000, Genova Ch.Grah University of Wuppertal 4 The basic structure of modules for the ATLAS Pixel Detector äSensor tile (16.4 mm x 60.4 mm active area) Main components which need to be contacted: ä16 read out IC´s, each providing 18 x 160 pixel unit cells (preamplifier, discriminator, digital readout; pixel cell size: 400 x 50 µm 2 ) 46080 connections in the pixel cell array per module with bump-bonding and flip-chipping as interconnection technique äone module controller chip äIdea of using a Thin Film technology to perform the signal interconnections and power distribution on the active sensor

5. Pixel 2000, Genova Ch.Grah University of Wuppertal 5 MCM-D, a Thin Film Technology äUp to 5 copper layers:  magnetron sputtered up to 2 m Ti/Cu/Ti  10 m/   additive electroplating up to 5  m Ti/Cu  Minimal width and spacing 10 and 20 m äFinal metallisation: âelectroless  5m Ni:P/ 200nm Au ä“Spin-on” polymer: BCB (Benzocyclobutene / DOW:CYCLOTENE ™ ) äPhotosensitive  Specific dielectric constant  r = 2.7 äProcess temperatures : 1h 220C per layer last layer 1h 250 C  Thickness / layer 4 - 10  m  Via  >20  m, Pad 30µm conductor layersdielectric layers Multi Chip Module Deposited

6. Pixel 2000, Genova Ch.Grah University of Wuppertal 6 MCM-D Module

7. Pixel 2000, Genova Ch.Grah University of Wuppertal 7 Advantages of modules in MCM-D technology äA robust, “easy-to-handle” module with bump-bonding as the only interconnection technique äSignal lines in µ-strip configuration, so with low crosstalk and well defined impedance äAllows routing in the pixel cell array to contact sensor and electronic cells which are not facing each other

8. Pixel 2000, Genova Ch.Grah University of Wuppertal 8 Schematic Cross-Section of a Bus System

9. Pixel 2000, Genova Ch.Grah University of Wuppertal 9 Some pictures of the MCM-D structures Feed-throughs signal bus power contact 50  m

10. Pixel 2000, Genova Ch.Grah University of Wuppertal 10 Feasibility Studies 10 äJust two exemplary plots äThe sensor properties are not affected by the MCM-D technology

11. Pixel 2000, Genova Ch.Grah University of Wuppertal 11 Yield Test - Thin Film Feed-through structures äDaisy-Chain interconnection äFour copper layers ä 1.1. 10 6 monitored vias with a diameter of 25µm äMeasured defect rate 8.13. 10 -6 (9 defects of 1 105 920 vias) äWe expect 1.5 unconnected pixel/module BCB etched for better visualisation

12. Pixel 2000, Genova Ch.Grah University of Wuppertal 12 Full Scale Prototype Module Frontend Chips Additional test pads contacted by wire bonding MCC

13. Pixel 2000, Genova Ch.Grah University of Wuppertal 13 Threshold and Noise (Untuned Full Scale Module) The MCM-D Module shows encouraging performance regarding Threshold distribution and Noise performance Module: MCM-D T1/Frontend B

14. Pixel 2000, Genova Ch.Grah University of Wuppertal 14 Single Chip Module Sensor cell array + MCM-D interconnections + Frontend chip äInvestigation of different Feed-through layouts, especially routing A Single Chip Module consists of: Picture: Frontend C on Single Chip PCB

15. Pixel 2000, Genova Ch.Grah University of Wuppertal 15 Feed-throughs in different layouts Class U (most common class) Class R 1 (to neighbouring pixel cell) Class R 2 (skipping one cell) Class R 3 (skipping two cells) Class U 400/600 (two columns at the border of the hybrid)

16. Pixel 2000, Genova Ch.Grah University of Wuppertal 16 Threshold distribution (Single Chip) Hybrid: MCM-D ST1/Frontend C

17. Pixel 2000, Genova Ch.Grah University of Wuppertal 17 Noise distribution (Single Chip) Hybrid: MCM-D ST1/Frontend C

18. Pixel 2000, Genova Ch.Grah University of Wuppertal 18 Summary of Noise measurements There is no influence on the performance, due to Feed-throughs in MCM-D. As expected, the crossing of copper lines in different layers (classes Ri) increases the Noise, due to the higher interpixel capacitance.

19. Pixel 2000, Genova Ch.Grah University of Wuppertal 19 Crosstalk Measurements Pixel N (masked to read out) Pixel N+1 (with threshold T) For Pixel N+i similar Crosstalk = T / Q Q hits Crosstalk = fraction of charge that couples into the neighbouring pixel through the interpixel capacitance

20. Pixel 2000, Genova Ch.Grah University of Wuppertal 20 Crosstalk distribution (Single Chip) Ri U 600 “ganged” Pixel: These electronic cells are connected to two sensor cells (by design).

21. Pixel 2000, Genova Ch.Grah University of Wuppertal 21 Summary of crosstalk measurements Note 1: There is no influence on the crosstalk, due to the Feed-throughs in MCM-D. Note 2: The performance of class R1 and R2 layouts is comparable to the 600µm long sensor cells (U 600 ).

22. Pixel 2000, Genova Ch.Grah University of Wuppertal 22 Source measurement Am 241 : Gamma-rays nr of hits Upper 3 cells not connected (by design) The MCM-D hybrid shows a uniform functionality. Defects were recognized as bad bump connections.

23. Pixel 2000, Genova Ch.Grah University of Wuppertal 23 Testbeam data äH8 Testbeam at SPS (CERN)  primary: 450 GeV protons äData was mainly taken with: 180 GeV pions äTelescope with 4 layers of strip-detectors (Resolution: 3 µm) H8 Telescope system All presented measurements: (MCM-D) SSG/Frontend B

24. Pixel 2000, Genova Ch.Grah University of Wuppertal 24 Reconstructed energy deposition No charge loss can be seen, due to the MCM-D structures Single hit events Double hit events (added charges) Conventional hybrid MCM-D hybrid

25. Pixel 2000, Genova Ch.Grah University of Wuppertal 25 Single hit resolution Conventional hybrid MCM-D hybrid Difference between predicted (Telescope) and measured particle track P2: sigma of gaussian tail P3: width of plateau

26. Pixel 2000, Genova Ch.Grah University of Wuppertal 26 Double hit resolution Conventional hybrid MCM-D hybrid Double hit resolution: 5µm (conventional and MCM-D hybrids)

27. Pixel 2000, Genova Ch.Grah University of Wuppertal 27 Multi Chip Module-Deposited Conclusion äIt is possible to build “easy-to-handle” Pixel Detector Modules with the MCM-D technique. äThe Sensor is not harmed / damaged by the processing. äThe signal and power distribution structures are able to drive full modules. äNo problems appeared due to the necessary connections between electronic and sensor cells. Outlook: äExplore the full potential of the MCM-D technique, modules with a homogeneous resolution may be build.

28. Pixel 2000, Genova Ch.Grah University of Wuppertal 28 Further possibilities of the MCM-D technology The possibility of integrating passive components in MCM-D is under investigation. R and C: Currently possible (due to the high process temperature this is not (yet) possible for our application!): 720 pF/mm 2 with Ta 2 O 5 as dielectric 10-100 /  with TaN as resistor material Inductor in MCM-D Technology