1. November 3-8, 2002D. Bortoletto - Vertex 20021 Silicon Sensors for CMS Daniela Bortoletto Purdue University Grad students: Kim Giolo, Amit Roy, Seunghee Son Engineering Physicist: Gino Bolla OUTLINE –Design consideration for Pixel sensors for the LHC: p-on-n versus n-on-n and p-stops versus p-sprays –Summary results from CMS Forward Pixel (FpiX) first prototype submission Sintef 1999 (received 2000) –Design improvements and results from Sintef 2001 submission (received 2002) –Irradiation studies up to 10 15 n eq /cm 2 –Barrel sensor design (Tilman Rohe) –Conclusions

2. November 3-8, 2002D. Bortoletto - Vertex 20022 FPIX COLLABORATION PSI (Horisberger) ETH U. Zurich U. Basel IHEP Wien RWTH Aachen US CMS UC Davis Northwestern Fermilab Purdue Johns Hopkins Rutgers Mississippi BARREL 2 Layers, 17(27) Mpixels FORWARD DISKS: 4 disks, 12 Mpixels 1.5<  <2.5

3. November 3-8, 2002D. Bortoletto - Vertex 20023 Design Considerations The LHC detectors will be hybrid pixels –Readout chip is very complex (500 K transistors) –Sensor are simpler (50k diodes) Irradiation changes silicon –Type inversion of the bulk material n  p –Increase of effective doping and full depletion voltage –Complex annealing and anti- annealing behavior –Undepleted bulk becomes high resistive –Increase trapping of signal charge

4. November 3-8, 2002D. Bortoletto - Vertex 20024 Radiation Hardness The CMS pixel design has been optimized for a dose of 6  10 14 n eq /cm 2 Fluence is dominated by  ’s. Oxygenation is expected to be useful Crucial to limit the periods without cooling because of anti-annealing Rose collaboration

5. November 3-8, 2002D. Bortoletto - Vertex 20025 Design Considerations p-on-n n-on-n

6. November 3-8, 2002D. Bortoletto - Vertex 20026 Design considerations p-on-n option –require sensors to be depleted for operation: High voltage after irradiation Complex guard ring design Difficult module construction Possible damage to the chip because of high  V and small distance between chip and the sensor Protection of unconnected pixels may be necessary To reduce trapping small gap between pixels Tilman Rohe pixel 2002

7. November 3-8, 2002D. Bortoletto - Vertex 20027 Design considerations n-on-n option: –Allows operation of undepleted sensors after type inversion –Requires double sided processing More expensive Lower yield Testing with bias grid (Atlas), resistive network (CMS) –N-side pixel isolation P-stops (CMS) P-spray (Atlas) –Design optimized for irradiation Guard rings Unbonded pixel protection Tilman Rohe pixel 2002

8. November 3-8, 2002D. Bortoletto - Vertex 20028 Guard ring Design Guard rings must satisfy two requirements: –Limit the lateral extension of the depletion region –Prevent breakdown at the device edge These goals can be achieved by: –Gentle potential drop towards the edge –Increasing gaps from inner to outer region –Field plates to reduce the field

9. November 3-8, 2002D. Bortoletto - Vertex 20029 Eleven guard ring design (Sintef 1999) After irradiation Guard Ring Performance  = 6  10 14 n eq /cm 2 Before irradiation No breakdown up to 800 V even after irradiation to  = 6  10 14 n eq /cm 2  Guard ring design frozen.

10. November 3-8, 2002D. Bortoletto - Vertex 200210 N-side isolation P-stops –Standard processing for most vendors –Additional mask –Alignment and design rules can lead to large gaps P-sprays –No extra mask –Lower cost –No alignment –Better performance after irradiation Moderated p-sprays –No additional mask –Good performance before and after irradiation

11. November 3-8, 2002D. Bortoletto - Vertex 200211 N-side isolation Charge trapping in Oxyde layer P-stops P-sprays N.I.M. A 377 (1996) 412 0.2 3.0

12. November 3-8, 2002D. Bortoletto - Vertex 200212 N-side isolation Sintef 1999 submission focused on double open p-stop ring (CMS Tracker-TDR baseline) We tested 8 p-stop options. Best designs have open p-stop rings (A, F and G) Opening between p-stops provides resistive network F: Single open ringG: Double open ring 2 A : Double open ring

13. November 3-8, 2002D. Bortoletto - Vertex 200213 P-stop performance Performance was measured before and after irradiation Before irradiation After irradiation IV measurements at -10 C after irradiation show: V bias < 300 V: Normal operation 300V< V bias <550 linear increase of the leakage current from the pixel area (soft breakdown) V bias >550V breakdown Design G T=-10 0 C

14. November 3-8, 2002D. Bortoletto - Vertex 200214 P-stop performance TDR Sensor was connected to prototype chip at PSI. “Soft breakdown” current is draw by few pixels that become noisy at around 300 V Noisy pixels are uncorrelated to missing bond connections

15. November 3-8, 2002D. Bortoletto - Vertex 200215 P-stop performance Design with one open ring (F): –Allows for smaller gaps –Shows improved performance after irradiation –No hard breakdown up to 800 V –Lower slope of leakage current increase after “soft breakdown”  = 1  10 14 n eq /cm 2  = 6  10 14 n eq /cm 2 Design F  = 1  10 14 n eq /cm 2  = 6  10 14 n eq /cm 2 A(TDR) at 300V ~5.0nA/pixel >10nA/pixel G at 300V ~1.9nA/pixel ~5.0nA/pixel F at 300V ~0.5nA/pixel ~4.0nA/pixel T=-10 0 C

16. November 3-8, 2002D. Bortoletto - Vertex 200216 Sintef 2001 submission Wafer Layout: –125x125 Finalize single pixel design (PSI-30 36  40 pixels Honeywell chip) –150x150 to match existing DMIL PSI-43 full size 52  53 pixels chip) –150x100 to match IBM 0.25  m compatible layout 15 wafers  Instrument 5 blades Bulk: (1,0,0) Resistivity=1-2 K   cm, thickness 275  m, several oxygenated wafers

17. November 3-8, 2002D. Bortoletto - Vertex 200217 Single pixel design P-stop Sintef 2001 (received in Summer 2002) submission focuses on single open p-stop. Small modifications: –improve yield (F design baseline). –Reduce inter-pixel regions to improve charge collection efficiency (FM design). –Field plates to improve breakdown FM Field Plate Average Breakdown voltage increases by 200 V

18. November 3-8, 2002D. Bortoletto - Vertex 200218 July 2002: Irradiated 85 structures (single ROC silicon sensors + diodes) at IUCF with 200 MeV protons. –15 pixels sensors and 10 diodes @  = 1x10 14 p/cm 2 –24 pixels sensors + 8 diodes @  = 6x10 14 p/cm 2 –20 pixel sensors + 8 diodes @  = 1x10 15 p/cm 2 We measured the properties of the chips at room T and -10 0 C –Half of the structures have been kept at -7.5 0 C at all time but for a few hours –Half of the structures were annealed for 4 minutes at 80 0 C following the procedure established by the Rose collaboration. Irradiation at IUCF

19. November 3-8, 2002D. Bortoletto - Vertex 200219 Single pixel design P-stop Measurements at T=-10 0 C Dose:1  10 14 n p /cm 2 Depletion voltage:20V Some pixel sensors show increased guard ring current at around 600 V

20. November 3-8, 2002D. Bortoletto - Vertex 200220 Single pixel design P-stop Dose:1  10 14 n p /cm 2 Several sensors showed “breakdown” before irradiation but not after irradiation. The guard current was higher than expected before irradiation

21. November 3-8, 2002D. Bortoletto - Vertex 200221 Single pixel design P-stop Sintef 2001 Dose: 6  10 14 n p /cm 2 Depletion voltage:220V Some pixel sensors show increased guard ring current at around 700 V

22. November 3-8, 2002D. Bortoletto - Vertex 200222 Single pixel design P-stop Sintef 2001 Dose: 1  10 15 n p /cm 2 Depletion voltage >500V Some pixel sensors show increased guard ring current at around 700 V

23. November 3-8, 2002D. Bortoletto - Vertex 200223 Calculate single pixel current increase due to radiation using:  I =  V   =4  10 -17 A/cm 3 (Rose Collaboration) We determine the expected current for  = 1x10 14 p/cm 2,  =6x10 14 p/cm 2 and  = 1x10 15 p/cm 2. –Expectations at -10 0 C for a single pixel  I= 0.85  10 -9,5.09  10 -9, 8.49  10 -9 A –Measurements at -10 0 C, –@V bias =300 V   I: = 0.62  10 -9, 3.59  10 -9, 5.75  10 -9 A –@V bias =500 V   I: = 0.65  10 -9, 3.82  10 -9, 6.10  10 -9 A –@V bias =1000V   I: = 0.78  10 -9, 5.08  10 -9, 7.39  10 -9 A Increase in leakage current

24. November 3-8, 2002D. Bortoletto - Vertex 200224 Performance of p-spray and open p-stop appears to be similar: Increase in leakage current P-spray –18 0 C P-stops –10 0 C

25. November 3-8, 2002D. Bortoletto - Vertex 200225 PSI sensors development PSI has made a submission with CIS, Erfurt, Germany. One wafer contains: –one full size barrel sensor with 150  m  150  m pixels (one open p-stop ring) –one full size barrel sensor with the "1/4 micron“ pitch of 100  m  150  m (p-spray design). –27 sensors with pitch 125  m  125  m to fit the old Honeywell PSI30 chip.

26. November 3-8, 2002D. Bortoletto - Vertex 200226 PSI sensors development Technology options aim to suppress soft breakdown –moderated p-spray (similar to ATLAS design). –"open p-stop" but with p-stop dose starting from 10 14 cm -2 down to 3  10 12 cm -2. Several design options were tried: –p-spray with different gap width 15, 20, 30  m –Standard p-stop –p-stop rotated by 90 0 between pixels. –crosses.

27. November 3-8, 2002D. Bortoletto - Vertex 200227 PSI sensors development PSI has received 10 wafers from CIS + 10 “dummies” (full size sensors are damaged). Five wafers were measured. Good yield on the small sensors (only 2 of 68 were bad V break <V dep +50V). Irradiation and beam test planned

28. November 3-8, 2002D. Bortoletto - Vertex 200228 Conclusions Probe station measurements indicates that the new p-stop design is robust up to fluence of 1  10 15 n eq /cm 2 We are currently evaluating the PSI43 chip 4 sensors wafers have been tested and they will be sent to bump bonding companies in November Beam tests and/or source data will be used to understand noise, and charged collection efficiency of the current design.