biasdependentclustercentroidshiftsintheLHCbVELO-CERN偏置依赖聚类质心的变化在LHCb腭-欧洲核子研究中心.ppt

1、Observations of anomalous(?) bias dependent cluster centroid shifts in the LHCb VELO detector.,Aras PapadelisNIKHEF Vertex 2005, Nikko, Japan,Outline,Silicon sensors for the LHCb VErtex LOcator. Sensors and FE electronics. Silicon specs. Beam test results Motivation Observed bias dependency of recon

2、structed cluster position. Observation of possible ballistic deficit in VELO. Anomalous centroid shifts.,The sensors in the VELO detector,Quick reminder(for more details, see talk of L.Eklund ) R and measuring sensors, 2048 strips per sensor Analogue FE read out done by the Beetle chip. 25 ns peakin

3、g time 128 strips read out in parallel.,Some silicon specifications,VELO sensor is of ”normal” design, no surprises are expected. Drift times well below 25 ns.,Background,Sept 2004: Steve Biagi (LHCb Liverpool) suggests that ballistic deficit in non-irradiated sensors can give offsets in reconstruct

4、ed cluster position for angled tracks. Subsequently investigated in VELO beam tests of Nov 2004.,Test beam setup,Hamamatsu R sensor (Telescope) Hamamatsu Phi sensor (Telescope) PR04 200 micron R sensor PR04 300 micron R sensor,Beam,CERN X7 SPS 120GeV pions.,Took data at 10, bias voltages 100V, 200V

5、and 300V. (Vfd is 34V and 55 V),Bias dependent mean residual shift,We are looking for a shift in the mean value of the track residual distribution. Align system at 100V Study mean residual at different bias voltages.,Residual (cm),Residual (cm),100 V,300 V,200 V,200 m,300 m,Residual shift vs radius,

6、Clear systematic shift in residuals between different voltages. Same size for both 200 and 300 m thick sensors. (Effect first observed by J. Palacios, CERN) ”Banana bend” for 200 m at high pitch, possibly due to warped sensor. Looks like sensors are misaligned!,200 m,300 m,8m,9m,pitch,Further checks

7、.,Widths of residual distribution more or less independent of bias voltage. -distributions look as expected for a misaligned detector.,200 m,300 m,Safety check,Data taking was done in one sequence during 10 hours.,Time ,200 m,300 m,VERY CLEAR: No time dependence big enough to cause misalignment,Ball

8、istic deficit?,Signal increases between 100V and 300V. Is this ballistic deficit? Also: 300V arrives about 2 ns before 100 V.,R300 R200,R200,Publications on ballistic deficit,Charge collection efficiency studies with irradiated silicon detectors. (Allport et al.) NIM A 501 (2003) 146-152,Similar res

9、ults, the ballistic deficit at 100 V is small.,Signal loss due to timing,*Simplified* way of viewing ballistic deficit. Assume: no charge trapping (unirradiated) homegenous E-field same everywhere in sensor. Electrons in the grey region do not contribute fully to the signal. shift in the cluster cen

10、troid! For a 300 m sensor, a 8 m shift corresponds to a 30 % signal loss between 300V and 100V. This is not what we observe! Sensor thickness dependence should be seen! Simulation needed!,Cluster centroid shift,300V,100V,n+,n-bulk,w,p+,NB: Proportions are not correct!,So, whodunnit?,Slowly moving se

11、nsors? NO! Sensor movement due to change in voltage? VERY UNLIKELY! Ballistic deficit? NO! Unclear if the signal loss that can be observed is ballistic deficit. Signal loss observed is not big enough to cause observed shift. No width dependence can be seen Looks like a surface effect Caused by chang

12、es in drift field when going from 100V to 300V? Effects related to shape of weighting field? But! Why is no pitch dependence seen? (Remember: strip width/pitch=constant) Timing effect Is the 2 ns peak shift big enough to cause 10 m shift?.,Conclusion,We see a distinct and systematic shift in mean residuals at 300V compared to 100V bias voltage. We have not understood it yet, despite a lot of effort. The question remains open Su