Bistable damage Si C meeting 28 09 2012

Bistable damage Si. C meeting, 28. 09. 2012.

Motivation N. Pacifico (Ph. D thesis, 2012) – 300 μm, 1 x 1016 MPP/HLL 150 μm, 5 x 1015 n/cm 2 • Bistable behavior (CM quenching) noticed after long-term annealing and current injection in FZ n+-p devices. • Significant drop in (expected) collected charge and leakage current in MPP/HLL 150 μm thin detector, after keeping it at Vbias for long period of time. • Reports of bias voltage effects on space charge in irradiated silicon detectors [NIMA, 1998 -2002, 2008]

MPP/HLL, 150 μm, 5 x 1015 n/cm 2, tann=40960 min @60ºC • • • Leakage current noticeable drop while searching for focus at 500 V. After the first CC measurements (Take 1), it was decided to repeat due to still present current drop. The next 2 measurements (Take 2&3) show extreme drop (nearly six-fold) in collected charge. After leaving the detector overnight (~12 h), the CC went back up. The rest were attempts to achieve highest CC, depending on duration of the measurements (voltage step and averaging), as well as the time of detector exposure to room temperature, with no bias applied.

MPP/HLL, 75 μm, 5 x 1015 n/cm 2, tann=20480 min @60ºC • The 75 μm sample annealed to 20480 min showed a very large increase (nearly tenfold!) in charge collection, owing to charge multiplication. • The detector seems to have larger thickness due to its very high sensitivity and relatively large distance (~2 mm) of the connected strip from the detector edge, causing signal before the beam waist (FWHM≈7 μm) entered the region below the strip. The signal before the entrance is due to reflections from Al housing. • However, after keeping the detector for more than 24 h at Vbias= 500 V a nearly sixfold decrease in collected charge was observed. • Before the onset of CM (~500 V) it was even lower than immediately after irradiation (this is due to increased space charge with annealing and consequent need for higher bias)

HPK-A 5, FZ, 300 μm, 5 x 1015 n/cm 2, tann=10240 min @60ºC • ALIBAVA readout system was used for study because it was possible to perform direct noise measurements as well. (St-90, rate~330 Hz, system gain ~400 e/ADC) • After 10240 min at 60ºC CC climbs to even ~30 k electrons of MPV (~42 k mean) at Vbias = 1000 V, comparing to ~11 k after only 80 min at same temperature. • Beneficial effect on the leakage current is also noticeable before higher CM. • However, after keeping the detector at Vbias=800 V for ~240 h, it results in nearly twofold decrease in both collected charge and correspondingly, the leakage current. • The exact time progression of this CC and current drop at 800 V follows:

HPK-A 5, FZ, 300 μm, 5 x 1015 n/cm 2, tann=10240 min @60ºC • Both CC and Ileak seem to fully stabilize after ~10000 min (first kink at ~2000 min!). After this time, obviously only the stable damage has influence on the Neff, which apparently is still sufficient CM, but at a much lower scale. • However, while CC and Ileak experience more than twofold decrease (>56%) in values at this voltage, the SNR drop is less (~41%). Even more interesting is the noise drop, which is even smaller (~31%). • This means that the SNR remains mostly due to larger contribution of CM to the signal, instead of the noise, supporting previous assumptions.

HPK-A 5, FZ, 300 μm, 5 x 1015 n/cm 2, tann=10240 min @60ºC • Since the collisions at LHC experiments run for around 17 h, then pause for 2 h before repeating the cycle without lowering the bias or increasing the temperature, it was interesting to observe what would happen if the detector would be left at -20ºC in this state, with no bias applied. • After 2 h and again after 12 h at -20ºC with no bias applied, no improvement was found. It seems as though this temperature ‘freezes’ the bistable damage effect. • This means that the bistable defects need certain amount of thermal energy in order to excite them again. The exposure time is presumed to be dependent on the temperature.

HPK-A 5, FZ, 300 μm, 5 x 1015 n/cm 2, tann=10240 min @60ºC • • Attempt to calculate activation energy – too unstable (current fall at very high rate). Attempt to restore the previous state (bistable damage ‘deactivated’) – failed*. Again, temperature T=-20 C was tested for ‘freezing’ the effect – confirmed. Activation of bistable damage by leaving the detector for a day at RT – optimal 48 h.

HPK-A 5, FZ, 300 μm, 5 x 1015 n/cm 2, tann=10240 min @60ºC • The detector was kept for a total of ~48 h at RT, with no bias applied, to activate all the bistable defects and repeat the measurements up to 5120 min at 800 V. The pattern was found to be very similar, proving that this damage remains bistable, however with well predictable behavior under bias. • This behavior was tested at higher bias voltage, Vbias=1000 V, and the results again show very similar pattern, but with the SNR remaining very high (the drop was less than 9%!), while the CC and Ileak drop is somewhat smaller than at 800 V (CC: ~30% and Ileak: ~42%). The noise drop was even greater (~36%). All this again confirms the beneficial effect of the CM. • Space charge concentration in the stable state is sufficient to invoke a sufficient amount of CM at high Vbias, enough to greatly overpower the induced noise increase!

HPK-A 5, FZ, 300 μm, 5 x 1015 n/cm 2, tann=10240 min @60ºC • CC/IV measurements for the whole bias range after 5120 min at 1000 V were performed and comparison with the previous measurement at 800 V show remarkable, non-expected resemblance. • The values at 800 V after 5120 min regardless of the voltage at which the detector was being kept (both 800 and 1000 V), show agreement to less than 5%! Keep in mind that the current was 1/3 greater at 1000 V! • This means that the amount of applied bias voltage/current through time does not influence neither the time constants, nor the magnitude of the values at any given stage. • Bistable damage evolution under Vbias=600 V was also examined and cross-checked with the previous results – a match of less than 3% in values was found*. However, the voltage is too low for any significant CM and therefore the SNR drops to < 10.

HPK-A 3, FZ, 300 μm, 1 x 1016 n/cm 2, tann=10240 min @60ºC • E-TCT measurements performed using the detector which was used for charge multiplication study and JINST (2012 JINST 7 P 06007) article. • After finding the focus, the detector was left overnight to re-activate any bistable defects deactivated during this process. The detector was then biased to 600 V (higher voltages produced very unstable leakage current) and CC/IV was performed at successive time intervals up to 5120 min. The results reveal a very similar pattern. As before, there is a significant drop in charge collection until the first kink at ~2000 min and afterwards seem to slowly stabilize (up to 10 k min probably). Greatest lost of charge, as expected, in the active/depleted bulk. • • • The signals however reveal exactly how bistable defects influence charge collection and what actually happens during this process.

HPK-A 3, FZ, 300 μm, 1 x 1016 n/cm 2, tann=10240 min @60ºC After 5120 min@600 V • • Bistable damage affects CM only! –The first peak remains the same! Thermally induced space charge = bistable?

HPK-A 3, FZ, 300 μm, 1 x 1016 n/cm 2, tann=10240 min @60ºC • This is also confirmed if the total charge collection before the bias effect is compared to the results afterwards. • Before multiplication kicks in (Vbias<300 V), there is virtually no difference in collected charge. However, above this point, even though there is some CM (as seen in previous slide), the total charge collected drops for more than 67% at 600 V! (almost twice as much than at 5· 1015 n/cm 2) • This means that the amount of bistable damage developed by thermal annealing is also irradiation fluence dependant. This in agreement with expectations, because neither CM, or bi-stability was experienced at lower fluences (<1· 1015 n/cm 2).

Conclusions • Bistable damage occurs in irradiated and long term annealed silicon detectors giving contribution to manifold increase in CC through CM mechanism (due to increase in Neff), and decrease if the detector is kept under bias for long periods of time. • Most of the bistable defects get de-activated after ~2000 min under bias, regardless of its magnitude. • The amount of the applied bias through time does not influence neither the time constants, nor the magnitude of the values at any given point/voltage during that time. • Bistable damage need certain amount of thermal energy/annealing in order to activate it again (~24 -48 h at RT). It was not possible to de-activate it with high leakage current. Temperature T=-20ºC has a ‘freezing’ effect. • The pattern is fully repeatable under any bias/temperature. • Higher voltage causes CM even in ‘stable’ state, therefore less impact on CCE, noise and SNR. • CM is shown to have more beneficial effect to CC than to increase of noise. • Bistable defects affect CM only (no change in signal peak from primary generated charge).

Backup