No Arabic abstract
Low Gain Avalanche Detectors (LGADs) are silicon sensors with a built-in charge multiplication layer providing a gain of typically 10 to 50. Due to the combination of high signal-to-noise ratio and short rise time, thin LGADs provide good time resolutions. LGADs with an active thickness of about 45 $mu$m were produced at CNM Barcelona. Their gains and time resolutions were studied in beam tests for two different multiplication layer implantation doses, as well as before and after irradiation with neutrons up to $10^{15}$ n$_{eq}$/cm$^2$. The gain showed the expected decrease at a fixed voltage for a lower initial implantation dose, as well as for a higher fluence due to effective acceptor removal in the multiplication layer. Time resolutions below 30 ps were obtained at the highest applied voltages for both implantation doses before irradiation. Also after an intermediate fluence of $3times10^{14}$ n$_{eq}$/cm$^2$, similar values were measured since a higher applicable reverse bias voltage could recover most of the pre-irradiation gain. At $10^{15}$ n$_{eq}$/cm$^2$, the time resolution at the maximum applicable voltage of 620 V during the beam test was measured to be 57 ps since the voltage stability was not good enough to compensate for the gain layer loss. The time resolutions were found to follow approximately a universal function of gain for all implantation doses and fluences.
We report results from the testing of 35 {mu}m thick Ultra-Fast Silicon Detectors (UFSD produced by Hamamatsu Photonics (HPK), Japan and the comparison of these new results to data reported before on 50 {mu}m thick UFSD produced by HPK. The 35 {mu}m thick sensors were irradiated with neutrons to fluences of 0, 1*10^14, 1*10^15, 3*10^15, 6*10^15 neq/cm^2. The sensors were tested pre-irradiation and post-irradiation with minimum ionizing particles (MIPs) from a 90Sr b{eta}-source. The leakage current, capacitance, internal gain and the timing resolution were measured as a function of bias voltage at -20C and -27C. The timing resolution was extracted from the time difference with a second calibrated UFSD in coincidence, using the constant fraction method for both. Within the fluence range measured, the advantage of the 35 {mu}m thick UFSD in timing accuracy, bias voltage and power can be established.
Low Gain Avalanche Detectors (LGAD) are based on a n++-p+-p-p++ structure where an appropriate doping of the multiplication layer (p+) leads to high enough electric fields for impact ionization. Gain factors of few tens in charge significantly improve the resolution of timing measurements, particularly for thin detectors, where the timing performance was shown to be limited by Landau fluctuations. The main obstacle for their operation is the decrease of gain with irradiation, attributed to effective acceptor removal in the gain layer. Sets of thin sensors were produced by two different producers on different substrates, with different gain layer doping profiles and thicknesses (45, 50 and 80 um). Their performance in terms of gain/collected charge and leakage current was compared before and after irradiation with neutrons and pions up to the equivalent fluences of 5e15 cm-2. Transient Current Technique and charge collection measurements with LHC speed electronics were employed to characterize the detectors. The thin LGAD sensors were shown to perform much better than sensors of standard thickness (~300 um) and offer larger charge collection with respect to detectors without gain layer for fluences <2e15 cm-2. Larger initial gain prolongs the beneficial performance of LGADs. Pions were found to be more damaging than neutrons at the same equivalent fluence, while no significant difference was found between different producers. At very high fluences and bias voltages the gain appears due to deep acceptors in the bulk, hence also in thin standard detectors.
Several thin Low Gain Avalanche Detectors from Hamamatsu Photonics were irradiated with neutrons to different equivalent fluences up to $Phi_{eq}=3cdot10^{15}$ cm$^{-2}$. After the irradiation they were annealed at 60$^circ$C in steps to times $>20000$ minutes. Their properties, mainly full depletion voltage, gain layer depletion voltage, generation and leakage current, as well as their performance in terms of collected charge and time resolution, were determined between the steps. It was found that the effect of annealing on timing resolution and collected charge is not very large and mainly occurs within the first few tens of minutes. It is a consequence of active initial acceptor concentration decrease in the gain layer with time, where changes of around 10% were observed. For any relevant annealing times for detector operation the changes of effective doping concentration in the bulk negligibly influences the performance of the device, due to their small thickness and required high bias voltage operation. At very long annealing times the increase of the effective doping concentration in the bulk leads to a significant increase of the electric field in the gain layer and, by that, to the increase of gain at given voltage. The leakage current decreases in accordance with generation current annealing.
A new module concept for future ATLAS pixel detector upgrades is presented, where thin n-in-p silicon sensors are connected to the front-end chip exploiting the novel Solid Liquid Interdiffusion technique (SLID) and the signals are read out via Inter Chip Vias (ICV) etched through the front-end. This should serve as a proof of principle for future four-side buttable pixel assemblies for the ATLAS upgrades, without the cantilever presently needed in the chip for the wire bonding. The SLID interconnection, developed by the Fraunhofer EMFT, is a possible alternative to the standard bump-bonding. It is characterized by a very thin eutectic Cu-Sn alloy and allows for stacking of different layers of chips on top of the first one, without destroying the pre-existing bonds. This paves the way for vertical integration technologies. Results of the characterization of the first pixel modules interconnected through SLID as well as of one sample irradiated to $2cdot10^{15}$, eqcm{} are discussed. Additionally, the etching of ICV into the front-end wafers was started. ICVs will be used to route the signals vertically through the front-end chip, to newly created pads on the backside. In the EMFT approach the chip wafer is thinned to (50--60),$mu$m.
Small-pitch 3D silicon pixel detectors have been investigated as radiation-hard candidates for the innermost layers of the HL-LHC pixel detector upgrades. Prototype 3D sensors with pixel sizes of 50$times$50 and 25$times$100 $mu$m$^{2}$ connected to the existing ATLAS FE-I4 readout chip have been produced by CNM Barcelona. Irradiations up to particle fluences of $3times10^{16}$ n$_{mathrm{eq}}$/cm$^2$, beyond the full expected HL-LHC fluences at the end of lifetime, have been carried out at Karlsruhe and CERN. The performance of the 50$times$50 $mu$m$^{2}$ devices has been measured in the laboratory and beam tests at CERN SPS. A high charge collected and a high hit efficiency of 98% were found up to the highest fluence. The bias voltage to reach the target efficiency of 97% at perpendicular beam incidence was found to be about 100 V at $1.4times10^{16}$ n$_{mathrm{eq}}$/cm$^2$ and 150 V at $2.8times10^{16}$ n$_{mathrm{eq}}$/cm$^2$, significantly lower than for the previous IBL 3D generation with larger inter-electrode distance and than for planar sensors. The power dissipation at -25$^{circ}$C and $1.4times10^{16}$ n$_{mathrm{eq}}$/cm$^2$ was found to be 13 mW/cm$^2$. Hence, 3D pixel detectors demonstrated superior radiation hardness and were chosen as the baseline for the inner layer of the ATLAS HL-LHC pixel detector upgrade.