New results from fast timing iLGAD sensor on Timepix4

TL;DR

The paper tackles the HL-LHC need for precise per-pixel timing by evaluating an inverse Low-Gain Avalanche Detector (iLGAD) with 55 μm pitch on a Timepix4 readout. Through lab calibration and a CERN H8 test-beam with a Timepix4 telescope, the study measures a uniform gain of about $G \\approx 4$, an efficiency of $99.6\%$, and timing improvements from ~750 ps to ~$377 \\pm 7~\\mathrm{ps}$ after timewalk and per-pixel corrections; grazing-angle measurements further yield ~$359~\\mathrm{ps}$ at a depth near $75~\\mu\mathrm{m}$. The results demonstrate the viability of iLGADs for high-rate timing with Timepix4 and highlight that sensor dynamics dominate the timing performance, suggesting that thinner iLGAD sensors (e.g., ~50 μm) could further enhance resolution. These findings inform future detector designs for HL-LHC timing-enabled tracking systems.

Abstract

With the High-Luminosity Large Hadron Collider (HL-LHC) the number of collisions per bunch crossing increases. To cope with these high rates in the pixel trackers, per-pixel time measurements are required, which implies the need for fast sensors. The inverse Low-Gain Avalanche Detector (iLGAD) is one of the fast sensor options that is being investigated. This paper will show the results of an inverse Low-Gain Avalanche Detector (iLGAD) with a pitch of 55 $μ$m, a thickness of 250~$μ$m and a large-area (2~cm$^2$), bump bonded to a Timepix4 ASIC. Timepix4 has 195~ps time binning on each pixel and therefore an excellent ASIC to test the sensor. The sensor is characterised with radio-active source measurements in the lab, and during beam test at the CERN SPS North Area H8 beamline, where the Timepix4 telescope was used. The telescope has a time reference of 12~ps and a pointing resolution of 2.4 $\pm$ 0.1~$μ$m. The iLGAD shows an almost uniform gain of approximately 4 and an efficiency of 99.6 $\pm$ 0.1\%. Without any corrections the obtained time resolution is about 750~ps. After timewalk and clock corrections the time resolution becomes 377 $\pm$ 7~ps. Grazing angle measurements have been done, which allow to measure the time resolution as function of depth of the charge deposition in the sensor. This provides more insight for the perpendicular time resolution.

Figures (7)

• Figure 1: A schematic view of two pixels of an iLGAD and a regular LGAD. The gain layer is on the backside for the iLGAD. The p+ pixel implant is connected via a bump bond to the ASIC. The traditional LGAD has an isolated area between the pixels where there is no-gain.
• Figure 2: Measured charge spectrum of the no-gain (left, col=0) and gain (right, col=245) pixels in a single column with an $^{241}$Am source. The deposited charge for a bias voltage of 200 V is shown. The mean value from a fit with a Gaussian function is listed in the plots. The uncertainty is the standard deviation of the fit.
• Figure 3: Schematic diagram of the eight telescope planes, the Device Under Test (DUT), and the two MCPs, adapted from timepix4telescope.
• Figure 4: Intra-pixel efficiency of a 250 µ m thick iLGAD at 210 V and a threshold of 1000 e-. The average efficiency is 99.6%.
• Figure 5: Time difference between MCP and Timepix4 hits for a bias voltage of 250V (left) and time resolution versus bias voltage (right) after per pixel correction. The black histogram in the left plot are the timestamps with a ToT cut. The yellow histogram is obtained after a global timewalk correction. The error bars in the plots show the spread in time resolution for different pixels