TCT-based monitoring of LGAD radiation hardness for ATLAS-HGTD production

TL;DR

This work addresses the need for reliable, large-scale QA of LGAD-based sensors for ATLAS-HGTD under HL-LHC radiation. It introduces a fast TCT-based Irradiation Test conducted on a $1\times2$ LGAD array embedded in a Quality Control Test Structure, extracting $V_{gl}$, $G$, leakage current, and interpad width, and calibrates these against CV and $^{90}$Sr charge measurements. The study defines wafer-acceptance criteria by mapping TCT observables to a Sr-90 charge threshold of 5 fC, demonstrates correlations across two sensor designs, and establishes a multi-stage wafer-acceptance procedure that yielded high pass rates in preproduction. This method enables rapid, wafer-level assessment of radiation hardness, supporting reliable delivery of over 21,000 ATLAS-HGTD sensors with consistent performance under HL-LHC conditions.

Abstract

Production of the High Granularity Timing Detector for the ATLAS experiment at High Luminosity LHC requires over 21000 silicon sensors based on Low Gain Avalanche Diode (LGAD) technology. Their radiation hardness is monitored as a part of the production quality control. Dedicated test structures from each wafer are irradiated with neutrons and a fast and comprehensive characterization is required. We introduce a new test method based on Transient Current Technique (TCT) performed in the interface region of two LGAD devices. The measurement enables extraction of numerous sensor performance parameters, such as LGAD gain layer depletion voltage, LGAD gain dependence on bias voltage, sensor leakage current and effective interpad distance. Complementary capacitance-voltage measurements and charge collection measurements with 90Sr on the same samples have been performed to calibrate the TCT results in terms of charge collection and define acceptance criteria for wafer radiation hardness in the ATLAS-HGTD project.

Figures (8)

• Figure 1: a) Schematics of an LGAD pixel designed for ATLAS-HGTD, illustrating the p$^+$ gain layer embedded between the n$^{++}$ collection electrode and the p-type silicon bulk. At the pixel edge the n$^{++}$ structure is extended to terminate the p-n junction and prevent electrical breakdown, and a p-stop structure is included for inter-pixel isolation (dimensions not to scale). b) Concept of the TCT measurement between two LGAD pixels used in the Irradiation Test. A focused laser beam enters the structure from the top and deposits charge either within the LGADs (Gain $>1$) or in the interpad region (PIN) with no gain ($\mathrm{G}=1$).
• Figure 2: a) Schematics of a Quality Control Test Structure (QCTS). The $1\times2$ LGAD array used in IT is on the far right side. The zoomed in $1\times2$ LGAD structure is shown for: b) IHEP-IME with a single optical window; and c) USTC-IME with three optical windows.
• Figure 3: a) Example LGAD transient signal waveform from the QCTS. In the TCT measurements the charge is extracted by integration over the interval [0, $3\,\mathrm{ns}$]. In the $^{90}\mathrm{Sr}$ measurement the charge is extracted from the waveform maximum in the interval [$-2\,\mathrm{ns}$, $5\,\mathrm{ns}$], and is converted to charge using a calibration factor of $80\,\mathrm{fC}/\mathrm{V}$. b) LGAD charge spectrum from the $^{90}\mathrm{Sr}$ measurement. The signal peak is fitted with the Landau-Gaussian function. The peak at $0\,\mathrm{fC}$ corresponds to triggered events where the particle does not pass the sensitive volume, and is related to noise level (typically $\sigma = 0.7\,\mathrm{fC}$). Its displacement from 0 is due to the bias in signal sampling.
• Figure 4: a) TCT charge collection profiles measured in an IHEP-IME sample at different bias voltages (the $y$-scale at $100\,\mathrm{V}$ is extended by a factor of 4 for better visibility). The interpad region is centered around $y=190\,\upmu\mathrm{m}$. The LGAD and PIN signals are fitted with constant functions in the corresponding intervals. At $y<50\,\upmu\mathrm{m}$ and $y>350\,\upmu\mathrm{m}$ no signals are generated in the sensor due to beam clipping on surface metallization. b) Charge profile in a USTC-IME sample. Three optical windows are scanned to measure charge in each LGAD and the PIN. Dots in both profiles at $V_\mathrm{bias}=100\,\mathrm{V}$ indicate the extracted rising and falling edges used for defining fit intervals and the arrows indicate the interpad distance.
• Figure 5: Example of TCT IT results on a QCTS sample irradiated to $2.5\times10^{15}\,\mathrm{n}_\mathrm{eq}\,\mathrm{cm}^{-2}$: a) Dependence of the LGAD signal size on $V_\mathrm{bias}$ in each LGAD pad in the $1\times2$ array. LGAD1 denotes the pad further away, and LGAD2 denotes the pad closer to the single cell LGAD on the QCTS. Intersections of the linear fits are used to determine $V_\mathrm{gl}$ (around $20\,\mathrm{V}$) and LGAD full depletion voltage (around $70\,\mathrm{V}$). The insets are focused on the region around $V_\mathrm{gl}$; b) PIN diode signal from the interpad region and extraction of the PIN full depletion voltage from a linear fit intersection; c) LGAD gain dependence on $V_\mathrm{bias}$ as the ratio of the LGAD and PIN signals (markers for LGAD1 and LGAD2 coincide almost completely).