The Strategic R&D Programme on Technologies for Future Experiments of the EP Department (EP R&D) develops the detector technologies needed for the next generation of High Energy Physics experiments and recently concluded phase I (2020-2023) of its ambitious work plan with a close-out report [1]. A second phase (2024-2028) has been launched [2] and represents CERN’s main scientific contribution to the European Committee for Future Accelerators (ECFA), Detector Research and Development (DRD) collaborations.
Among the pursued detector technologies, advanced semiconductor detectors are key, and are covered by four Work Packages in the EP R&D programme: hybrid (WP 1.1) and monolithic (WP 1.2) silicon pixel detectors, silicon modules (WP 1.3), and, the subject of this article, a work package on the characterization, modelling, and radiation hardening of silicon detectors (WP 1.4).
The focus of the activities within Work Package 1.4 is on developing and providing characterization tools and modelling approaches for semiconductor detectors under development in the community, as well as the fundamental understanding of radiation damage to detectors and the underlying physics processes. This work also entails the optimization, modelling, and radiation hardening of sensors for strong radiation fields.
The WP 1.4 team consists of about ten students and post-docs, as well as CERN EP-DT staff contributing part-time to the activity. The team collaborates closely with several external groups through projects in the framework of the RD50/DRD3 collaboration and the European AIDAinnova project.
Activities - Highlights
We report here on a selection of highlight activities, while a comprehensive overview can be found in the last annual report and the publications cited there [1]. Below, we first describe the work on radiation damage and mitigation techniques and then highlight some of the characterization techniques and tools we are developing and using. The sections are written in a stand-alone readable form, but all activities are inherently connected: on the one hand, the need to develop new characterization tools arises from the need to develop and understand new devices with increasing complexity, and on the other hand, innovation in the area of characterization can foster a deeper understanding, resulting in new and innovative device developments.
Radiation Damage Simulations
The silicon detectors in the inner layers of the LHC experiments, the future HL-LHC experiments, and even more in future hadron-collider experiments are exposed to a large particle flux originating from the interaction points of the circulating particle beams. The passage of those particles through the detector material results in radiation damage to the detectors in the form of performance degradation and, in the most severe cases, failure of detectors [3, 4]. To develop mitigation techniques, it's crucial to understand radiation damage at both microscopic (material) and macroscopic (device/system) levels to ensure long-term detector operation.
Within the work package, we performed a profound study on the Non-Ionizing Energy Loss (NIEL) Hypothesis, which states that radiation damage produced by different particles with varying energies can be predicted by summing up the energy that those particles give in the form of displacement damage to the crystal structure of the detector material. This classical concept forms the basis for all radiation damage predictions in today’s experiments, and has proven to work very well in most cases. However, there are some exceptions (i.e., NIEL Hypothesis violations) that have been experimentally observed and cannot be explained without studying the types of defects formed in the crystal structure of the detector bulk.
A comprehensive simulation was performed: using Geant4, the primary particle silicon interaction (coulombic, elastic, inelastic) was studied for various particles and energies. The resulting primary knock-on atom (PKA) recoil energy distributions for silicon and the nuclear fragments were then simulated in TRIM (Transport of Ions in Matter), resulting in initial atomic displacements. The latter is shown in Figure 1 for two exemplary events with PKAs of (a) 10 keV and (b) 100 keV. Finally, either a custom-made program based on the OPTICS clustering algorithm or a custom-developed kinetic Monte Carlo atomic movement simulation was used to predict the generation of different defect types.
Within the project, a comprehensive set of PKA distributions for protons, neutrons, electrons, and gammas of various energies was achieved. With these, the classical NIEL displacement damage functions could be fully reproduced and improved for very high particle energies.
Figure 1: Distribution of displaced silicon atoms after the generation of a single Primary Knocked-On atom (PKA) with (top) 10 keV and (bottom) 100 keV at the position 0-0-0 in the displayed coordinate system. The displaced atoms are named ”Insterstitials” and the remaining empty lattice sites ”Vacancies”. The given units are in form of multiples of the inter-atomic distance in the silicon crystal with a lattice constant of 2.35 Ångström (1˚A= 0.1 nm = 10−10m).
As a further outcome, a separation of the produced defects into point-like and clustered defects was achieved, which is now used to better understand the aforementioned NIEL Hypothesis violations. One prominent example is the significantly enhanced radiation damage in LGAD timing detectors when exposed to protons compared to neutrons. A PhD thesis providing more details on this work is in preparation [5].
Defects in Silicon and Silicon Carbide Detectors
The radiation-induced performance degradation of solid-state detectors can most often be traced back to electrically active defects in the crystal structure of the bulk material. In our projects, we use the Thermally Stimulated Currents (TSC) and Deep Level Transient Spectroscopy (DLTS) techniques to identify these defects and characterize their properties. Figure 2 provides an example of DLTS measurements on electron- and gamma-irradiated p-type silicon detectors, along with a photo of the experimental setup used to acquire them.
The spectra show the most prominent defects, including the Carbon-interstitial-Oxygen-interstitial (CiOi) defect, which is a donor-type defect, and the Boron-interstitial-Oxygen-interstitial (BiOi) defect, which is an acceptor-type defect. They have different impacts on detector performance. In general, we can state that the CiOi defect is not harmful, as it is not charged at room temperature, while the BiOi defect has a detrimental impact. It is negatively charged at room temperature, and the formation of one BiOi defect removes one of the boron atoms (Bs) that have been intentionally introduced into the crystal (i.e., a dopant) to achieve the desired conductivity. Therefore, the formation of BiOi should be avoided, whereas the formation of CiOi is not harmful.
Figure 2: (a) The experimental setup used for DLTS and TSC measurements in the temperature range from 10K to 300K. The inset shows the sample holder ceramics with a silicon sensor mounted. The DLTS spectra for irradiated silicon sensors after hole injection (b) and after injection of electrons and holes under forward bias (c) are shown. More details can be found in the text and in [6].
This finding led to the idea of adding a high carbon concentration to the sensitive boron doping of devices (e.g., LGAD gain layers). With this defect-engineering approach, more CiOi and fewer BiOi defects are formed, as both defects are generated in the same fundamental defect formation chain. The approach has proven to be highly successful and has led to the production of LGAD sensors that can survive the high radiation fields in the HL-LHC timing layers, whereas their non-carbon-enriched counterparts would fail due to radiation damage.
Our team was deeply involved in the development of the radiation-hard LGAD devices and is still working on further improvements to the technique, as well as the development of new LGAD timing detector technologies [7]. Additionally, we apply our expertise in studying defects in silicon carbide devices. These devices are currently under study, as their wide bandgap makes them an interesting candidate for future applications in high-radiation or high-temperature environments. Unlike silicon sensors, the radiation-induced increase in leakage current does not play a significant role in these devices.
Two Photon Absorption - Transient Current Technique
The Transient Current Technique (TCT) is a widely used tool to study signal formation in silicon detectors, probe features like charge carrier lifetime, or even evaluate the electric field inside the bulk of the device. In this technique, short laser pulses, typically lasting a few hundred picoseconds, are used to generate excess charge carriers along the optical path inside the detector volume. These carriers then drift inside the detector towards the electrodes, thereby generating the detector signal—i.e., they produce a current transient. This process is very similar to the detection mechanism in the LHC silicon detectors, where traversing charged particles create the excess charge carriers.
The Two Photon Absorption - TCT (TPA-TCT) builds upon the TCT technique but offers true three-dimensional spatial resolution in the deposition of the charge carriers, and accordingly, in the mapping of device parameters. The technique was pioneered at CERN, born from a Knowledge Transfer grant to produce the very first tabletop TPA-TCT system. The setup (see Fig. 3) was commissioned in 2021 and has since been continuously improved. A pulsed fiber laser with a wavelength of 1550 nm and a temporal pulse width of 430 fs was developed with an external company. Silicon is fully transparent to light at this wavelength, and under normal conditions, the light would pass through the detector without any effect. However, by confining the light with a highly focused objective into a tiny volume of about 70 m and by restricting the time into the femtosecond regime. The photon density becomes so high in the focal plane that two photons can be combined to deliver enough energy to create an electron-hole pair in the semiconductor (i.e., excess charge carriers). The result is a TCT technique with true three-dimensional spatial resolution, in contrast to all previously available TCT techniques, which were based on standard light absorption (i.e., SPA – Single Photon Absorption).
Figure 3: The TPA-TCT setup at the Solid State Detectors (SSD) lab of the EP-DT group.
An example of a TPA-TCT measurement on a picoAD sensor is given in Fig. 4. The picoAD is a multi-junction monolithic silicon pixel detector under development by the Monolith project. The example visually demonstrates that even in a sensor only 15 microns thick, differences in charge collection can be observed when depositing charge close to the collecting electrodes (Fig. 4a) or close to the backside of the sensor (Fig. 4b). Further experiments performed with the TPA-TCT setup involve in-depth characterization of electric fields in various silicon sensors, including radiation-damaged sensors, testing of Single Event Upsets (SEU) in readout electronics, and gain measurements in Low Gain Avalanche Detectors (LGADs). These are documented in a PhD thesis [8].
Figure 4: xy-scan of a picoAD sensor at two different depth (z-coordinate) within the 15 μm thick device; close to the electronics on the top side (a) and close to the back side of the sensor (b).
Caribou DAQ
The Caribou flexible DAQ system for laboratory and test-beam measurements of silicon detectors [9] has been developed, maintained, and extended through a collaborative effort within WP 1.4, the RD50/DRD3, and AIDAinnova collaborations.
The universal Control and Readout (CaR) board constitutes the core of the modular Caribou hardware (Fig. 5) and provides resources and interfaces for a variety of detector prototypes connected via custom-designed chip boards. The DAQ FPGA firmware (Boreal) and software (Peary) are executed on a commercial Zynq ZC706 System-on-Chip (SoC) board connected to the CaR board.
The modular hardware, firmware, and software architecture allows for efficient integration of new devices, reusing developments from already existing implementations in the Caribou community, thereby minimizing development time and costs for the participating projects. This also applies to hardware and software integration in the major test-beam telescope systems at DESY and CERN, which require only minimal adaptations for new devices.
The flexible Caribou structure enables a wide variety of applications in different domains. Example use cases beyond pixel-detector R&D include the implementation of a Time-to-Digital-Converter (TDC) for picosecond timing measurements [10] and a test system for Analog-to-Digital Converters (ADCs) interfaced with the Caribou hardware [11].
To date, more than 15 different pixel-detector prototypes with dedicated chip boards have been interfaced with Caribou. Two productions coordinated by WP 1.4 have resulted in more than 50 CaR boards being delivered to the 14 participating institutes.
Ongoing work within the Caribou developer community focuses on upgrading the system to the high-performance Ultrascale+ FPGA platform and on a more compact and cost-efficient System-On-Module (SoM) hardware concept (CaR board 2.0), which is currently under design at Brookhaven National Labs.
Figure 5: Caribou DAQ hardware setup with a CLICpix2 device under test.
Test-beam infrastructure, reconstruction, and simulation
The WP 1.4 scope includes support and further development of several hardware and software tools for test-beam characterizations, which were initially developed within CERN’s Linear Collider Detector project [12].
The pixel-detector testing activities pursued in WP 1.4 are centered around a high-rate test-beam telescope installed in the SPS H6 beam line [Fig. 6]. The telescope is based on a system of Timepix3 reference tracking and timing planes. The high-rate capability of the telescope (up to 40 Mhits/cm²/s) and its high precision (< 2-micron pointing resolution) enable efficient testing of small-scale high-resolution pixel-detector technology demonstrators.
The setup is continuously being improved and extended to keep up with the increasing demands of new detectors developed in the EP R&D silicon work packages and collaborating projects. Recent additions include picosecond timing planes based on Micro-Channel-Plate Photo-Multiplier Tubes (MCP-PMT), as well as an air-flow cooling system for the telescope planes and the devices under test [13]. Future gradual upgrades to the telescope system are foreseen in the coming months. A key focus of the 2025 activities will be on preparations for the LHC Long Shutdown 3 from 2026, when beam tests at CERN are expected to be on hold for two years and alternative testing infrastructure needs to be in place.
Figure 6: Aerial view of the WP 1.4 beam telescope setup in the SPS H6 beam line (left) and a close-up of an H2M 65 nm CMOS sensor device under test, mounted on a translation and rotation stage between three upstream and three downstream Timepix3 reference planes (right).
In addition to the testing hardware infrastructure, the corresponding software tools for data reconstruction and simulation are also part of the project scope. Participants of WP 1.4 have contributed significantly to the code base of the Corryvreckan test-beam reconstruction and analysis framework [14, 15] and continue to provide support to new users. The tool has since found widespread use in the pixel-detector test-beam community.
Strong contributions have also been made to the development of the Allpix-Squared simulation toolkit, which uses detailed TCAD simulation results as input to high-statistics Geant4-based Monte Carlo simulations [16]. The achievements of WP 1.4 include the validation of transient simulations for monolithic CMOS sensors based on test-beam data [17, 14].
Future plans
The rich R&D program of EP-RD WP1.4 will continue for the next 4 years as described in the EP-RD phase 2 (2024-2028) planning document [2].
In the context of radiation damage to solid-state detectors, further simulations and measurements to extend, improve, and/or revise the NIEL scaling hypothesis are underway. Experimental data are to be gathered for silicon and silicon carbide-based detectors and LGAD sensors, and a particular project in the context of the DRD3 collaboration will be the impact of extreme fluence exposures up to 10¹⁸ n/cm².
For the characterization tools within the Solid State Detectors (SSD) lab of the EP-DT group, a setup to measure defects in wide bandgap semiconductors is under construction. For the TPA-TCT technology, it is planned to guide the laser beam into a cryostat to allow for very low-temperature measurements and to develop a dispersion correction using a Spatial Light Modulator (SLM) technology. In parallel, we explore the potential of TPA-TCT (with, however, a different wavelength) for silicon carbide-based detectors.
The pixel-detector DAQ and characterization infrastructure and tools will continue to evolve towards higher measurement performance, keeping up with the ever-increasing demands of the testing activities.
References
[1] G. Aglieri et al., Strategic R&D Programme on Technologies for Future Experiments - Annual Report 2023 and Phase-I Closeout, 2024. CERN-EP-RDET-2024-001.
[2] C. Joram et al., Extension of the R&D Programme on Technologies for Future Experiments., 2023. CERN-EP-RDET-2023-001.
[3] Michael Moll, Displacement Damage in Silicon Detectors for High Energy Physics, IEEE Transactions on Nuclear Science, Vol. 65, No. 8, August 2018, 1561-1582 https://doi.org/10.1109/TNS.2018.2819506
[4] CERN Yellow Report Vol. 1 (2021), Radiation effects in the LHC experiments: Impact on detector performance and operation, https://doi.org/10.23731/CYRM-2021-001
[5] Vendula Subert, PhD thesis 2024 (in preparation), University of Hamburg.
[6] Anja Himmerlich, Nuria Castello-Mor, Esteban Curras Rivera, Yana Gurimskaya, Vendula Maulerova-Subert, Michael Moll, Ioana Pintilie, Eckhart Fretwurst, Chuan Liao, Jörn Schwandt; Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment, Volume 1048, March 2023, 167977 https://doi.org/10.1016/j.nima.2022.167977
[7] Esteban Curras, Albert Doblas; Marcos Fernandez, David Flores, Javier Gonzalez, Salvador Hidalgo, Richard Jaramilo, Michael Moll, Efren Navarrete, Giulio Pellegrini, Ivan Vila; Nuclear Instruments and Methods in Physics Research Section A: Accelerators Spectrometers Detectors and Associated Equipment, Volume 1055, October 2023, 168522; https://doi.org/10.1016/j.nima.2023.168522
[8] Sebastian Pape, Characterisation of Silicon Detectors Using the Two Photon Absorption – Transient Current Technique, PhD thesis 2023. CERN-THESIS-2023-336.
[9] Tomas Vanat et al., Caribou - A versatile data acquisition system, PoS TWEPP 2019, 2020, https://doi.org/10.22323/1.370.0100.
[10] E. Buschmann, Status and recent extensions of the Caribou DAQ system for picosecond timing with an FPGA TDC, JINST 18 C02005, 2023, doi:10.1088/1748-0221/18/02/C02005.
[11] A. Quinn, An Open-Source Framework for Rapid Validation of Scientific ASICs, FERMILAB-CONF-24-0206-ETD, 2024, arXiv:2406.15181.
[12] D. Dannheim, K. Krüger, A. Levy, A. Nürnberg, E. Sicking (eds.), Detector Technologies for CLIC, CERN–2019–001, 2019, doi:10.23731/CYRM-2019-001
[13] J. Braach, Performance Evaluation of the FASTPIX Silicon Pixel Sensor Technology Demonstrator for High-Precision Tracking and Timing, PhD thesis, Hamburg University, to appear, 2024.
[14] K. Dort, Simulation Studies and Characterisation of Monolithic Silicon Pixel-Detector Prototypes for Future Collider Detectors & Unsupervised Anomaly Detection in Belle II Pixel-Detector Data, PhD thesis, Giessen Univ., 2022, https://cds.cern.ch/record/2813457.
[15] J. Kroeger, Characterisation of a High-Voltage Monolithic Active Pixel Sensor Prototype for Future Collider Detectors, PhD thesis, Heidelberg Univ., 2021, https://cds.cern.ch/record/2784385.
[16] D. Dannheim et al., Combining TCAD and Monte Carlo methods to simulate CMOS pixel sensors with a small collection electrode using the Allpix2 framework, NIMA 964 (2020) p.163784, doi:10.1016/j.nima.2020.163784.
[17] R. Ballabriga et al., Transient Monte Carlo simulations for the optimisation and characterisation of monolithic silicon sensors, NIMA 1031 (2022) p.166491, doi:10.1016/j.nima.2022.166491.