4H-SiC Fast Luminosity Detector

• 4H-SiC fast luminosity detectors are devices built from 4H-silicon carbide, offering a wide bandgap, high breakdown field, and thermal stability for robust operation in harsh environments.
• They employ varied architectures including PIN diodes, graphene-enhanced electrodes, and LGAD-based designs to overcome low charge generation and improve timing precision.
• Implementations in collider systems, such as at CEPC, demonstrate sub-50 ps timing resolution and millisecond-scale luminosity monitoring through optimized readout and simulation techniques.

Searching arXiv for papers on 4H-SiC fast luminosity detectors and related 4H-SiC timing devices. A 4H-SiC fast luminosity detector is a luminosity-monitoring device built from 4H-silicon carbide sensing elements and optimized for fast timing, low leakage current, and radiation-hard operation. In the recent literature, the term covers several closely related implementations: PIN diodes for direct fast charge collection, graphene-optimized PIN structures for improved field uniformity and reduced timing non-uniformity, and LGAD-derived devices that introduce internal multiplication to compensate for the comparatively low charge generation in SiC. These detectors are being developed for high-luminosity collider feedback, precise vertex timing and time-of-flight layers, and other harsh-environment measurements in which timing precision and long-term stability must be maintained under irradiation (Li et al., 31 Jul 2025, He et al., 13 May 2026).

1. Material platform and detector families

4H-SiC is used for fast luminosity detection because it combines a wide bandgap, high breakdown field, fast saturated electron drift velocity, thermal stability, and high atomic displacement energy. Several studies frame these properties explicitly against silicon: silicon detectors often require complex low-temperature systems when operating in irradiation environment, whereas 4H-SiC is investigated for operation under extreme irradiation and high temperature environments (Wang et al., 2023, Švihra et al., 12 Apr 2025).

The central limitation of 4H-SiC for particle timing is not the speed of carrier transport but the comparatively low charge generation in SiC relative to standard silicon. This is why the literature has split into two main device strategies. The first is the fast PIN approach, which relies on thin depleted regions, high electric fields, and low capacitance to maximize slew rate without internal gain. The second is the LGAD approach, which adds a gain layer to provide controlled avalanche multiplication and thereby raises the signal amplitude while preserving fast leading edges (Kráčmar et al., 16 May 2026, Novotný et al., 10 Mar 2025).

Within this landscape, several distinct detector families have emerged. Conventional vertical PIN detectors are used for direct timing and irradiation studies. Graphene-optimized PIN detectors replace or supplement conventional metal electrodes with monolayer graphene in order to homogenize the electric field and reduce timing variation across the active area. LGAD development includes first-generation onsemi devices, segmented strip and pixel geometries, and ultra-thin AC-LGAD concepts studied with WeightField2 for collider conditions (Jiang et al., 28 May 2025, Kalani et al., 23 Jan 2026).

2. Representative device architectures

Recent work spans single-pad timing diodes, pixelated luminosity monitors, graphene-electrode devices, and segmented avalanche detectors. Representative structures and reported results are summarized below.

Architecture	Representative structure	Representative result
PIN detector (He et al., 13 May 2026)	$2 \times 2\ \mathrm{mm}^2$, 50 $\mu$m N-epi, 300 V	40 ps before irradiation; 45 ps after irradiation
Graphene-optimized PIN (Xiao et al., 28 Mar 2026, Jiang et al., 28 May 2025)	$2 \times 2\ \mathrm{mm}^2$, monolayer graphene transparent electrode over the active region	21.2 ps at maximum scanning distance; rise time reduced by 24% at 200 V
CEPC PIN-array monitor (Li et al., 31 Jul 2025)	Two modules, each with 12 pixels of $5\,\mathrm{mm}\times5\,\mathrm{mm}$ and 100 $\mu$m thickness	$\nu=1/\sqrt{2594}\approx2.0\%$ at 1 kHz
Segmented LGAD (Kráčmar et al., 16 May 2026)	Strip pitch 80 $\mu$m; pixel pitches 55 and 110 $\mu$m	40 ps jitter at $M\approx6$; 25 ps at $M\approx10$
Ultra-thin AC-LGAD (Kalani et al., 23 Jan 2026)	20 $\mu$0m 4H-SiC sensor at 243 K	Total $\mu$1 ps; simulated $\mu$2 ps up to $\mu$3

For PIN devices, the canonical structure is a vertical stack on a conductive 4H-SiC substrate. One reported detector consists of a 350 $\mu$4m thick heavily doped n-type substrate, a 50 $\mu$5m lightly doped n-type epitaxial layer with nominal nitrogen doping of $\mu$6, and a 0.6 $\mu$7m P$\mu$8 contact layer aluminum doped to $\mu$9; the device area is $2 \times 2\ \mathrm{mm}^2$0, with 400 nm PECVD SiO$2 \times 2\ \mathrm{mm}^2$1 passivation and Ni/Ti/Al ohmic contacts on both p and n faces (He et al., 13 May 2026).

The CEPC fast luminosity monitor adopts a different geometry. At Position 1, 10 m from the interaction point, two identical modules are mounted at $2 \times 2\ \mathrm{mm}^2$2 mm in the horizontal direction on the 62 mm-outer-diameter vacuum chamber. Each module is an array of 12 PIN-type 4H-SiC pixels with $2 \times 2\ \mathrm{mm}^2$3 pixel size, 100 $2 \times 2\ \mathrm{mm}^2$4m thickness, and active-volume doping of $2 \times 2\ \mathrm{mm}^2$5 (Li et al., 31 Jul 2025).

Graphene-optimized detectors retain a PIN architecture but modify the front electrode. In one implementation, a monolayer graphene transparent electrode is transferred on top of 500 nm PECVD SiO$2 \times 2\ \mathrm{mm}^2$6, while a metal ring and bond pad provide biasing. In another, monolayer graphene replaces metal over the 1.4 mm-diameter central region, leaving a peripheral metal ring for biasing. These structures are intended to suppress the non-uniform field associated with conventional window electrodes (Xiao et al., 28 Mar 2026, Jiang et al., 28 May 2025).

3. Signal formation, charge collection, and timing

The signal model used across the literature is the Shockley-Ramo picture. In the CEPC design study, the induced current is written as $2 \times 2\ \mathrm{mm}^2$7, where $2 \times 2\ \mathrm{mm}^2$8 is the charge, $2 \times 2\ \mathrm{mm}^2$9 the carrier drift velocity, and $5\,\mathrm{mm}\times5\,\mathrm{mm}$0 the weighting field. For timing, the jitter-dominated expression is written as $5\,\mathrm{mm}\times5\,\mathrm{mm}$1 (Li et al., 31 Jul 2025, He et al., 13 May 2026).

For the irradiated 4H-SiC PIN detector studied with $5\,\mathrm{mm}\times5\,\mathrm{mm}$2Sr $5\,\mathrm{mm}\times5\,\mathrm{mm}$3-particles, $5\,\mathrm{mm}\times5\,\mathrm{mm}$4–$5\,\mathrm{mm}\times5\,\mathrm{mm}$5 measurements give effective doping concentrations of $5\,\mathrm{mm}\times5\,\mathrm{mm}$6 before irradiation and $5\,\mathrm{mm}\times5\,\mathrm{mm}$7 after irradiation. The full depletion voltage is $5\,\mathrm{mm}\times5\,\mathrm{mm}$8, and at $5\,\mathrm{mm}\times5\,\mathrm{mm}$9 the depletion width is approximately $\mu$0, corresponding to an approximately uniform electric field of about $\mu$1. At the same bias, the reverse leakage currents are $\mu$2 and $\mu$3 for the unirradiated and irradiated detectors, respectively. In the detailed exposition, these pA-level leakage currents are linked to low parallel noise, with effective $\mu$4 exceeding 20 and supporting jitter below 40 ps (He et al., 13 May 2026).

The corresponding timing measurement uses a $\mu$5Sr source, a Si-LGAD reference channel with known 37 ps resolution at 200 V bias, a UCSC readout board with on-board charge sensitive amplifier, a PE15A1008 main amplifier with 20 dB gain and $\mu$6 GHz bandwidth, and a 25 GS/s, 2.5 GHz oscilloscope. Time pick-off is performed with constant-fraction discrimination at 50% of amplitude, and the detector resolution is extracted from $\mu$7. In this setup, coincidence time resolution broadens from 53.1 ps to 58.5 ps after irradiation, corresponding to single-channel resolutions of 40 ps and 45 ps (He et al., 13 May 2026).

Graphene-based field engineering changes the timing problem from one of pure speed to one of speed uniformity. In UV-TCT measurements at 120 V, a reference ring-electrode detector shows FWHM time resolution degrading from 16.1 ps at point 0 to 38.1 ps at point 5, whereas the graphene-optimized device changes only from 18.4 ps to 21.2 ps. The same study reports that carriers in the graphene-optimized geometry drift vertically to graphene, then laterally in graphene in less than 0.1 ns to the ring, while the noise standard deviation remains stable at approximately 3.40 mV; by contrast, the reference detector noise rises from 3.52 mV to 3.63 mV across the scan. The reported stability improvement in time resolution is 87% (Xiao et al., 28 Mar 2026).

LGAD-based timing pushes the same metrics further by internal multiplication. For segmented 4H-SiC LGADs, TPA-TCT measurements give a 10–90% rise time of about 180 ps and FWHM of about 450 ps, with jitter approximated as $\mu$8. Under $\mu$9 mV RMS, the measured slope is about 25 mV/ps at $\nu=1/\sqrt{2594}\approx2.0\%$0 and about 40 mV/ps at $\nu=1/\sqrt{2594}\approx2.0\%$1, corresponding to about 40 ps and 25 ps jitter, respectively. In ultra-thin AC-LGAD simulations, a 20 $\nu=1/\sqrt{2594}\approx2.0\%$2m 4H-SiC device at 243 K, $\nu=1/\sqrt{2594}\approx2.0\%$3 V, and $\nu=1/\sqrt{2594}\approx2.0\%$4 gives $\nu=1/\sqrt{2594}\approx2.0\%$5, a 10–90% rise time of about 350 ps, jitter of about 7 ps, and total $\nu=1/\sqrt{2594}\approx2.0\%$6 ps (Kráčmar et al., 16 May 2026, Kalani et al., 23 Jan 2026).

4. Luminosity-monitor implementations in collider systems

The most explicit collider implementation appears in the CEPC fast luminosity monitor study. CEPC is described as employing a crab waist scheme to achieve ultrahigh $\nu=1/\sqrt{2594}\approx2.0\%$7 luminosity at Higgs mode, and the detector is designed to support a luminosity-driven dithering system for horizontal beam stabilization. The monitoring principle uses primary radiative Bhabha electrons with average energy about 24 GeV, which strike the beam pipe and initiate electromagnetic showers; secondary $\nu=1/\sqrt{2594}\approx2.0\%$8 rays and electrons then exit the pipe wall and deposit energy in 4H-SiC pixels. Using BBBREM for radiative Bhabha generation, SAD for transport, and RASER for detector response, the study evaluates three candidate locations at 10 m, 84 m, and 90.5 m from the interaction point; 84 m is excluded for mechanical-space conflict, and 10 m is selected because of the highest loss rate and symmetric horizontal distribution. The optimized configuration uses a 3 cm vertical detector length with a 1 $\nu=1/\sqrt{2594}\approx2.0\%$9A threshold, giving approximately 2594 counts per millisecond and therefore about 2.0% relative precision at 1 kHz. The same work defines a Total Sample Current over $\mu$0 ms,

$\mu$1

and finds a nearly perfect linear relation with normalized luminosity,

$\mu$2

with $\mu$3 and $\mu$4. The combined uncertainty is estimated as $\mu$5, dominated by counting statistics (Li et al., 31 Jul 2025).

The luminosity-monitor role is not limited to CEPC. The heavy-ion irradiation study of a 4H-SiC PIN detector explicitly links sub-45 ps time tagging and high charge collection to high-luminosity collider experiments requiring precise vertex timing and time-of-flight layers. The segmented LGAD program makes the same connection in a different architectural regime, arguing that strip and pixel geometries with internal gain are suitable for fast luminosity measurements in harsh environments (He et al., 13 May 2026, Kráčmar et al., 16 May 2026).

A practical implication of these studies is that luminosity detection in 4H-SiC is no longer defined by a single sensor topology. The CEPC concept emphasizes millisecond-scale relative luminosity precision and fast orbit feedback, whereas PIN and LGAD timing studies emphasize per-hit time tagging in the 20–50 ps regime. This suggests a layered instrumentation strategy in which 4H-SiC devices can serve both feedback and precision-timing roles, depending on geometry and readout.

5. Radiation hardness, thermal stability, and operational robustness

Radiation hardness is one of the principal arguments for 4H-SiC fast luminosity detectors, but the literature treats it in several distinct regimes. Under irradiation with $\mu$6 of 16.5 MeV/u $\mu$7Ta$\mu$8, the 4H-SiC PIN detector shows less than 2% change in effective doping concentration, negligible leakage increase from $\mu$9 A to $\mu$0 A, and time-resolution degradation from 40 ps to 45 ps. The same source contains a notable internal difference in its charge-collection reporting: the abstract states a CCE of 99.24% under Ta Heavy Ion Irradiation, while the detailed exposition gives 97.4% at 300 V from a most-probable collected charge of 2.63 fC after irradiation. In both presentations, the conclusion is that electrical performance, CCE, and timing remain stable before and after irradiation (He et al., 13 May 2026).

Graphene-optimized PIN detectors address a lower-fluence regime and focus on electrode robustness as well as bulk response. At $\mu$1, the reported rise time is 344 ps with $\mu$2 ps and the CCE is about 90.6%, with no appreciable rise-time degradation; at $\mu$3, the rise time becomes 372 ps with $\mu$4 ps and the CCE is about 87.6%. That study states that graphene has a certain radiation resistance (Jiang et al., 28 May 2025).

For avalanche detectors, the emphasis shifts to fluence-driven gain degradation and bias recovery. In ultra-thin 4H-SiC AC-LGAD simulations, acceptor removal in the gain layer is modeled as $\mu$5 with $\mu$6, and carrier trapping follows $\mu$7. At $\mu$8, total collected charge in a 20 $\mu$9m device falls from about 4 fC to 3 fC, while $M\approx6$0 degrades from about 18 ps to 20–25 ps; at $M\approx6$1 V, $M\approx6$2 ps is maintained (Kalani et al., 23 Jan 2026). For segmented 4H-SiC LGADs, preceding single-pad devices irradiated up to $M\approx6$3 are reported to show less than 20% gain loss and full charge collection, with gain recovery after $M\approx6$4 min annealing cycles (Kráčmar et al., 16 May 2026).

Thermal stability is likewise presented as a differentiator. The CEPC design summary includes room-temperature operation and radiation tolerance to MGy-level $M\approx6$5 doses. Other LGAD studies state that performance is stable from $M\approx6$6 to $M\approx6$7 without significant annealing effects, and that leakage remains below 1 nA per channel at $M\approx6$8 V even between $M\approx6$9 and $M\approx10$0 because of the wide bandgap (Li et al., 31 Jul 2025, Švihra et al., 12 Apr 2025, Kráčmar et al., 16 May 2026).

6. Design constraints, readout bottlenecks, and future directions

A recurrent misconception is that the sensor material alone determines luminosity-detector performance. Several studies show instead that the readout chain is often the limiting element. In measurements on 4H-SiC detectors for high-intensity ion beams, minimum-ionizing particles cannot be detected reliably with the used discrete electronics setup in combination with the single-channel sensor, while a strip sensor combined with an ASIC-based readout electronics from the CMS/Belle-II experiments makes it possible to recover a certain part of the signal. In a separate study of thin 4H-SiC PIN detectors, the front-end readout electronics used for 4H-SiC detectors are described as a bottleneck, which motivated the development of a high-frequency readout board with an intrinsic bandwidth of 10 GHz (Christanell et al., 2021, Gsponer et al., 15 Sep 2025).

This hardware dependence is matched by increasingly elaborate simulation environments. RASER is used both for CEPC detector optimization and for simulated 4H-SiC LGAD timing, where it couples DEVSIM for electrical properties and internal fields, Geant4 for MIP energy deposition, FEniCS for weighting-potential drift equations, and ROOT for pulse analysis. The drift-velocity extraction study adds a complete environment combining TCAD, Allpix$M\approx10$1, and SPICE simulations, in good agreement with measured data. Ultra-thin AC-LGAD design uses WeightField2 and benchmarks it against FBK LGAD irradiation data (Wang et al., 2023, Gsponer et al., 15 Sep 2025, Kalani et al., 23 Jan 2026).

Another misconception is that PIN structures are inherently excluded from the highest timing tier. Conventional 4H-SiC PIN detectors already report 40–45 ps single-channel time resolution under $M\approx10$2-particle timing, and graphene-optimized PIN devices report 21 ps in TCT at the maximum scanning distance. The graphene study explicitly states that 21 ps is comparable to state-of-the-art 4H-SiC LGADs, which typically exhibit time resolutions better than 35 ps under single minimum ionizing particle equivalent injection conditions (He et al., 13 May 2026, Xiao et al., 28 Mar 2026).

The current development directions therefore separate into three technically distinct lines. One line pursues simpler PIN devices with increasingly optimized fields, transparent electrodes, and high-bandwidth readout. A second line pursues moderate-gain LGADs and segmented LGADs for sub-30 ps timing and fine spatial granularity. A third line pursues system-level luminosity monitors such as the CEPC array, where millisecond counting precision, TSC linearity, and feedback integration are primary. A plausible implication is that the mature 4H-SiC fast luminosity detector will not be a single canonical detector, but a class of devices spanning feedback monitors, timing layers, and segmented high-rate trackers, all built around the same wide-bandgap transport platform.