Ultra-Thin 4H-SiC LGADs: Fast, Rad-Hard Sensors

• Ultra-thin 4H-SiC LGADs are silicon carbide-based radiation sensors featuring a 10–50 μm active layer and a precisely engineered gain region for moderate avalanche multiplication.
• They leverage 4H-SiC's wide bandgap and high breakdown field to deliver sub-50 ps timing resolution and enhanced radiation hardness in collider and space applications.
• Advanced fabrication techniques—including controlled CVD growth, targeted ion implantation, and optimized edge termination—ensure high breakdown voltages and reliable performance.

Ultra-thin 4H-SiC Low Gain Avalanche Detectors (LGADs) are silicon carbide-based solid-state radiation sensors featuring an active region of reduced thickness—typically between 10 and 50 μm—with a finely engineered internal charge multiplication layer. Leveraging the wide bandgap and high breakdown field of 4H–SiC, these devices realize moderate avalanche gain, sub-50 ps timing, and exceptional radiation hardness, targeting applications in collider experiments, space instrumentation, and harsh environments where conventional silicon LGADs degrade. The following sections detail layer architectures, doping strategies, device fabrication, avalanche physics, performance metrics, and prospects for further optimization.

1. Physical Structure and Layer Engineering

The defining characteristic of ultra-thin 4H-SiC LGADs is the highly controlled epitaxial stack on a thick mechanical substrate. Key architectural elements include:

• Substrate: Typically n⁺ or n⁺⁺ 4H-SiC with thickness 300–500 μm and doping N ≃ 1×10¹⁷–10¹⁹ cm⁻³, serving as a mechanical handle and ohmic contact (Novotný et al., 10 Mar 2025, Švihra et al., 12 Apr 2025).
• Drift (epi) region: Grown by CVD; active thickness typically 20–50 μm, with N-type (or p-type) doping N_D ≃ 10¹²–10¹⁵ cm⁻³. Fully depleted at bias V_dep ≈ 50–250 V depending on thickness (Švihra et al., 12 Apr 2025, Onder et al., 16 Oct 2025, Kalani et al., 23 Jan 2026).
• Gain layer: A shallow high-concentration region, typically p⁺ (implant), or n⁺ (for n-in-p device) placed ≈0.5–1.0 μm below the entrance surface, with thickness t_GL ≤ 1.5 μm, and peak doping N_A,peak ≃ 10¹⁶–10¹⁷ cm⁻³ (Švihra et al., 12 Apr 2025, Yang et al., 2022).
• Edge termination: Power electronics-style Junction Termination Extension (JTE) rings, and, in advanced designs, deep trench-isolation or guard-ring structures extend breakdown voltage above 1–2.4 kV (Onder et al., 16 Oct 2025).
• Metallization and passivation: Ti/Al, Ni, or Ti/Pt/Au stack for ohmic and Schottky contacts; front grid for UV compatibility; surface passivation by SiO₂, SiN, or polyimide (Švihra et al., 12 Apr 2025, Novotný et al., 10 Mar 2025, Onder et al., 16 Oct 2025).

Typical values for layer thickness and doping are summarized below (representative; variations exist per design):

Layer	Thickness (μm)	Doping (cm⁻³)
P⁺⁺ Contact	0.3	5×10¹⁹ (Al)
Gain layer (p⁺/n⁺)	0.5–2.4	1–8×10¹⁶
Drift (epi/bulk) region	20–50	1–5×10¹³ (n/p)
Substrate (handle)	350–500	≥1×10¹⁸

No post-fabrication wafer thinning is used in most studies; epi-on-substrate processing is standard (Švihra et al., 12 Apr 2025, Novotný et al., 10 Mar 2025).

2. Doping Profiles, Field Control, and Multiplication Layer Design

Avalanche gain is realized within the gain layer via sharply tailored doping:

• In n-in-p (“AC-LGAD”) or p-in-n (classic) topologies, the gain layer is created by ion-implant (e.g., B or Al), with dose and profile set by implant energy and post-activation anneal (Švihra et al., 12 Apr 2025, Kalani et al., 23 Jan 2026).
• The LGAD’s multiplication region forms a narrow, high-field zone at the p⁺/n (or n⁺/p) junction, where the local electric field $E_{\text{peak}}$ exceeds the critical field $E_{\text{crit}} \approx 2 \times 10^{6}$ V/cm (4H–SiC) (Švihra et al., 12 Apr 2025, Yang et al., 2022).
• Variants include “triangle”-type (high $E_\text{peak}$ at abrupt junction) and “trapezoid”-type (field plateaus via additional shallow doping for robustness) (Yang et al., 2022). Trench-isolated and step-profile (multi-layer) gain designs enhance field uniformity and breakdown margin (Onder et al., 16 Oct 2025, Yang et al., 2022).

Field shaping through JTE/trench and optimizing the gain-layer dose (+20–50% variation) allow tuning from M ≈ 3–25, balancing timing performance with breakdown and radiation robustness (Švihra et al., 12 Apr 2025, Onder et al., 16 Oct 2025).

3. Fabrication Methodology

The baseline process for ultra-thin 4H-SiC LGADs consists of:

1. Epi growth: CVD deposition of high-resistivity layer of desired thickness; uniformity verified (ΔT/T ≲ 4–5%).
2. Gain layer formation: Ion implantation (e.g., B, Al, N) to create a sharp doping peak ≈1 μm from the surface; doses in the range 10¹²–10¹³ cm⁻²; subsequent anneal for activation (Švihra et al., 12 Apr 2025, Novotný et al., 10 Mar 2025).
3. Edge termination: JTE rings (dose 10¹³–10¹⁴ cm⁻², range ≈150–300 nm) and, for >kV designs, deep etched trenches with passivation (Onder et al., 16 Oct 2025).
4. Metallization: Deposition/liftoff of ohmic contacts (Ni, Ti/Al, etc.) for p⁺ and n⁺ regions. Optional front “metal-grill” for optical tests (Švihra et al., 12 Apr 2025, Wang et al., 2023).
5. Surface passivation: SiO₂/SiN/polyimide to reduce surface generation and edge breakdown.
6. Dicing, packaging: Mechanical saw, die attach, wire bonding, PCB mount.

No backgrinding or total device thinning is employed at volume; future optimization aims for total thicknesses ≲50–100 μm (Novotný et al., 10 Mar 2025).

4. Avalanche Multiplication Physics and Analytical Models

Avalanche gain calculations for 4H-SiC LGADs utilize both analytical and TCAD methods:

• Multiplication factor: $$M(V) = \exp\left(\int_0^w \alpha[E(x, V)]\, dx\right)$$, where $w$ is the gain layer thickness.
• Ionization coefficients (Townsend/Okuto-Crowell/MPL):

$\alpha(E) = A \exp\left[-\frac{B}{E}\right]$

For 4H-SiC, typical values are $A \sim 10^{6}$ cm⁻¹, $B \sim 10^{7}$ V/cm (Yang et al., 2022, Švihra et al., 12 Apr 2025).

• Depletion width: $$W(V) = \sqrt{2 \epsilon_{\text{SiC}} (V_{\text{bi}}+V)/q N_{\text{eff}}}$$, $\epsilon_{\text{SiC}}\approx 9.7\,\epsilon_0$.
• Impact of bias: $E_{\text{peak}}$ increases with bias; gain becomes significant for $V_\text{bias} ≳ 300–400$ V; $M(V)$ rises exponentially until controlled by field-shaping or termination design (Švihra et al., 12 Apr 2025, Onder et al., 16 Oct 2025).
• Radiation effects: Acceptors in the gain layer are deactivated following $N_A(\Phi) = N_{A,0} \exp(-c \Phi)$, leading to increased required bias for a given M after irradiation; $c \approx 5.6 \times 10^{-16}$ cm² (SiC) (Kalani et al., 23 Jan 2026).

Simulations (Sentaurus, WF2, RASER) consistently predict high-field localization, strong avalanche gain, and fast transient response for thin active regions (Kalani et al., 23 Jan 2026, Wang et al., 2023).

5. Electrical, Transient, and Timing Performance

Measured and simulated performance in ultra-thin 4H-SiC LGADs demonstrates:

• Leakage current: $I_\text{rev} < 1\,\mu$A at $V=100$–$300$ V (as low as 30 pA at 500 V in advanced designs); weakly temperature-dependent due to $3.26$ eV bandgap (Švihra et al., 12 Apr 2025, Onder et al., 16 Oct 2025).
• Capacitance: $C(V)$ in accord with $1/C^2 \propto (V+V_{bi})^{-1}$, typically $\sim1.5$–$20$ pF depending on pad size and thickness (Švihra et al., 12 Apr 2025, Onder et al., 16 Oct 2025).
• Gain: $M=3$–$25$ for $20$–$50$ μm devices at $V=400$–$800$ V (Švihra et al., 12 Apr 2025, Kalani et al., 23 Jan 2026).
• Timing resolution ($\sigma_t$): Projections and simulations for 20–30 μm thick devices yield $\sigma_t \lesssim 25$ ps (post-irradiation), with best-case $\sigma_t \lesssim 10$ ps unirradiated (Kalani et al., 23 Jan 2026). Ultra-thin (<10 μm) designs are predicted to achieve $\sigma_t \sim 30$ ps at moderate gain (Wang et al., 2023).
• Breakdown voltage: Enhanced by optimized edge termination; V_BR > 500 V (JTE), up to >2.4 kV (trench+JTE) (Onder et al., 16 Oct 2025, Švihra et al., 12 Apr 2025).
• Temperature and radiation stability: Gain, leakage, and timing performance degrade only weakly for $-50$ to $+150$°C, and SiC devices maintain M > 10 at fluences 5×10¹⁵ $n_{eq}$/cm² with appropriate bias (Kalani et al., 23 Jan 2026).

Timing jitter is dominated by electronics noise and signal slope (dI/dt), with transit time below 200 ps for 30 μm at $v_{sat} ≈ 2 \times 10^7$ cm/s (Onder et al., 16 Oct 2025). Trench and field-control designs further mitigate capacitive and edge noise contributions.

6. Comparative Advantages, Limitations, and Materials Perspective

4H-SiC LGADs, especially in ultra-thin form, offer significant advantages:

• Radiation tolerance: 4H-SiC is inherently harder to displacement damage and surface defect accumulation (improvements by ×5–10 over silicon LGADs) (Novotný et al., 10 Mar 2025, Kalani et al., 23 Jan 2026).
• High breakdown field: 4H-SiC tolerates 2.0–2.4 MV/cm, allowing operation at substantially higher bias without microplasma effects (Yang et al., 2022, Onder et al., 16 Oct 2025).
• Temperature robustness: Leakage doubles per +15°C in SiC vs. +7–8°C for Si (Novotný et al., 10 Mar 2025).
• Material comparison: Compared to Si and diamond (see table below), SiC presents a balanced combination of timing, charge collection, and gain retention under irradiation (Kalani et al., 23 Jan 2026).

Property	4H–SiC	Si	Diamond
Bandgap [eV]	3.26	1.12	5.5
$v_{sat} [\text{cm/s}]$	2×10⁷	1×10⁷	2.5×10⁷
Intrinsic gen.	57 eh/μm	75 eh/μm	40 eh/μm
Breakdown field	2.0 MV/cm	0.3 MV/cm	10 MV/cm
Rad. hardness	high	moderate	very high

Limitations and challenges: Large area thinning below 20–30 μm is limited by epitaxial technology and process control. Low charge collection (Q₀) due to wide bandgap necessitates higher gain (M ~ 20–50) for adequate SNR under MIP signals. High bias operation and gain tuning are constrained by breakdown margin, surface passivation, and field crowding (Švihra et al., 12 Apr 2025, Kalani et al., 23 Jan 2026, Novotný et al., 10 Mar 2025).

7. Prospects, Optimization Strategies, and Future Directions

Active research is pursuing several axes of improvement:

• Epi thinning and process control: Refinement of CVD for epi layers <20 μm, total device thinning, and tighter doping control (N_eff < 1×10¹⁴ cm⁻³) (Onder et al., 16 Oct 2025, Novotný et al., 10 Mar 2025).
• Multi-step and graded gain layers: Dual or graded implants to optimize field distribution and flatten gain vs. bias (Švihra et al., 12 Apr 2025).
• Edge and surface optimization: Wider/deeper trench JTE, surface passivation, and field plates to suppress microplasma and enhance V_BR (Onder et al., 16 Oct 2025).
• Radiation benchmarking: Ongoing proton and neutron irradiation up to 1×10¹⁶ $n_{eq}$/cm², with in-beam timing measurements (Švihra et al., 12 Apr 2025, Kalani et al., 23 Jan 2026).
• Simulation-framework validation: WF2 and TCAD have shown close agreement with SiC and Si LGAD data (within 10% for gain, rise time), enabling predictive design without full-scale prototyping (Kalani et al., 23 Jan 2026, Wang et al., 2023).
• Material systems comparison: Continued benchmarking of SiC, Si, and diamond for dictate design choices in HL-LHC and beyond.

A plausible implication is that full convergence of LGAD architecture and epitaxial SiC technology could yield sub-20 ps, rad-hard, ultra-thin timing sensors feasible for deployment in the most challenging collider, space, and industrial environments.

---

References:

(Švihra et al., 12 Apr 2025, Onder et al., 16 Oct 2025, Kalani et al., 23 Jan 2026, Novotný et al., 10 Mar 2025, Wang et al., 2023, Yang et al., 2022)