n-type LGADs: Inverted Silicon Avalanche Detectors

• n-type LGADs are silicon sensors with an inverted p++–n+–n structure that facilitates electron-initiated impact ionization.
• They exhibit distinct electrical signatures, including a near-surface gain layer field peak, making them ideal for detecting low-penetrating particles like UV photons and soft X-rays.
• Irradiation studies reveal that managing annealing and bias adjustments is critical to counteracting space charge sign inversion, leakage current increases, and shifting depletion voltages.

to=list_mcp_resources ฝ่ายขายข่าว=json {} to=list_mcp_resource_templates 彩神争霸大发快三=json {} n-type Low Gain Avalanche Detectors (nLGADs) are silicon sensors with a highly doped gain layer that enables controlled charge multiplication via impact ionization. In the silicon implementation studied most directly in recent irradiation work, the devices invert the conventional p-type LGAD doping and are realized as a $p^{++}$–$n^{+}$–$n$ structure on high-resistivity float-zone n-type bulk, with the shallow $n^{+}$ gain layer placed near the top side so that electron-initiated multiplication is favored for charge deposited in the first few microns. This topology makes nLGADs well suited for low-penetrating particles such as UV photons and soft X-rays, while also exposing a radiation-response phenomenology that differs qualitatively from that of p-type LGADs because the n-type bulk undergoes space charge sign inversion (SCSI) under irradiation (Kraus et al., 8 Aug 2025, Kraus et al., 4 Apr 2025).

1. Device definition, architecture, and terminology

In the recent silicon nLGAD studies, the device concept is a $p^{++}$–$n^{+}$–$n$ stack. The n-type high-resistivity float-zone bulk provides the drift volume, while a thin, highly doped $n^{+}$ layer immediately below the $p^{++}$ topside contact creates a localized high electric field under reverse bias. A heavily doped $p^{++}$ top contact and an $n^{+}$0 backside contact complete the structure. Reported test structures include single-pad devices with full-cover metallization on a $n^{+}$1 active area and square devices with active area $n^{+}$2 and a small opening in the top metallization to permit laser illumination. In both electrical studies, the guard ring was grounded during measurement, and reference PiN diodes from the same wafer were used as controls in the low-fluence study (Kraus et al., 8 Aug 2025, Kraus et al., 4 Apr 2025).

The defining operational distinction from conventional p-type LGADs is that the dominant multiplying carrier is changed. In p-type LGADs, the architecture is typically $n^{+}$3–$n^{+}$4–$n^{+}$5 and the devices are optimized for minimum ionizing particles. In nLGADs, the high-field region is formed at the $n^{+}$6–$n^{+}$7 junction in an n-type bulk, and electron-initiated impact ionization is exploited. Because electrons have higher ionization coefficients than holes in silicon, near-surface generation by UV photons or soft X-rays produces electrons that drift into the high-field $n^{+}$8 gain layer and multiply efficiently (Kraus et al., 4 Apr 2025).

The term “nLGAD” is not fully uniform across the literature. In community parlance summarized in the compensated-gain-layer work, “nLGADs” can denote LGAD-like devices whose gain layer is net n-type and whose geometry may target hole impact ionization or altered field distributions (Sola et al., 2022). In standard segmented LGAD literature, “nLGADs in the conventional sense” can also refer to n-on-p devices with $n^{+}$9 segmented electrodes above a buried $n$0 gain layer on p-type bulk, as in the FBK UFSD-3.2 arrays (Darby et al., 2021). In current irradiation studies of silicon nLGADs, however, the term refers specifically to the inverted $n$1–$n$2–$n$3 concept (Kraus et al., 8 Aug 2025, Kraus et al., 4 Apr 2025).

2. Electrostatics, depletion, and multiplication

For a uniformly doped one-sided junction, the depleted bulk is described by Poisson’s equation,

$n$4

with $n$5 the signed effective space-charge density. The depletion width and full depletion voltage are given by

$n$6

These relations govern the global depletion of the bulk, while the gain layer superimposes a local field peak. For nLGADs the avalanche gain is modeled as

$n$7

where $n$8 is the electron ionization coefficient. Commonly used forms for ionization coefficients include $n$9 and $n^{+}$0, although specific parameterizations were not extracted in the irradiation studies (Kraus et al., 4 Apr 2025).

Before irradiation at $n^{+}$1, the gain layer depletes at about $n^{+}$2 and breakdown occurs in the range $n^{+}$3–$n^{+}$4. The low-fluence study further quantified the temperature dependence of breakdown using the metric $n^{+}$5: breakdown shifts from below $n^{+}$6 at approximately $n^{+}$7 to above $n^{+}$8 at $n^{+}$9, with an approximately linear temperature coefficient of about $p^{++}$0. The stated explanation is reduced carrier mean free path from phonon scattering, which lowers impact-ionization probability (Kraus et al., 8 Aug 2025).

The pre-irradiation field profile is the canonical LGAD profile. Depletion starts from the top-side junction via the gain layer, and a pronounced field peak forms near the topside. In UV-TCT, where a $p^{++}$1 pulsed laser probes near-surface generation, the reported pre-irradiation gain is approximately $p^{++}$2 at about $p^{++}$3 bias. This directly reflects efficient electron multiplication in the near-surface high-field region (Kraus et al., 4 Apr 2025).

3. Irradiation response and space charge sign inversion

Recent irradiation studies span both low-fluence and higher-energy proton exposures. In the low-fluence campaign, devices were irradiated with neutrons at the TRIGA reactor in Ljubljana and with $p^{++}$4 protons at the AIC-144 cyclotron in Krakow. Neutron fluences were $p^{++}$5, $p^{++}$6, $p^{++}$7, $p^{++}$8, and $p^{++}$9, with accuracy of about $n^{+}$0. Proton fluences were $n^{+}$1, $n^{+}$2, $n^{+}$3, and $n^{+}$4, with accuracy well below $n^{+}$5. Using leakage-current increases in irradiated PiNs and the relations

$n^{+}$6

the study obtained a hardness factor $n^{+}$7, yielding converted proton fluences of $n^{+}$8, $n^{+}$9, $n$0, and $n$1 (Kraus et al., 8 Aug 2025).

A separate study irradiated IMB-CNM nLGADs with $n$2 protons at the CERN PS-IRRAD facility to fluences of $n$3, $n$4, $n$5, $n$6, and $n$7, followed by $n$8 minutes of annealing at $n$9 (Kraus et al., 4 Apr 2025). Across these studies, the central bulk effect is SCSI. In high-resistivity n-type silicon, SCSI is known to occur at fluences on the order of $n^{+}$0 for protons and neutrons. The proton-irradiation study states that inversion has occurred by $n^{+}$1, and that gain and electric-field signatures suggest the transition sets in between $n^{+}$2 and $n^{+}$3 (Kraus et al., 4 Apr 2025).

The standard parameterization used for the evolution of effective space charge is

$n^{+}$4

where $n^{+}$5 is the initial net donor density, $n^{+}$6 is the removal coefficient, and $n^{+}$7 is the introduction rate of radiation-induced defects that act as acceptors. The inversion criterion is $n^{+}$8. In physical terms, donor removal and the creation of acceptor-like defects drive the bulk from net n-type to net p-type. In nLGADs, this generates a second junction at the backside near the $n^{+}$9 contact and reverses the depletion direction from backside to front after inversion, thereby reconfiguring the internal electric field (Kraus et al., 4 Apr 2025, Kraus et al., 8 Aug 2025).

The irradiation studies further state that donor removal in nLGADs appears more pronounced at lower fluences than acceptor removal in traditional p-type LGADs. This suggests that the gain-layer field in nLGADs can degrade rapidly once irradiation-induced donor deactivation and SCSI set in (Kraus et al., 4 Apr 2025).

4. Electrical and transient signatures after irradiation

All post-irradiation electrical measurements in the low-fluence study were performed at $p^{++}$0 to mitigate increased leakage current, and a short annealing step of $p^{++}$1 minutes at $p^{++}$2 was applied prior to measurements to equalize short-term annealing across samples (Kraus et al., 8 Aug 2025). After irradiation, nLGAD $p^{++}$3–$p^{++}$4 curves show four characteristic features: increased leakage current with fluence; a steep current rise associated with gain-layer depletion that becomes less pronounced and shifts to higher bias; a characteristic dip in leakage current before full depletion; and an increase in breakdown voltage with irradiation. The dip is tied to guard-ring pinch-in when the depletion front reaches the top side, so that only the area inside the guard ring contributes to pad current. Reference PiN diodes show an initial current rise with bias, then a decrease once the guard ring pinches in, and finally a plateau; comparison of pad current and total device current confirms the guard-ring origin of the dip (Kraus et al., 8 Aug 2025).

After SCSI, the steep $p^{++}$5–$p^{++}$6 rise no longer marks the gain-layer depletion voltage in the pre-irradiation sense. Instead, it reflects a more complex field reconfiguration and depletion from the backside. The macroscopic meaning of the curve has therefore changed: the same measurement feature no longer corresponds to the same internal field topology once the bulk has inverted (Kraus et al., 8 Aug 2025).

The post-irradiation $p^{++}$7–$p^{++}$8 response is equally distinctive. Measured at $p^{++}$9, the irradiated nLGADs exhibit a distinct peak near full depletion. This peak shifts to higher bias and broadens with increasing fluence, and it is higher and narrower after neutron irradiation than after proton irradiation. The stated origin is a resonance of defect charges when two depletion regions merge at full depletion: one develops from the backside after SCSI and the other from the top-side gain layer. The resonance is frequency dependent. In a $p^{++}$0 proton-irradiated nLGAD at $p^{++}$1, changing the frequency from $p^{++}$2 to $p^{++}$3 modifies the peak shape and position (Kraus et al., 8 Aug 2025).

Laser-based transient techniques provide direct confirmation of the field redistribution. UV-TCT with a $p^{++}$4 pulsed laser probes near-surface generation and electron multiplication. TPA-TCT with a $p^{++}$5 femtosecond laser operated in the quadratic absorption regime creates a localized generation volume at controlled depth and records prompt current about $p^{++}$6 after generation. The induced current follows Ramo’s theorem, $p^{++}$7. Before irradiation, TPA-TCT shows a pronounced field peak at the gain layer near the topside and depletion progressing from top to back. After $p^{++}$8, depletion starts from the backside and the gain-layer peak is significantly reduced. UV-TCT shows that gain of approximately $p^{++}$9, obtained before irradiation at about $n^{+}$00, is recovered only by increasing the bias to about $n^{+}$01 after $n^{+}$02 (Kraus et al., 4 Apr 2025).

These observables support a consistent interpretation: increasing fluence reduces the electric field in the gain layer, reduces impact-ionization probability, raises the bias needed to reach a given gain, and tightens breakdown margins. While timing measurements were not reported in the low-fluence irradiation study, that study explicitly states that the trends imply degradation of charge multiplication and timing performance with fluence (Kraus et al., 8 Aug 2025).

5. Annealing behavior and kinetics

Annealing was investigated in both isochronal and isothermal modes. A uniform first annealing step of approximately $n^{+}$03 minutes at $n^{+}$04 shifted full depletion to lower bias, with the full-depletion point extracted as the voltage of the $n^{+}$05–$n^{+}$06 resonance peak. This was identified as beneficial annealing, and the effect was more pronounced at higher fluences (Kraus et al., 8 Aug 2025).

In the isochronal protocol, a $n^{+}$07 proton-irradiated nLGAD at $n^{+}$08 was annealed for $n^{+}$09 minutes at successively higher temperatures from $n^{+}$10 to $n^{+}$11 in $n^{+}$12 steps. The $n^{+}$13–$n^{+}$14 peak shifted from approximately $n^{+}$15 to over approximately $n^{+}$16 with increasing temperature, indicating that the beneficial regime was bypassed and reverse annealing dominated in that temperature range and timing (Kraus et al., 8 Aug 2025).

In the isothermal protocol, a neutron-irradiated sample at $n^{+}$17 and a proton-irradiated sample at $n^{+}$18 were annealed at $n^{+}$19 for progressively longer times from $n^{+}$20 to $n^{+}$21 minutes. The $n^{+}$22–$n^{+}$23 dip became more pronounced, and a corresponding valley formed in the $n^{+}$24–$n^{+}$25 curves, reflecting significant reconfiguration of internal fields during annealing. The evolution of the $n^{+}$26–$n^{+}$27 peak voltage showed a slight initial reduction followed by an increase. Beneficial annealing was visible for about $n^{+}$28 minutes at $n^{+}$29 before reverse annealing set in (Kraus et al., 8 Aug 2025).

The stated kinetics follow an Arrhenius law,

$n^{+}$30

which explains why higher temperatures accelerate defect evolution. The observed behavior is reported to align with the established three-component model of $n^{+}$31: a short-term beneficial component, a stable damage term, and a reverse-annealing component. Operationally, the low-fluence study concludes that a short initial anneal at about $n^{+}$32 is beneficial, whereas longer or hotter anneals favor reverse annealing; control of storage and annealing temperature-time history is therefore essential (Kraus et al., 8 Aug 2025).

6. Comparison with other LGAD families and ongoing design directions

The contrast with p-type LGADs is structural and phenomenological. In p-type LGADs, gain-layer acceptor removal with fluence tends to shift gain-layer depletion toward lower bias and ultimately suppress gain. In nLGADs, the recent silicon studies emphasize SCSI of the n-type bulk, reversal of depletion direction so that depletion starts from the backside, merging of two depletion fronts at full depletion, a $n^{+}$33–$n^{+}$34 peak associated with that merger, and breakdown shifting to higher voltages with fluence. The macroscopic $n^{+}$35–$n^{+}$36 and $n^{+}$37–$n^{+}$38 signatures are therefore qualitatively different (Kraus et al., 8 Aug 2025).

One design direction relevant by contrast is the compensated LGAD gain layer. That approach uses overlapping $n^{+}$39 and $n^{+}$40 implants with

$n^{+}$41

so that the effective gain-layer doping remains p-type, with a target near $n^{+}$42, while irradiation removes both species according to

$n^{+}$43

The paper explicitly states that this compensated structure does not create a net n-type gain layer and is not a “pure” nLGAD; multiplication remains electron-dominated under an $n^{+}$44 readout in an n-in-p sensor (Sola et al., 2022). This distinction is important because compensated LGADs address radiation tolerance by balancing acceptor and donor removal, whereas silicon nLGADs in the $n^{+}$45–$n^{+}$46–$n^{+}$47 sense are defined by inverted polarity and by the SCSI-driven field reconfiguration documented experimentally (Kraus et al., 4 Apr 2025, Kraus et al., 8 Aug 2025).

A second direction is the extension of nLGAD concepts to wide-bandgap materials. First-generation 4H-SiC LGADs were fabricated on n-type substrate/epitaxial wafers and use a buried n-type donor gain implant centered approximately $n^{+}$48 below the top surface. The first batch used $n^{+}$49 n-type epitaxial layers with doping down to approximately $n^{+}$50, gain-implant energies of $n^{+}$51–$n^{+}$52, and doses of $n^{+}$53–$n^{+}$54. The gain-layer depletion and multiplication onset occur between approximately $n^{+}$55 and $n^{+}$56, stable operation was observed up to $n^{+}$57 at wafer level, and the capacitance of fully depleted $n^{+}$58 devices is $n^{+}$59. No irradiation results were reported in that first study, but the work highlights 4H-SiC’s high field capability, thermal stability, and the use of a Junction Termination Extension designed for breakdown voltage above $n^{+}$60 (Novotný et al., 10 Mar 2025).

Within silicon, the recent low-fluence and high-energy proton studies establish a consistent picture. nLGADs retain their near-surface multiplication advantage for UV photons and soft X-rays, but irradiation changes the internal field profile, shifts full-depletion and operating bias upward, increases leakage current, and reduces gain-layer field strength. With appropriate biasing, low-temperature operation, and controlled annealing, the low-fluence study states that the devices remain viable for applications involving low-penetrating radiation, while their radiation response must be actively managed within an evolving operating envelope (Kraus et al., 8 Aug 2025).