Fast Reverse Recovery in Ge‑Doped Vertical GaN Schottky Barrier Diodes

Introduction

Wide‑band‑gap semiconductors, particularly GaN, have become pivotal for next‑generation high‑frequency, high‑power, and high‑performance electronics [1]–[6]. Recent advances in hydride vapor phase epitaxy (HVPE) now enable commercial growth of low‑dislocation GaN substrates (≤10⁶ cm⁻²) [7]–[10]. Vertical device architectures built on these substrates offer larger current densities, reduced leakage paths, and enhanced reliability compared with lateral designs [11], [12].

GaN‑based Schottky barrier diodes are critical components in power‑switching circuits. Their unipolar nature eliminates minority‑carrier storage, yielding high switching speeds and low reverse‑recovery losses [13]–[17]. However, systematic studies of reverse‑recovery behavior in vertical GaN SBDs remain scarce. This work addresses that gap by fabricating Ge‑doped vertical SBDs on free‑standing GaN and rigorously evaluating their reverse‑recovery performance.

Ge is an attractive n‑type dopant for GaN, offering shallow activation (≈20 meV) comparable to Si while inducing less lattice distortion due to its ionic radius [19]–[20]. Ge doping is reported to yield smoother surfaces and reduced defect densities [21], [22], and enhances thermal stability, a key requirement for high‑temperature devices. Here, we present the first systematic investigation of reverse‑recovery characteristics in Ge‑doped vertical GaN SBDs.

Methods and Experiments

The device architecture is illustrated in Fig. 1a: a 390‑µm thick free‑standing n⁺‑GaN substrate overlain by a 9‑µm n⁻‑GaN drift layer. The substrate, grown by HVPE, incorporates 1×10¹⁸ cm⁻³ Ge and exhibits a dislocation density of 1×10⁶ cm⁻² [7]. The epitaxial layer was deposited by metal‑organic chemical vapor deposition at ≈2 µm h⁻¹.

Device fabrication involved Ti/Al/Ni/Au ohmic contacts on the back surface and Ni/Au Schottky electrodes on the front, with diameters of 100, 200, 300, 400, and 500 µm (Fig. 1b). Full process details are available in our earlier reports [23], [24].

a Schematic cross‑section of the vertical SBD. b Optical microscopy image of the five electrode diameters. c Panchromatic CL image of the epitaxial layer.

CL imaging was performed with a Quanta 400 FEG SEM at 10 kV to map dislocation distribution. Capacitance‑voltage (C‑V) and current‑voltage (I‑V) measurements employed a Keithley 4200 analyzer. Temperature‑dependent tests spanned 300–500 K using a custom setup.

Results and Discussion

Figure 1c shows the CL map of the epitaxial layer. Dark spots indicate non‑radiative dislocation centers; their uniform distribution yields an average density of ≈1.3×10⁶ cm⁻², confirming high‑quality crystal growth.

C‑V and G‑V curves were recorded at 1 MHz (Fig. 2a–b). From the 1/C²–V plots, the donor concentration N_d is ≈6.2×10¹⁵ cm⁻³. The phase angle θ, calculated via tan⁻¹(2πfC/G), remains ≈90°, indicating minimal leakage through the Schottky barrier (Fig. 2b inset).

J‑V characteristics (Fig. 2c) reveal I_on/I_off ratios ranging from 9.9×10⁷ to 1.3×10¹⁰ and V_on values between 0.70 and 0.78 V across all diameters, demonstrating excellent electronic performance.

a Room‑temperature C‑V curves for each electrode size. b G‑V curves with θ versus V inset for 300‑µm device. c J‑V curves for all diameters.

Reverse‑recovery behavior was measured using the test circuit in Fig. 3a. A square‑wave voltage (+20 V to –20 V) drives the device; the transient current is captured by a high‑speed probe on a Tektronix MDO 4104‑3 oscilloscope. The reverse‑recovery waveform (Fig. 3b) defines storage time t_a, reverse‑current delay t_b, and recovery time T_rr (time to 10% of peak reverse current). The reverse‑recovery charge Q_rr is obtained by integrating the reverse current up to T_rr.

a Test circuit for reverse‑recovery measurement. b Schematic of the reverse‑recovery waveform.

Figure 4 displays the reverse‑recovery currents for all electrode diameters. The T_rr values are 15.8, 16.2, 18.1, 21.22, and 24.5 ns for 100, 200, 300, 400, and 500 µm devices, respectively. Corresponding Q_rr values are 0.0127, 0.0536, 0.150, 0.280, and 0.405 nC. All devices achieve sub‑25‑ns recovery times, and the smallest device exhibits the fastest recovery. Both T_rr and Q_rr increase with electrode area.

Reverse‑recovery current for each electrode diameter.

To probe voltage dependence, we repeated the measurement with ±10 V. T_rr remains essentially unchanged, while Q_rr continues to scale with the square of diameter (i.e., electrode area) as shown in Fig. 6.

Reverse‑recovery time T_rr versus electrode diameter d.

Reverse‑recovery charge Q_rr versus electrode diameter d.

The dominant mechanism behind the observed behavior is trap‑mediated storage rather than parasitic inductance. Uniform trap distribution leads to a Q_rr proportional to contact area. Consequently, the RC time constant grows with device diameter, which explains the linear increase in T_rr. Reducing electrode size or drift‑layer thickness should further improve recovery performance, as corroborated by our earlier work [29].

Temperature dependence was evaluated for the 500‑µm device (Fig. 7). Reverse‑recovery time and charge remain unchanged across 300–500 K, indicating that trap concentrations are temperature‑insensitive in GaN, unlike Si‑based SBDs, where T_rr can increase by >190% between 300 and 425 K [17]. The superior thermal stability of GaN underpins its suitability for high‑temperature power switching.

Reverse‑recovery characteristics for 500‑µm device at 300, 400, and 500 K.

Conclusions

We fabricated vertical GaN Schottky barrier diodes on Ge‑doped free‑standing substrates grown by HVPE. Cathodoluminescence and electrical measurements confirm high crystal quality and excellent electronic performance. Reverse‑recovery tests reveal times from 15.8 to 24.5 ns for devices ranging from 100 to 500 µm in diameter, with both T_rr and Q_rr scaling positively with electrode area. These findings provide a benchmark for optimizing GaN‑based SBDs for high‑speed, high‑temperature power applications.

Abbreviations

CL

Cathodoluminescence

C‑V

Capacitance‑voltage

DUT

Device under test

FS

Free‑standing

GaN

Gallium nitride

HVPE

Hydride vapor phase epitaxy

I‑V

Current‑voltage

SBDs

Schottky barrier diodes

SEM

Scanning electron microscope