Reliable electrical performance of β-Ga₂O₃ Schottky barrier diode at cryogenic temperatures

β-Ga₂O₃ is attracting increasing attention for power conversion applications due to its ultrawide bandgap (4.9 eV), large critical electric field (8 MV/cm), and high electron mobility (250 cm²/V s).^1–3 A β-Ga₂O₃ Schottky barrier diode (SBD) incorporating a novel junction termination extension has reportedly exhibited a high breakdown voltage of over 2.5 kV, a specific on-resistance of 5.9 mΩ cm², and a high Baliga’s power figure of merit surpassing 1 GW/cm².⁴ 

Several studies have reported the characteristics of the β-Ga₂O₃ SBD in extreme environments such as high reverse voltage stress,⁵ high temperature,^6,7 neutron irradiation,^8–10 proton irradiation,¹¹ and electron irradiation.¹² Temperature-dependent on-resistance was derived from the forward current characteristics of a β-Ga₂O₃ SBD from 25 to 175 °C.¹³ Decreased carrier lifetime and diffusion length with increasing fluences were also observed and documented for a β-Ga₂O₃ SBD after 1.5 MeV electron irradiation.¹² 

Nevertheless, the characteristics of β-Ga₂O₃ SBDs at cryogenic temperatures have not been thoroughly investigated. Several issues still need to be solved before the implementation of the β-Ga₂O₃ SBD for cryogenic temperature applications, such as satellite electronics.¹⁴ 

• Compared with β-Ga₂O₃ SBD directly on a bulk material,^15,16 the β-Ga₂O₃ SBD fabricated on a high-quality epilayer has been regarded as a viable path to improve the electrical performance of devices. However, the static and dynamic performance of the Ga₂O₃ SBD with a homogeneous epilayer at cryogenic temperature is still unknown. The lack of such results would mask the potential of the β-Ga₂O₃ SBD for low temperature applications.

• There have been some studies reporting the dynamic performance of the β-Ga₂O₃ SBD by applying various forward and reverse stresses.^5,17,18 Forward stress up to 14 V has been applied to investigate the instability of turn-on voltage (V_th), on-resistance (R_on), and n.¹⁷ However, the relationship among dynamic performance, low frequency noise, and specific deep-level trap in the β-Ga₂O₃ SBD at both cryogenic temperature and room temperature is still unavailable.

In this paper, the electrical and trap characteristics were investigated for a large-size (2 × 2 mm²) β-Ga₂O₃ SBD at both room temperature and cryogenic temperature. Static electrical characteristics were studied by temperature-dependent capacitance-voltage (C-V) and current density-voltage (J-V) characteristics from 50 to 350 K. Dynamic performance and low frequency noise were applied to elucidate the impact of bulk traps at both room temperature and cryogenic temperature. Via deep-level transient spectroscopy (DLTS), an electron trap was discovered in the β-Ga₂O₃ epilayer. The comprehensive study paves a solid path toward practical applications of the large-size β-Ga₂O₃ SBD in extreme environments.

Figure 1(a) illustrates the schematic cross section of the β-Ga₂O₃ SBD. The device structure contains a 7 μm-thick β-Ga₂O₃ epilayer grown on a β-Ga₂O₃ substrate. Ti/Au and Ni/Au were deposited as the backside ohmic contact and the square Schottky contact (2 × 2 mm²), respectively.

Stressed J-V measurement is applied to study the dynamic performance of the β-Ga₂O₃ SBD at 300 and 100 K, as shown in Figs. 3(a) and 3(b). The inset in Fig. 3(a) shows the minor variation in J when the device is exposed to an U_s of −100 V, for various stressing durations. J-V curves were positively shifted during the stressing process, indicating an emission process of bulk trap at 300 K. However, as exhibited in the inset in Fig. 3(b), J-V curves remain stable during the same stressing process, inferring a reliable electrical performance at 100 K.

On-the-fly measurement is also performed to further investigate the dynamic performance of the β-Ga₂O₃ SBD at 300 and 100 K. Figure 3(c) shows the variation in J at 0.67 V when the device is exposed to various U_s, for a stressing time of 500 s. With increasing stressing time, J declines from 0.38 to 0.36 A/cm² for a reverse stressing voltage of −50 V. When U_s is enhanced to −100 V, J decreases to 0.35 A/cm². For the recovery process (the shadow region), within 20 s, J gradually recovers from 0.36 to 0.38 A/cm² after the device is biased to the U_s of −50 V for 500 s. When U_s is enhanced to −100 V, J only recovers to 0.37 A/cm² within 20 s. Compared to the monotonical reduction in J at 300 K, J fluctuates within a limited range around the original current density at 100 K, as shown in Fig. 3(d). At 300 K, the variation in J is associated with the occupied state of bulk traps in the depletion region. With longer stressing time, the number of empty bulk traps rises, leading to increased J.²⁶ The variation in J is also related to the change in U_s, for the depletion region is widened with enhanced U_s, which increases the number of empty bulk traps. At 100 K, the inertness of bulk traps leads to slight fluctuation of J.

Figures 3(e) and 3(f) show the dynamic on-resistance ratio as a function of stressing time at 300 and 100 K. The dynamic on-resistance ratio is defined as R_on,d/R_on,s, where R_on,d and R_on,s are the on-resistance measured in on-the-fly measurement and fresh state, respectively. When the device is exposed to U_s of −50, −70, and −100 V for a stressing time of 500 s, the ratio is measured to be 1.05, 1.06, and 1.09 at 300 K, respectively, as exhibited in Fig. 3(e). Meanwhile, the ratio dramatically increases from 0 to 5 s, indicating that most of the bulk traps are emptied during the first 5 s. During the recovery process, as depicted in the shadow region of Fig. 3(e), the ratio can recover to a fresh state after recovery times of 20 and 70 s when U_s is −50 V and −70 V at 300 K, respectively. However, after a recovery time of 100 s, the ratio only decreases to 1.01 when U_s is −100 V, indicating that 100 s is not enough for R_on,d to completely recover. Figure 3(f) depicts the ratio of 300 and 100 K during the stressing process. Compared to the results at 300 K, the ratio is limited to 1 ± 0.03 with all U_s at 100 K. The small variation in the ratio demonstrates good dynamic switching performance of the β-Ga₂O₃ SBD at cryogenic temperature.

Figure 4 displays the low frequency noise characteristics of the β-Ga₂O₃ SBD at 300 and 100 K. Figures 4(a) and 4(b) depict the noise current density spectra at 300 and 100 K from 1 Hz to 10 kHz. The low frequency noise consists of flicker noise and Lorentzian noise, and S_I spectra have a 1/f component caused by flicker noise and a 1/f² component caused by Lorentzian noise.²⁷ As shown in Fig. 4(a), S_I at 0.8 V is five or six orders of magnitude larger than S_I at 0.5 V due to the high forward current density with high forward bias. A Lorentzian hump, which is associated with a generation-recombination center in the bulk of the β-Ga₂O₃ SBD, can be observed at 300 K. S_I at 1.1 V is twelve or thirteen orders of magnitude larger than S_I at 0.8 V, as exhibited in Fig. 4(b). At the fixed bias of 0.8 V, S_I at 300 K is twelve or thirteen orders of magnitude larger than S_I at 100 K. The large low frequency noise level is considered to be related to the large forward current density at high temperature.²⁸ Meanwhile, the Lorentzian hump disappears at 100 K. Figures 4(c) and 4(d) show S_If as a function of f to eliminate the influence of the 1/f component. The Lorentzian hump in Fig. 4(c) could be acutely observed, indicating that bulk trap plays an important role in the electrical performance at room temperature. Nevertheless, as shown in Fig. 4(d), no Lorentzian hump could be found, inferring low activity of bulk trap at cryogenic temperature.

Trap characteristics of the β-Ga₂O₃ SBD are investigated by DLTS measurement. Figure 5(a) exhibits a temperature-scanning DLTS spectrum with a reverse bias U_R = −20 V, a filling pulse U_P = −0.5 V, a filling pulse width t_P = 0.1 s, and a measurement period T_W = 4 s. A majority carrier (electron) trap called E1 in this paper is revealed by a distinct peak around 305 K. Nevertheless, the DLTS signal reaches almost zero when the temperature is lower than 200 K, inferring that trap hardly affects the electrical performance of the β-Ga₂O₃ SBD at cryogenic temperature, matching well with the results in Figs. 3 and 4. It can be concluded that E1 is more likely to cause the dynamic performance and Lorentzian hump. As shown in Fig. 5(b), the activation energy for emission (E_emi) and capture cross section (σ_n) of E1 are extracted to be 0.82 eV and 1.32 × 10⁻¹³ cm² by the Arrhenius plot. During the DLTS measurement, the depletion region width with a U_R of −20 V is calculated to be 1.45 μm at 305 K, smaller than the thickness of the β-Ga₂O₃ homogeneous epilayer (7 μm). Therefore, the depletion region situates in the β-Ga₂O₃ epilayer, indicating that E1 is associated with a β-Ga₂O₃ homogeneous epilayer.

E1 has been found in the β-Ga₂O₃ epilayer in our previous work, together with a similar E_emi of 0.80 eV, a smaller σ_n of 6.40 × 10⁻¹⁵ cm², and a comparable N_Ta of 4.95 × 10¹³ cm⁻³,³⁰ indicating that E1 has larger capture capability in this study. E1 is also widely found in other literature studies,^31–35 demonstrating the universality of E1 and the essentiality of characterization. According to the studies, E1 is found in samples before and after proton irradiation,^31,33 and E1 found in this work is considered to be compensating centers in β-Ga₂O₃ and is associated with complexes involving native defects.^31,33

Figure 5(d) reports the temperature-scanning DLTS spectrum with different U_R. The positive peak corresponding to E1 stabilizes around 305 K, thereby supporting the conclusion that E1 is more likely to be a bulk trap rather than an interface trap.³⁶ Meanwhile, it can also be revealed that the emission process of E1 is independent of the electric field resulting from the fixed positive peak position.³⁶ 

Figure 5(e) investigates the temperature-dependent emission process by isothermal DLTS from 285 to 315 K. The positive peak corresponding to the emission time constant (τ_e) shifts toward a lower T_W as the temperature increases. τ_e decreases from 8.98 s at 285 K to 0.35 s at 315 K, indicating a temperature-enhanced emission process.

Figure 5(f) displays the temperature-scanning DLTS spectrum with various t_P. It can be observed that the peak DLTS amplitude remains stable with increasing t_P, as shown in the inset in Fig. 5(f). The saturation of the peak DLTS amplitude demonstrates that traps are almost filled within a duration of 0.05 s.

In summary, electrical properties of a relatively large-size (2 × 2 mm²) β-Ga₂O₃ SBD have been explicitly investigated for both room temperature and cryogenic temperature. Through J-V characteristics, n decreases from 1.34 at 50 K to 1.02 at 350 K, demonstrating the stable near-ideal Schottky characteristics. For leakage current characteristics, with a bias of −50 V, J at 100 K is one order of magnitude lower than that at 300 K, leading to outstanding blocking performance at low temperatures. From 50 to 350 K, qΦ_B extracted from C-V characteristics slightly declines from 0.70 to 0.65 eV. A variation in J can be readily observed during the stressing and recovery process at 300 K, while insignificant J fluctuation was recorded at 100 K. Meanwhile, a Lorentzian hump exists in low frequency noise spectra at 300 K but disappears at 100 K. The results of static characteristics, dynamic performance, and low frequency noise analysis point to high static performance and reliable switching properties of the β-Ga₂O₃ SBD at cryogenic temperature. The study establishes a solid foundation for the utilization of the large-size β-Ga₂O₃ SBDs as high power rectifiers for extreme low temperature environments.

This work was supported by ShanghaiTech University Startup Fund under Grant No. 2017F0203-000-14, the National Natural Science Foundation of China (Grant No. 52131303), the Natural Science Foundation of Shanghai (Grant No. 22ZR1442300), and, in part, by the CAS Strategic Science and Technology Program under Grant No. XDA18000000.

The authors have no conflicts to disclose.

Haolan Qu and Wei Huang contributed equally to this work.

Haolan Qu: Conceptualization (lead); Formal analysis (lead); Investigation (lead); Methodology (lead); Visualization (lead); Writing – original draft (lead); Writing – review & editing (equal). Wei Huang: Conceptualization (supporting); Methodology (supporting); Resources (equal); Writing – review & editing (supporting). Yu Zhang: Methodology (supporting); Writing – review & editing (equal). Jin Sui: Methodology (supporting); Writing – review & editing (supporting). Jiaxiang Chen: Writing – review & editing (supporting). Baile Chen: Methodology (supporting); Resources (supporting). David Wei Zhang: Methodology (supporting); Resources (supporting). Yuangang Wang: Methodology (supporting); Resources (supporting). Yuanjie Lv: Methodology (supporting); Resources (supporting). Zhihong Feng: Methodology (supporting); Resources (supporting). Xinbo Zou: Conceptualization (supporting); Funding acquisition (lead); Resources (equal); Supervision (lead); Writing – review & editing (equal).

The data that support the findings of this study are available from the corresponding author upon reasonable request.