Access Official Technical PDF

Share Summary:LinkedInEmail

For engineers developing β-Ga2O3 power devices, the gate stack must provide suitable electrical insulation while maintaining stable operation under applied bias. A study of an Al-rich HfAlO/β-Ga2O3 MOS capacitor examines both baseline electrical properties and charge trapping during constant-voltage stress (CVS). Its results highlight why oxide defects and interface states matter when assessing a dielectric for a transistor gate.

Why evaluate HfAlO in a Ga2O3 MOS capacitor?

β-Ga2O3 is of interest for power electronics because of its ultra-wide bandgap and high theoretical breakdown field. However, material properties alone do not determine device performance. The gate dielectric and its interface with the semiconductor can influence parameters such as threshold voltage, channel mobility, leakage, and stability.

HfAlO combines hafnium oxide and aluminum oxide in a high-permittivity dielectric. In the reported structure, the film was deposited by atomic layer deposition (ALD) using alternating oxide cycles, with an HfO2-to-Al2O3 cycle ratio of 1:4. The resulting Al-rich film was approximately 30.71 nm thick. The capacitors were built on β-Ga2O3 substrates and evaluated using capacitance-voltage (C-V), current-voltage (I-V), and CVS measurements.

High-permittivity dielectrics can provide a lower equivalent oxide thickness (EOT) than a thicker, lower-permittivity film with comparable capacitance. That makes dielectric constant useful to consider alongside leakage and interface behavior. In this study, the extracted relative permittivity of HfAlO was 10.74, corresponding to an EOT of 11.15 nm.

Baseline electrical properties and interface quality

Before stress testing, the C-V measurements showed hysteresis of 0.92 V and negligible frequency dispersion between the reported 1 MHz and 1 kHz measurements. The extracted interface state density was approximately 1.16 × 1012 eV-1 cm-2. These measurements provide a baseline for evaluating how the gate stack responds when a voltage is applied for an extended period.

The study also reported a gate leakage current density of 1.18 × 10-5 A/cm2 at a gate voltage of +10 V. From the C-V-derived parameters, the effective electric field at that bias was estimated as 5.87 MV/cm. These are measurements from the capacitor structure; they should not be treated as direct predictions of transistor performance or long-term device lifetime.

The pre-stress flat-band voltage was higher than its calculated ideal value. The authors attributed this shift to negative effective defects in the dielectric and/or at the HfAlO/β-Ga2O3 interface. Their analysis considers several charge populations rather than assigning the shift to a single defect type.

What constant-voltage stress reveals about charge trapping

During CVS testing, capacitors were held at positive gate biases and their C-V response was measured over time. The curves shifted in the positive direction as stress voltage or duration increased. In this structure, that shift was consistent with an increase in negative effective charge in the dielectric and/or at the interface.

The response is relevant to device operation because trapped charge can change the electrostatic conditions of the gate stack. A change in flat-band voltage is therefore a useful indicator of how the capacitor responds to electrical stress, although it does not by itself identify where every trapped charge resides.

The analysis separates effective charge into contributions from:

  • Border traps: defects in the near-interface oxide that can exchange charge with the semiconductor, including through tunneling.
  • Interface traps: electronic states at the HfAlO/β-Ga2O3 boundary whose occupancy can change with the surface Fermi level.
  • Fixed and deep oxide charges: charge populations that also contribute to the measured effective oxide charge.

As stress time increased, the extracted effective charge density rose, particularly over the reported 10-to-200-second range, and then approached saturation at longer durations. The authors also found that the relative importance of border and interface traps varied with stress bias. Border traps contributed more to the flat-band shift at medium stress bias, while interface traps had greater influence at larger bias. The measurements therefore point to a stress response shaped by multiple defect populations, not a single uniform trap mechanism.

Implications for gate-stack evaluation

For system designers and device developers, these findings reinforce the importance of testing gate dielectrics under more than an initial C-V or leakage measurement. A useful evaluation should consider how the flat-band voltage and extracted trap-related parameters change with both applied bias and time.

The reported HfAlO capacitor showed a combination of moderate-to-high permittivity, low measured gate leakage at the stated bias, and measurable stress-induced charge trapping. Those properties make the material a candidate for further investigation in β-Ga2O3 MOS devices, but the observed shifts also underline the need to account for interface and oxide defects when designing and qualifying a gate stack.

In practical terms, CVS and frequency-dependent C-V measurements can help distinguish baseline electrical characteristics from stress-related changes. For a Ga2O3 MOS capacitor intended to inform power-transistor development, tracking these changes provides a more complete view of dielectric behavior than considering permittivity or leakage alone. The study supports continued evaluation of Al-rich HfAlO while identifying charge trapping as a key factor for device stability.

Access Official Technical PDF

About The Author