Charge Traps in Wide-Bandgap Semiconductors for Power Electronics Applications

Abstract

Wide-bandgap semiconductors have been envisioned for power electronics applications because of their ability to operate at higher temperatures and higher applied voltages without breakdown. However, the presence of defects may cause device failure, necessitating a comprehensive understanding of material defects. This review provides a fingerprint of known defects in three envisioned semiconductors for power electronics: 4H-SiC, GaN, and β-Ga₂O₃. Via a detailed discussion of defects—the origins of electrically active charge traps—through their activation energies and capture cross-sections, we provide important insights into defect parameter distributions. This review not only serves as a reference but also offers a strategic roadmap for distinguishing between similar defects. Such knowledge is key for the development of more robust and efficient power electronic devices that can fully exploit the potential of wide-bandgap semiconductors.

1. Introduction

Although silicon (Si) and other narrow-bandgap semiconductors are exclusively used in consumer electronic devices, researchers have shown interest in wide-bandgap (WBG) semiconductors owing to their envisioned application in power electronics. Unlike standard silicon-based semiconductors, WBG materials have already demonstrated exceptional electrical characteristics that allow for operation at higher voltages, frequencies, and operating temperatures. Among the most envisioned WBG semiconductors are silicon carbide (SiC) and gallium nitride (GaN), with bandgaps of around 3.3 eV and 3.4 eV, respectively [1,2,3,4,5]. This wider energy bandgap, in contrast to silicon’s bandgap of 1.1 eV, allows for the application of higher voltages without the presence of electric breakdown, leading to devices with thinner drift regions and reduced on-state resistance.

The potential of WBG semiconductors to operate at elevated temperatures is crucial not only at higher power densities but also for applications in challenging environments, such as aerospace and automotive systems. In comparison, silicon carbide (SiC) devices can operate consistently at temperatures higher than 300 °C, while silicon devices are often constrained to approximately 150 °C [1]. This nature of electronic systems and the elimination of the need for complex cooling solutions improve the overall system efficiency and decrease fabrication and operation costs.

Moreover, the ability to achieve greater switching frequencies with WBG devices results in substantial enhancements in the efficiency of power conversion systems [1,2,3,6]. Note that increased operating frequencies enable the implementation of smaller passive components, which leads to more compact and lightweight power electronic devices and modules. This is essential in applications such as electric vehicles and renewable energy generation systems, where limitations in size and weight play a key role [7].

Notwithstanding their benefits, the commercial application of WBG semiconductors faces serious challenges, such as increased material costs and the requirement for reliable manufacturing techniques. Therefore, research is primarily motivated by the advantages of WBG semiconductor devices in terms of energy efficiency, system downsizing, and performance under extreme conditions.

Given the growing demand for greater efficiency and smaller footprints in power electronics, especially in the fields of renewable energy integration and e-mobility, it is expected that WBG semiconductors will have a significant impact on the advancement of power electronics and energy conversion components. Nevertheless, the semiconductor production process has not been completely optimised, and these materials still have high concentration of structural defects. It should be noted that defects such as imperfections in the crystal lattice behave as charge carrier traps and can be interpreted as energy states localised within the energy bandgap [8,9]. A comprehensive list of charge traps is presented in this paper, together with an analysis of the correlations between the activation energies and cross-sections of these charge traps. A complex study was conducted on notable WBG materials, namely silicon carbide (4H-SiC), gallium nitride (GaN), and gallium oxide (β-Ga₂O₃). All these materials are envisioned for power electronics applications since they have outstanding material properties [10,11,12]; see

Table 1. Summary of fundamental material properties: energy gap ($E_g$), mobility (µ), saturated velocity ($v_{sat}$), critical electric field ($E_c$), and thermal conductivity (λ) [10].

	${\mathit{E}}_{\mathit{g}}$ (eV)	µ (cm2/Vs)	${\mathit{v}}_{\mathit{s}\mathit{a}\mathit{t}}$ (107 cm/s)	${\mathit{E}}_{\mathit{c}}$ (MV/cm)	λ (W/mK)
Si	1.1	1400	1.0	0.3	150
4H-SiC	3.3	1000	2.0	2.5	370
GaN	3.4	1200	2.5	3.3	130
β-Ga2O3	4.9	300	1.1	8.0	13~29

and Figure 1. All of the above-mentioned materials have larger energy gaps than Si; however, each of them has slightly different exceptional properties that make it crucial for future power applications. 4H-SiC has the highest thermal conductivity, which makes it suitable for higher-temperature operations. GaN has outstanding charge carrier mobility and saturated velocity, which suppresses dynamic switching losses. On the other hand, β-Ga₂O₃ has an extremely high critical electric field, which drastically reduces the gate-drain distance, enabling the fabrication of compact and efficient semiconductor switches.

Note that even though there are other WBG semiconductor materials, such as diamond, aluminium nitride (AlN), boron nitride (BN), zinc oxide (ZnO), or other polymorphs of silicon carbide (e.g., 6H-SiC), the device fabrication technology is still immature [10,13,14].

It is important to note that the presence of structural defects and charge traps has a significant impact on the performance and reliability of semiconductor devices. Silicon carbide (4H-SiC) commonly exhibits structural defects such as silicon vacancies, carbon vacancies, and antisite defects, which occur when silicon atoms occupy carbon lattice positions (or vice versa) [15]. Similar structural defects occur in gallium nitride (GaN) [16]. Defects of this nature can introduce energy states that are localised deep within the energy bandgap and can negatively affect the lifetime of charge carriers and effective mobility. Defects of this nature have the ability to induce current collapse in high-electron-mobility transistors (HEMTs) and, hence, affect the breakdown voltage of power devices.

Gallium oxide (β-Ga₂O₃) is a newly developed material belonging to the WBG family which faces specific challenges [17,18]. The oxygen vacancies in this oxide semiconductor are highly concentrated and serve as donors, therefore influencing the controlled doping of the semiconductor. Furthermore, the existence of self-trapped holes and cation vacancies can lead to substantial alterations in the electrical properties of a material by introducing subsurface acceptor levels. An accurate determination of the activation energies of these traps is essential for understanding their influence on device performance. Deep-level traps with activation energies located far from the band edges can serve as recombination sites, thereby reducing the lifetime of carriers and the effectiveness of the device. In contrast, shallow traps may have less pronounced impacts but can still affect carrier transport (the Poole–Frenkel effect) and device characteristics. Equally important are the capture cross-sections of these charge traps, as they describe the rate of charge carrier capture and release processes. Larger capture cross-sections can result in fast trapping and de-trapping, which cause transient effects after voltage pulse application. Therefore, analysing the correlation between activation energies and capture cross-sections is crucial for predicting trap behaviour and its impact on device operation. This understanding can guide strategies for defect engineering and device design to minimise the negative impacts of these traps.

To fully analyse these defects and traps, advanced characterisation methods, such as deep-level transient spectroscopy (DLTS), thermally stimulated current (TSC), and photoluminescence (PL) spectroscopy, are usually applied [18,19,20,21]. The methods mentioned here offer deep insights into the energy levels and trap concentrations and capture the kinetics of different trap states. It should be noted that all the mentioned characterisation techniques exhibit several variations based on the excitation signal, operating conditions, or evaluation method. The TSC technique involves cooling a semiconductor device under voltage bias, which fills traps with charge carriers. This is followed by heating the device (without any applied voltage) at a constant rate while measuring the resulting current. This technique is particularly useful for the investigation of shallow traps and defects in high-resistivity materials. DLTS is a highly effective method used to analyse electrically active defects in semiconductors. As proposed by Lang, this technique is based on the application of voltage pulses to a semiconductor junction and the subsequent measurement of the capacitance transients as a function of temperature. DLTS offers data on defect energy levels, capture cross-sections, and concentrations. Several modifications of the DLTS technique include measurements of charge or current transients. PL spectroscopy uses the excitation of semiconductors with light and analysing the emission spectrum. This technique is particularly useful for studying radiative recombination processes and optically active defects. Among these methods, DLTS offers the highest sensitivity to charge traps. DLTS offers quantitative information on defect parameters, including activation energies and capture cross-sections. TSC provides valuable information on trap distributions and activation energies. PL spectroscopy provides data on the energy levels of radiative transitions without requiring any changes in temperature. On the other hand, DLTS requires the formation of a space charge region, which restricts its use to specific device types, whereas TSC and PL impose fewer limitations on device topology and can be used with a broader range of materials. The mentioned characterisation techniques frequently offer complementary information. For instance, DLTS can measure the electrical activity of defects, while PL can expose their inherent optical characteristics. TSC can bridge the gap by providing information on both shallow and deep traps.

Comprehensive knowledge of the charge traps in WBG semiconductors is essential for the progress of technology and achieving the full potential of these materials in power electronics and other high-performance applications. Gaining this understanding will open the path for future development of more reliable and efficient WBG semiconductor devices.

2. Charge Traps in Semiconductors

The charge trap classification is not unified and can be based on various criteria. The primary parameter describing the trap is the polarity of the trapped charge carrier. The traps can capture electrons or holes, preventing them from contributing to the electric current. Regarding the energy level of the trap, we recognise shallow and deep traps. Shallow traps have energy levels close to the conduction band (electron traps) or valence band (hole traps). If the energy level of the trap is lower than 3$k_{B}T$ ($k_{B}T$ is the thermal energy, where $k_B$ is the Boltzmann constant, and $T$ is thermodynamic temperature), the charge carriers can easily be thermally excited out of shallow traps and contribute to conduction. In contrast, deep traps have energy levels deep within the bandgap, far from the conduction or valence bands. Captured carriers are less likely to escape thermal excitation and remain trapped for longer periods, affecting the device performance.

The origin of defects is another popular classification of traps [22]. Material defects can be intrinsic or extrinsic. Intrinsic traps are essential properties of semiconductor materials and can result from imperfections in the crystal structure or impurities that occur during material growth. On the other hand, extrinsic traps are introduced during device fabrication processes, such as ion implantation, deposition, or thermal treatments. Another crucial factor to consider is the spatial arrangement of traps within a device. Traps may be uniformly distributed over the whole material or strongly concentrated at interfaces and surfaces. Interface traps, which lie at the interface between various materials (e.g., the semiconductor–oxide interface), can have a significant impact on the mobility of the effective mobility and the subthreshold swing in transistors. On the other hand, surface traps have the potential to impact the velocity of surface recombination and accordingly modify the performance of optoelectronic devices.

Another popular classification of defects is related to the defect dimensionality [23,24]. Here, we recognise point defects and extended defects. Point defects include vacancies, interstitials, antisites (atoms occupying the wrong lattice site in compound semiconductors), Frenkel defects (paired vacancy and interstitial), and Schottky defects (paired cation and anion vacancies). Extended defects are linear (1D), planar (2D), or bulk (3D) defects. Common linear defects are dislocations, whereas typical planar defects are grain boundaries. The voids or precipitates (empty spaces or clusters of different phases within the crystal) represent the bulk defects.

Their capture cross-sections and emission rates describe the capture and emission kinetics of the traps. The capture cross-section ${\sigma }_n$ represents the effective area within which a free carrier can be captured by the trap. Larger capture cross-sections indicate a higher probability of carrier capture during its transport through the semiconductor. The emission rate $e_n$, conversely, describes how quickly trapped carriers are released from the charge traps. As a result, in the case of electrons, the relationship between these parameters is expressed using the Arrhenius equation [20,25]

$$e_n={\sigma }_n{v}_{th}{N}_c\exp (-E_a/k_{B}T),$$

(1)

where $v_{th}$ is the thermal velocity of electrons, $N_c$ is the effective density of states in the conduction band, and $E_a$ is the activation energy of the trap defined as a difference between the bottom edge of the conduction band $E_c$ and the trap energy level $E_t$. Substitution of the effective mass of the free electron $m_n^{*}=v_{th}{N}_c/T^2$ provides us with a linearisation of Equation (1) as follows

$$\mathrm{l}\mathrm{n}(e_n/T^2)=\left({\sigma }_n{m}_n^{*}\right)-E_a/k_{B}T,$$

(2)

and gives us an opportunity to estimate the activation energy and capture cross-section. It is interesting to note that while the Arrhenius equation is empiric only, a more theoretical approach for reaction rates provides the Eyring equation [26,27]

$$e_n=(k_{B}T/h)\exp (-{\Delta G}_n^{*}/RT),$$

(3)

where $h$ is the Planck constant, $R$ is the gas constant, and ${\Delta G}_n^{*}$ is the Gibbs free energy of activation, which consists of ${\Delta H}_n^{*}$, the enthalpy of activation, and ${\Delta S}_n$, the entropy of activation, as

$${\Delta G}_n^{*}={\Delta H}_n^{*}+T{\Delta S}_n.$$

(4)

The enthalpy of activation represents the energy required to move an electron from its localised state in the trap to a delocalised state at the bottom edge of the conduction band. Contrary to activation energy, the enthalpy of activation is temperature-independent, making it a more fundamental parameter for describing trap behaviour across different temperatures. The entropy of activation reflects the change in the number of available microstates or the degree of disorder as the charge moves between the localised trap state and the delocalised band state. Unlike the enthalpy of activation, the entropy of activation can have a temperature dependence, especially for traps with strong lattice coupling.

In WBG materials, where strong electron–phonon coupling is often observed, the entropy of activation can play a particularly important role in determining trap behaviour.

Moreover, a combination of Equations (1) and (4) gives us the following relationships:

$$E_a={\Delta H}_n^{*}+RT.$$

(5)

$${\sigma }_n={\sigma }_0\exp ({\Delta S}_n/R).$$

(6)

where ${\sigma }_0$ is a temperature-independent prefactor. Note that the capture cross-section provides a possibility to distinguish between traps that might have similar activation enthalpies but different emission or capture kinetics. Hence, both activation energy and capture cross-section are needed to represent a charge trap fingerprint.

Charge traps in semiconductor devices may have various origins and significantly impact device performance, particularly in power electronics applications. Intrinsic traps arise from lattice imperfections such as vacancies or dislocations. Impurities are extrinsic defects which can also be recognised as unintentional doping. In the case of heterostructure, the interface always generates interfacial states due to symmetry breaking. It should be noted that even the device operation may cause additional defects. Hot carriers are produced when the current flowing through the device attains high energy beyond the lattice temperature and can create lattice defects due to impact ionisation. Also, mechanical, thermal, or voltage bias stress can induce defects that can serve as charge traps.

Semiconductor physics usually understands the impact of charge traps only through suppressing mobile charge carrier density; however, the answer is more complex in electronic devices. Trapped charge influences the surrounding region by its electrostatic field and can cause threshold voltage instability. This effect is crucial in power metal–oxide–semiconductor field-effect transistors (MOSFETs) and insulated-gate bipolar transistors (IGBTs), where a stable and well-defined turn-on voltage is essential for reliable switching operations. Furthermore, localised electric field enhancements can reduce the breakdown voltage of power devices. This effect is particularly significant in high-voltage applications.

Defects in the drift region of power devices also can act as scattering centres, increasing the on-resistance. This parameter is critical in power electronics as it directly affects conduction losses and overall device efficiency. Also, the dynamic nature of charge traps is important. Charge trapping and de-trapping processes can influence the dynamics of the depletion region formation and collapse, affecting the switching speed of power devices. Last but not least, there is the device reliability issue. The accumulation of charge in traps over time can lead to the drift of device parameters and eventual device failure. This phenomenon is critical in power electronics, where long-term voltage stress is common [28,29].

3. Trap Parameters

All three discussed semiconductors have been envisioned for power electronics applications; however, they differ in all material or electronic properties. The 4H-SiC structure belongs to the hexagonal crystal system with a space group of P63mc. GaN crystallises in the wurtzite structure under ambient conditions, belonging to the hexagonal crystal system with a space group of P63mc, similar to 4H-SiC. β-Ga₂O₃, the most stable polymorph of gallium oxide, crystallises in a unique monoclinic structure with a space group of C2/m.

Regarding the charge transport, high carrier mobilities and saturation velocities characterise drift diffusion as the main charge transport mechanism in 4H-SiC. Under low electric fields, the mobility is limited by the scattering of phonons, whereas at higher fields, the initiation of velocity saturation occurs due to the emission of optical phonons. The anisotropic nature of the 4H-SiC crystal structure results in direction-dependent mobility, with higher mobility along the c-axis compared to the basal plane. Note that GaN-based devices are utilised in AlGaN/GaN heterostructures where the formation of a two-dimensional electron gas (2DEG) at the interface leads to high-mobility electron transport. Nevertheless, drift diffusion is the primary transport mechanism in bulk GaN, analogous to 4H-SiC. β-Ga₂O₃ primarily relies on drift diffusion transport, although it exhibits considerably lower mobilities compared to 4H-SiC and GaN because of robust electron–phonon coupling. The anisotropic mobility of β-Ga₂O₃ is characterised by the strongest mobility in the [010] direction and the lowest mobility in the [100] direction. Even though the semiconductor materials differ, this review aims to find a common way of trap state evaluation.

3.1. Silicon Carbide (SiC)

Silicon carbide, particularly its 4H polytype (4H-SiC), has been proposed as a promising WBG semiconductor for high-power, high-temperature, and high-frequency applications. The 4H-SiC polytype, with its wide bandgap of 3.23 eV, high breakdown electric field of up to 4 MV/cm [30], and superior thermal conductivity of 3.47 W/cm·K [31], offers significant improvement in contrast with conventional silicon-based devices. However, the application of 4H-SiC in electronic devices is suppressed by the presence of various defects that can significantly impact material properties and device performance [32,33,34,35]. Defects in 4H-SiC can be broadly categorised into point defects, extended defects, and impurity-related defects. Point defects, including silicon vacancies (V_Si), carbon vacancies (V_C), antisites (Si_C and C_Si), and interstitials (Si_i and C_i). Extended defects, such as stacking faults, dislocations, and in some cases, micropipes, can act as efficient recombination centres and current leakage paths, degrading device properties. High-quality 4H-SiC crystal and epitaxial layer formation is difficult because of the high-temperature procedures and the complex stoichiometry of the material. Such growth conditions can result in the incorporation of both point defects and extended defects. Moreover, subsequent processing procedures after growth, such as ion implantation and high-temperature annealing, have the potential to introduce novel defects or modify existing ones.

The Z₁/Z₂ centre in 4H-SiC represents one of the most reported and extensively studied deep-level trap states in SiC [36,37,38,39,40]. Initially observed by DLTS measurements, this defect is assumed as two closely spaced levels, denoted as Z₁ and Z₂. Due to their proximity in energy (0.65~0.7 eV) and similar behaviour, they are often collectively referred to as the Z₁/Z₂ centre. The most common hypothesis, supported by a combination of experimental observations and first-principles calculations, attributes this defect to the carbon vacancy in 4H-SiC. Specifically, the Z₁ component is associated with the vacancy’s double-negative to neutral (=/0) transition at the hexagonal lattice site, while Z₂ corresponds to the same transition at the cubic site. One of the most challenging aspects of the Z₁/Z₂ centre is its remarkable thermal stability. In detail, the defect is observed even after high-temperature annealing treatments where the material is heated up to 1700 °C, making its elimination particularly difficult once it is formed [37,41]. This stability necessitates careful control during crystal growth and innovative post-growth treatments for defect mitigation.

The EH6/7 centre in 4H-SiC is a complex deep-level trap state consisting of two closely spaced levels known as EH6 and EH7 [39,40,42]. From a spectroscopic perspective, the EH6/7 centre generates energy levels at around 1.55–1.65 eV. The complex deep levels, located closer to the valence band than the Z₁/Z₂ centre, greatly influence the electrical properties of 4H-SiC, namely in terms of carrier trapping and recombination kinetics. Several investigations have examined the atomic configuration and source of the EH6/7. Initially, it was believed to be associated with a carbon vacancy, analogous to the Z₁/Z₂ centre. Nevertheless, recent theoretical simulations and empirical data indicate a more complicated scenario. Presently, the EH6/7 centre is attributed to the carbon antisite–vacancy pair (C_Si-V_C) [43].

An EH4 trap in 4H-SiC is a notable defect characterised by an activation energy of around 1 eV. Preliminary investigations indicated a potential correlation with defects associated with carbon. Nevertheless, new studies integrating experimental findings with theoretical calculations have suggested a potential connection between the EH4 trap state and a di-carbon antisite defect (C_Si-C_Si) or carbon antisite–vacancy pair [44]. The validity of this concept is substantiated by its documented occurrence in environments enriched in carbon and its response to specific thermal treatments. Defects of this nature have the ability to induce current collapse in high-electron-mobility transistors (HEMTs) and, hence, affect the breakdown voltage of power devices. Yellow luminescent emission, often observed in GaN, is also attributed to specific defect configurations. The Ci1, Ci3, Ci4, and Ci5 defects in 4H-SiC represent a series of carbon interstitial-related defects [39,40,42,45]. These structural defects stand for simple carbon interstitials, more complex carbon interstitial clusters or carbon interstitial–vacancy pairs, and carbon di-interstitial configurations [39].

Figure 2 illustrates the distribution of charge traps within the energy bandgap. The impurity-related defects occupied shallow states, while intrinsic defects represented deep states. Interstitial defects, in particular, create a state with an activation energy of 1 eV or more. Noteworthy, the capture cross-section is proportional to the disorder caused by the defect. The broad range of Z₁/Z₂ centre capture cross-sections illustrates different charge states of the trap.

3.2. Gallium Nitride (GaN)

Gallium nitride (GaN) has been suggested to be a next-generation semiconductor for optoelectronics and power electronics. With its direct bandgap of 3.4 eV, high electron mobility, and excellent thermal conductivity, GaN offers ideal performance in light-emitting diodes (LEDs), high-electron-mobility transistors (HEMTs), and power electronic devices [46]. Despite these advantages, the presence of various structural defects in GaN significantly impacts its material properties and device performance, presenting ongoing challenges for technologists and device engineers [47,48].

The wurtzite crystal structure of GaN, together with the difficulties experienced in epitaxial growth processes, results in the formation of a complex group of defects. These defects can be classified as point defects, extended defects, and impurity-related defects. All of them significantly influence the electrical and optical characteristics of GaN-based devices.

Many studies have documented point defects in GaN, such as vacancies (V_Ga, V_N), interstitials (Ga_i, N_i), and antisites (Ga_N, N_Ga) [48,49]. Continuing to provide a major difficulty in GaN epitaxial layers are extended defects, primarily threading dislocations with densities generally ranging from 10⁶ to 10⁹ cm⁻². These dislocations, which include edge, screw, and mixed types, can function as non-radiative recombination centres, pathways for current leakage, and places for charge carriers to scatter, therefore significantly reducing the efficiency and reliability of electronic devices. The optimisation of GaN crystals and epitaxial layers of superior quality is further hindered by the absence of native substrates, hence requiring heteroepitaxial growth on foreign substrates such as sapphire, silicon carbide, or silicon. The mismatch in lattice and differences in thermal expansion coefficients result in strain and the formation of additional defects.

Recent progress in growth methodologies, such as metal–organic chemical vapour deposition (MOCVD) and molecular beam epitaxy (MBE), has greatly enhanced the quality of GaN material. Nevertheless, some significant problems remain, including the existence of carbon impurities and their role as compensatory centres, the development of Mg-H complexes in p-type GaN, and the difficulty of attaining consistent p-type doping [50,51]. Nitrogen vacancies introduce deep donor levels within the bandgap of GaN. The nitrogen vacancy can exist in multiple charge states, commonly denoted as a positive (V_N⁺), neutral (V_N⁰), and negative state (V_N⁻). The ability to transition between these charge states is fundamental to its role as a charge trap. As a result, the defect may be observed with various activation energies. Also, gallium vacancies in GaN represent a significant class of intrinsic point defects since they can exist in multiple states from neutral (V_Ga⁰) to triple negative (V_Ga³⁻). The ability to accommodate up to three negative charges makes gallium vacancy a potentially significant charge trap, especially in n-type GaN [48,49,52].

The nitrogen antisite defect (N_Ga) in GaN represents a point defect where a nitrogen atom occupies a gallium lattice site. The N-antisite defect can exist in multiple charge states, ranging from negative (N_Ga⁻) to double positive (N_Ga²⁺). This ability to accommodate different charge configurations makes N_Ga a versatile charge trap capable of interacting with both electrons and holes [48,49,53].

Gallium and nitrogen interstitials in gallium nitride (GaN) represent important point defects. Nitrogen interstitial (Ni) introduces deep levels that can act as both electron and hole traps, depending on the charge state, similar to the antisite defect [48,49,54].

Figure 3 depicts the distribution of charge traps in GaN. Interestingly, there are no shallow trap states, and all states are above 0.5 eV. Impurity-related defects are mostly between 1.0 and 1.35 eV (except oxygen located at 0.6 eV). Interstitial atoms create states from 0.65 to 1.22 eV, while vacancies range from 1.45 to 1.62 eV. The dislocations generate states at a wide range of the energy bandgap.

3.3. Gallium Oxide (β-Ga₂O₃)

Gallium oxide (β-Ga₂O₃) has recently been introduced as a promising candidate for WBG oxide semiconductors, attracting significant attention in the field of power electronics. With its remarkable properties, including a wide bandgap of 4.8–4.9 eV, high breakdown electric field of 8 MV/cm, and good electron mobility, β-Ga₂O₃ offers potential advantages over traditional wide-bandgap semiconductors such as SiC and GaN [55]. However, like all semiconductor materials, the presence of various defects in β-Ga₂O₃ significantly influences its electrical properties [56].

The exceptional anisotropic features of the monoclinic crystal structure of β-Ga₂O₃ give rise to both possibilities and challenges in relation to the formation and behaviour of defects. Point defects in β-Ga₂O₃, such as vacancies (V_Ga, V_O), interstitials (Ga_i, O_i), and antisites (Ga_O, O_Ga), are essential factors in influencing the electrical properties of the material [57,58]. Notably, the oxygen vacancy has been recognised as a primary shallow donor, playing a significant role in the intrinsic n-type conductivity of β-Ga₂O₃. Conversely, gallium vacancies function as deep acceptors and they may contribute to compensatory effects.

Although the concentrations of extended defects, such dislocations and stacking faults, in β-Ga₂O₃ are typically smaller than those of other WBG semiconductors, they have been detected. Defects of this nature can function as carrier traps and recombination sites, therefore potentially compromising the functioning of the device.

Developing high-quality β-Ga₂O₃ crystals and epitaxial layers has many constraints. Although melt-growing techniques such as edge-defined film-fed growth (EFG) and Czochralski procedures have achieved success in generating bulk crystals [56], the management of point defects and impurities maintains a crucial challenge. Active development of epitaxial growth techniques, such as molecular beam epitaxy (MBE) and halide vapour phase epitaxy (HVPE), is being undertaken to attain layers of superior quality with precise doping [59].

Furthermore, faults associated with impurities also have a substantial impact on β-Ga₂O₃. The deliberate introduction of components such as Si, Sn, or Ge for n-type conductivity, and Mg or Fe for semi-insulating behaviour, brings further intricacies in the phenomenon of defect development and interaction. Optimising device performance requires a thorough understanding of the behaviour of these dopants and their possible connections with intrinsic defects.

Experiments using DLTS and PL techniques have identified the E1 defect in β-Ga₂O₃ as an intrinsic point defect. It is described as a deep donor level with an activation energy ranging from 0.55 to 0.65 eV. A precise microscopic origin of the E1 defects is still unknown. Various hypotheses have been suggested, such as either (i) the presence of oxygen vacancies, (ii) the existence of gallium interstitials, or (iii) the existence of defect complexes that include several point defects or impurities (e.g., Si_Ga-H or Sn_Ga-H complexes) [58,60].

This study characterises the E2(Fe) defect in β-Ga₂O₃ as a deep trap level linked to iron impurities in the crystal lattice. The replacement process is made relatively simple by the comparable ionic radii of Fe³⁺ and Ga³⁺, which enables the relatively effortless integration of iron impurities during crystal formation or subsequent contamination after growth. The activation energy of the E2(Fe) defect is commonly observed to range from around 0.78 to 0.85 eV [58,60]. Denoted as E2*, the defect has comparable energy to the E2 trap state, but it varies in terms of the capture cross-section [58,61]. The common assumption is that it is an intrinsic defect, such as a gallium vacancy. It should be noted that the crystal structure of β-Ga₂O₃ has two distinguishable Ga sites (tetrahedral and octahedral), which may result in various configurations of V_Ga with slight differences in their energies and capture cross-sections [58].

Preliminary theories ascribed the E3 defect to external contaminants, suggesting that transition metals including titanium (Ti), magnesium (Mg), and cobalt (Co) could be prospective candidates. Nevertheless, the fact that related trap states can be generated by exposing particles to protons and neutrons implies a more complex source, maybe including intrinsic defects or complexes of defects [58]. The simultaneous existence of the E3 defect in both as-grown samples and its generation by irradiation emphasises the complex procedure of defect formation and stability in β-Ga₂O₃.

The E4 defect, exclusively detected after material irradiation, offers an interesting study in the impact of radiation on β-Ga₂O₃ [58,61]. Typically reported at an energy level of 1.0–1.2 eV, the E4 centre is crucial when determining the radiation hardness of devices based on β-Ga₂O₃. The unique appearance of this defect after exposure to radiation implies a causative mechanism associated with displacement damage or electronic excitation processes caused by high-energy particles [58].

On the other hand, there is a significant association between the E5 and E6 defects and the level of germanium (Ge) doping in β-Ga₂O₃ [57]. The mentioned correlation offers significant insights into the interactions between dopant atoms and the host lattice. Given its group IV nature, germanium has the capacity to occupy both gallium and oxygen sites in the crystal structure of β-Ga₂O₃, which may result in the development of complex defect configurations. The existence of these defects associated with germanium (Ge) can greatly impact the effectiveness of Ge as a dopant and, therefore, contribute to compensating processes that influence the overall electrical characteristics of the material [57,58].

Figure 4 illustrates the distribution of charge traps in β-Ga₂O₃. Obviously, Ge-related defects are shallow, while other impurity-related defects create deep levels. Interestingly, all reported trap states are quite distinguishable, even though their origin is still unclear.

4. Conclusions

In this comprehensive review, trap states in 4H-SiC, GaN, and β-Ga₂O₃, three semiconductors intended for power electronics applications, are critically analysed. Using cataloguing and characterisation of defects based on their activation energies and capture cross-sections, we provide a unique fingerprint for each documented defect, hence establishing a significant reference library for researchers and industry experts.

This paper not only examines the origins of observable defects but also offers a systematic approach to correctly identify what is causing them. A detailed understanding of the concepts drives the development of fabrication technology and the implementation of process optimisation procedures. By introducing a full defect landscape, this review acts as a starting point for future research, supporting more focused strategies for defect mitigation. The broad spectrum of activation energies and capture cross-sections emphasises the complex nature of defect physics in wide-bandgap semiconductors. Future investigations should prioritise the development of precise defect characterisation techniques to mitigate charge traps. Each material presents unique challenges. 4H-SiC suffers from intrinsic complex defects such as Z₁/Z₂, while the impurity-related localised states are shallow levels only. GaN exhibits a wide range of deep levels caused by immature fabrication technology, allowing for the presence of impurities as well as structural defects. Especially vacancies and dislocations illustrate extremely wide range of material disorder level represented by the capture cross-sections. Novel materials such as β-Ga₂O₃ are not investigated in detail; however, specific impurities such as Fe, Ti, or Ge generate deep levels. As the investigation of charge traps matures, there is a growing need for standardised measurement and reporting protocols for charge trap characteristics. This will facilitate more meaningful comparisons across different research groups and accelerate progress in the field. This review aims at an introduction of charge trap maps using activation energies and capture cross-sections. Note that even though WBG semiconductors discussed in this review differ in material properties, electric properties, or even in the charge transport mechanism, the charge trap maps can give chance to compare these materials. While this review focused on 4H-SiC, GaN, and β-Ga₂O₃, attention should also be directed towards emerging ultra-wide-bandgap materials such as AlN, diamond, and BN. Revealing charge trap behaviour in these materials will be crucial as they progress towards practical device applications.

The trapping phenomenon in power electronics devices has significant impact on other device-related fields. The high concentration of charge traps is not a challenge for fabrication technologies, also the device architectures must be introduced. Vertical GaN structures or novel β-Ga₂O₃ transistor designs will elucidate how charge traps manifest and impact performance in these geometries. The investigation of the long-term evolution of charge traps under different stress conditions is still an issue worthy of considerable focus. Future development of accurate lifetime prediction models that account for trap-related degradation mechanisms will be essential for the commercial application of WBG materials. Fundamentally, this understanding is necessary for fully exploiting the capabilities of WBG semiconductors in power electronics and enabling the development of more effective, reliable, and high-performance devices that have the potential for revolutionising energy conversion and management in various fields.