Electrically active defects in Ta-doped β-Ga₂O₃ grown using the optical floating zone method Open Access

Beta-phase gallium oxide (β-Ga₂O₃) is an ultrawide bandgap (UWBG) semiconductor (∼4.8 eV) that has gained considerable interest in recent years, owing to its impressive Baliga figure of merit (BFOM),¹ a theoretical breakdown field of 8 MV cm⁻¹, and ability to grow large-area single crystals from the melt.2,3 These advantages make β-Ga₂O₃ a strong candidate for deep ultraviolet photodetectors4,5 and high-power electronics,⁶ potentially supplementing more established wide bandgap materials, such as gallium nitride (GaN) and silicon carbide (SiC).7,8 Conventional group IV dopants, Si, Ge, and Sn, can be readily introduced to produce n-type material, with free electron concentrations in the range of 10¹⁶–10¹⁸ cm⁻³.9–13 However, there has been growing interest in exploring alternative dopants that can improve doping efficiencies while inducing minimal strain. Tantalum (Ta) has emerged as a promising candidate as a donor doping impurity due to its favorable properties.14–16 First principle calculations based on density functional theory (DFT) indicate that Ta-doping can be utilized to enhance the conductivity of β-Ga₂O₃.¹⁷ This is primarily due to Ta possessing five valence electrons (Ta⁵⁺) rather than four when doping with Si (Si⁴⁺) or Sn (Sn⁴⁺), which facilitates more effective donor substitution at Ga³⁺ sites.17–20 In addition, Ta has an ionic radius of 0.064 nm, which is closer to that of Ga (0.062 nm) when compared to both Si (0.040 nm) and Sn (0.069 nm),17–20 reducing the likelihood of structural defects. Recently, Hall effect measurements on β-Ga₂O₃:Ta revealed a superior carrier concentration and mobility in comparison to Sn-doped β-Ga₂O₃ material with the same nominal doping concentration.¹⁶ 

Most commercially available β-Ga₂O₃ substrates are grown via the edge-defined film-fed growth (EFG) method, which can maintain high growth rates for large diameter crystals while being cost-effective.21,22 In 2023, Novel Crystalline Technology, Inc. (NCT) reported β-Ga₂O₃ substrates with diameters of 6 in. grown using the vertical Bridgman technique.²³ However, substrates grown using these methods require large crucibles that are prone to contamination, and the overall manufacturing cost is highly dependent upon the cost of iridium,²⁴ thereby incentivizing growth in crucible-free environments using methods such as Optical Floating Zone (OFZ).16,25 Single crystals with high purities can be achieved using OFZ with precise control of the doping level.²⁶ This is particularly beneficial for defect-related studies within research settings, which do not require large-area wafers.

In this work, both optical and electrical characterization techniques have been applied to Ta-doped β-Ga₂O₃ to investigate the electrically active states present in the material. This has been achieved through deep-level transient spectroscopy (DLTS)²⁷ in conjunction with high-resolution Laplace-DLTS (L-DLTS),28,29 which can provide information about electron traps within ∼1 eV from the conduction band minimum (CBM).30,31 In addition, photoluminescence (PL) has been applied to investigate the broad luminescence bands observed in β-Ga₂O₃ at ∼3.0–3.6 eV.32–36 Despite the inherent advantages of OFZ growth, numerous studies have reported the presence of deep-level defects that may hinder device control and reliability. Perrier et al.²⁶ applied DLTS to Si-doped substrates grown via OFZ and observed four deep-level traps—so called, E_s, E1, E2, and E3. Cui et al.¹⁹ reported a single defect level located at 0.73 eV relative to the conduction band, using conventional-DLTS on Ta-doped β-Ga₂O₃ grown by the OFZ method. This level possesses a similar electron activation energy for electron emission as the E2 trap, which has been firmly associated with Fe_Ga defects.31,37–40 It has been argued that the E2 trap has acceptor properties, and it is commonly reported in bulk β-Ga₂O₃ in significant concentrations of ∼10¹⁶ cm⁻³,37,39,41 making it a prominent compensating center. From our results, deep levels with activation energies for electron emission close to those for the E_s, E9, E1, E2a, E2b, and E3 traps have been observed, as per the labeling convention used by Perrier et al.²⁶ and Wang et al.⁴² The origin and nature of each of these traps will be discussed.

Using the optical floating zone method (CSC made FZ-4000 system), Ta-doped β-Ga₂O₃ crystals were grown - growth details are also provided in Ref. 16. For the initial material preparation, high-purity Ga₂O₃ (5N) and Ta₂O₅ (4N) powders were sourced from Alfa Aesar. The precursor materials were precisely weighed and thoroughly homogenized through grinding to ensure uniform Ta distribution throughout the mixture, with Ta concentration set at 0.05 mol. %. The homogenized powder was subsequently loaded into a latex tube to create a cylindrical feed rod (10 mm in diameter and 60 mm in length), followed by hydrostatic compression at 70 MPa. The feed rod was extracted from the latex tube and subjected to sintering at 1350 °C for a 24-h period, resulting in a highly compact ceramic rod with minimal void content. The crystal growth process utilized a previously grown β-Ga₂O₃ crystal with (100) orientation as the seed. The OFZ system’s 4000 W halogen lamps, coupled with focusing mirrors, generated the necessary heat to melt the feed rod. To maintain optimal growth conditions and ensure a stable crystal–melt interface, both the seed and feed rods were rotated in opposite directions at 20 rpm. Growth proceeded at a controlled rate of 10–12 mm/h under compressed dry air flow (2 l/min). The resulting crystal boule was sliced into wafers using diamond wire cutting technology, and the wafer surfaces were prepared through chemical mechanical polishing to achieve the surface quality required for device fabrication.

For electrical measurements, topside Ni/Au (30/150 nm) Schottky contacts and large area backside Ti/Au (30/150 nm) ohmic contacts were fabricated to form vertical Schottky barrier diodes (SBDs). Prior to metal deposition, the samples were cleaned using acetone, isopropyl alcohol, and de-ionized water. The samples were subsequently cleaned with HCl and then H₂O₂ held at 85 °C. Following this, ohmic contacts were produced using e-beam evaporation, with a post-deposition heat-treatment for 5 min at 450 °C in a N₂ atmosphere. Subsequently, circular topside Schottky contacts were deposited with diameters between 200 and 500 μm using a lithography and lift-off process. For optical analysis, as-grown samples were studied with no post-growth processing, except standard cleaning procedures. The PL measurements were performed in a closed cycle cryostat with helium exchange gas, using a 255 nm UV-LED for excitation at a power density of 71 μW/cm².

Figure 1(a) shows the current density-voltage (J–V) characteristics of a typical SBD, which were measured in the voltage range from −10 to 1 V using a HP 4140 pA meter at temperatures of 295 and 40 K on a 400 μm diameter device. Using the same diode, capacitance–voltage (C–V) measurements [Fig. 1(b)] were conducted using a HP 34192A impedance analyzer with voltage varying from −8.0 to 0 V at a 1 MHz frequency of the probing AC signal. Across this voltage range, at both 295 and 40 K, the phase angle was between 84° and 89°, indicating a good quality of the diode. The spatial profile of the concentration of uncompensated donors [$Nd+(W)$] was extracted from the C–V dependencies [Fig. 1(c)], using a static dielectric constant (ε) value for (100) oriented β-Ga₂O₃ of 10.2.⁴³ This yielded a carrier concentration in the range (1.0–1.2) × 10¹⁸ cm⁻³ down to ∼100 nm from the metal–semiconductor (M–S) junction. The free carrier concentration was also determined for an unintentionally doped (UID) β-Ga₂O₃ sample, which was grown under identical conditions and from the same source powder batch used to produce the β-Ga₂O₃:Ta material. This UID crystal underwent the same SBD processing steps outlined in the section titled Experimental Details, and subsequent C–V measurements were performed on a 400 μm diameter SBD. The UID sample exhibits a lower free carrier concentration in the range (2.5–6.5) × 10¹⁷ cm⁻³, indicating that Ta-doping increases the n-type conductivity. It is important to note that the C–V measurements on the UID material, using the same voltage range, probed deeper into the sample because of the lower carrier concentration. Following this, a capacitance–temperature (CT) scan was conducted [Fig. 1(d)] by cooling the sample down to 20 K without any applied bias, then at low temperature, a bias of −5.0 V was applied. Thereafter, the temperature was increased with a ramping rate of 2 K/min up to 430 K. The results shown in Figs. 1(a) and 1(d) reveal no significant sign of carrier freeze out down to 20 K, confirming that any doping donor levels in the β-Ga₂O₃:Ta sample are shallow. It is important to note, however, that additional unintentional shallow donor impurities, such as Si,44,45 may be contributing to the free carrier concentration as well as to the CT dependence shown in Fig. 1(d). However, a comparison with the UID material shows that Ta is the dominant donor in the doped material. Similarly, Li et al.⁴⁶ conducted capacitance vs temperature measurements on unintentionally doped (UID) β-Ga₂O₃ grown via EFG, with a doping concentration of 1 × 10¹⁷ cm⁻³, and observed a significant drop in capacitance in the temperature range from 40 to 20 K. Subsequent admittance spectroscopy analysis revealed a level at about 11 meV below E_c, potentially related to Si, which could be a source of unintentional doping.

The L-DLTS analysis is shown in Fig. 3, obtained using the “double” L-DLTS technique,⁵¹ in which two filling biases and a fixed reverse bias are utilized to probe a narrow window in the depletion region. This technique minimizes the influence of the electric field on the electron emission rate. For the case of the peak at 170 K (E_s), an analysis of the conventional DLTS spectra has been used to obtain ΔE_em and σ_app values, due to the proximity of the defect to the M–S interface complicating effective “double” L-DLTS analysis—shown in the supplementary material. Figure 3(a) shows the L-DLTS spectrum recorded at 270 K using parameters U_b = −6 V, U_p1 = −2 V, U_p2 = −3 V, and t_p = 1 ms. From this, the broad conventional-DLTS signal can be separated into two discrete electron emission signals with ΔE_em values of 0.46 eV (E9) and 0.52 eV (E1). These activation energies are close to those reported in the literature for the traps referred to as E9 and E1, respectively.⁴²  Figure 3(b) represents the L-DLTS spectrum measured at 440 K using U_b = −6 V, U_p1 = −3 V, U_p2 = −4 V, and t_p = 10 ms, which enables the broad E2 peak to be separated into its constituent components—E2a, E2b, and E3—as initially reported by Zimmermann et al.³⁹ 

The resulting Arrhenius plots are given in Fig. 4 for each of the six observed traps, with their corresponding properties summarized in Table I. The ΔE_em, σ_app, and N_T values for E_s are consistent with those reported by Perrier et al.,²⁶ relating to a sub-surface point defect that originates from polishing-induced scratches. More specifically, E_s was observed in their Si-doped β-Ga₂O₃ sample, which underwent a mechanical polishing process. This agrees with our observation, as E_s is only present in the “sub-surface” DLTS scan; the Ta-doped material studied in this work was also subjected to a chemical–mechanical polishing process. In another report, Polyakov et al.⁵² revealed that a trap with the activation energy for an electron emission of 0.28 eV has been observed following exposure to 18 MeV α-particles, implying a native defect that could have a similar position of an energy level as E_s within the bandgap. A more detailed comparison of the characteristics of the E_s trap and the irradiation-induced defect detected by Polyakov et al. is yet to be made.

For the deep level at 0.46 eV relative to the CBM, it is inconclusive as to whether this trap relates to the E9 trap with a similar activation energy of ∼0.4 eV reported in Ref. 42. E9 has been detected in β-Ga₂O₃ epilayers grown via metal–organic chemical vapor deposition (MOCVD), with evidence suggesting it is an extrinsic point defect that acts as a deep acceptor level.53–55 In our previous work,⁵⁵ we showed that the concentration of E9 increased with depth from the M–S junction in the UID and Si doped epilayers grown by MOCVD, implying it could originate from the melt-grown EFG material that the epilayers were grown upon. However, this trap is not commonly reported in DLTS studies of bulk β-Ga₂O₃ materials. Notably, Seyidov et al.⁵⁶ reported on a trap of 0.41 eV from the CBM in a material grown using the Cz method, which they also concluded is likely to be extrinsic in nature. In addition, Kruszewski et al.⁵⁷ observe a trap at E_c−0.46 in Cz-grown material, with a trap concentration of ∼2.0 × 10¹⁴ cm⁻³. It is important to note that Perrier et al.²⁶ did not observe the E9 center in Si-doped β-Ga₂O₃ grown using the OFZ method. This could imply that, in our β-Ga₂O₃:Ta material, the E9 electronic signature originates from a defect associated with the Ta source powder. Further investigation is required to determine whether the E_c−0.46 eV electron trap detected in our study relates to the E9 defect reported in the literature.

The E1 trap is commonly reported throughout Refs. 38, 47, 55, 56, and 58–63. Notably, Langørgen et al.⁶³ observed that annealing β-Ga₂O₃ in hydrogen increases the concentration of E1, implying it is linked to a hydrogen-related complex. Subsequent hybrid-functional calculations suggested its properties are consistent with singly hydrogenated divacancies (⁠$VGaibH−VO1$ and V_Ga1H − V_O1).⁶³ This suggestion is supported by Seyidov et al.,⁵⁶ who observed that E1 was present in their bulk Cz material after it had been subjected to a heat-treatment up to 650 K upon a DLTS scan. To explain this observation, Seyidov et al.⁵⁶ propose a rearrangement of the doubly hydrogenated divacancy (V_Ga12H−V_O2), which is assigned to the E4 defect, into the more stable $VGaibH−VO1$ complex during their DLTS measurements. Contrastingly, Polyakov et al.⁶⁰ argued E1 to be a deep donor level, with Si_Ga1-H or Sn_Ga2-H being likely candidates.

In bulk β-Ga₂O₃, E2 and E3 are commonly observed DLTS signals, and it is well established that they are related to deep acceptor and deep donor levels, respectively.⁴¹ Through a comparison between DLTS measurements and results from secondary ion mass spectrometry (SIMS), Ingebrigtsen et al.³⁷ showed that E2 is associated with Fe_Ga defects. Similarly to our findings, E2 was present in significant concentrations of 10¹⁶–10¹⁷ cm⁻³, making Fe a prominent compensating impurity in n-type β-Ga₂O₃. The ΔE_em values for E2a (0.69 eV) and E2b (0.75 eV) obtained using L-DLTS are consistent with the values reported by Zimmerman et al. (E2a = 0.66 eV; E2b = 0.73 eV),³⁹ in addition to the corresponding ΔE_em values predicted by theoretical calculations, which yielded E2a = 0.68 eV and E2b = 0.78 eV for Fe atoms at the tetrahedrally and octahedrally coordinated Ga sites, respectively.³¹ Using a similar analysis, E3 was linked to Ti via a combination of SIMS results with the results of DLTS measurements.³⁹ However, it is important to note that Seyidov et al.⁵⁶ only observe E3 in Cz grown β-Ga₂O₃ with Au and Pt Schottky contacts, whereas for Ni-based SBDs, E3 is only present after a thermal load up to 650 K. This observation extends the potential candidates for E3 to V_O-related defects, which could induce a DLTS signature overlapping with that of the Ti_Ga point defect. This suggestion by Seyidov et al.⁵⁶, that an intrinsic defect state is close to the Ti_Ga level, is supported by an observation by Polyakov et al.⁵², whereby an increase in the E3 concentration following irradiation with α-particles occurred—implying that E3 is sensitive to irradiation. Moreover, theoretical predictions based on DFT reveal that the transition level of V_OIII is ∼1.0 eV from the CBM, and it acts as a deep donor.⁶⁴ Therefore, it is difficult to distinguish the emission signals from Ti_Ga and V_OIII in DLTS.

The origin and nature of all these observed deep-level defects are reported extensively in reviews by Langørgen et al.,³¹ Polyakov et al.,⁴⁰ and Wang et al.⁴² 

From the DLTS and L-DLTS analysis, no peaks have been observed that can be directly related to Ta_Ga. This observation suggests that tantalum has a donor level that is too shallow (<E_c−0.1 eV) to be detected using DLTS measurements down to ∼40 K. Combining this result with the lack of significant carrier freeze-out down to 20 K and the increase in the free carrier concentration in Ta-doped samples compared to UID material, which is shown in Fig. 1(c), it can be argued that tantalum is an effective shallow donor impurity in β-Ga₂O₃.

To obtain further information about the defects in the investigated as-grown 0.05 mol. % β-Ga₂O₃:Ta (100) material, temperature-dependent (TD) PL measurements were conducted, as shown in Fig. 5. The PL measurements were performed under excitation using a 255 nm UV-LED, with a power density of 71 μW/cm². Figure 5 reveals that the PL signal intensity decreases gradually with increasing temperature but without any significant change in the peak energy. Onuma et al.³² reported similar behavior for Si-doped (100) β-Ga₂O₃, whereby cathodoluminescence (CL) analysis was applied that demonstrated the growth direction influences the peak position with respect to measurement temperature. This provides a possible explanation for the observed red shift of the peak in the TD PL conducted by Liu et al.¹⁵ for Ta-doped (010) material as well as in the temperature-resolved CL performed by Huynh et al.⁶⁵ on single crystals grown in the (−201) direction. The inset of Fig. 5 presents the normalized TD-PL spectra and shows that while the peak position remains nearly unchanged with temperature, the full width at half maximum (FWHM) broadens significantly, expanding from 0.7 eV at 15 K to 0.9 eV at 300 K.

Previous work has shown that the PL spectra of β-Ga₂O₃ are largely composed of four emission bands, which correspond to two ultraviolet bands (UV at around 350 nm and UV’ at around 400 nm), one blue band (BL at around 450 nm), and a green band (GL at around 500 nm).36,66 It has been reported by Shimamura et al.³⁶ that three emission bands are present at 2.5–2.8, 3.1, and 3.6 eV, which correspond to BL, UV, and UV’, respectively. However, there is a discrepancy in the literature on the number of UV bands, either one or two. Liu et al.¹⁵ previously analyzed Ta-doped material and proposed two UV bands, which are attributed to recombination involving self-trapped holes (STHs) on the trigonally coordinated oxygen lattice sites, O_Ⅰ and O_Ⅱ.15,32,66 Whereas Binet and Gourier³⁵ and Huynh et al.⁶⁵ reveal a single UV band that is also related to loosely bound electrons that form self-trapped excitons. Through deconvolution of the PL spectra measured at 15 K, our β-Ga₂O₃:Ta (100) material shows two dominant yet broad emission bands at 3.10 and 3.40 eV, as shown in Fig. 6. This suggests that two UV bands are present in our material, which could be related to recombination via STHs. Hybrid functional calculations by Frodason et al.⁶⁷ reveal that the $STHO1++eCBM−$ and $STHO2++eCBM−$ transitions have peak positions at 3.11 and 3.04 eV, respectively. These PPs are particularly close to our UV band, and Frodason et al. note that these PPs could even be underestimated. Other experimental work has also categorized UV-related emission in the range of 3.1–3.6 eV,32,33,36,66–72 which is consistent with our observations.

It could also be argued that the band at 3.10 eV in Fig. 6 is associated with BL, which is reported in the range of 2.5–3.0 eV,15,32,72 as opposed to UV. For instance, Huynh et al.⁶⁵ observed a peak at 3.05 eV, which they attributed to BL, based on CL measurements. This BL has been linked to the recombination of a trapped electron from a donor level with a trapped hole at an acceptor level, i.e., a donor–acceptor pair (DAP) transition,³² whereby the donor level is related to the doping impurities and the acceptor level is tentatively assigned to V_Ga²⁻ vacancies at the tetrahedral site,15,36 but some studies suggested that a divacancy (V_O − V_Ga) could also be a potential candidate.32,36 Work by Liu et al.¹⁵ on Ta-doped β-Ga₂O₃ identified two UV bands in addition to a BL band at 2.73 eV, which they assigned to a DAP transition involving Ta located at E_c−0.04 eV.

The low temperature PL spectrum in Fig. 6 shows two small shoulder peaks at 2.17 eV (GL) and 4.29 eV (UVB), with Gaussian areas approximately two orders of magnitude lower than those of the dominant UV and UV′ emission bands. The intensity of the green band in β-Ga₂O₃ crystals depends on the concentration of specific unintentionally incorporated impurities, such as Si, Be, Ge, Sn, Zr, and Li,34,36 and is correlated to the O partial pressure during growth in low purity (4N) undoped crystals.36,73–75 The low relative intensity of the impurity-dependent GL emission band suggests that the OFZ grown β-Ga₂O₃:Ta material contains low concentrations of the impurities responsible for the GL. Theoretical predictions by Frodason et al.⁶⁷ propose a wide range of impurities that could induce an emission band near the 2.17 eV peak that we observe, including Zn_Ga1 (2.15 eV), Mg_Ga2 (2.32 eV), Be_Ga1 (1.71 eV), Ca_Ga2 (2.40 eV), N_O1 (2.28 eV), N_O3 (1.87 eV), N_O1H (2.25 eV), and N_O3H (2.10 eV). In addition, numerous intrinsic-related states have predicted PL PPs close to that of our observed GL band, such as $VGaib$ (1.93 eV), $VGaib2H$ (2.12 eV), $VGaib3H$ (2.54 eV), $VGaic2H$ (1.95 eV), and $VGaicSii2H$ (2.55 eV). Further discussion about these defect states and their luminescent properties is provided in Ref. 67. The other shoulder peak, centered at 4.29 eV and labeled UVB, has a PL intensity about half that of the GL band, and its origin remains unclear. Thapa et al.⁷⁰ observed peaks at 4.08 and 4.6 eV, which are close to our UVB band, but the authors stated that these PL peaks are still under investigation. We have found no conclusive explanation for the emission mechanisms responsible for this peak, and further investigation is needed to elucidate the nature of the UVB band.

To study the broad PL spectrum in Fig. 6 further, the activation energies of quenching of the feature at 3.10 eV and the UV’ band at 3.40 eV were determined. The GL and UVB peaks were excluded from this analysis due to their relatively low intensities compared to the peaks at 3.10 and 3.40 eV. In the supplementary material, Fig. S2 shows the temperature dependence of the PL spectra on a semi-log scale, highlighting the thermal quenching of the GL and UVB.

The activation energy values (⁠$EA$ extracted for the 3.10 eV feature and the UV′ emission band are 20 ± 1 and 79 ± 4 meV, respectively, with corresponding R-squared (R²) values of 0.989 and 0.995. It can be observed that the fitting for both UV-related emission bands closely follows the integrated PL intensity. This observation is consistent with the R² values, which are both close to 1. For the peak at 3.10 eV, the E_A value aligns with previous studies that have calculated activation energy values for donor states in β-Ga₂O₃ within the range of 0.02–0.03 eV.⁷⁷ Onuma et al.⁷² propose a mechanism for the UV band observed in heavily Si-doped β-Ga₂O₃, involving a DAP transition, where Si_Ga serves as the donor state in the range E_c−16–36 meV. A similar DAP mechanism has been reported for the BL, involving acceptors such as V_Ga, V_O − V_Ga, Mg_Ga, or N_O.15,36,72 For our β-Ga₂O₃:Ta material, it remains inconclusive as to whether the E_A value of 20 meV corresponds to a DAP involving Si,⁷² Ta,¹⁵ or potentially another mechanism altogether. For the UV’ band at 3.40 eV [Fig. 7(b)], the fitting calculation yields an activation energy E_A of 79 ± 4 meV, which is comparable to the ∼72 meV value reported by Lukman and Bergman⁷¹ for a PL peak at ∼3.5 eV. They proposed a non-radiative mechanism involving STHs and phonons to explain the thermal-quenching of the PL signal. This conclusion was supported by correlating phonon energies determined via Raman scattering analysis with their PL E_A values.

We have studied electrically active defects in Ta-doped (100) β-Ga₂O₃ single crystals grown using the optical floating zone method. From C–V and CT analysis, a relatively uniform carrier concentration in the range (1.0–1.2) × 10¹⁸ cm⁻³ was derived, with no sign of significant carrier freeze out down to 20 K. These findings confirm the theoretical predictions that Ta can be an effective n-type dopant in β-Ga₂O₃. Both conventional- and L-DLTS were applied to Ni-based SBDs to extract the electronic properties of the six observed DLTS defect signals, referred to as E_s, E9, E1, E2a, E2b, and E3. For the E_s, E9, and E1 traps, their concentration decreases significantly when probing further away from the sample surface. The concentration of the Fe-related E2 center is of the order of 10¹⁷ cm⁻³ in the bulk, which makes it an important compensating defect in this material. The PL spectra reveal two dominant emission bands centered at 3.10 and 3.40 eV at 15 K, with extracted activation energy values of 20 ± 1 and 79 ± 4 meV, respectively, for temperature-induced PL quenching.

See the supplementary material for Figs. S1 and S2. To obtain the data in the Arrhenius plot (Fig. 4) for the E_s trap, a series of conventional-DLTS scans were taken with varying rate windows. The proximity of the E_s trap to the surface prevented effective “double” L-DLTS from being applied. These conventional-DLTS spectra are presented in Fig. S1. The temperature-dependence of the PL intensity (Fig. 5) is also presented on a semi-log scale in Fig. S2. This highlights the thermal quenching of the GL (2.17 eV) and UVB (4.29 eV) bands.

The authors in Manchester would like to acknowledge the support by the EPSRC-UK under Contract Nos. EP/T025131/1 and EP/S024441/1. Christopher A. Dawe would also like to acknowledge the support from the EPSRC CDT in compound semiconductor manufacturing. We acknowledge the partial financial support from the UKRI Innovation and Knowledge Center (IKC) REWIRE under Grant No. EP/Z531091. M. Kuball acknowledges the financial support from the Royal Academy of Engineering through the Chair in Emerging Technologies Scheme. We would like to greatly acknowledge the Commonwealth Scholarship Commission (CSC) for funding this research under the Commonwealth Split-Site Ph.D. scholarship (2023–2024) under Grant No. INCN-2023-373 and the SJSGC fellowship by the University Grant Commission, Ministry of Education, Government of India for V. L. Ananthu Vijayan under Grant No. UGCES-22-OB-KER-F-SJSGC-7611.

The authors have no conflicts to disclose.

Christopher Dawe: Conceptualization (lead); Data curation (lead); Formal analysis (lead); Investigation (lead); Methodology (equal); Project administration (equal); Visualization (equal); Writing – original draft (lead); Writing – review & editing (lead). Lijie Sun: Conceptualization (equal); Data curation (equal); Formal analysis (equal); Investigation (equal); Methodology (equal); Visualization (equal); Writing – original draft (equal); Writing – review & editing (equal). V. L. Ananthu Vijayan: Conceptualization (equal); Formal analysis (supporting); Investigation (supporting); Methodology (equal); Resources (equal); Visualization (supporting); Writing – review & editing (supporting). Vladimir Markevich: Data curation (supporting); Formal analysis (supporting); Investigation (supporting); Methodology (supporting); Supervision (equal); Validation (equal); Visualization (supporting); Writing – review & editing (equal). Janet Jacobs: Resources (equal); Supervision (equal). Ian Hawkins: Resources (equal); Supervision (equal). Matthew Halsall: Conceptualization (supporting); Formal analysis (supporting); Funding acquisition (equal); Investigation (supporting); Methodology (supporting); Project administration (equal); Resources (equal); Supervision (equal); Validation (equal); Visualization (supporting); Writing – review & editing (supporting). Anthony Peaker: Formal analysis (supporting); Funding acquisition (equal); Investigation (supporting); Resources (equal); Supervision (equal); Validation (equal); Visualization (supporting); Writing – review & editing (supporting). David Binks: Formal analysis (supporting); Investigation (supporting); Supervision (equal); Validation (equal); Visualization (supporting); Writing – review & editing (supporting). Sai Charan Vanjari: Investigation (supporting); Supervision (supporting); Validation (equal); Writing – review & editing (supporting). Sridharan Moorthy Babu: Funding acquisition (equal); Project administration (equal); Resources (equal); Supervision (equal); Validation (equal); Writing – review & editing (supporting). Martin Kuball: Conceptualization (equal); Data curation (supporting); Formal analysis (supporting); Funding acquisition (lead); Investigation (supporting); Project administration (equal); Resources (equal); Supervision (equal); Validation (equal); Visualization (equal); Writing – review & editing (equal).

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