Low $114cm−3$ free carrier concentration in epitaxial $β$-Ga₂O₃ grown by MOCVD

Monoclinic β-Ga₂O₃ has recently captivated extensive interest as a promising ultra-wide bandgap semiconductor for power electronics applications owing to its attractive material properties, including a large bandgap of ∼4.9 eV,¹ a high critical breakdown field of ∼6–8 MV/cm,^2,3 and a substantially large Baliga’s figure of merit (BFOM) of ∼3400.² A prevalent advantage of β-Ga₂O₃ in the realm of power devices is the availability of free-standing high quality β-Ga₂O₃ substrates grown directly from the melt,^4–7 making it economically viable as compared to the present technologies based on GaN and SiC. Controllable n-type doping of Ga₂O₃ has been achieved using shallow donors, including Si,⁸ Sn,⁹ and Ge¹⁰ with conductivity varying over many orders of magnitude. This provides a superior platform for power switching applications using β-Ga₂O₃.

Many promising β-Ga₂O₃ based power device architectures have been reported and include MESFETs,² SBDs,^11–15 and MOSFETs.^16–18 Among these devices, vertical power device structures with thick lightly doped drift regions are desired to fully exploit the high Baliga’s figure of merit (BFOM $=ϵμEc3$⁠, where ϵ is the dielectric constant, E_c is the critical breakdown field, and μ is the carrier mobility).¹⁹ High breakdown voltage and low on-resistance are required to minimize the conduction loss in the vertical power devices, which necessitates low net charge concentration and high electron mobility in the drift region of the material. For example, the state-of-the-art breakdown voltage of 21 kV for SiC BJTs was demonstrated by using a 186 µm thick lightly N-doped SiC layer with the net charge concentration of ∼2 × 10¹⁴ cm⁻³.²⁰ Comparably, in β-Ga₂O₃, even though the formation of a PN junction has not yet been realized due to the lack of p-type doping,²¹ the growth of thick epitaxial film with low net charge density will facilitate the advancement of unipolar power devices.

Numerous researchers have reported on the homoepitaxy of β−Ga₂O₃ on various substrate orientations by using molecular beam epitaxy (MBE),^3,22 metalorganic chemical vapor deposition (MOCVD),^23–28 and halide vapor phase epitaxy (HVPE)^29,30 growth methods. With growth dominantly on (001)-oriented β-Ga₂O₃ substrates, the HVPE method has shown films with net background charge concentration as low as ∼10¹³ cm⁻³ based on CV measurements.³⁰ However, the high growth rate associated with the technique results in films with rough surface morphology, rendering the film useless for device fabrication, unless chemical–mechanical polishing (CMP) of the epi-surface is used.¹¹ This makes it challenging to realize sharp doping profiles and abrupt heterointerfaces. MBE of β-Ga₂O₃ is currently limited by the slow growth rate and difficulties in achieving <1 × 10¹⁶ cm⁻³ net background concentration.^3,10 MOCVD, however, has the advantage of growing the epitaxial films at a high growth rate (∼10 µm/h) with sub-nanometer surface roughness.²⁴ The latest results in the growth of β-Ga₂O₃ films using MOCVD showed its viability for the growth of device quality epitaxial films with RT electron mobilities between ∼150 cm²/V s and 184 cm²/V s and net background carrier concentrations between ∼3 × 10¹⁵ cm⁻³ and ∼3 × 10¹⁶ cm⁻³.^25,27,31,32 Recently, a high quality MOCVD grown unintentionally doped (UID) β-Ga₂O₃ film with a RT bulk mobility of 176 cm²/V s and one of the lowest background carrier concentrations of ∼7 × 10¹⁵ cm⁻³ was demonstrated.²⁵ Further optimization of the MOCVD growth settings has produced an epitaxial film with a record high electron mobility >11 000 cm²/V s at 54 K, but the RT background concentration was ∼5.4 × 10¹⁵ cm⁻³.^33,34 These results are based on films grown using pure oxygen sources. In this study, we explore the use of nitrous oxide (N₂O) as an oxygen source for the growth of β-Ga₂O₃ thin films. The purity of the N₂O grown β-Ga₂O₃ films is found to be strongly dependent on process conditions, including the N₂O flow rate, growth pressure, and temperature.³⁵ Using optimized growth conditions, β-Ga₂O₃ films with high RT electron mobility and record low net background carrier concentrations were demonstrated.

β-Ga₂O₃ films were grown on Fe-doped (010) β−Ga₂O₃ substrates (Novel Crystal Technology) using Agnitron Technology’s Agilis R&D MOCVD system.^12,24,25,34 Triethylgallium (TEGa) was used as a precursor for Ga, and N₂O or oxygen (5N) was used for oxidation. Ar (6N) gas was used to carry the TEGa vapor into the reactor. Point of use purifiers were used for Ar, O₂, and N₂O gases to reduce the impurity to below ppm level. HRXRD, AFM, and SEM techniques were used to characterize the crystal and surface qualities of the films. SIMS (Evans Analytical Group) measurements were used to monitor the concentration of N, H, and C in the films. The electrical properties of the films were analyzed using mercury probe CV and temperature-dependent Hall measurement methods.

The growth of Ga₂O₃ using N₂O requires entirely different growth conditions than typical growth settings used with pure oxygen.^24,25 Our initial attempts to use optimal process conditions established to grow high quality Ga₂O₃ using pure oxygen by merely substituting N₂O for oxygen did not lead to any growth. The growth of Ga₂O₃ using N₂O, generally, requires high growth pressure (>75 Torr) and temperature (>700 °C).^36–38 Such requirements for N₂O were desired due to its kinetically inert characteristics at low pressure and temperature.³⁹ The details of the exact process conditions contain proprietary information and will not be fully disclosed. Rather, we will discuss the results by providing a range of growth pressures and temperatures.

The N₂O grown films were compared to a known pure oxygen grown film to determine their quality. Figure 1 compares the surface morphology and ω-2θ HRXRD scan of ∼1.0 µm thick epitaxial β-Ga₂O₃ grown films using pure oxygen and N₂O gases. For the N₂O grown film, a substrate temperature of T₁, a chamber pressure of P₁, and a N₂O/TEGa ratio of 4300 were used (Table I). The 2D AFM images for the oxygen and N₂O grown layers are shown in Figs. 1(a) and 1(b), respectively. Both the oxygen and N₂O grown layers showed smooth surface morphology with an rms roughness of ∼0.8 nm, as determined from a 5 µm × 5 µm scan area of the images. This rms value of the layers is comparable to the best values available in the literature for β−Ga₂O₃ layers grown by MBE⁴⁰ or MOCVD.^25,26 In Fig. 1(c), the HRXRD (020) peaks of the epilayers are compared. The XRD full width at half maximum (FWHM) of the (020) peak for the N₂O grown film is ∼41 arc sec, which is comparable to bulk substrates.⁴¹ Given the small thickness the epilayers, the extracted XRD linewidth may be affected by the contribution from the substrates, but the XRD data presented here show the similarity in the quality of the Ga₂O₃ films grown using N₂O and pure oxygen for the same film thickness. The similarity observed in the surface and structural quality for the films grown using oxygen and N₂O gases shows that the latter can be used as an alternative oxygen source to grow epitaxial Ga₂O₃ films with the same quality as the films grown using the pure oxygen source.^25,26

The N₂O grown Ga₂O₃ thin film showed an extremely high resistivity compared to the pure oxygen grown films. To measure the Hall mobility, free carrier concentration, and I–V characteristics of the N₂O grown film, Ti (∼20 nm)/Au (∼130 nm) ohmic contacts were sputter deposited on the film using the Van der Pauw geometry. The deposited contacts underwent a 60 s rapid thermal annealing at 470 °C in an N₂ atmosphere to reduce the contact resistance.⁴² The I–V characteristics and Hall measurements of the film were performed using an Ecopia (HMS-3000) Hall measurement system. The N₂O grown films were very resistive with >10⁵ $Ω$-cm resistivities (or sheet resistance of >10⁹ $Ω$/sq), making the RT Hall mobility measurements on the films unsuccessful.

This high resistivity N₂O grown layer was implemented as a buffer layer to reduce the background carrier concentration for pure oxygen grown films. For this, two UID Ga₂O₃ epitaxial films were grown using pure oxygen on (010) oriented Ga₂O₃ substrates with [sample I, Fig. 2(a)] and without [sample II, Fig. 2(a)] a N₂O grown Ga₂O₃ buffer layer. The pure oxygen grown layers are ∼3.0 µm thick and were deposited using optimal growth conditions, which already led to a net background concentration as low as mid-10¹⁵ cm⁻³.¹² The N₂O grown Ga₂O₃ buffer layer was ∼120 nm thick. The background net charge density (N_d–N_a) of the films was measured using a mercury probe CV system with a probe diameter of 855 µm. Figures 2(b) and 2(c) show the capacitance (C) vs bias voltage and the charge distribution (N_d–N_a) depth profiles, respectively, for UID layers (samples I and II) measured at an excitation frequency of 100 kHz. The average N_d–N_a values for samples I and II are ∼6.0 × 10¹⁵ cm⁻³ and 2.0 × 10¹⁵ cm⁻³, which are extracted assuming a dielectric constant of 10.⁴³ Evidently, the use of the N₂O grown buffer layer helped to reduce the background carrier concentration in the pure oxygen grown layer by as much as three times. This suggests that transitioning from N₂O to pure O₂ growth in the same MOCVD run retained high crystal quality and helped lower the background charge concentration.

The observed high resistivity in the N₂O grown Ga₂O₃ layers and the drop in the background carrier concentration of the oxygen grown Ga₂O₃ layer resulting from the use of a N₂O grown buffer layer could be attributed to the incorporation of nitrogen into the films from the N₂O source (vide infra). First-principles studies showed that N incorporates into the sites of both Ga and O, but the substitution of N on an O site acts as a compensating center with a deep energy level at >1.3 eV above the valance band maximum.^44–46 The deep acceptor nature of N was also demonstrated by ion implanting it into a Ga₂O₃ substrate.⁴⁷ In the N₂O grown Ga₂O₃ films in this work, N could be uniformly incorporated into the film, causing donor compensation which leads to the observed high resistivity in the films.⁴⁸ 

One common challenge recognized in many epitaxial growth methods used for homoepitaxial Ga₂O₃, regardless of process conditions, is the accumulation of Si at the film/substrate interface with SIMS measured concentration more than 1 × 10¹⁹ cm⁻³.^26,49 This high and unintentionally localized interface Si acts as an active donor, which adversely affects the quality of the epitaxial film (e.g., unable to achieve a low background concentration in the film) and subsequently the device performance.³⁵ For example, in field effect transistors (FETs), it creates a parasitic channel at the film/substrate interface, preventing the transistor from pinching off.

Some researchers ascribed this interface Si to ambient contaminants adsorbed on the substrate surface before the film growth, which is then, in the initial growth stage, incorporated into the growing layer.^16,50 However, our recent attempts to manage the interface Si using buffered hydrofluoric acid (BHF)-based substrate surface treatment indicate that the origin of the interface Si could be largely from the substrate polishing process not from the pre-growth adsorbed contaminants. A brief (∼10 s) etching of the Ga₂O₃ substrate in BHF (2%) led to the reduction of the interface Si by more than one order of magnitude as confirmed by SIMS,³² suggesting that the BHF solution etches off a pre-existing SiO_x on the substrate surface. While increasing the etch time or using concentrated HF may help to completely remove the Si from the substrate surface, an alternative method to manage the interface Si is to compensate it by growing thin buffer layers that contain deep acceptor dopants, such as Mg, Fe, and N.^47,51

To study whether N incorporates into the N₂O grown layers, four samples with various multilayer structures were grown at different growth conditions and SIMS measurements were conducted on them. Figure 3 shows the schematics of a multilayered Ga₂O₃ thin film (SIMS stack) grown using N₂O. The N₂O grown layers are separated using the pure oxygen grown layers. In Fig. 3(a), which describes the SIMS stack of sample No. 1 in Table I, the N₂O grown layers (i.e., layers A, B, C, and D) are grown by varying the growth pressure and N₂O/TEGa ratio, but at a constant substrate temperature of T₁. Layers A and B, and C and D were grown at a chamber pressure of P₁ and P₂ (P₁ > P₂ > 75 Torr), respectively. The N₂O/TEGa ratio for A/C and B/D were 4300 and 2150, respectively. The same SIMS stack schematically shown in Fig. 3(b) was used for sample sets, No. 2, No. 3, and No. 4 in Table I, grown at constant chamber pressure of P₁, but at various substrate temperatures of T₂, T₃, and T₄ (700 °C < T₁ < T₂ < T₃ < T₄ < 1100 °C) and different N₂O/TEGa ratios. The layers E, F, and G were grown at N₂O/TEGa ratios of 4300, 2150, and 1290, respectively, for each of the three samples. Details of the SIMS samples and growth conditions for the individual layers are shown in Table I. The UID β-Ga₂O₃ separating the N₂O grown layers is grown using optimized process conditions which led to a RT Hall mobility of 176 cm²/V s.²⁵ 

Figures 4(a)–4(d) show the SIMS depth profiles of N, H, and C for the N₂O grown β-Ga₂O₃ SIMS stack samples, No. 1, No. 2, No. 3, and No. 4, respectively. Contrary to the pure oxygen grown UID β-Ga₂O₃ films, the films grown using N₂O as an oxygen source showed the incorporation of nitrogen and hydrogen into the β-Ga₂O₃ films. A clear contrast in both N and H incorporations are observed from the SIMS profiles. As shown in Fig. 4(a) (sample No. 1), high concentrations of N and H were observed in the N₂O-grown films (i.e., A, B, C, and D layers), while they tailed off during the oxygen-grown interlayers.

The concentration of both N and H impurities was found to be strongly dependent on the MOCVD process settings used during the growth of the film. The highest N incorporation (∼1.9 × 10¹⁹ cm⁻³) is achieved for layer B, which is grown at a higher growth pressure of P₁ and a lower N₂O/TEGa ratio of 2150. With the decrease in growth pressure from P₁ to P₂, but using a constant substrate temperature of T₁, the incorporation of N into the N₂O grown layer showed slight reduction (layers C and D). The N concentration in the UID Ga₂O₃ layers between the N₂O grown layers is close to the detection limit (detection limit [N] ∼8 × 10¹⁶ cm⁻³) of the SIMS instrument, which indicates that there is very little to no diffusion of N out of the N₂O grown layers. This result is consistent with lower thermal diffusivity of N (than Mg) observed during the thermal annealing process in an N implanted Ga₂O₃ substrate.⁴⁷ 

The N doping, however, is accompanied by the incorporation of H, which virtually follows a similar profile as N (see Fig. 4). The reason for the incorporation of H into the N₂O grown films is unclear, but it appears to be exclusively due to the N₂O oxidant since the layers grown using pure oxygen showed no detectable H (detection limit [H] ∼2 × 10¹⁷ cm⁻³) in the film. This is clearly shown in the SIMS traces [Fig. 4(a)], at a depth >2 µm, where the N, H, and C profiles extended ∼0.5 µm into the oxygen grown ∼1.8 µm thick UID β-Ga₂O₃ layer embedded between layer A and the substrate [see Fig. 3(a)]. The use of N₂O may create a volatile chemical byproduct that promotes the incorporation of H into the growing films, but the exact cause requires a detailed study on the chemical interaction routes of the metalorganic source and the nitrous oxide gas, which is beyond the scope of this study. Interestingly, there is no measurable incorporation of C into the Ga₂O₃ layers, and they remain to be unaffected both by the process conditions and by the identity of the oxygen source used in the growth of the layers.

It is known that H in Ga₂O₃ acts as a shallow donor.^52,53 Therefore, its incorporation along with N into the N₂O grown film is undesirable as the two impurities potentially compensate each other, but the effect may not be significant since N showed dominant incorporation. In addition, potential formation of hydrogenated gallium vacancies (V_Ga-nH, n = 1, 2, 3)^54–56 could lead to electrical passivation of the incorporated H as well. This is likely the case since the formation of gallium vacancies in an MOCVD growth of Ga₂O₃ is efficient.⁵⁷ Consequently, high resistivity has been observed for the films grown under the growth conditions used in sample No. 1. However, future systematic study is necessary to understand the behavior of the simultaneous incorporation of N and H impurities.

Using optimal MOCVD process conditions, N₂O grown films without N and H can be realized. Figures 4(b)–4(d) presents the SIMS depth profiles for N, H, and C of sample No. 2, No. 3, and No. 4 grown at a constant chamber pressure of P₁ but at variable substrate temperatures T₂, T₃, and T₄ (700 °C < T₁ < T₂ < T₃ < T₄), respectively. The N₂O/TEGa ratios used for the growth of layers E, F, and G in each sample were 4300, 2150, and 1290, respectively. The concentration of N and H in the films showed a consistent decrease with the increase in the growth temperature. When the “best” growth conditions are met, films with no detectable N and H are obtained, as shown by SIMS in Fig. 4(d). This was achieved for the growth condition used for sample No. 4 (Table I), which showed undetectable N and H concentrations regardless of the N₂O/TEGa ratio. The uptake of N and H for layer E for all the samples is likely an interface issue.

The dependence of N and H incorporation on the process conditions, including chamber pressure, substrate temperature, and N₂O/TEGa ratio (fixed TEGa molar flow) for the four samples is summarized in Figs. 5(a) and 5(b), respectively. The incorporation of both N and H impurities into the films is inversely proportional to the N₂O flow rate. When the N₂O/TEGa ratio decreased by half from 4300 (layer A) to 2150 (layer B) at constant P₁ and T₁, the N incorporation increased by 1.3 times, while at the same time, the H incorporation was reduced by 3.2 times [see Fig. 4(a)]. Similarly, in Fig. 4(b), the increase in N and H is observed when the N₂O/TEGa ratio decreases. Comparing the layers grown at P₁ and P₂ (P₁ > P₂) for sample No. 1, but constant substrate temperature and N₂O/TEGa ratio, higher N and H incorporation was observed at the higher pressure, P₁. For a fixed pressure and N₂O/TEGa ratio, the incorporation of the two impurities decreases with the increase in substrate temperature. Therefore, one can effectively grow an unintentionally doped Ga₂O₃ film by using a properly tuned growth temperature, chamber pressure, and N₂O flow rates, which leads to the complete removal of both N and H from the grown films.

A ∼3.2 µm thick β-Ga₂O₃ thin film was grown using N₂O by using the process conditions used for the growth of layer F in sample No. 4. This optimal condition suppressed the N and H incorporation into the film below the detectable limit as observed from the SIMS data in Fig. 4(d), effectively growing UID β-Ga₂O₃. The surface morphologies of the film were characterized by SEM and AFM. Figures 6(a) and 6(b) represent the 2D AFM and field emission scanning electron microscopy (FESEM) images of the sample. The film shows a featureless surface with a 2.9 nm rms roughness. HRXRD measurements for the on-axis (020) plane reflection of the epitaxial films indicate a narrow full width at half maximum (FWHMs) of 43.4 arc sec, which is comparable with the FWHM values measured for the bulk substrates, suggesting high structural quality in the N₂O grown β-Ga₂O₃ by MOCVD.

Hall measurement on the film was conducted using the Van der Pauw configuration. To obtain ohmic contact with the N₂O grown β-Ga₂O₃ film, regrowth of a heavily doped β-Ga₂O₃ contact layer was deposited by plasma-assisted molecular beam epitaxy (PAMBE). Sn (with the doping concentration of ∼7 × 10¹⁹ cm⁻³ as measured by SIMS) was used as an n-type dopant for contact regrowth. The high doping concentration prevents carrier freeze out in the regrown contact layer and ensures ohmic contact in a wide temperature range, making it feasible to obtain reliable Hall measurements on the N₂O grown β-Ga₂O₃ film over a wide temperature range, including the cryogenic temperatures. Details of the regrowth contact fabrication can be found in Ref. 25.

The temperature-dependent Hall mobility and charge density of the film are presented in Figs. 7(a) and 7(b), respectively. The Hall mobility increased from 153 cm²/V s at 300 K to 550 cm²/V s at 100 K, below which the ohmic contact failed due to low charge concentration in the Ga₂O₃ film. The obtained free carrier concentration at room temperature is 2.4 × 10¹⁴ cm⁻³, which sets a record for β-Ga₂O₃ epitaxial films grown by MOCVD. The recently reported pure oxygen grown unintentionally doped β-Ga₂O₃ has the lowest net background concentration of 3 × 10¹⁵ cm⁻³ with RT electron mobility of ∼150 cm²/V s.³² Mobility and charge density calculations were performed to fit the temperature-dependent Hall mobility and charge density, but the reliability of the fit was compromised due to the narrow data range, resulting from the failed ohmic contact below 100 K. However, the slow increase in the Hall mobility with temperature suggests that impurity scattering in addition to phonon scattering could be limiting the electron mobility. This scattering is likely due to high compensation levels related to both intrinsic and extrinsic acceptors, such as Ga vacancies⁵⁷ and N incorporation into the film (although undetectable by SIMS). At room temperature, however, the electron mobility in β-Ga₂O₃ is dominantly limited by the polar optical phonon scattering mechanism.⁵⁸ Thus, the effect of the N impurity is negligible, resulting in RT electron mobility comparable to oxygen grown films with low background concentration. Further optimization of the process conditions to reduce the compensation charge density is critical for improving the electron mobility in drift layers. A recent result presented in the 2019 IWGO³³ with a record mobility of 11 200 cm²/V s at 50 K for an MOCVD grown high purity β-Ga₂O₃ is attributed to the low acceptor compensation density in the film, which was only ∼10% compensation. The repeatability of the growth condition leading to the obtained low free carrier density was confirmed by growing three different films. The SIMS measurement for N and H impurities in each sample showed concentrations below the instrument detection limit (data not shown), and the resistivity was comparable to the film under study. However, transport measurement was not conducted due to the time-consuming nature of the contact fabrication process used in this work (i.e., regrowth method).

Figure 8 compares the RT Hall mobility vs carrier concentration of the N₂O grown β-Ga₂O₃ film alongside the previously reported pure oxygen grown UID β-Ga₂O₃ films by MOCVD. As shown in this figure, pure oxygen grown UID Ga₂O₃ epitaxial films with RT electron mobilities ranging between ∼150 cm²/V s and 176 cm²/V s have routinely been demonstrated with n-type background carrier concentration between 3 × 10¹⁵ cm⁻³ and 3 × 10¹⁶ cm⁻³.^25,32,59,60 By using N₂O as an oxygen source, however, the free carrier concentration was more than one order of magnitude lower than the lowest carrier concentration attained by using the pure oxygen source, with no effect on the RT electron mobility.

In summary, β-Ga₂O₃ epitaxial films were grown by MOCVD using N₂O as an oxygen source. N and H were found to be incorporated into the N₂O grown films, but their incorporation efficiency was strongly dependent on the process conditions, including the substrate temperature, pressure, and N₂O flow rate. Optimal process conditions allowing the incorporation of the two impurities from the highest possible concentration to their complete removal from the films were identified by using SIMS measurements. At high substrate temperature, high pressure, and high N₂O flow rate, complete removal of N and H from the N₂O grown films was realized. The film grown without N and H incorporation showed a record low room temperature carrier concentration value of ∼2 × 10¹⁴ cm⁻³ with an electron mobility of 153 cm²/V s. A thin β-Ga₂O₃ buffer layer grown using N₂O was also found to reduce the net background charge concentration in a pure oxygen grown film and is attributed to the compensation of Si accumulated at the film/substrate interface by N. The results show that the use of the N₂O oxidant can lead to low background concentration and high electron mobility films that are required for the realization of high-performance power electronic devices.

The work at Agnitron Technology was supported by the Air Force Office of Scientific Research (AFOSR, Program Manager Dr. Ali Sayir) and the Office of Naval Research (ONR, Program Manager Mr. Lynn Petersen) through Grant Nos. FA9550-19-C-0030 and N6833518C0192, respectively. Support for UCSB was provided by AFOSR through Program No. FA9550-18-1-0059 and GAME MURI Award No. FA9550-18-1-0479, monitored by the project manager Dr. Ali Sayir. Additional support from UCSB was provided through a subcontract from Agnitron Technology through ONR Program No. N00014-16-P-2058 and from DTRA through Program No. HDTRA 11710034.