SiC substrate moccvd gallium oxide epitaxial wafer

Abstract

This invention discloses a SiC substrate MOCVD gallium oxide epitaxial wafer, belonging to the field of wide-bandgap semiconductor epitaxial material preparation technology. The epitaxial wafer comprises, from bottom to top, a silicon carbide substrate, a β-Ga2O3 carbon getter layer, and a β-Ga2O3 drift layer. The carbon concentration ratio of the β-Ga2O3 carbon getter layer to the β-Ga2O3 drift layer is not less than 100, forming a decreasing carbon concentration gradient. The β-Ga2O3 drift layer contains periodically distributed annealing repair interfaces. The preparation method employs a precursor spatiotemporal separation alternating gas supply mode to grow the β-Ga2O3 drift layer, with periodic insertion of in-situ flash annealing steps. A three-stage carbon purification mechanism synergistically removes carbon from three stages: carbon source generation, defect complex formation, and cumulative stabilization, solving the technical bottleneck of uncontrollable carrier compensation ratio caused by organometallic precursor carbon contamination during MOCVD growth of ultra-thick drift layers.

Description

Technical Field

[0001] This invention relates to the field of wide bandgap semiconductor epitaxial material preparation technology, specifically to a SiC substrate MOCVD gallium oxide epitaxial wafer. Background Technology

[0002] β-Ga₂O₃ is an ultra-wide bandgap semiconductor material with a bandgap of 4.8 eV and a theoretical breakdown electric field strength exceeding 8 MV / cm. Its Baliga quality factor is approximately 3000 times that of silicon, demonstrating its potential to replace silicon carbide and gallium nitride in high-voltage power electronic devices. However, β-Ga₂O₃ has a thermal conductivity of only 10.9–27 W / m·K, far lower than silicon carbide (490 W / m·K) and gallium nitride (130 W / m·K), resulting in severe self-heating under high-power operating conditions. Epitaxially growing β-Ga₂O₃ on a silicon carbide substrate, leveraging the high thermal conductivity of silicon carbide to address heat dissipation, is a recognized core technology approach in academia and industry. Among various epitaxial growth techniques, metal-organic chemical vapor deposition (MOCVD) is considered the most promising method for mass production of β-Ga₂O₃ due to its wide process window, good scalability, and flexible precursor selection.

[0003] However, MOCVD growth of β-Ga₂O₃ faces an inherent technical contradiction: MOCVD requires the use of organometallic compounds (such as trimethylgallium™Ga or triethylgallium TEGa) as gallium precursors. These precursors inevitably release carbon-containing groups (methyl or ethyl) during high-temperature pyrolysis, some of which are incorporated into the β-Ga₂O₃ lattice, occupying oxygen sites and forming carbon substitutional oxygen defects (C₂O). The C₂O combines with gallium vacancies (V₂Ga) generated during growth within a timescale of minutes to form thermodynamically stable C₂O-V₂Ga deep-level compensation defect complexes with binding energies as high as 2.0 eV. These complexes act as deep-level electron traps, with each complex capable of capturing one free electron. In high-voltage power device drift layers requiring thicknesses of 10–20 μm, the long-term accumulation and exposure of carbon impurities allows for C₂O-V₂Ga complex densities reaching up to 10⁻⁶. 16 cm -3 The order of magnitude is higher than that of the target n-type doping concentration, which is only 10. 15 cm -3 The resulting loss of control over the carrier compensation ratio (which can exceed 10) leads to a free electron concentration in the epitaxial layer that is far below the design value. The carrier mobility is also significantly reduced due to the trap scattering effect, becoming the core technical bottleneck for the expansion of β-Ga2O3 to high-voltage power devices of 1200V and above. Summary of the Invention

[0004] Technical Objective: To address the problem that carbon impurities introduced by the decomposition of organometallic precursors coupled with gallium vacancies to form stable deep-level compensation defect complexes during the growth of ultra-thick drift layers in β-Ga2O3 / SiC heteroepitaxial wafers via MOCVD, leading to uncontrollable carrier compensation ratios in the drift layer, this invention discloses a gallium oxide epitaxial wafer grown on a SiC substrate via MOCVD.

[0005] Technical solution: To achieve the above technical objectives, the present invention adopts the following technical solution:

[0006] This invention provides a SiC substrate MOCVD gallium oxide epitaxial wafer, comprising, from bottom to top:

[0007] silicon carbide substrate;

[0008] A nucleation buffer layer is disposed above a silicon carbide substrate;

[0009] A β-Ga2O3 carbon getter layer is disposed above the nucleation buffer layer;

[0010] An undoped transition layer is disposed above the β-Ga2O3 carbon getter layer;

[0011] A β-Ga2O3 drift layer is disposed above the undoped transition layer;

[0012] The carbon concentration ratio of the β-Ga2O3 carbon getter layer to the β-Ga2O3 drift layer is not less than 100, and a decreasing carbon concentration gradient is formed from the β-Ga2O3 drift layer to the β-Ga2O3 carbon getter layer. The β-Ga2O3 drift layer contains multiple periodically distributed annealing repair interfaces along its thickness direction, and the drift layer sublayer is located between adjacent annealing repair interfaces. The deep-level trap concentration at the annealing repair interface, as determined by deep-level transient spectroscopy, is lower than the deep-level trap concentration within the drift layer sublayer.

[0013] Furthermore, the thickness of the β-Ga2O3 nucleation buffer layer is 50~100nm; the thickness of the β-Ga2O3 undoped transition layer is 100~200nm; and the carbon concentration of the β-Ga2O3 undoped transition layer is between the carbon concentration of the β-Ga2O3 carbon getter layer and the carbon concentration of the β-Ga2O3 drift layer.

[0014] Furthermore, the thickness of the β-Ga2O3 carbon getter layer is 20~50 nm, and the carbon concentration is 1×10⁻⁶. 18 ~1×10 19 cm -3 The carbon concentration of the β-Ga2O3 drift layer is no greater than 5 × 10⁻⁶. 15 cm -3 .

[0015] Furthermore, the thickness of the β-Ga2O3 drift layer is 10~20 μm, and the n-type doping concentration is 5×10⁻⁶. 14 ~5×10 16 cm -3 The thickness of the sublayer between adjacent annealing repair interfaces is 200~500nm.

[0016] This invention also provides a method for preparing gallium oxide epitaxial wafers via MOCVD on a SiC substrate, specifically including the following steps:

[0017] S1. A core buffer layer, a β-Ga2O3 carbon getter layer, and an undoped transition layer are sequentially grown on a silicon carbide substrate.

[0018] S2. Above the undoped transition layer, a β-Ga2O3 drift layer is grown using a precursor spatiotemporal separation alternating gas supply mode. The precursor spatiotemporal separation alternating gas supply mode is as follows: in a single gas supply cycle, only the gallium precursor and carrier gas are introduced first, without the oxygen source, so that gallium atoms are adsorbed on the growth surface to form a sub-monolayer of metallic gallium; then only the oxygen source and carrier gas are introduced, without the gallium precursor, to oxidize the sub-monolayer of metallic gallium in situ to β-Ga2O3.

[0019] S3. During the growth of the β-Ga2O3 drift layer, after each sublayer of a preset thickness is grown, the precursor spatiotemporal separation alternating gas supply mode is paused, and an in-situ flash annealing step is performed. The in-situ flash annealing step is as follows: the temperature is raised to an annealing temperature higher than the growth temperature and held for a preset time before being lowered back to the growth temperature. Then, the precursor spatiotemporal separation alternating gas supply mode is restored to continue growth.

[0020] Repeat steps S2 and S3 until the β-Ga2O3 drift layer reaches the target thickness.

[0021] Furthermore, the single gas supply cycle of the precursor spatiotemporal separation alternating gas supply mode is 6~12s; wherein, the time for introducing gallium precursor is 2~5s, the time for introducing oxygen source is 2~5s, and a pure carrier gas purification interval of 0.5~1s is set between introducing gallium precursor and introducing oxygen source and between introducing oxygen source and the next introduction of gallium precursor.

[0022] Furthermore, the growth temperature is 880~920℃; the annealing temperature is 960~1000℃; the preset time is 60~90s; and the preset thickness is 200~500nm.

[0023] Furthermore, in step S1, the growth of the β-Ga2O3 carbon getter layer uses trimethylgallium as a gallium precursor, the V / III ratio is 50~100, and the growth temperature is 830~870℃.

[0024] Furthermore, in step S3, during the holding phase of the annealing temperature, oxygen source pulses are intermittently introduced. The duration of each oxygen source pulse is 3-5 seconds, and the number of pulses is 2-3 times. Adjacent oxygen source pulses are separated by pure carrier gas.

[0025] Beneficial effects: The SiC substrate MOCVD gallium oxide epitaxial wafer provided by this invention has the following beneficial effects:

[0026] (1) This invention establishes a gentle carbon concentration decreasing gradient from the β-Ga2O3 drift layer to the β-Ga2O3 carbon getter layer by setting a β-Ga2O3 carbon getter layer below the β-Ga2O3 drift layer and isolating the β-Ga2O3 carbon getter layer and the β-Ga2O3 drift layer with an undoped transition layer. This gradient provides a clear diffusion direction for the residual carbon atoms in the β-Ga2O3 drift layer. The presence of the undoped transition layer avoids the carbon in the β-Ga2O3 carbon getter layer from directly diffusing upward to the bottom of the β-Ga2O3 drift layer during high-temperature long-term growth, and at the same time avoids the formation of an electroactive defect band at the interface between the β-Ga2O3 carbon getter layer and the β-Ga2O3 drift layer due to an excessively steep concentration gradient.

[0027] (2) This invention forms a periodic annealing repair interface within the β-Ga2O3 drift layer, periodically geothermally activating the early carbon substitution-gallium vacancy associations formed during epitaxial growth, promoting their dissociation back to the free state, and simultaneously promoting the migration of free gallium vacancies to extended defects, thereby reducing the number of precursors for deep-level traps. The formation of the annealing repair interface can be characterized and confirmed by deep-level transient spectroscopy (DLTS).

[0028] (3) The preparation method of the present invention completely isolates the gallium precursor and the oxygen source in the time dimension through the spatiotemporal separation of the precursor and the alternating gas supply mode, thereby eliminating the path of generating carbon-containing particles by gas phase pre-reaction from the source, and using the oxygen source to directly oxidize and remove the residual carbon on the growth surface without competition during the oxidation half-cycle, thereby reducing the carbon doping rate by more than one order of magnitude.

[0029] (4) The above three mechanisms act on the generation stage, defect complex formation stage, and accumulation stabilization stage of the carbon impurity life cycle, respectively, forming a complete carbon purification cascade. There is a deep synergy among the three: spatiotemporal separation gas supply reduces carbon input, making the β-Ga2O3 carbon getter layer less prone to saturation; in-situ flash annealing provides kinetic activation energy for carbon diffusion; and the concentration gradient of the β-Ga2O3 carbon getter layer provides a directional migration direction for the free carbon released by flash annealing. The effect of the three-level synergy is far greater than the sum of the effects of implementing them individually.

[0030] (5) The carbon concentration of the β-Ga2O3 drift layer of the epitaxial wafer of the present invention can be reduced to 5×10⁻⁶. 15 cm-3 Below this, the carrier compensation ratio can be controlled within 2, and the carrier mobility can reach 150 cm⁻¹. 2 With a voltage of / V·s or higher, β-Ga2O3 provides a reliable material basis for its application in high-voltage power devices of 1200V and above. Attached Figure Description

[0031] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below.

[0032] Figure 1 This is a schematic diagram of the cross-sectional structure of a gallium oxide epitaxial wafer on a SiC substrate provided in an embodiment of the present invention.

[0033] Figure 2 This is a timing diagram of the precursor spatiotemporal separation alternating gas supply mode in an embodiment of the present invention.

[0034] Figure 3 This is a schematic diagram of the overall process of β-Ga2O3 drift layer growth in an embodiment of the present invention.

[0035] Figure 4 This is a schematic diagram comparing the carbon concentration depth distribution of the epitaxial wafer provided in this embodiment of the invention with that of a conventional MOCVD epitaxial wafer.

[0036] Figure 5 This is a schematic diagram of the temperature-time curve of the in-situ flash annealing step in an embodiment of the present invention.

[0037] Figure 6 This is a schematic diagram comparing the depth distribution of SIMS carbon concentration in the β-Ga2O3 carbon getter layer before and after the growth of the β-Ga2O3 drift layer in an embodiment of the present invention.

[0038] The markings in the figure are as follows: 10-Silicon carbide substrate, 20-Nucleation buffer layer, 30-β-Ga2O3 carbon getter layer, 40-Undoped transition layer, 50-β-Ga2O3 drift layer, 51-Drift layer sublayer, 52-Annealed repair interface. Detailed Implementation

[0039] The present invention will now be described more clearly and completely by way of a preferred embodiment in conjunction with the accompanying drawings, but this does not limit the invention to the scope of the described embodiment.

[0040] See Figure 1The SiC substrate gallium oxide epitaxial wafer provided in this embodiment of the invention comprises, from bottom to top: a silicon carbide substrate 10, a nucleation buffer layer 20, a β-Ga2O3 carbon getter layer 30, an undoped transition layer 40, and a β-Ga2O3 drift layer 50. The β-Ga2O3 drift layer 50 is composed of multiple drift layer sublayers 51 and annealing repair interfaces 52 stacked alternately.

[0041] The silicon carbide substrate 10 is of the 4H crystal form with n + The conductive silicon carbide single crystal substrate has a resistivity of 0.015~0.025 Ω·cm and a thickness of 350~500 μm. The conductive silicon carbide substrate was chosen because the target application of this epitaxial wafer is a vertical structure power device (such as a Schottky barrier diode or power MOSFET). The conduction current needs to travel longitudinally through the entire epitaxial structure to reach the back electrode of the substrate; therefore, the substrate is required to have good longitudinal conductivity. The silicon carbide substrate 10 has a crystal orientation of (0001) plane and an off-axis angle of 4°.

[0042] The nucleation buffer layer 20 is a β-Ga₂O₃ thin film with a thickness of 50-100 nm. Since silicon carbide is hexagonal (4H-SiC, space group P63mc) and β-Ga₂O₃ is monoclinic (space group C2 / m), their crystal symmetries are incompatible. If the β-Ga₂O₃ carbon getter layer 30 is grown directly on the silicon carbide surface, it will exhibit a polycrystalline or strongly textured state due to the lack of an orientation template, and none of the epitaxial layers above it will achieve single-crystal quality. The nucleation buffer layer 20 first forms β-Ga₂O₃ nuclei with a dominant orientation on the silicon carbide substrate 10 under low-temperature conditions, providing an orientation template for subsequent layers. The nucleation buffer layer 20 tolerates higher defect and carbon impurity levels, but the established epitaxial orientation relationship is a fundamental prerequisite for the feasibility of the entire epitaxial wafer.

[0043] β-Ga₂O₃ carbon getter layer 30 is a β-Ga₂O₃ thin film with a thickness of 20~50 nm and a carbon concentration of 1×10⁻⁶. 18 ~1×10 19 cm -3 The intentionally introduced high concentration of carbon in the β-Ga₂O₃ carbon getter layer 30 creates a concentration gradient of at least two orders of magnitude between it and the low carbon concentration in the upper β-Ga₂O₃ drift layer 50. According to Fick's first law, the diffusion flux J is proportional to the concentration gradient:

[0044]

[0045] Where D is the diffusion coefficient of carbon in β-Ga₂O₃, C is the carbon concentration, and x is the coordinate along the thickness direction. Since the carbon concentration increases from the β-Ga₂O₃ drift layer 50 (low concentration end) to the β-Ga₂O₃ carbon getter layer 30 (high concentration end), the concentration gradient... The concentration gradient is positive, while the diffusion flux J is negative, indicating that the net migration direction of carbon atoms is from the β-Ga2O3 drift layer 50 to the β-Ga2O3 carbon getter layer 30. This concentration gradient provides a continuous thermodynamic driving force for the directional removal of residual carbon from the β-Ga2O3 drift layer 50.

[0046] The upper limit of carbon concentration for the β-Ga2O3 carbon getter layer 30 is set to 1×10⁻⁶. 19 cm -3 Rather than higher, this is based on the following considerations: when the carbon concentration exceeds 1×10 19 cm -3 At this point, the solid solubility of carbon atoms in the β-Ga₂O₃ lattice approaches its limit. Excess carbon will exist in the form of carbon clusters or graphite phase precipitates, causing stress concentration and additional defects, which in turn degrades the crystal quality of the upper undoped transition layer 40 and the β-Ga₂O₃ drift layer 50. The lower limit of carbon concentration is chosen to be 1×10⁻⁶. 18 cm -3 This is to ensure the connection with the β-Ga2O3 drift layer 50 (carbon concentration not greater than 5 × 10⁻⁶). 15 cm -3 The concentration ratio between the two is not less than 100 (in practice, it can reach 200~10000), providing sufficient driving force for the concentration gradient.

[0047] The reason for choosing a thickness of 20~50nm for the β-Ga2O3 carbon getter layer 30 is that when the thickness is too thin (<20nm), the total amount of carbon contained in the β-Ga2O3 carbon getter layer 30 is insufficient to receive the carbon atoms released by the β-Ga2O3 drift layer 50 during the entire growth process, which poses a risk of saturation; when the thickness is too thick (>50 nm), the high defect density of the β-Ga2O3 carbon getter layer 30 itself (caused by high carbon doping) will affect the initial crystal quality of the β-Ga2O3 drift layer 50 by extending upward through dislocations, and also increases unnecessary series resistance.

[0048] The technical concept of the β-Ga2O3 carbon getter layer 30 is inspired by gettering techniques in silicon-based CMOS integrated circuit processes. In silicon-based CMOS processes, by introducing mechanical damage or high phosphorus concentration regions on the back side of the silicon wafer, metal impurities (such as iron and copper) in the active region on the front side of the chip are directionally attracted to the back side of the silicon wafer using concentration or stress gradients, thereby improving the purity of the active region. This invention transplants this cross-disciplinary gettering concept to the β-Ga2O3 heteroepitaxial system: the β-Ga2O3 carbon getter layer 30 is similar to the back-side getter layer in silicon processes, the β-Ga2O3 drift layer 50 is similar to the active region, and the carbon impurities are similar to the metal impurities in silicon processes. However, the difference lies in the fact that silicon-based gettering utilizes the high diffusion rate of metal impurities in silicon (which can migrate at room temperature), while the diffusion coefficient of carbon in β-Ga2O3 is extremely low (only 10 at 900 °C). -15 cm 2 (on the order of / s), therefore, in-situ flash annealing is needed to provide instantaneous high temperature to activate carbon diffusion—this is the physical basis for the deep synergy between the β-Ga2O3 carbon getter layer 30 and the annealed repair interface 52.

[0049] The undoped transition layer 40 is a β-Ga2O3 thin film with a thickness of 100~200nm. No intentional doping is introduced, and its carbon concentration is between that of the β-Ga2O3 carbon getter layer 30 and the β-Ga2O3 drift layer 50. The undoped transition layer 40 is an indispensable component of this epitaxial wafer structure, and it has two functions: First, it establishes a gentle transition region of carbon concentration between the β-Ga2O3 carbon getter layer 30 and the β-Ga2O3 drift layer 50, avoiding the formation of electroactive defect bands due to excessively steep concentration gradients at the interface. If such local enrichment occurs at the bottom of the drift layer, it will lead to the generation of high-density deep-level traps, which will worsen the reverse leakage current of the device. Second, it provides spatial isolation to prevent the high concentration of carbon in the β-Ga2O3 carbon getter layer 30 from directly diffusing upwards to the bottom of the β-Ga2O3 drift layer 50 during the long-term (more than 10 hours) high-temperature growth of the drift layer. If the undoped transition layer 40 is missing and the β-Ga2O3 carbon getter layer 30 is in direct contact with the β-Ga2O3 drift layer 50, during the growth process at 900 °C for more than 10 hours, the carbon in the β-Ga2O3 carbon getter layer 30 will inevitably diffuse into the bottom tens of nanometers of the β-Ga2O3 drift layer 50 through thermal diffusion, forming a bottom region with an abnormally high carbon concentration. This not only fails to achieve the carbon gettering effect but also becomes a "release source" of carbon.

[0050] β-Ga₂O₃ drift layer 50 is a β-Ga₂O₃ thin film with a thickness of 10~20 μm, n-type doping (doping element is Si), and a doping concentration of 5×10⁻⁶. 14 ~5×10 16 cm-3 The β-Ga₂O₃ drift layer 50 is the core functional layer in the power device that withstands reverse blocking voltage and provides a current path during forward conduction. The carbon concentration of the β-Ga₂O₃ drift layer 50 is no greater than 5 × 10⁻⁶. 15 cm -3 This is guaranteed by a three-tiered carbon purification and synergy mechanism.

[0051] The internal structure of the β-Ga2O3 drift layer 50 is not a homogeneous thin film, but rather consists of multiple drift sublayers 51 and annealing repair interfaces 52 stacked alternately. Each drift sublayer 51 has a thickness of 200~500nm and is formed by a precursor spatiotemporal separation alternating gas supply mode. The annealing repair interface 52 between adjacent drift sublayers 51 is a characteristic interface formed during the in-situ flash annealing step performed during the growth process.

[0052] The structural characteristics and characterization methods of the annealed repair interface 52 are as follows: The microscopic differences between the annealed repair interface 52 and its adjacent drift sublayer 51 can be confirmed by the following two complementary characterization methods. The first characterization method is deep level transient spectroscopy (DLTS). Layer-by-layer peeling DLTS testing is performed on the β-Ga2O3 drift layer 50 along its thickness direction (i.e., firstly, a Schottky contact is prepared on the surface of the complete β-Ga2O3 drift layer 50 and DLTS measurement is performed to record the concentration and energy level position of the deep level traps; then, a certain thickness is removed by dry etching and a Schottky contact is re-prepared and measured again; this process is repeated until the full thickness of the β-Ga2O3 drift layer 50 is covered), which can obtain the distribution curve of the deep level trap concentration along the thickness direction. The distribution curve exhibits periodic concentration modulation: at the depth corresponding to the annealed repair interface 52, the concentration of deep-level traps (electron traps with activation energies approximately 1.0–1.5 eV below the conduction band) associated with the C_O-V_Ga complex in the DLTS signal shows a local minimum; while in the central region of the drift sublayer 51, the concentration of these traps is relatively high. The spatial period of this periodic modulation is consistent with the preset sublayer thickness (200–500 nm). The second characterization method is secondary ion mass spectrometry (SIMS) carbon depth distribution measurement. High-resolution SIMS testing of the β-Ga2O3 drift layer 50 reveals a local decrease in carbon concentration at the depth corresponding to the annealed repair interface 52, with its spatial period also consistent with the sublayer thickness.

[0053] Taking a 10 μm thick β-Ga2O3 drift layer 50 as an example, when the sublayer thickness is selected as 300 nm, the β-Ga2O3 drift layer 50 contains 33 drift layer sublayers 51 and 33 annealing repair interfaces 52. The basis for selecting a sublayer thickness of 300 nm is the balance between epitaxial efficiency and annealing effect: if the sublayer thickness exceeds 500 nm, the influence depth of a single annealing is insufficient to cover the central region of the sublayer, resulting in the defect associations in the center of the sublayer not being effectively thermally activated; if the sublayer thickness is less than 200 nm, too frequent flash annealing interruptions will significantly reduce epitaxial efficiency (each annealing interruption consumes about 3 minutes) and cause the surface of the epitaxial layer to roughen due to multiple temperature rises and falls.

[0054] See Figure 3 The preparation method of this embodiment includes the following specific steps:

[0055] Step S0: In-situ cleaning of the substrate.

[0056] The 4H-SiC substrate was placed in the MOCVD reaction chamber. The reaction chamber temperature was raised to 1100°C and held for 300 seconds in a hydrogen atmosphere to perform in-situ thermal cleaning of the silicon carbide substrate 10, removing the native oxide layer and organic contaminants. The temperature was then lowered to 750°C to prepare for the growth of the nucleation buffer layer 20.

[0057] Step S1a: Growth of nucleation buffer layer.

[0058] β-Ga₂O₃ nucleation buffer layers 20 were grown at 750 °C using triethylgallium (TEGa) as the gallium precursor and O₂ as the oxygen source in a conventional continuous MOCVD mode (i.e., TEGa and O₂ were simultaneously introduced into the reaction chamber). The flow rate of TEGa was 15–30 μmol / min, the flow rate of O₂ was 500–1500 sccm, the V / III ratio (molar flow ratio of group V elements to group III elements) was 1000–2000, the growth rate was 0.2–0.5 μm / h, and the growth time was 6–30 min, resulting in nucleation buffer layers 20 with a thickness of 50–100 nm.

[0059] The reason for choosing to grow the nucleation buffer layer 20 at a low temperature of 750 °C is that the low temperature is conducive to increasing the nucleation density, forming a high density of β-Ga2O3 nuclei on the surface of the silicon carbide substrate 10, and reducing the probability of rotating domain formation in the subsequent high-temperature growth stage. The reason for using the conventional continuous MOCVD mode (rather than the precursor spatiotemporal separation alternating gas supply mode) for the nucleation buffer layer 20 is that this layer allows for a higher carbon content without the need for special carbon purification; and the gas-phase pre-reaction degree of TEGa and O2 is relatively light at low temperature, so using the continuous mode can simplify the process flow.

[0060] Step S1b: Growth of β-Ga2O3 carbon absorber layer.

[0061] The temperature was raised to 830-870℃ (850℃ was selected in this embodiment), and the gallium precursor was switched from TEGa to trimethylgallium (TMGa). TMGa was chosen instead of TEGa because the methyl radicals (CH3·) generated by TMGa pyrolysis are far more chemically active than the ethyl radicals (C2H5·) generated by TEGa pyrolysis. Methyl radicals are more likely to participate in surface reactions and be incorporated into the lattice, thus achieving efficient carbon incorporation. The V / III ratio was reduced to 50-100 (far lower than the conventional 500-2000) to create an oxygen-deficient surface, further promoting carbon occupation of oxygen sites (forming C_O). The TMGa flow rate was 20-40 μmol / min, the O2 flow rate was 100-200 sccm (standard milliliters per minute), and the growth time was 2-5 min, resulting in a β-Ga2O3 carbon getter layer 30 with a thickness of 20-50 nm.

[0062] Under the above conditions, the carbon concentration of the β-Ga2O3 carbon getter layer 30 can reach 1×10⁻⁶. 18 ~1×10 19 cm -3 Carbon exists primarily in the form of C_O (carbon replaced by oxygen) in the β-Ga2O3 lattice, serving as the endpoint of the concentration gradient convergence of carbon atom migration during the subsequent growth of the β-Ga2O3 drift layer 50.

[0063] Step S1c: Growth of undoped transition layer.

[0064] Switch the gallium precursor back to TEGa (low-carbon precursor), raise the temperature to 900℃, and enter the precursor spatiotemporal separation alternating gas supply mode (STAS mode, see step S2 for details). In STAS mode, grow a 100-200 nm undoped β-Ga₂O₃ transition layer 40 at a growth temperature of 880-920℃; in this embodiment, 900℃ is selected. No SiH₄ dopant source is introduced.

[0065] The undoped transition layer 40, grown in STAS mode, has a significantly lower carbon concentration than the β-Ga₂O₃ carbon getter layer 30, but may still be slightly higher than the target value for the β-Ga₂O₃ drift layer 50. The presence of the undoped transition layer 40 causes the carbon concentration to decrease along the depth direction from 10 times that of the β-Ga₂O₃ carbon getter layer 30. 18 ~10 19 cm -3 Passing through layer 40, 10 16 ~10 17 cm -3 The drift layer to β-Ga2O3 has a maximum length of 5 × 10⁻⁶. 15 cm -3 This forms a gradual transitional distribution with a three-stage decreasing pattern, effectively suppressing the local enrichment of carbon at the interface.

[0066] Steps S2 and S3: Growth of β-Ga2O3 drift layer (STAS mode + periodic in-situ flash annealing).

[0067] See Figure 2 and Figure 5 The β-Ga2O3 drift layer 50 was grown using a combined process of precursor space-time separation alternating gas supply mode (STAS mode) and periodic in-situ flash annealing (IFA). The growth temperature of the β-Ga2O3 drift layer 50 was 880~920℃, and 900℃ was selected in this embodiment.

[0068] The working principle of the precursor space-time separation alternating gas supply mode is as follows:

[0069] In conventional continuous MOCVD growth, gallium precursor (TEGa) and oxygen source (O2) are simultaneously introduced into the reaction chamber. They undergo a pre-reaction in the gas phase before reaching the substrate surface, generating GaO. x Nanoparticles and carbon-containing organic gallium oxides. After these gas-phase reaction products settle onto the growth surface, they reduce the utilization rate of the effective precursor (to only about 30%) and directly introduce carbon into the epitaxial layer lattice.

[0070] The precursor spatiotemporal separation alternating gas supply mode of the present invention eliminates the aforementioned gas-phase pre-reaction at its source by isolating the gallium precursor from the oxygen source in the time dimension. See also Figure 2 A single gas supply cycle T is 6~12s, consisting of the following four stages:

[0071] Stage A (Gallium Deposition Stage): Lasts 2-5 seconds. Only TEGa and N2 carrier gases are introduced; O2 is not introduced. In an oxygen-free environment, TEGa decomposes on the substrate surface at 900°C, and gallium atoms adsorb onto the growth surface to form a sub-monolayer of metallic gallium. The ethyl group (C2H5·) produced by TEGa decomposition undergoes a β-H elimination reaction to generate ethylene (C2H4) and H2, which are then desorbed from the surface in gaseous form. Since O2 is absent, no gas-phase pre-reaction occurs, and no carbon-containing GaO is formed. x Particles. A small amount of carbon remaining on the surface (from incompletely desorbed ethyl fragments) is temporarily adsorbed between gallium atoms at this stage.

[0072] Phase B (First Purification Interval): Lasts 0.5~1s. Only pure N2 carrier gas is introduced, while TEGa and O2 are shut off. This interval removes residual TEGa vapor from the reaction chamber and pipelines, preventing it from cross-reacting with the subsequently introduced O2 upstream of the pipeline or reaction chamber.

[0073] Stage C (Oxidation + Carbon Purification Stage): Lasts 2-5 seconds. Only O2 and N2 carrier gases are introduced; TEGa is not introduced. After O2 reaches the substrate surface, two parallel surface reactions occur: First, the sub-monolayer gallium adsorbed in Stage A is oxidized to β-Ga2O3, completing the lattice construction of one deposition cycle; second, the surface carbon remaining in Stage A is oxidized to CO or CO2 gas and desorbed from the surface. This stage is the first stage of the carbon purification mechanism of this invention—source cutoff. Since TEGa is not present in the reaction chamber at this time, O2 can directly act on the surface carbon without competing with the gallium precursor, and the carbon oxidation and removal efficiency is much higher than that of conventional continuous MOCVD mode (in the conventional mode, most of the O2 is consumed by the gallium precursor to form Ga2O3, and only a very small amount of O2 can react with the surface carbon).

[0074] Phase D (Second Purification Interval): Lasts 0.5~1s. Only pure N2 carrier gas is introduced. This interval removes residual O2 to prevent it from undergoing gas-phase pre-reaction with TEGa in the next cycle.

[0075] The four stages described above constitute a complete STAS cycle, with each cycle depositing approximately 0.3–1 nm of β-Ga₂O₃. Assuming a single cycle of 8 seconds and a single-cycle deposition thickness of 0.5 nm, the equivalent growth rate of the STAS mode is 0.5 nm × 3600 / 8 = 225 nm / h = 0.225 μm / h. After optimization with a 4–8x speedup, the rate is approximately 1–2 μm / h. In actual growth, the growth rate can be increased by increasing the TEGa single-pulse flow rate and shortening the purge interval.

[0076] The working principle of the in-situ flash annealing process is as follows:

[0077] See Figure 5 After growing drift sublayers of 200–500 nm each in STAS mode, the STAS cycle is paused, and an in-situ flash annealing (IFA) step is performed. The temperature-time process of the IFA step is as follows:

[0078] Heating stage: Turn off the TEGa and SiH4 doping sources, retain the N2 carrier gas, and raise the substrate temperature from the growth temperature (900℃) to the annealing temperature (960~1000℃) at a rate of 15~20℃ / s. The heating time is approximately 3~5s.

[0079] Holding stage: Maintain the annealing temperature for 60-90 seconds. This stage is the window for IFA to play its core role. The annealing temperature of 960-1000℃ is chosen based on the following physical basis: the thermal dissociation activation energy of the C_O-V_Ga complex is approximately 2.0 eV, corresponding to a characteristic temperature of approximately 950℃. When the temperature exceeds 950℃, the dissociation rate of the C_O-V_Ga complex exceeds its formation rate, and the complex begins to undergo net dissociation. The upper limit of the annealing temperature is chosen to be 1000℃ rather than higher because above 1050℃, β-Ga2O3 begins to decompose, producing highly volatile Ga2O intermediates, leading to surface roughening of the epitaxial layer and deviation from the stoichiometry.

[0080] During the heat preservation phase, short O2 pulses (lasting 3-5 seconds each) are intermittently introduced 2-3 times. The function of the O2 pulses is to oxidize the free carbon released from the IFA dissociation from the outermost surface of the epitaxial layer into CO / CO2 gaseous desorption—this constitutes the second stage of carbon purification (blocking complex stabilization). The reason for separating the O2 pulses with pure N2 is that, in the O2-free window, the free gallium vacancies released by the IFA can diffuse and migrate sufficiently to dislocation nuclei or extended defects without being recaptured by oxygen, where they are absorbed and annihilated. The diffusion coefficient of gallium vacancies at 980℃ is approximately 10. -12 cm 2 / s, the diffusion length within 60s is approximately 80nm. At a dislocation density of 10... 8 cm -2 In the epitaxial layer, the average distance from any point to the nearest dislocation nucleus is approximately 560 nm. Therefore, the 80 nm diffusion length of a single IFA is insufficient for a vacancy to reach the dislocation nucleus. However, through the cumulative effect of multiple IFAs (one IFA every 300 nm sublayer, requiring 33 IFAs for a 10 μm drift layer), the total diffusion distance of the vacancy after 33 "multi-step relays" can reach 33 × 80 nm = 2640 nm, exceeding the average dislocation spacing, ensuring that most vacancy eventually reach the dislocation nucleus and annihilate.

[0081] During the IFA (In-Fluid Aeration) heat treatment phase, the diffusion coefficient of carbon in β-Ga₂O₃ decreases from approximately 10 at 900℃. -15 cm 2 / s increased to approximately 10 at 980℃ -13 cm 2Within 60 seconds, the diffusion length increases from approximately 2.4 nm to approximately 24 nm. While this 24 nm single-diffusion length is much smaller than the total thickness of the β-Ga₂O₃ drift layer 50 (10~20 μm), the concentration gradient established by the β-Ga₂O₃ carbon getter layer 30 imparts a net downward bias to each diffusion of carbon atoms. Without a concentration gradient (i.e., without the β-Ga₂O₃ carbon getter layer 30), carbon atoms undergo random walk (Brownian motion), resulting in zero net displacement after N IFAs. However, driven by the concentration gradient established by the β-Ga₂O₃ carbon getter layer 30, each diffusion of carbon atoms has a net downward component (towards the β-Ga₂O₃ carbon getter layer 30). After 33 IFAs, the total net downward displacement of carbon atoms can reach hundreds of nanometers—this constitutes the third stage of carbon purification (concentration gradient convergence).

[0082] Cooling phase: The temperature is reduced back to 900℃ at a rate of 10~15℃ / s to resume the STAS cycle and continue the growth of the next drifting sublayer 51.

[0083] The n-type doping of the β-Ga₂O₃ drift layer 50 is achieved by simultaneously introducing SiH₄ (silane) in stage A of the STAS mode. SiH₄ is ​​introduced during the Ga deposition stage of stage A, allowing Si atoms and Ga atoms to be simultaneously adsorbed onto the growth surface. In the oxidation stage of stage C, Si atoms are oxidized and fixed at Ga sites in the Ga₂O₃ lattice, forming Si_Ga donors. The flow rate of SiH₄ is ​​precisely adjusted using a mass flow controller to control the Si doping concentration at 5 × 10⁻⁶. 14 ~5×10 16 cm -3 Within the range.

[0084] The three-stage carbon purification mechanism of this invention works synergistically to address the entire lifecycle of carbon impurities during epitaxial growth:

[0085] First stage (source cutoff): The spatiotemporal separation alternating gas supply mode of the precursor isolates the gallium precursor from the oxygen source in the time dimension, eliminates gas-phase pre-reaction, and performs non-competitive direct oxidation removal of residual carbon on the surface during the oxidation half-cycle. This stage acts on the generation stage and surface adsorption stage of the carbon impurity life cycle, reducing the carbon incorporation rate from 10 in conventional MOCVD. 17 cm -3 The magnitude dropped to 10 16 cm -3 Magnitude.

[0086] The second stage (blocking stabilization): Periodic in-situ flash annealing dissociates the carbon substitution-gallium vacancy complex back into its free state through instantaneous high temperature before the complex completes thermodynamic stabilization. O2 pulses then oxidize and desorb the released free carbon and drive the released free gallium vacancies to dislocation nuclei for annihilation. This stage acts on the defect complex formation stage of the carbon impurity lifecycle.

[0087] The third stage (concentration gradient convergence): The decreasing carbon concentration gradient established by the carbon getter layer provides the thermodynamic driving force for the directional migration of free carbon released in the second stage, causing carbon atoms to migrate directionally towards the carbon getter layer along the concentration gradient direction during the instantaneous high-temperature window of each flash annealing. This stage acts as the cumulative stabilization phase of the carbon impurity lifecycle.

[0088] It should be noted that the actual migration distance of carbon in the third-level mechanism within a single annealing window is limited (due to annealing time and temperature). The direct migration of carbon from the upper part of the β-Ga2O3 drift layer to the β-Ga2O3 carbon getter layer cannot be completed in a single annealing. The net migration of carbon towards the β-Ga2O3 carbon getter layer is a cumulative effect of multiple annealing processes, and this cumulative effect is influenced by multiple factors (including the initial carbon concentration distribution in the β-Ga2O3 drift layer, annealing temperature and time, sublayer thickness, and the capacity of the carbon getter layer, etc.), making it unsuitable for simple linear superposition-based quantitative inference. This invention, through the segmented comparative experiments (experimental groups A~D) and the pre- and post-SIMS comparison experiments of the carbon getter layer, confirms the overall effect of the three-level synergy and the fact of net carbon migration towards the carbon getter layer based on actual test results.

[0089] There is an inseparable causal dependency among the three stages: the first stage reduces the total carbon input, preventing the carbon getter layer from becoming saturated due to excessive carbon influx, thus maintaining the continued effectiveness of the third stage; the second stage provides the kinetic activation energy required for carbon atoms to be released from the lattice fixed state to a mobile state, enabling the concentration gradient driving force of the third stage to act on the free carbon—without the activation of the second stage, carbon is locked in the lattice as a stable C_O-V_Ga complex, and the concentration gradient of the third stage cannot drive the migration of the locked carbon; the third stage provides a definite migration direction and final state convergence point for the free carbon released in the second stage—without the concentration gradient of the third stage, the free carbon released in the second stage will wander randomly and will be recaptured by nearby gallium vacancies after cooling at the end of flash annealing, resulting in zero net carbon removal effect.

[0090] Experimental verification:

[0091] Segmented comparative experiments (to verify the independent and synergistic contributions of each level of mechanism):

[0092] To verify the independent contribution of each stage in the three-stage carbon purification mechanism and the overall synergistic effect of the three, the following four sets of comparative experiments (Groups A to D) were set up on the same MOCVD equipment and 4H-SiC substrate, with a target Si doping concentration of 1×10⁻⁶. 15 cm -3 The drift layer thickness is 10 μm:

[0093] Experimental Group A (Control Group, Conventional MOCVD): TEGa and O2 were continuously supplied, the growth temperature was 900℃, the V / III ratio was 1500, there was no carbon getter layer, no undoped transition layer, and no in-situ annealing.

[0094] Experimental Group B (Level 1 only): Drift layers were grown using only the STAS mode, without carbon getter layers, undoped transition layers, or in-situ annealing. Other conditions were the same as in the examples.

[0095] Experimental group C (first stage + second stage): The drift layer was grown in STAS mode and IFA (980℃ / 60s) was performed every 300 nm, but there was no carbon getter layer and no undoped transition layer.

[0096] Experimental Group D (Three-level synergy, complete scheme): Fully implements the scheme of this invention, including a nucleation buffer layer and a carbon getter layer (30nm, carbon concentration 5×10). 18 cm -3 ), undoped transition layer (150nm), STAS mode growth drift layer (8s cycle), periodic IFA (once every 300nm, 980℃ / 60s).

[0097] The carbon concentration in the drift layer of each group was determined by secondary ion mass spectrometry (SIMS), the carrier concentration and mobility were measured by the van der Bauer Hall effect, and the concentration of deep-level traps (1.0–1.5 eV) below the conduction band in the drift layer was measured by layer-by-layer exfoliation DLTS. The test results are as follows:

[0098] Experimental Group A: Carbon concentration in the drift layer is approximately 3 × 10⁻⁶ 17 cm -3 The concentration of deep-level traps (EC-1.2eV) is approximately 5×10⁻⁶. 16 cm -3 The carrier concentration is approximately 2 × 10⁻⁶. 14 cm -3 Carrier mobility is approximately 45 cm⁻¹ 2 / V·s.

[0099] Experimental group B: Carbon concentration in the drift layer is approximately 4 × 10⁻⁶ 16 cm -3 The concentration of deep-level traps is approximately 1×10⁻⁶. 16 cm -3 The carrier concentration is approximately 3 × 10⁻⁶.14 cm -3 Carrier mobility is approximately 70 cm⁻¹ 2 / V·s. Compared to experimental group A, the STAS mode reduced the carbon concentration by about an order of magnitude (from 3 × 10⁻⁶ V·s⁻¹). 17 Reduced to 4×10 16 This confirms the independent contribution of the first-level source cutoff.

[0100] Experimental group C: Carbon concentration in the drift layer is approximately 1.5 × 10⁻⁶. 16 cm -3 The concentration of deep-level traps is approximately 2 × 10⁻⁶. 15 cm -3 The carrier concentration is approximately 5 × 10⁻⁶. 14 cm -3 Carrier mobility is approximately 100 cm⁻¹ 2 / V·s. Compared with experimental group B, the carbon concentration was further reduced, and the reduction in deep-level trap concentration (from 1×10⁻⁶) was also greater. 16 Reduced to 2×10 15 The decrease in carbon concentration was greater than that of carbon (from 4 × 10⁻⁶). 16 Reduced to 1.5×10 16 This indicates that IFA not only reduces carbon concentration by promoting carbon desorption, but also reduces the probability of incorporated carbon forming deep-level complexes by promoting associative dissociation and vacancy annihilation. However, the carbon concentration in the drift layer remains at 10¹. 6 The magnitude is higher than the target value (not greater than 5 × 10). 15 cm -3 ).

[0101] Experimental group D: Carbon concentration in the drift layer is approximately 3 × 10⁻⁶ 15 cm -3 The concentration of deep-level traps is approximately 3 × 10⁻⁶. 14 cm -3 The carrier concentration is approximately 8 × 10⁻⁶. 14 cm -3 Carrier mobility is approximately 155 cm⁻¹ 2 / V·s. Compared to experimental group C, the carbon concentration increased from 1.5 × 10⁻⁶ V·s. 16 Reduced to 3×10 15 (Reduced by 5 times), deep-level trap concentration from 2×10 15 Reduced to 3×10 14 (Reduced by approximately 7 times). This additional reduction cannot be explained by the carbon getter layer alone (the carbon getter layer itself does not directly remove carbon from the drift layer), but is the result of the synergistic effect of the carbon getter layer and the IFA: the concentration gradient established by the carbon getter layer gives the free carbon released by the IFA a preferred migration direction, making the net carbon efflux efficiency higher than that of random diffusion without a gradient.

[0102] See Figure 4 The comparison of carbon concentration depth distribution in these four sets of experiments intuitively presents the progressive and synergistic effects of the three-level mechanism.

[0103] Verification of carbon concentration changes in the carbon gettering layer (CGL) (direct evidence):

[0104] To directly verify whether carbon does indeed migrate and converge from the drift layer towards the carbon getter layer, the following comparative tests were performed on the epitaxial wafer of experimental group D:

[0105] See Figure 6 The SIMS carbon depth distribution in region 30 of the β-Ga2O3 carbon getter layer in experimental scheme D was measured with high resolution and compared with the following reference sample.

[0106] Reference sample E: Under the same conditions, only steps S0 to S1c are completed (i.e., after completing the growth of nucleation buffer layer 20, β-Ga2O3 carbon getter layer 30 and undoped transition layer 40, the β-Ga2O3 drift layer 50 is not grown, and the sample is directly cooled and removed).

[0107] Comparison results: The peak carbon concentration of the β-Ga2O3 carbon getter layer 30 in reference sample E is approximately 5 × 10⁻⁶. 18 cm -3 The total integrated carbon content (area under the peak) is used as a baseline value. In experimental group D, after the growth of all 10 μm β-Ga₂O₃ drift layers, the peak carbon concentration in region 30 of the β-Ga₂O₃ carbon getter layer is approximately 7 × 10⁻⁶. 18 cm⁻³, the total integrated carbon content increased by approximately 35% compared to the reference sample E. Meanwhile, the carbon concentration peak shape of the β-Ga₂O₃ carbon getter layer 30 in experimental group D exhibits an upward broadening characteristic (i.e., in the direction of the undoped transition layer 40) – this asymmetric broadening indicates that the additional carbon increase comes from above rather than from the substrate side below.

[0108] The comparative results directly demonstrate that during the growth of the β-Ga2O3 drift layer 50, carbon atoms do indeed migrate from above (in the direction of the drift layer and transition layer) and converge into the β-Ga2O3 carbon getter layer 30, resulting in a higher total carbon content in the β-Ga2O3 carbon getter layer 30 after the entire drift layer has grown compared to its initial growth stage. This experimental fact supports the practical effectiveness of the third-stage carbon purification mechanism (concentration gradient convergence) of this invention.

[0109] DLTS confirmation in the annealing repair interface:

[0110] Layer-by-layer exfoliation DLTS testing was performed on the β-Ga2O3 drift layer 50 of the epitaxial wafer in experimental group D. After each dry etching process removing approximately 100 nm, a new Ni / β-Ga2O3 Schottky contact was fabricated, and DLTS scanning was performed within the temperature range of 77–400 K. The results showed that in the depth distribution of the β-Ga2O3 drift layer 50, the concentration of deep-level traps (EC -1.2 eV, corresponding to defects in the C_O-V_Ga complex) exhibited a modulation period of approximately 300 nm—at the depth location corresponding to the annealed repair interface 52, the trap concentration was approximately 2 × 10⁻⁶. 14 cm -3 In the central region of sublayer 51 of the drift layer, the trap concentration is approximately 5 × 10⁻⁶. 14 cm -3 The ratio of the two is approximately 2.5:1, confirming a measurable difference in deep-level trap concentration between the annealed repair interface 52 and the drift layer sublayer 51.

[0111] As a control, in experimental group A (conventional MOCVD, without IFA), the concentration of traps at the same deep level in the drift layer was approximately uniformly distributed along the depth direction, without exhibiting periodic modulation characteristics, and the concentration was approximately 5 × 10⁻⁶. 16 cm -3 .

[0112] The DLTS results confirm the existence of the annealed repair interface 52 as a characterizable and identifiable structural feature, and provide a quantitative criterion: by performing layer-by-layer peeling DLTS tests, the periodic modulation of the deep-level trap concentration in the drift layer depth distribution (the modulation period is consistent with the preset sublayer thickness) can confirm the existence of the annealed repair interface.

[0113] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A SiC substrate MOCVD gallium oxide epitaxial wafer, characterized in that, From bottom to top, they include: silicon carbide substrate; A nucleation buffer layer is disposed above a silicon carbide substrate; A β-Ga2O3 carbon getter layer is disposed above the nucleation buffer layer; An undoped transition layer is disposed above the β-Ga2O3 carbon getter layer; A β-Ga2O3 drift layer is disposed above the undoped transition layer; The carbon concentration ratio of the β-Ga2O3 carbon getter layer to the β-Ga2O3 drift layer is not less than 100, and a decreasing carbon concentration gradient is formed from the β-Ga2O3 drift layer to the β-Ga2O3 carbon getter layer. The β-Ga2O3 drift layer contains multiple periodically distributed annealing repair interfaces along its thickness direction, and the drift layer sublayer is located between adjacent annealing repair interfaces. The deep-level trap concentration at the annealing repair interface, as determined by deep-level transient spectroscopy, is lower than the deep-level trap concentration within the drift layer sublayer.

2. The SiC substrate MOCVD gallium oxide epitaxial wafer according to claim 1, characterized in that, The thickness of the β-Ga₂O₃ carbon getter layer is 20~50 nm, and the carbon concentration is 1×10⁻⁶. 18 ~1×10 19 cm -3 The carbon concentration of the β-Ga2O3 drift layer is no greater than 5 × 10⁻⁶. 15 cm -3 .

3. The SiC substrate MOCVD gallium oxide epitaxial wafer according to claim 1, characterized in that, The thickness of the β-Ga₂O₃ drift layer is 10~20 μm, and the n-type doping concentration is 5×10⁻⁶. 14 ~5×10 16 cm -3 The thickness of the sublayer between adjacent annealing repair interfaces is 200~500nm.

4. The SiC substrate MOCVD gallium oxide epitaxial wafer according to claim 1, characterized in that, The thickness of the nucleation buffer layer is 50~100nm; the thickness of the undoped transition layer is 100~200nm, and the carbon concentration of the undoped transition layer is between the carbon concentration of the carbon getter layer and the carbon concentration of the drift layer.

5. The SiC substrate MOCVD gallium oxide epitaxial wafer according to claim 1, characterized in that, The carbon concentration of the β-Ga2O3 carbon getter layer after the completion of the entire β-Ga2O3 drift layer growth is higher than the carbon concentration when the initial growth is completed.

6. A method for preparing gallium oxide epitaxial wafers via MOCVD on a SiC substrate, characterized in that, Specifically, the following steps are included: S1. A core buffer layer, a β-Ga2O3 carbon getter layer, and an undoped transition layer are sequentially grown on a silicon carbide substrate. S2. Above the undoped transition layer, a β-Ga2O3 drift layer is grown using a precursor spatiotemporal separation alternating gas supply mode. The precursor spatiotemporal separation alternating gas supply mode is as follows: in a single gas supply cycle, only the gallium precursor and carrier gas are introduced first, without the oxygen source, so that gallium atoms are adsorbed on the growth surface to form a sub-monolayer of metallic gallium; then only the oxygen source and carrier gas are introduced, without the gallium precursor, to oxidize the sub-monolayer of metallic gallium in situ to β-Ga2O3. S3. During the growth of the β-Ga2O3 drift layer, after each sublayer of a preset thickness is grown, the precursor spatiotemporal separation alternating gas supply mode is paused, and an in-situ flash annealing step is performed. The in-situ flash annealing step is as follows: the temperature is raised to an annealing temperature higher than the growth temperature and held for a preset time before being lowered back to the growth temperature. Then, the precursor spatiotemporal separation alternating gas supply mode is restored to continue growth. Repeat steps S2 and S3 until the β-Ga2O3 drift layer reaches the target thickness.

7. The method for preparing a SiC substrate MOCVD gallium oxide epitaxial wafer according to claim 6, characterized in that, The single gas supply cycle of the precursor spatiotemporal separation alternating gas supply mode is 6~12s; wherein, the time for introducing gallium precursor is 2~5s, the time for introducing oxygen source is 2~5s, and a pure carrier gas purification interval of 0.5~1s is set between introducing gallium precursor and introducing oxygen source and between introducing oxygen source and the next introduction of gallium precursor.

8. The method for preparing a SiC substrate MOCVD gallium oxide epitaxial wafer according to claim 6, characterized in that, The growth temperature is 880~920℃; the annealing temperature is 960~1000℃; the preset time is 60~90s; and the preset thickness is 200~500nm.

9. The method for preparing a SiC substrate MOCVD gallium oxide epitaxial wafer according to claim 6, characterized in that, In step S1, the growth of the β-Ga2O3 carbon getter layer uses trimethylgallium as a gallium precursor, the V / III ratio is 50~100, and the growth temperature is 830~870℃.

10. The method for preparing a SiC substrate MOCVD gallium oxide epitaxial wafer according to claim 6, characterized in that, In step S3, during the holding phase of the annealing temperature, oxygen source pulses are intermittently introduced. The duration of each oxygen source pulse is 3-5 seconds, and the number of pulses is 2-3 times. Adjacent oxygen source pulses are separated by pure carrier gas.

Images