Carbon Doped Gallium Nitride: Advanced Semiconductor Material For High-Performance Electronic And Optoelectronic Devices

Fundamental Properties And Doping Mechanisms Of Carbon Doped Gallium Nitride

Carbon doped gallium nitride exhibits distinctive electronic properties arising from carbon's behavior as a deep acceptor when substituting nitrogen sites in the GaN lattice. The incorporation of carbon introduces energy levels approximately 0.9 eV above the valence band maximum, creating semi-insulating characteristics with resistivity exceeding 10^7 Ω·cm at room temperature 14. This high resistivity makes C-doped GaN invaluable for buffer layers in power electronic devices, where vertical leakage current suppression is critical.

The doping mechanism involves carbon atoms preferentially occupying nitrogen sites (C_N) during epitaxial growth, particularly under nitrogen-rich conditions. Carbon concentration in GaN layers can be precisely controlled from 10^16 to 10^19 atoms/cm³ through adjustment of growth parameters including V/III ratio, growth temperature (typically 900-1100°C), and carbon precursor flow rates 24. In p-type applications, carbon co-doping with magnesium has demonstrated synergistic effects, where carbon concentrations of 1-5×10^17 cm^-3 combined with magnesium doping enhance hole activation efficiency while maintaining crystallographic quality 219.

Key material characteristics include:

• Electrical resistivity: 10^7-10^10 Ω·cm for semi-insulating applications, tunable through carbon concentration 14
• Optical absorption edge: 3.4 eV (365 nm) for bandgap transitions, with sub-bandgap absorption features at 2.2-2.8 eV related to carbon deep levels 1
• Thermal stability: Carbon-related deep levels remain stable up to 1200°C, significantly higher than hydrogen-passivated magnesium acceptors 20
• Breakdown field strength: Enhanced to 3.5-4.0 MV/cm in carbon-doped buffer structures compared to 2.8 MV/cm in undoped GaN 4

The carbon doping profile critically influences device performance. In HEMT structures, extrinsically carbon-doped GaN buffer layers with concentrations of 5×10^18 cm^-3 reduce on-state resistance by 15-25% while improving breakdown voltage from 650V to over 900V compared to unintentionally doped buffers 4. For photoconductive switches, carbon concentrations of 2-8×10^17 cm^-3 enable sub-nanosecond optical quenching when illuminated with infrared light (850-1550 nm), a unique property not observed in undoped or silicon-doped GaN 1.

Synthesis Routes And Processing Techniques For Carbon Doped Gallium Nitride

Metalorganic Chemical Vapor Deposition (MOCVD) Growth

MOCVD represents the dominant synthesis method for carbon doped gallium nitride, offering precise control over carbon incorporation through manipulation of growth chemistry. Carbon doping is achieved through two primary routes: intentional introduction via carbon-containing precursors (e.g., CCl₄, CBr₄, or propane) or parasitic incorporation from metalorganic precursors such as trimethylgallium (TMGa) 24.

Optimized MOCVD growth parameters for C-doped GaN include:

1. Growth temperature: 950-1050°C for semi-insulating layers; 1000-1100°C for p-type co-doped structures 2
2. Reactor pressure: 100-300 Torr, with lower pressures favoring carbon incorporation 4
3. V/III ratio: 500-2000 for n-type suppression; 1000-3000 for p-type co-doping applications 2
4. Carbon precursor flow: 0.5-5 sccm CCl₄ or 10-50 sccm propane for controlled doping levels 14
5. Growth rate: 1.5-3.0 μm/h to maintain surface morphology and minimize point defect incorporation 2

For p-type carbon-magnesium co-doped GaN, a modified growth sequence enhances dopant activation 2. The process involves alternating supply of group-III precursors (TMGa) and group-V precursors (NH₃) with magnesium source (Cp₂Mg) supplied during both phases. This approach yields p-type layers with hole concentrations of 3-7×10^17 cm^-3 and resistivity of 1.5-3.5 Ω·cm, representing 40-60% improvement over conventional continuous-flow growth 2. Carbon concentration in these layers ranges from 8×10^16 to 3×10^17 cm^-3, with the carbon-to-magnesium ratio maintained at 0.2-0.4 for optimal electrical performance 219.

Hydride Vapor Phase Epitaxy (HVPE) And Bulk Crystal Growth

For applications requiring thick carbon-doped GaN layers (>50 μm) or freestanding substrates, HVPE provides growth rates of 50-200 μm/h while maintaining carbon doping control 1314. Carbon incorporation in HVPE occurs through reaction of gallium chloride (GaCl) with ammonia in the presence of carbon-containing species, typically introduced via methane or carbon tetrachloride at concentrations of 0.1-1.0% in the carrier gas 13.

HVPE-grown carbon-doped GaN substrates exhibit:

• Dislocation density: 5×10^5 to 2×10^6 cm^-2 for freestanding substrates, two orders of magnitude lower than heteroepitaxial films 1314
• Carbon concentration uniformity: ±15% across 2-inch diameter substrates when growth temperature is maintained at 1020±5°C 13
• Resistivity: 10^8-10^9 Ω·cm for semi-insulating grades; 0.02-0.05 Ω·cm for n-type oxygen co-doped variants 1314

Post-Growth Thermal Processing And Dopant Activation

Thermal annealing plays a crucial role in optimizing the electrical properties of carbon-doped GaN, particularly for p-type co-doped structures. Rapid thermal annealing (RTA) at 1200-1400°C for 30-300 seconds in nitrogen or forming gas (N₂/H₂) atmospheres activates magnesium acceptors while preserving carbon-related deep levels 20. Advanced annealing protocols employ heating rates exceeding 100°C/s to minimize surface decomposition and gallium desorption, with silicon carbide susceptors placed in close proximity (<500 μm) to the GaN epilayer to maintain surface stoichiometry through vapor-phase equilibrium 20.

For carbon-doped buffer layers in HEMT structures, post-growth annealing at 800-900°C for 10-30 minutes enhances semi-insulating properties by reducing residual shallow donor concentrations from 5×10^16 to below 1×10^16 cm^-3 4. This thermal treatment does not significantly alter carbon concentration but promotes carbon-vacancy complex formation, further increasing resistivity 4.

Electronic Structure And Charge Transport In Carbon Doped Gallium Nitride

Deep Level Characteristics And Compensation Mechanisms

Carbon in gallium nitride introduces deep acceptor levels that fundamentally alter charge transport and optical properties. Secondary ion mass spectrometry (SIMS) combined with deep-level transient spectroscopy (DLTS) reveals that carbon substituting nitrogen sites (C_N) creates an acceptor level at E_C - 0.9 eV, while carbon-vacancy complexes (C_N-V_Ga) introduce additional levels at E_C - 0.6 eV 19. These deep levels effectively compensate residual shallow donors (primarily oxygen and silicon impurities at concentrations of 10^16-10^17 cm^-3), resulting in semi-insulating behavior when carbon concentration exceeds 3×10^17 cm^-3 9.

The compensation ratio (N_A/N_D) in carbon-doped GaN typically ranges from 0.8 to 1.2 for semi-insulating applications, with precise control achieved through carbon concentration adjustment 9. Temperature-dependent Hall effect measurements demonstrate that electron concentration decreases from 10^16 cm^-3 at 300K to below 10^14 cm^-3 at 77K in optimally compensated material, confirming the deep nature of carbon-related acceptors 9.

Optical Quenching Phenomena In Carbon Doped GaN Photoconductive Switches

A unique property of carbon-doped gallium nitride is its optical quenching behavior, enabling sub-nanosecond switching in photoconductive semiconductor switches (PCSS) 1. When C-doped GaN (carbon concentration 2-8×10^17 cm^-3) is illuminated with ultraviolet light (355-385 nm, photon energy >3.2 eV), photoexcited carriers are generated, transitioning the material from an insulating state (resistivity >10^8 Ω·cm) to a conductive state (resistivity 10^2-10^4 Ω·cm) within picoseconds 1.

The optical quenching mechanism involves infrared illumination (850-1550 nm) that selectively excites electrons from carbon deep levels to the conduction band, where they rapidly recombine with holes at carbon acceptor sites, effectively "quenching" the photoconductivity 1. This process occurs with time constants of 200-800 picoseconds, enabling switching frequencies exceeding 1 GHz 1. Key performance metrics include:

• On-state current density: 500-2000 A/cm² at 10 kV bias for 100 μm gap devices 1
• Off-state leakage: <1 nA/cm² at 10 kV bias 1
• Optical trigger energy: 50-200 μJ/cm² at 365 nm for turn-on; 100-500 μJ/cm² at 1064 nm for quenching 1
• Switching speed: <500 ps rise time; <800 ps fall time with IR quenching 1

This optical quenching capability is absent in undoped or silicon-doped GaN, making carbon doping essential for high-speed, high-power PCSS applications in pulsed power systems and directed energy weapons 1.

Carbon Doped Gallium Nitride In Power Electronic Devices

High-Electron-Mobility Transistors (HEMTs) With Carbon-Doped Buffer Layers

Carbon-doped GaN buffer layers have become the industry standard for high-voltage AlGaN/GaN HEMTs, addressing critical challenges of vertical leakage current and dynamic on-resistance degradation 4. In typical HEMT structures, a carbon-doped GaN buffer (1-3 μm thick, carbon concentration 3-8×10^18 cm^-3) is grown on a transition layer stack (AlN/AlGaN superlattice or composition-graded AlGaN) above a silicon, silicon carbide, or sapphire substrate 4.

The extrinsically carbon-doped buffer provides multiple performance enhancements:

1. Vertical breakdown voltage improvement: Devices with carbon-doped buffers (5×10^18 cm^-3) exhibit breakdown voltages of 900-1200V compared to 600-750V for unintentionally doped buffers, representing a 40-60% improvement 4
2. Reduced buffer leakage: Vertical leakage current density decreases from 10^-4 A/cm² to below 10^-7 A/cm² at 600V bias 4
3. Dynamic on-resistance stability: Carbon doping suppresses electron trapping in buffer states, reducing dynamic R_on degradation from 35-50% to less than 10% under hard-switching conditions 4
4. Improved current collapse immunity: Drain lag and gate lag effects are reduced by 60-80% in carbon-doped buffer HEMTs 4

Optimization of carbon doping profiles within the buffer layer further enhances performance 4. A graded carbon profile with peak concentration (8×10^18 cm^-3) positioned 200-500 nm below the channel interface and decreasing to 2×10^18 cm^-3 at the buffer-substrate interface provides optimal trade-off between vertical breakdown and lateral conductivity modulation 4.

P-Type Carbon-Magnesium Co-Doped GaN For Vertical Power Devices

For vertical GaN power devices such as p-n diodes, p-i-n diodes, and junction field-effect transistors (JFETs), carbon co-doping with magnesium in p-type layers addresses the challenge of high p-type resistivity 2619. Conventional magnesium-doped p-GaN exhibits resistivity of 5-15 Ω·cm due to low hole activation efficiency (2-5% at room temperature) and hydrogen passivation of acceptors 2.

Carbon-magnesium co-doping employs carbon concentrations of 1-3×10^17 cm^-3 alongside magnesium concentrations of 5-15×10^18 cm^-3, achieving:

• Hole concentration: 5-9×10^17 cm^-3, representing 50-80% increase over Mg-only doping 26
• Resistivity: 1.2-2.8 Ω·cm, a 60-75% reduction compared to conventional p-GaN 26
• Activation energy: 140-165 meV, slightly reduced from 170-190 meV in Mg-only films due to carbon-induced lattice strain effects 2

The mechanism involves carbon atoms creating local strain fields that reduce magnesium-hydrogen binding energy, facilitating hydrogen desorption during post-growth annealing at 700-850°C 26. Additionally, carbon incorporation suppresses formation of magnesium-nitrogen vacancy complexes that act as compensating donors 2. Secondary ion mass spectrometry depth profiling reveals that carbon distribution remains uniform (±10%) throughout the p-type layer, while magnesium exhibits slight surface accumulation (20-30% higher concentration in top 50 nm) 6.

Applications Of Carbon Doped Gallium Nitride Across Industries

Optoelectronic Devices: LEDs And Laser Diodes

In III-nitride optoelectronic devices, carbon doping serves dual roles: as a co-dopant in p-type cladding layers and as a compensating dopant in n-type layers to control carrier concentration 919. For blue and ultraviolet light-emitting diodes (LEDs), carbon-magnesium co-doped p-GaN contact layers (carbon: 8×10^16-2×10^17 cm^-3; magnesium: 8×10^18-2×10^19 cm^-3) reduce operating voltage by 0.3-0.6V at 20 mA drive current compared to Mg-only doped layers, translating to 8-15% improvement in wall-plug efficiency 19.

In laser diode structures, precise carbon concentration control in n-type cladding layers prevents parasitic p-type conversion while maintaining low resistivity 9. Carbon concentrations below 1×10^18 cm^-3 in n-AlGaN cladding layers (aluminum composition 5-15%) ensure electron concentration remains above 5×10^17 cm^-3, critical for current spreading and optical mode confinement 9. Temperature-dependent photoluminescence studies confirm that carbon-related deep levels do not introduce significant non-radiative recombination centers when concentration is maintained below this threshold 919.

Specific performance improvements in carbon-optimized optoelectronic devices include:

• LED forward voltage: Reduced from 3.4V to 2.9V at 20 mA for 450 nm blue LEDs 19
• Laser threshold current density: Decreased from 3.5 kA/cm² to 2