Gallium phosphide

Studies investigating the interactions of gallium phosphide with other materials have revealed insights into its reactivity and stability. Theoretical investigations using density functional theory have explored the bonding characteristics of gallium phosphide with various ligands, indicating that certain ligands can enhance the stability of gallium phosphide monomers . Additionally, interactions with metal carbonyls have been studied to understand potential catalytic applications.

Metal-Organic Chemical Vapor Deposition (MOCVD) on Silicon Substrates

MOCVD remains the dominant technique for GaP epitaxy on silicon due to its scalability and precision in layer-by-layer growth. The process employs volatile precursors such as trimethylgallium (TMGa) and phosphine (PH₃), which decompose on heated silicon substrates to form crystalline GaP. Early challenges in lattice mismatch (≈4%) between GaP and Si were mitigated through graded buffer layers and substrate offcutting. For instance, GaP grown on Si (001) substrates with a 2° miscut demonstrated a 30% reduction in threading dislocations compared to on-axis substrates.

Recent studies highlight the role of growth temperature in defect formation. At 540°C, GaP nanostructures with diameters of 40–80 nm and heights of 5–15 nm were achieved, exhibiting quantum confinement effects evidenced by a 0.15 eV blue shift in photoluminescence spectra. However, residual oxygen incorporation (up to 5×10¹⁷ cm⁻³) from precursor decomposition remains a concern, necessitating in-situ purification steps.

Table 1: MOCVD Growth Parameters for GaP on Silicon

Molecular Beam Epitaxy (MBE) for Nanowire Doping Optimization

MBE enables precise doping control in GaP nanowires (NWs), critical for tailoring electronic properties. Beryllium (Be) and silicon (Si) serve as primary p-type and n-type dopants, respectively. Be-doped GaP NWs exhibit hole concentrations up to 2×10¹⁹ cm⁻³, while Si doping achieves electron densities of 5×10¹⁸ cm⁻³. A 2024 study demonstrated that external electric fields (1–5 kV/cm) applied during MBE growth enhance dopant incorporation by 40%, suppressing self-compensation mechanisms.

Axially heterostructured GaPN/GaP NWs grown via plasma-assisted MBE show promise for flexible optoelectronics. These structures, embedded in silicone membranes, exhibit room-temperature photoluminescence at 1.45–1.50 eV, with nitrogen content varying between NWs and parasitic layers due to vapor-solid vs. vapor-liquid-solid growth dynamics.

Key Challenges:

• Dopant Segregation: Be tends to accumulate at NW tips, creating non-uniform axial profiles.
• Radial vs. Axial Growth: High Be fluxes (>1×10¹⁸ cm⁻³) shift growth from diffusion-limited (L ∝ 1/D²) to nucleation-limited regimes, altering NW aspect ratios.

Liquid-Encapsulated Czochralski (LEC) Crystal Growth Dynamics

LEC growth produces bulk GaP crystals with diameters exceeding 75 mm, essential for substrate manufacturing. Indium doping (1.7 at%) reduces dislocation densities by 50% through lattice hardening, while crystal rotation rates (10–30 rpm) modulate oxygen incorporation. Undoped GaP crystals exhibit oxygen concentrations of 2×10¹⁷–5×10¹⁷ cm⁻³, increasing radially from edge to center.

Table 2: LEC-Grown GaP Crystal Properties

Thermal stress-induced cracking remains a bottleneck, mitigated by post-growth annealing at 900°C under phosphorus overpressure.

Nanoheteroepitaxial Growth on Silicon Nanotip Arrays

Nanoheteroepitaxy on CMOS-compatible Si nanotips (NTs) enables monolithic integration of GaP devices. Using gas-source MBE, GaP islands (200–500 nm diameter) were selectively grown on NTs fabricated via reactive ion etching. These islands adopt a zinc-blende structure with (111) orientation, exhibiting room-temperature mobilities of 16 cm²/V·s and photoluminescence intensities comparable to bulk GaP.

Advantages:

• Strain Relaxation: NT geometry reduces interfacial strain by 70% compared to planar growth.
• Scalability: Demonstrated on 200 mm Si wafers, compatible with high-volume manufacturing.

Density Functional Theory Analysis of Nanocrystalline Systems

Density functional theory has become indispensable for predicting the electronic properties of GaP nanocrystalline systems. Traditional DFT methods using local density approximation (LDA) or generalized gradient approximation (GGA) functionals often underestimate bandgaps due to their incomplete treatment of exchange-correlation effects [6] [7]. However, hybrid functionals and range-separated hybrids, such as HSE06, improve accuracy by incorporating exact Hartree-Fock exchange. For instance, studies comparing DFT with quantum Monte Carlo (QMC) benchmarks reveal that range-separated hybrids reduce bandgap errors in GaP nanocrystals from ~30% (LDA) to <10% [4].

A key challenge lies in modeling quantum confinement effects in low-dimensional GaP structures. DFT simulations of GaP nanowires with diameters below 5 nm predict a transition from indirect to direct bandgap behavior due to spatial carrier localization [2] [3]. These findings align with experimental observations of strong photoluminescence at 594 nm in wurtzite GaP nanowires, which exhibit a direct bandgap of ~2.1 eV [3]. Computational studies further highlight the role of surface passivation: AlGaP shells grown on GaP cores suppress surface states, enhancing photoluminescence intensity by reducing non-radiative recombination [3].

Generalized Gradient Approximation for Valence Band Modeling

The generalized gradient approximation remains widely used for modeling GaP’s valence band structure despite its limitations in bandgap prediction. GGA-PBE calculations accurately reproduce the valence band maximum (VBM) at the Γ-point, which is dominated by phosphorus 3p orbitals hybridized with gallium 4s states [6] [7]. However, spin-orbit coupling (SOC) corrections are critical for refining these predictions. For GaP, SOC reduces the bandgap by 0.3–0.4 eV, primarily by lowering the conduction band minimum (CBM) at the X-point [6].

Recent advancements in meta-GGA functionals, such as SCAN and r2SCAN, improve upon GGA by including kinetic energy density dependencies. These functionals achieve better agreement with experimental bandgaps while maintaining computational efficiency [7]. For example, meta-GGA predicts a GaP bandgap of 2.2–2.3 eV, closely matching empirical values [1] [7]. Table 1 summarizes key computational results for GaP’s electronic structure.

Table 1: Bandgap Predictions for Gallium Phosphide Using Different DFT Functionals

Quantum Confinement Effects in Low-Dimensional Architectures

Quantum confinement in GaP nanostructures enables direct bandgap behavior by discretizing electronic states. Wurtzite GaP nanowires, synthesized via vapor-liquid-solid (VLS) growth, exhibit a direct bandgap of 2.18–2.25 eV due to structural anisotropy and reduced dimensionality [2] [3]. High-resolution transmission electron microscopy (HRTEM) confirms the hexagonal crystal structure, with fewer than one stacking fault per micrometer [3]. These nanowires emit light at 594 nm (yellow-orange) with a short radiative lifetime of ~1 ns, characteristic of direct transitions [2].

Alloying further tunes emission wavelengths: incorporating aluminum (Al) into GaP increases the bandgap to 2.97 eV (blue-violet), while arsenic (As) alloying reduces it to 1.8 eV (infrared) [3]. Such ternary systems (e.g., AlₓGa₁₋ₓP, GaAsᵧP₁₋ᵧ) enable spectral coverage from 555 nm (green) to 690 nm (red), addressing the “green gap” in light-emitting diodes [2] [3].

Defect-State Engineering for Intermediate Band Formation

Defect engineering introduces intermediate bands within GaP’s bandgap, enhancing photon absorption and carrier multiplication. Nitrogen doping, for instance, creates a trap state 0.3 eV above the valence band, enabling sub-bandgap photon absorption [1]. Similarly, sulfur (S) and tellurium (Te) act as n-type dopants by introducing shallow donor levels, while zinc (Zn) generates p-type conductivity via acceptor states 0.06 eV above the valence band [1] [5].

Recent theoretical studies explore ligand-stabilized GaP monomers as precursors for defect-engineered materials. Cyclic alkyl(amino) carbene (cAAC) ligands stabilize GaP units with bond dissociation energies of 137 kcal/mol, facilitating controlled defect incorporation during synthesis [5]. These defects form intermediate bands that enable multi-exciton generation, potentially increasing solar cell efficiency beyond the Shockley-Queisser limit.

The integration of gallium phosphide with silicon presents unique challenges despite the relatively small lattice mismatch between these materials. Silicon possesses a lattice constant of 5.43 Å, while gallium phosphide exhibits a lattice constant of 5.50 Å, resulting in a lattice mismatch of approximately 0.37-0.4%. Although this mismatch is significantly smaller than other third-fifth semiconductor combinations, it still poses substantial challenges for heteroepitaxial growth.

The heterovalent nature of the gallium phosphide silicon interface creates additional complexity beyond the simple lattice mismatch consideration. The transition from the non-polar silicon substrate to the polar gallium phosphide epilayer generates interfacial charge imbalances that lead to the formation of localized electronic states within the common bandgap. These states arise from the excess charge associated with heterovalent bonding at the silicon gallium and silicon phosphorus interfaces.

Stacking fault pyramids represent a significant defect mechanism in gallium phosphide nucleation layers on silicon. These defects originate at the gallium phosphide silicon interface and propagate along all four {111} planes. Initial nucleation processes can generate stacking fault pyramid densities as high as 6 × 10^7 cm^-2, which subsequently interact with gliding misfit dislocations during lattice-mismatched growth, thereby enhancing the threading dislocation density.

The pseudomorphic critical thickness for gallium phosphide on silicon typically ranges from 30-100 nanometers, beyond which plastic relaxation occurs through the formation of misfit dislocations. This critical thickness depends on various factors including growth temperature, surface preparation, and doping conditions. Nitrogen-doped gallium phosphide has been demonstrated to achieve improved lattice matching to silicon, with the possibility of creating gallium phosphide nitrogen alloys that are nearly lattice-matched to silicon.

Threading dislocation densities in gallium phosphide on silicon are strongly influenced by doping type and concentration. N-type gallium phosphide exhibits significantly higher threading dislocation densities of approximately 3.1 × 10^7 cm^-2, nearly thirty times higher than both p-type and unintentionally doped gallium phosphide at approximately 1.1 × 10^6 cm^-2. This dramatic difference has been attributed to electronic effects that dominate over solute effects on dislocation velocity at typical doping concentrations.

Complementary Metal Oxide Semiconductor Compatible Wafer Bonding Techniques

Direct wafer bonding represents a promising approach for integrating gallium phosphide with silicon substrates while maintaining complementary metal oxide semiconductor compatibility. This technique involves bringing clean, activated surfaces into intimate contact, followed by thermal annealing to strengthen the bond. The process typically employs hydrophilic surface activation through oxygen plasma exposure or chemical treatments to populate the surface with hydroxyl groups at sufficient density to initiate bonding.

Surface activated bonding offers advantages for gallium phosphide silicon integration by enabling room temperature bonding without requiring high-temperature annealing processes. This technique utilizes argon fast atom beam sputtering to clean and activate the surfaces in ultra-high vacuum conditions, followed by application of high bonding pressures of several tens of megapascals. The resulting bonds achieve strength close to the bulk fracture energy of the materials without thermal treatment.

Gallium phosphide on insulator structures have been successfully fabricated using direct wafer bonding of gallium phosphide aluminum gallium phosphide gallium phosphide heterostructures onto silicon dioxide on silicon wafers. The process involves atomic layer deposition of aluminum oxide as a bonding layer, followed by megasonic cleaning with ozone-rich deionized water to ensure hydrophilic, contamination-free surfaces. Subsequent annealing at 300°C for 2 hours increases the bonding energy while remaining compatible with complementary metal oxide semiconductor processing.

Micro-transfer printing has emerged as a novel approach for versatile gallium phosphide silicon integration. This technique enables selective, localized integration with complementary metal oxide semiconductor circuits without requiring large-scale material coverage. The process involves photolithographic patterning and plasma etching to create gallium phosphide structures, followed by elastomeric transfer printing onto target substrates. This approach addresses the limitations of traditional wafer bonding techniques that require full wafer coverage.

Multilayer transfer processes have been developed for integrating gallium phosphide with silicon complementary metal oxide semiconductor structures. These processes typically involve double bonding and layer transfer procedures, where plasma-enhanced chemical vapor deposition oxide layers serve as bonding interfaces. The technique requires careful control of densification processes at temperatures around 600°C to eliminate residual gas molecules and byproducts incorporated during oxide deposition.

Strain Relaxation Mechanisms in Monolithic Integration

Plastic relaxation in gallium phosphide silicon heterostructures occurs through several distinct mechanisms that depend on layer thickness, growth conditions, and doping levels. The formation of misfit dislocations at the heterointerface represents the primary strain relaxation mechanism once the critical thickness is exceeded. These dislocations typically form along the ⟨011⟩ directions with a characteristic spacing of approximately 50-60 nanometers depending on the exact lattice mismatch and growth conditions.

Elastic strain relaxation occurs through lateral expansion and buckling mechanisms when gallium phosphide layers are grown on compliant substrates. The strain energy associated with the lattice mismatch can be accommodated through three-dimensional deformation of the epilayer, particularly when the interface area is minimized through nanostructured growth approaches. This mechanism is particularly effective for nanoheteroepitaxy where gallium phosphide islands are grown on silicon nanotips.

Dislocation filtering mechanisms have been developed to reduce threading dislocation propagation in gallium phosphide silicon structures. Compressively strained superlattices can effectively trap and annihilate threading dislocations through dislocation blocking and interaction processes. The implementation of strained short-period superlattices enables stepwise strain accommodation while maintaining two-dimensional growth modes that suppress three-dimensional island formation.

Thermal strain relaxation represents an additional consideration in gallium phosphide silicon integration due to the mismatch in thermal expansion coefficients. Silicon exhibits a thermal expansion coefficient of 2.6 × 10^-6 °C^-1, while gallium phosphide has a coefficient of approximately 5.7 × 10^-6 °C^-1. This mismatch can generate additional stress during thermal cycling processes, potentially leading to crack formation and dislocation multiplication.

Anisotropic strain relaxation has been observed in gallium phosphide containing heterostructures, where relaxation rates differ significantly between perpendicular crystallographic directions. The major relaxation axis typically corresponds to the [1̄10] direction regardless of the strain sign, while the dislocation density direction can vary between * and [1̄10]* depending on whether the strain is tensile or compressive.

Selective Area Growth on Patterned Silicon Substrates

Selective area epitaxy represents a sophisticated approach for achieving high-quality gallium phosphide integration on silicon substrates through substrate patterning and localized growth. This technique involves creating periodic arrays of growth sites using lithographic patterning and etching processes, followed by selective deposition of gallium phosphide only in the unmasked regions.

Nanoheteroepitaxy on patterned silicon substrates enables strain energy reduction through compliance effects and reduced interface areas. Silicon nanotips fabricated using complementary metal oxide semiconductor technology on 200 millimeter wafers can achieve island densities exceeding 10^8 per cm^2. The reduced lateral dimensions of these growth sites minimize the driving force for plastic relaxation and extended defect formation.

Microsphere lithography combined with anisotropic silicon wet etching provides a scalable method for creating wafer-scale surface patterning. This approach enables the formation of both planar and pyramidal pit nucleation sites with controlled dimensions and spacing. The technique can produce ordered arrays of gallium phosphide based nanostructures with compositional homogeneity across large substrate areas.

Mask selectivity in selective area growth depends critically on the growth temperature, precursor flux ratios, and surface preparation conditions. Silicon dioxide masks are commonly employed due to their chemical inertness and thermal stability under typical growth conditions. The selectivity can be enhanced through surface treatments and growth parameter optimization to suppress nucleation on masked areas while promoting growth in the openings.

Aspect ratio dependent growth in patterned substrates influences the final morphology and crystalline quality of gallium phosphide structures. Narrow openings with high aspect ratios tend to promote single-domain growth and reduced defect densities due to geometric confinement effects. However, complete filling of the patterned features requires careful optimization of growth kinetics and mass transport processes.