Optimal 1.2‑nm Al₂O₃ Passivation Layer for Silicon p‑n Junctions: Correlated Structural and Electrical Analysis

Introduction

Reducing surface recombination in silicon p‑n junctions is essential for maximizing photocurrent generation in photovoltaic devices. Front‑side passivation protects the illuminated, non‑metallized region, while rear‑side passivation mitigates recombination at the contact interface. Conventional metal‑silicon contacts exhibit high recombination, prompting the use of either small‑area contacts with low local doping or thin tunneling dielectric layers for local passivation. Recently, carrier‑selective passivation layers have shown promise, allowing one carrier type to tunnel while blocking the other.

Al₂O₃ deposited by ALD is the industry standard for surface passivation because it delivers sub‑nanometer thickness control and introduces negative fixed charges that provide strong field‑effect passivation. The optimal Al₂O₃ thickness, however, remains debated, with reported values ranging from 0.24 nm to 30 nm. Many studies focus on surface recombination velocity (Sₑ) rather than device‑level metrics, underscoring the need for a direct correlation between interface structure and electrical performance.

In this work we deposit Al₂O₃ layers of 0.24–1.9 nm on implanted n⁺‑p Si junctions, employ HRTEM/EDX to confirm atomic‑scale thicknesses, and measure serial resistance, ideality factor, carrier lifetime, EQE, and PCE. The data reveal that 1.2 nm Al₂O₃ provides optimal passivation while keeping tunneling resistance low.

Methods

Device Fabrication

Implanted n⁺‑p Si junctions were fabricated on 4‑inch, boron‑doped p‑type (100) wafers (5–10 Ω·cm). Phosphorus ions (10¹⁴ at cm⁻², 180 keV) were implanted, followed by 900 °C annealing for 5 min to activate the dopants. An Al₂O₃ tunneling layer (0.24–1.9 nm) was deposited by thermal ALD (TMA/H₂O, 290 °C) in a PICOSUN R200 system; each cycle yields ~0.12 nm. A 1.5 nm native SiO₂ layer formed naturally on the Si surface. An 80 nm SiNₓ:H overlayer was deposited by PECVD (340 °C, 1 Torr, 10 W) and annealed at 650 °C for 10 min to drive H into the Si. Ti/Au (20/800 nm) finger electrodes were sputtered on the front side, and a 400 nm Ti/Au back contact was evaporated. Final annealing at 400 °C for 10 min ensured ohmic contacts.

Characterization

Cross‑sectional TEM samples were prepared by FIB on a FEI Helios dual‑beam Nanolab 600i. HRTEM, STEM‑HAADF, and STEM‑EDX were performed on a JEOL ARM200F microscope (200 kV). Images were acquired along the Si[011] direction to align the electron beam with the Al₂O₃/Si interface. Electrical measurements were conducted under AM 1.5 G illumination using an Oriel Sol3ATM simulator. To isolate series‑resistance effects, Sinton’s Suns‑Voc method was employed, providing an open‑circuit voltage sweep with a xenon flashlamp and neutral‑density filters.

Results and Discussion

HRTEM imaging confirmed the presence of an amorphous Al₂O₃ layer atop the native SiO₂, with thicknesses of 0.9, 1.2, and 1.9 nm for the 0.24, 1.0, and 1.9 nm ALD cycles, respectively. STEM‑HAADF combined with EDX maps verified the elemental distribution: bright contrasts correspond to Al‑rich Al₂O₃, while Si signals are absent within the oxide. Intensity profiles across the interface yielded accurate thicknesses consistent with the ALD deposition model.

Series resistance (Rₛ) remained constant (~1.1 Ω) for Al₂O₃ thicknesses up to 1.2 nm, then spiked to ~3.1 Ω at 1.9 nm, indicating the onset of significant tunneling resistance. Modeling with the tunneling probability equation showed that the barrier height (ϕ_B) of 2.08–3.5 eV (SiO₂/Al₂O₃) reproduces the observed Rₛ trend, confirming that thin layers (<1.2 nm) maintain negligible Rₜʰᵤ while providing effective field‑effect passivation.

The ideality factor (n) decreased from >2 (unpassivated) to ~1.5 as Al₂O₃ thickness increased, reaching a peak at 1.2 nm. At 1.9 nm, n rose sharply to 4, reflecting the detrimental impact of increased Rₛ. Carrier lifetime, derived from open‑circuit voltage decay, mirrored this behavior: lifetimes improved with thickness up to 1.2 nm and declined at 1.9 nm, likely due to reduced H diffusion from the SiNₓ layer during annealing.

EQE measurements showed the highest response in the 600–900 nm range for the 1.2 nm Al₂O₃ sample, with a pronounced drop at longer wavelengths due to rear‑side recombination. The measured PCE peaked at 5 % for 1.2 nm under standard illumination, but the corrected PCE (Rₛ = 0) remained around 11 % across all thicknesses, underscoring the intrinsic passivation quality.

Conclusions

ALD‑grown Al₂O₃ layers provide an effective tunneling passivation for silicon n⁺‑p junctions. Through combined HRTEM/EDX structural analysis and comprehensive electrical testing, we identify 1.2 nm as the optimal thickness that delivers strong field‑effect passivation while avoiding excessive series resistance. Although device efficiency was not fully optimized, the demonstrated 1.2 nm passivation strategy offers a clear pathway to enhance high‑efficiency silicon solar cells.