Silicon Nanotexture Surface Area Mapping Using Ultraviolet Reflectance

Giuseppe Scardera, D.n. Payne, Muhammad Umair Khan, Yù Zhang, Anastasia Soeriyadi, Shuai Zou, Daqi Zhang, Rasmus Schmidt Davidsen, Ole Hansen, Bram Hoex, Malcolm AbbottIEEE Journal of Photovoltaics 2021 · DOI

Abstract

The enhanced surface area of silicon nanotexture is a critical metric for solar cell integration, influencing optical reflectance, dopant diffusion, and surface recombination. This property is typically quantified by the Enhanced Area Factor (EAF)—the ratio of surface area to projected area. However, traditional EAF measurement techniques lack the statistical throughput needed for comprehensive correlation studies. This article demonstrates that silicon's dominant ultraviolet reflectance peak, $R(E_2)$ at ~280 nm, is highly sensitive to surface etching and correlates strongly with EAF. By applying this correlation to $R(280\text{ nm})$ reflectance mapping, the researchers achieved accurate, high-resolution full-wafer EAF spatial mapping. This technique is shown to be sensitive enough to correlate nanotexture variations with post-diffusion sheet resistance, a finding supported by finite-difference time-domain (FDTD) simulations.

What Problem Does This Paper Address?

Nanotexturing silicon, often referred to as "black silicon," is a primary method for reducing optical reflection in solar cells to near-zero levels. While beneficial for light trapping, this process significantly increases the silicon's surface area. This increase, measured as the Enhanced Area Factor (EAF), presents a trade-off: it improves optical performance but simultaneously increases surface recombination—which leads to efficiency losses—and complicates dopant diffusion during manufacturing.

Schematic illustrating the relationship between texture volume, void volume, and the projected area used to define the Enhanced Area Factor (EAF).

Measuring EAF accurately across a production-scale wafer is notoriously difficult. Current methods are either gravimetric (providing only a single average value for an entire wafer) or microscopic (using AFM or SEM to provide high detail over tiny, non-representative areas). Without spatial mapping, it is impossible for engineers to identify non-uniformities in the etching process or correlate localized texture variations with electrical performance.

Methods

The authors propose using the $E_2$ reflectance peak of silicon as a non-destructive proxy for surface area. This peak, occurring in the deep UV range (~270–285 nm), is a fundamental electronic "fingerprint" of silicon that is exceptionally sensitive to surface disturbances.

  • Sample Preparation: The team generated various nanotextures using two industry-standard methods: Reactive Ion Etching (RIE) and Metal Catalyzed Chemical Etching (MCCE).
  • Characterization: Atomic Force Microscopy (AFM) was used to establish "ground truth" EAF values for the different textures.
  • Optical Mapping: A scanning reflectance tool (PVTools LOANA) mapped full wafers at 280 nm and 550 nm (visible light) for comparison.
  • Simulation: 3D Finite-Difference Time-Domain (FDTD) simulations modeled 35 distinct nanopyramid geometries to theoretically validate the relationship between void volume, surface area, and UV reflectance.
SEM images showing the physical variety of silicon nanotextures studied, ranging from (a) coarser structures to (b) fine, high-aspect-ratio "black silicon" features.

Key Insights and Ideas

  • UV Sensitivity vs. Visible Light: The $E_2$ peak at 280 nm is significantly more sensitive to surface area changes than visible light (550 nm). While visible reflectance often saturates or becomes noisy at high EAF, UV reflectance follows a predictable exponential decay.
  • The "Decay Band": FDTD simulations revealed that while different physical shapes (morphologies) affect reflectance slightly differently, they all fall within a narrow "decay band." This suggests the UV method is robust across various etching recipes.
  • Predicting Electrical Performance: The study demonstrated that EAF maps could predict variations in sheet resistance after doping. Higher EAF leads to higher phosphorus uptake, resulting in lower sheet resistance—a variation captured clearly by the UV map.
  • Grain Sensitivity: In multicrystalline silicon, the technique was sensitive enough to resolve grain-dependent EAF variations caused by different crystal orientations etching at different rates.

Key Results

  • Strong Correlation: The study established a clear exponential decay relationship between $R(280\text{ nm})$ and EAF with an $R^2$ value of 0.987.
  • High Resolution: The method produced full-wafer EAF maps with 250 $\mu$m spatial resolution, far exceeding traditional microscopy.
  • Validation: UV-derived EAF values showed excellent agreement with AFM measurements, with most values falling within a 4% deviation.

Limitations and Future Work

While the "decay band" is narrow, the authors note that the relationship is not perfectly universal; different etching technologies (e.g., very deep vs. very shallow pores) might require specific calibration curves. Additionally, AFM tips may struggle to reach the bottom of extreme high-aspect-ratio features, potentially underestimating EAF in those specific cases. The next step involves integrating UV reflectance mapping into inline production tools to allow real-time monitoring of texture uniformity during solar cell manufacturing.

Summary updated

Abstract

The enhanced surface area of silicon nanotexture is an important metric for solar cell integration as it affects multiple properties including optical reflectance, dopant diffusion, and surface recombination. Silicon nanotexture is typically characterized by its surface-area-to-projected-area ratio or enhanced area factor (EAF). However, traditional approaches for measuring EAF provide limited statistics, making correlation studies difficult. In this article, silicon's dominant ultraviolet reflectance peak, R(E2), which is very sensitive to surface etching, is applied to EAF spatial mapping. A clear decay correlation between R(E2) and EAF is shown for multiple textures created using reactive ion etching and metal catalyzed chemical etching. This correlation is applied to R(280 nm) reflectance mapping to yield accurate, high-resolution full-wafer EAF spatial mapping of silicon nanotextures. R(280 nm) mapping is also shown to be sensitive enough to correlate the impact of nanotexture spatial variation on post-diffusion sheet resistance. Finite-difference time-domain simulations of several nanoscale pyramid textures confirm a decay band for R(E2) versus EAF, consistent with our measurements. We suggest that R(E2) mapping may prove useful for other silicon nanotexture properties and applications where EAF is important.
BibTeX
@article{scardera2021silicon,
  author = {Giuseppe Scardera and D.N. Payne and Muhammad Umair Khan and Yù Zhang and Anastasia Soeriyadi and Shuai Zou and Daqi Zhang and Rasmus Schmidt Davidsen and Ole Hansen and Bram Hoex and Malcolm Abbott},
  title = {{Silicon Nanotexture Surface Area Mapping Using Ultraviolet Reflectance}},
  journal = {IEEE Journal of Photovoltaics},
  year = {2021},
  doi = {10.1109/jphotov.2021.3086439},
}
Figure