🔬 applied physics

Tracing boron diffusion into a textured silicon solar cell using electron beam induced current in a scanning transmission electron microscope

This paper presents a combined experimental and simulation framework using scanning transmission electron beam induced current (STEM-EBIC) to successfully trace and quantify boron dopant distributions beneath the pyramid-like textured surfaces of silicon solar cells, accounting for surface recombination effects to enable future measurements of weak electric fields.

Original authors: Tobias Meyer, David A. Ehrlich, Peter Pichler, K. L. Skrollan Detzler, Christoph Flathmann, Valeriya Titova, Tim Böckendorf, Hartmut Bracht, Jan Schmidt, Michael Seibt

Published 2026-07-14

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Tobias Meyer, David A. Ehrlich, Peter Pichler, K. L. Skrollan Detzler, Christoph Flathmann, Valeriya Titova, Tim Böckendorf, Hartmut Bracht, Jan Schmidt, Michael Seibt

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine a silicon solar cell not as a flat, boring sheet of glass, but as a microscopic mountain range. To catch more sunlight, scientists etch the surface into tiny, pyramid-shaped peaks. It's a brilliant trick for harvesting energy, but it turns the solar cell into a tricky maze for scientists trying to peek inside. They want to see exactly where the "boron" atoms (the secret ingredients that make the electricity flow) have settled after being baked into the silicon.

The problem? The usual tools for looking inside are like trying to map a cave with a flashlight from miles away—they just aren't sharp enough to see the details of these tiny pyramids. Other tools that are sharp enough often break the delicate structure or can't tell the difference between boron atoms that are actually doing their job and those that are just sitting there.

Enter the heroes of this story: a super-powerful microscope called a Scanning Transmission Electron Microscope (STEM) and a technique called Electron Beam Induced Current (EBIC). Think of the electron beam as a tiny, super-precise flashlight. When it shines on the solar cell, it wakes up the charge carriers (the electricity-makers), and the resulting current tells the scientists exactly where the action is happening.

The Big Discovery
The team, led by Tobias Meyer and his crew, did something they say is a "first": they managed to take a full, 2D picture of the boron distribution underneath these bumpy, pyramid-textured surfaces and compare it directly to a computer simulation. It's like taking a high-resolution photo of a hidden treasure map and then drawing the exact same map on a computer to see if they match.

They found that their computer model and the real-world photo matched up beautifully in the "neutral" zones—the flat, calm areas between the electrical junctions. The simulation showed that the boron diffused (spread out) in a way that created a "retrograde" profile, meaning the highest concentration of boron wasn't right at the surface, but about 0.12 µm (that's 0.00012 millimeters) deeper down.

The "Shift" and the Mystery
However, the match wasn't perfect everywhere. When they looked at the "space charge region"—the busy intersection where the electrical magic happens—they noticed a shift. The peak of the electrical signal in the experiment was about 80 nm (0.00008 mm) away from where the chemical junction (where the boron meets the silicon) was predicted to be in the simulation.

The authors suggest this isn't a mistake in their math, but rather a hint that something else is going on. They argue that surface charges (tiny electric charges sitting on the surface of the material) might be messing with the electric field, pushing the signal slightly off-center. They also note that their simplified model of how atoms recombine (cancel each other out) might be missing some complex details about different types of defects. They explicitly rule out the idea that the sample was just misaligned with the electron beam as the cause, because the geometry of the trench makes that unlikely.

How They Did It
To get these results, they had to be incredibly careful. They sliced a tiny piece of the solar cell (a "lamella") using a focused ion beam, which is like using a microscopic scalpel. They made sure to cut vertical trenches to prevent electrical short circuits, kind of like building a moat around a castle to keep the water from flooding the whole kingdom.

They simulated the boron diffusion process step-by-step, mimicking the real furnace conditions: heating it up to 941 °C, flowing in gases to deposit boron, and then cooling it down. They tuned their simulation until the sheet resistance (a measure of how hard it is for electricity to flow) matched a flat reference sample they made at the same time, and until the position of the chemical junction matched their measurements.

What They Are Sure Of (and What They Aren't)
The paper is very sure that their method works for mapping the boron in these textured areas. They proved that by using an "effective model" that projects the 3D world into a 2D simulation, they can get a great match with the experimental data.

However, they are careful not to claim they have solved the mystery of the space charge region completely. They suggest that the difference between the simulation and the experiment in that specific area might be due to surface charges or more complex recombination rules that their current model doesn't fully capture. They propose that in the future, combining this technique with other methods like electron holography could help measure these weak electric fields even more precisely.

In short, they've built a powerful new way to "see" inside the microscopic mountains of a solar cell. They've shown that their computer models can predict where the dopants are with high accuracy, but they've also found a small, intriguing discrepancy that suggests there's still a little bit of electrical magic left to uncover.

Technical Summary: Tracing Boron Diffusion in Textured Silicon Solar Cells via STEM-EBIC

Problem Statement
The characterization of dopant diffusion in modern optoelectronic devices, particularly those with strong surface textures like pyramid-like structures in silicon solar cells, presents significant challenges. Macroscopic techniques such as electrochemical capacitance-voltage, SIMS, or plasma time-of-flight mass spectrometry lack the lateral resolution required for heterogeneous structures. Conversely, high-resolution methods like atom probe tomography or nano SIMS detect dopant atoms regardless of their ionization state, failing to capture the electrically active species critical for device performance. While Scanning Spreading Resistance Microscopy (SSRM) and Scanning Capacitance Microscopy offer high resolution, they suffer from calibration difficulties due to tip wear. Electron Beam Induced Current (EBIC) offers high spatial resolution and sensitivity to active dopants but is complicated by strong surface recombination effects, making quantitative interpretation in complex geometries difficult. Furthermore, a quantitative model specifically applicable to the Scanning Transmission Electron Microscope (STEM) geometry—utilizing thin, electron-transparent samples—has only recently been established.

Methodology
The authors employed a self-consistent framework combining experimental STEM-EBIC measurements with finite element method (FEM) simulations of boron diffusion and charge carrier transport.

Sample Preparation: A textured silicon solar cell with a boron-diffused emitter was prepared. A focused ion beam (FIB) was used to extract an electron-transparent lamella from the aluminum-covered area between two pyramids. Vertical cuts were made during preparation to prevent short circuits from redeposited material.
Experimental Characterization:
- STEM-EBIC: Experiments were conducted in an FEI Titan 80-300 E-TEM at 300 kV. An electron beam (42 pA) scanned the sample, and the short-circuit current was measured to generate 2D EBIC maps.
- SSRM: Cross-sectional SSRM measurements were performed on a cleaved and polished sample edge to provide a rough estimate of the chemical junction position.
- Reference: A planar reference sample was processed simultaneously to determine sheet resistance via four-point probe (4PP) measurements.
Simulation Framework:
- Diffusion Simulation: Using Sentaurus Process, the authors modeled the boron diffusion process (deposition, drive-in, post-oxidation, cool-down). They accounted for the growth of a boron glass layer and the segregation of boron into silicon and oxide. The model was calibrated using the planar sheet resistance and the chemical junction position derived from experimental data. An oxygen-enhanced diffusion (OED) scaling factor was applied to match experimental observations.
- EBIC Simulation: A 2D drift-diffusion model (omitting the beam propagation direction) was implemented in COMSOL Multiphysics. The model solved the van-Roosbroeck equations (Poisson and continuity equations) for electrostatic potential, electron, and hole concentrations.
- Recombination Modeling: To account for surface recombination in the thin lamella, an effective recombination model was employed. This model projects the system along the electron beam direction, treating surface and bulk recombination via effective lifetimes ( $\tau_h$ and $\tau_e$ ) within a Shockley-Read-Hall framework. This approach allows the comparison of entire 2D experimental maps with simulations.

Key Results

Experimental Observations: The STEM-EBIC map revealed a signal well below the aluminum contact, reflecting the surface texture but appearing flattened. Analysis of vertical profiles yielded effective diffusion lengths of 63 nm (p-side) and 103 nm (n-side), values significantly lower than bulk expectations due to surface recombination. SSRM measurements indicated a chemical junction depth of approximately 500–550 nm below the aluminum contact.
Simulation Agreement: The simulated boron distribution, calibrated to match the experimental sheet resistance and junction position, showed a retrograde profile with a maximum concentration ~0.12 µm below the surface.
EBIC Comparison: There was good agreement between the experimental and simulated 2D EBIC maps regarding signal shape and the position of the EBIC maximum. The decay lengths of the signal tails in the neutral regions also matched, validating the effective recombination model.
Discrepancies: A shift of approximately 80 nm was observed between the position of the maximum electric field (chemical junction) and the EBIC maximum. Additionally, deviations were noted within the space charge region, particularly on the n-side. The authors attribute these discrepancies to potential surface charges (e.g., from FIB preparation), simplified recombination terms, or 3D heterogeneity not captured in the 2D model.

Key Contributions

First Numerical Modeling of 2D STEM-EBIC Maps: The paper demonstrates, for the first time, that entire two-dimensional STEM-EBIC maps of heterogeneous, textured systems can be numerically modeled. This is achieved by using an effective recombination model that projects the 3D thin-sample geometry into a 2D framework.
Self-Consistent Framework: The authors established a workflow where diffusion simulations provide the dopant distribution input for EBIC simulations, which are then directly compared to experimental maps. This allows for the tracing of dopant distributions underneath complex surface textures.
Validation of Effective Models: The study confirms that effective models accounting for bulk and surface recombination can successfully describe EBIC signals in neutral semiconductor regions of thin samples, enabling quantitative comparisons between experiment and simulation.

Significance and Claims
The paper claims that this approach opens "unprecedented possibilities" for observing excess carrier dynamics, weak electric fields, and dopant distributions in nanoscale functional devices. The authors emphasize that while the current model shows excellent agreement in neutral regions, the discrepancies in the space charge region suggest that the effective recombination parameters may depend on the electric field.

The authors modestly suggest that once a model for electric field dependence is established, STEM-EBIC could serve as a highly sensitive method for measuring weak electric fields, potentially surpassing the sensitivity limits of complementary techniques like electron holography or 4D-STEM, particularly in low-doped substrates. The work highlights the value of obtaining feedback on dopant distributions in heterogeneous geometries at the nanometer scale, bridging the gap between high-resolution imaging and quantitative electronic property analysis.

More like this