🔬 materials science

Single photon emitters in hBN: Limitations of atomic resolution imaging and potential sources of error

This study demonstrates that identifying single-photon emitters in hexagonal boron nitride using atomic-resolution ADF-STEM is fundamentally limited by sample thickness (beyond ~17 layers) and susceptible to misidentification due to residual threefold astigmatism, rendering the technique unreliable for the thicker flakes typically used in photonic research.

Original authors: David Lamprecht, Shrirang Chokappa, Alissa M. Freilinger, Barbara Maria Mayer, Maximilian Melchior, Jana Dzíbelová, Darwin Lorber, Luiz H. G. Tizei, Mathieu Kociak, Clemens Mangler, Lado Filipovic, Ja

Published 2026-07-22

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: David Lamprecht, Shrirang Chokappa, Alissa M. Freilinger, Barbara Maria Mayer, Maximilian Melchior, Jana Dzíbelová, Darwin Lorber, Luiz H. G. Tizei, Mathieu Kociak, Clemens Mangler, Lado Filipovic, Jani Kotakoski

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 you are trying to find a specific, tiny typo in a massive, multi-layered book. This book isn't made of paper, but of a super-thin, super-strong material called hexagonal boron nitride (hBN). Scientists are obsessed with this material because it might hold the secret to "single-photon emitters"—tiny light bulbs that flash just one photon at a time. These are the building blocks for the future of quantum computers and ultra-secure communication. The problem is, these light bulbs are hidden inside the book's pages, and we don't know exactly which letter (or atom) is causing the flash. To find them, researchers use a super-powerful microscope called ADF-STEM. Think of this microscope as a flashlight that can see individual atoms. If an atom is missing (a vacancy) or swapped for a different type (like carbon), the flashlight should see a change in brightness, like a dimmer spot on a page.

However, there's a catch. The "book" of hBN is stacked in a very specific, alternating pattern, like a sandwich where the top and bottom slices are slightly offset. When the book gets too thick, the flashlight beam gets messy, and the shadows start to blur. The big question scientists have been asking is: "Can we still spot these tiny atomic typos if the book is thick, like the ones used in real quantum experiments?" This new paper acts like a reality check for that question. It suggests that for many of the thick samples people are using, the answer might be a hard "no," and worse, the microscope might be tricking us into seeing things that aren't there.

The Great Atomic Hide-and-Seek

So, you want to find a single carbon atom hiding inside a stack of boron and nitrogen atoms? That's the mission. Scientists have been using a high-tech electron microscope (ADF-STEM) to take pictures of these stacks, hoping to spot the "dimmer" or "brighter" spots that would reveal a defect. The idea is simple: if you swap a heavy atom for a lighter one, or remove one entirely, the image should change. But this paper, led by David Lamprecht and his team, says that for the thick samples usually used in these studies, the game is rigged.

The Thickness Trap
The researchers ran detailed computer simulations to see how deep they could "see" into the hBN stack. They found that the microscope works great for thin slices—like a single layer or a few layers. But as soon as the stack gets past about 17 atomic layers (which is roughly 6 nm thick), the picture gets fuzzy. At this point, the difference between a boron atom and a nitrogen atom becomes so small that the microscope can't tell them apart anymore. It's like trying to distinguish between two shades of blue paint when they are mixed together in a thick bucket; the colors just blur into one.

Even worse, if you are looking for a single carbon atom (which is a prime suspect for creating those quantum light flashes), the microscope loses the ability to spot it even sooner. The simulations showed that single carbon atoms become invisible or indistinguishable from the background at thicknesses of just 7 layers (about 2.3 nm). This directly challenges recent claims that scientists have successfully identified single carbon atoms in flakes as thick as 210 layers. The paper suggests those identifications might be optical illusions caused by the thickness.

The "Ghost" Contrast
Here is where it gets really tricky. You might look at a picture of a thick hBN sample and see a clear pattern of bright and dark spots, thinking, "Aha! I found the defects!" The paper argues that this pattern might be a ghost.

The culprit is a sneaky microscope error called threefold astigmatism. Imagine looking through a pair of glasses that are slightly bent in a weird, three-pointed star shape. Even if the glasses are mostly clear, they can make a round object look like a triangle, or make one side of an object look brighter than the other. In the microscope, this error can create "artificial contrast." It can make a stack of atoms look like it has alternating bright and dark columns, even if the atoms are all identical and perfectly arranged.

The team proved this by intentionally messing with their microscope's settings. When they tweaked the "threefold astigmatism" knob, they could make the atoms look like they had a specific pattern, or even make a thick sample look like a thin one. It's like a magician using a special mirror to make a solid wall look like it has a secret door. The paper warns that this error is so subtle that even experts might miss it, leading them to "find" defects that don't actually exist.

The Real-World Test
To make sure their computer simulations weren't just guessing, the team went into the lab. They took real hBN flakes, some as thin as a single layer and others as thick as 15 nm (about 39–48 layers).

Thin samples: They could clearly see the atoms and even spot carbon atoms in the single-layer samples.
Thick samples: When they looked at the thick, "near-bulk" samples, the images were a mess. The atoms were blurry, and the patterns they saw were often just interference from the sample being slightly tilted or the microscope having those sneaky aberrations.

They also checked the "light" these samples emit. They found that the brightness of the light (cathodoluminescence) drops off sharply as the sample gets thinner, vanishing almost completely around 10 nm. This means that while the thick samples are bright and easy to study with light, they are terrible for finding the specific atomic cause of that light using a microscope.

The Bottom Line
The paper concludes that trying to match a specific atomic defect to a light flash in thick hBN samples using this type of microscope is a losing battle. The "signal" (the real defect) gets drowned out by the "noise" (thickness and microscope errors).

The authors aren't saying we should give up on finding these quantum emitters. Instead, they are saying we need to change our strategy. We can't just look at thick samples and hope the microscope will do the work. We need to use much thinner samples (where the atoms are easier to see) and combine the microscope with other tools, like spectroscopy (analyzing the energy of the electrons) or scanning probe techniques, to confirm what we are seeing. Without these extra checks, we might be chasing ghosts in the machine, mistaking a glitch in the lens for a breakthrough in quantum physics.

Technical Summary: Single Photon Emitters in hBN – Limitations of Atomic Resolution Imaging and Potential Sources of Error

Problem Statement
Hexagonal boron nitride (hBN) is a promising host material for single-photon emitters (SPEs) due to its robust, room-temperature emission and photostability. However, the microscopic origin of these emitters remains elusive, with candidates ranging from boron and nitrogen vacancies to carbon substitutional dopants. A primary challenge in identifying these defects is the spatial resolution mismatch between optical techniques (micrometer scale) and atomic-scale microscopy. Consequently, researchers often rely on annular dark-field scanning transmission electron microscopy (ADF-STEM) to correlate optical emission with specific atomic structures.

This approach faces significant hurdles:

Sample Thickness: Strong SPE emission is frequently observed in "near-bulk" hBN flakes (10–70 nm thick, or 30–210 layers), whereas atomic resolution imaging is most reliable in monolayers.
Stacking Complexity: hBN possesses an AA' stacking order with vertically alternating boron (B) and nitrogen (N) atoms. This creates periodic contrast modulations that complicate the differentiation of B and N columns as thickness increases.
Instrumental Artifacts: Residual aberrations, particularly non-radially symmetric ones like threefold astigmatism, can introduce artificial contrast patterns that mimic genuine structural features, leading to potential misidentification of defects.

Methodology
The authors employed a combined approach of advanced image simulations and experimental validation to systematically investigate these limitations:

Simulations: Multislice STEM image simulations were performed using the open-source tool abTEM. These simulations modeled pristine hBN and hBN containing specific defects (carbon substitutions and boron vacancies) across varying thicknesses (1 to 20+ layers). The simulations controlled variables such as kinetic energy (60 keV), probe size, and thermal diffuse scattering (frozen phonon model) while introducing specific residual aberrations (astigmatism, coma, threefold astigmatism) to isolate their effects.
Experimental Validation: Experiments were conducted using a 60 keV Nion UltraSTEM 100. The study utilized:
- ADF-STEM Imaging: Both medium-angle (MAADF) and high-angle (HAADF) detectors were used to image monolayer, bilayer, and multilayer (up to ~15 nm) hBN samples.
- Electron Energy-Loss Spectroscopy (EELS): EELS mapping around the carbon K-edge (290–310 eV) was used to chemically verify the presence of carbon atoms in defect clusters.
- Controlled Aberration Introduction: To demonstrate the impact of threefold astigmatism ( $A_{23}$ ), the authors manually detuned the corrector to introduce specific $A_{23}$ amplitudes and phases, observing the resulting contrast changes in bilayer and multilayer regions.
- Cathodoluminescence (CL): Thickness-dependent CL measurements were correlated with EELS thickness estimates to characterize emission intensity as a function of sample thickness.

Key Results

Thickness-Dependent Visibility Limits:
- Simulations and experiments demonstrate that under idealized conditions, the intensity difference between B and N columns becomes indistinguishable at thicknesses beyond 17 atomic layers (~6 nm).
- Carbon Substitutions: Single carbon atoms replacing boron become highly unreliable to identify beyond 7 layers (~2.4 nm). This challenges recent claims of identifying single carbon atoms in flakes as thick as 210 layers.
- Vacancies: While boron vacancies ( $V_B$ ) remain theoretically detectable up to ~20 layers in simulations, their detection becomes unreliable at thicknesses >10 nm (30 layers) due to noise and diminishing contrast.
Artificial Contrast from Threefold Astigmatism ( $A_{23}$ ):
- The study identifies residual threefold astigmatism as a critical source of error. Even small, well-aligned $A_{23}$ components can induce significant artificial contrast between neighboring atomic columns.
- Depending on the phase alignment relative to the crystal orientation, $A_{23}$ can reverse the apparent intensity contrast (making B appear brighter than N or vice versa) or flatten the contrast entirely, mimicking even-layer stacking in odd-layer samples.
- Experimental detuning confirmed that introducing $A_{23}$ creates artificial "B/N" contrast in near-bulk samples where none should exist, and can even create the visual appearance of alternative stacking orders (e.g., AB-stacking).
Limitations of Post-Processing and EELS:
- Standard image processing techniques (Gaussian and double Gaussian filtering) amplify artificial signals caused by aberrations, potentially leading to false-positive identification of defects.
- EELS mapping, while effective for monolayers, faces limitations in thicker samples. The carbon K-edge signal from a single substitutional atom is obscured by the background from plasmon tails and the boron K-edge, and is indistinguishable from surface hydrocarbon contamination.
Thickness-Dependent Emission:
- CL measurements show that emission intensity drops quadratically with decreasing thickness, vanishing around 10 nm. This suggests that the "bright" emitters often studied in thick flakes may not be present or detectable in the thin samples required for reliable atomic imaging.

Significance and Claims
The paper argues that the current reliance on ADF-STEM for atomic-resolution identification of SPEs in near-bulk hBN is fundamentally flawed due to intrinsic physical limits and instrumental artifacts. The authors claim:

Intrinsic Limits: Even under ideal conditions, atomic resolution imaging cannot reliably distinguish specific defect types (like single carbon substitutions) in hBN flakes thicker than a few nanometers.
Artifact Risk: Residual threefold astigmatism is a pervasive and insidious error source that can generate convincing but entirely artificial contrast, leading to the misidentification of defect types and stacking sequences.
Methodological Necessity: The correlation of atomic structures with optical properties requires a shift away from thick samples. The authors emphasize that reliable structural attribution necessitates the use of atomically thin samples combined with complementary techniques (such as STEM-EELS or scanning probe microscopy) under strictly controlled aberration conditions.
Generalizability: These limitations are not unique to hBN but apply to other layered materials with hexagonal symmetry, such as transition metal dichalcogenides and graphene, where residual aberrations can similarly distort contrast and lead to misinterpretation of phase or stacking.

In conclusion, the authors advocate for extreme caution when interpreting ADF-STEM images of multilayer hBN and call for advanced methodologies to bridge the gap between optical emission and atomic structure without relying on thick, artifact-prone samples.

The Great Atomic Hide-and-Seek

More like this